Global 19177

Page 1

Population, Family and Society in Pre-Modern Japan

12:27:16:11:09 Page 1 Page 2

Akira Hayami

12:27:16:11:09 Page 2 Page 3

The Collected Papers of Twentieth-Century Japanese Writers on Japan

VOLUME 4

Collected Papers

AKIRA HAYAMI

Population, Family and Society in Pre-Modern Japan

12:27:16:11:09 Page 3 Page 4

Series: COLLECTED PAPERS OF TWENTIETH-CENTURY JAPANESE WRITERS ON JAPAN

Volume 4 Akira Hayami: Population, Family and Society in Pre-modern Japan

First published in 2009 by GLOBAL ORIENTAL LTD PO Box 219 Folkestone Kent CT20 2WP UK

www.globaloriental.co.uk

ISBN 978-1-906876-09-8

All rights reserved. No part of this publication may be reproduced or transmitted in any form or by any electronic, mechanical or other means, now known or hereafter invented, including photocopying and recording, or in any information storage or retrieval system, without prior permission in writing from the publishers.

British Library Cataloguing in Publication Data A CIP catalogue entry for this book is available from the British Library

Set in Plantin 10 on 11.5pt by ReﬁneCatch Ltd, Bungay, Suﬀolk Printed and bound in England by CPI Antony Rowe Ltd, Chippenham, Wilts

12:27:16:11:09 Page 4 Page 5

Contents

Preface vii Foreword by Osamu Saito¯: Akira Hayami: a historiographical appraisal ix Introduction by Akira Hayami xvi

Prelude Two Historical Landmarks: [I] Philip II and Toyotomi Hideyoshi; [II] Francisco Xavier and Japan 3

Part I: Tokugawa Japan 9 1. Japan: Sixteenth, Seventeenth and Eighteenth Centuries 11 2. A Great Transformation: Social and Economic Change in Sixteenth and Seventeenth-century Japan 42 3. Japan in the Eighteenth Century: Demography and Economy 52 4. Industrial Revolution versus Industrious Revolution 64

Part II: Demography through the Telescope 73 5. The Population at the Beginning of the Tokugawa Period – An Introduction to the Historical Demography of Pre-industrial Japan 75 6. Population Trends in Tokugawa Japan: 1600–1868 99 7. Population Growth in Pre-industrial Japan 113 8. Population and Family in Crisis: A Study of Northeastern Japan in the Late Eighteenth Century 119 9. Japan in Transition from Tokugawa to Meiji: Population Changes 132

Part III: Demography through the Microscope 163 10. The Shu¯ mon Aratame Cho¯: Japan’s Population Registers (with L.L. Cornell) 165 11. Demographic Aspects of a Village in Tokugawa Japan 185 12. Class Diﬀerences in Marriage and Fertility Among Tokugawa Villagers in Mino Province 204 13. Labor Migration in a Pre-industrial Society: A Study Tracing the Life Histories of the Inhabitants of a Village 219

12:27:16:11:09 Page 5 Page 6

CONTENTS

Part IV: Family and Household 237 14. Size of Household in a Japanese County Throughout the Tokugawa Era (with Nobuko Uchida) 239 15. Regional Diversity in Demographic and Family Patterns in Pre-industrial Japan (with Kurosu Satomi) 280 16. Population and Household Dynamics: A Mountainous District in Northern Japan in the Shu¯ mon Aratame Cho¯ of Aizu, 1750–1850 (with Aoi Okada) 303

Part V: Epilogue 343 17. Another Fossa Magna: Proportion Marrying and Age at Marriage in Late Nineteenth-century Japan 345 18. Illegitimacy in Japan 362 19. Koji Sugi and the Emergence of Modern Population Statistics in Japan: the Inﬂuence of German Statistics 369

Index 377

12:27:16:11:09 Page 6 Page 7

Preface

his volume comprises a collection of essays in English that I published Tfrom 1966 to 2005 (the four collaborative essays are included with the permission of my co-authors). In addition to the collection of essays, I also published The Historical Demography of Pre-modern Japan 1 as a separate book, and Popu- lation, Society and Economy in Pre-modern Central Japan 2 at nearly the same time as the present volume. Also, as one of the authors and editors of Emergence of Economic Society in Japan, 1600–1859, 3 I wrote the Introduction4 and Chapter 7, ‘Demography and living standards’ (with Hiroshi Kito), but these two essays are not included in the present volume in their entirety. The essays in this book have been printed as chapters in books, or in magazines and journals published in Japan and in various publications abroad. Although I am the author of these works, it still takes me too much time to find my own essays on my shelves. This is, of course, due to my own disorganization, but it has simply become too difficult for me to find passages for quotations, and it is too much trouble to leaf through magazines where my articles appear in order to look for something. So when a friend of mine, Professor Erich Pauer of Philipps-University in Marburg, Germany, suggested having my work published by Global Oriental, it was just like the Japanese proverb Jigoku de hotoke ni au (‘Meeting Buddha in the midst of Hell’). I seized the opportunity without hesitation. Words cannot express the appreciation I feel to Professor Pauer and to Paul Norbury of Global Oriental. While I was honoured to receive such courtesy from these kind people, when I began reading through my collection of essays, I realized that there were several problems. The biggest was unavoidable, stemming from the fact that this collection of essays encompasses a very broad range. Putting them all into a single book made it almost impossible to find a suitable title. Thus, while the title of this book itself does not necessarily exactly match its content, several related essays have been grouped into a ‘Prelude’ and five parts. Without this, the contents of this book would have been as jumbled as an over- turned toy box, so I felt that some amount of organization was necessary. The early modern period is a division of Japanese history from the end of the sixteenth century — after the establishment of regimes under Oda Nobunaga and Toyotomi Hideyoshi to the Meiji Restoration in 1868, via the Tokugawa or Edo period. Of course, this classification is based upon the political events; actual history went on without those political distinctions. Given this, as the author of this book, I have included three essays that are concerned with incidents after 1868. It must, however, be added that timeframes of those three works bridge the early modern and modern periods.

vii

12:27:16:11:09 Page 7 Page 8

PREFACE

The essays on the social and economic history of Japan, which cover several centuries (Part I–1, 2, and 3) reflect my basic understanding of the structure and changes of the Edo period. The majority of the essays were related to the historical demography of the Edo period (1603–1868), my field of specialization (esp. Parts II and III); but there are also essays on family history, a closely related area (these are included in Part IV–14, 15, and 16). Part V deals with the population or family, but after the Meiji Restoration (1868). The second biggest problem was that the earliest essays were written in 1966 and 1967, while the most recent were published in journals in 2005, thereby covering a period of nearly forty years. There may be no researcher who, over the course of forty years, has not made some changes to his or her methodologies, outlook or historical perspectives. At the very least, the researcher acquires knowledge over time, resulting in changes in his or her methodologies and concepts. I am no exception. In particular, when dealing with history, new materials are discovered and new methodologies are developed, so the conclusions derived must change accordingly. It has taken courage for me to include my inexperienced essays from the 1960s in this book. The essays from the 1980s to the first half of the 1990s include successive new discoveries, and were written while I was overwhelmed by them. Over the past ten years my work has been based on fairly stable concepts, and can be linked to the standpoint I take today. I have received much support and cooperation from a number of acquaintances in producing to write and translate the essays included here; thus I would need a whole page just to list the names of those involved. If I am allowed to mention one name, I do not hesitate to identify Dr Jeong Mi Lee. Without her assistance, this book would never have got started.

Akira Hayami Tokyo, March 2009

NOTES

1. Tokyo: University Press of Tokyo, 2001 (originally published in Japanese by Iwanami Shoten, Publishers, Tokyo, 1977). 2. Nichibunken Monograph Series 10; Kyoto: International Research Center for Japanese Studies, 2009. 3. Edited with Osamu Saito and Ronald Toby, Oxford: Oxford University Press, 2004. 4. I invite readers to refer to this Introduction too.

PUBLISHER’S NOTE

Readers will readily appreciate that the contents of this volume have been drawn from many diverse sources, published over a period of some forty years. Consistency in style and presentation, romaniza- tion, the use of macrons and spelling, therefore, has only been possible on a chapter by chapter basis. It should also be noted that every eﬀort has been made by both Publisher and Author to seek clearances for republication where required. Where this has not been possible, the Publisher would welcome correspondence from such individuals, companies or organizations.

viii

12:27:16:11:09 Page 8 Page 9

Foreword

Akira Hayami: a historiographical appraisal

Osamu Saito¯ Professor of Japanese and Asian Economies, Institute of Economic Research, Hitotsubashi University, Tokyo

kira Hayami, the author of the essays assembled in this volume, is known Aas the doyen of Japanese historical demography. However, he started his academic career just as an economic and social historian, not a specialist in social science-oriented historical studies. In order to understand a role he played as the pioneer in this ﬁeld of social science history, therefore, I have to begin this appraisal by taking a cursory look at the state of historical studies in general at the time of his debut as a historian.

I Since Meiji times Japan’s historiography has been dominated by two traditions. The ﬁrst is the German tradition of historical research. It was introduced during the Meiji period into the Imperial University of To¯kyo¯’s curriculum by Ludwig Riess (1861–1928) who was a disciple of the celebrated nineteenth-century German historian Leopold von Ranke. As illustrated by his phrase that history should be nothing but to ‘show what actually happened’, Ranke’s method of historical research placed much emphasis on the use of primary data, as against that of secondary literature, and also on the critical study of the data through rigorous source criticism. It is this mode of historical research that Riess brought into Meiji Japan’s history circles, the spirit of which bore tangible fruit in the form of projects to collect and compile primary sources, on the one hand, and of the rise of professionalized, source-based history writing, on the other. All this, no doubt, played a crucial role in establishing the empirical study of history as a professional discipline in pre-war Japan. However, Ranke’s rejection of grand theories and, more importantly, his obsession with the individuality of each historical situation and with the uniqueness of every human agency, be it a person or a nation, eventually gave rise to a rather distinct tendency in professionalized history writing, one often ridiculed as ‘mindless empiricism’. As such, this Ranke-Riess tradition has tended to discourage the Japanese history student to conceptualize his or her ﬁndings and thus to get him- or herself divorced from emerging social sciences.1

12:27:16:11:09 Page 9 Page 10

FOREWORD

Such gaps were filled almost exclusively by Marxist thinking, and this has become the second tradition in Japanese historiography. Although the teaching of Marxist doctrines in higher education was often suppressed by the pre-war and wartime governments, Marxist literature attracted a number of intellectuals of the 1920s and 30s. The Great Depression and its disastrous impact on the countryside, the coexistence of capitalist and pre-capitalist sectors, growing income inequality—all called for explanation and conceptualization. It was, in those scholars’ eyes, Marxism that could offer them both. Obviously interpretations could differ even among Marxist intellectuals. There was, for example, an active debate between two schools of Marxists concerning how to interpret the development of Japanese capitalism since the Meiji Restoration of 1868.2 However, Marxian influence was not confined in modern economic history. Soon after the collapse of the wartime regime, Marxist scholarship grew in all areas of historical research. In ancient and medieval history, for example, the Marxian stage theory offered historians a framework of how to interpret early histories of a non- Western country, according to which, for example, one Marxist historian proposed to view measures such as nation-wide cadastral surveys (called kenchi or Taiko¯ kenchi ) conducted by Taiko¯ Hideyoshi a unifier of the country of the 1580s, a ‘feudal revolution’ that must have acted as a break from the medieval past.3

II What was lacking among Japanese historians in the period immediately after the Second World War was the knowledge about non-Marxian social sciences and research methodologies for social-science oriented history. It was against this historiographical background that Akira Hayami started his career as an economic and social historian at Keio¯ University. Probably he was in a rather unusual position. His mentor, Kentaro¯ Nomura, was a student of W. J. Ashley at Cambridge, who turned to Japanese economic and social history. Although he was never considered a social-science oriented or quantitative historian, he made efforts to collect as many as possible of the particular kinds of data, such as Goningumi-cho¯ (records of five household units), Mura meisai-cho¯ (village reports) and Shu¯ mon aratame-cho¯ (population registers),4 suggesting that he became more interested in patterned observations than in the individuality of a historical event or fact. And, while he showed no interest in Marxian approaches, it was he who first tried to make use of information contained in Shu¯ mon aratame-cho¯ in order to construct population statistics of a particular village.5 As emphasized in his own Introduction, Hayami’s encounter with fresh findings derived from studies that employed a new methodology of nominal record linkage, called ‘family reconstitution’, applied to large quantities of data, i.e. Christian parish registers of baptism, marriage and burial, first developed at the Institute National d’Etudes Démographiques in France and then elaborated by the Cambridge Group for the History of Population and Social Structure in England, had a great impact on his subsequent research. He realized how valuable Tokugawa Japan’s village population register could be as a source material for historical demographic research and thus decided to undertake a project on Shu¯ mon aratame-cho¯ demography, for which a new, powerful as well as

12:27:16:11:09 Page 10 Page 11

FOREWORD

time-consuming methodology is crucially important as the need for quantifica- tion is always evident in population studies. The area he chose for his first project was a county in central Japan, Suwa-county in Shinano province; based almost entirely on data from the village registers, he published a pioneering monograph in this social-science history field of historical demography.6 The collection of essays is divided into two parts. Part I employs an aggregative approach and is concerned with long-term trends in totals of Suwa county’s population and households and their demographic components. This is made possible by a rare collection of village population registers, which cover more or less the whole county, enabling us to go back to the seventeenth century. This superb set of registers allows us to calculate not only crude birth and death rates, but age structures, child-woman ratios and mean household size figures as well. The results show that, in contrast to the trend after the 1720s (for which national totals are available), population had grown at a considerable rate during the seventeenth century, a fact that was not explicitly recognized by any other historians.7 Part II of the book is a detailed account of one single village, Yokouchi, and it is this that set a standard for Tokugawa historical demography. The Yokouchi account is based on a good series of population registers which covers a long period from 1671 to the end of Tokugawa rule, but not beyond. Unlike European parish registers, Tokugawa Japan’s register gives us a base population because each registered book lists all the households and their members in a given community. This also implies that there is virtually no need to ‘reconstitute’ the family in the Tokugawa case, so that demographic measures such as age at marriage, completed family size, marital fertility rates, and age-specific mortality rates (and hence life tables) can easily be computed from the registers. Thus, despite a couple of serious drawbacks such as the one that infant deaths which took place between the two compilation dates are likely to have been unregistered, most of the measures supplied from European family reconstitution studies now became available for the first time in Japanese history circles as far as the period before the Meiji Restoration is concerned. There are a number of interesting findings.8 Two of such findings merit attention here. One is a secular decline in early childhood mortality, and the other a decline in marital fertility from the eighteenth century on. In this Suwa book, he even went on to suggest a possibility of the existence of a deliberate family limitation by means of infanticide in order to account for the latter observation.9 It is worth noting that there is a set of characteristics in this and other studies that followed. They use village population registers as principal data, cover a period up to the Meiji Restoration, and employ a family reconstitution-type methodology in which individual demographic events are aggregated in order to compute various rates. The results of such exercises are presented with a descriptive account accompanied by a number of tables and charts, followed by a speculation or two about relationships between population and economy. Here I should hasten to add that Hayami made an equally invaluable contribution to the quantitative study of family and household in the traditional period, since the very nature of the village population register enables us to conduct what the Cambridge Group has done on the basis of English listings of inhabitants and census enumerators’ books.10 Again, much of his effort in this area of inquiry was

12:27:16:11:09 Page 11 Page 12

FOREWORD

directed towards the placing of findings in broad contexts of economic history. As in population trends, observed changes were more substantial for the seventeenth than for later centuries, which according to Hayami’s view, constituted the core of what he calls the ‘Great Transformation’ of seventeenth-century Japan.11 Perhaps, this style of history writing reflects the fact that Hayami started his career as an economic and social historian. For example, the two findings mentioned above were interpreted in relation to changing economic conditions and living standards of the peasantry: the declining tendency in early childhood mortality must have reflected an improvement in the standard of living in Suwa, while the possibility of family limitation might be accounted for by Yokouchi families’ desire to keep the living standards they had attained in the course of economic change of the previous century. Hayami’s second project, which had started well before the publication of the Suwa book, was on a region called No¯bi in central Japan. The No¯bi region, located on river deltas, is much larger in size than Suwa, and collected sets of village population registers are generally better in quality. Probably for these reasons, it took him a much longer time to finish the work: the book on No¯bi was published in 1992.12 The No¯bi book, too, has both aggregative and detailed village-level chapters, and thus exhibits a similar set of characteristics: the use of population registers as primary sources of information, covering a period up to the Meiji Restoration, and the application of family reconstitution-type methodology, the results of which are presented with a number of tables and charts showing demographic rates. However, the No¯bi monograph is less speculative in terms of economic history-related interpretations; instead, a couple of new analytical innovations are introduced. The population register lists all individuals in the village, so that by tracing them from one register to another, we can compile a fairly complete record of mobility of a person over his or her life-cycle stages. By focusing on a single village called Nishijo, the quality of whose series of registers is exceptionally high, Hayami applied this exercise of ‘tracing the life histories’ to the study of migration, both out and return, and provided for us a number of fresh findings. He also took a close look at the effect of out-migration on demographic consequences such as the age at marriage and fertility rates, thus addressing the issue of interplay between variables within a demographic system.13 Another innovation Hayami introduced into the No¯bi project is the use of a landholding measure (called kokudaka, expressed in rice yields) as a proxy for wealth class of the peasantry in demographic analysis. The kokudaka itself has long been used widely by historians as a key variable in Japanese agrarian history, but by classifying demographic rates by landholding class he showed us many useful findings— useful when interpreting the observations, both temporal and cross-sectional.14

III In what sense, then, were the historiographical gaps that had existed in the period after the Second World War ﬁlled by the two books and the essays assembled in this collected volume? In his Introduction, Hayami notes that recent developments in the use of ‘sophisticated’ demographic and statistical methodology ‘far surpass the capabilities of a low-tech historian like myself’.15 This is too modest a

xii

12:27:16:11:09 Page 12 Page 13

FOREWORD

statement. It is true that he never showed interest in more ‘sophisticated’ measures developed by formal demographers, such as the Princeton indices of fertility

(If , Ig, Im and Ih), the Coale-Trussell indices of family limitation (M and m), and Brass’s logit model of life-table mortality, but given the situations of the 1940s and 50s, the introduction by Hayami of family reconstitution-type methods into Japan’s historical research was a revolution. His laborious compilation of data and calculation of rates and other measures produced a number of findings, which were far more robust as historical findings than any other observations based on documentary and other non-quantitative pieces of evidence. It was those empirical findings—such as the secular decline in child mortality—that served as antidotes against the conventional picture of Tokugawa history. Undoubtedly, therefore, the role Hayami played is to be found in research methodology.16 Admittedly, some of the propositions Hayami proposed still remain debatable or are no longer a consensus. For example, the practice of infanticide is no longer considered even by Hayami a universally observable phenomenon. Indeed, in his No¯bi book a criticism is made on Thomas C. Smith’s hypothesis of sex-selective infanticide.17 As for a scenario Hayami offered in order to account for observed changes in household and family in the seventeenth century, I set out a different interpretation. It is true that the average size of what was listed on the registers declined; however, some of those large units on the registers my have been multi- household compounds as farming units, while within each compound were separable living spaces for different family households. Thus, I suggest, it was the multi-household compound as a farming as well as living unit that disintegrated in the course of the seventeenth century, while the way in which the household was formed did not change—it remained a stem one.18 Another example is about Hayami’s estimate of population in 1600. Now several scholars do question its robustness since it would imply too high a growth rate for the seventeenth century.19 Given the situations in the 1970s and 80s, however, no one can deny that a number of empirical findings Hayami brought about were made possible by the revolution in research methodology. One may ask if Hayami’s methodological revolution led him to a new level of conceptualization, which had also been missing in Japan’s historiographical tradition in the pre-war and wartime periods. He did borrow concepts and notions from social sciences, especially economics, from time to time, but it seems to me that the levels of conceptualization as well as debating skills displayed in his past publications are not markedly high. One example is his discussion of the ‘industrious revolution’. Based on observations derived from the No¯bi project, he put forward a hypothesis that there was a conversion from ‘horsepower’ to ‘manpower’ in Japanese farming in the seventeenth and eighteenth centuries.20 He argues that despite technological progress, farming became more labour intensive and people had to work much longer hours towards the end of the Tokugawa period. I believe that many would agree as far as the latter proposition is concerned. However, we can question if the former statement is really necessary to explain the latter. What actually happened may well have been that while there was a modest increase in livestock and land infrastructure, it was followed by an even greater increase in labour input, which is indeed what Thomas C. Smith noted in his 1959 book.21 It is my view, therefore, that Hayami’s role as a pioneer

xiii

12:27:16:11:09 Page 13 Page 14

FOREWORD

in social science history was much greater in research methodology than in the area of conceptualization. A couple of additional points may be made in relation to what the Hayami revolution left out. First, I should like to emphasize that the richness of information contained in Japan’s population register has not yet been fully exploited. As noted earlier, the population register is a census-type list of village population taken every year. This means that whereas the rate of infant death calculated from this material is always understated, we can have far better information about death rates for higher ages. Also available is the information about divorce and remarriage, so that it is not difficult, for example, to relate the death of the partner to the question of remarriage in the Tokugawa past. These are the issues never pursued fully in European historical demography since Christian parish registers are unable to provide us with reliable information about death and remarriage in adult ages. By applying the family reconstitution-type methods straight to the Japanese population registers, such important issues went simply unnoticed in Hayami’s days. In other words, here is a very promising area for future research in both population and family history, and I should like to note that scholars of younger generations have now started to explore some of the issues in these research areas.22 Second, family reconstitution-type demography has often been considered a micro approach to the history of population as it ‘reconstitutes’ individual family histories in order to examine birth and death rates as well as ages at marriage. It is worth noting, however, that it is not a micro analysis in any methodological sense. Once families are ‘reconstituted’, all individual life histories are aggregated to compute demographic rates and averages such as the mean age at marriage. They will be disaggregated again by various factors, such as age, sex, cohort and wealth class, but such exercises are methodologically different from what is now called micro-data analysis. In micro-data analysis, we need not aggregate individual data: we can examine one variable in relation to another with respect to the individual observations as units of analysis. This approach, which benefits from recent advances in statistical methods such as event history techniques, actually enhances the analytical rigour when we explore the functional relationships between demographic, social and economic variables involved. Again, younger generations have launched a project along this new line of investigation. The project, which Hayami himself helped get started, is an international one called the Eurasian Population and Family History project and has already produced the first fruit. This is also a promising area of research for historians of younger generations who follow in Hayami’s footsteps.23

NOTES

1. For the classical German school and its crisis in the early twentieth century, see G.G. Iggers, Historiography in the twentieth century: from scientiﬁc objectivity to the postmodern challenge (Hanover, NH, 1997). He notes that ‘professionalization’ of historical studies in Japan took place, just as in most Continental European countries and the United States, at the end of the nineteenth century, ahead of Great Britain and the Netherlands (ibid. p. 27). 2. The two schools were called Ko¯za-ha and Ro¯no¯-ha, both of which had profound inﬂuences on economic and social historians even in the post-war period.

xiv

12:27:16:11:09 Page 14 Page 15

FOREWORD

3. Discussions that followed this proposal are called the Taiko¯ kenchi debate, in which the young Akira Hayami once took part (see his Introduction below). 4. Nomura compiled and published the first two sets of data he collected as Goningumi-cho¯ no kenkyu¯ (Tokyo, 1943) and Mura meisai-cho¯ no kenkyu¯ (Tokyo, 1949). 5. The results of this effort was published posthumously as Nomura Kenkyu¯ -kai Ko¯mi-mura Kyo¯do¯- kenkyu¯ -han, ‘O¯¯ gaki-hanryo¯ Mino-no-kuni Motosu-gun Ko¯mi-mura no koko¯ to¯kei’, Mita gakkai zasshi, vol. 53, nos. 10–11 (1960), pp. 978–1020. For more on the Shu¯ mon aratame-cho¯, see Part III, Chapter 10 below. 6. Kinsei no¯son no rekishi jinko¯-gaku-teki kenkyu¯ (Tokyo, 1973). The following paragraphs draw on Section II of Saito¯ Osamu, ‘Historical demography: methods and interpretations’, in Historical studies in Japan (VIII), 1988–1992, ed. The National Committee of Japanese Historians (Tokyo, 1995), pp. 77–89. 7. See also Part II, Chapter 5 below, which is an attempt to estimate the size of Japan’s population in 1600. 8. See Part III, Chapter 11 below, which summarises much of Part 2 of the Suwa book. 9. A similar but not exactly identical suggestion—that sex-selective infanticide was practised according to the parity and sex composition of existing children—was made by Thomas C. Smith in his analysis of population registers of a village in No¯bi, a region Hayami chose for his second field. See Thomas C. Smith, Nakahara: family farming and population in a Japanese village, 1717–1850 (Stanford, 1977). 10. See for example Part V, Chapter 14 below, which addresses the issues in the history of household and family based on Suwa data. See also Chapters 15 and 16 in Part V. 11. For this thesis, see Part 1, Chapter 2 below. 12. Kinsei No¯bi-cho¯ no jinko¯, keizai, shakai (Tokyo, 1992). 13. The results of this exercise were published in 1973 as an English-language article, which is included as Chapter 8 in Part II below. 14. See Part III, Chapter 12 below. 15. See below. 16. A similar ‘revolution’ was made in economic history research too. In the 1960s a research group called the Quantitative Economic History Group was formed, and Hayami was one of its founding members. Eventually the work done by its members re-wrote the Japanese economic history, resulting in the much acclaimed Iwanami series of Nihon keizai-shi, 8 vols (Tokyo, 1988–90). Its four-volume English-language series of The economic history of Japan, 1600–1990 is being published by Oxford University Press. So far published are: Volume 1: A. Hayami, O. Saito and R.P. Toby, eds., Emergence of economic society in Japan, 1600–1859 (Oxford, 2004); and Volume 3: T. Nakamura and K. Odaka, eds., Economic history of Japan, 1914–1955: a dual structure (Oxford, 2003). 17. Hayami Akira, Kinsei No¯bi-cho¯ (op. cit. n. 6 above), pp. 218–22. For a survey of evidence, see O. Saito, ‘Infanticide, fertility and “population stagnation”: the state of Tokugawa historical demography’, Japan Forum, vol. 4, no. 2 (1992), pp. 369–81, and Saito, ‘Methods and interpretations’ (op. cit. n.6 above). See also O. Saito, ‘Historical demography: achievements and prospects’, Population studies, vol. 50, no. 3 (1996), pp. 537–53, which places Japanese data and findings in more general frameworks of historical demography. 18. Saito¯ Osamu, ‘Dai-kaikon, jinko¯, sho¯no¯ keizai’, in Hayami Akira and Miyamoto Matao, eds., Nihon keizai-shi, vol. 1: Keizai shakai no seiritsu (Tokyo, 1988), pp. 171–215. 19. See for example Angus Maddison, The world economy: a millennial perspective (Paris, 2001), p. 237, and W.W. Farris, Japan’s medieval population: famine, fertility, and warfare in a transformative age (Honolulu, 2006). 20. See Part I, Chapter 4 below. It should be remembered that Hayami’s notion of an ‘industrious revolution’ is very different from Jan de Vries’ (see his The industrious revolution: consumer behavior and the household economy, 1650 to the present, New York and Cambridge, 2008). 21. Thomas C. Smith, The agrarian origins of modern Japan (Stanford, 1959). 22. To cite but a few from remarriage studies: O. Saito and K. Hamano, ‘The death of the partner, remarriage and family continuation in Tokugawa Japan: a village study’, Revista de Demografia Histórica, vol. 24, no. 2 (2006), pp. 155–78, and S. Kurosu, ‘Remarriage in a stem family system in early modern Japan’, Continuity and change, vol. 22, no. 3 (2007), pp. 429–58. 23. T. Bengtsson, C. Campbell, J. Lee, et al., Life under pressure: mortality and living standards in Europe and Asia, 1700–1900 (Cambridge, MA, 2004). See my review of this book in Population studies, vol. 59, no. 3 (2005), pp. 393–5. MIT Press have announced that the volumes on fertility, nuptiality and migration will follow.

12:27:16:11:09 Page 15 Page 16

Introduction

Akira Hayami

I o start with, I would like to look back over my research since its beginning. In TSeptember 1950, when Japan was still experiencing the effects of defeat in the Second World War, I graduated from the Department of Economics of Keio¯ University in Tokyo. I was twenty years old. For my senior paper, I studied the mercantile system of England in the seventeenth century, comparing the writings of John Wheeler, a representative of the concept of bullionism, which aimed to accumulate precious metals within a country, with those of Thomas Mun, who promoted mercantilism, which did not object to funds leaving a country through exports as long as the funds were eventually returned through trade. Every day I travelled to the library, which had been restored after the war, and there I experienced the fascination of research through my first real experience with academia. As I worked on my paper, my desire to conduct research using primary historical materials became insatiable. So upon graduation, I became a research assistant at the Attic Museum,1 a private research organization dealing with economic history, fishing history and folklore. At the institute, I worked on Fisheries Agency projects to study fishing villages and industries and to collect historical documents related to fishing people throughout Japan. I visited various localities to examine historical documents, transcribing document lists and important historical documents. At that time, we did not have microfilm technology, but eventually cameras that had been ‘made in occupied Japan’ became available on the market, and could be used to photograph historical documents. I stayed at the institute for three years, during which I worked with primary handwritten manuscripts from the Edo period on a daily basis. I became very interested in the way of life of the common people of that era, but the greatest thing I attained during that time was the ability to decipher the manuscripts of the era, many of which still exist. The Korean War ended in 1953, and austerity measures were applied to government finances. Funds for studying fishing villages were cut, and researchers had to find other work. I became an assistant in the Department of Economics at Keio¯ University, and for the following thirty-six years, until 1989, I worked there in research and education. The university’s economic history laboratory housed a large number of primary historical documents from the Edo period that had been

xvi

12:27:16:11:09 Page 16 Page 17

INTRODUCTION

collected by Kentaro¯ Nomura. We attended graduate seminars with this professor where these documents were read, and thanks to my three years of experience in reading such documents, it was not so difficult for me to understand the seminars. For a short time after transferring to Keio¯ University, I used these historical materials2 to study the history of farming villages in the Kanto¯ region (i.e. Tokyo and the surrounding prefectures), and I presented several papers. In regard to my study3 of Musashi-no-kuni, Toshima-gun, Tsunohazu-mura, located in what is now a newly emerging metropolitan centre of Tokyo with many high-rise buildings, I felt it was particularly significant as a rare example of a farming village that had been absorbed into an urban region. I also participated in the ‘Debate on the Cadastral Surveys by Taiko¯’,4 which shook up Japan’s academic community. Studies of land survey ledgers from early modern times eventually came to the conclusion that the unit for levying land taxes was the ‘village’,5 and in some cases, the unit for levying work service was also the ‘village’ – so the village was used as a unit for managing both land taxes and work servitude imposed on the land and the people.6 This is not only significant in terms of the land surveys; it is also of decisive importance with respect to understanding the fundamental characteristics of ‘pre-modern’ society. However, there has been almost no discussion of this matter in the half-century that has passed since. My interests later turned to the study of population and demography,7 and since 1960, no work has been done in Japanese or English on land survey in and of itself. Part I, Chapters 3 and 4 touch on the new administrative system that was established in conjunction with the land surveys, but information about this matter was published only in an essay in French.8

II The series of essays on land surveys was followed by a short blank period, but in 1963–64 I was given the opportunity by Keio¯ University to study abroad. I spent the latter part of this term at Ghent University in Belgium. During that time, I met J. Craeybeckx of the Free University of Brussels, but little did I know then that this meeting would determine the research that I would undertake. One day, he gave me two books to read that had been published by INED (Institut National d’Etudes Démographiques) of France about a field of study that was then very active in Western Europe. These books were on historical demography and were published in collaboration with Louis Henry, also of INED. One was a manual on research methodology, and the other was a study of a parish in Normandy.9 As I read the books, it did not take me long to realize that they contained revolutionary methodologies that would greatly advance the field from being simply a study of population history to become historical demography. Of course, materials on historical demography in Christian societies were primarily parish registers that chronologically recorded the baptism, marriage and burial of each individual in the parish, so differed from the historical materials of Japan. For the Edo period, however, there are superb historical demographic materials (Part III, Chapter 12) in the form of Shu¯ mon aratame-cho¯ (register of religious faith, SAC) and Nimbetsu aratame-cho¯ (population register, NAC). I thought that if a series of these records was available covering a long period, and if xvii

12:27:16:11:09 Page 17 Page 18

INTRODUCTION

these records could be used, they could serve as research materials that would be at least as good as the parish registers. Early on, attention was being given to these historical demographic materials. In particular, Kanetaro¯ Nomura, my mentor, had even considered publishing a work that utilized them, and I had assisted him in this effort, so I was well aware of their potential. However, the fact that historical demography had been established in Europe and was prospering there was completely unknown to researchers in Japan. This was because there was a type of adverse reaction to the idea of incorporating population into historical research, and also because quantitative research involving population or commodity prices and labour wages was virtually being ignored. After returning to Japan in 1964, I immediately began collecting information from historical materials, starting with the historical demographic materials in the collection of the National Archive of Japan in Tokyo. At that time, I had to directly copy data from the historical materials onto notepaper, or to photograph them. So in this way, I began gathering materials, and started my research with the village SACs that I had already begun organizing. I first applied family reconstitution to the SAC of a village, and was able to obtain an extraordinary result on fertility and marriage. However, because there was almost no interest at all in historical demography in Japan at that time, and consequently it was even difficult to present essays on the subject, I decided to write my essays in English and sent them to researchers abroad for their comments. One person who criticised my work was Thomas C. Smith, then at Stanford University, who had already done research on Japan using Japanese language material. Like me, Smith was interested in the SACs, and he had later published Nakahara,10 a great work that used an original method to uncover the existence of birth restrictions. Smith and I continued our association almost until his death. In September 1968, the Fourth International Conference of Economic History was held at Indiana University, Bloomington, in the United States. I was very surprised to receive an invitation from the program committee asking me to present a paper at the session on Population and Economy. This invitation may have been due to some information from Smith. I presented a paper on the analysis of the SAC of Shinano-no-kuni, Suwa-gun, Yokouchi-mura, in Central Japan, the subject of my current research (Part III, Chapter 13 and Part IV, Chapter 16). This was before Japan had experienced high economic growth, so I could not pay my own travel expenses; the conference covered them instead. This was the first time I had presented a paper at an international conference, and I was quite nervous. On returning to my seat, I was sure my presentation had not been a success. Afterwards, though, a gentleman came up to me and asked me to submit it to the journal he edited. I was surprised to learn that the journal was the famous Annales économie société civilisations of France, and the editor was the famous historian Emmanuel Le Roy Ladurie. I was able to take advantage of my beginner’s luck. I later learned that he was the same age as me, and he has since become one of my closest friends. My paper was translated into French and published in the journal.11 While in New York, I was also fortunate to meet Peter Laslett of Cambridge University, and this meeting also greatly affected my subsequent scholastic activities. Laslett was one of the leaders of the recently formed Cambridge Group for

xviii

12:27:16:11:09 Page 18 Page 19

INTRODUCTION

the History of Population and Social Structure. He asked me to present a paper at an international conference on population and family the following year, in 1969. There I presented a paper on changes in household size in Shinano-no-kuni, Suwa-gun, Yokouchi-mura (Part IV, Chapter 16). At that time, there were various conflicts and uprisings at universities in Japan, and Keio¯ University was no exception. My own laboratory had been occupied by extremist students, but I was able to find a way in and get my briefing papers out. Another researcher from Japan who attended this Cambridge conference was Chie Nakane. The 1960s were a decade of transformation for me. By the 1970s, international organizations related to historical demography were being formed, and international workshops were being held somewhere almost every year. This research field requires patience to get past the long preparatory period. We researchers have all gone through similar difficulties, and while we may debate academically, we also form close, lasting personal relationships. I have received immeasurable benefit from these friends. In 1973, I used the SACs of Shinano-no-kuni, Suwa-gun, Yokouchi-mura to create Japan’s first monograph entitled ‘Historical Demography’.12 This book was recognized in academic circles, and historical demography was at last able to take its place in Japan. Young researchers began to appear in this field, and it became somewhat easier to obtain funds for research and publishing. Looking back over my own research activities, I realize that it took a great amount of time and difficulty to collect, organize, and decipher the historical data, but once the data had been prepared and the analytical process was begun, new facts began to unfold one after the other, and I became engulfed in the interesting aspects of this bottom-up history. Of course, historical research also requires top-down research, but there is actually an abundance of historical demographic data from the Edo period of Japan, and for that society, bottom-up research offers a comparative advantage over other types of research. Many of the essays in this book are related to historical demography and were written in English. There were also a few essays in French.13 Looking back over these works, however, I realize that the international contributions I made to this field were insignificant. Furthermore, recent developments in historical demography are being made using sophisticated demographic and statistical methodologies, which far surpass the capabilities of a low-tech historian like myself. I did not learn from an organized system, but rather by observing others. Such laxness may no longer be permitted in this field. In regard to other Asian countries as well, much research is indeed being published,14 regardless of who is actually doing the research.

III What have we been able to see through research of historical demography? The work in Part V, Chapter 17, was done in collaboration with Satomi Kurosu. Research using micro-historical data is currently being organized, and we are attempting to get a complete overview of the data for the nation as a whole. Through this, we see that there are clear diﬀerences across regions in factors such as the average age of ﬁrst marriage, the number of births per woman, the age at

xix

12:27:16:11:09 Page 19 Page 20

INTRODUCTION

which women stopped giving birth, the number of generations in a household, and demographic shifts. Characteristics of north-east Japan were early marriage, a small number of births, early end of childbearing, multiple-generation households, and little migration. Central Japan tended to have late marriages, a large number of births, late end of childbearing, fewer generations in a household, and a large amount of migration. South-west Japan was almost the same as central Japan, but because there were no large cities there, there was less migration than in central Japan. There were also large numbers of divorces and remarriages following initial marriages, and a large number of children born out of wedlock, which, if taken in a positive light, could indicate that there was a relatively free and open attitude towards relationships between men and women. Although it is only a theory, I consider that these results reflect environmental characteristics as well as the characteristics of the people settled in each region. It has been said that north-east Japan’s population included people who had migrated from north-east Asia. These hunters and gatherers tended to not increase their population beyond the carrying capacity of the resources, and the harsh natural environment of north-east Japan created a situation where people needed to marry early and have no more children than necessary, and for women to return to the labour force after bearing children. There were many generations in one household; the average was a three-generation stem family, but some families had four or more generations. Through simulations run using these conditions, we see that the structure was one that maintained the maximum, stable number of people of productive ages within the household. Of course it may be that the people were not actually aware of this, and that this system had simply become established as a tradition or custom. Central Japan was the most urbanized region of the country in the Edo period, and there was a demographic shift from rural villages to cities in this region (Part IV, Chapter 15). Birth rates were high in farming areas, but people migrated to urban areas, where there was a high death rate, so there was no population increase for the region as a whole. There was an obvious influence of negative feedback involving economic development and population stagnation. This feature is evident after the start of the Edo period, and it is difficult to understand conditions prior to that. A simulation for north-east Japan shows similar results as for the centre and south-west, with large fluctuations in the population of productive age within households, and large differences between the maximum and minimum levels. This is connected to shifts in the possession of agricultural land – when a household had few members of productive age, it would pawn or sell the land; and when the number of productive members increased, it would buy it back. The village documents for this region contain vast amounts of information on the purchase, sale and pawning of land, providing subjects for future research. In my opinion, south-west Japan shares certain characteristics with the rim of the East China Sea. This region includes western and southern Kyushu, Iki, Tsushima, the Goto¯ Islands, and the Seinan Islands, as well as the Ryu¯ kyu¯ Kingdom (independent until 1878), Korea’s Jeju Island, and perhaps the coastal area of China. Furthermore, the demographic characteristics described above are common to South-East Asia. It is possible that the demographic and family patterns of south-western Japan were brought by immigrants from South-East Asia.

12:27:16:11:09 Page 20 Page 21

INTRODUCTION

But we must wait for future studies to determine whether this was so. Because these people travel by boat, they have been able to move long distances faster than is possible on land. If we give our imagination full rein, we might consider that around the East China Sea lived a people with a common culture and language; but with the emergence of countries, the area was divided among Japan, Korea, Ryu¯ kyu¯ and China. Today, it is inhabited by people speaking different languages. But with respect to language as well as family and demographic characteristics, it has aspects more common to the East China Sea coastal cultural area than to each country. It is like an Atlantis of East Asia: as if a place once sank into the sea, existing today only as a rim of its former self. There are many ruins on the seafloor of the Okinawa islands, and if this concept is not simply fantasy, we may need to conduct a scientific study of the possibility. What I would like to emphasize here is that Japan is by no means alone or unique. The idea of Japan as a country of one people and one language is a myth. The Japan described by the SACs of the Edo period is one of people with a diverse sense of values. At first glance, the population may appear unified, but this is due to various structural elements in a country without a strong, monotheistic religion, a chain of islands which had not been infiltrated from outside.

IV Lastly, let us consider whether it is possible to extract a total image (histoire totale) of Tokugawa Japan (i.e. seventeenth to mid-nineteenth century) from the observations I have described. Such attempts are made in Parts I and II. However, there are still too many blank areas, so I would like to briefly explain my concepts about Tokugawa Japan that are not discussed in this book. The Tokugawa government became the central authority in Japan in 1603. It had learned many things from the preceding Toyotomi government (late- sixteenth century), including how to ascertain the value of taxed lands throughout the country in terms of rice,15 thanks to a cadastral survey using the kokudaka system (an estimate of the annual yield of farmland, measured in koku [180 litres] of unpolished rice). The Shogun generals would exchange property, expressed in kokudaka, for the loyalty of their daimyo¯ territorial lords of various ranks. The daimyo¯ formed the same type of relationship with their vassals, and all the land in Japan was assigned a value, which served as the basis for assessing land tax. This was significant. Before, the local lords forcibly collected land tax and work servitude from the farmers. In the Tokugawa period, however, a standard was established, and land tax was imposed based on it. Work servitude also remained in some cases, but not as a general rule. Basically, the harvests were surveyed each year by the warriors who came to the villages to determine the land tax rates. The warriors and farmers were separated, with the samurai living together in urban areas, and it was not difficult for the farmers to prevent them from accurately ascertaining crop amounts. The kokudaka system was an effective means of collecting land tax when the volume of production remained consistent for long periods of time. But as productivity gradually increased, the warrior class became unable to tell how much had been produced. Starting around the middle of the Tokugawa era, all territorial lords became aware of the costs and ineffectiveness of

xxi

12:27:16:11:09 Page 21 Page 22

INTRODUCTION

harvest surveys, so they started setting consistent land tax rates. Nevertheless, it was inconceivable to the lords to try to restructure the cadastral survey, because every daimyo¯ had to behave according to his kokudaka rank. Any change in the daimyo¯ would break the status quo. Thus the kokudaka was fixed throughout the period. The farmers were now able to know how much annual land tax would be collected, and if they paid that amount, they could keep the remainder. This subtle class conflict between warrior and farmer ended with the farmer (and merchant) as the winner. In the seventeenth century, a farmer’s average life expectancy was thirty or less, but by the middle of the nineteenth century, it had risen to nearly forty. Farmers were living ten years longer not because of medical advances, but rather due to progress in food, clothing and shelter. Their increased life expectancy was undoubtedly the result of improved standards of living.16 Another significant aspect of the kokudaka system was that it treated the ‘village’ as the unit, and the burden of land tax was borne by the village. The farmers may have had de facto land possession, but their ownership had not been established by law (de jure). The establishment of private land ownership rights resulted from land tax reforms implemented by the Meiji government, which promoted modernization immediately after the Meiji Restoration (1868) at the end of the Tokugawa period. Before that, there had been no system for registering posses- sions, including land. Viewed from abroad, it must have appeared very odd that Japan had advanced financial and commerce systems in the Tokugawa period, but lacked a system of registration; instead it used a system of implied certification by the members of society. But for all that, the Tokugawa period was peaceful. In this way, the rulers of Tokugawa Japan were unable to amass wealth, but there were those among the ruled who were able to do so. In some cases, people even bought the right to become a samurai, so entering the samurai class. In that sense, social mobility did exist in the latter half of the Tokugawa period. The SACs have numerous examples of children from upper-class farming families being adopted into the samurai class. The farmers were by no means bound to the land. Some formed groups with which they travelled, and some young people left for nearby cities, particularly those with a high demand for labour, where they worked and in many cases settled. Some became indentured servants of merchants, advanced to assistant manager and managerial positions, and then went on to establish their own independent shops. However, approximately half of those who went to the cities returned to their villages after about ten years (Part IV, Chapter 16). Once these commoners began to experience both urban and village life, it became easier for the cultures of both to propagate. For example, this undoubtedly helped increased literacy rates in the villages. While Tokugawa society was not perfect, it was very dynamic and in various ways rich in transformation. Tokugawa Japan has left us with many historical materials covering a wide range. It is difficult to study societies for which few or no historical materials exist, but it is also hard to study societies with too many historical materials. When one uses only the materials from a specific village, one runs the risk of not being able to see the forest for the trees. At the very least, there is a tendency to fall into a narrow range of study. To overcome this, the most important things are a solid

xxii

12:27:16:11:09 Page 22 Page 23

INTRODUCTION

historical perspective and research framework. This is something I would very much like to emphasize as a lesson learned by one who has worked wholeheart- edly in the research of Tokugawa Japan for the past half century.

NOTES

1. This research organization was established by Keizo¯ Shibusawa, the famous businessman and politician, who also was a researcher who understood this field. 2. The documents from the collection of the late Kanetaro¯ Nomura are currently kept in the Faculty of Letters, Keio¯ University. 3. The results were published as ‘Toshi kinko¯ son no sho mondai – Musashi-no-kuni, Toshima-gun, Tsunohazu-mura (Issues for villages neighbouring urban areas – Musashi-no-kuni, Toshima-gun, Tsunohazu-mura)’, Mita Gakkai Zasshi, 47(3), 1954. 4. In Japan referred to as Taiko¯ kenchi ronso¯. This discussion relates to the significance of the cadastral surveys of Hideyoshi Toyotomi in the late-sixteenth century. With a background of the release of agricultural lands in post-war Japan, this was a very active and lively debate. The issue eventually faded, with none of the theories remaining, but its wake provided an opportunity to consider the ‘middle ages’ and ‘early modern age’ of Japan as periods, thus playing a role in the history of research. 5. The unit for the assessment of land taxes was not the ‘individual’ or ‘house’. One of the first acts of the government of the Meiji Restoration (1868) was land tax reform, which changed the unit for levying land taxes from the village to the individual. 6. In more than a few cases, the end of the cadastral survey ledgers contained the muradaka (‘village total’) as a standard for levying land taxes, and also included the number of yakuya (‘servitude households’; many different expressions were actually used) as a unit for levying work servitude. When this information was not included in the cadastral survey ledgers, the number of homes for which work servitude could be levied was handled separately. 7. The primary reason for the transition to population study was that the cadastral survey ledgers also included a survey of the number of houses. For a fairly large number of clans at the beginning of the Edo period, surveys were conducted on the number of houses and the number of people. The daimyo¯ considered that it was the units of ‘land’ and ‘individual person’ for which they could use the survey ledgers. 8. The essay is titled ‘Epanouissement du “Nouveau regime seigneurial” aux 16e et 17e siecles’, published in Keio Economic Studies 1, 1963, pp. 21–52. 9. Louis Henry et Michel Fleury, Des registres paroissiaux a l’histoire de la population: Manuel de dépouillement et d’exploitation de l’etat civil ancient, Paris: INED, 1956. Louis Henry et Etienne Gautier, La population de Crulai: paroisse normande. Etudes historique, Paris: INED, 1958. 10. Thomas C. Smith, Nakahara: Family farming and population in a Japanese Village, 1717–1830, Stanford: Stanford University Press, 1977. 11. Akira Hayami, ‘Aspects demographiques d’un village japonais, 1671–1871’, Annales economie societe civilization 3, 1969. 12. Akira Hayami (in Japanese), Kinsei no¯son no rekishi-jinko¯-gakuteki kenkyu¯ , Tokyo: To¯yo¯ Keizai Shimpo¯-sha, 1973. 13. In addition to the article mentioned in footnote 15, there are ‘La demographie historique japonaise. Bibliographie selective’, Annales de démographie historique, 1970; ‘Mouvements de longue durée et structures japonaises de la population à l’époque Tokugawa’, Annales de démographie historique, 1971; ‘La démographie du Japon a l’époque des Tokugawa’, Population et famille 54, 1981–3; ‘Le report Eurasia au Japon: La creation d’une banque de données historiques de la démographie japonaise et son utilization pour la recherche’, Annales de démographie historique 2, 1998; ‘Population, ménage et succession au nord-est du Japon: Aizu 1750–1850’, Ebisu 36, 2006 (collaborateur Aoi Okada). 14. Tommy Bengtsson et al. (eds), Life under pressure. Mortality and living standards in Europe and Asia, 1700–1900, Cambridge (Mass.) and London: MIT Press, 2004. 15. Currency amount in some rare cases. 16. See Akira Hayami and Hiroshi Kito, ‘Demography and living standards’, in Akira Hayami et al. (eds), Emergence of economic society in Japan 1600–1859, Oxford: Oxford University Press, 2004, pp. 213–46.

xxiii

12:27:16:11:09 Page 23 Page 24

12:27:16:11:09 Page 24 Page 1

PRELUDE TWO HISTORICAL LANDMARKS

12:27:16:11:09 Page 1 Page 2

12:27:16:11:09 Page 2 Page 3

First published in The Palace, vol. 3 (Autumn), pp. 14–17, Tokyo, 1992.

PHILIP II AND TOYOTOMI HIDEYOSHI

he year 1992 will be observed (though not necessarily celebrated) Tthroughout the world as the 500th anniversary of both Columbus’s arrival in the West Indies and the expulsion of the Moslems from the Iberian peninsula. However, 1992 also marks the 400th anniversary of two incidents of great significance to Japan’s international relationships but which are usually overshadowed by the European events of a century earlier. In 1592 Japanese forces invaded Korea under the orders of the general Toyotomi Hideyoshi. Two years earlier Hideyoshi had gained control over the entire Japanese nation which had been engaged in civil war for more than a century. Hideyoshi planned to conquer China, and demanded that Korea grant free passage to his forces. When the Korean government refused his request. Hideyoshi ignored it and, in the spring of 1592, ordered his forces to land at the southern end of the Korean peninsula. Forces from China mobilized immediately, and Hideyoshi’s invasion of Korea developed into a full-scale war that continued until his death. The financial burden on the local lords who had sent troops was so great that after his death power quickly slipped from his successors into the hands of Tokugawa Ieyasu, who did not engage in the invasion. Ieyasu later abandoned the campaign. Korea was swiftly overrun. The Manchus, a northern people who invaded Korea after Hideyoshi, overthrew the financially exhausted Ming China and established the Ch’ing dynasty in the 1640s. In Korea, the Yi dynasty was forced to pay tribute to the Manchus, who enforced even stricter isolation on Korea than did Tokugawa Japan. The Ch’ing dynasty remained in power in mainland China for over two and a half centuries. Yet it was so decentralized and so far removed from local sources of power that when the great European and American powers—now economic and military giants by virtue of the industrial revolution—pressed China for concessions, dynasty leaders could not formulate a coherent foreign policy and reacted weakly. Hideyoshi’s invasion of Korea was therefore a principal source of the instability that ensued throughout northeast Asia, and China’s loss of control over the region had a strong and lasting effect on subsequent Asian history.

12:27:16:11:09 Page 3 Page 4

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

The other incident of 1592 was Japan’s encounter with Spain. That summer, at his military headquarters in northern Kyushu, Hideyoshi received a diplomatic mission from Manila, which is on the island of Luzon in the Philippines and at the time, Spain’s base for its activities in the Pacific region. The subject of their meeting was the rebuff received the previous year, when emissaries from Luzon refused Hideyoshi’s outrageous demand for tribute from the Philippines, a demand linked to his ambitions to conquer that territory as well as China. The few letters exchanged by Hideyoshi and the governor of Manila reveal much about the development of relations between Japan and Spain. Each letter from Japan contained a lengthy preamble on the greatness of Hideyoshi, while each letter from Manila contained a lengthy preamble on the greatness of King Philip II. Not surprisingly, Hideyoshi and the mission from Manila did not come to any new understanding as a result of this conference. However, it did result in Spanish access to missionary work in Japan by Franciscans and Dominicans, with Spain being granted much the same status that Portugal had obtained for its Jesuits. The slander that these missionary sects aimed at each other only served to strengthen the Tokugawa shogunate’s distrust of Christianity. Eventu- ally this overzealous competition resulted in the arrest, exile, or execution of many missionaries and converts. Laws enacted in 1639 outlawed all forms of Christian worship, restricted foreign trade, and prohibited Japanese from travelling overseas, thus heralding Japan’s entry into a prolonged period of ‘isolation’. These encounters between Japan and Spain during Hideyoshi’s reign four hundred years ago well represent the distinctive characteristics of the Tokugawa period. Philip II and Toyotomi Hideyoshi never met, but the letters they exchanged certainly demonstrate that they were aware of each other’s presence. How did the Spanish king in El Escorial react to Hideyoshi’s outrageous demands from Osaka Castle? We can never know for sure. But even though Japan and Spain were on opposite sides of the world, with a letter taking more than a year to get from one country to the other, Hideyoshi was undoubtedly awed by the vast Spanish territory as shown in the world maps brought to Japan by the Portuguese. Hideyoshi and Philip II both died in September 1598, according to the Gregorian calender, but thanks to these magnificent maps we can still travel to the past and imagine talking with them.

12:27:16:11:09 Page 4 Page 5

First published in The Palace, vol. 4 (Winter), pp. 14–17, Tokyo, 1992.

FRANCISCO XAVIER AND JAPAN

he first Europeans arrived in Japan in 1543. The year 1993 will there- Tfore mark the 450th anniversary of the first contact between Japan and the Western World. The first encounter between the Japanese and Europeans actually took place entirely by chance. A Portuguese crew engaged in trading along the coast of China was caught in a storm and drifted ashore on an island in southern Japan. The guns the Portuguese brought with them were soon in great demand among Japanese feudal lords, for whom warfare was a daily occupation. Guns were not the only thing the Portuguese brought with them. Portuguese and Spanish expansion throughout the world at this time was characterized by profitable trading and fervent attempts to disseminate Christianity. After sailing along the west coast of Africa and around the Cape of Good Hope to reach India and Southeast Asia, however, the Portuguese apostolate met with little success. Other monasteries had already been established in these regions, and extremely harsh natural conditions made it impossible for missionaries to continue their activities over extended periods of time. Of all the Christian orders scrambling to be the first to spread the word, one had a significant edge: the newly formed Society of Jesus, thanks to its close ties with the Portuguese royal family. These ties enabled the Society to send its missionaries into areas allocated specifically to them under the treaties of Tordesilhas and Zaragosa. Francisco Xavier, a high-ranking priest in the Society of Jesus, was not Portuguese, but he did manage to lay the foundations of Christianity in the two years he was in Japan. Christianity was to remain a major religion within Japan until it was finally completely banned in 1639 by the Tokugawa Shogunate. In the meantime, Japan developed its reputation as an attractive trading partner and producer of high-quality silver. The price of silver in relation to gold was lower in Japan than in China, and traders could profit by buying silver in Japan and exchanging it for gold in China. Since Japan and China had no diplomatic ties at the time, trade restrictions were virtually nonexistent. The Europeans, who had harvested huge quantities of gold and silver from the New World, were by now quite sensitive to slight differences in

12:27:16:11:09 Page 5 Page 6

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

the relative values of gold and silver. They were therefore amazed at the huge price gaps that existed between the Japanese and Chinese markets and lost no time in exporting massive quantities of silver from Japan to China. Christian missionary work at this time intensified to the extent that even a few feudal lords in western Japan were converted. A contemporary report by missionaries compiling the number of baptized Japanese indicates that Christians accounted for 10% of the Kyushu population, although this figure is probably something of an exaggeration. The first person to link the burgeoning silver trade to the success of Christianity in Japan was Luis de Camões, an acclaimed Portuguese poet of the 16th century. In the Os Lusiadas (Canto X, 131), which eulogized Portuguese expansion abroad, he gave this description: Much of the World being now conceal’d from You A time will come when it shall all be show’d. But by all means the Islands thou must view, Where Nature sees most cost to have bestow’d This, shadow’d half, which China answers to, Japan is, yielding the best Silver-mine: Which th’Evangellick Furnace shall refine. However, history did not unfold according to Camões’s vision. Toyotomi Hideyoshi and Tokugawa Ieyasu, who brought Japan out of civil war and unified the nation at the close of the 16th century, realized that Christian- ity did not mesh with the principles of their newly united nation. What’s more, they both held a personal distaste for all things Christian. Missionary work was eventually prohibited by law, and it soon became illegal to practice the faith. Finally, when the Christian rebellion was suppressed in 1638, the Tokugawa Shogunate imposed a strict ban on the religion. Japanese were further prohibited from traveling abroad, and those living abroad were not allowed to return home. This isolationist policy was continued under the Meiji government, and not until 1873 was the ban on Christianity finally lifted. At the turn of the 17th century, Japanese mineral resources dried up and domestic demand for silver increased. As a result, the export of silver slowed significantly, with Japan actually becoming a net importer of silver by the second half of the 18th century. Once proclaimed by Marco Polo to be the island of gold and silver, Japan had now become an unattractive prospect for European colonization due to its scarcity of resources. ‘The Examination of the Sects’ (Shumon-Aratame)—a system used to scrutinize religious practice in the wake of Tokugawa’s strict ban on Christianity—has proved an extremely valuable reference point for demographic historians. Under this system, which began in 1638, all residents in Japan were forced to pledge that they were Buddhists, i.e. not Christians, in order to obtain official certification from the temples. In fact, this turned out to be more of an inquisition than an examination. Although the 1638 examination was only carried out in areas under the direct control of the Tokugawa Shogunate, a further decree in 1671 extended the Examination throughout the rest of Japan.

12:27:16:11:09 Page 6 Page 7

FRANCISCO XAVIER AND JAPAN

Records of the Examination were produced every year until the ban on Christianity was ﬁnally lifted. These records, however, were not standardized throughout Japan. Nevertheless the names of individual family members, household addresses, and the names of the temples to which a family belonged were always recorded. In many cases, the ages of family members were recorded and explanations given when families moved from one place to another. Financial records were sometimes maintained on the estimated value of each family’s land and livestock. In practice, these records had much the same function as today’s census. One special feature of the Examination was that all Japanese citizens were subject to oﬃcial scrutiny. Records thus make it possible to trace the entire lives of ordinary people who lived during the Tokugawa era and to create demographic indexes on births, deaths, marriages, and relocations. Recently developed techniques in historical demography have allowed us to conduct detailed sociological analysis in these areas also. No other country of this era produced such detailed records on individuals, married couples, and family behavior. The shogunate and feudal lords who ordered that these statistics be compiled were unaware of their true value, and limited their use of this information to such purposes as determining the size of the population in their own domains. Most of the records were thus destroyed. The data we can use today are mostly copies or drafts of records which were preserved in towns and villages more by chance than anything else. By carefully researching this data, we have been able to form a far richer and more realistic picture of Japan under the Tokugawa Shogunate than had previously been imaginable. This historical breakthrough would never have been possible if the Examination of the Sects had not been carried out. In this sense, Japan is a country rich in historical resources, especially for demographic historians. Perhaps we have Francisco Xavier to thank for his contribution to the turn of historical events that led to the compilation of these records.

12:27:16:11:09 Page 7 Page 8

12:27:16:11:09 Page 8 Page 9

PART I TOKUGAWA JAPAN

12:27:16:11:09 Page 9 Page 10

12:27:16:11:09 Page 10 Page 11

First published in History of Humanity (P. Burke and H. Inalcik (eds)), vol. 5, pp. 344–357, UNESCO, 1999.

1 Japan: Sixteenth, Seventeenth and Eighteenth Centuries

hile 1492 was not in itself a year of particular significance in Japan’s history, Wit roughly marked the start of the greatest transformation Japan had ever experienced. A transformation with many facets – political, economic, social and cultural – that was to continue through the sixteenth and seventeenth centuries. Indeed, it was this transformation that set the guidelines for creating the special characteristics of modern Japan. In short, prior to that transformation, the main elements constituting Japan were greatly influenced by systems and culture either in China or originally brought over from China. Japan was in many respects a member of the ‘Chinese World Order’, and a very faithful member at that. Again, prior to that transformation, there was, first, no clear distinction in Japan between the religious and the secular; second, political power and economic wealth were often to be found in the hands of the same individual. Thus, in the religious sphere, ‘art’ would be for the benefit of Buddhism, while in the political sphere, shoguns wielding political power would invest in their own foreign trade operations. However, after the transformation, not only was Japan divorced from the ‘Chinese World Order’, but the values constituting its society were separate and independent. Subsequently these started interacting with each other to produce a society having pluralistic values. Inhabitants of the Japanese Archipelago became worldly as opposed to religious, and formed social groups based on a mentality emphasizing economic values. In this sense, one is justified in calling it ‘a great transformation’ when looking at Japan from a historical perspective (Hayami, 1986). Of course, a transformation of this nature did not come about overnight. One might say 100 to 200 years were required. Moreover, it did not progress simultaneously throughout the entire Japanese land mass. It varied in nature from region to region, and the sequence of the various facets was not the same everywhere. Thus it at times involved an incoherent state that could be called a temporary state of confusion, chaos or crisis. In this context, the new experience of diverse (outward and inward) international relationships added an extra dimension, making the situation more and more complex. Below, we discuss how this transformation came about and the Tokugawa Japan that was the final result.

12:27:16:11:09 Page 11 Page 12

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

THE SIXTEENTH CENTURY – JAPAN WHEN THE TRANSFORMATION BEGAN First let us consider the political administration and land administration aspects of the situation prevailing in the sixteenth century when the transformation began. In pre-sixteenth century Japan, neither the political system nor the land administration system were simple. There was instability, and in some cases coexistence of systems having mutually incompatible principles. However, in the circumstances none of the political power players had the power to create a new system – a situation that continued up until the country became embroiled in the great confusion of the Sengoku period (Warring States period) that began in the latter half of the fifteenth century (1467–1573). Though there were still vestiges of the Ritsuryo authority and shoen authority centred on Kyoto in central Japan in the Sengoku period, there were also upstart petty local lords at the local level, and political control became weak. One aspect of this was the explosion of economic activity, which was not limited to regional commerce, since foreign trade also developed through traders in port cities such as Sakai and Hyogo. Also, with the inflow of money from China, farmers did not have to work just to feed themselves and make their annual tax contribution, but also began to work to produce for the market. Thus economic development took place in a ‘bottom up’ direction. In the towns, a self-governing system based on the upper urban classes was established. The existent enfeebled political authority was not allowed to inter- vene, with people guaranteeing their own economic activity themselves. Even in the villages, it was found that in the face of the political vacuum, communes called so were established and autonomous social groups based on a hierarchy with the temples and shrines as a nucleus grew up (Tonomura, 1992). These autonomous organizations that grew up in central Japan from the latter half of the fifteenth century and through the sixteenth century were historically speaking extremely exceptional for Japan in that they were the result of the development of economic activity in the absence of strong political authority. It was just when such autonomous organizations started spreading from central Japan to the remoter regions that a new political authority from elsewhere established itself in the central zone and nipped the populist organization of society in the bud. In contrast, in the outlying regions, the Sengoku daimyo (Civil War period local lords), exercising power locally and independently vis-à-vis both the Ashikaga Shogunate and the Ritsuryo establishment, continued to grow. Using their military might, they fought with the neighbouring lords, extended their territory, and used brute force to eradicate the various traditional systems. Besides laying down the law of the land on their own, they then initiated ‘top down’ economic development by establishing irrigation and water-use facilities, building roads and bridges, and developing mines. However, social construction on that basis left the social structure in its traditional form. Thus two types of farming enterprise coexisted. The first were large-scale businesses based on dogo (powerful provincial or village landowners) with great recourse to forced labour. The second were small farms normally based on a husband and wife unit. In either case they were

12:27:16:11:09 Page 12 Page 13

JAPAN: SIXTEENTH, SEVENTEENTH AND EIGHTEENTH CENTURIES

producing exclusively to feed themselves and pay their annual tax. The dogo in normal times lived in the village to supervise agricultural business. For short periods, they sometimes had to rush off with at the head of all their clan to partake in a campaign on behalf of their lord and master. The Sengoku daimyo tried to extend their power, keeping things as they were and without attempting to change this existing form of farming family. Once these Sengoku daimyo in outlying areas achieved a balance of power, they set their sights on achieving national hegemony, and started to seek to advance into central Japan. This was because it was the centre of political power and because economic wealth was concentrated there. However, just then, they found a force had developed right in their path. This was a force that had grown up in the zone between the centre and the outlying regions. It was a new form of ‘territorial lord system’ that had been able cleverly to absorb the economy that had developed in a ‘bottom up’ direction, free from the framework of traditional political rule. The special feature of this system was the separation of soldiering and agriculture. This consisted of having the warriors segregated from the farmers and obliged to live grouped together in one place. Thus a professional army was created with the peasants concentrating on agriculture. The dual benefits derived from this separation of the roles were great military power and high agricultural productivity in the fief. Very conveniently for those with the newly found power, a new weapon had come to hand. It was the gun – brought in by Portuguese landing on an island in the southern part of Japan in 1543 and in next to no time produced in Japan. However, the first person to use it effectively in battle was Oda Nobunaga (1534–82) from the intermediate zone lying between central Japan and the outlying regions, who had established a standing army on the warrior/peasant separation principle. He had managed to build this up into a small but potent force by carrying out massed battle training. The method he devised for using the guns was to have three rows of men lined up in ‘steps’ firing in rotation, so that the matchlock gun was used as efficiently as possible, taking into account the time required for ‘firing’, ‘cleaning’ and ‘loading’ in the case of such a weapon. Use of this firing method was some seventy years in advance of Europe (Parker, 1988). A typical example of success using this method can be seen in the Nagashino battle of 1575, where a small force was able to overcome the cavalry of Sengoku daimyo Takeda, which at the time had a fearsome reputation throughout Japan. Oda Nobunaga overcame the surrounding Sengoku daimyo one after another. In 1570 he finally entered Kyoto and drove out the Shogun. Although he had reached the final phase marking the end of the long Sengoku period (Warring States period), he himself was assassinated by a vassal in Kyoto in 1582. Oda Nobunaga was succeeded by Toyotomi Hideyoshi (1536–98) who destroyed the said assassin’s force within a few days. By appropriating Oda Nobunaga’s newly created military power and his new system for taxation and administration described below, Toyotomi Hideyoshi was able to pacify the whole of Japan over a period of just eight years, building a large castle at Osaka and bringing the economy of highly developed central Japan under his control (see Map 1.1). The new system for taxation and administration established by Hideyoshi was subsequently taken over by the Tokugawa Shogunate (1603–1868) and became

12:27:16:11:09 Page 13 Page 14

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

Map 1.1 Map of Japan (after A. Hayami).

the foundation for Japan’s administrative system prior to the Meiji Restoration in 1868. This form of administration was something now called the kokudaka system (assessment tax base system), whereby farmland and residential land were measured ‘ﬁeld by ﬁeld’ and ‘parcel by parcel’ and an evaluation made in accordance with the grade of the land. This evaluation was the base for levying the annual tax. The term ‘kokudaka system’ also derives from this. However, in many cases the evaluation does not indicate a quantity in terms of money, but in terms of a volume of rice (koku is a unit for measuring rice by volume and daka means amount). Furthermore, in the case of villages, this was calculated not on the basis of the single individual’s estimated land holding but with respect to the total kokudaka for the whole village. Thus, the person responsible for paying the annual tax was not the individual but the village as a whole. Thus one might go as far as to say that the problem of who actually paid what did not involve the local lord. However, this kokudaka system (replacing the levying of an annual tax that

12:27:16:11:09 Page 14 Page 15

JAPAN: SIXTEENTH, SEVENTEENTH AND EIGHTEENTH CENTURIES

had existed from the seventh century) was a system peculiar to Japan. At least, at its inception one can say that it functioned as a device for establishing a stable political authority. A key feature of administration via this kokudaka system was that it enabled the Shogunate to switch the postings of the daimyos. To be more specific, under the kokudaka system of administration, each daimyo was classified in terms of the kokudaka which should be his due in accordance with his family ranking. He could then be posted to any fief in Japan assessed at that level. Although transfers from one posting to another did not often occur in practice, the daimyo was always conscious that his being ‘local lord’ of a given area of land was arbitrary. As a result the daimyo’s relationship with his fief and its people was rendered tenuous. This means that the local lord or fiefdom system in Tokugawa Japan cannot be explained in terms of the concept of feudalism as established in Europe. Of course, as explained below, this kokudaka system of administration also had defects. However, these very defects had positive aspects over the long term by enriching a certain fraction of those being governed – notably some of the farmers and merchants. This situation deriving from the kokudaka system was to become an important condition for making Tokugawa society one of the types where politics and the economy were separate.

SO CALLED ‘ISOLATION’ – ESTABLISHMENT OF THE JAPAN-CENTRED WORLD ORDER Another singular aspect of sixteenth and seventeenth-century Japan is the tremendous fluctuation in its international relations. The movements in international relations during this period, which was to conclude with the generally misunderstood ‘isolation’ or closed-door policy, were actually closely linked to the history of Europe and East Asia. Furthermore, one should not take the word ‘isolation’ too literally, as recent studies are making clear (Toby, 1984). Japanese merchants were able to engage in trade freely. However, up until about the middle of the sixteenth century, one cannot ignore the traditional China trade, such as that carried out by the Ashikaga Shogunate. The shogun Ashikaga Yoshimitsu sought a monopoly over the Ming China trade. He was granted Japanese King status by the Ming Emperor, becoming a member of the Ming ‘Chinese World Order’. However, this trade, consisting of importing copper coins from China, was very profitable at a time when coins were not being minted in Japan. Naturally this lucrative trade finally came to be coveted by the daimyo holding the real power and their attempts at usurping the right to this trading monopoly made it one of the political triggers for the strife during the Sengoku period. However, it simultaneously played an economic role, stimulating development by providing a circulation element for the money economy that was gradually developing in central Japan. This licensed trade authorized by the Ashikaga Shogunate heralded the interruption of relations at State level with Ming China. Thereafter, relations between Japan and China were to continue for three centuries without any inter-governmental agreement. With this situation prevailing, the people finally to come on the scene were the Europeans. The Portuguese came to Japan by chance in the 1540s. However, they

12:27:16:11:09 Page 15 Page 16

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

were already aware that Japan offered better prospects than any other place they had been to as regards achieving their twin goals in advancing overseas – that is trading and spreading Christianity. For Japan up to that time had no monotheist faith representing a tough opponent for Christianity; its people were seeking spiritual stability because of the continuous fighting, while the guns that had been brought in were avidly sought and fought over by Sengoku daimyo. Furthermore, as there was no formal trade between Japan and China and there was a differential in the value of gold and silver between the two countries, so that an enormous profit could be made merely by exporting silver from Japan (where it was cheap) and exchanging it for gold in China. Macao was the transit point for the two-way trade between Japan and China, the main items being silver from Japan and raw silk and woven silk from China. In the latter half of the sixteenth century, the opening of silver mines in Japan proceeded at a great pace and a considerable amount of silver was produced. Its fame was such that it is even alluded to in Os Lusiadas, the epic poem by the world-famous Portuguese poet, Camões (Camões, 1963). Since the daimyos in western Japan found they were able to profit economically when they engaged in trade in their fief, they vied for visits by the Portuguese boats, and therefore allowed the proselytizing of Christianity within their fief. Some even became ‘Christian daimyos’, being baptized themselves. The success of Christianity in western Japan was such that there were even reports that Christians amounted to as many as 20 per cent of the total population. In the final years of the sixteenth century, the missionary movement reached central Japan. A period had arrived in Japan that could be called the Christian Century (Boxer, 1951). However, at the end of the sixteenth century, Japan’s national unification was progressing. In particular, with the coming to power of Toyotomi Hideyoshi, Japan’s international relations were suddenly transfigured. Firstly, in the process of unifying the country, Hideyoshi learnt that some land in Nagasaki had been donated to the Jesuits by a Christian daimyo. As well as confiscating it, he issued an order expelling all missionaries from the country on the grounds that Christianity was not only not needed in Japan, but harmful as well. Admittedly, this order was not carried out strictly. Besides which there was the matter of Hideyoshi’s curiosity which made him subsequently often meet the envoy from Portugal seeking to get evangelization restarted. At around that time, evangelization in Japan was faced with a new situation in that the Franciscans and Dominicans – with the backing of the Spanish crown – had began proselytizing. Using Manila in the Philippines as a base, Spain began seriously to advance into Asia with the support of Philip II who was also King of Portugal. Hearing that the propagation of Christianity in Japan was going relatively well, they sent a succession of missionaries. For them, the Jesuits who had taken precedence were the real enemy, and they took every opportunity to say bad things about each other. When these rumours reached the receptive ears of Hideyoshi, who was already adopting an anti-Christian stance, they were enough to bring about an incident involving the arrest and execution of the missionaries. This was where the period of difficulties for Christianity began (Elison, 1973). Of all of Toyotomi Hideyoshi’s international ventures, the invasion of Korea

12:27:16:11:09 Page 16 Page 17

JAPAN: SIXTEENTH, SEVENTEENTH AND EIGHTEENTH CENTURIES

was the one that had the greatest effect within the country and externally. As soon as Hideyoshi had gained control of the whole of Japan in 1590, he immediately began to look abroad to expand the territory under his jurisdiction. Looking at it from today, this reckless action seems to have no rational explanation. Hideyoshi’s dream was to build a dominant sphere in East Asia from Korea and China to the Philippines. He was due to start by attacking China, and asked the Korean government to allow his troops to pass through that country. Naturally they were not going to accede to this request, and in April of 1592 his force landed in Korea. At the time, Yi Dynasty Korea was a civilian state without an army worthy of the name. So in the summer of that year, Hideyoshi’s forces advanced from Seoul to Pyongyang. Yi asked Korea’s suzerain state, Ming China, to send reinforcements, and as Ming China responded favourably, it became a war between the Japanese and Ming military with Korea as the stage. In addition, as the Koreans were inflicting defeats on the Japanese at sea, the Japanese army’s advance was halted. In these circumstances, the war became a stalemate. An attempt at peace also ended in failure, and in the end the war was to continue up until the death of Hideyoshi in 1598. While Hideyoshi’s Korean invasion brought about the financial impoverishment of the daimyo who had to sustain troops, conversely it increased the relative strength of the Tokugawa family who did not send troops. Furthermore, the loyalty of the daimyo themselves to the Toyotomi house following the death of Toyotomi Hideyoshi became even more questionable because the heir was young and no proper system for ensuring the succession had been set up. What brought things to a head was perhaps that Japan had been excommuni- cated from the Chinese World Order centred on China. The fact that Japan had fought with the army of the Chinese World Order’s suzerain state meant that Japan was completely ignoring its position as a state owing allegiance. If China had at that time had the necessary power, it would without doubt have taken sanctions against Japan. For better or worse, the Ming did not have the strength to do that, so conversely it was being destroyed by one more ‘barbarian’. With this state of confusion regarding political authority prevailing on the Chinese mainland, Japan was able to escape from the Chinese World Order without retribution from China. The Tokugawa administration that took over power from the Hideyoshi administration initially appeared to want to restart country-to-country exchanges with China. Having formal relations with China on a country-to-country basis signified – as regards northern East Asia – entering the Chinese World Order. To this, the Chinese maintained a state of indifference. One reason for this was that they could not simply forgive Japan’s barbaric act. Another reason was that the Ming dynasty itself was too concerned with its own crisis to concern itself with Japan. The Tokugawa Shogunate – aware of the political confusion on the Chinese mainland – decisively moved to leave the Chinese World Order once the foundations of their own administration became firmly established with the coming of the 1630s, and at the same time established a World Order centred on Japan (Toby, 1984). What complicated the international relations Japan was to have were problems with Europe, and particularly with regard to Christianity. Ever since the time of

12:27:16:11:09 Page 17 Page 18

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

Hideyoshi, those holding the reins of power had become progressively more hostile towards Christianity. At the beginning of the Tokugawa administration there appears to have been an aim to have a positive approach towards maintaining relationships with the European countries due to the proﬁt to be derived from trade. However, they knew that in the case of the two Iberian countries, trade and proselytizing were inseparable in their overseas advances. Furthermore, the revolt by Christians in 1630 hardened the Tokugawa Shogunate’s attitude, and various measures involving a certain amount of ‘overkill’ were taken relating to international relations. Not only was Christianity forbidden, and the two Iberian countries proscribed, but foreign travel by Japanese and the return of Japanese already abroad were forbidden. As a result Japan was a closed country; and one that had isolated itself, and not one that had been isolated. However, if one observes Japan’s international relations at that time with great attention, one can see that Japan’s relations with Europe and choice of a position in northern East Asia, and more particularly with China, proceeded simultaneously. One need not say that the crucial thing for Japan at that moment was the latter (northern East Asia and China). This was because no one then could imagine that Europe would experience a political and economic change that would confer great power on it. Though this policy forbade Japanese from travelling overseas, trade with Korea and the Ryukyus Liugius was delegated to the Tsushima clan and to the Satsuma clan respectively. Near Pusan in Korea, Tsushima had the Japanese residential area called the ‘Waegwan’, and it was to there they sailed to carry out trade and negotiations between the two countries (Tashiro, 1981). Similarly, the Satsuma had a ‘Ryukyuan Lodge’ in the Ryukyus, where political negotiations were carried out abroad in addition to the primary purpose which was trading (Sakai, 1968). It is true that in Tokugawa Japan there was no state with a modern centralized concentration of authority, but it was only one step away. Thus it was an age for which the term ‘Early Modern’ is appropriate (Hall and Jansen, 1968).

PAX TOKUGAWA When the Tokugawa took over the government of Japan, what they sought most of all was the stable continuation of their rule. By keeping in mind the good and bad points of the short rule of their predecessors in office, Nobunaga and Hideyoshi, the Tokugawas were able to maintain control over a long period. In order to achieve continuity of rule, the controlling authorities systematized their record keeping. Even though doing so was unbelievably complicated, they tried to maintain legal and political precedence. This maintenance of precedence provided stability of rule in as much as it was able to function in the context of social and economic change. The Tokugawa Shogunate first of all rendered absolutely powerless people and institutions who might be hostile to them. This included the Imperial Household and nobility, and also the temples and shrines. For example, saying that the emperor’s work was scholarship, they only recognized his right to decide the appellation of the year and to give titles to warrior Houses. The nobles were only allowed to carry out official functions effectively devoid of any political content – these being carried out as laid down in the Official Calendar. Thus, although the

12:27:16:11:09 Page 18 Page 19

JAPAN: SIXTEENTH, SEVENTEENTH AND EIGHTEENTH CENTURIES

emperor and nobles were important from a ceremonial point of view, they did not actually have any political power. Had either Tokugawa Ieyasu or his two pre- cursors really used their power they probably could have eliminated the Imperial Court and nobles at Kyoto by cutting them off at the roots. However, at that juncture they made it clear to the ‘ruling’ classes and to the public at large that they were the rulers of Japan by ‘pinning down’ the Kyoto-based emperor and nobles who had anyway over a period of a thousand years only been rulers in name. The attitude towards temples and shrines was also astute. The Tokugawa Shogunate were well aware of how difficult it was to get control of the situation when dealing politically with groups of people linked by religious faith. Christian- ity was thoroughly suppressed, while Buddhism was brought within the government system. With Buddhism made ‘official’ any divergence of view within a religious group was exploited by the Shogunate doing all it could to perpetuate any denominational splits. By intervening forcefully even in the case of the most anodyne incident, they let it be known that the Shogunate could control the religious world too just as it liked. There was also the ‘religious inquisition system’, which was originally put into effect in order to eradicate Christianity, whereby all Japanese were decreed to be Buddhist and affiliated with a specific temple of a specific sect. Under this system, the temples came to adopt the attitude that there was no point in tiring themselves out seeking converts since they were guaranteed a set number of believers even without any effort on their part. Buddhist activity in Japanese society lost its momentum, and so much so, that to put it extremely one could argue that it had become ‘Funeral Buddhism’. The Tokugawa Shogunate was a military organization based on a Shogun at the top of a warrior family. Its organization required very strict rules. More than anything, and whatever the level, treason was strictly guarded against and measures taken to prevent it. The daimyo had to send their wives and children to the Shogun’s residence at Edo (Tokyo) as hostages, while they themselves had to go back and forward at vast expense to live alternate years at their own base and at Edo. This system was generalized, and was even given the name sankin kotai system. In fact, it meant that the daimyo were almost constantly at Edo. Marriage by any of the daimyo’s children required permission, in order to preclude the formation of anti-Shogunate alliances that might form through marriage. The Shogunate had one-quarter of total land in the country, and put the major towns and mines under its direct control. Daimyo in Tokugawa Ieyasu’s direct lineage, and accordingly daimyo in whom the most trust could be placed, were allotted fiefs at important places. Also, daimyo who had been vassals before the Tokugawa family took over the country were differentiated from ‘daimyo who had sworn allegiance later’. In addition to cleverly posting the daimyo of these two groups throughout the country, the Shogunate made the former responsible for the Shogunate’s policy decisions (including the tasks and duties these implied), and gave the latter the burden of large-scale construction works. In this way, the Tokugawa administration commanded 200 or more daimyo all over the country and posted them to fiefs very much in accordance with their expectations. However, in the event of any trouble the daimyo would be

12:27:16:11:09 Page 19 Page 20

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

transferred and the size of the fief even reduced. In extreme cases, the family could be eliminated. If we concentrate on this aspect, it would seem that the Tokugawa Shogun was ‘an absolutist ruler’. However, from a legislative point of view, the edicts promulgated by the Shogunate may have been put into effect in the directly administered areas, but it was the daimyo’s laws that were applied in the daimyo’s territory. Thus it was not a matter of laws being applied generally throughout the country. Though the Shogunate had judicial power to cope with disputes arising between persons living in the fiefs of different daimyo, problems restricted to the daimyo’s territory fell under his jurisdiction. Looked at in this way, Tokugawa Japan could be said to have been a society with decentralization of authority (see Plate 1.1). Although the Tokugawa Shogun had ‘Regulations for Samurai’ supported by a Confucian ideology, there was a clear distinction between public and private. Since transgression of this rule met with severe punishment, it is said that this had a profound effect on the ethics of later Japanese society (Nakai, 1988). For example, Samurai who sought personal economic gain were severely censured. For that reason, one can reasonably say that not one Samurai of the Tokugawa period became rich. Rather the opposite happened: the level of consumption of the society went up generally, while the income of the Samurai was fixed and in fact decreased in real terms – it not being rare for them to fall into poverty. The fundamental characteristic of the period is apparent from the fact that the warrior classes with political power were actually faced with economic poverty. The fundamental social system adopted by the Tokugawa Shogunate was kokudaka system mentioned above. The kokudaka system had the following two special characteristics. The first special characteristic was tied to an annual taxation system with the hamlet as unit. In the Tokugawa period, the official responsibility and burden of paying the annual tax was the hamlet’s, and not the individual’s. This does not directly signify there was no private ownership of land, but to put it extremely, it did mean that the lord of the territory could not have cared less what happened about ownership of land by each individual farmer. What the local lord needed was the annual tax based on the kokudaka, and so long as they did not prevent this, he did not interfere in matters such as the private ownership of land by individual farmers. For that reason, it was necessary to wait for the land-reforms such as those introduced soon after the establishment of the Meiji government (1868–1912) for the private ownership of land to be put on a legal basis. It is somewhat surprising for a society with a developed economy such as that of Japan in the Tokugawa period not to have public guarantees regarding the private ownership of land. To arrive at that situation, the society had to be peaceful and without pillagers. Also the people constituting the village had to have relationships based on mutual trust. As a corollary, it was necessary for ‘human goodness’ to become a commonly accepted concept. Rights did not depend on a system devised by the state, but on the goodwill of the neighbours. Indeed, this way of thinking is one that permeates the mentalité of the present day Japanese. The second special characteristic was on the economic front, where the productivity of the land rose, but the local lord was not able to increase the kokudaka in keeping with this. If the rise in productivity during the Tokugawa period can be

12:27:16:11:09 Page 20 Page 21

JAPAN: SIXTEENTH, SEVENTEENTH AND EIGHTEENTH CENTURIES

Plate 1.1 The Himeji Castle, built by Daimyo Ikeda in 1601–9 (from Tokuji Kato, ed., Himejijo (Himeji Castle), Nihon Meijo Shusei [Collection of Japanese Fine Castles], Tokyo, 1984).

12:27:16:11:09 Page 21 Page 22

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

considered to be due to the increased yield per unit area of land, one can say that in a society where land productivity is rising, the kokudaka system was not a suitable system from the point of view of those levying the annual tax. However, from the point of view of those burdened with the annual tax the reverse is true. This is because the kokudaka is a system where a local lord cannot grab the increase in productivity for himself. In consequence, if the yield per unit area increases due to intensive use of the land, the resultant increase may remain in the hands of the farmer, or at the very least not all end up in the hands of the local lord as annual tax. On the other hand, the peasants constituting the class being ruled who had had their weapons confiscated, in accordance with the farmer/warrior separation policy applied since Hideyoshi’s time, had lost the ability to fight head-on with the samurai armies made up of professional military men. With the samurai concentrated in the cities, and hardly any samurai in the village areas, the peasants were nevertheless able to maintain their own safety and stability in their districts by themselves. All guns, including hunting guns, owned by the peasants had to be declared without exception. Though it depended on the local lord, even daggers and the like could be subject to registration. Thus methods of resistance for the peasants were limited to gathering virtually weaponless in large numbers and shouting their demands; or getting the amount they had to contribute as annual tax reduced on some pretext. Although the former are often cited by researchers in the context of riots, one can find in the latter the basic configuration of the period’s social movements as represented in the daily conduct of the peasants. Textbooks say that the peasants of that time were subjected to many constraints including being placed under mutual surveillance based on the goningumi system (5-family-unit mutual surveillance); being forced to lead a frugal life in accordance with the laws; not being allowed to buy or sell land; limitations on the type of crops they could grow; and even no freedom of movement. Although one must admit that an abundance of edicts forbidding this and that were promulgated, there was no executive organization to ensure that the peasants kept to them. One might say that these restrictive laws represented what the leaders ideally would have liked, with the reality being quite different. The actual life of the peasants was, it seems, as if there were a complete absence of such restrictive laws. They carried on rationally according to the economic rationality: buying and selling land, producing items from which they expected to make a profit, and even moving. In the event of a lawsuit, the law would be resuscitated, but otherwise it was left dormant. The merchants in the towns looked at daily life from the standpoint of the ideology of the samurai who saw themselves as the ‘rulers’ as opposed to the ‘ruled’. That is, though they were allotted a social position below that of the farmers according to Confucian philosophy as they made a profit without making anything, they were actually showered with economic privileges, and came to amass social power. Many of present day Japan’s zaibatsu (industrial and financial conglomerates or combines) that had financial business as a nucleus originated from activities during this period. There was also great disparity between merchants. These ranged from two groups of merchants at the top – the big merchants closely involved in the financial affairs of the Shogunate and

12:27:16:11:09 Page 22 Page 23

JAPAN: SIXTEENTH, SEVENTEENTH AND EIGHTEENTH CENTURIES

daimyo together with the merchants with the power to create credit by distribut- ing promissory notes as if they were money – down to small-time traders doing deals in the street. In the case of the great merchant families, a tradition or code was established, and a ladder for the promotion of the numerous employees institutionalized. Also, since, as a rule, they were allowed to engage freely in their activities, they were able to make maximum profits at that time despite their lowly status in society (according to Confucian values) and even become supporters of culture. A phenomenon specially characterizing Tokugawa Japan was the increase in population. Taking the country as a whole, and extrapolating from the figures available for one area, the population at the beginning of the seventeenth century was 12 million (±2 million). According to the first general census carried out in 1721 by the Tokugawa Shogunate, the population had reached 30 million. Though the population stagnated somewhat over almost the next century, it reverted to the upward trend in the nineteenth century, linking up with the population increase of modern times (Hayami, 1971). Why did the population increase? Evidently, the increase in the seventeenth century was linked with the increase in the area of land under cultivation. During this century, referred to as the ‘Great Land Reclamation Period’, the bringing under cultivation of flat areas throughout Japan continued apace. In accordance with the new local lord system, the local lords came to have the right to exclusive local lordship rights for a given domain, and set up flood control and irrigation facilities, and succeeded in turning deltas – which up to then had only seen unstable production – into fertile agricultural land. The peasants too were free from social instability, and a family unit centred on husband and wife became an agricultural production unit. Seen from another angle, it represented the formation of a peasant society. With such a family format, the marriage rate also increased, and the number of single persons decreased. A kind of baby boom took place, and an explosive increase in population occurred throughout the country (Hayami and Miyamoto, 1988). This increase in population was absorbed by the rapidly established towns. For instance, in the seventeenth century, Japan’s population increased two and a half times, while that of the towns and cities increased tens of times. Processes such as the above were limited geographically, but documentation on population does remain to confirm it. Thanks to this documentation one can observe how the old-style households encompassing collateral relatives and a great number of unmarried people gave way to small families made up of direct relatives. Again according to the documentation, this change spread in concentric circles centred on the cities, its speed being roughly 200 m per year (Hayami, 1973b). As towns grew up throughout the country, even that speed was enough for the whole of the country to have become a peasant society by the middle of the eighteenth century with the exception of remote places in the mountains. The peasants were producers not only of food but also of a large portion of the raw materials for industry. In addition to their traditional production destined to meet their own needs and permit payment of the annual tax, the peasants produced for the market – doing so efficiently and seeking the best way to make a profit. The peasant economy had arrived (Smith, 1959).

12:27:16:11:09 Page 23 Page 24

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

The central market for the whole country was indisputably Osaka’s Dojima rice market (see Plate 1.2). There the annual taxation rice (warehouse rice) and rice purchased directly from the farmers (barn rice) were traded amongst the merchants. This trading was not limited to spot goods (spot trading) but also involved futures trading equivalent to present day ‘hedging’. It was as if it were playing the role of the present day commodity and stock exchanges (Miyamoto, 1988). Using semaphore, the ‘central price’ was sent to the top of a nearby mountain, from where the information was transmitted from peak to peak using smoke of different colours. It is said that it reached Edo, a distance of more than 500 km, in 24 hours. There are records showing that all sorts of methods were used to transmit these prices to the rice collection and distribution centres in the areas producing branded rice which had become particularly price-sensitive. Thus it was that the rice prices all over the country fluctuated almost in parallel (Shinbo, 1978; Iwahashi, 1981). If we look at the commodity distribution route, we see that traders were organized at each level, there being the following levels: traders buying from the producers; local transit traders, central brokers, wholesalers, transit traders at the place of consumption and retailers. Along that route there flowed commodities, money and information. Transactions at the central market were not carried out on a cash basis, but concluded on a credit basis. By creating credit, it was in fact possible to minimize the effect on the money transport. The central merchants with their increasing economic power were able to effect purchases at monopol- istic prices by advancing money and so on. Also, around the middle of the eighteenth century when the expansion of the consumer market levelled off with the blunting of the population increase, they showed a tendency to try to exploit their

Plate 1.2 Rice trading in Osaka’s Dojima market (from Settsu Meisho Zue [An Album of Noted Places in Settsu Province], Osaka, 1796–8).

12:27:16:11:09 Page 24 Page 25

JAPAN: SIXTEENTH, SEVENTEENTH AND EIGHTEENTH CENTURIES

monopoly to maintain their market share even in the area of the sale of goods. The Shogunate’s policies, reflecting the economic difficulties of the time, attempted to raise taxes by granting recognition to such monopoly groups. As a result, a kind of guild-like organization called kabu nakama rapidly developed (Hauser, 1974) (see Plate 1.3). Another example of merchant activity worth mentioning was the ‘money- exchanging business’. Although they were called money-changers, their activity could more aptly be described as ‘mainly financial’. What happened was due to the complexity of the Shogunate’s currency system, for the nature of the currency differed from place to place: in the area of eastern Japan centred on Edo it was gold coins; in the area of western Japan centred on Osaka and Kyoto silver ingots were the fundamental tender. A market rate differing from the official rate of exchange evolved. The biggest problem for the Shogunate and the lords was how to send the money obtained from the sale of the annual taxation rice on the Osaka market to Edo. While there was the problem of the different gold and silver currencies, there was the much greater problem of the risk associated with the transport of large amounts of cash. The people dealing with this problem were the people working in the finance business called Okawasegumi (the Exchangers Union). At the Osaka market they took in the silver the Shogunate and lords had received from the sale of the taxation rice and paid out the equivalent in gold at their Edo branch. Their number included the Mitsui ‘company’ which has become one of the leading industrial conglomerates in Japan today (see Plate 1.4). Conversely, they sought items on the Osaka market that ordinary merchants

Plate 1.3 Odenmacho Cotton Wholesale Dealer in Edo cotton store (from Edo Meisho Zue [An Album of Noted Places in Edo], Edo, 1834–6).

12:27:16:11:09 Page 25 Page 26

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

Plate 1.4 Mitsui drapery store in Edo (from Edo Meisho Zue [An Album of Noted Places in Edo], Edo, 1834–6).

could not procure on the Edo market and transported them to Edo, selling them for gold at their Edo branches. By accepting the shipper of the goods at the Osaka oﬃce, they were able to avoid the trouble of transporting cash. Many of the leading city banks of present-day Japan can trace their origins to this economic activity during the Edo period. With the explosion of this economic activity, concepts and various systems to achieve economic rationality evolved. In this way, commerce in the Tokugawa period was to lead the economy as a whole (see Figure 1.1). This is ﬁrst because of Japan’s complicated topography, which means that every tiny region used its geographically comparative advantage in some aspect to produce special commodities, and this resulted in the need for the expansion of consumer markets in order to trade them. Another factor was that with the establishment of large cities, such as Edo, Osaka and Kyoto, having populations attaining anything from 500,000 to 1 million, the low transaction cost there for commercial activities became a condition for surviving the competition, and the growth of commerce began. Large merchants set out norms to maintain their own family business according to the precepts of the household, sometimes employing anywhere from several dozen to several hundred people. They established the human resources development systems required for a business enterprise. This included systematizing promotions, internal nurturing of cadres, and helping to set people up in business (Saito, 1987). Though books of account were generally of the ‘single entry’ type, the concept of double-entry bookkeeping was already germinating, and they were totally prepared for the introduction of modern accounting systems brought from the Western World in the latter part of the nineteenth century (Ogura, 1962). Even in farming village

12:27:16:11:09 Page 26 Page 27

JAPAN: SIXTEENTH, SEVENTEENTH AND EIGHTEENTH CENTURIES

Figure 1.1 Tokugawa economy (after A. Hayami). The local lord levied an annual tax in rice or money on the peasants. Any rice not consumed was converted into money; not an ‘economic act’ as the local lord alone decided its value. However, its value progressively became aligned with market forces, and selling oﬀ the rice became an economic act. As the purchase of various products was done at market values, these were also economic acts.

society, farmers’ mutual financial organizations called ‘ko’ were established and fulfilled that function perfectly satisfactorily. The area of manual industry saw very considerable development in such fields as spinning, weaving and food-processing. The spread of raw cotton production from the plains of the main island of Japan to the sandy sea-shore area of the Seto Inland Sea permitted the spread of cotton thread and cotton cloth production. However, this did not become a pole of industrial activity as in England because the fibres of Japanese raw cotton were short and unsuitable for mechanical spinning. In contrast, raw silk was a raw material for high added value silk cloth. Particularly during the eighteenth century, the import of high quality raw silk and silk cloth decreased and the domestic production of raw silk expanded in humid areas in the valleys and at the foot of mountains as an import substitution. The cocoons from which the raw silk is unreeled are the chrysalides of the silkworm. When the silkworms are at the larval stage they consume large quantities of feed consisting of mulberry leaves, so the cultivation of mulberry was the number one problem. The process of unreeling the raw silk from the cocoon was even then done with a silk-reeling device. Often carried out as agricultural by-employment, it developed in places close to where the cocoons were produced. Furthermore, the process of producing silk cloth from the raw silk was such that the value of the product varied greatly depending on whether there was an expert technician involved or not. As a result the operation could be carried out for instance in the city of Kyoto – far from where the raw silk was produced and on a fair scale for the times. Thus the production of raw silk and silk cloth saw the development of specialization and improvement through competition in many fields, from the

12:27:16:11:09 Page 27 Page 28

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

production of mulberry, silkworm-egg paper and so on by the farmers to the production of high quality silk cloth by skilled technicians in the cities. Thus when Japan became open to the world market in the latter half of the nineteenth century, it had internationally competitive silk products able to compete with raw silk and silk cloth produced in China. As opposed to high-class goods such as silk cloth, there was a striped cloth made by blending silk and cotton, which became a mass consumption product in the last half of the Tokugawa period, and whose production spread to everywhere in the land. Thus ‘self-sufficiency’ was even disappearing from the peasants’ cloth. The food processing industry also developed with much of it being centred on brewery products such as sake, miso and soy sauce. The production of sake traditionally flourished in the vicinity of Osaka and Kyoto, with high class branded products being produced. These products were appreciated in the Edo con- sumers’ market, and a local sake produced near Edo was also good price-wise. This fermentation techniques reached high technological levels that led on to the modern industry of today. The pottery industry was another field where development occurred. At the time Toyotomi Hideyoshi was invading Korea, technical knowledge of a high order was gained by summoning Korean potters to Japan to get information. In the western part of Kyushu, ceramics and porcelain of high repute were produced and exported to Europe from Nagasaki through the Dutch East India Company. Also, all kinds of pottery products ranging from crockery and home items used by ordinary people to products used in construction works were produced wherever there were raw materials and fuel. The forestry industry must also be mentioned. One can easily imagine the vast quantities of wooden materials required during the initial construction phase of the cities and for large-scale construction works. Even subsequently, consumer markets were established to meet normal requirements for wood, for the construction and maintenance of housing in the cities and for social infrastructure. Wood differs from other products by virtue of being difficult to transport. It therefore had to be transported by water. For this, the wood was either made up into rafts or allowed to float down river as logs to the distribution centres down- stream. From there they were forwarded to Edo and Osaka on the sea lanes. In the production areas, initial needs could be met by large-scale felling of the natural forest, but they were faced with the problem of the exhaustion of woods suitable for felling and transportation, and regulations set by the lords of the fief concerning the use of forests and woodland started to be instituted. Specially designated woodlands were established, where the felling of trees by ordinary people came to be forbidden on the grounds that the woodlands belonged to the local lord (Totman, 1989). Then there were members of the public, often living in regions where wood grew quickly or where transporting it to market was easy, who set up forestry businesses. These forestry businesses did not involve felling the natural forest, but the replanting of trees in areas where they had been felled. This was a tree- nurturing industry. The tree planting and growing industry which appeared from the eighteenth century showed that the forestry industry was economically viable

12:27:16:11:09 Page 28 Page 29

JAPAN: SIXTEENTH, SEVENTEENTH AND EIGHTEENTH CENTURIES

in that form. When one considers that it is necessary to invest over a relatively long period of fifty years from the time of planting to felling, one can see it is not something anyone can do. However, families engaged hereditarily in the forestry business did spring up here and there, with their business continuing in some form even today. At first sight it would appear that the attitude of such local lords and the public to the forestry industry would be different, but in terms of results, both sides wanted to protect Japan’s forests and avoid destruction through felling, so in that sense they had much in common. At that time in Japan, wood was virtually the only resource not only for building materials but also for civil engin- eering projects, boat-building and fuel. If the local lords had not instituted protection policies and if the forestry industry run by the members of the public had not proceeded with the planting of trees, Japan’s forests would have been depleted during the Tokugawa period, leaving it covered with bald mountains. Of course one must not forget that Japan had the advantage of an often warm and very humid climate in which trees grow quickly. The mining industry suffered from a technical bottle-neck, with an early falling off of output. With the increasing depth of the tunnels, getting rid of water and supplying air became a problem, and with no external energy it was insoluble. Precious metals such as gold and silver were in most cases mined under the direct control of the Shogunate, with the work being done by convicts. This was known as ‘hard labour’. For copper, there was investment and extraction all over the country by the daimyo and the merchants, and until quite a while later it was exported. Copper extracted at the mines in Shikoku, and then refined in Osaka for shipment to The Netherlands from Nagasaki by The Dutch East India Company in the form of rods had a set role of making prices fluctuate on the European copper market (Glamann, 1978). However, no one in Japan knew this at the time. For iron there was the ‘foot bellows iron-making method’ with iron sand as the raw material. Since the raw material and charcoal for fuel were easy to come by, the operation could be carried out almost anywhere. Though large amounts of charcoal were required as fuel, forestry resources were not depleted because trees grew quickly on account of the high temperature and humidity at the latitude where Japan is situated. Although coal was finally used as fuel in such industries as salt-making at the end of the Edo period, it was not used prior to that. Overall, mining resources were varied, but quantities were always small. Even if one takes gold, for which Japan was famous in the sixteenth and seventeenth centuries, we find that Japan had become a net importer by the latter half of the eighteenth century. The demand in a pre-industrialized society could only just be met; but with the coming of the age of real industrialization the insufficiency of resources was immediately felt. From this we can see that even as regards construction materials, instances of the use of mineral resources or large quantities of cut stone were rare, and that construction using wood, earth and stones, and ceramic products was the usual practice. Rather than giant multi-storey buildings, constructions that were low with many complicated features were preferred – together with gardens – whether it were for temples and shrines or for private houses. However, these buildings

12:27:16:11:09 Page 29 Page 30

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

resist poorly to earthquakes, fires, lightning and damp, so the only available way of storing property safely was in thick-walled earth warehouses. Although anti-fire organizations and thinking about fire-prevention had developed, there was no way to stop a fire when it was dry with a strong wind, so the possibility of cities being totally burnt in almost an instant was a constant threat.

THE EIGHTEENTH CENTURY – STABILITY AND CHANGE The seventeenth century was truly the ‘Great Land Reclamation Period’. Although there are no accurate statistics, the area under cultivation would seem to have increased by close to two times. The population had also increased to somewhere between two and a half to three times its former level. Population growth in the cities was remarkable. It can be seen that cities (roughly 200 in number) with populations of 5,000 or more represented 15 per cent of the total population at the outset of the eighteenth century, with Edo having a population of 1 million, Osaka 500,000, and even Kyoto also with 500,000 (Rozman, 1973). The important thing was that these cities had an economic function by forming a network between the cities themselves and between the cities and the farming villages. People, goods, money and information went back and forth. Economically, a national economy had been established; people became economy-minded; and it had become a society acting economically. Of course, the level of production technology was low. In agriculture, human-power rather than live-stock was the main source of energy for ploughing the ﬁelds. In manual industries, apart from the water-wheel, there really was not any incorporation of external energy. Manual industries that had developed at that time, such as spinning, weaving, miso-making and ceramics-making, all used machines operated by human-power. However, the development of labour-intensive farm industries and production industries depending on such human-power held the danger of ‘reinforcing’ labour, and made people hard-working. Of course, this is subject to the important proviso that extremely hard work over a long period oﬀers the prospect of being rewarded. In societies where all the fruits of hard work over a long period end up in other people’s hands, labour is only penal servitude and people would do all they could not to work. In such societies the motivation to work is compulsion. Societies where compulsion is the motivation have very low productivity. However, the situation in Tokugawa Japan was not like that (see Plates 1.5–1.7). In such conditions, extremely hard work was considered not to be ‘slave labour’ but a virtue because economic gain was obtained exclusively through labour. So among the ordinary people the work ethic of labour being a ‘virtue’ was born. Moreover, inside the family this concept of work being a virtue passed from parent to child. Outside the family, this began to be taught as a norm to follow in the educational institutions that had sprung up throughout the country from around the eighteenth century, such as the private schools in the cities and the ‘temple schools’ in the countryside. One may call this an ‘industrious’ revolution! (Hayami, 1989). In the world of literature, at the end of the seventeenth century, there was Ihara Saikaku (1642–93) who wrote about the Osaka merchants’ activities and their

12:27:16:11:09 Page 30 Page 31

JAPAN: SIXTEENTH, SEVENTEENTH AND EIGHTEENTH CENTURIES

Plate 1.5 Ploughing in Early Spring (from Nogyo Zusetsu [An Illustration of Agriculture].

view of the world. In his works he enthusiastically describes the process whereby they became rich through hard work to finally become very rich merchants. The underlying theme of his works is how the common folk seek worldly happiness as an extension of their daily lives. Saikaku’s literature was widely read during the Edo era. If one regards the Genji Monogatari (The Tale of Genji) in the early eleventh century, whose main theme is life at the palace of the nobles, and Tsurezure-gusa (Essays in Idleness) in the fourteenth century, about refusing the present world, as representative of Japanese literature before the seventeenth century, one can see just how great was the change in the people’s values and mentalité. With the advent of the eighteenth century following the seventeenth century with its expansion, changes came about in various domains. Firstly, the expansion of the area under cultivation had reached the extreme limit possible with the then technology. Almost all of Japan’s arable land (about 15 per cent of the total area) had been put to use. To increase production even further, the only option was to increase production per unit area. At the same time, taking the country as a whole, the population per se was close to its ceiling and the urban population ratio had reached its limit. The farmers were forced to utilize the land more intensively, with deep cultivation and heavy fertilizing. Even fish, sardine and herring, were used as a fertilizer after drying and pounding. This stimulated the fishing industry throughout Japan, not only to catch fish as a food but also as a fertilizer. Fishing

12:27:16:11:09 Page 31 Page 32

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

Plate 1.6 Rice Planting in Late Spring (from Nogyo Zusetsu [An Illustration of Agriculture]).

villages flourished and the fish markets in the urban areas were established with high-spirited call. Right at the end of the seventeenth century, the supply of mineral resources – which had even had surplus capacity for exports after meeting increasing internal demand – suddenly fell. Not only did exporting become difficult, but what is more, supplying enough to meet the needs for the currency used in Japan became difficult. As a result, in 1696 the Shogunate engaged in a currency debasement operation, reducing the amount of gold or silver in the metal used in minting to 80 per cent of the level it had previously been. From the start of the eighteenth century this debasement process accelerated, so that silver coins struck in 1708 were of extremely bad quality – being, it is said, only 20 per cent pure. Thus, at a stroke, trust was lost in the Japanese gold and silver currencies which up until then had been the most trusted currency in East Asia, and distributed as a kind of international currency. Finally, the Shogunate again started making high quality currencies, but the amount of precious metal in a money of a given value was lower than before the debasement process started. Furthermore, in order to meet the increasing demand for currency, non- metallic currency came to be issued by the daimyo. There was a form of currency, called hansatsu (domainal paper money) whose distribution was restricted to within the territory. Though the Shogunate’s permission was required, it had the

12:27:16:11:09 Page 32 Page 33

JAPAN: SIXTEENTH, SEVENTEENTH AND EIGHTEENTH CENTURIES

Plate 1.7 Rest after the Good Harvest – a holy day after cultivation (from Nogyo Zusetsu [An Illustration of Agriculture]).

following two results in conjunction with the Shogunate’s debasement of the currency. The first result was that from that time on, the people living in Japan ceased to pay attention to the quality of the currency. Especially when debasement first took place, new and old coins were in circulation at the same time and confusion occurred. However, when they got used to it, they accepted low quality coins and hansatsu. There is a link between this and the fact that the new Meiji Government was later domestically able to overcome its first financial crisis brought about by the issue of a large quantity of paper money unbacked by precious metal. The second result was that the Shogunate were denied the exclusive right to issue money, even though the issuing of hansatsu was subject to certain conditions. Prior to the issuing of hansatsu, the Shogunate held the monopoly regarding the issuing of silver and gold coinage. The Edo Shogunate that had prided itself on having the real power as the central government largely thanks to that monopoly, found that being denied the right to that monopoly was the first step in its downfall. In brief, the Shogunate lost the power to control the economy through the issuing of currency. Even in the cities, the situation changed radically when the seventeenth century population increase and the associated rise in demand stopped. During the seventeenth century the Shogunate did not interfere at all, on the principle that ‘trade

12:27:16:11:09 Page 33 Page 34

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

should be free’. Powerful merchants gathered in the cities, while a network of wholesalers, brokers and retailers was formed. However, when the increase in demand and increase in population stopped in the eighteenth century, the merchants tried to hold onto their existing share and started to set up exclusive monopoly groupings. The complexities so characteristic of Japan’s present-day product distribution system are to be found in the complex distribution- organization established during this period. The Shogunate and even the daimyo were able to procure ‘recognition fees’ by giving them official recognition. This was to try and compensate for the lack of increase in annual taxation income from the villages then unable to pay more. In the eighteenth century, many changes took place in the Shogunate and in the daimyo domains. With the various systems institutionalized at the beginning of the Edo period being unable fully to cope with the subsequent changes in situation – and notably those brought about by economic development – there were many cases where revision was inevitable. However, if one looks at it in terms of financial policies one can find a very interesting feature. This is the way the financial expansion and contraction policies repeated in a thirty to forty year cycle. For example, in the middle of that century, the Shogunate carried out a great number of public projects, promoted foreign trade, minted coins in keeping with the needs of the time and approved merchants’ associations. In brief, they pursued positive economic policies. However, natural calamities followed, and the persons who had figured prominently in promoting those policies were dismissed on the grounds that they were against tradition. A conservative faction then grasped the real power. At that juncture, commerce regressed and policies stressing farming were adopted. ‘Belt-tightening’ and a return to the spirit prevailing at the time the Shogunate was established was stressed. But the already developed market economy would not allow the pursuance of retrenchment policies, and the promoter of those policies retired in disgrace and expansionist policies were reverted to. Finally, this wave motion must have repeated itself with a certain amplitude between limits. On the one hand, there was the accepted principle underlying the formation of the pax Tokugawa, and defying it to return to a simply self-sufficient society was impossible so a market economy had to be allowed to a certain degree. On the other hand, they could not threaten the pax Tokugawa by permitting the spontaneous development of the market economy. As a result, there were pendulum-like swings between retrenchment policies and expansionist policies, or between spiritualism and materialism. The conservative factions and progres- sive factions were battling it out everywhere, with inter-personal battle lines drawn. Also, at this time the Shogunate’s policies did not extend throughout the country, and daimyo were able to pursue the policies independently of the Shogunate. There were many instances of the lords implementing expansionist policies while the Shogunate was adopting retrenchment policies. Consequently, the Shogunate’s policies rarely fully achieved their objectives. Very often they simply terminated prematurely half-way, and the cyclic movement repeated itself. Following the re-coinage operation in the 1730s, for about one century prices remained almost stable. The annual interest rate remained unchanged at about

12:27:16:11:09 Page 34 Page 35

JAPAN: SIXTEENTH, SEVENTEENTH AND EIGHTEENTH CENTURIES

10 per cent. One can say that economically, long-term stable conditions lasted. In that sense one can think of this period as being a histoire immobile period. How- ever, this was only the superficial appearance; in fact changes were taking place internally. For example, let us take the question of the population. It is said that Japan’s population levelled off at about 30 million. The generally accepted view used to be that this stagnation was caused by the farmers resorting to birth control because of poverty. But there is another view that it represented the farmers ‘positive’ means of increasing their standard of living (Hanley and Yamamura, 1977). However, what must be borne in mind is that there were great differences from region to region. In northern Japan, eastern Japan and in the vicinity of the big cities the population decreased, but in south-west Japan the population increased slowly but surely. The decrease in population in northern Japan and eastern Japan in the latter half of the eighteenth century can be explained by the cold weather conditions. Furthermore, cold temperatures continued to prevail during the (rice) maturing period in the summers for successive years. The impact of this cold weather was particularly great because much of the land in question had been brought into cultivation during the period of great expansion in the seventeenth century and was land close to the ‘northern limit’ for rice cultivation. Japanese agriculture consisted essentially of rice cultivation and with the technology available at the time, northern Japan had a northern limit for rice cultivation which therefore made it particularly sensitive to low temperatures in summer and not possible to find alternatives. Although historical documents tend to exaggerate when they mention hunger, the population in this area twice went down by 20 per cent because of cold weather – that is in 1756 and 1783. It took a long time for it to recover. Examination of the population trend for this region, shows that over the Edo era the peak was at the beginning of the eighteenth century, and that the population in the Meiji Restoration period (1868) did not return to the peak level. There are cases where the fall in population was directly caused by hunger, such as where people actually died of hunger. However, in many cases people died because the decrease in nutrition during famines reduced people’s resistance to disease. Ironically, in Northern Japan early marriage was the custom, and the average age of marriage for girls was around 15 years old. If couples continued to procreate throughout their lives they would have more than ten children. In fact, only three or four infants would grow into adults, because the marriage would sometimes be terminated through the death of one of the partners or because of the high infant mortality rate. During periods of famine and subsequently, there might well be a force at work making people try to limit the population. Although the marriage age became a little higher in this region, direct population control, usually by abortion or infanticide, had become a custom. Some of the daimyo became so worried that the population recovery was not going well that they strictly prohibited abortion and infanticide, and undertook a survey of pregnant women to preclude it. They even adopted a policy of giving subsidies to help those who gave birth bring up the child. Furthermore, the population in the vicinity of Edo, Kyoto and Osaka did not increase because of the many people migrating to the big cities. This phenomenon

12:27:16:11:09 Page 35 Page 36

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

can be explained by the urban ‘graveyard’ theory. That is to say, the urban mortality is higher than the fertility. Without considerable migration from the countryside they could not maintain their population levels. Other factors are involved in addition to mortality and fertility. They are imbalance of the sexes; a high ratio of single people; residential instability, and so on. Anyway, as the population replica- tion ratio in the city was negative, migration from the surrounding farming villages was surely necessary. The high population density in the cities meant that once an epidemic decease struck, the chances of dying were much greater than in the countryside. In big cities like Edo, people’s daily food was often such that (good) taste was given precedence over nutritional value. Lack of vitamin B and of minerals often led to symptoms similar to beriberi, and even to the extent that this was referred to as the ‘Edo sickness’. Migration to the city from the countryside was important. According to research on rural populations, one can see that there was very considerable migration. This did not always represent people intending to become permanent residents; often it was just people migrating for temporary work. According to a case study in one village located between Kyoto/Osaka and Nagoya with a population of 100,000 at that time: of the men and women having attained 10 years of age, 50 per cent of the men and 60 per cent of the women left for temporary labour migration. Both men and women of 13 and 14 years of age left the village to go to the city, and continued working there for twelve or thirteen years, after which half of the men and two-thirds of the women returned to the village. Men and women who did not return might have died in the city before then, or have decided to stay on permanently. This the phenomenon of temporary labour migration controlled the village population in two ways. The first was that those deciding to stay permanently in the city or dying while working away from home represented a direct factor reducing the absolute population of the villages. The second was that the men and women who returned from working away from home married late with a consequent reduction in offspring. If one considers the difference in average female marriage age, between the returnees and those having stayed in the village as being about five years and representing a difference of about two children, the degree of population limitation occurring as the indirect effect of working away from home – that means a kind of indirect population regulation effect – may be surprising (Hayami, 1992). Therefore, the presence or absence of such a regulatory action – that means the presence or absence of a city in the vicinity to absorb the population – produces big differences in the population trends, and moreover in the relation between population and society in the regions in question. In central Japan where the population regulation action worked – there are big cities and also a higher ratio of urban population – population pressure was relatively weak, and so the degree of social instability was relatively low. In contrast, in western Japan where there were no big cities and the urban population ratio was lower, it became a factor in making population increase continue. Later, population pressure built up and was to become a cause of social instability. Though based on a limited geographical area, a study of the average longevity indicates that at the beginning of the Edo era, it was roughly 30 years, and that

12:27:16:11:09 Page 36 Page 37

JAPAN: SIXTEENTH, SEVENTEENTH AND EIGHTEENTH CENTURIES

half of live-birth offspring were dead by the age of 10. In the latter part of the Edo era, life expectancy in normal times had risen by close to ten years, and the child death rate was two-thirds of that at the beginning (Hayami, 1973b). In the context of the period prior to the introduction of modern health-care and public hygiene, it is difficult to see any other reason for this improvement in anything other than in improved living conditions. The clothes worn by ordinary people, changed to cotton from the previous ‘mixed’ fibre. The mixed fibre cloths were ‘unhealthy’ in that they could only be coarsely woven, lacked resistance, behaved poorly with regard to the cold and humidity, and could not be washed often. In contrast, clothes made from cotton do not have these drawbacks. They wash well, and have high resistance to mois- ture. Even this brought about revolutionary progress in the lives of the people, since clothing was, like housing and food, of key importance in their lives. Although specifically forbidden by law, the wearing of clothes made of silk became generalized on festive occasions. Also, to produce the raw materials for making cotton, raw silk and flax, various industries such as spinning and weaving developed throughout the country subject to local conditions. The production of raw cotton was concentrated around the sandy areas, like those in the vicinity of Osaka or the coast around the Seto Inland Sea. Raw silk was mostly produced in eastern Japan on gentle slopes near where the mulberry trees grew. At the beginning of the eighteenth century, the amount of silk cloth imported from China had decreased and the demand for domestically produced raw silk and silk cloth grew. As a result a good quality product with high added value began to be produced widely as a local speciality. For dyeing the cloth, production of safflower and Japanese indigo was started in specified places, also as a local speciality. As regards food, there is lack of data, but it seems that rice-based food had become generalized, with animal protein coming from chicken eggs. Fish was dried and salted and one can find traces of fish merchants even in the most remote places. The development of the brewing industry was remarkable and sake, miso and soy sauce could be found in most kitchens. An essential ingredient was salt and its production by the new ‘salt farm method’ in the Seto Inland Sea went ahead. It is said that the Seto Inland Sea area alone produced 4.5 million koku (1 koku = approx. 180 litres). With the commandments of the Buddhist religion saying meat should be avoided, the only meat might be the occasional chicken, so nutrition in the Japanese meal was mainly centred on carbohydrates. Caloric intake was rather low. In eighteenth century Japan people could enjoy continuous peace and it was a period when development of various aspects of ‘consumer’ life occurred (Hanley, 1990). For example, in the publishing sphere, the religious books and highly academic books published in large quantities in the seventeenth century gave way to newly compiled and published material. This included dictionaries, literature, maps, travel books, geography books, encyclopaedias and directories. The publishing centre of Japan, which up to then had been Kyoto and Osaka, shifted to Edo in the course of that century. Travel, ostensibly to visit temples and shrines, became organized, and a travel industry became established with intermediaries gathering clients together and arranging accommodation on a contract basis in tourist spots. As regards

12:27:16:11:09 Page 37 Page 38

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

communications, two large private postal companies provided a country-wide service, though most of their operations were between Edo and Kyoto/Osaka. Thus even before Japan became a modern country, it had a domestic communications network. Transportation by water had limited possibilities in Japan because of the topography. It was developed to its maximum, and included an inland network for the distribution of goods. However, maritime transport was very developed. Between Edo and Osaka two maritime transport groups were in severe competition. Another busy route to Osaka was a roundabout route via the Japan Sea and the Seto Inland Sea. As far as the transportation of goods was concerned, the development of coastal transportation routes in the Edo period meant that the needs of markets established throughout Japan could be met. Japanese sailing ships had no centre-board and weak keels, so sailing into the wind was difficult and there were many shipwrecks in consequence. The construction of large ocean-going sailing ships did not come about because the Japanese themselves were not allowed to engage in overseas trade. In contrast, road transportation for cargos was poor, and apart from in a few areas, transportation facilities were never used by the general public. And since the Shogunate’s policy was not to build permanent bridges across the large rivers, ordinary people could only travel by foot. As regards eating and drinking, high-class ryotei (special Japanese restaurants) sprang up in the three major cities and in big cities built around castles. For the ordinary people, establishments serving light meals and tea appeared. Thus the cities and sightseeing spots became very convenient. In the latter half of the eighteenth century, this was accompanied by the production of local specialities in many places. These often were cakes or other delicacies. Thus a food culture was established, and was reckoned to be ‘one of the pleasures of life’. Facilities for the ordinary people’s amusement were mainly concentrated in the cities; they include all the amusements so characteristic of present Japan such as theatrical plays, sumo (Japanese wrestling), raku-go (comic story-telling), naniwa- bushi (narrative ballad), kodan (serious story-telling). The town citizenry would have their favourites, and followed the doings of popular performers avidly. Com- petitive games such as go and shogi (Japanese chess) became the most popular of these. However, ‘poetry games’ just like mikasa-zuke that were half-educational, half-play were popular. These often involved using parts of famous poems as a starting point for a group of competitors to invent additional parts, thus producing a game in which contestants would learn the original poem (see Plate 1.8). Also, during this period, practising waka (31-syllable verse), haiku (17-syllable verse), calligraphy, flower arrangement, the tea ceremony, traditional dancing, and musical instruments such as the koto or shamisen, became part of a proper education, and particularly for girls. Mastering these came to be seen as the proper accoutrement for a cultured girl. Becoming an instructor in such a field came to be regarded as a proper profession, and such instructors came to represent a set proportion of the citizens of the cities. However, the ultimate in amusement was going to the ‘amusement quarters’. In Japan the amusement quarter was not just a brothel quarter. People also went there to enjoy high-level conversation with cultured women, and to eat and drink.

12:27:16:11:09 Page 38 Page 39

JAPAN: SIXTEENTH, SEVENTEENTH AND EIGHTEENTH CENTURIES

Plate 1.8 The Kabuki Theatre in Osaka (from Settsu Meisho Zue [An Album of Noted Places in Settsu Province], Osaka, 1796–8).

A great feature of these establishment was that people could really relax because, once inside, differentiation between samurai and ordinary people, between rich and poor, disappeared, and only differentiation between those with ‘class’ (amus- ing themselves in style) and the uncouth boors remained. These ‘amusement quarters’ sprang up in the large castle towns as well as in the three major cities. To conclude, in eighteenth century Japan, a consumer culture, which had not existed in the seventeenth century and before, developed just like the social system and economic systems. Although these developments started in the cities, they soon spread throughout the country due to the daily human exchange between cities and the countryside, and between the centre and outlying regions. Upper class farmers could even achieve the same high standard of intellectual life as city residents. It can be summed up as an age of mass culture. Of course this does not mean a total absence of high culture (Keyes, 1988; Mason, 1992; Nishi, Hozumi and Torton, 1985; Noma and Webb, 1966; Schneider, 1973; Swan, 1979). Artists exclusively hired by the daimyo produced highly artistic works, and some nobles even became patrons of artists though they themselves were not rich. However, if one had to choose a single phrase to sum up the character of the art and literature of the Edo period one would have to say it was ‘popular’. For better or worse, it produced the prototype for the cultural state of present- day Japan. It was something developed independently, without influence of the Western World, and after having broken free from the influence of China. Would it not be appropriate to consider it as being a form of new cultural creation? Writing like this may give the impression that Tokugawa Japan was some kind of paradise. In fact, it was not necessarily so by any means. In the shadow of this prosperity there lay much discrimination, which had always existed, but only really became apparent during this period. For instance, those in charge of

12:27:16:11:09 Page 39 Page 40

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

slaughtering (animals) or handling leather products – who were discriminated against because Buddhist philosophy said these tasks should be avoided – were forcibly cut off from ordinary people and even obliged to form their own villages besides being allotted the lowest rank in society. Another example can be seen in the sacrifice of the indigenous Ainu tribe in Japan’s northern island of Hokkaido when the Japanese came to extend their sovereignty effectively to them. Notwith- standing the Great Revolt at the end of the seventeenth century, the Ainu lost their land, and were forced to work for the Japanese under harsh conditions. Without doubt their population decreased considerably. Japan also lost much through the developments in the Edo period described above. First, isolation meant that the people the Japanese encountered in their daily lives were limited to Japanese. A mode of communication between people having a common language and the same customs developed, but as they had no chance to be in contact with people not sharing the same language and customs, they became very bad at international negotiation. The effect can still be seen today. The Japanese developed a flair for coexisting in society, while making the most of restricted territory and limited resources. But when they left the territory, they found their ability to act freely to have diminished. One could say that the present nature of the Japanese – namely, to be good at operating systematically in groups, but to be not good at individualistic activities – was a product of those days. These may be perceived as problems in terms of the values of the West which took the lead in forming modern society. However, as no one at that time could predict the course of history, the development of Japan in the Edo period should perhaps be regarded as having been a genuine cultural creation that should be judged in its own right.

BIBLIOGRAPHY Boxer, C. 1951. The Christian Century in Japan 1549–1650. Berkeley/London. Camões, L. de. 1963. The Lusiads. Bullough, G. (ed.). Carbondale, Illinois. Elison, G. 1973. Deus Destroyed. The Image of Christianity in Early Modern Japan. Cambridge, Mass. Glamann, K. 1978. Nihon do to 17 seiki yoroppa no pawa poritikkusu [Japanese Copper and European Power Politics in the Seventeenth Century]. Tohogaku. (Tokyo), no. 56, July 1978, pp. 103–11. Hall, J. W.; Jansen, M. B. (eds). 1968. Studies in the Institutional History of Early Modern Japan. Princeton. Hanley, S. B. 1987. Everyday Things in Premodern Japan: The Hidden Legacy of Material Culture. Berkeley and Los Angeles. Hanley, S. B.; Yamamura, K. 1977. Economic and Demographic Change in Pre-Industrial Japan, 1600–1868. Princeton. Hauser, W. B. 1974. Economic Institutional Change in Tokugawa Japan. Osaka and the Kinai Cotton Trade. Cambridge. Hayami, A. 1971. Mouvements de longue durée et structures japonaises de la population à l’époque de Tokugawa. Annales de Démographie Historique (Paris), pp. 247–63. —— . 1973a. Nihon niokeru keizaishakai no tenkai. [Development of the Economy- Minded Society in Japan]. Tokyo.

12:27:16:11:09 Page 40 Page 41

JAPAN: SIXTEENTH, SEVENTEENTH AND EIGHTEENTH CENTURIES

—— . 1973b. Kinsei noson no rekishi jinkogakuteki kenkyu. Shinshu suwa chiho no shumon-aratame-cho bunseki. [An Historical Demography of Early Modern Japanese Community of Suwa, Shinano]. Tokyo. —— . 1986. A Great Transformation: Social and Economic Change in Sixteenth and Seventeenth Century Japan. Bonner Zeitschrift für Japanologie. Vol. 8, pp. 3–13. Bonn. —— . 1989. Kinsei nihon no keizaihatten to Industrious Revolution. [Economic Devel- opment and Industrious Revolution in Early Modern Japan]. In: Hayami, A. et al. (eds). Tokugawa shakai karano tenbo, pp. 19–32, Tokyo. —— . 1992. Kinsei nobi-chiho no jinko – keizai – shakai. [Population, Economy and Society in Early Modern Japan: A Study of the Nobi Region]. Tokyo. Hayami, A.; Miyamoto, M. 1988. Keizaishakai no seiritsu. 17–18 seiki. Nihon-keizaishi 1. [Emergence of Economy-Minded Society. Economic History of Japan. Vol. 1]. Tokyo. Iwahashi, M. 1981. Kinsei nihon bukkashi no kenkyu. Kinsei beika no kozo to hendo. [A Study of the Price History of Early Modern Japan. Structure and Fluctuations of the Price of Rice]. Tokyo. Keyes, R. S. 1988. The Male Journey in Japanese Prints. Berkeley. Mason, P. 1992. History of Japanese Art. New York. Miyamoto, M. 1988. Kinsei nihon no shijo keizai. Osaka kome shijo no kenkyu. [Market Economy in Early Modern Japan. A Study of Osaka Rice Market]. Tokyo. Nakai, K. W. 1988. Shogunal Politics. Arai Hakuseki and the Premises of Tokugawa Rule. Cambridge, Mass. Nishi, K.; Hozumi, K. 1985. Torton, H. M. (trans.). What is Japanese Architecture? Tokyo/New York. Noma, S. 1966. Webb, G. T. (trans.). The Arts of Japan. Late Medieval to Modern. Vol. 2. Tokyo/New York. Ogura, E. 1962. Goshu nakai-ke choai no ho. [Accounting Books of Nakai Merchant Family in Omi Province in Early Modern Japan]. Kyoto. Parker, G. 1988. The Military Revolution. Military Innovation and the Rise of the West, 1500–1800. Cambridge. Rozman, G. 1973. Urban Networks in Ch’ing China and Tokugawa Japan. Princeton. Saito, O. 1987. Shoka no sekai – Uradana no sekai. Edo to Osaka no hikaku toshi-shi. [The Merchant Life in Early Modern Japan. A Comparative Urban History of Edo and Osaka]. Tokyo. Sakai, R. K. 1968. The Ryuku (Liu-ch’iu) Islands as a Fief of Satsuma. In: Fairbank, J. K. (ed.). The Chinese World Order, pp. 111–34. Cambridge, Mass. Schneider, E. H. 1973. A History of Japanese Music. London. Shinbo, H. 1978. Kinsei no bukka to keizaihatten. Zenkogyoka shakai heno suryoshiteki sekkin. [The Price and Economic Growth in Early Modern Japan. A Quantitative Historical Approach to a Pre-Industrial Society]. Tokyo. Smith, T. C. 1959. The Agrarian Origins of Modern Japan. Stanford. Swan, P. C. 1979. A Concise History of Japanese Art. Tokyo/New York. Tashiro, K. 1981. Kinsei nitcho tsuko boekishi no kenkyu. [Trade and Diplomacy between Japan and Korea in Early Modern Period]. Tokyo. Toby, R. 1984. State and Diplomacy in Early Modern Japan. Asia in the Development of the Tokugawa Bakufu. Princeton. Tonomura, H. 1992. Community and Commerce in Late Medieval Japan: The Corporate Villages of Tokuchin-ho. Stanford. Totman, C. 1989. The Green Archipelago: Forestry in Pre-industrial Japan. Berkeley.

12:27:16:11:09 Page 41 Page 42

First published in Silkworms, Oil, and Chips . . . (ed. E. Pauer), (Bonner Zeitschrift für Japanologie, vol. 8. pp. 3–13), Bonn, 1986.

2 A Great Transformation: Social and Economic Change in Sixteenth and Seventeenth-century Japan

INTRODUCTION apan’s emergence as an ‘economy-minded society’, the most crucial pre- Jrequisite of its industrialisation and modernisation, is the theme of this paper. When did this emergence occur? Should we phrase our discussion in terms of the ‘ancient’ (kodai: 7th to 12th cent.) and ‘medieval’ (chu¯ sei: 12th to 16th cent.) periods into which Japanese history is usually divided? Without doubt, currency was in circulation and money was being lent and borrowed during these periods; people lived, produced, consumed, and performed other economic activities. According to world economic history as it is broadly interpreted, we could even begin our account several millenia B.C., when people lived in hunting and gathering societies. But for the reasons which follow, I believe that every nation’s economic history should have its own particular periodisation with the emergence of ‘economy-mindedness’. In all periods of history, human beings have laboured in order to live; they produced and consumed. They exchanged and distributed goods after realizing the benefits that accrued from a division of labour. But in earlier periods of Japan’s history, as elsewhere in the world, people did not behave in an ‘economically-minded’ way. Human beings are multifaceted creatures, and in some historical periods they view politics, religion or customs higher than economic rationality in determining their actions. By behaviour that is ‘economically-minded’, I mean specific actions rationally adopted with the aim of maximising profits or economic utility and minimising loss at a given level of technology. The following simple example should clarify this. Imagine that there are two stores selling the same quality of some commodity in the same quantities, but at different prices. Few people today would knowingly buy from the more expensive store. But people have not always acted in such an ‘economically- minded’ manner. In earlier periods of history, people might have refused to buy from the less expensive store on the grounds, for instance, that its owner believed in a different religion. Even today, not all people are total homo economics: in a fit of passion or for irrational reasons, they may behave in a non-economic, or even anti-economic manner. Pluralistic values lead human beings toward one or another form of utility

12:27:16:11:09 Page 42 Page 43

A GREAT TRANSFORMATION

maximisation suited to each particular time and situation. Economic-mindedness is not the only form of utility to be maximised. Yet in the modern world we discover that economic utility accounts for a great deal in people’s lives and that once they have created an economically-minded society, it is virtually impossible for them to regress to earlier forms of thought and action. Economic utility has enriched people’s material culture, lengthened their life-span, and, in various ways, fulfilled the increased expectations which it created in men’s minds. Industrialisation could then emerge and further strengthen this tendency. However, the world’s various nations and peoples have not advanced to industrialisation by the same route. Generally speaking, Western European countries, new nations such as the United States, Canada and Australia, which derived from Western Europe, and Japan achieved industrialisation through a market economy or what may be called capitalism. On the other hand, states like the USSR or the People’s Republic of China achieved it through socialism or what John Hicks has called a ‘command economy’.1 Since the balance of military power or other factors in world affairs often do much to determine which methods a particular nation will adopt, this choice cannot be explained solely in terms of history. Even so, in countries which began industrialisation prior to the late nineteenth century or where the necessary preconditions (among which the precedence of economic utility over other forms of utility is the most important) were fulfilled, a market economy and economic development, spurred on by the workings of price mechanism, appeared. For that reason, it is permissible to seek the preconditions of industrialisation in Japan’s history. Indeed, this is a central issue which economic historians should pursue.

SHARP CONTRAST IN JAPANESE SOCIETY BEFORE AND AFTER 1600 First, I would like to determine when the maximisation of economic utility in Japan began. With two reservations, I assert that it appeared first in the late sixteenth and seventeenth centuries. We should realize that in Japan this transformation to an economically-minded society and to industrialisation did not take place all at once. In Western Europe, countries such as Britain, France, and the Netherlands were locked in fierce competition: if one of them was outdistanced by the others, she faced the threat of extinction. For this reason, all these countries pushed ahead with economic-mindedness and industrialisation more or less at the same time and relative speed. In contrast, Japan, an isolated island nation free from threatening neighbours, transformed itself at a more leisurely pace. Moreover, Japan did so at widely varying rates in different regions, and retained some traditional values and social customs. All of this precludes any precise statements about when the country as a whole attained economic- mindedness and began to industrialise and modernise. Second, since there are few literary source materials in Japan much before 1600 that tell us about social conditions, such as how peasants lived and produced, no real contrast can be made between pre-1600 and post-1600 socio-economic conditions through primary historical sources. Not until Japan’s ‘early modern’ (kinsei: late sixteenth century to 1868) period, which begins with the three unifiers, Oda Nobunaga (1534–82), Toyotomi Hideyoshi (1536–98),

12:27:16:11:09 Page 43 Page 44

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

and Tokugawa Ieyasu (1542–1616), did source materials assume a new character which enables researchers to discern with accuracy what village and town life was like. This fact itself tells us that there would be sharp contrasts in attitude between the medieval and early modern periods, and lends support to our hypothesis that a crucial transformation in mentalité and behaviour occurred in the late sixteenth and seventeenth centuries. After this transformation, the Japanese people assumed a decidedly this-worldly, secularised attitude towards life; few gave their hearts over to Buddhism, the religion which had enthralled their ancestors when it arrived from the Asian mainland a thousand years earlier. One cannot help noting the vastly different values reflected in the works of early modern literature produced by Ihara Saikaku (1642–93) or Chikamatsu Monzaemon (1653–1724) and ancient classics like the Manyo¯shu and Genji-Monogatari or medieval masterpieces such as Tsurezuregusa and the Taiheiki. Although the early modern Tokugawa period produced no towering masterpieces on the scale of Genji, it witnessed a higher degree of literacy among all social strata and a vaster outpour- ing of literary publications than any previous period in Japanese history. With the Tokugawa era, a form of mass culture appeared in Japan. In O¯¯ saka, a highly sophisticated marketing and credit system was established by the end of the seventeenth century. Sardines caught and dried in the fishing villages of eastern Japan could be transported more than 500 kilometres (300 miles) to O¯¯ saka to be used as fertilizer for cotton cultivation in the surrounding villages. A large proportion of the cotton fabrics made in O¯¯ saka area were shipped to eastern Japan to make clothes for peasants.2 This type of commerce which was conducted over long distances and in large quantities did not exist before 1600. Moreover, in Tokugawa Japan, large-scale sea transport flourished and geographical advantages were utilised by various localities. This early modern period formed customs and attitudes that now are deemed ‘distinctively’ Japanese; it made the Japanese economy, society, and culture what they are today. As economic historians, the periodisation we adopt can and should differ from that derived from other types of history. A change in regimes or ruling structures does not necessarily entail rationalised socio-economic behaviour. Instead we stress the continuities rather than the dissimilarities between Tokugawa Japan and the supposedly ‘modern’ period which followed it. And we hold that the decisive turning point (although it cannot be pinpointed) heralding the start of Japan’s ‘modernity’ occured in the late sixteenth and seventeenth centuries rather than after the Meiji Restoration and the industrialisation which followed.

ECONOMIC CHANGE IN TOKUGAWA JAPAN During the Tokugawa era, roughly speaking, the population trebled, the amount of cultivated land doubled, and agricultural output quadrupled.3 Such far- reaching changes were unknown prior to the Tokugawa era, and together constitute a ‘Great Transformation’4 in Japanese history. It was a qualitative as well as a quantitative change. Although these ﬁgures are not absolutely precise, one should note that there are obvious diﬀerences in degree of growth. That output grew more rapidly than population means that the per capita income could grow, and

12:27:16:11:09 Page 44 Page 45

A GREAT TRANSFORMATION

the fact that the area of cultivated land grew less rapidly than output means that the land must have been used more effectively or intensively. Such differential growth should have produced structural changes in production. The greatest degree of land and labour-intensive agriculture was achieved in Tokugawa Japan. The seed/yield ratio of rice reached about 1:50, and the average area of cultivated land per agricultural household fell to under 100 ares (2.5 acres). Yet this was still enough to maintain foodstuffs for a total population of 35,000,000 in the last decades of the Tokugawa era without imports. However, a recent study makes clear the decreasing use of lifestock for cultivation. Considering the growth of output, there must have been a conversion from ‘horsepower’ to ‘manpower’ in rural Japan. The term industrious revolution can be applied to this change, which I will examine on another occasion. In over-simplified terms, this transformation was from a static to a dynamic economy. Hitherto, Japanese behaviour was neither economic nor rational in the sense of a landed proprietor or a cultivator seeking to increase productivity or actual production. Production was for the purpose of self-sustenance and for the compulsory payment of annual taxes. In such a society, labour was directly linked to survival; if a cultivator did not work, he and his family starved or would face severe punishment. Since a cultivator lacked personal motivation, it was not unusual for him to try to get by without working if possible. This situation was exacerbated by the structure of the economy. There was a wide range of types of production units in medieval rural Japan. There were both relatively large engagement units based on the use of subordinate labour, as well as small stem family units. Economically speaking there was no uniform production function. Political rulers or lords also lacked economic incentives. Due to the complex system of rights of tenure and the fact that the estates were geographically scattered, efforts to increase productivity proved difficult even when attempted. Instead, rulers tried to increase the number of estates under their control or expand their rights of tenure, in order to increase the amount of tax they could collect. This situation began to change in the sixteenth century due to the commercial- isation and urbanisation of the central Kinai area and the appearance elsewhere of territorial lords known as sengoku daimyo¯. As markets developed with the circulation of money in the Kinai area, cultivators began to produce for the market. Productivity subsequently increased with the expectation of profit, since cultivators realized that they could improve their level of material life through added income. Of course not all cultivators proved skilled in the ways of the market; there were also failures. And since no one had sufficient political authority to guarantee property rights, a kind of anarchy existed in the Kinai area before the late sixteenth century. Still the Kinai was the centre of a bustling trade in goods and specie; merchants developed a network for trade in daily necessities and not just luxury goods as before. These changes gradually spread out from the Kinai to neighbouring provinces, but in regions farther from the central Kinai area, conditions were different. In those regions we find the rise of powerful territorial lords (sengoku daimyo¯) who controlled their domains by brutal force. While their domains were not as

12:27:16:11:09 Page 45 Page 46

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

economically developed as the Kinai area, these lords constructed large-scale irrigation projects and invested in the economic infrastructure of their domains5 largely by means of a corvee labour extracted from resident cultivators. As a result, these lords contributed to an increase in actual production and a rise in productivity. The sengoku daimyo¯, however, did not attempt to implement structural changes such as the separation of warrior and cultivator. Since they relied on existing agricultural units, we do not find the kind of remarkable trade development which occurred in the Kinai area. We can divide mid-sixteenth century Japan, therefore, into two different types of areas. The first was the Kinai which, although economically developed, suffered from a kind of political anarchy. The others were the more remote areas where, apart from an expansion of economic output, there was no structural change. In these latter areas we find the rise of the new territorial lords, the sengoku daimyo¯, who acquired exclusive political authority over their domains. In the late sixteenth century, however, we find a third type of area proved of critical importance. In this third area the lords could combine the efficient political authority of the sengoku daimyo¯ and the economic development of the Kinai area. This development was the result of three unifiers, who rose to power in the provinces of Mino, Owari, and Mikawa located in the eastern part of the Kinai area. The fact that they unified Japan at this particular time was not a historical accident. What did they accomplish? If we were to chronicle their many achievements, the effort would require many volumes. Thus we will take up their story only in connection with our subject, Japan’s ‘Great Transformation’.

DECISIVE POLICIES It was under Oda Nobunaga that the policy of separating warrior and cultivator was implemented. He could do this because growing productivity enabled him to remove his vassal-farmers from the villages and then assemble them in a single place where they could be trained as professional troops. This policy was followed by Toyotomi Hideyoshi, who succeeded Nobunaga after the latter’s assassination in 1582. Hideyoshi further strengthened this policy through a sword hunt katana- gari with the aim of disarming the peasantry. By confiscating their weapons, Hideyoshi turned the peasantry into a ‘professional’ class of land tillers. A sharp social division resulted when the warrior class was removed from the land; they relinquished their role as cultivators and took on the responsibility of military affairs and government. Their removal from the land meant that the warrior class could be transferred from their ‘domain’, and they lost their connection with the land. We find that through this process retainers lost their fiefs and became bureaucrats receiving stipends.6 Begun in Hideyoshi’s time, the practice of arbitrarily transferring daimyo¯ fiefs, became especially prominent under the third unifier, Tokugawa Ieyasu. Further- more, under Ieyasu and the Tokugawa government, the separation of the warrior class and the peasantry was instituted throughout Japan, with the important exception of Satsuma. Such political reorganisation, which began with the military action of a single daimyo¯, Nobunaga, was extended throughout Japan within a few decades (i.e. from the end of the sixteenth to the beginning of the

12:27:16:11:09 Page 46 Page 47

A GREAT TRANSFORMATION

seventeenth centuries). The systems of class division and daimyo¯ and retainer transfer, which characterised Tokugawa Japan, produced a number of very important social and economic changes and, at the same time, created a society that cannot be labelled ‘feudal’ in the European sense. An astounding spree of urbanisation resulted when the warrior class was removed from the countryside and stationed in a particular castle together with craftsmen and merchants. Thus, around two hundred newly-built castle towns mushroomed throughout the nation with surprising speed. Castle towns played a tremendously important economic role emanating from both the tax and social systems of Tokugawa Japan. Resident warriors received a fixed stipend in rice and money that was collected as annual tax from the peasantry. A warrior used part of the rice to feed himself and his household, but the remainder had to be sold to merchants for cash. With this revenue, warriors purchased essential goods, other than rice, from merchants. Merchants and craftsmen, of course, paid for their necessities of life with the cash which they earned from their occupational activities. This meant that a large consuming population resided in castle towns who possessed purchasing power in terms of money. By labelling this power ‘effective demand’, Tokugawa Japan instituted a type of Keynesian policy three hundred years before its real time. These developments exerted a decisive influence on society as a whole. People, goods, and money began to flow between urban and rural areas throughout the nation. What began as a tranquil eddy, soon became a swirling whirlpool that enveloped the whole country. The construction of castle towns required the outlay of enormous resources and several decades to complete them, but the accompanying changes which slowly developed in the Kinai area spread quickly to the rest of the nation. The impact of the market economy emanated outward from the central castle town in a concentric circle. Each domain covered a small area, and the market economy covered all of Japan by the end of the seventeenth century. One of the most important changes in this context was the transition to a homogeneous rural society based on the stem family as the primary unit of agriculture.7 Finally, the practice of alternate attendance (sankin-ko¯tai), already pioneered by Hideyoshi, was firmly established in the 1630s. Under this system, a daimyo¯ left his family members in permanent residence as hostages in the sho¯gun’s capital, Edo. The daimyo¯ himself, accompanied by his retinue of vassals, was obliged to make a bi-annual trip from his domain to Edo to pay his respects to the sho¯gun. The system also had a great influence on Japanese society. First, Edo was not simply the castle town of the bakufu; it was also a metropolitan capital in every sense of the word. At the beginning of the eighteenth century, it was the world’s largest city with an estimated population of a million residents, evenly divided between warriors and commoners. Regardless of living standards, a population of one million meant that Edo had an enormous consuming demand.8 All the essentials of urban life such as food, fuel, clothing, building materials and other necessities had to be transported into the city. However, the Kanto¯ area, where Edo was located, proved incapable of supplying all necessary goods for the city in the early Tokugawa period. Since neither a trade nor information network had yet been established in the region, Edo was forced to rely on the shipment of goods

12:27:16:11:09 Page 47 Page 48

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

from the developed area of Kinai. O¯¯ saka was the most logical geographic terminal for the stockpiling and shipment of goods to Edo, not only from the Kinai but also from the west and from the Japan Sea area of Japan. O¯¯ saka was known as the ‘entrepôt of the realm’ (tenka no daidokoro), flourishing with a population of half a million in its peak years. As a result of the relationship between Edo, as consumer, and O¯¯ saka, as supplier, a national economy developed in Japan. The prices of staple commodities particularly of rice, began to fluctuate in parallel elsewhere in Japan, at least in normal years. In addition, the symbiotic relationship between the two cities produced a very sophisticated financial system based on trade, credit and finance.

THE UNEXPECTED RESULTS OF POLICIES The policies discussed above were undertaken with strictly political and military objectives in mind; the subsequent economic results were totally unexpected. Although unintentional, no conscious policy could have proved as effective. Also at the same time, peasants began to produce goods for sale in addition to the former requirements of self-sustenance and taxes. Their behaviour became increasingly economically-oriented. Interestingly, the political rulers considered such economic activity by the peasant class to be ideologically repugnant. For them, the portrait of the ideal peasant, as depicted in the edicts of the period, was one who produced only enough to pay taxes and survive; the ‘obedient’ peasant was not supposed to produce for profit. But as a result of the policies adopted by the bakufu, the peasant had no choice but to produce for and participate in the market system. Thus the portrait of the peasant as depicted in bakufu laws, for the most part, differed completely from reality.9 Peasant participation in the market also had political implications. Neither the bakufu nor the daimyo¯ possessed means of turning back the clock to realize the idealised portrait of the peasant, after these economic forces were set in motion. The everyday administration of the village was left to the residents themselves; the presence of the warrior class at the village level was limited to a surprisingly few sub-administrative offices which performed the routine work of territorial administration. As has often been said, Tokugawa society was largely divided into four classes consisting of warriors, peasants, craftsmen and merchants. Other occupations such as court nobles, members of religious orders, fishermen and miners were also present, but they were small in number compared with the four main classes. In order to simplify our discussion of economic consequences we will omit them. Originally, when the warrior class sold its stipend rice to the merchants, the price was determined by the bakufu or daimyo¯ themselves. To borrow the words of Hicks again, this was a type of ‘command economy’. Gradually, the initiative in determining the price of rice passed from the warrior class to the merchants, since the increasing expenditures of the samurai exposed them to a costly urban life. They were forced to receive money in advance from the merchants who were becoming powerful. The amount of trading rice which came directly to the market from farming villages gradually increased. As the advantage passed

12:27:16:11:09 Page 48 Page 49

A GREAT TRANSFORMATION

to the merchant class, the purchase of rice as a commodity definitely became a buyer’s market. The situation was also made more complex by the tension between political and economic factors. While the collection of the tax rice by the warrior class from the peasantry was based on political differences between the two classes, the conversion of tax rice into money by the warriors, so that they could buy goods, represented economic activity, as did the activity of the peasant and merchant classes. The exchange of goods and money between classes as well as within a class indicated increasingly commercial behaviour in a society where commercial behaviour was ostensibly denied. The exception to this tendency occured when the warrior class used its political power to default on loans and to force compulsory donations from wealthy merchants and peasants. Another question is, how did the warrior class disburse the collected taxes. If the Tokugawa warrior class had used all the land tax as personal income just as medieval estate proprietors had done, then the Tokugawa land tax would have constituted a private rent and would and would not have differed from that of the medieval period. And if the Tokugawa land tax had been used solely for such public purposes as administration, defence, education and social security, then it would have represented the tax characteristics of a modern state. In reality, the Tokugawa system blended both private rent and public tax. These dual characteristics were the result of taiko¯ kenchi, or the cadastral surveys conducted by Toyotomi Hideyoshi between 1582 and 1598,10 which were taken over by the Tokugawa government. At issue here is the nature of the land tax rather than the amount of tax paid by the peasant class. Since about one third of the tax was paid in money, a large amount of cash and rice flowed from the peasantry to the warrior class, or from rural to urban society, a process that necessarily involved the merchant class. The extent of this inter-class trade was such that everyone in society was affected by the burgeoning market economy. It also meant that regardless of one’s social class, an individual possessed the chance to improve his economic status. In contrast, however, the warrior class found itself in increasingly straitened economic circumstances, despite its political and institutional superiority. No warrior in Tokugawa Japan acquired enormous wealth, this is one of the most significant aspects of the period. In Tokugawa Japan, political power did not translate into economic wealth for the ruling warrior class. This situation resulted in a polarisation between political and economic power. Since this polarisation also followed class lines, it meant that Tokugawa society was divided into a small, impoverished elite represented by the warrior class and the rest of the society. Although political authority was the sole possession of the warrior class, there was considerable intra-class economic mobility – both up and down. And there was even inter-class mobility, if not so much. Thus Tokugawa Japan had many ‘intermediates’, who were impoverished warriors, or wealthy peasants, craftsmen and merchants. Of course not all of the last three were wealthy, but the opportunity of becoming well-to-do did exist. How and why such a complex and contradictory system developed in Japan remains a puzzling problem for historians. At present there is no definitive answer to these problems, but one can advance a few points which may prove helpful in illuminating them. First, political rule

12:27:16:11:09 Page 49 Page 50

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

was not based strictly on the possession of specific land. Under the kokudaka system, a daimyo¯’s ‘domain’ was temporary. The political rule of the bakufu and daimyo¯ was limited to the taxation of the rural community; they did not hold actual demesne entailing direct responsibility over the management of agricultural or industrial production. Villages were self-regulating units, left alone as long as peace and order were maintained. The bakufu and the daimyo¯ taxed a certain amount of rice and money based on figures produced by cadastral surveys, not from the peasants personally but from the community. Eventually, there was no legally prescribed status of ‘unfree men’ or serfs as in medieval Europe. Although this did not mean that the peasants of Tokugawa Japan were not bound by restrictions in their daily lives, they had no levy labour service for tillage. They could marry at will without paying the jus primae noctis; they were free to move if they reported this to the authorities; and inheritance was free from added imposts. Taxation was more flexible in the urban areas. The residents did not pay land tax or income tax to the political authorities. Although rich merchants frequently were forced to donate money to the impoverished warrior class, they could obtain certain privileges in return. Economic development led to the loss of political control over the economy by the warrior class.

CONCLUSION Due to this ‘great transformation’, Japan became a nation characterised by economic-mindedness, pluralistic values and effective government. It was not a modern state, but certain necessary pre-conditions were already present. Even the so-called sakoku (literally, ‘closed country’) policy could be carried out by the government because it had the power and authority needed to enforce this measure. But in economic history, the most important point was the emergence of an economically-minded society in which market mechanisms appeared and economic values became independent of other values. Through this transformation, the Japanese, particularly the common people, realized that their material life could be improved through economic activities. Industriousness became a virtue – the most essential secularised value. It was handed down to posterity through the ie, or family system. In terms of modernisation, Japan’s experience was unique, if not paradoxical, compared with the West. Certainly, the course of industrialisation differed in Japan and elsewhere. And, as frequently noted, the contemporary Japanese economy has its own peculiarities – Japan Inc., lifetime employment, the seniority system and loyalty to the company, etc. But no one holds that the economy of Japan has not developed through the market economy or capitalism. Although we should not ignore the differences in economic development between Japan and the West, it is more important to search for the historical origins and conditions which allowed Japan to catch up with the West economically, despite a different cultural setting. From the facts discussed above, the reader may easily recognise that the pre-conditions for modern Japan’s economic success lay in her ‘great transformation’.

12:27:16:11:09 Page 50 Page 51

A GREAT TRANSFORMATION

NOTES

1. John Hicks, A Theory of Economic History, Oxford, 1969 2. On the economy of O¯¯ saka and Kinai, William B. Hauser, Economic Institutional Change in Tokugawa Japan. Cambridge, 1974 3. Hayami Akira, Nihon keizaishi e no shikaku. To¯kyo¯, 1968. p. 39 4. I borrow this term from Karl Polanyi, The Great Transformation. New York. 1957. Polanyi marks the end of ‘transformation’ at the passing of the laissez-faire economy. In eﬀect, then, my usage is totally the reverse of his. 5. The most famous case was the construction of ﬂood prevention works by the daimyo¯ Takeda in Kai Province. 6. A detailed analysis can be found in Yamamura Kozo A Study of Samurai Income and Entrepreneur- ship. Cambridge, 1974 7. Thomas C. Smith, The Agrarian Origins of Modern Japan. Stanford. 1959, chap. 4. Also see Hayami Akira and Uchida Nobuki, ‘Size of household in a Japanese county throughout the Tokugawa era’, in Peter Laslett (ed.), Household and Family in Past Time. Cambridge, 1972. pp. 473–515 8. Gilbert Rozman, ‘Edo’s Importance in the Changing Tokugawa Society’, The Journal of Japanese Studies. Vol. 1, No. 1. 1974. pp. 91–112 9. For the economic activities of peasants and merchants in Tokugawa Japan, Susan B. Hanley and Yamamura Kozo, Economic and Demographic Change in Preindustrial Japan, 1600–1868. Princeton, 1977 10. On the cadastral surveys, Hayami Akira, ‘Epanouissement du nouveau régime seigneurial aux 16e et 17e siècles’. Keio Economic Studies. Vol. 1. 1963. pp. 21–51

12:27:16:11:09 Page 51 Page 52

First published in On the Eighteenth Century as a Category of Asian History: Van Leur in Retrospect (eds. L. Blussé and F. Gaastra), pp. 131–146, Hampshire, 1998.

3 Japan in the Eighteenth Century: Demography and Economy

INTRODUCTION fter a ‘great transformation’ in the seventeenth century,1 and before the busy A ‘transition’ of the nineteenth century,2 eighteenth-century Japanese society has been seen as relatively quiet, static and sleepy in every component. There were almost no international and domestic events that shook Japan nationwide and although the isolationist policy of Japan was enforced in the 1630s, this did not mean closing the door entirely to other countries. Japan even enjoyed its own world order in its relations with the kingdoms of Korea and Ryukyu, and saw merchants travel to her shores from China and the Netherlands.3 This world order was very small compared to that of the Chinese, but it served to keep Japan independent, particularly by restricting relations to a minimum with its giant neighbour, China. Under this international security policy in the eighteenth century, the Tokugawa government was able to concentrate its energy on establishing a firm administration. The essential characteristic of this era was its observation of a status quo policy, as the government had succeeded in keeping Japan peaceful without any major disturbances. However, during this period several developments occurred which affected Japan’s industrialization and established the conditions for modern economic growth.4 In this paper, the author wishes to examine this ‘quiet’ eighteenth century in Japan. Like almost all other pre-modern societies in this period, Japan has not left us with nationwide statistics, either quantitative or qualitative; except in the area of population figures. In that respect Tokugawa Japan is really one of the most well-documented pre-modern societies in the world. In addition to this, plenty of data was produced by the village more realistic population figure. At any rate, the remarkable stability in the total number has caused people to misinterpret Tokugawa Japanese society as a stagnated one, both in terms of demographic and economic growth. However, when we take a closer look at the population trends by province (Figure 3.1), we immediately realize that there were striking differences between the different regions. Roughly speaking, the population declined in the northeastern provinces, increased in the south-western provinces, and stagnated in the central provinces. The most serious decline was in the three provinces of northern Kanto, where the population fell from 1.84 million to 1.39 million, that is −25 per cent, followed by eastern-Ou, Mutsu Province, which fell from 1.96 million to

12:27:16:11:09 Page 52 Page 53

JAPAN IN THE EIGHTEENTH CENTURY

Figure 3.1 Population change by province between 1721 and 1846.

1.60 million, that is −18 per cent. In contrast, the San’in region, which consisted of the seven provinces of western-most Honshu, witnessed a population increase from 0.98 million to 1.16 million, that is +18 per cent, while in southern-Kyushu (four provinces), the population increased from 1.09 million to 1.25 million, that is +15 per cent. There was an important population decline in the most urbanized areas, the ﬁve provinces of southern Kanto and Kinai. Edo was located in the former with a population of one million, and Kyoto and Osaka were located in the latter, each with a population of 0.4–0.5 million. The proportion of urban population might be 20–30 per cent in these regions. In southern Kanto, however, population declined 12 per cent and in Kinai 10 per cent. These two regions were regarded as the most economically advanced, growing commercial crops such as cotton, rape-seed and indigo which were distributed through commercial networks; these are referred to later. Yet during the eighteenth century the population in these two key regions did not increase, but actually decreased. The same applied to the population of the surrounding areas. As already noted, the population decline in the northern Kanto area was the steepest, while the population in the neighbouring Kina region declined by 8 per cent. Regions where the population increased during the eighteenth century were the eastern part of the central area and the west of Japan. Both areas had very low urbanized proportions, probably even less than 10 per cent. Because the population trend in eighteenth-century Japan showed very explicit local diﬀerences, we

12:27:16:11:09 Page 53 Page 54

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

may conclude that there was no stagnation to be witnessed on a nationwide level. We should therefore examine what the reason for this population trend was in each region. During 1753–1756 and 1782–1786 the northern part of Japan was hit by a series of bad harvests. Both periods witnessed low summer temperatures which resulted in meagre rice yields. Because of the extraordinarily high population growth in the preceding century, even remote and marginal lands came to be used for rice cropping, and as a matter of course bad weather dealt its first blows to these areas. Even in the flat areas starvation was reported, leading to a sharp drop in the population in this region. If we look in more detail at this drop, using more reliable data from the population surveys conducted by local lords, we see however that the population decline occurred almost throughout the eighteenth century, with two sharp drops in the late 1750s and mid-1780s.5 In the Aizu domain, from 1718 onwards, the population also began to decline from 169,000 to 116,000 in 1787. In the Yonezawa domain it is not clear when the decline started but the population in 1692 was 133,000, and 99,000 in 1760 and 1787–1790. Both domains thus experienced a decline of over 30 per cent. These two domains were located in Dewa Province, where the regional decline was much milder than that of their eastern neighbour, Mutsu. We can assume that the decline in Mutsu Province during the famine years may have been more serious, possibly over 40 per cent. Such declines, due to environmental crises, affected the demographic trend in northern Kanto as well. This region was exposed to cold and wet winds from the Okhotsk high-pressure area in summer and the agricultural production suffered a telling blow just as the northern region did. The temperature did not rise enough to make rice growing possible and the misty weather resulted in a lack of sunshine. These phenomena were even more severe on the Pacific side of the island. In addition to this, Mt Asama, located near this region, erupted in 1783 and the Kanto Plain was covered with volcanic ash. The agricultural production dropped drastically and the combination of a lack of food stocks, the absence of effective relief by the administration, and poor transportation, exposed northern and north-eastern Japan to the Great Tenmei Famine, the severest famine ever recorded. The Kinki region and its vicinity was, in economic terms, certainly the most advanced area, followed by the southern Kanto region. In the eighteenth century, the residents of the central cities of these regions, Kyoto, Osaka, and Edo enjoyed a flourishing culture. There was no indication of an impending decline. However, the regional population of these urban conglomerates did not grow: it actually decreased. This is an example of the ‘urban graveyard’ effect every highly urbanized area experienced in the pre-industrial period.6 This affected not only the urban population itself but also the population trend of surrounding areas, from which a huge migration into the central city took place. Consequently, the northern Kanto region suffered both from famines and the ‘urban graveyard’ effect. In contrast, the population in Southwest Japan steadily increased. The annual growth rate was 1.5 per cent, lower than modern standards, but nevertheless the difference with other regions was significant. If any population grows at this rate, it will increase 20 per cent in 100 years, 45 per cent in 200 years. Using

12:27:16:11:09 Page 54 Page 55

JAPAN IN THE EIGHTEENTH CENTURY

pre-industrial standards, this growth rate can be said to be remarkable, particularly considering the length of time over which it continued. Why did the population grow in this area? There are several possible answers to this question. First, as Japan became a peasant stem family society, almost all peasants married and established families. The proportion of those married reached very high levels which, even when mortality was high, resulted in population growth. From my own study, using micro-data in central Japan, reproduction rates were positive for a female who married before the age of twenty-four. In the south-western region, the age at marriage for females was rather high, yet the average age was still younger than twenty-four, possibly around twenty-two to twenty-three years old. This then produced an excess number of children, who could be hired in the manufacturing or commercial sectors. Second, the natural environment in Southwest Japan was much more favourable for economic activities compared to that of Northeast Japan. The temperature in winter was above zero, and there was rarely any snow. The long navigable sea coast enabled easy access to the central market, Osaka. In particular, the Inland Sea (Setonai-kai), with its many ports, played a definitive role in sea-transportation, just as the Mediterranean did in Europe. Thus in south-western Japan, there was sufficient employment for peasant children, and we can be confident that there was no need for population limitation in this area. Third, the population distribution was favourable to this region’s growth. At the terminal of the above-mentioned sea routes was Osaka, the urban centre of the Kinai region, which had a population of half a million, together with the city of Kyoto. Eventually, the Kinai region was highly urbanized; it is likely that over 30 per cent of the population dwelt in the urban area. This means the demand for consumer commodities was very high: thus the south-western region became its supplier, as there was no large city within that region. Within the south-western region the proportion of urban population was, in contrast surprisingly low, probably under 10 per cent, in spite of its lively economic activity. The ‘graveyard effect’ did not function in this area. Thus demographic trends took different directions in eighteenth-century Japan. The appearance of ‘stagnation’ in the national total should not be seen as evidence of omnipresent stagnation in Japan. Certainly living conditions in the north-eastern part of Japan worsened due to a hostile natural environment which, in turn, influenced the population decline through the widespread practice of abortion and infanticide. Nevertheless urban life remained colourful and the people living in the western part of Japan continued to enjoy an improving standard of living.

GEOGRAPHICAL AND INTER-CLASS MOBILITY AND THE LABOUR MARKET With the help of micro-data on Tokugawa Japan, we should be able to follow the behaviour of families, married couples and even individuals. The number of cases on which such studies can be carried out are limited, but in the west of Mino Province, located in the centre of Japan, we have discovered several good data sets for this study.7 On the basis of these data, we found that almost half of the

12:27:16:11:09 Page 55 Page 56

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

children who survived beyond the age of ten, in both sexes, left their native village for another place at least once, mainly travelling to urban areas to work. They generally started to work away from home at the age of thirteen or fourteen. Then, after thirteen to fourteen years, about half of the migrants returned home to marry, and the other half remained in the urban areas, staying there until their death. This kind of temporary migration played a dual role in the population limitation. Firstly, it directly sent excess children to urban areas, and secondly, when some of these returned to the village they were already over the age of twenty-five, hence their age of marriage was late. For example, a female who experienced this kind of migration would marry five years later than those who did not migrate. On average this difference led to a decrease of two in the number of children born – causing a critical difference in the maintenance of the family succession. The largest suppliers of this type of migration were the lowest class of peasantry, and because it often happened that such families could not find a successor, these became extinct. On the other hand, among the upper-class families in rural areas, relatively few children went to work outside the village, so they tended to get married earlier and produced more children. The ‘excess children’ established, in turn, branch families with a little or no inheritance, which in effect meant that they created new families in the lower classes. Since the extinction of households in the lower classes and the establishment of new branch households offset each other, the number of households and the size of the population remained roughly the same. This mechanism could also be seen in central areas where a high demand for labour existed in the urban areas within accessible distance of the countryside. The Edo-centred area and the Kyoto-Osaka-centred area made up circles with a diameter of about two hundred kilometres around each city. In eighteenth-century Japan labour markets were established in limited geographical areas, and many young men and women commuted between rural and urban areas to obtain jobs. There was no political or administrative barrier to this migration, only distance or natural obstacles blocked the formation of a nationwide labour market.

THE ECONOMY As shown graphically in Figure 3.2, goods, money and services were exchanged between three social groups; the ruling warrior class, farmers, and merchants/ artisans, circled respectively. Within each group, of course, there were different social strata, for instance in the warrior class, territorial lords (daimyo) and vassals, or in the farming group, landowners and tenants. As the figure shows, however, the initial flow was the land tax which was levied on the farmers by the warrior class in money and in kind. This transfer was not a modern payment of tax, but a pre-modern reciprocal arrangement based on the different social status of both sides. The ruling warrior class collected tax by virtue of its social status as ‘overlord’. The transfer was carried out in money and rice. The total amount of this operation is not precisely clear, but we may assume it to have been equivalent to approximately ten million ryo in gold coins (one ryo contained 8.6 grams of gold

12:27:16:11:09 Page 56 Page 57

JAPAN IN THE EIGHTEENTH CENTURY

Figure 3.2 Tokugawa Economy

in the eighteenth century) and ten million koku (one koku equalled about 180 litres). The fact is this transfer was the largest one both by money and goods, it was the tribute, not done through the market. How far the other dimension of economy characterized as ‘market economy’, insofar this kind of transfer exists, Tokugawa Japan cannot be labelled a modern society. Yet it should not be called a ‘feudal society’ either. This is because Japanese society was not ‘feudal’ in the sense European history describes feudalism. The lordship was rather weak; almost all the warrior class dwelt in castle towns and their ties with the land and its rural occupants were not very solid. In the Japanese system there existed no such thing as a demesne in which villein-labour was requisitioned. In Japan the rate of taxation even dropped as low as 25 per cent for the total income of the peasantry.8 The lordship of Tokugawa society thus received money and rice by ﬁxed amounts or rates depending on the assessment of the land. This land tax could be transformed without much diﬃculty into public loans after the Meiji Restoration (1868), and as a consequence there was no brutal abolishment of the lordship during the transition period, similar to that experienced by European countries. At any rate, once the warrior class received the land tax from the agricultural sector, they used part for private consumption but also spent a considerable amount for administrative purposes. They had to maintain Bushi law and order in their territory and make reasonable investments towards such social overheads as the construction and repair of roads, bridges and irrigation systems. At the same

12:27:16:11:09 Page 57 Page 58

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

time, they had to provide for the stability and protection of the territory by way of police, justice and education. They were by no means simple exploiters, but administrators who were responsible for the effective rule of the territory. In this sense, the land tax of Tokugawa Japan had a dual function, that is as a tribute, and as a tax as found in modern society. Compared with the land-tax in sixteenth- century Japan, which was simply collected for direct consumption9 by the ruling class, that of Tokugawa Japan was very different. Except for the transfer of land taxes from the agricultural sector to the warrior class, all other transfers of goods, money and services between the different social groups, and within them, were made commercially. The most advanced trade could be witnessed on the Osaka (Dojima) rice market, where over two thirds of the transacted rice in Japan was handled.10 The exchange system developed in a very sophisticated way, the exchange even having a market in futures. Price information was transmitted in Japan as quickly as the available techniques permitted at the time. For example, smoke signs on hill- and mountaintops were used as transmitters, in much the same way as nowadays parabolic antennae are used. This illustrates how the local merchants could stay posted on the price fluctuations of the central Osaka rice market. A marketing system also developed. Local merchants, acting as middlemen, collected goods in their native area and sent them on towards Osaka. In Osaka, the goods reached the wholesaler and were then distributed to other middlemen who in turn sent these commodities to other local markets. This marketing system developed as early as the eighteenth century, and still continues today as a system of marketing which is very difficult to change. It was in the eighteenth century that the Osaka-centred marketing system functioned most effectively. The price of commodities remained stable between 1730 and 1820, with only a few exceptions due to bad harvest years. In the seventeenth century, until 1690, prices increased because of a shortage of supply against growing demand, and between the 1690s and 1730, there were sharp fluctuations in price based on recoinages of the standard currencies of gold and silver. The currency system was very complicated in Tokugawa Japan. The minting of gold coins and silver ingot was monopolized by the Tokugawa government, but copper coins were minted by the Tokugawa government as well as by powerful daimyo. By the end of the seventeenth century, the production of gold and silver in mines was diminishing, but nevertheless silver was still exported abroad, mainly to China through Nagasaki, Tsushima and Satsuma.11 On the other hand, the demand for silver coins was increasing until finally the Tokugawa government realized that there was no choice but to stop the drain of silver through foreign trade, and to carry out a debasement of silver ingots. Before 1695, the silver ingot which was used as currency contained 80 per cent pure silver. In 1695 it dropped to 64 per cent and in the first decade of the eighteenth century the debasement continued and reached as inconceivable a low as 20 per cent pure silver in 1711. Gold coins followed this pattern but not in such a drastic fashion. Such violent debasement was a product of ignorance on the part of the officials on the one hand, but at the same time, at least in the beginning, it was a necessary move for meeting the demand in society. However, once the debasement took

12:27:16:11:09 Page 58 Page 59

JAPAN IN THE EIGHTEENTH CENTURY

place, it jumped back with increasing prices; then the government suffered from a deficit in expenditure, resulting from the debasement. When we look at the fluctuations in the price of rice in the Osaka market during this confused period, in 1691 just before the debasement, rice of one koku was 41 to 53 momme in silver. But in 1716, after the last debasement, it soared to a price ranging from 130 to 162 momme, that is three times higher.12 The government realized it had to stop such inflation and, in 1716, new recoin- age in the reverse direction was carried out, with an 80 per cent silver content ingot, accompanied by a reduction of circulation and a reduced budget. Yet, as silver resources were short and the demand for money with growing economic activities was high, such drastic treatment was not supported by merchants to whom the finance of government was indebted. In 1736 the government began to mint new currency with lower quality again, this time with 46 per cent silver content, and the economic confusion was subdued. Through these events, we can see that Japan was and is a society where economic law functions; in this case, the over-supply of currency resulted in inflation. More important, this phenomenon was observed by scholars of the time; it stimulated their intellectual curiosity and led them to analyse it. Experimental economics was born in the eighteenth century. A ‘Gresham’s law’ was found by Miura Baien (1723–89), a scholar living in a town in a remote area, who completed a literary work, Kagen (origin of value), in which he referred to the relations between labour and wage, as Adam Smith did in The Wealth of Nations in the very same year.13 During the eighteenth century, the economic policy of the Tokugawa government fluctuated between a reduced and an expanded budget. The reduced budget, ideologically linked with conservative doctrine, denied the development of a market economy. The expanded budget, linked with enlightened thought, tried to develop economic activity as much as possible. For example, in the latter case, the Tokugawa government practised the expansion of foreign trade through monopolizing the sale of staple export goods to China, such as fishery products, by first building up and then subsidizing the producers. Through expanding exports the government was able to import silver which was badly needed to meet the growing demand. Ideologically, however, this policy was very dangerous to the traditional attitude towards foreign relations that Japan had established, and so was replaced by the conservative reduced budget within a couple of decades. But both policies had their limitations. As Japan became a sufficiently developed commercialized society, getting back to a ‘natural economy’ was impossible. On the other hand, unlimited economic expansion could not be actualized because it would shake the basis of the Tokugawa government. The economic policy fluctuated between these two limits in the eighteenth century. It should be noted, though, that in Tokugawa society there was no truly central government. The policy of the Tokugawa government was valid in its territory, but not in another lord’s. When the Tokugawa government ordered the reduced budget, any local lord could order the opposite. Therefore, in this sense, a national economy did not fully exist at this time. Even in currency, although gold and silver money minting was monopolized by the Tokugawa government, the exchange rate between gold, silver and copper

12:27:16:11:09 Page 59 Page 60

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

fluctuated every day. Adding to this, the local daimyo could issue paper money (hansatsu) which was valid only within their territory, if the Tokugawa government permitted it. In fact, during the eighteenth century, paper money was issued by all powerful daimyo. Its face value was always lower than that of gold and silver, but for ordinary people it was more convenient for daily use and its circulation became an everyday affair. Thus by the end of the eighteenth century, the Japanese became accustomed to using poor-quality gold and silver money as well as paper money. This was effective in tiding over the critical situation just after the Meiji Restoration when the new government had no stock of gold and silver and was forced to issue paper money. In Tokugawa Japan, in spite of its commercial development, about 85 per cent of the population lived in rural areas, and agriculture commanded the biggest production. Farming was carried out by family labour, sometimes with the help of an indentured servant. Cultivated land per household was very small, 2.5 acres. The peasant family had to work this small plot through manual labour: as all the land had to be used to provide for the increased population, the feeding of livestock was a very costly business and a peasant working with spade and hoe could turn up the soil more effectively than would a traditional plough driven by livestock. The traditional Japanese horses and cows were smaller than those in Europe and were incapable of drawing a plough which could dig deeper into the field. The author has come upon the fact that in Owari and Mino Provinces (central Japan, near the Nagoya area), in the last decades of the seventeenth century, the population was 343,000 with 18,000 livestock, but in the early nineteenth century this changed to 444,000 population with 8,000 livestock.14 This indicates that there must therefore have been a switch from horse-power to man-power in agriculture. Another remarkable feature of Tokugawa agriculture was the use of a great deal of fertilizers, the most striking being dried sardines and herring. The fishing for this, mainly done on the north-east coast, became big business and the fish were transported to Osaka for marketing to use in cotton plantations at first, and later for every type of cultivation. This kind of indirect use of fish for food production did not exist at all before the Tokugawa era, but in the eighteenth century, fishery, transportation, and marketing were established all over Japan. In contemporary literature, such fertilizers along with many kinds of strained lees that could be purchased, were called ‘gold fertilizer’ (kin-pi), and undergrowth and ash which could be obtained by the labour of the peasantry themselves was called ‘green fertilizer’ (ryoku-hi). Both fertilizers were used heavily in cultivation and made the soil very fertile. This resulted in land productivity being high on the one hand, but on the other it favoured the growth of weeds. The peasants were then faced with having to pluck the weeds from the cultivated land by hand. Putting these characteristics together, it appears that the Tokugawa peasantry had to work harder than before. We have no written records to compare the working hours, but we can assume that the peasants in Tokugawa Japan were working for long hours doing painful manual labour. During the eighteenth century, such type of work in rural Japan eventually created a work ethic of industriousness. Hard work was not rewardless drudgery but a virtue, resulting in a better life in this world. This development, which took place in eighteenth-

12:27:16:11:09 Page 60 Page 61

JAPAN IN THE EIGHTEENTH CENTURY

century Japan, can be called the ‘industrious revolution’, which has continued to the present day as a mentalité of the Japanese attitude to work. These ethics were transmitted from one generation to the next within the family rather than in religious institutions, as was the case with Protestantism in the West. In the taxation system of Tokugawa Japan, the surplus produced by hard work was not confiscated as an extra tax by the ruler. Merchants and landowners were the main recipients of the surplus, but the peasantry could avail of it too. The peasantry held hopes of living in a house with mats, dressing in silk, and eating delicious fish caught far away from their village. The living standard, housing, foodstuffs, clothing and life expectancy of the peasantry became steadily better, if not extraordinary. Hard work was the price of progress. Thus the Japanese came to realize that their work could bring them higher standards of living, when it led to higher income. They now behaved economically and efficiently. Many inventions and improvements were carried out; almost all of these had a labour-intensive character but contributed to increased production. In agriculture, many kinds of hoes were developed, and peasants could choose the best one for their soil. In rice cultivation, seed selection was introduced and careful planting and weed-plucking became common for peasants. In addition to this, the seed/yield ratio of rice reached more than 1:50, the highest in a rice-producing society at that time.15 Land productivity also reached the same heights. In eighteenth-century Japan, except for the irrigation system, there was little capital investment in agriculture and industry. Compared to Western development, Japanese development was highly labour intensive. The characteristics of technology were geared towards maximizing the production within a limited, small plot. Many books on agriculture were written by farmers themselves and circulated all over Japan. As literacy of the Japanese reached 50 per cent for males just before the Meiji Restoration, the diffusion of advanced knowledge by the written word was fast and efficient.

CONCLUSION Of course Tokugawa Japan was not a paradise. There were a lot of problems. Peasant uprisings were frequent, infant and child mortality was high, urban life was unhealthy, and there were too many in the warrior classes to feed by the land-tax alone. It is likely that half of the warrior class had no work, other than being idle. But still they ruled. When the standard of living was rising in every class of society, only the warrior class was left behind. This became obvious in the nineteenth century, but even in the eighteenth century it was already painfully clear. Several warriors lived their lives pursuing hobbies, without obtaining the necessary arms and weapons. Impoverishment led to stealing by warriors, a reaction to changing circumstances that was in striking contrast with that of the merchants and farmers they ruled. Thus in Tokugawa society, the social group who had political power lost their economic wealth, and one which had no political power gained economic wealth. This disruption was unique in Asian societies. As in European societies, Japan was not a despotic society before the modern period, but a decentralized society

12:27:16:11:09 Page 61 Page 62

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

in the religious and secular worlds, with decentralized political and economic values as well. Upon such a value system and under such conditions, Japan was preparing quietly for industrialization in the late nineteenth century after encountering the West. Firstly, demographically, Japan had already experienced a ‘population explosion’ in the seventeenth century. The people realized that an unlimited population increase might bring disastrous results and that population limitation, in some form, should be introduced. Some areas, faced with crises created by the natural environment, saw abortion as a positive check and infanticide was practised, and in other less affected areas the preventive check of late marriage was introduced. People could come and go between rural and urban societies. This meant that the life styles of urban and rural societies were not separate and information reached rural areas by word of mouth. Secondly, Japan became an ‘economic society’ with well-developed commerce and urban networks as early as the eighteenth century. It had no large-scale agricultural or manufacturing equipment at all, but had highly developed, small- scale, labour-intensive tools. In any society which was so advanced economically that it could be called an ‘economic society’ before facing industrialization, the process would inevitably lead to a market economy. On the other hand, if a society had not yet reached the level of ‘economic society’ when it faced industrialization, it might take on a kind of command economy. Socialism was one of the ways to industrialize, although it seems not to have been particularly successful. Thirdly, Japan in the eighteenth century was a multi-value society, that is a society in which values were divided. The people became secular, and mass culture flourished. Political power and economic wealth were shared by different social groups; no single group or class had despotic authority. In other words, Japanese society was ‘soft’ compared with other ‘solid’ societies. This flexibility would help Japan to adapt to Western civilization, developing geographically and qualitatively different pre-conditions. Finally, Japan experienced its ‘industrious revolution’ well before industrialization. Because this work ethic was transmitted through the family when society became secular, it had the ability to survive, even up to the present. What is most important in describing the character of pre-modern Japan is the fact that the people realized their work would be rewarded in their material life, and they maintained the desire to improve themselves. Private schools were established away from urban centres, in remote areas, without any assistance from the political authorities. This means the demand to read, write and count was extremely high among the ordinary people. Certainly Japan’s path to a modern society was different from that of the West, and perhaps it would be true that if Japan had not met with the industrialized West in the middle of the nineteenth century, she could not have industrialized by herself. Japan borrowed and imitated advanced Western technology and institutions: these were necessities for industrializing the country at that time. But it can be said too, that Japan was fully ready to receive them under the conditions mentioned above. If such conditions had not existed, the path of Japan’s industrialization may have been very different. It is certain that Japan’s

12:27:16:11:09 Page 62 Page 63

JAPAN IN THE EIGHTEENTH CENTURY

experiences that took place in the eighteenth century were the keys to her achievement of industrialization.

NOTES

1. Akira Hayami, ‘A Great Transformation: Social and Economic Change in Sixteenth and Seven- teenth Century Japan’, Bonner Zeitschrift für Japanologie 8 (1986) 3–13. 2. On the transition, see Marius Jansen and Gilbert Rozman eds, Japan in Transition: From Tokugawa to Meiji (Princeton 1986). 3. Ronald Toby, State and Diplomacy in Early Modern Japan: Asia in the Development of the Tokugawa Bakufu (Princeton 1983). 4. All ﬁgures given, are from Akira Hayami, ‘Meiji zenki jinko tokeishi nenpyo: hu, bakufu kunibetsu jinkohyo’ (A Chronological Table of the History of Statistics in Early Meiji Japan; with the Table of Provincial Populations Compiled by Tokugawa Bakufu), Nihon Kenkyu (Bulletin of the International Research Center for Japanese Studies) 5 (1994). 5. Akira Hayami, ‘Kinsei Ou Chiho jinko no shiteki kenkyu joron’ (Introductory Remarks to the Population History of the Ou Region in Early Modern Japan), Mita Gakkai Zasshi 75/3 (1982) 70–92. 6. W. Farr found that mortality in late nineteenth-century urban England depended on the population density. See E.A. Wrigley, Population and History (London 1969). 7. Akira Hayami, ‘Rural Migration and Fertility in Tokugawa Japan: The Village of Nishijo, 1773–1868’ in: Susan B. Hanley and Arthur P. Wolf eds, Family and Population in East Asian History (Stanford 1986) 110–132. 8. Shunsaku Nishikawa calculated this rate through a quantitative analysis of ‘Bocho fudo chushi- n’an’, a survey of the territory of Suwo and Nagato carried out by Choshu-han in 1840. See Shunsaku Nishikawa, ‘The Economy of Choshu on the Eve of Industrialization’, The Economic Studies Quarterly 38 (1987) 323–337. 9. The author owes the use of this term ‘direct consumption’ to B.H. Slicher van Bath, De Agrarische Geschiedenis van West-Europa (500–1850) (Utrecht and Antwerpen 1962). 10. For the Osaka market, see William B. Hauser, Economic and Institutional Change in Tokugawa Japan: Osaka and the Kinai Cotton Trade (London 1974). 11. Kazui Tashiro, ‘Exports of Gold and Silver during the Early Tokugawa Era, 1600–1750’ in: Eddy H.G. van Cauwenberghe ed., Money, Coins and Commerce: Essays in the Monetary History of Asia and Europe (Leuven 1991) 75–93. 12. For prices in Tokugawa Japan, see Mitsui Bunko ed., Kinsei koki ni okeru shuyo bukka no dotai (The Trends of the Price of Staple Commodities in Late Tokugawa Japan), (2nd edn.; Tokyo 1989). 13. Shunsaku Nishikawa, ‘Edojidai no porichikaru ekonomi, (Political Economy of the Edo Period), Nihon hyoransha (1979) 94–113. 14. Akira Hayami, Kinsei nobi chiho no jinko, keizai, shakai (Population, Economy and Society in Early Modern Japan: A Study of the Nobi Region), (Sobunsha 1992). 15. Mitsuo Oka, ‘Nihon nogyo gijutsushi: kinsei kara kindai he’ (A History of Agriculture in Japan: From Early Modern to Modern Period), Minerva shobo (1988) 24, ﬁgure 8.

12:27:16:11:09 Page 63 Page 64

First published in Journal of Japanese Trade & Industry, November/December 2001.

4 Industrial Revolution versus Industrious Revolution

INTRODUCTION lthough it has recently become somewhat questionable, the Japanese are Afrequently referred to as being hard working. Among those who agree with the view are some who contend that it is part of the “national character” of the Japanese. The author feels that one ought not to use the notion of “national character” in any situation. The French term “mentalité” is used to indicate a fundamental state of mind of a social group that continues over a long period. The English term “mentality” is a different concept. The term “mentalité” is difficult to translate into Japanese, but it is not the same as “national character.” The assiduity of the Japanese falls within the purview of the term “mentalité.” That is to say, it persists over a long period of time, but it is not the peculiarity of a particular ethnic group. Unless one takes this into account, there is an ever-present danger that the theory of the industriousness of the Japanese may occasionally lapse into a theory of ethnic superiority. However, no one can deny that for the past several hundred years the Japanese have been diligent. Indeed, foreign people who visited Japan during that period attested the fact in unison. Even in the present day, among the advanced industrial nations, Japan’s workers put in the most work hours per year, with a total of over 1,800 hours. If overtime were included, this figure would increase even further. Needless to say, long working hours do not automatically indicate industriousness. One has to consider the degree of concentration on labor during working hours, and one cannot over- look effectiveness, that is, productivity. However, it is also a fact that working long hours is not mandatory in Japan. What I am saying is that for one reason or another, the Japanese at one time became assiduous, and that at some point they will cease to be so. It may be that after more than 400 years, the period when the Japanese worked hard may be coming to an end. Setting aside considerations of the future for the time being, as an historian, I would like to investigate when and for what reasons the Japanese became so industrious.

INDUSTRIOUS REVOLUTION Recently in Western academic circles, the term “Industrious Revolution” is being used as a concept in the study of history. For example, Professor Jan de Vries, a

12:27:16:11:09 Page 64 Page 65

INDUSTRIAL REVOLUTION VERSUS INDUSTRIOUS REVOLUTION

U.S. economic historian, presented at his inaugural meeting as president of the Economic History Association a speech titled “The Industrial Revolution and the Industrious Revolution,” which was later published in the association’s The Journal of Economic History.1 Furthermore, Britain’s Windfall Films produced a television program called “The Day the World Took Off” for Channel Four Television in the United Kingdom, which was broadcast from May through July 2000 and received highly favorable reviews. The program investigated the historical roots of Britain’s Industrial Revolution. During the program, well- respected University of Cambridge cultural anthropologist Alan Macfarlane com- mented that in the rice paddy cultivation in agricultural villages in Asia, in which there is investment of the labor of large numbers of people, one can see the characteristics of the “industrious revolution.”2 Actually the author is the one who coined the phrase “industrious revolution.”3 In later conversations with my friends de Vries and Macfarlane I communicated my ideas on the subject. In view of the fact that they have used the term, they must have concluded that it was appropriate as a term of reference. However, when they employ the phrase “industrious revolution,” they do not always use it in the sense in which I use it. As is often the case, the term already seems to have a life of its own. First, in the case of de Vries, as in the title of the article “The Industrial Revolution and the Industrious Revolution,” the two concepts are not seen as contrasting with one another. He argues that first in a society in which commercial- ization (also referred to as a market economy) has advanced, machines and then the steam power that moves those machines is introduced into industry, and the “industrial revolution” takes hold. In other words, there is a continuity from “industrious revolution” to “industrial revolution” which is not part of my own original conception. Macfarlane, in a comment during the television program, uses the expression in regard to Asian agricultural villages – which judging from the context is the villages of Nepal which he has continued to study – to describe the situation where a large number of peasants are working in wet rice cultivation. Indeed, he then makes a valuable assertion and is certainly justified in pointing out that compared to wheat, rice has a high harvestable volume to planted volume ratio, and that given the same size plot of land, rice was able to support a greater number of people. Here in Japan Umesao Tadao, through Kawakatsu Heita, has responded posi- tively to this notion. Here the concept of “industrious revolution” is understood appropriately and it is inserted in the image of world history of both of these scholars.4 As can be assumed from the title of the present article, my own conception of these two revolutions is that they take different courses and contrast with one another. Let me explain my idea in detail.

FROM ANIMAL POWER TO MANPOWER: CONVERSION OF THE AGRICULTURAL PRODUCTION SYSTEM IN TOKUGAWA JAPAN (EDO PERIOD: 1603–1867) I have done research on population and agriculture in Owari (now the western part of Aichi Prefecture) and Mino (now the southern part of Gifu Prefecture)

12:27:16:11:09 Page 65 Page 66

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

provinces during the Edo period. Of extreme interest was the following change. The most powerful daimyo (territorial lord) of Owari and Mino provinces was Owari Han 5 who held the castle at Nagoya. This domain carried out a survey in the 1670s and 1820s of the population, and the number of households, oxen and horses of each village within its territories. The 1670s survey was carried out under the auspices of the han, while the 1820s survey was ordered by a vassal of the han. Although the 1820s survey was not official, and they are of different characters, they do allow us to compare statistical surveys such as population, and the number of residences, horses and oxen. The two surveys clarify the number of mura 6 (villages) in existence at the time. The survey of the 1670s – which is slightly incomplete because one gun is not included – shows that there were 725 in Owari and 209 in Mino, making a total of 935. The total population of these mura was 266,071 in Owari and 77,315 in Mino. One hundred and fifty years later these figures had increased to 332,258 and 112,298, respectively. Within all of the eight gun of Owari – excluding the missing gun – there was an increase in population. The same was true of the 18 gun of Mino. However, at the level of the mura, some showed a decrease of population. Of special interest is the fact that in many of the mura in the vicinity of the castletown of Nagoya (population of 100,000), we can observe a reduction in population. This is considered to have been the result of the “urban graveyard effect.”7 Such exceptions did exist, but when we look at the overall region, we see that the population increased in Owari by approximately 25% and in Mino by approximately 45%. Considering the length of the period of 150 years, this degree of population growth amounts to approximately 2 to 3 permillage per annum, which compared to the standards of the modern period is not particularly high. However, if it continues over a long period of time, even this relatively low level of growth leads to a large increase in population, and in actuality during the Tokugawa (Edo) period the Owari and Mino provinces were one area which did show higher growth. From these observations, the author obtained the concept of industrious revolution, which neither has in view an industrial revolution, nor merely indicates the large number of workers in Asian agriculture. The industrious revolution, more than anything else, is a method by which the peasants as a group of producers increased the amounts they produced in order to cope with the market economy. It was precisely this kind of motivation that was a necessary as well as sufficient condition for the industrious revolution. This is not simply a way of referring to the rice cultivation for which many workers were required. That is, what it refers to is the phenomenon realized by the fact that increasing the amount of production (Y) was not an increase in the factor of capital investment (K = capital = livestock), but rather it was an increase in the investment of another factor in production: labor (L). The phenomenon of the increase of Y was actualized despite a decline in the ratio of K/L. When there is an increase in the amount of capital (K), the eventual result is an industrial revolution. In other words, when there was an increase in the amount of labor (L) invested, it did not lead to an industrial revolution, although there might be a development of “soft techniques.” As a result, Japan during the Edo period was unable to achieve an industrial revolution on its own. Instead, as a result of its

12:27:16:11:09 Page 66 Page 67

INDUSTRIAL REVOLUTION VERSUS INDUSTRIOUS REVOLUTION

encounter with the nations of the West, who had already experienced an industrial revolution, Japan imported one. Hence, my use of the term industrious revolution is restricted to mean that “increasing the level of productivity (Y) became an opportunity.”

CHANGES IN THE NUMBER OF LIVESTOCK In addition to population, the two surveys by Owari Han investigated the livestock numbers of each mura. Traditionally the livestock used for agriculture were horses in eastern Japan, oxen from central into western Japan and horses in the far western region. Owari and Mino provinces were at the western edge of the eastern region where horses were employed, and if one crossed a mountain range one came to the province of Omi (now Shiga Prefecture), where oxen were employed. As a result, in the surveys of livestock, horses were far and away in the majority. It is not clear, however, exactly what type of horses were in use in this region during the periods investigated. Compared with the Western horses imported to Japan beginning in the Meiji period (1868–1912), the native horses which remain in the extreme western portions of Japan were much smaller and could not possibly exert one horsepower. Nonetheless, they were stronger and faster than oxen, and required more feed. A survey from the 1670s shows 12,337 horses and oxen for the 635 mura of Owari. The figure was 4,197 in an 1820s survey, a reduction to approximately one-third. The most conspicuous reduction was in the area around Nagoya and the flatlands along the coast, where the figure decreased to about 20%. The decrease in Mino was not as large as that in Owari, but even so, the number in the flatlands decreased to about 45%. The mountainous areas in Mino saw a reduction to only 78%. As can be seen, the decrease in livestock varied by topography, but given the fact that the population increased, the number of livestock per capita was dramatically reduced in all areas. In Owari in the 1670s, the number of livestock per capita was 0.05, and this shrank to 0.013 in the 1820s, a decrease to approximately one-fourth. In Mino in the 1670s survey the per capita figure in the flatlands was 0.065, which declined to 0.021 in the 1820s, amounting to about one-third. This major change in the K/L ratio resulted from the fact that the human population increased while the number of livestock declined. What does this reduction of the K/L ratio indicate? Needless to say, it shows that the work previously carried out by domesticated animals came to be carried out by humans. Manpower was substituted for animal power. Of course, the above-mentioned observations are phenomena limited to Owari and Mino provinces, and we do not know the situation in other areas. Especially in the areas where oxen were dominant there is the possibility that they did not decrease, because the efficiency of oxen was higher than that of horses.8 However, the decrease in the numbers of livestock apparent in Mino and Owari provinces cannot be seen as a phenomenon entirely unique to that region, so at the very least there seems very little doubt that the number of horses decreased in eastern Japan and in the far west region of Japan. Ordinarily, “modern development” in agriculture indicates the increase of labor

12:27:16:11:09 Page 67 Page 68

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

productivity by; 1) the investment of more capital (K) in production, 2) the ratio of labor (L), another factor in production, declines as a result, and 3) consequently the per capita productivity (Y/L) increases. If that is the case, then the phenomena observed in Tokugawa Japan is a change to the reverse. Within such a transformation, with the decrease of labor productivity, did the per capita income of peasants rise? Or did it decline; that is, did they become poorer? This is not an easy question to answer. However, at a minimum, it does not appear that the rural population’s standard of living was lowered. On the contrary, there seem to have been improvements in all aspects of their lives: clothing, food and housing. As a direct result of this, life expectancy at birth rose. In the case of the Suwa district of Shinano Province (now Nagano Prefecture) in the late 17th century, the life expectancy was 25, and in the early 19th century it had extended to 35 years. During the period prior to the introduction of modern medicine and public sanitation, such a lengthening of life expectancy can only have come from an improvement in the standard of living, and certainly could not have occurred under conditions of poverty. Let us now consider the benefits of domesticated livestock in agriculture. Espe- cially in the case of Japan, where paddy rice cultivation is dominant, domesticated animals were of greatest use during cultivation in spring. Crushing hardened soil as one plows up makes it possible to cultivate crops at a specific depth, increasing the vertical direction of cultivation within the planting acreage. The deeper cultivation is, the greater the increase in productivity. Prior to the early Edo period, plows for domestic animals were small, and deep plowing was virtually impossible. On the contrary, the use of man-power with plows and hoes made deep cultivation possible. The use of barnyard manure, made from such materials as livestock dung and weeds, when spread over soil increased the nutrients. In other words, it increases the amount of soil that is used in cultivation in a solid way. For this point, livestock were doubly beneficial in the work of cultivation and in the production of fertilizer. However, when water was introduced into the farmyards and sprouts were transplanted from rice-seedling beds in the process of cultivating rice, livestock became redundant. The work of cultivation in paddies filled with water included weeding and supplementary fertilizing, but these were mainly done by manpower and there was no room for the introduction of livestock. After the spring cultivation, livestock were used solely for transportation. Livestock were of considerable merit, while fertilizer was primarily made by taking material from the villagers’ communal land, woodlands and grass cutting places, and mixing grass with animal dung. In addition, following the harvest, these animals were necessary for the transportation of the crops. In the Edo period, however, materials including the dregs from making sake and fish (especially dried sardines and dried herrings) were substituted as fertilizer. Due to the population increase, woodlands were developed into areas for cultivation so that food and materials for human consumption were grown there, and consequently land for growing livestock fodder gradually shrank. Necessarily, it became more expensive to obtain feed for livestock. Furthermore, peasants began to purchase fertilizer from merchants. In a phrase of the period, fertilizer was referred to separately as “natural manure” (self-sufficient manure)

12:27:16:11:09 Page 68 Page 69

INDUSTRIAL REVOLUTION VERSUS INDUSTRIOUS REVOLUTION

and “money manure” (purchased manure). Dependence on purchased manure required a variety of investment, and being interdependent on the peasants’ market-oriented production, agriculture became an enterprise managed by a family. In an enterprise, the principle of maximum benefits for minimum expense comes into play. Livestock which offered minimal benefits and were costly to raise became relatively useless. This transformation is eloquently illustrated by the change in the K/L ratio of Owari and Mino provinces. Within these provinces, the area where the number of livestock decreased the least was the area where pottery was produced, because livestock were most beneficial in the transportation of raw materials and fuel. There was also a smaller decrease in the mountainous areas with lots of woodlands. In these areas the expenses for raising livestock were lower. However, in the flatlands, livestock disappeared almost entirely, and even where a small number remained, they decreased so greatly it is inconceivable that they were used in cultivation.

CHANGES IN CULTIVATION METHODS AND STRUCTURE OF HOUSEHOLDS As a result of this livestock decrease, it is probable that peasants had to work harder and for longer hours. We have no statistics regarding the daily working hours of the peasants or the number of days they worked per year. Nor have we found long-term journals or records of agricultural operations that related to the decline in livestock numbers. However, despite it being in a different region, we can trace the great change that took place in the structure of farming families. In the Suwa district of Shinano Province, we can trace through records the process by which the joint family households that were numerous at the end of the 17th century were dismantled into stem or nuclear family households.9 In these joint family households lived large numbers of unmarried servants. Of the children of the household head families, these were people who did not succeed to the household or did not go to other houses as adopted children or in marriage, and they spent their entire lives in that household. In all likelihood, their numbers were decided in line with the labor demands of the busiest agricultural season, and when work fell off, either there was no work for them or they simply performed miscellaneous chores. It need hardly be mentioned that their productivity was low, and when the primary factor of market productivity was introduced to this region, they gradually disappeared. The joint family household also dissolved into conjugal family units. Why did this dismantling occur? In the same way that the number of livestock declined in the Owari and Mino provinces, it resulted from the farming households choosing a more efficient means of management. In a society centered on rice cultivation like Japan, rather than choosing an operation that aims at large-scale areas of cultivation by employing a large quantity of labor, peasants achieved higher economic efficiency by pursuing small-scale cultivation operated by the family. During the busy agricultural season, there developed a custom of farming households exchanging labor. Regardless of whether the peasants

12:27:16:11:09 Page 69 Page 70

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

possessed their own land for cultivation or whether they did not, they pursued operations that relied upon family labor. There are no records, but there is no doubt that in the Suwa district of Shinano Province, there was a decrease in the number of livestock per capita accompanying the change in family structure. The raising of livestock and the collection and transporting of manure was work given to the servants of large families. Following the changes in agricultural methods, it was no longer necessary to maintain large numbers of servants. In terms of the necessity or lack of necessity of agricultural labor, there is a strong element of seasonality. Looking at the busy seasons of rice cultivation, spring is when soil is plowed and rice-seedlings are transplanted; summer is a time of weeding; and autumn is harvest time. After the harvest, there is an off-season. Consequently, the traditional form of manpower employed to match the busy agricultural season and that maintained throughout one’s life did not contribute to productivity during the off-season. On the contrary, they had to be supported in the off-season, so they became low-efficiency laborers. When the principles of a low-cost high-return market economy were introduced to the agricultural villages, this traditional form of farming household broke up and by necessity there was a transition to the form of the conjugal family, which became the central unit of agricultural households. However, this change did not take place overnight. Rather, it occurred over a long period of time, and from a modern perspective the change spread slowly. In the case of the Suwa district of Shinano province, if we observe the process with the indices of the average size of the households in each mura, we will find that this change developed in concentric circles spreading outward from the castle town of Suwa. The average household size did not go below 4.5 and when it reached this figure, the traditional method of farming completely disappeared, and one can set that point as the establishment of the Tokugawa agricultural method by which the family, mainly the conjugal family, is the provider of labor. Suwa gun had a diameter of approximately 25 km, and it took almost 120 years for the average household size to reach 4.5 throughout the gun. This was a pace of about 200 m per year. Whether this is considered slow or fast is difficult to determine because of a lack of comparison. Compared to the changes of our contemporary information- oriented society, it is of course slow. However, when one considers that it took place without the means of information orientation provided by modern technology and without the issuance of legal statutes, but only by word-of-mouth circulation among families or between villages, such a pace cannot be judged particularly slow. In addition, in terms of the peasants who lived at that time – especially those in charge of managing traditional agricultural methods – it is not difficult to imagine that there must have been hesitation and resistance to fundamental change in agricultural methods. It was no easy thing to dissolve the operation method of using the collateral families and lifetime servants, which had been passed down for generations, and make a transition to the system that left agricultural operations to each family’s management and charged land rents. It is probable that they saw a family close at hand which made this transition successfully and determined to dissolve the old pattern, and it might take several generations to complete the process.

12:27:16:11:09 Page 70 Page 71

INDUSTRIAL REVOLUTION VERSUS INDUSTRIOUS REVOLUTION

CONCLUSION In Japan, this industrious revolution had no religious background. In this sense, it was diﬀerent from the “Protestant ethic” of the West. In the West, industriousness was propagated through the churches, but in Japan it was primarily the families which passed along the spirit from parent to child to grandchild. The Meiji administration made industriousness a ﬁxed line of national policy, setting up statues of Ninomiya Kinjiro in the courtyards of each elementary school as a symbol of diligence. At present, in the West, accompanying the weakening of the church in society, the channel for propagating industriousness has grown feeble. In Japan, a pouring out of criticism following the defeat in the war of national government controlling education and the loosening of family bonds weakened the channel by which industriousness is promoted. In the period of high growth after the end of the war, a particular generation worked furiously to raise the level of the Japanese economy which had been severely lowered by wartime defeat, creating in the eyes of the world an image of the “industrious Japanese.” However, once the goal had been reached and criticism of excessive working hours grew, and as the trend toward a social welfare state came into view, the Japanese gradually came to be not so particularly devoted to work. The “industriousness” of Japanese has achieved its historic mission during the past 400 years and is now gradually leaving the spotlight. Now that families have fewer children and society is graying, and in a few years the population begins to decrease, it will be extremely interesting to see what choices will be made by the Japanese in the future.10

NOTES

1. Jan de Vries, “The Industrial Revolution and the Industrious Revolution,” in The Journal of Economic History, Vol. 54, No. 2, 1994, pp.249–270. 2. The subtitle of the program was “The Roots of the Industrial Revolution.” The series of eight 50-minute programs was broadcast biweekly beginning May 28, 2000 and it is a really high quality production. The text version of the program by Sally and David Dugan is The Day the World Took Oﬀ (Macmillan, 2000). 3. Hayami Akira, the introductory chapter “Keizai Shakai no Seiritsu to sono Tokushitsu – Edo-jidai Shakai Keizai-shi eno Shiten (The formation and characteristics of economics and society; the point of view of the history of Edo period society and economics),” in Atarashii Edo-jidai Zo wo Motomete – Sono Shakai Keizai-shi teki Sekkin (The Search for New Images of Edo Period History: A Socio-economic Historical Approach), Toyo Keizai Inc., 1977, pp.3–18. See also “Kinsei Nihon no Keizai Hatten to Industrious Revolution (Modern Japanese Economic Development and the Industrious Revolu- tion),” in Shinbo Hiroshi (eds.), Suryo Keizaishi Ronshu 2: Kindai Ikoki no Nihon Keizai (Essays on Quantitative Economic History Vol 2: The Japanese Economy during the Period of Modern Transformation), Nihon Keizai Shimbun Inc., 1979, pp.3–14. 4. Umesao Tadao (eds.), Bunmei no Seitaishikan wa Ima (Current Perspectives of an Ecological View of History), Chuo Koronsha, 2001. 5. Toward the end of the Tokugawa period, a daimyo (territorial lord), when referred to as an organization of administration, came to be described as a han. 6. For an explanation of the terms “mura” and “gun,” see the quotation below from Hayami Akira, The Historical Demography of Pre-modern Japan, University of Tokyo Press, 2001, pp. 179–180: “There were approximately 70,000 villages throughout Japan during the Edo Period, and some villages had several lords, . . . [but] the Japanese village was small and had much community character. . . . The local level above . . . mura is the gun, which might be translated as ‘county.’ However, during the period we are studying, the gun had no meaning as an administrative unit. The

12:27:16:11:09 Page 71 Page 72

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

term was used simply as an expression to indicate a geographical area; moreover, the ‘gun’ was entirely different from the English ‘county,’ ...” 7. In general, before the outcome of modern industrial technology was applied to urban living, the death rate was higher than the birth rate, and the natural population growth was negative in urban areas, especially in terms of the concentration of the populace in the major cities. To maintain and expand the population of urban centers, it was necessary to have a major influx of population from the neighboring agricultural villages, which resulted in a stagnation of population growth in the areas surrounding the cities. In Europe this is referred to as the “urban graveyard effect,” and this author refers to the Japanese case as the “Toshi Ari-jigoku Setsu (urban ant lion theory).” 8. Oxen were not used as a food source during the Edo period, but their hides and bones were important materials used in handicrafts. In addition, in certain areas from central to western Japan where cattle were bred as domestic animals, we can even observe the increase of their number. 9. Hayami, Historical Demography, chapter 4, pp.66–119. 10. If we restrict consideration to the population of productive age (15 to 64 years), when people have the greatest income and also the greatest needs, Japan reached a peak in 1995 of 87,260,000 in that age group (69.4% of the total population), and the figure has declined since then. It is this author’s opinion that this special feature of the population structure is one of the true causes of the present depression in current-day Japan, but no prominent figures or politicians have touched upon it.

12:27:16:11:09 Page 72 Page 73

PART II DEMOGRAPHY THROUGH THE TELESCOPE

12:27:16:11:09 Page 73 Page 74

12:27:16:11:09 Page 74 Page 75

First published in Keio Economic Studies, vol. 4, pp. 1–28, 1966–7.

5 The Population at the Beginning of the Tokugawa Period

An Introduction to the Historical Demography of Pre-industrial Japan

I okugawa Japan bequeathed to us an abundant supply of materials for Tpopulation study. Although these do not constitute vital statistics in the modern demographic sense, Japan in the Tokugawa period was a rare example of a well-documented pre-modern state. There are two reasons for this fortunate situation: (1) When Hideyoshi Toyotomi succeeded in the unification of the country in the last part of the 16th century, the Sho¯gun and the provincial feudal lords became very much interested in the quantitative aspect of their subjects and conducted surveys on the breadth of their domains and the size of their population, the first attempt of the sort since the eighth and ninth centuries when a census was carried out. Such being the situation, Japan is fairly well provided with demographic materials. (2) As a result of the strict ban put on Christianity by the Tokugawa government, the registration of individual’s religious faith was enforced. The people were ordered to report what religious faith they embraced, what religious body they belonged to, and to avow themselves to be Buddhists. The people in general, both urban and rural, were all ordered to enter their family temples in the registration book prepared on the basis of the town or village unit. It was about 1670 that this registration system was put in effect on a nationwide level. At first, the system had nothing to do with census-taking, but later it came to be utilized for that purpose too. Besides the names and the family temples of individuals, their sex, age and relation to their household heads, and such matters as the reason and the time for the change of their domiciles were entered. These registrations were not nationally uniform in form. The fact, however, that census-like registrations existed in many provinces, and especially, the fact that they were conducted annually, should be considered a significant phenomenon facilitating the demographic study of the Tokugawa period. Besides the material mentioned above, there exist such first-hand materials as the demographic information found in various research papers and the

12:27:16:11:09 Page 75 Page 76

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

genealogical records that were filed quite commonly in those days. It is also possible to obtain some materials concerning various social classes, as the literacy of the masses regardless of class distinction was rather high in Tokugawa Japan. Such an abundance of demographic research material is very encouraging, but at the same time we should be aware that it contains many pitfalls. In other words, it is important to pay close attention to the reliability or the exactness of material to be used as statistical data. Otherwise, we are likely to fall into erroneous conclusions. In spite of the importance of the population problem in Japan and the abundance of material for research, the historical study of it has been rather retarded, as historians were generally interested in other phases of the matter. It does not follow, however, that the historical study of Japanese population has been totally neglected. For example, Mr. Naotaro¯ Sekiyama and Mr. Bonsen Takahashi published scholarly books and special reports on the population in the Tokugawa period; the late Professor Kanetaro¯ Nomura1 brought out a short essay on the subject in English; and Mr. Ayanori Okazaki2 and Mrs. Irene B. Taeuber3 also touched on it in their works, the former in French and the latter in English. Seeing, however, that the study of historical demography has recently reached a higher level in the advanced countries of Europe, especially with reference to economic development, the Japanese scholars in the line have awakened to the importance of approaching the demography of the Tokugawa period from a new angle, from that of the latest professional techniques. Influenced by this trend, some scholars, reflecting a fresh perception, have come forward to publish their works. The author4 of this essay, cooperating with some experts in the field, is now engaged in the study of the demographic history of the Tokugawa period as viewed from the standpoint mentioned above, and has published the results in Japanese. Seeing, however, the pressing importance of the problem, I am now ready to take any possible opportunity to publish them in English. This essay is an introductory attempt to realize my intention, presenting at the same time critical views on the ways the national population of Japan in early Tokugawa days has been treated in the past, as well as the size of this population as estimated from lately discovered local materials.5

II It was in 1721 when the national population of Tokugawa Japan was enumerated for the first time. As one of the measures of the Kyo¯ho Reform to strengthen the administrative power of the Shogunate, Sho¯gun Yoshimune ordered the feudal lords to report the size of the population in their domains, according to their respective customary methods. This order was repeated in 1726, with a ruling that a report be submitted every 6 years. Not all the results of these orders are available now, but it is clear that this measure was continued in some way or other as far down as 1846. However, the figures thus obtained, generally ranging around 26,000,000 or 28,000,000, can not be called very significant. The new government of the Meiji period began a census system (the so-called Jinshin Census) in 1872, and found the national population to be about

12:27:16:11:09 Page 76 Page 77

POPULATION AT THE BEGINNING OF THE TOKUGAWA PERIOD

34,000,000, later adjusting it to 35,000,000. The discrepancy of 7 or 8,000,000 between the size of the population as counted by the Tokugawa and by the Meiji governments was not due to an increase in population but was due to the imperfect census-taking of the Tokugawa government. It is a common view that the population of Japan in 1720 would amount to about 30,000,000 if the omissions were added, and reached the 35,000,000 mark in the following 150 years. A demographic situation such as this is generally interpreted in terms of a stationary state of population movement or is taken as a case of Malthusian equilibrium in population. It is, however, important that this so-called stationary state be examined in two lights. In the first place, we generally assume an omission in census-taking to have taken place in a demographic situation like this, but in the particular case of Tokugawa Japan, duplication rather than omission could be the case. It has been discovered that duplication occurred in the early census- taking, especially when the census was carried out along with the registration of religious faith, as was the case with the Tokugawa government. It seems, therefore, quite reasonable to infer that the results of the national census in 1720 were closer to actuality than those of later years. In other words, the so-called stationary state of the population in the latter part of the Tokugawa period need not be taken as rigidly as is generally done. Secondly, even if we yield a point and admit that population movement in Japan was slack in the latter part of Tokugawa period, it is still not correct to grant that this occurred uniformly throughout the country, as different regions must have had unique conditions affecting this stationary trend in population. Whatever the case may be, there is little danger that the reported national estimate of Japanese population after 1721 was very far from the truth, since a national census was effected time and again after 1721. The situation is different, however, when we deal with the population of Japan prior to 1720. There is nothing to depend on but estimates when we try to look into the nationwide demographic condition of Japan in that period, as no positive evidence exists now to prove that an attempt was factually made to establish the size. A national estimate of Japanese population in the early days of the Tokugawa period was made by Togo Yoshida at the beginning of the 20th century. His method was very simple as will be explained later. Having succeeded in the unification of Japan, which had been in a chaotic state of civil war in the last part of the 16th century, Hideyoshi Toyotomi conducted a national land survey (Taiko¯ Kenchi), and found that the total area of Japan amounted to 18,500,000 koku. Koku is a quantitative estimation unit for cereals, equivalent to about 5 bushels. In this case, the national land, subject to taxation, was found to be 18,500,000 koku estimated in terms of rice crops. The national kokudaka (the amount of koku) was reestimated many times in the Tokugawa period which succeeded the Toyotomi period; it was set at 26,000,000 koku in the last part of the 17th century and at 28,000,000 koku in the middle of the 18th century. The last figure bespeaks the annual quota of one koku of rice per person. In other words, according to the report of the national census conducted by the Tokugawa government about that time, the population of Japan then was

12:27:16:11:09 Page 77 Page 78

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

estimated to be 28,000,000. Applying the relationship 18,500,000 koku = a 18,500,000 population for Japan in the 17th century, that is, 1 koku to one person, we may deduce the population of Japan in the first part of the 17 century to have been about 18,000,000. This estimate is generally accepted without much criticism. And it is the generally accepted view that the population of Japan was increasing to some extent up to the middle of the Tokugawa period but ceased to grow after that time. This estimation by Yoshida, however, involves a great hypothesis. The relationship: “1 koku to 1 person” may be significant. But he applied this relationship in his depiction of the national population of Japan 100 years later. I wonder if this is right. In the first place, let us examine the kokudaka. As a matter of fact, the kokudaka often did not represent the actual amount of the cereals produced. Not to mention the deficiencies in the survey methods used in those days, or the omissions in the survey, the figure the kokudaka expresses, being institutionalized, became gradually fixed apart from the actual harvested amount of products with the passage of time. In the Tokugawa period, it was officially forbidden to resurvey land even when it was obvious that it had gained in yield, except when it was newly reclaimed land. It seems important here to explain the purpose for which the land survey was conducted in those days. The land survey was enforced in the feudal age in order to find the proper standard for the annual assessment on the agricultural villages and that of the rice-stipend to be paid to the retainers of a provincial lord. The first standard was ascertained by multiplying the village kokudaka by a tax rate. Thus it was possible to increase the annual assessment by altering tax rates, while the village kokudaka estimated by a land survey was kept as officially set. On the other hand, the kokudaka of a provincial lord signified his status, the basis of the feudalistic social order. Any change, therefore, in the officially granted kokudaka of a feudal lord meant a disturbance in the existing social order, which was avoided if possible. Thus kokudaka remained unvaried in spite of the changes there may have been in actual productivity. The only change made in kokudaka was an occasional addition of a negligible piece of taxable land to the lord’s domain. In short, kokudaka expressed a dead value, statistically speaking. Thus, when the Meiji government made a nationwide survey of the total amount of the cereal products of the country about 1870, it was found that the amount far exceeded the one calculated on the basis of the kokudaka reported by the Tokugawa administration.6 This does not mean a sudden increase of cereal products just prior to the Meiji period, but testified to a continuous development of farming throughout the Tokugawa period. Such being the situation, it is erroneous to apply a relationship between the kokudaka and the population of Japan found at a specific time in the Tokugawa period to any other time. In other words, the estimate made by Yoshida of the population of Japan at the beginning of the 17th century seems to have no authenticity. Is it not strange that his work has been constantly quoted for more than 50 years?

12:27:16:11:09 Page 78 Page 79

POPULATION AT THE BEGINNING OF THE TOKUGAWA PERIOD

III Since it is impossible to obtain any directly and nationally collected statistics on population prior to 1721, the only way to calculate it is to estimate it from locally gathered material. It is fortunate, however, that the feudal lords who were assigned to rule provincially after the advent of Hideyoshi Toyotomi took interest in knowing the quantitative aspects of their domains and populace. Com- pared with the medieval lords who were rather indifferent to their provinces, the feudal lords who appeared in and after the 16th century seem to have been more rationally inclined in ruling their people. They were eager to ascertain the strength of their domains in order to administer their people wisely, to stabilize their provinces financially, to conduct business properly, and to win in war. Feudal lords who had been utterly indifferent as to how many houses or people there were in their domains since the 7th century came to pay attention to them. At first, they had to be satisfied with a knowledge of the number of houses only. With the coming of the 17th century, however, they became interested in knowing the number of their people, as they wanted to find out how much labor service they could exact in their domains. Of course, their attempt was far short from what is called vital statistics in the modern sense of the word. The survey was gradually improved, however, from an inquiry which took the household for its basic study unit and the aim of which was to find out how many houses there were which qualified in order to furnish a prescribed amount of statute labor, into one which, assuming a more rationalized procedure, took the individual for the unit of inquiry so as to decide the burden of service labor to be imposed according to sex and age. This point was referred to previously.7 The oldest and the best arranged demographic record of this sort was made in the beginning of the 17th century. Lord Hosokawa, whose domain centered around Kokura of Kyu¯ shu¯ , made surveys in 1609 and 1611 of the kokudaka, the number of houses, the number of people and the number of oxen and horses of each village in his provinces.8 There is another piece of evidence, similar in nature, which came out about the same time as the above, although the exact date of its issue is unknown. Lord Uesugi, whose headquarters were at Yonezawa in the To¯hoku district made a study of his domain on items practically the same as Hosokawa’s. A part of this work has been printed and published. It is dubious, however, how much scientific value is to be found therein if such a crude form of investigation is scrutinized statistically. So, as a more reliable source material, we cite here one which was obtained by Lord Hosokawa when he made an investigation of his provinces, Buzen and Bungo of Kyu¯ shu¯ , in 1622. Titled The Registration of Men and Domestic Animals in Kokura Domain (Kokura-han Jinchiku Aratamecho¯), it was printed and published.9 We will now examine this collection of materials, which is composed roughly of two types of historical material. One contains the total amounts of village kokudaka, the number of houses, people, oxen and horses, by the villages of the 10 counties in the entire domain, with the exception of the two counties for which no kokudaka is reported. The other reports on the families within a village, being

12:27:16:11:09 Page 79 Page 80

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

very much like the village register of the present. All the materials of the last type now available are ones concerned with Hayami County, Bungo Province. Coming across these two types of material that have been handed down, we students of historical demography feel quite encouraged, as they prove that some sort of personal study within a village was made in the past, and that they are fairly reliable for use as positive evidence. Before we proceed to analyze the material in general, we must first check its reliability. Table 5.1 below shows an unnatural distribution of ages, if judged by the end numerals of different ages obtained from the material of the Hayami County type. The persons whose ages end with 0, 5, or 8 are 4 times more numerous than the persons whose ages end with 4, 7, or 9. This uneven distribution of persons at different ages seems due to the low intellectual level of the general populace and the undeveloped training in statistical investigation that are generally prevalent in pre-modern society. But this particular case in Japan with the large discrepancies seems to involve another factor: the superstitious aversion of anything which sounds or hints at “death” and “suffering”, that is, “shi” and “ku” in Japanese. Hence, some uneducated persons in old Japan were prejudiced against 4 which reads “shi”; 7, “shichi”; and 9, “ku”.

IV Chart 5.1 represents 6,444 persons excepting 14 whose ages are unknown, by 5 years age divisions; these people are registered in the historical materials of 76 villages in Yufuin, Yokonada, Kitsuki districts of Hayami County. For 5 villages out of the 76, Beppu, Ishigaki, Hamawaki, Onokodaira, and Tateishi in the so-called Yokonada district, practically no people at and above 61 are entered. It cannot be that no persons of those ages lived there; probably, the investigation was conﬁned to those below 60 years of age. The population of these ﬁve villages is represented by the shaded parts to show it distinctly.

Table 5.1 The Age Distribution of the Population of Beppu and Ishigaki Villages, Hayami County, Bungo Province, 1622, as Judged by Numeral in Ten’s Place, 1622

Numeral in Ten’s Place Total Population (1) Population at and Above Percentage for (2) 21 Years Old (2)

0 128 118 21.4% 1 75 67 12.2 2 53 45 8.2 3 66 61 11.1 4 23 18 3.3 5 83 76 13.8 6 38 37 6.7 7 31 20 3.6 8 98 82 14.9 9 34 27 4.9 Total 629 551

12:27:16:11:09 Page 80 Page 81

POPULATION AT THE BEGINNING OF THE TOKUGAWA PERIOD

Chart 5.1 The Population Composition by Age, Hayami County, Bungo Province, 1622

Glancing at Chart 5.1, we are struck by two points: (1) an unbalanced ratio of the sexes and (2) a very small number of young people, these two combined greatly reducing the number of females below 20 years of age. This is again something inconceivable, bespeaking a possible blunder in the registration of females and youngsters. Should any population have this composition, it would be called a “vanishing type.” It may be possible with one or two villages, but it could never happen with so many as 76 as was the case with Hayami County. What was the reason for these results? As was mentioned before, it was in the nature of the object for which the registration was enforced that a demographic deformation like this occurred. As it aimed at finding out how much labor service could be secured in a domain, the investigation naturally confined its search to the males between 16 and 60 years of age, when examined on the basis of the individual instead of the family as its survey unit. Careless handling of the persons who were not the direct object of study accounted for the results shown in Chart 5.1. Should we conclude then that this chart is entirely useless? Examining the chart closely, however, we find it properly representing conceivable age compositions of the persons above a certain age group. It seems we can definitely give credit to the chart on this point. For example, the sex ratio of the persons at and above 31 years

12:27:16:11:09 Page 81 Page 82

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

of age is 120; this is a possible numerical value in the early years of the Tokugawa period. Using this as the basis, is it impossible for us to estimate the total population including the persons that are not mentioned in the historical material? In proceeding with our discussion on this basis, it is essential for us to grant the hypothesis that the population at and above 31 years of age, male and female, in the chart are actual numbers. Of course, this is a groundless hypothesis. Should we succeed, however, in showing that they represent certain percentages of the actual numbers, a theoretical estimation of the total population adjusted from them would not be a very difficult task. What we must do here is to substitute a proportionately calculated age composition of the people who are under as similar conditions as possible. It is, however, practically impossible to find the materials which will properly meet this requirement from among the ones that have been handed down to us, not only with reference to area conditions but to chronological time, since the one under our consideration now is the oldest of its type. Further, it is necessary to limit the selection of the age group population composition tables to the exact or the fairly exact ones which may be somewhat defective but which can be adjusted by other conditions or by some additional historical materials. Looking for the materials that will satisfy these conditions, we found that there was after all no way but to resort to the annually conducted religious faith registration materials. Thus, we have decided to use the annually conducted shu¯ mon-aratame-cho¯ (religious faith registration materials) that were taken in the years closest to 1622. Even these materials with very few exceptions, were by no means adequately analyzed to meet our purpose. Imperfect as these materials are, we have arranged them in Table 5.2 opposite. It is doubtful in the case of Ko¯mi village whether the actual number of servants was included. Also, it is not clear whether the low ratio of the population below age 5 was due to a low birth rate or a defective description of the material. While entertaining such doubts on this material, we raise another question, that of whether the composition ratio for the population at and above 31 years of age is too high. On the other hand, the figures for Yokouchi village are the results after necessary adjustments were made. In both cases, however, the number of the infants one year old is not included.11 The population composition ratio by age varies according to natural and social conditions. While putting aside the question of increase or decrease due to social conditions, it is natural that the composition ratio for the young rises with a rapid increase in the birth rate. The total population of Ko¯mi village during this period increased from 440 to 471 or from 361 to 374 if servants are excluded, although it tended to decrease temporarily in this period. But the average birth rate for the period was 15‰ (18‰, if exclusive of servants); this is unbelievably low. Since the average death rate for the period was 18‰ (21‰, if exclusive of servants), with the death rate thus exceeding the birth rate, there should have been a decrease in population. But in actuality, an increase in population occurred. This is a strange phenomenon. There must have been something wrong with the birth registration. Here again the Ko¯mi village information should be considered problematic. Now, taking up Yokouchi village, we find its population to have increased from 189 to 219 in this period with an average rate of increase of 1.2% a year. If judged

12:27:16:11:09 Page 82 Page 83 Total includes Females mi and Yokouchi Villages mi and Yokouchi ¯ 10 Males 5Below 31 & up 5 Below 31 & up 5 Below 31 & up c Age Groups in Ko ﬁ Sex Ratio Age both servants and emigrant employees. (annual average) 1688 356.5 99 10.4% 43.0% 8.4% 41.3% 9.3% 42.1% 16881681 433.0 201.4 109 113 9.0 12.3 40.2 34.5 7.4 10.2 37.8 42.5 11.4 8.2 38.2 37.8 – – – The Population Composition Ratios of Speci Composition Ratios The Population Table 5.2 Table mi village two types of population are given: one which excludes servants but includes emigrant employees, and the other which one which excludes servants includes emigrant but employees, are given: types of population mi village two ¯ * Ko For mi* 1674 mi inclusive of servantsmi inclusive 1674 ¯ ¯ VillagesKo Years Population Total Ko Yokouchi 1671

12:27:16:11:09 Page 83 Page 84

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

by the birth rate of 30‰, this increase seems to have come from a fairly rapid natural increase of births. In other words, since the population of this village seems to be on the increase, the population ratio below age 5 is high, and that of 31 and up is low, if compared with Ko¯mi village as presented in Table 5.2. Despite the considerations discussed above, we apply this population composition by age of Yokouchi village to Hayami County, since we find no better example to use for our purpose. Another point to be considered in this connection is the question of sex ratio. The sex ratio both of Ko¯mi and Yokouchi villages is lower than that of Hayami County. Since the sex ratio varies with different age groups, we hesitated in deciding which of them to use. Thus, the population composition by age is presented in two different ways in Table 5.3: one which utterly disregarded sex ratio, and the other which took it into consideration. According to the table, it is clear that the real size of the population in Hayami County could be obtained by multiplying the one which appeared in the historical material by 1.56 if the sex ratio is disregarded, and by 1.76 if it is to be given weight. In other words, the total population of Hayami County would be the number mentioned in the documents of 1622 plus an addition of 56% or 76%. Next, let us examine the composition of the family. First, it is important here to clarify the meaning of the word “family” used as the unit of investigation in the historical material. To state the conclusion first, it is very doubtful if we can safely depend on the expression “number of houses” which often appears as the caption of a column of totals. The “number of houses” is used in confusion of two ideas: one signifying the members comprising a family, and the other identifying the number of houses or structures such as “rooms”, “stables”, etc. Thus, the expression “number of houses” as used in the historical material is meaningless, statistically speaking. The situation, however, of Hayami County is different as an individual investigation of each member of a family was conducted. In Hayami County, there existed families of different scales, all the way from a small family consisting of a married couple and their children to the large family embracing scores of members.

Table 5.3 The Adjusted Population of Hayami County

Males Females Total Sex Ratio

a Population of 76 villages, Hayami County 3,691 2,739 6,430 135 b Population of Hayami County exclusive of 3,024 2,238 5,264 136 Beppu and other 4 villages c Population of Hayami County at 31 years of 1,710 1,425 3,135 120 age and up d Composition ratio of the population at 31 34.5% 42.5% 38.2% 92 years of age and up, Yokouchi village e c/d 4,910 4,380 9,290 f c/d 8,210 g e/b 1.76 h f/b 1.56

12:27:16:11:09 Page 84 Page 85

POPULATION AT THE BEGINNING OF THE TOKUGAWA PERIOD

The scale of a large family is determined by the extent to which it includes nago or genin (subordinate persons) as members. It is very difficult to tell if these nago or genin were quite subordinate to their masters or practically independent. Mr. Tokoro writes that since the livestock of each nago or genin is entered individually, “it is possible to say that they were economically independent in spite of their subordination to their masters”.12 This is certainly a sensible interpretation of the matter. Seeing however that almost all the farmers, who are the heads of large families, constitute a squire class of people, we shall take the families that are treated as unit entities in the documents to be independent families. (The number of the people entered here is without the additions discussed in a previous paragraph.) The Yokonada area includes such large villages as Beppu, Ishigaki and Hamawaki along the seacoast; the Kitsuki area is in the south of the Kunisaki Peninsula; the Yufuin area includes a small basin and hilly places separated from Beppu by a mountain. As is clear at one sight, there are distinct differences in the family compositions of these three areas. They are not so conspicuous with the average sizes of families, but coming to their rates, we find them very interestingly marked. The Yufuin area, for example, is not only large in point of the average size of its families but their distribution is extensively scattered compared with other places. Indeed, the families of 11 members and up comprise as many as 16% of all households, and one family has as many as 61 members. Certainly, there must be some factors working in this area which are peculiar to it. With the Kitsuki area we find a noticeable tendency for a smaller family. The family size distribution of this place is very much like that of the middle part of the Tokugawa period. Now, the family composition consists roughly of the blood relatives and the subordinate members. The number of persons who belong to the first category, including the master of a house, his wife, and their lineal relatives, is rather limited, and this number is very little influenced by area conditions. Thus it is obvious that the relative size of a family varies according to whether it has a large number of subordinate members or not. These figures should be adjusted for the reasons mentioned above. But generally speaking, the numerical variance in the family size between the Yufuin type and the Kitsuki type should be attributed to the difference in the developmental concept of family. The factor which regulates the family conditions of the Yufuin type is a remaining powerful medieval clannishness in the management of affairs, calling for a large number of subordinate laborers, whereas the Kitsuki type has progressed to a stage of small scale management based on the labor force of the family. The fact that the investigations, carried out by the same lord, in the same county, and in the same year, exhibited such a marked difference in their results should be considered characteristic evidence showing the transition from the ancient to an advanced stage of social development. Table 5.5 presents females from 16 to 50 years of age by area, according to whether they are married or not. Following the example of the documents, their status in the family is classified into two categories: the blood relations of the household head including wife, mother, daughter, daughter-in-law and sisters, and the other members including nago and genin. There is a possibility that the

12:27:16:11:09 Page 85 Page 86

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

Table 5.4 Family Composition by the Number of Members, Hayami County, 1622

Actual Numbers Rates

Areas Yokonada Kitsuki Yufuin Yokonada Kitsuki Yufuin Number of Villages* 5 45 27

Composing Family Membership: Number Household Household Household % % % 1 1 5 2 0.5 0.9 0.5 2 16 82 18 7.4 15.1 4.9 3 42 131 42 19.6 24.1 11.4 4 46 119 44 21.4 21.9 11.9 5 29 68 33 13.5 12.5 8.9 6 24 55 93 11.2 10.1 10.5 7 24 33 36 11.2 6.1 9.8 8 9 16 33 4.2 2.9 8.9 9 5 13 30 2.3 2.4 8.1 10 6 6 21 2.8 1.1 5.7 11 2 4 14 0.9 0.7 3.8 12 3 2 10 1.4 0.4 2.7 13 3 17 0.6 4.6 14 1 1 2 0.5 0.2 0.5 15 3 1 5 1.4 0.2 1.4 16 1 1 6 0.5 0.2 1.6 17 1 3 0.2 0.8 18 1 1 4 0.5 0.2 1.1 19 1 1 1 0.5 0.2 0.3 21 1 0.3 22 2 0.5 23 1 1 0.5 0.3 25 1 0.3 28 1 0.3 30 2 0.5 61 1 0.3 Total 215 543 369 100% 100% 100% Average Number 5.4 4.5 7.6

* In accordance with the feudatory nature of administration in the Tokugawa period, one village was sometimes reported as two tenures of land, or two villages, being put together as one tenure of land for a local provincial lord. These feudal tenures were adjusted in this study on the basis of the existing village unit.

unmarried includes some women who had been divorced or widowed. The document gives no information on this point. The marriage rate of the blood relations is very high. Among the subordinates, the nago shows as high a rate as the blood relations, whereas genin shows a low rate; this fact betrays the conditions of life which bind these people. Seen from the standpoint of demographic history, this low rate of the people of a lower status must have aﬀected to quite some extent the birth rate of the people in general. Of course, the word “marriage” creates a problem here. It is possible that some

12:27:16:11:09 Page 86 Page 87 Total Rate of the MarriedRate Genin 50 years old) and Marriage Rates Hayami County, 1622 50 years old) and Marriage County, Hayami Rates – Nago Number of Females (16 Number of Females Table 5.5 Table Rate of the MarriedRate (%) Blood relations Nago Genin Total Blood relations Nago Genin Total Number of the Married Married Unmarried Total Married Unmarried Total Married Unmarried Total Married Unmarried Total Area Blood relations Yokonada 186KitsukiYufuin 440Total 2 365 27 991 2Area 31 188 82 467Yokonada 82 367Kitsuki 0 Composition Rate Family 1,022 249Yufuin 413 5 50.0Average 2 7 68.0 82 47.8 55.6 87 1 251 21.8 24 420 9 34 105 12.7 109 32.5 28.2 144 22.9 358 106 19.4 133 19.8 100.0 153 21.4 392 269 546 100.0 107 1,438 98.8 623 100.0 141 100.0 396 148 94.3 99.5 376 97.0 1,834 687 100.0 771 94.2 0.9 99.2 98.3 18.0 71.5 5.9 8.7 79.6 81.8 78.4

12:27:16:11:09 Page 87 Page 88

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

births occurred out of wedlock, but it is quite certain that the unmarried had less chance for childbearing than the married. Assuming, therefore, that about 20% of the females capable of child-bearing were denied such a chance in this case, the general childbirth rate would be reduced to that extent. So, we may safely say that there exists a certain proportional relationship between the number of females who are denied the chance of childbearing and the birth rate in general. It is important, therefore, that we look for some positive evidence while studying the historical demography of Tokugawa period before we can come to any deﬁnite conclusion. Lastly, we shall examine the relationship between village kokudaka and population. In order to obtain insight into the relationship between a local kokudaka and its population, we picked the following areas: (A) the plains along the lower reaches of the rivers in Miyako County and Nakatsu County, (B) Kunisaki

County (Kunisaki Peninsula), (C1) the Yufuin area of Hayami County, (C2) the Kitsuki area of Hayami County, (D) the mountainous area of Usa County. In selecting these places, we mainly went by their natural conditions, excluding the urban villages and ﬁshing hamlets, though these were very few in number. Except for the last mentioned, we missed no village which appears in the still existing map attached to this historical material. Consequently, (A) is purely agricultural; (B) is characterized by comprising the

seacoast of a peninsula and some low hills; (C1) is a basin among mountains; (C2) is very much like B in its physical features; and (D) bears the features of mountain villages. To find out how these differences in the natural conditions of various areas affect kokudaka, the distribution of population, and the relationship between kokudaka and population is the aim of our analytical study presented in the following paragraphs.

Chart 5.2 including A, B, C1, C2, D shows the correlations between the village kokudaka and the population in each area. The numbers used in this Chart are as they were found in the documents. They should have been multiplied by 1.56 or 1.71 as was explained before. Disregarding, however, this necessary adjustment, we used the original numbers as they were, for our purpose was to seek the correlations among them. The straight line in each chart represents the regression line as analyzed by computer. A computer was also used in calculating correlation equations and coefficients.13 Looking at these results, we clearly perceive that there is a fairly high degree of correlation between the kokudaka and the population of every village. This high correlation, varied as it is according to different areas, means the existence of a correspondence between a certain amount of kokudaka and a certain size of population in a village. This characteristic may have weakened with the passage of time, but for the early years of Tokugawa period, before the commodity production had been generally developed, it seems correct to regard village kokudaka as corresponding to the cultivated area and its population as corresponding to the potential labor force. Thus at least some sort of correspondence, if not a rigid relationship, between the two factors, arable land and population, will have to be admitted.14 Next, we notice sharp differences in the slants of the regression lines; this means that the ratio between kokudaka and population varies according to

12:27:16:11:09 Page 88 Page 89

POPULATION AT THE BEGINNING OF THE TOKUGAWA PERIOD

Chart 5.2 A Correlation between Village Kokudaka and Population

12:27:16:11:09 Page 89 Page 90

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

12:27:16:11:09 Page 90 Page 91

POPULATION AT THE BEGINNING OF THE TOKUGAWA PERIOD

area. Chart 5.3 presents the regression lines adjusted from the ones shown in Chart 5.2. The first adjustment was the elimination of constants. The second was the multiplication of the numbers appearing in the historical material by the previously mentioned figures 1.56–1.76; 1.66 in this particular case for the purpose of obtaining the practical population. According to Chart 5.3, Area A shows the lowest ratio between village kokudaka and population; in other words, a definite amount of kokudaka there corresponds to a number of people which is smaller than that of any other area fi (1,000 koku to 276). Examining the increasing order in the ratio, we nd B and C2 take almost the same slant, D deviates from them in some measure, and C1 exhibits a marked contrast to them. The fact that Area A is open farm land, B and

C2 are on peninsulas, D is mountainous, and C1 is a small basin among hills seems to account for the observed variation in the relationship between kokudaka and population. In our study of these areas, we noticed that the means for life other than agriculture were amply provided for in descending order from A to D; in other words, judged by social structure, A should be considered the most advanced.

In Area C1 which was an exception to this general run of progress, “one koku per person” seems to have been the controlling standard of the people’s life there. Socially speaking, therefore, this place should be considered the most backward among the areas under investigation. While discussing family composition above we became aware that this area had more large families than any other area. Considering these facts, it seems permissible to observe that Yufuin, Hayami County, was slow to react against the feudal way of life because of its geographical isolation from other places, and continued strengthening clannishness among the people, especially among the large farm holders. When a land survey was carried

Chart 5.3 The Ratio between Kokudaka and Population in Diﬀerent Areas (1622)

12:27:16:11:09 Page 91 Page 92

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

out, for example, the oﬃcials in charge often made terms with these large family masters and reported the latter’s crops to be smaller than the actual amount, resulting in an apparent high ratio of village kokudaka against the population in

general. Whatever the situation, area C1 should be treated as an exceptional case. It is clear at any rate that there was a high degree of interrelationship between village kokudaka and population in the Kokura domain. In the following section, we are to examine this discovery to see if it can be used as a foothold in estimating the national population of Japan in early Tokugawa.

V It is clear that there was a definite correlation between village kokudaka and the size of population as was revealed by land surveys. Since, however, different areas exhibit a large difference in their coefficients of this relationship, we were undecided as to which of them should be used in the estimation of the nation’s population. For example, the national population would be 5,180,000, if we adopted the coefficient of Miyako-Nakatsu County, and it would be 8,140,000, if we calculated by that of the mountain solitudes of Usa County, as against the 18,500,000 koku of crops. As we were unable to learn to what extent of the nation each of these criteria should be applied, we took them as expressive of the minimum and the maximum, and decided to judge the population of Japan at the time of the surveys at from 5,180,000 to 8,140,000 estimated on the basis of national kokudaka. We assume, however, that this is merely “rural population”, and that it is important to add other demographic elements if we want to obtain the number of the entire population. In other words, it is essential to take the urban population— the warrior class, merchants and artisans—and the people in fishing villages and mining towns into consideration. It is, of course, impossible to calculate this. Mr. Naotaro Sekiyama estimates the urban population in the latter half of Tokugawa period at 3,700,000 or 3,800,000 out of a total population of 30,000,000, that is, “12% of the entire nation or one-seventh of the rural population”.15 This estimation of his, which is supported by the urban census taken in the early years of Meiji period and other similar reports, should be considered fairly accurate. Of course, this estimate cannot be applied just, as it is, to the early Tokugawa period. Keeping this fact in mind, however, we seem justified in assuming that the urban population and the provincial population which had no connection with kokudaka was 20% of the rural farming population. Considering the difference in the size of population between the early and the later part of the Tokugawa period, this percentage may appear somewhat high, but it is not too low. Adding this to the rural population, we have come to conclude that the population of Japan in early Tokugawa was from 6,220,000 to 9,800,000. On the basis of the 1721 survey, the population of Japan about that time is usually considered to have been 30,000,000; in other words, the population of Japan had increased 3.06 to 4.82 times, that is, at an annual rate from 0.9% to 1.3% during the 120 years up to 1721. This is by no means an impossible growth rate. In pre-modern society where such demographic factors as a high birth rate and

12:27:16:11:09 Page 92 Page 93

POPULATION AT THE BEGINNING OF THE TOKUGAWA PERIOD

a high mortality rate prevail, even a slight variation in these have a great effect on population trends. Let us assume here that the long term mortality rate of the Tokugawa period was 30‰ Thus an annual population increase of 0.4% coupled with the above mortality rate requires a 34‰ birth rate. Similarly, an annual increase rate of 0.9% requires a birth rate of 39‰, and an annual increase rate of 1.3% necessitates a birth rate of 43‰. (It is not necessary to consider emigration and immigration of people in discussing the demography of the Tokugawa period. The trend of the national population in that period was influenced by nothing but the difference between births and deaths.) Thus, an annual birth rate as high as 40‰ was possible. If expressed inversely, it is not at all out of proportion to suppose that the 30,000,000 population of Kyo¯ho¯ Era (1716–1735) was the result of an annual increase rate of 1% prevalent during the 120 years up that time. Chart 5.4 represents a sketch of the national population trends in the Tokugawa period as described above. The increase rate in this author’s estimation is rather conspicuous for the first half of the period (the part enclosed by the two lines in Chart 5.4, which indicate the 0.9% and the 1.3% annual increase) but it will be seen that this is by no means an impossible rate of increase. However, we as yet do not know when such a sudden change began and when it came to an end. Judging by the chart, however, we may tentatively consider that it ended in the Kyo¯ho¯ era.16 Before we begin examining the significance of the estimations we have obtained, we shall once again reflect on the process we have come through. To begin with, in estimating, we made various assumptions. They are: (1) that the population registration at Hayami County in 1622, at least the part which deals with the males of 31 years of age and up, is correct; (2) that in estimating the total

Chart 5.4 The National Population Trend

12:27:16:11:09 Page 93 Page 94

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

population from part of it, the classification model by age of the far-off Yokouchi village of Shinshu¯ province can be used; (3) that the ratio between population and kokudaka obtained from Buzen and Bungo provinces can be extended to a nationwide estimate; and (4) that the urban population (inclusive of fishermen; to be exact, the non-agricultural population) should be 20% of the agricultural population. These assumptions should be examined. First, we take up assumption (1). It is true that there is no guaranteeing the exactitude of this assumption. For that matter, no census can be perfect in numerical preciseness. Taking into consideration the rate of probable error in this case, there is ample reason to believe that the population calculated here is smaller than the actual number, and this will result in a reduced estimate of the national population. If it represents only 80% of the actual population, the national population estimated on that basis should be multiplied by 1.25. The problem with assumption (2) is that the model adopted in this case is one which was originally made for a village of Shinano Province which differs in chronological time and local conditions. As was mentioned previously, this model is of an expanding character as it is a population model classified by age at a time of rapid increase under a high birth rate and a high mortality rate. In other words, through the application of this model which represents a low percentage of population of working age, the entire population of Hayami County is apt to be estimated as greater than it probably was. Sex ratio is another question which should be considered here. The margin of its influence, plus or minus, will fall within 10 percent of the population. Taking up assumption (3), we recall that the Taiko¯ Land Survey played a decisive role in the systematization of land surveys with the village kokudaka as its central element. Generally speaking, it is safe to say that the calculated amount of the village kokudaka did not deviate much from the national average. The question to be considered here lies rather in the fact that this area is situated in such a remote place as Kyu¯ shu¯ . Compared with a small area centering around Kidai which was the most advanced area at that time, a large part of Japan in those days was either in a worse condition of backwardness or in a more advanced state. Therefore, in estimating the population on a national scale, it seems correct not to take this area as an exceptional case. Concerning assumption (4), it was definitely stated above that the 20 percent version of urban population may be too high, but it is definitely not too low. Thus, after checking the 4 assumptions, we come to the conclusion that assumption (1) may bring about an underestimation of the national population, and assumptions (2) and (4) may produce an overestimation. These two tenden- cies would tend to balance each other, and thus minimize probable errors in calculation, which would be less than 30 percent at worst. The real problem is with assumption (3). If the ratio of kokudaka of a backward rural village to its population differs widely from one of an advanced rural village, this sample is certainly not suitable to be used for the calculation of the national population, and it is important that some analytical study of similar material at another place be conducted. The author foresees that such a study would show a ratio of population to village kokudaka in an advanced area which is somewhat greater than in a backward area. But since there are places that are more backward than Buzen

12:27:16:11:09 Page 94 Page 95

POPULATION AT THE BEGINNING OF THE TOKUGAWA PERIOD

and Bungo, the ratio between the village kokudaka and the population of these areas can be considered not very far from the national average. Having examined the above assumptions, we may conclude that the estimated figures presented in these assumptions should be accredited with some sort of validity, although the margin between their maximum and minimum is often very large. At least, they are better founded than the relationship 18,000,000 koku of rice = 18,000,000 in population. Future studies will gradually reduce the distance from the truth. At the present stage of this type of research, we will have to be satisfied with the present degree of accurateness. We should next look into the significance of these estimations as related to historical facts. To begin with, let us consider them in connection with the population of the Kyo¯ho era. In discussing this, it is necessary to obtain an exact figure of the national population in the Kyo¯ho era, but here we will use the old one for expedience. These estimates do not wholly deny the common view that the population of Japan in the Tokugawa period increased rapidly in the first half of the period but was slow in increasing in the second. They even strengthen the view that there was a rapid increase of population in the first half of the Tokugawa period. Furthermore, the adjusted increase rates are by no means too high, as they fall within the limits of possibility. With the common view of the Japanese population thus established, in other words, a marked contrast in population trends between the first and the second half of the Tokugawa period, we naturally become interested in the reasons for such a situation. First, we will raise the question of what caused such a rapid increase of population in the first half of the Tokugawa period? We know of a small number of similar population movements which occurred in some disparate rural villages prior to the Kyo¯ho era. Thus it seems correct to assume that in the first half of the Tokugawa period the population increase in rural villages, together with a sudden development of cities followed by a rapid increase of urban population, enabled Japan to maintain a constant annual increase rate of the population of about 1%, a rate which should be considered rather high. Briefly stated, there existed few conditions which restricted population increase at that time. People enjoyed a long period of peace, the cultivated area was expanded, and towns sprang up centering around the castle-towns. In other words, rural folk had no cause for limiting their family size. These were the external conditions, social as well as economic, which resulted in the increase of population. The author, however, would like to rather call attention to the significance of the internal conditions that were closely connected with a change which took place in agricultural management. As was explained above, the agricultural management in the backward regions was mainly dependent on the possession of a large number of subordinate laborers. Demographically considered, this means a low marriage rate among the persons of working age, as was revealed in this article. No sudden increase of population due to an increase of births can be expected of a society where the marriage rate is low. The society prior to the Tokugawa

12:27:16:11:09 Page 95 Page 96

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

regime seems to have been one in this situation. There is no question that the total population of Japan then was only slowly increasing, or was stationary. In the beginning of the Tokugawa period, however, the management of agriculture underwent a sudden change, transforming itself into small scale farming based on the single family unit which was primarily composed of the nuclear family. It became important, therefore, to reproduce the necessary labor force within the limited capacity of an individual peasant family, thus bringing about a rise in marriage rates and birth rates. The death rate may have stayed unchanged, but a sudden increase of population was realized.

VI. CONCLUSION On the basis of the documents of 1622, an analytical study of Japanese population from various aspects has been undertaken. Of course, not all the contents of this document were used. More than anything else, we must recall that this work covers “domestic animals” as well as “men”. We should not, therefore, neglect the parts dealing with the numbers of oxen and horses, especially as related to the relationship between the types of agriculture and the number of oxen and horses necessary for each of them. But this essay has confined the use or the study of this work to the part dealing particularly with the “registration of men”. Needless to say, it is impossible to draw any general conclusions from what we have discussed at the present moment when the demographic study of Tokugawa Japan is just beginning. What we have learned by this study should rather be called a preparatory lesson for a further inquiry into the question of Tokugawa demography. After all, this is merely a trial study which requires much more detailed investigation of various positive evidence before it can be considered complete. Be that as it may, we now feel rather confident of the estimation we made of the national population of Japan which is considerably smaller than 18,000,000, the number generally held by scholars in the past as a legitimate inference of the population. This commonly held view is based on the assumption that the kokudaka and the population of Tokugawa Japan were in direct relationship to each other both during and after the Kyo¯ho era, and that this can be applied retrospectively to the early years of the Tokugawa period. The estimation presented here is the result of a nationwide hypothetical expansion of the situation in Buzen and Bungo Provinces. A hypothesis is after all only a hypothesis, however. Based on it, our estimations may be overestimated or underestimated. But it is likely that the marginal errors involved in them have been offset or narrowed down. Thus we may be permitted to claim a higher reliability for our estimations than for the estimations that were commonly used in the past. If our assumptions and hypotheses are granted, the national population in the early part of the Tokugawa regime most probably was less than 10,000,000. If we believe that no double reporting was committed, and that the exclusion of 3,000,000 or 4,000,000 persons was involved in the national census of the Kyo¯ho era, we naturally come to see a high increase rate of population prevailing in the early part of the Tokugawa regime. Was such a sudden increase of population

12:27:16:11:09 Page 96 Page 97

POPULATION AT THE BEGINNING OF THE TOKUGAWA PERIOD

possible in those days, in a society where nothing like modern medicine, public sanitation or an environmental study of life was known? Of course, this question can not be easily answered. But there is one thing which may serve as a possible explanation for a high fertility of the people at that time, an increase in the number of small families throughout the country. As was mentioned above, the marriage rate of genin, an important factor in the supply of service labor, was low; hence their birth rate was also low. Thus it should be generally surmised that the greater the proportion of genin in the composition of the population, the lower was the fertility of a population. Once freed, however, genin proved a vital agency in the increase of population. The early part of the Tokugawa regime seems to have been a typical case of this. Also, domains became fixed and provincial lords took an interest in the efficient management of their lands, often trying to increase the productivity of their farms, expanding the area of cultivated land, facilitating irrigation, and promoting conservation works. The situation also encouraged the people to open up new fields, which had a stabilizing effect on the older fields and was thus helpful in supporting an increasing population in a village. Another factor in connection with the population increase of the early Tokugawa period is the emergence of towns, mostly castle-towns built by ambi- tious feudal lords. This seems to have been a powerful agency in the rise of the birth rate. Thus both in urban and rural areas the population exhibited a sudden increase. The above explanation for the movement of Japanese population, depicting its size at the beginning of Tokugawa period as smaller than the generally recognized one, and admitting its sudden increase soon afterwards, is merely a hypothesis not yet proved by sufficient positive evidence. For this, we await similar studies on other areas.

NOTES

1. Kanetaro¯ Nomura: On Cultural Conditions Aﬀecting Population Trends in Japan, Economic Series No. 2, Division of Economics and Commerce, the Science Council of Japan, 1953, To¯kyo¯. 2. Ayanori Okazaki: Histoire du Japon, l’economie et la population, Paris. 3. Irene B. Taeuber: The Population of Japan. 4. Akira Hayami: “Demographic History of Tokugawa Period (Tokugawa Jidai no Jinko¯-shi Kenkyu¯ )”, Shakai-Keizai-Shigaku, Vol. 32, No. 2, 1966. Akira Hayami: “The Vital Statistics in a Late-Tokugawa Village—A Quantitative Analysis of the Shümon-aratame-cho¯, 1778–1871, (Tokugawa-ko¯ki Owari Ichi No¯son no Jinko¯ To¯kei)”, Mita Gakkai Zasshi, Vol. 59, No. 1, 1966. Akira Hayami: “The Demographical Aspects of a Rural Village in Tokugawa Japan, 1671–1871” (Shu¯ mon-aratame-cho¯ o tsu¯ jite mita Shinshu¯ Yokouchimura no cho¯ki jinko¯-to¯kei) Keizaigaku Nenpo¯, No. 10, 1966. Yo¯ichiro¯ Sasaki: “An Urban Population in the later Tokugawa Period” (Tokugawajidai ko¯ki toshi jinko¯ no kenkyu¯—Settsu-no-kuni, Nishinari-gun, Tenno¯ji-mura) Shikai, No. 14, 1967. Yo¯ichiro¯ Sasaki: “Population Trends in Musashi-no-kuni, 1823–1876 (Bakumatsumeiji Shoki Musashi-no-kuni Jinko¯ Su¯ sei ni kansuru Ichiko¯satsu,”) Mita Gakkai Zasshi, Vol. 59, No. 3, 1966. 5. Akira Hayami: “An Estimation of the Gross Population in Japan at the Beginning of the 17th Century (Kokurahan Jinchiku Aratame-cho¯ no Bunseki to Tokugawa Shoki Zenkoku Jinko¯ Suikei no Kokoromi), Mita Gakkai Zasshi, Vol. 59, No. 3, 1966. 6. For example, the total cereal production of 47.2 million koku, including 31.5 million koku of rice,

12:27:16:11:09 Page 97 Page 98

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

11.8 million koku of wheat, and 3.9 million koku of other grains, estimated at the beginning of the Meiji Era (1870s) would be 94.4 million koku, if estimated on the basis of koku unit used in the Tokugawa period. Calculated from K. Ohkawa, M. Shinohara and M. Umemura (ed.), Estimates of Long-term Economic Statistic of Japan since 1868. (Cho¯ki Keizai To¯kei), 9, Agriculture and Forestry. 1966, Tokyo. Chap. 3 “Agricultural Output.” 7. Akira Hayami: The Registration of Population and the Yakuya System in the 17th Century. (Kinsei Shoki no Iekazu-hitokazu Aratame to yakuya ni tsuite) Keizaigaku Nenpo No. 1, 1957. 8. The Registration of Men and Beasts of Kokura Domain (Kokura-han Jinchiku Aratamecho), 5 vols., The Japanese Historical Material of Tokugawa Regime (Dai-Nihon Kinsei-shiryo¯). 9. Fukushima Prefectural History (Fukushima-ken-shi), Vol. 5, 1965. 10. A Study Group of the Prof. Nomura Seminar, “Statistical Survey on the Population of Ko¯mi Village, Motosu County, Mino Province, O¯¯ gaki Domain—1674–1872—”, The Mita Gakkai-zasshi, Vol. 53, Nos. 10, 11, 1960. Akira Hayami: “The Demographical Aspects of a Rural Village in Tokugawa Japan.” 11. This is an error due to the old Japanese way of counting age, that is, one year is added on every New Year. Throughout this essay this old method has been used. 12. Mitsuo Tokoro: “Labor Service as Carried on by Peasants in Early Tokugawa Days”, (Kinsei Shoki no Hyakusho¯ Honyaku) Shohei Takamura and Yoshitaka Komatsu, ed.,: Feudalism and Capitalism (Ho¯kensei to Shihonsei). 13. For these calculations, the author is indebted to Mrs. Yo¯ko Sano, the Institute of Management and Labor Studies in Keio University. 14. In Chart 5.2, A→D, the values of the constants are not large. They may be quite disregarded, especially in estimating the national population of Japan as is discussed in the following section. “Statistical Survey on the Population of Ko¯mimura Motosu-gun, Mino-no-kuni.” 15. Naotaro¯ Sekiyama: Population Structure of Tokugawa Japan (Kinsei Nihon no Jinko¯ Ko¯zo¯), 1958, p. 239. 16. Strictly speaking, we are quite unaware which part of the Tensho¯-Genna Era (1573–1623) is referred to by the expression “18,000,000 population=18,000,000 koku of crops”, whether or not the population increase in Kyo¯ho era was in a straight upward movement, and of the general population trend up to the Meiji period after the Kyo¯ho era. In a way, we may consider that a considerable population increase occurred in some areas of Japan around the closing years of the shogunate and the Meiji Restoration.

12:27:16:11:09 Page 98 Page 99

Invited Paper 3.2, 46th Session of the ISI (International Statistics Institute), Tokyo, 1987.

6 Population Trends in Tokugawa Japan: 1600–1868

INTRODUCTION okugawa Japan (1600–1868) is often portrayed as a good example of popu- Tlation stagnation or the “Malthusian trap”. It has also been pictured as a society where the “immoral checks on population”, abortion and infanticide, were customary. Population surveys undertaken by the Tokugawa government from 1721 to 1846—to be examined in detail later—show that the population of Japan in this period ranged from a minimum of 24.9 million in 1792, to a maximum of 27.2 million in 1828. The average was 26.1 million, and the difference between the minimum and maximum figures was 1.2 million, or only 5 percent. This suggests a stagnating population, but, if we extend the period of observation and carry out careful demographic analysis, we find that population trends can by no means be disposed of with the word “stagnation”, nor “histoire immobile”. This period abounded in dynamic changes that paved the way for industrialization.

POPULATION TREND IN THE 17th CENTURY There is no direct proof of the size of the Japanese population before 1721, but, following an estimate made in the early 20th century, the figure of 18 million in 1600 has long been used.1 However, there is no scientific basis for this conclusion, and more recent work estimates the population to have 12 million plus or minus 2 million.2 While this estimate was based on a sample of local documents, it still has better empirical grounding than the earlier 18 million figure. Local records from the last half of the 17th century, which have higher degree of reliability, do verify that there was a remarkable increase in population in many daimyo territories and in towns and villages. Population growth in the territory of several daimyo (local lord) before 1732 can be seen in Table 6.1. The total population of this table contains about 12 percent of that of whole Japan, and we can say there was a fairly high population growth in the 17th century. The annual growth rates of 1 percent are by no means rare. If a population increases continually at a rate of 1 percent per year, it will triple within a century. The survey of 1721 showed a population of 26 million, and since it was probably underestimated by about 15 percent, the actual population was 30 million. Almost all of the remarkable increases of the 17th century followed a logistic

12:27:16:11:09 Page 99 Page 100

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

Table 6.1 Population changes before 1732 in several domains available

Domain (province) Period Population Annual growth rate (per 1000)

Morioka (Mutsu) 1669–1732 245,635 – 322,109 4.3 Ichinoseki (Mutsu) 1686–1725 25,756 – 27,694 1.9 Sendai (Mutsu) 1690–1732 599,241 – 647,427 1.8 Nihonmatsu (Mutsu) 1685–1732 73,351 – 70,614 −0.8 Aizu (Mutsu) 1669–1732 142,527 – 159,849 1.8 Nakamura (Mutsu) 1681–1717 87,840 – 86,502 −0.4 Tsuruoka (Dewa) 1694–1732 126,383 – 131,164 1.0 Yonezawa (Dewa) 1692–1732 133,259 – 113,711 −4.0 Mito (Hitachi) 1698–1732 276,922 – 315,996 3.9 Nagoya (Owari)* 1702–1732 626,822 – 821,614 9.2 Tsu (Ise)* 1665–1732 252,061 – 287,242 1.9 Okayama (Bizen) 1686–1721 185,043 – 223,959 5.5 Tokushima (Awa) 1665–1732 308,880 – 470,512 6.3 Kochi (Tosa) 1681–1732 327,971 – 418,498 4.8 Izuhara (Tsushima) 1665–1712 23,900 – 29,503 4.5 Kumamoto (Higo) 1634–1734 203,678 – 531,248 10.0 Kagoshima (Satsuma)* 1698–1732 260,961 – 339,955 7.8

Provinces with asterisks are where the lord had considerable size of domain besides the provinces parenthesized. Sources: Naotaro Sekiyama, Kinsei nihon no jinko kozo, Tokyo. 1969. pp.125–129. Bonsen Takahashi, Nihon jinkoshi no kenkyu dai 2, Tokyo, 1941. Yoshinobu Yoshida, Okitama minshu seikatsushi, Tokyo, 1973, pp.113–114.

curve, so it is not unreasonable to consider that the population in 1600 numbered about 40 percent of the later total. An important characteristic of this rapid increase in population is that it must have begun at a particular point. Otherwise, had the increase continued at the rate of tripling in 100 years, the population of Japan 500 years earlier, about the time when the Kamakura bakufu was established in the year 1200, would have been only 120,000, a figure which is completely improbable. Research shows that Japan’s population had already reached 5.6 million by the 8th century.3 Consequently, the question is when and why the rapid population increase of the 17th century began. My understanding is as follows. First, in the 15th century, the central region, Kinai, pioneered in the development of cities and the spread of the circulation of money among peasants. Subsistence production changed to market-oriented production, and peasants began to produce at a high level of efficiency. Agricultural management using family labor was chosen over traditional forms of agricultural management, which depended on subordinate labor. As a consequence, the conjugal family became the unit of production in agricultural villages, family formation occurred at a great rate, and the population began to grow.4 Until the end of the 16th century this phenomenon was limited to the urban areas of the Kinai, such as the residential cities of Kyoto and Nara, where seigniorial lords received the land taxes of their estate in currency, and to the commercial and trading cities of Hyogo and Sakai. Consequently, the population of Japan as a whole was not greatly affected. The great change in population resulted from Toyotomi Hideyoshi’s unification

100

12:27:16:11:09 Page 100 Page 101

POPULATION TRENDS IN TOKUGAWA JAPAN: 1600–1868

of the country in the last two decades of the 16th century, and from the decisive execution, by the Tokugawa government which followed him, of its major policies: functioned and geographical separation warriors from agriculture (heino bunri), the establishment of castle towns (jokamachi) and the system of alternate attendance at the shogun’s court (sankin kotai). Cities, more than 200 of them, sprang up, like mushrooms after rain, all over Japan, and their residents obtained the necessities of daily life with money. There is almost no reliable information about the earliest urban populations, but it is certain that they made up, at a minimum, 10 percent of the total population. This change, which had been progressing gradually in Kinai during the previous two centuries, advanced swiftly, and penetrated throughout the country. It provided economic incentives all over the country, made peasants produce more efficiently, and established the family as the unit of agricultural management. Family formation in peasantry became common, a population explosion was kindled, and the 17th century saw a baby boom in villages. Servants who spent their lifetimes unmarried gradually disappeared and the proportion married increased. A single household came to be composed of a single married couple and their lineal relations, and, as a result, mean household size decreased significantly. This phenomenon is clearly evident in village population registers from 1670 on. Nuclearization of the family—if we can use that term—provided energy for the explosion of population.5 This population growth, probably first in Japan’s history, combined with the expansion of arable land in villages, sent large numbers of men and women into the abruptly created cities. The increase ended soon in some areas, and continued for a long time in others, depending on the scope for expansion of arable land. As a result, during the Tokugawa period, almost all of the arable land, which makes up about 15 percent of Japan’s total area, was converted into cultivated land, and only a very few areas of level pasture and forest remained. When one looks at the platy parts of Japan today, the most salient difference from other countries is that there is almost no land that is not used for either cultivation or for residences. This landscape was created during the Tokugawa period. Population growth brought about, first, expansion of arable land, and, secondly, increase in the amount of production from a given area of land. To the extent that Japanese agriculture was centered on producing rice on paddy fields, expansion of arable land necessitated investment in the irrigation system. Increased land productivity required deep tillage, use of large amounts of fertilizer, and careful control. On small areas of land this necessarily met with diminishing returns. Population growth meant that the marginal gains of the peasants in agricultural production were pursued to far lower levels than in other fields of economic activity. In Kinai, where the scope for expanision of arable land was small, and where population growth began early, it ended somewhat before 1700. Most of the other areas stopped growing by 1750. Of course, there were also areas—such as parts of Western Japan—where growth did continue throughout the Tokugawa period. Another characteristic of population increase in the 17th century is the growth of the urban population. The number of cities themselves increased, for reasons

101

12:27:16:11:09 Page 101 Page 102

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

mentioned above, and by about 1700, Edo’s population had surpassed one million, and Osaka’s and Kyoto’s populations had each reached 400,000. These three urban centers merited the special term san-to, the three great metropolises, and far outdistanced the other cities. Nagoya and Kanazawa were in the 100,000 population class, and the castle towns of Akita, Sendai, Fukui, Okayama, Hiroshima, Tokushima, Kumamoto, and Kagoshima, and the port towns of Sakai and Nagasaki were ranked with populations around 50,000. There were 22 other cities which are thought to have had populations over 20,000.6 Adding all these together gives us an urban population of about 3 million. Including the populations of other castle towns, and post towns, temple towns, and local market or handicraft towns adds an additional urban population of over 1 million. Rough calculation indicates that, in about 1700, urban population made up about 15 percent of the total Japanese population of 30 million. This urban population grew rapidly between 1600 and 1700. Most of the castle towns, of which Edo is typical, were established in places which had not been cities before. Let us assume that the population of the area of later Edo before 1590, when Tokugawa Ieyasu constructed a castle there, had been a couple of thousand. In order for the population to reach one million by 1700, 110 years later, an arithmetic annual net increase of 9,000 persons is necessary. However, since, as will be explained later, the rate of replacement of population in preindustrial cities was a negative figure, the influx into the city must have been much greater. The difference between fertility and mortality in this kind of urban population means that an annual influx of at least about 1 percent of the population is necessary for the city to maintain its size, and a greater suflux is needed for it to grow. If we assume that the urban population of the entire country was 4.5 million, then, in order to just maintain this population, 45,000 people must have moved from villages to the city each year. If we add to this the additional influx which must have occurred when the urban population was increasing, perhaps 70–80 thousand people were pulled to the cities each year throughout the whole country. Gilbert Rozman has estimated that 17th century Edo received a net immigration from villages of 10,000–15,000 persons per year.7

POPULATION TREND IN THE 18th CENTURY Ignoring, for the moment, limited local areas, there were no nationwide famines or epidemics recorded for the 17th century. Almost everywhere, population showed straightforward growth. With the coming of the 18th century, the situation changed. In various localities there are appropriate time-lags, but, overall, population growth abated, the population stagnated, and in some areas, it even declined. The urban population—especially the population of castle towns—reached its peak in the ﬁrst half of the 18th century. The Tokugawa government began investigating the total national population at precisely the period when this national population appears to have stopped growing. These surveys, which began in 1721 and were carried out once every six years from 1726 until 1846, are the only available source material from which to derive national statistical data in the Tokugawa period. Among these 22 surveys, all but three give sum totals for

102

12:27:16:11:09 Page 102 Page 103

POPULATION TRENDS IN TOKUGAWA JAPAN: 1600–1868

the entire country, 12 give data by province (kuni), and surprisingly enough, 10 give data by province with sex. As with the statistics of most pre-modern societies, these surveys are not of extremely high reliability. However, these surveys have great meaning for understanding Tokugawa Japan’s population trends through quantitative data. The Tokugawa government carried out administrative reforms in the 1720s and these surveys were inagurated as a result of those reforms. However, these surveys were not carried out like modern censuses. They had no fixed basis or method, were not taken throughout the nation on a single chosen day, and did not use a standard blank form. Rather, they are the results derived from each daimyo’s or magistrate’s investigations of the population—many or which were shumon- aratame-cho, religious faith register described earlier.8 Counting methods were not uniform: people under a certain age were not counted, the timing of migration was determined in different ways, both underestimation and some double- counting took place. The most significant lack is that the samurai, warrior class, population itself, including members of samurai households, were outside the scope of the investigations. Early Meiji statistics show a samurai population of about 2 million, but the extent of other sources of under-estimation, and of changes in the population omitted from the surveys, remain unresolved problems. It is certain that, in cities, the number of uncounted vagabonds increased, which even led to attempts to register them in Edo. Despite these shortcomings, these statistics do make it possible to know at least something of each province, and hence, to observe population trends in various areas from 1721 through 1846. When one analyzes the statistics from this perspective, the data which otherwise create little interest, showing in national totals, a stagnant population, come to life and reveal colorful local differences in the degree of population change. Figure 6.1 shows the extent of change in the population in each province from 1721 to 1846. There is a remarkable contrast between the northeastern region and the Kinai, where population declined, and Hokuriku and Western Japan, where it increased. Let us compare actual populations. The three provinces of northern Kanto—Kozuke, Shimotusuke, and Hitachi—and six provinces in the Sanyo area—Bizen, Bitchu, Bingo, Aki, Suo, and Nagato—each had a population of 1.8 million in 1721. By 1846, the former area had declined to 1.3 million, and the latter had increased to 2.3 million. The annual rates of change for these areas were about 0.2 percent, but totally opposite directions. As a result, over the space of 125 years, a difference in population of more than 1 million was created; if we combine the two areas, the result is no change at all. This is the meaning of “stagnation”: a chance result produced by totaling regional figures to obtain an aggregate national population count. There was no force operating to preserve the national population at a given level. Let us begin by examining areas where the population decreased. The decrease was most conspicuous in the two provinces of Ou, and the three provinces of northern Kanto, especially between 1721 and 1786, when the population of these five provinces declined by 18 percent. The annual rate was about minus 0.3 percent. In this area there are many examples of a daimyo noticing a decline in population of his own territory, investigating and recording the population for each territory every year, in order to establish various population maintaining

103

12:27:16:11:09 Page 103 Page 104

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

Figure 6.1 Population Change: 1721–1846

policies. When one examines these records, one sees that, in every place, the population entered the period of decline sometime between 1700 and 1720, and, after experiencing large declines midway, in 1755–56, and, again, in the 1780s, continued to decrease until about 1800. From about 1820 population gradually

104

12:27:16:11:09 Page 104 Page 105

POPULATION TRENDS IN TOKUGAWA JAPAN: 1600–1868

began to increase, and, after another large drop in the 1830s, growth was restored. However, the population in 1870 was no larger than it had been in 1700. Other than the temporary decline of the 1830s, these changes resulted from a long-term deterioration of climate during the 18th century and from terribly poor harvests, especially in 1755–56 and 1783–86, the periods termed, respectively, the Horeki and Tenmei famines. This area was one where cold and wet air sweeping in from the Odhotsk high pressure zone during the period of grain ripening in the summer brought about a fall in air temperature and a lack of sunshine, resulting in a precarious harvest. This pattern continued for a long time in the 18th century and, during it, particularly in the two periods mentioned above, its effects were terrible. Tales, some exaggerated, are told of hell-like conditions where enormous numbers of people lay on the ground, dead of starvation. There were no cold-resistant varieties of rice, and the population increase of the 17th century meant that cultivation had been extended even to marginal lands, so, in this area, the northern limit of rice cultivation using contemporary agricultural techniques, the smallest decline in temperature or decrease in sunlight undoubtedly brought about great damage. Neither lords nor peasants had effective means to counteract these conditions. The policy of maintaining the population was the only countermeasure which the lords could adopt to prevent the decrease of the agricultural population within the domain and preserve tax recepts. Included among these policies were providing incentives for people to migrate from areas, such as Hokuriku, where population was increasing and paying sustenance allowances for children who were born, in order to prevent the peasants from committing abortion and infanticide, among others. How did the peasants respond? The most direct response, when famine threatened, was to escape to an area where food was likely to be had. In reality, large- scale movement was difficult, there were no daimyo who would welcome refugees, and many turned sufferers away. In all probability, tens of thousands of people, once turned adrift, died of starvation when winter came. Abortion and infanticide were also practiced to reduce the number of mouths to feed. For peasants in this area in this period, population control was a choice made in a struggle to survive. How can we explain the gradual decline which occurred in the other two areas, the five provinces of southern Kanto and the five provinces of Kinai? During the same period, population in southern Kanto decreased by 12 percent, and in Kinai, by 9 percent. Edo and Kyoto-Osaka were the respective centers of these two areas where urbanization had progressed the farthest, where cash crops were cultivated in agricultural villages, and where opportunities for employment were abundant in handicrafts and by-employments. According to statistics for 1875, when the conditions of the Tokugawa period still prevailed, the percentage of urban population (the rate of population resident in administrative units of over 5000 to the total population of the area) was 20 percent in southern Kanto and 33 percent in Kinai.9 The effect on the climate was also not as severe as it was in northern regions. I believe that the high livel of urbanization was a factor explaining the population decrease in these two areas. Generally speaking, in preindustrial cities, the death rate is higher than the birth rate, and the urban population cannot replace itself endogenously. If there is

105

12:27:16:11:09 Page 105 Page 106

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

no large influx of population from the outside of the city, that is, from rural areas, the maintenance and growth of its population is impossible. Therefore, when the economy develops in a given area and the percentage of urban population increases, it serves as a brake on the increase of population in the area as a whole, and the population stagnates. This is the negative feed-back function between economic development and population growth in a preindustrial society discovered by E. A. Wrigley.10 The following facts explain the operation of this function in these two areas. Investigations of the birthplaces of commoners resident in Edo were carried out six times from 1843 through 1867. During that period, from 22 to 30 percent of the populace were born outside Edo—in all probability, in rural areas.11 Surveys of the residents of neighborhoods in Kyoto and Osaka carried out in the bakumatsu period also show that on the order of 20 to 30 percent were born elsewhere.12 Yoichiro Sasaki, who analyzed the registers of the town of Takayama in Hida Province, shows that if one carried out a simulation using figures for fertility, mortality, and age at marriage taken from a century of records from 1770 onwards, the population would have declined by two-thirds within a hundred years had there been no immigration.13 Especially in castle towns, large amounts of land in residential areas was secured for bushi residences, Buddhist temples, and Shinto shrines, so land for the common people to live on was extremely limited. The population resident on one of Kyoto’s busiest streets changed hardly at all between 1685 and 1863.14 This indicates that the population density of this street had already reached its marginal ceiling by the earliest part of that period. Buildings were constructed of wood, and fire and earthquakes were frequent, so high buildings could not be constructed. Increased population brought about urban sprawl. In Edo, for example, residential land extended out along radiating roads, went outside of Edo’s limits as an administrative area, and a megalopolis took shape. The same phenomenon occurred in other castle towns. As a result of the rise in the standard of living of town residents, population density decreased, possibly resulting in suburbanization. As Susan Hanley and Kozo Yamamura emphasize, another possible explanation for the decline of population in these “advanced areas” is that because the people who lived in these areas were well-educated, well-informed, and economically-oriented, they may have traded off the number of children for a better standard of living.15 Certainly, peasants in these areas, especially in Kinai, were already far from worrying about how to survive, and it is possible that the population stagnation and slight decline in this area was brought about by their limiting the number of children they had in order to secure a better life. However, there are almost no detailed demographic studies pertaining to this locality, so I will leave theirs as only one possible explanation. How can we explain the increasing population in Hokuriku and western Japan which contrasts with Kinai and Edo? When Japan became a peasant society and the conjugal family became the unit of agricultural production, the potential for an increase in population was created. If celibacy declined, most adults formed families, and since population control was unnecessary, the population would naturally increase. In such a society, there is no need to seek special reasons for the increase; a lack of increase is what must be explained.

106

12:27:16:11:09 Page 106 Page 107

POPULATION TRENDS IN TOKUGAWA JAPAN: 1600–1868

To be more specific, we can present the examples of Ise Bay, the Inland Sea, and the western coast of Kyushu, where draining of the seaside and the building of poulders progressed and the amount of arable land expanded. There were many shallow shelves extending a great distance from the shoreline in this area, and arable land could be created easily. For example, in the innermost part of Ise Bay drainage of the seashore proceeded throughout the Tokugawa period, and the shoreline advanced into the sea at an average of 30 yards per year. The registers of a village located in this area show that, from 1778 to 1871 the population grew from 162 to 410.16 If one omits those who entered the village for marriage—balanced by those who left the village for the same reason—the average annual natural rate of increase is a high figure of 1 percent. In western Japan, climate and transportation were good to continue the population growth, if it was not so high. The cultivation of various cash crops, by-employments and rural handicraft industries progressed. Through experience, the peasants learned to make use of the principle of comparative advantage: when specialization of production occurred, productivity increased, and the carrying capacity of the population grew. However, the contrary explanation—that industries developed in response to low labor costs that resulted from increased population—is also a possibility. However, since population and economics are mutually interconnected, research providing much more corroborative evidence is needed. In any case, population increase and the growth of industry occurred together in the western part of Japan. Suo province showed the highest rate of population growth in the nation between 1721 and 1846—a rate of increase of 66 percent, or 0.4 percent annually. Furthermore, Shunsaku Nishikawa and Thomas C. Smith’s econometric studies show that the non-agricultural income of rural residents in this area around 1830 was plus or minus 50 percent of their total income.17 Without a doubt, the western part of Japan developed as a center supplying raw materials and finished products to the handicraft industries that were gradually growing. Economic development occurred primarily in the rural areas, so population growth avoided becoming entangled in the negative feedback function such as occurred in southern Kanto and Kinai. Everywhere in the western part of Japan the urban population was less than 10 percent of the total. In light of this, how can we interpret population growth in Hokuriku, where non-agricultural output was not especially great? This coastal plain on the Japan Sea, including the province of Dewa, specialized in commercialized rice production. The amount of production did not fluctuate as much as in areas further north because specialization productivity was higher, and relatively large areas of agricultural land were used. Therefore, the Hokuriku achieved a high level of rice production per person. This high productivity and relatively easy coastal naviga- tion (except in winter) meant that in 1875, Hokuriku supported a high urban population of 16 percent. The province of Dewa came fourth in the nation, with an urban population ratio of 15 percent. Hence, examination of population trends by region shows individual patterns of growth, decline, and stagnation, but when these are added together they offset each other and become zero growth. This leads to the question “Was there some force which tended to produce a population growth rate of zero operating in Japan

107

12:27:16:11:09 Page 107 Page 108

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

in this period?” I believe not; zero population growth in aggregate terms was entirely accidental. The reason for believing this is that, if the factor market is fully established, the stagnation of the national population would be related to the national economy, particularly if labor can migrate freely. Although migration was extensive in the Tokugawa period, the common people could not move freely over long distances.18 Kanto and its surrounding area was the source of in-migration into Edo, and Kinai and its surrounding area, that for Kyoto and Osaka. The radius of the area of voluntary labor migration was typically about 100 miles. It was not possible for the economically distressed peasants of northeastern Japan to move to prosperous western Japan in order to obtain jobs. The population ﬂuctuated independently, probably with the locality as the unit of change. Hence the force of a demographic cause-and-eﬀect relationship reciprocal between localities, which could have operated nationally, was limited to neighboring localities, to the extent that it existed at all.

POPULATION TREND IN THE EARLY 19th CENTURY Population surveys carried out by the Tokugawa government in 1840, which give data by province, have been discovered recently.19 Their figures provide additional confirmation of such regional population trends during the first half of the 19th century by allowing us to trace population changes in each province at six-year intervals from 1822 through 1846. From this we learn that there was a fairly large decline between 1834 and 1840. During this period the national population declined 4.2 percent, an annual rate of 0.7 percent. Decline occurred in 62 of the 68 provinces of Japan and was most precipitous in Ou, Hokuriku, Kinai, and San’in, while Kanto, Shikoku, and Kyushu were comparatively stable. Other documents show that in 1837 and 1838 Osaka’s population declined by 11 percent; in Owase town in the province of Kii, 20 percent of the population aged 8 and over died;20 and, in 1837 alone the death rate in Takayama in Hida province reached 87 per thousand.21 This is clearly due to epidemics, and actual documents from Owase show that the most common causes of death in this period were digestive diseases accompanied by high fever and diarrhea—probably typhoid or dysentery. It is now clear that the decline of population in 1837 and 1838 was brought about by the Tenpo famine, but that death was due, not to starvation, but to epidemics.22 But it is also possible that famine resulted from the decrease in the agricultural labor force brought about by epidemics. Between 1804 and 1828, the national population increased by 6 percent, an annual rate of 0.2 percent. Even in Ou, Kanto, and Kinai, areas of conspicuous decline until that period, the population increased. Epidemics caused temporary declines in Japan’s population which, with the coming of the 19th century, had begun to increase once again. The rebound from those declines was both high and swift. Population increased in 64 of Japan’s 68 provinces from 1840 to 1846, at a rate of 3.8 percent, or 0.6 percent per year. The effects of this famine, unlike that of 1780, did not continue long. In short, we can say that with the coming of the 19th century, Japan’s population once again began to grow. The epidemic of 1837–1838 was a completely

108

12:27:16:11:09 Page 108 Page 109

POPULATION TRENDS IN TOKUGAWA JAPAN: 1600–1868

exogenous event, and it did not disrupt the underlying structural causes leading to the growth trend. There is no conclusive evidence as to whether the Tokugawa government carried out population surveys after 1846 until its collapse in 1868. Documents from some daimyo show that they did report the populations of their domains to the government, which seemed to lose the capability to compile national population surveys. The next opportunity to calculate the national population does not come until after the Meiji government enacted a completely new registration system in 1872. Consequently, there is a vacuum of a quarter of a century. No part of the population is consciously excluded from the Meiji figures, and although there is some underestimation, they are highly reliable. The total from the 1872 registers is 33.1 million, later corrected to 34.8 million. There are also other estimations on early Meiji population, but, at any rate, there is statistical discontinuity between these figures and the Tokugawa government surveys, and the two cannot be linked. Observation of changes in population after 1872 shows that it increased at a rate of 0.5 percent per year until about 1890, and it is impossible to believe that this increase began suddenly with the Meiji Restoration. There is no doubt that it is a continuation of the trend of population growth which commenced at the beginning of the 19th century. During this quarter century of blank space, various events important from a demographic and economic standpoint occurred. Cholera was rampant from 1858 to 1861. The pandemic which occurred at the time was a result of the opening of Japan’s ports, which brought it to Japan. Cholera twice hit the cities and seems to have caused a considerable decline in the population of central Japan. Western medicine, particularly vaccination against smallpox, was introduced to a certain extent, and it lowered infant and child mortality especially. The wealth of merchants and the upper-class peasants was economically important in contrast to the poverty of the samurai. The establishment of textile and food-processing industries in villages and country towns created a certain shift in population distribution. After the opening of full-scale trade with foreign countries in the 1850s, areas producing raw silk, silk fabrics, tea, and other major export goods, and commercial market towns dealing with them experienced a boom, and population tended to concentrate there. The Tokugawa government and the daimyo lost their governing power, severe inflation forced samurai class’s purchasing power into a recession, and with the relaxation of the sankin kotai system at the beginning of the 1860s, economic demand from Edo greatly decreased. This undoubtedly affected Osaka as well. Finally, internal warfare accompanying the Restoration—albeit on a relatively small scale for such a political change—killed tens of thousands of people, and the Restoration’s reforms changed the distribution of population. For example, the Tokugawa vassals, who had lived in Edo, moved, along with the shogun, to Shizuoka, in Suruga province, and Shizuoka and its environs had to accommodate tens of thousands of people at a stroke. During the blank quarter century from 1846 to 1872 many factors caused fluctuation in population size and rearranged its distribution, so we cannot assume that the population trends of the Tokugawa period continued unchanged.23 How- ever, my own research on the Ou area shows that, with the exception of the

109

12:27:16:11:09 Page 109 Page 110

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

interruption caused by the prevalence of cholera in 1861, population in a number of daimyo territories grew at an average annual rate of 0.4 percent. This is approximately the same as the average annual rate of increase of 0.5 percent from 1872 to 1890. Thus it is clear that there is a continuity between the population trends of these two periods. In conclusion, we can say that although the trend of population growth established at the beginning of the 19th century was disturbed by the epidemics in 1837–1838 and 1858–1861 and by the disorder of the Restoration, this long-term trend was not affected, and growth continued at an annual rate of from 0.2 percent at the beginning of the period to 0.5 percent in the 1880s. How- ever, the trends in specific localities differed from those earlier in the Tokugawa period. The calamities which brought about large declines in population in Tokugawa Japan are often termed the three great disasters: The Kyoho famine, which occurred in western Japan in 1733 and was caused by infestations of rice insects; the Tenmei famine, which spread from northeastern Japan to central Japan in the 1780s; and the Tempo famine, which stemmed from epidemics that flourished in northeastern and central Japan in 1837–1838. These years were ones of bad harvests, so rice prices also jumped. I believe that the Horeki famine (1750s) should also be included in this list of disasters. However, the important fact is that people did cope with these great calamities. From the Kyoho famine the peasants learned the use of whale oil to stamp out rice insects, and though there is no doubt that in later periods these insects often did fly in from the Asian continent, the subsequent damage they caused was minimized. The Horeki and Tenmei famines dealt an especially heavy blow to the daimyo and peasants of northeastern Japan, but the daimyo responded by inducing migration from areas of population growth (Hokuriku, for example), and by giving subsidies for the expenses of raising children in order to curb abortion and infanticide. As yet there are no quantitative measures of the effects of these policies, but it is worthy to note that though strapped by financial difficulties, daimyo nonetheless made such disbursements. They also carried out the policy of ordering villages and towns to provide for the storage of grain. Peasants who experienced these famines began endeavoring to ensure a stable harvest, and many empirical works such as agricultural handbooks (nosho) were published and circulated. With the exception of the Tempo epidemics which stemmed from uncontrollably bad harvests, there was little death due to starvation after the Tenmei famine. These Tenpo epidemics seem to have provided the Japanese some motivation to conquer disease. The enthusiasm for Dutch medicine increased. Vaccination against smallpox was introduced from Nagasaki to Kyoto at the end of the 1840s, and Kyoto bacame a center for medical schools. The significant decline in child mortality in all areas from this period onward is probably a result. Various calamities attacked Japan in the latter half of the Tokugawa era and reduced its population. However, we should place more emphasis on the fact that the people learned from these calamities and forged countermeasures, thus minimizing population declines.

110

12:27:16:11:09 Page 110 Page 111

POPULATION TRENDS IN TOKUGAWA JAPAN: 1600–1868

A CONCLUSION Until now, abortion and infanticide—measures frequently mentioned as means of population control—have been interpreted by Japanese scholars as the most salient evidence of poverty and misery in Tokugawa Japan. But as demographic studies of Tokugawa Japan have progressed, a few people have come to hold the opinion that its relatively low growth rate—so unlike the high population growth which aﬄicted contemporary China and oppresses most of the less industrialized countries today—raised per capita income and contributed to Japan’s industrialization. Thus, the implications of the term “stagnation”, and its relation to industrialization, have become completely reversed. I agree that over the long term, this explanation applies, but feel that a number of reservations are necessary. In the 17th century, what is tantamount to a population “explosion” occurred, but in the 18th century, the population patterns of northern Japan cannot be explained without using the word “misery”. On the other hand, in the western part of Japan, population growth continued at an annual rate of 0.3 percent, and remarkable economic development occurred in the area. From 1800 onward, population growth became a national phenomenon, although it was interrupted by epidemics, the national population had begun long-term growth which continued as modern population growth.

SUMMARY Tokugawa Japan has been frequently described as a model of “Malthusian trap”, with population stagnation, abortion and infanticide. Poverty and misery have been labeled to this society due to this stagnation. Recent studies, however, press us to reexamine such a stereotyped assessment to Tokugawa Japan. Rather, through the new views, it was the decisive pre-condition to Japan’s success of industrialization. Demographically, there was nothing of a “stagnation”, but several dimensions by time and space. The first phase, the 17th century, was characterized by very high population growth with urbanization. The second, the 18th century, had a “stagnation” when we looked national population. But, if we examine it locally, “stagnation” might have happened by chance, offsetted by totally different trends of local population. The third, the early 19th century, obviously told us steady population increase nationwide, followed by modern growth.

RESUME Jusqu’à récemment, les chercheurs japonais ont expliqué l’avortment et l’infanticide—moyens de contrôle des naissance—sont les preuves de la pauvreté et de la misère au Japon, à la période de Tokugawa. Mais nos recherches ont progressé, et maintenant, certains sont d’avis que le taux d’augmentation relativement bas de la population—contre le taux d’augmentation relativement haut des pays en voie de développment et de le Chine—a éleve le revenu par tête, et a aidé le développement économique du Japon. Ainsi, l’interprétation de “la stagnation” et la relation entre la stagnation et l’industrialisation est renversée.

111

12:27:16:11:09 Page 111 Page 112

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

Je partage cet avis si l’interprétation est appliqué pour une longue durée, mais avec quelques réservees. Au XVII siècle, il y a eu une “explosion” démographique. Au XVIII siècle, a eu lieu dans le Nord du Japon une crise demographique qu’on ne peut pas expliquer sans le mot “misère”. Mais, dans l‘ouest du Japon, l’augmentation de la population a continué à raison de 0.3% par an, et la region a connu un développement économique tout à fait impressionnant. Après 1800, on note une augmentation de la population dans toutes les régions du Japon, et la population totale augmente malgré des interruptions dues à des épidemies, selon le modèle démographique moderne.

NOTES

1. Togo Yoshida, Ishinshi Hakko. Tokyo, 1910. p.26. 2. I added 20 percent diﬀerence to 12 million estimated by Shakai Kogaku Kenkyusho. Shakai Kogaku Kenkyusho, Nihon Retto niokeru Jinkobunpu no Choki Jikeiretsu Bunseki. Tokyo. 1974. p. 58. 3. Goichi Sawada, Naracho Jidai Minsei Keizai no Suteki Kenkyu. Tokyo, 1927. p.146. 4. Akira Hayami, “A Great Transformation” in Bonner Zeitschrift fur Japanologie, Band 8, 1986. pp. 1–13. 5. Akira Hayami and Nobuko Uchida, “Size of household in a Japanese county throughout the Tokugawa era” in Peter Laslett (ed.), Household and Family in Past Time. Cambridge, 1972. pp. 473–515. 6. Gilbert Rozman, Urban Networks in Ch’ing China and Tokugawa Japan. Princeton, 1973. 7. Gilbert Rozman, “Edo’s Importance in the Changing Tokugawa Society” in Journal of Japanese Studies, Vol.1, No.1, 1974. pp.91–112. 8. Akira Hayami, “La demographie du Japon a l’epoque des Tokugawa” in Population et Famille, 54, 1981–3. pp. 81–101. Laurel L. Cornell and Akira Hayami, “The Shumon Aratame Cho: Japan’s Population Registers”. in Journal of Family History, Vol.11, No.4, 1986. pp.311–328. 9. Rikugunsho Sanbokyoku, Kyobu Seihyo. Tokyo, 1875. 10. E. A. Wrigley, Population and History. London, 1969. 11. Naotaro Sekiyama, Kinsei Nihon no Jinko Kozo. Tokyo, 1969. p.221. 12. Akira Hayami, “Kyoto Machikata no Shumon Aratame Cho” in Kenkyu Kiyo (Tokugawa Rinseishi Kenkyusho), Showa 55 nendo, 1981. pp.502–541. 13. Yoichiro Sasaki, “Urban Migration and Fertility in Tokugawa Japan: The City of Takayama, 1773–1871” in Susan B. Hanley and Arthur P. Wolf (ed.), Family and Population in East Asian History. Stanford, 1985. pp. 133–153. 14. Akira Hayami, “Kyoto Machikata”. 15. Susan B. Hanley and Kozo Yamamura, Economic and Demographic Change in Preindustrial Japan, 1600–1868. Princeton, 1977. 16. Akira Hayami, “The Demographic Analysis of a Village in Tokugawa Japan: Kando-Shinden of Owari Province” in Keio Economic Studies, Vol.5, 1968. pp. 50–88. 17. Shunsaku Nishikawa, “Protoindustrialization in the Domain of Choshu in the Eighteenth and Nineteenth Centuries” in Keio Economic Studies, Vol. 18, No.2, 1981. pp.13–26. Thomas C. Smith, “Farm Family By-employments in Preindustrial Japan”. in Journal of Economic History, Vol.29, No.4, 1969. pp.687–715. 18. Akira Hayami, “Labor Migration in a Pre-industrial Society: A Study Tracing the Life Histories of the Inhabitants of a Village” in Keio Economic Studies, Vol.10, No.2, 1973. pp.1–18. 19. Kazuo Minaki, Bakumatsu Edo Shakai no Kenkyu. Tokyo, 1978. pp.164–185. 20. Akira Hayami, “Kishu Owase-gumi no Jinko Susei” in Kenkyu Kiyo, Showa 43 nendo, 1969. pp.304–348. 21. Yoichiro Sasaki, “Hida-no-kuni Takayama no Jinko Kenkyu” in Shakai Keizaishi Gakkai (ed.), Keizaishi niokeru Jinko. Tokyo, 1969. pp.95–117. 22. Recent work, however, suggests that this loss was caused by famine. Ann Jannetta, Epidemics and Mortality in Early Modern Japan. Princeton, 1987. 23. I have described the population change in this period in following work. Akira Hayami, “Population Changes” in Marius B. Jansen and Gilbert Rozman (ed.), Japan in Transition: from Tokugawa to Meiji. Princeton, 1986.

112

12:27:16:11:09 Page 112 Page 113

First published in Evolution agraire et croissance démographique (ed. A. Fauve-Chamoux), Liège, pp. 33–9, 1987.

7 Population Growth in Pre-industrial Japan *

opulation growth and economic development have been related to each other Pin almost all known societies, but scholars have been unable to explain which came first: it is a typical chicken-and-egg problem. Moreover, since population growth within a closed geographic area may take place either through increased fertility or decreased mortality, the relationship between population and the economy is a complex one. For example, on the one hand, we could assume that population growth first took place when ecological changes that produced good harvests induced an increasing birth rate. This in turn may have led to economic development through the extension of effective demand in a growing population that mobilized more resources. On the other hand, economic development may have first raised the standard of living and consequently brought down mortality, thereby creating a population increase. Generally speaking, economists prefer the latter explanation, while other social scientists prefer the former; and historians waver between the two alternatives. But recent scholarship reveals that population growth, whether it stemmed from rising fertility or declining mortality, has produced economic development from early to recent periods of history. The Ester Boserup thesis1 holds that from the period of slash-and-burn agriculture to that of the medieval European three-field agrarian system, population growth or even pressure has forced humankind to use land more efficiently, and consequently, has engendered economic improvement. B.H. Slicher van Bath2 collected much evidence to indicate that in medieval to early modern Western Europe, the development of agriculture, technology, and more efficient land use, contributed to the population growth that preceded economic development. The most forceful theory of “population growth preceding economic development” has been put forth by Douglas C. North,3 who holds that prior population change was the most decisive factor in economic change in Western European history. A neglected study by R.G. Wilkinson4 also declares that population growth is the key factor behind social change in world history. However, in the contemporary world, we know that developing nations are suffering from high population growth that prevents economic development. China, for example, which was once proud of its huge population and its high growth rate, has made a complete turn-about in policy and is now conducting a “one child per family” campaign. This is a great paradox: in the past, population growth preceded economic development, but today it obstructs such development. What is the truth in the relationship between population and economic

113

12:27:16:11:09 Page 113 Page 114

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

growth in the Northeast Asian countries? Since I lack a thorough knowledge of the several millienia-long history of this region, I will limit my comments to Tokugawa Japan (1603–1868) and Ch’ing China (1644–1912).

TOKUGAWA JAPAN The population of Tokugawa Japan as a whole probably tripled, but it increased after many ups and downs during that 265 year period and evinced marked regional diversities.5 Roughly speaking, the 17th century was one of rapid population growth; the 18th, one of so-called “stagnation”; and the 19th, one of gently sloping increase. From fragmentary local records, I have calculated that the national population in 1600 might have been 12 million plus or minus 2 million. We know from the population surveys conducted by several territorial lords (daimyo) and regional studies in 17th century rural Japan, that the population increased at a high annual rate of 0.5 to 0.8 percent. It may be rash to conclude that these results indicate the population trend of Japan as a whole because this increase resulted from the establishment of the small family unit in rural society during the 17th century, which was accompanied by a high married proportion compared to the situation in the preceding period when there was a low proportion and when non-married subordinate males and females were the main resource of labor power.6 But I want to stress that Japan obtained the potential for an increase in population because of the emergence of this small family farming system: thereafter the population could increase at any time as long as there was no need or force to check it. In the 18th century, national population surveys were conducted by the central government (shogunate) every 6 years from 1721 to 1846, and these provide the demographer with unusually good data. Japan’s total population in the years covered by these surveys did not vary above 28 million or below 24 million, it is true. But we should note that these surveys contained serious underestimations. They excluded the warrior class population and the child populations in some domains. Though we have but imperfect supplementary data, an underestimation of 5 million is by no means unrealistic, and when this figure is added, we obtain a population of 29 to 33 million in this century. Fortunately several surveys recording population province by province (there were 68 provinces in total) were carried out, and these permit us to discover regional population trends in this century.7 These trends show clear regional differences. In the northeast, the population declined by 30 percent, but in the southwest, it increased considerably. It is very interesting that in the most economically advanced areas, that surrounding Edo and that surrounding Kyoto and Osaka, the regional population almost stagnated. Village data in these regions show that rural areas had a natural rate of increase; that is, higher fertility and lower mortality. The very economic development of urban areas reduced populations there through lower fertility and higher mortality. In short, the negative feed-back function between economic-urban development and regional population trends in pre-industrial societies noted by E.A. Wrigley was evident.8 In rural areas, almost all daughters and sons who did not or could not succeed to the family headship left for urban areas to find work. Some settled there

114

12:27:16:11:09 Page 114 Page 115

POPULATION GROWTH IN PRE-INDUSTRIAL JAPAN

permanently; some died at a rather early age due to the unhealthy conditions of pre-industrial town life. Even those who returned to the rural area could get married at a later age. Huge streams of immigrants from rural areas were needed if urban populations were to be maintained because many persons returned to their native or neighbouring villages after working for a decade or so in the towns. Thus the population in urbanized areas could be controlled by economic development. In northeast Japan, however, long-term climate change affected harvests so badly that people had to check the size of population to what permitted them to maintain their living standard. Abortion and infanticide were introduced, and the rural population decreased drastically. But we should notice that in spite of frequent bad harvests in this area, the urban population was growing. This tells us that even under such conditions, the economic activities did not shrink but could develop considerably. In southern Japan, on the contrary, population continued to grow but at a reduced rate, 0.2–0.3 percent. In this area, climate change did not affect the harvest at all, and people could enjoy a comparative advantage from cultivating and then developing various rural industries and by-employments. Population growth became slower than that of earlier years through late marriage and migration to the urban area. People now came to realize that population limitation could bring them a higher standard of living.9 In the 19th century, Japan’s population increased again. Although this increase was interrupted by severe epidemics between 1837 and 1861, the growing trend was steady and irreversible. If we omit the influence of the short term population decline in these years, the growth rate would be 0.3 or 0.4 percent per year. And this growth was undoubtedly a prelude to Japan’s modern population growth which began in the 1890s with about 1 percent, continued to the 1970s. Through recent studies of Japan’s proto-industrialization, we are sure that this growth was closely related to the development of cash crop cultivation, mainly cotton, rapeseed and sericulture; and rural industries, mainly textile, food-processing and brewing. An economic analysis of a region where such development was remarkable, shows that about a half of the income of a peasant family consisted of by- employment to these industries or the tertiary sector.10 And since these incomes were free from taxation, their net value must be specially high. These favorable economic conditions stimulated the population to grow, particularly in the area where it had stagnated in the preceding period.

AGRICULTURAL DEVELOPMENT In the 17th century, the development of agriculture in Japan was achieved mainly in extensive phases. Land reclamation, construction of irrigation systems and other infrastructures were pushed on by both lords and peasants. We have no accurate ﬁgure, but it is not unlikely that during the ﬁrst 100 years of this period, the cultivated area might have doubled. This development was initiated by a “great transformation” in the 17th century,11 which was introduced by rapid urban growth caused through the policy of sudden building up of 200 castle towns where the warrior class, merchants and craftsmen dwelt. Because their necessities of daily life could be supported through the market,

115

12:27:16:11:09 Page 115 Page 116

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

these towns had an effective demand to stimulate the agricultural production in rural areas. At first, peasants would try to expand their cultivated land, and to some extent, this was maximised by the time when diminishing returns set in. With varied time lag, the ceiling was actualized within 100 years, because Japan did not have enough arable land and the population density was already rather high. Peasants came to realize that they had to change the direction of development from land-extensive to land- and labor-intensive. The 18th century in Japan produced a crisis because the balance of population and land deteriorated and in the northern part of Japan a “little ice age” came around. The agricultural development clearly changed its nature to use land more effectively and use more human labor. The two crop system, rice as summer cropping and other cereals as winter cropping, became common where it could be done. This threatened land productivity so that the peasants had to use much fertilizer. The traditional fertilizers, grasses and leaves or heap, called “green fertilizer”, had been obtained by peasants by their own labor in uncultivated common lands, but as a greater part of these lands had now been allocated to the peasants for cultivation, they faced difficulties in obtaining fertilizer even though the demand for it became crucial. The solution was found by the introduction of the “cash fertilizer”; dried fish and lees and grounds.12 These were not peasant’s products, but of another occupation group and reached the peasants through merchants. But at any rate, one of the remarkable characteristics of Japanese agriculture, heavy fertilizing, started in such a way. Another solution was vertical land use, deep plowing, but without a plow. Compared with the European plow, the traditional plow in Japanese agriculture was small and the draught animals were poor as well. But in the rural areas there had been plenty of horses and cattle before the end of the 17th century; however after that, their number obviously declined. Instead of a plow drawn by livestock, a hoe or spade using human labor became the main plowing tool. This means that the labor that had been carried out earlier by horse-power now came to be done by man-power. Heavy fertilizing forced peasants to work on weeding, otherwise the land productivity distinctly declined. Thus the peasants in later Tokugawa Japan came to work very hard. I think the contemporary Japanese “work horrific” nature can be explained by this historical transformation. But, in another word, this is the cost of an “emancipation” of peasants from their former miserable status. Now their family formation was actualized, their decision-making in agricultural production enabled them to make money, if it was scarce. Hard working grew into an ethic and was transmitted through a mundane channel: family. I have termed this transformation an “industrious revolution”,13 which, in contrast with the industrial revolution taking place on the opposite side of the Eurasian Continent, did not reduce labor in human activity but raised it. But Japan could only narrowly escape from the crisis of the 18th century through hard working. The population of Tokugawa Japan, in the 18th century, did not stagnate at all. We must note differences according to region, and say that a “stagnation”, which many scholars have mentioned, occurred for the population as a whole quite accidentally. There was no force maintaining the nation’s population at a zero

116

12:27:16:11:09 Page 116 Page 117

POPULATION GROWTH IN PRE-INDUSTRIAL JAPAN

growth rate in the 18th century. In the most economically advanced areas, rural families may have held down family size to an optimum level by artificial means, but in the north, people were forced to limit the size in order to survive. And in the west, population continued to grow if at a moderate rate. This introduced us to an opinion that population stagnation was neither a necessary nor sufficient condition for economic development. This is speculative, but recognizing the historical reality of Japan’s 19th century economic development, I hold that moderate population increase, accompanied by an effective mobilization of resources, was the most crucial pre-condition for agricultural development and industrialization. No one has succeeded in determining quantitatively just what a “moderate” growth rate is, or even how to measure it in the societies with which we are concerned. To solve this fundamental problem in the process of modernization, we must collect the findings of many regional quantitative studies in which more sophisticated methods are applied.

CHINA AND JAPAN Since I lack a thorough knowledge of Chinese affairs, I will limit my comments to a very brief comparison of these two northeast Asian countries. In Ch’ing China, population as a whole almost quadrupled, even under more severe conditions than Japan, such as political instability, frequent famines and epidemics, and lower productivity and efficiency in agriculture and land use. There were sharp contrasts in the conditions of population and the economy between China and Japan. First, in Japan, international migration was strictly forbidden and, in fact, no international migration existed between 1639 and 1858. But in China, although international migration was prohibited in law as well, plenty of Chinese people went out from the mainland and settled Chinese towns elsewhere in Southeast Asia. Chinese population need not therefore have been limited within its domestic production. Second, in Japan, the inheritance custom was of various types,14 but to some extent, impartible inheritance was dominant, particularly in the warrior class and in the eastern part of the country. In China, influenced by Confucianism and the wider scale of land, partible inheritance custom was common.15 It was the highest virtue in the Chinese ethic that an elder lived surrounded by a great number of his descendants. Thus in China people did not realize population limitation and would not put it into force. In Japan, population pressure did exist, but its size was controlled through various ways if its meaning was not correctly recognized. Both countries had a relatively higher population density than that of European countries, but their demographic process in the time just before industrialization was totally different, and made them achieve modernization through a quite different avenue.

NOTES

* This paper was an introductory comment for the session on Asia. 1. Ester Boserup, The Conditions of Agricultural Growth, London, 1965. 2. B.H. Slicher van Bath, The Agrarian History of Western Europe, London, 1963.

117

12:27:16:11:09 Page 117 Page 118

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

3. Douglas C. North and R.P. Thomas, The Rise of the Western World, Cambridge, 1973. 4. R.G. Wilkinson, Poverty and Progress, London, 1973. 5. Akira Hayami, “Mouvements de longue durée et structures japonaises de la population à l’époque de Tokugawa”, Annales de Démographie Historique, Paris, 1971. 6. Akira Hayami and Nobuko Uchida, “Size of Household in a Japanese County throughout the Tokugawa Era”, in Peter Laslett (ed.), Household and Family in Past Times, London, 1972. 7. Akira Hayami, op. cit supra note 5. 8. E.A. Wrigley, Population and History, London, 1969. 9. Susan B. Hanley and Kozo Yamamura, Economic and Demographic Change in Preindustrial Japan, 1600–1868, Princeton, 1977. 10. Shunsaku Nishikawa, “Productivity, Subsistence, and By-Employment in the Mid-Nineteenth Century Choshu”, Explorations in Economic History, Vol. 15, N° 1, 1978. 11. Akira Hayami, “A Great Transformation”, Bonner Zeitschrift für Japanologie, Band 8, 1986. 12. Thomas C. Smith, The Agrarian Origins of Modern Japan, Stanford, 1959. 13. Akira Hayami, “Keizaishakai no seiritu to sono tokushitsu” (The Emergence of Economy Minded Society and Its Characteristics in Japan), in Shakai-Keizaishi-Gakkai (ed.), Atarashii edo-jidaishizo wo motomete (Toward a New Image of Edo Period), Tokyo, 1977. 14. Akira Hayami, “The Myth of Primogeniture and Impartible Inheritance in Tokugawa Japan”, Journal of Family History, Spring, 1983. 15. James Nakamura and Matao Miyamoto, “Social Structure and Population Change: A Compara- tive Study of Tokugawa Japan and Ch’ing China”, Economic Development and Cultural Change, Vol. 30, N° 2, 1982.

118

12:27:16:11:09 Page 118 Page 119

Paper delivered at International Symposium no. 6, International Research Center for Japanese Studies (eds.), Nature and Humankind on the Age of environmental Crisis, 1992, pp. 139–151.

8 Population and Family in Crisis: A Study of Northeastern Japan in the Late Eighteenth Century

INTRODUCTION AND POPULATION TRENDS n 1721, the Tokugawa government decided to carry out a population survey I throughout Japan, by collecting population reports that had been investigated by every local lord and magistrate in their respective administrative area. The second survey was conducted in 1726 and after that it was carried out every six years. The last survey was undertaken in 1846. Among these 22 surveys, 12 have the figures for each province (kuni), and 10 have the population in each province by sex. This was the first national population survey in Japan for which there is an available record. However, as with almost all pre-modern statistics, the population records were not at all accurate. There are several weakness including the lack of the warrior class population, the lack of children in certain local domains, and the intentional reporting of incorrect data. The national population was recorded as stagnating between 24 and 28 million, however, we now know that this number is an underestimate by about 5 million. Still these shortcomings can be ignored when we utilize it for drawing the population trends for each province or area. Dividing Japan into three, according to the survey the population of the northeast part was in decline, the central part was stagnant, and the south-west part was increasing. [Fig. 8.1] When we look at Japan in detail, the population trends of the 14 districts were as follows. On the Pacific side of Ou, Kanto, Kinai and surrounding Kinki districts, population had decreased, and in the Hokuriku, Chugoku, Shikoku and Kyushu districts, population had increased. The decreases in Ou and northern Kanto were due to bad weather, the resulting poor harvests continued thoughout the late eighteenth century. The decreases in southern Kanto, Kinai and surrounding Kinki districts was the result of the over-urbanization of these areas. In Kanto, Edo had a population of one million, and in Kinai, Kyoto’s and Osaka’s respective populations were 400 to 500 thousand. These “megalopolis” were the product of the developing economic activities that Tokugawa Japan experienced. But more importantly they also had their urban network, with the middle- and small-size urban communities that were located in the same area.

119

12:27:16:11:09 Page 119 Page 120

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

Figure 8.1 Population Trend of Japan 1721–1846.

By pre-modern standards, such huge urban populations create an urban “graveyard” effect, or a negative feedback function between economic development and population growth in the area, so that population stagnated or even declined. Since the crowded urban conditions were very vulnerable to famine, epidemics and disasters, the natural growth rate for an urban population was negative before modern technology was adapted to urban living conditions. In the western part of Japan, on the contrary, where population growth was obvious, there were no huge cities and the urbanization ratio was low. There was no “graveyard.” The situation of the Ou district was similar to the western part, however, its population also declined. There must be some particular reason: the deterioration of the natural environment. In the 1750s and 1780s, the climate of northern Japan was unusual. Both the summer temperature and the hours of sunlight were insufficient for rice croping. In 1783, Mt. Asama erupted and the Kanto Plain was covered by volcanic ash. Agricultural production fell drastically, and coupled with the lack of food storage, ineffective administrative relief, and poor transportation led northern Japan to the great Tenmei famine. The severest ever recorded. The author does not know if these unusual climatic conditions were peculiar to Japan or not. In the seventeenth century, when Europe was hit by a climatic crisis, Japan and North-east Asia enjoyed very high population growth which had never previously occurred. Perhaps the natural environment was suitable for such a development. The population of Japan increased from 12 million to 30 million and was accompanied by a huge amount of land reclamation for cultivation, which resulted in the extension of agricultural undertakings into marginal land.

120

12:27:16:11:09 Page 120 Page 121

POPULATION AND FAMILY IN CRISIS

The eighteenth century crisis affected such marginal lands, particularly those in northern or high altitude parts of Japan. Under pax Tokugawa, everybody enjoyed prosperous economic development, and did not prepare for such a disaster. We can trace the population trends in Mutsu and Dewa provinces, which consisted of Ou district, and also in the territories of 6 local lords. In Mutsu and Dewa, [Fig. 8.2] population declined throughout the eighteenth century, though it began to recover in the 1790s. The sharp decline of 1830s was the result of epidemics and famine, but we will not examine it in this paper. Mutsu had the more serious decline, between 1721 and 1786 there was a 20% loss. In Dewa it was milder with a 8% loss, and its recovery was quicker than in Mutsu. Mutsu had the higher initial population, but Dewa’s population recovered the quicker and by 1822 it had returned to its initial population level. Because Mutsu is mostly on the Pacific coast and Dewa is on the Japan Sea coast, the crisis seems to be much serious on the Pacific side. Six local lords population records are available. [Fig. 8.3] Setting the population in 1750 at 100, we can see that all the population was declining in the eighteenth century, with particularly sharp declines in the 1750s and 1780s. All initial population levels were higher than the last recorded populations. At the same time there were local variations. Soma decreased most drastically, as the 1780s population was at 60% of the 1750s. It was located on the Pacific sea coast and we can imagine that the cold summer weather hit this area most severely. In contrast, the population of Yonezawa was not affected so seriously, and again we should mention that Yonezawa was located in the Japan sea side.

Figure 8.2 Population Trends of Mutsu and Dewa Provinces – 1721–1846.

121

12:27:16:11:09 Page 121 Page 122

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

Figure 8.3 Population Trends in 6 Daimyo Territories.

The population decrease in 1750s can be seen in every domain but Nihon- matsu. These population declines are attributed to the so-called “Horeki famine,” but the author would call attention to the fact that the decline occurred well before the 1750s and had continued for several decades. The “Horeki famine” hit during the long term trend of decrease and the subsequent “Tenmei famine” of 1780s visited here when the people were exhausted by a long series of continued disasters. The famine in the northern part of Japan takes place because of cool summers caused by cold and wet wind from the Okhotsk high pressure system. Temperature does not rise to the required degree for rice growing and the misty weather causes a shortage of sunshine. These phenomena are much harder on the Pacific coast. Next we will look at the population trends of an individual town and some villages in the Nihonmatsu domain. There are a very good series of population registers for the town of Koriyama-kamimachi, and the three villages of Shimomoriya, Hidenoyama and Niita. Koriyama was growing very rapidly into the local town. (The total population reached over 5,000 at the end of Tokugawa period.) Shimomoriya was located at the foot of high land and would probably the most directly affected by the bad climate. Hidenoyama was a neighboring village of Koriyama, and when Koriyama was urbanizing it became a suburban community. Niita was located in the flat area between Koriyama and the castle town Nihonmatsu, and thus had relatively good conditions for agriculture. Thus the natures of these four communities were different, but we can assume that these are representative of the Nihonmatsu domain. The population survey

122

12:27:16:11:09 Page 122 Page 123

POPULATION AND FAMILY IN CRISIS

began as early as the 1680s, compiled every year, and recording very minute items, perhaps one of the best population registers in Tokugawa Japan. Recently two monographic studies have been published for Shimomoriya and Niita. According to the population trend of the three villages, [Fig. 8.4], Shimomoriya and Niita were almost synchronized, but Hidenoyama declined in the early eighteenth century and recovered quickly. The most severe decline was in Shimomoriya, at its bottom in the late eighteenth century its population was arround two thirds of its 1740s peak. Niita was similar, but its population began to deline in the 1770s and continued to the 1820s, which was followed by a remarkable recovery that rapidly attained its population peak of the 1740s. Hidenoyama declined just before the “Horeki famine,” but after that the population grew continiously for 90 years. Thus the three villages population trends had diﬀerent shapes, but as we can assume them as representing the trends of population in the Nihonmatsu domain, we can combine these trends and compare them with that of Koriyama- kamimachi. [Fig. 8.5] It obviously clear that the population of Koriyama grew remarkably, apart from two drops in 1780s and 1830s. At the beginning of register in 1709, the population was 799, but in the last year of 1868, it reached 2,462; 3.1 times larger, on in other words, the annual growth rate was around 1%. On the other hand, in the three villages population continued to decline until the 1830s, which is the bottom of population trend of the period concerned. If we

Figure 8.4 Population Trends of 3 Villages (Shimo-moriya, Hidenoyama and Niita).

123

12:27:16:11:09 Page 123 Page 124

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

Figure 8.5 Population Trends. (Ko¯riyama-Kamimachi and 3 Villages).

compare the population trends with that of the Nihonmatsu domain, the trend of the three villages was almost similar. In this area, the ratio of urban population was not high, possibly under 10%, thus the population growth of the towns would not explicitly aﬀect the domain population trend. As the Nihonmatsu domain was located at the southernmost area of the Ou district, it was probably not representative of the population of Ou district. But, it is very remarkable that in spite of population decline as a whole, the urban population continued to increase. Such population increase would not occur if the economic activity stagnated.

POPULATION BY SEX AND SEX RATIO Age and sex are the fundamental components of population. However, we can refer only to sex here, because age structure must be discussed after a long time consuming project examining registers that is now in progress. Here I will investigate the trend of population by sex followed by sex ratio. The population surveys of the Tokugawa government left us 10 records of the results of population survey in each province by sex. In Mutsu and Dewa, the population of both sexes can be seen as follows. [Fig. 8.6] In both provinces, the diﬀerence of population by sex became closer towards the end of the survey. In Mutsu, where the population decline was sharp, the female population was rather stable and population size seems to be controlled through the male population. Especially in the crisis years, the decline of male population was

124

12:27:16:11:09 Page 124 Page 125

POPULATION AND FAMILY IN CRISIS

Figure 8.6 Population by Sex in Mutsu and Dewa.

much greater than female population decline. In Dewa, although the whole population almost stagnated, if anything there was an increase in the female population. Only for the Yonezawa domain do we have population trend by sex. During 170 years, the difference between male and female populations was also diminishing. [Fig. 8.7] The female population started to grow from 1750, much earlier than the post 1830 increase of the male population. In Yonezawa, there was no serious decline after 1750, so that its decline seems to have taken place without the influence of the environmental crisis. In the Nihonmatsu domain, Koriyama had a large initial male population, by the 1770s there was no difference between both sexes, and after 1810, female population overtook the male population. [Fig. 8.8] In the combined three villages [Fig. 8.9], the difference was also disappearing. But it was only in 1820s that both sexes reached the same level. In the crisis years of the 1750s, 1780s and 1830s, the decline of the male population was much larger than that of the female population, but since in other periods there were minimal differences there is no direct relationship between the crisis periods and the change in the population structure by sex. Regardless of population size, the component of population by sex can be

125

12:27:16:11:09 Page 125 Page 126

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

Figure 8.7 Population by Sex in Yonezawa-han Territory.

Figure 8.8 Population by Sex. (Ko¯riyama-Kamimachi).

described by the sex ratio. Here I express the sex ratio as the male index number when the female number is 100. Through the population survey conducted by Tokugawa government [Fig. 8.10], the sex ratio trend of Mutsu and Dewa paral- leled each other beautifully. In the domain population, the case of Yonezawa

126

12:27:16:11:09 Page 126 Page 127

POPULATION AND FAMILY IN CRISIS

Figure 8.9 Population by Sex in Three Villages.

[Fig. 8.11] indicates a drop from 150 in the initial period to 110 in the mid-nineteenth century. When this is juxtaposed with the provincial trend, they indicate a splendid synchronization. This tendency must be the common feature of the Ou district. The decline of sex ratio can be seen in Koriyama and the three surrounding villages. It parallels Mutsu, but to a lower degree. In any rate, as previously indicated, there was a significant reduction in the population difference by sex, and a decline of sex ratio in this district. The female population was rather stable, but the male population experienced wider fluctuations. Several factors determine the decline of sex ratio. The first is the possible vanishing of the under-registration of the female population. However, according to a careful investigation of the population register in Nihonmatsu domain, there was no such under-registration. If such kind of under-registration was usual, there has to be numerous new entries of female population in the register that have no clear reason, but almost all entries appear to have peculiar reasons. We can discard under-registration as an explantation of the decline of sex ratio in this domain. Secondly, there is the decline of female mortality or extension of life expectancy. There might be a differential mortality or life expectancy between the sexes,

127

12:27:16:11:09 Page 127 Page 128

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

Figure 8.10 Sex Ratio in Mutsu and Dewa 1721–1846.

but before we can reach such a conclusion, we must calculate the mortality and life expectancy by sex from a fairly large slice of the population. The author is now undertaking this work using the population registers in Nihonmatsu and other domains in this district. At the moment, the differential mortality and life expectancy rate appear to be fairly small, and if they had any function their effect on the sex ratio would be slight. But this is a very important issue because it is connected with the changing position of females in society. Thirdly, female infanticide was practised in Tokugawa Japan as a means of population control. Much has been spoken about this and the author does not deny its existence. However, there should be some reservations. Through the investigation of the registers of Nihonmatsu domain, the sex ratio at birth was rather normal: 102.7 in Shimomoriya and 109.2 in Niita. This contradicts the myth of the custom of female infanticide. If infanticide was undertaken, it was not done sex-selectively. Lastly, we should take note of migration. If the difference by sex in out- migration and in-migration is large enough to affect the existing sex ratio, migration can be a factor in dividing the sex ratio. But it seems that in terms of population drain there was no large sex bias. However, we should wait for the result of the investigation of the registers. It is true that a considerably large number of the male population once went to Edo or other urban districts from the

128

12:27:16:11:09 Page 128 Page 129

POPULATION AND FAMILY IN CRISIS

Figure 8.11 Sex Ratio in Yonezawa-han Territory.

villages of the Nihonmatsu domain, but at least half of them later returned to the village: Nihonmatsu domain was located with easy access to Edo. At the moment, with the ﬁrst reason already discarded, we have not found any decisive reason to explain the changing population by sex or sex ratio. As a whole, in Tokugawa Japan, in the middle of the eighteenth century, the sex ratio was very high in the nothern and westernmost areas, but it was normalizing during the late eighteenth and early nineteenth centuries. Perhaps a combination of the second factor and a decline in the third factor would push the ratio down to the normal rate.

CRUDE DEATH RATE The fundamental factors to determine population dynamics are birth, death and migration. Among them, birth and migration in the concerned districts can be only obtained through the investigation of the registers. Here we will observe death by the year end record of each register, and through them we will calcuate a crude death rate (hereafter simply expressed as “death rate”). The accurate rate can be calculated only after a careful study of the registers. The death rate in Koriyama and the three villages [Fig. 8.12] can be seen on the graph, calculated on a ten-year moving average. These ﬁgures are far too low to

129

12:27:16:11:09 Page 129 Page 130

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

Figure 8.12 Crude Death Rate.

accept as the mortality level for a pre-modern society. This occurs because of a weakness of the register which lacks infant deaths between birth and the first investigation. The number of infant deaths must have been high. A study which investigated pregnancy in the early-nineteenth century near to Nihonmatsu esti- nated that about 20% of babies died before the first register was complied after their birth. Using this study we should add 20% more deaths (and births) to the number on the register. The original figure can be estimated using the left hand scale, and the revised figure uses the right handscale. At any rate, there is a strong correspondence between the mortality in Koriyama and its surrounding villages. The degree of fluctuation is rather high, between 20 and 40‰, in spite of being averaged on a 10-year period. But the decades having low death rate came just after the decades having high rates. This enables us to assume that in the decades with high death rates which can be called mortality crisis years, the physically weak, the old and sick died, and in next decades their numbers would shrink. But again we badly need to obtain more accurate figures from further investigations of the register. In Koriyama, the average death rate in whole period was 32.2‰, and in the three villages 29.9‰. The sample size is as follows: the number of deaths in whole period for Kiriyama was 5,353, and for the three villages it was 3,739. Can we consider this slight difference as the “urban graveyard” effect? The answer is yes and no. Considering that the population of this town consisted of more age groups which had a low age specific mortality, even such a slight difference may

130

12:27:16:11:09 Page 130 Page 131

POPULATION AND FAMILY IN CRISIS

make sense. On the other hand, on the graph, there are several crossings of the rate between Koriyama and the villages, so that the author will reserve judgement on the decisive answer to this question. We should examine the age composition and age specific mortality in Koriyama and the three villages. The differential death rate by sex has no significance. The synchronization is impressive again. The residents of Koriyama and neighboring villages seemed to behave demographically in a similar fashion. The sex ratio of the death rate is generally higher for males than for females. Koriyama fluctuated in a wider range, and this might reflect the characteristics of an urbanizing population.

CONCLUSION These observations are just the starting remarks of a historical demography of the district now being undertaken by the author, and there can be no difinite conclusion. This same proviso applies to the relation with the change in environment. However, we can say that we should examine with great care the respective local data, because in the Ou district, there were numerous variations in population tendency, sex ratio, and so on. Even in the 1780s Tenmei famine, frequently regarded as a nationwide disaster, while some places were affected, some were not. Population in this district, generally speaking, declined in the late-eighteenth century. Still this contrasts with the urban population growth in a town such as Koriyama, a flourishing market town, in which the residents were engaged in numerous economic activities. Population did not increase as a result of the inflow of displaced persons. Considering such variation, the preconception we might have that the Ou district was backward and poverty-stricken should be discarded. Finally, this paper refers to just the southern half of Ou district. If the good records of the northern counterpart are found, it could alter the picture significantly.

BIBLIOGRAPHY 松枝茂『会津藩の人口政策』山一書房, 1943年. 福島県『福島県史第2巷通史編1 1 』 1971年. 吉田義信『置賜民衆生活史米沢藩の社会経洛史的研究』国書刊行会, 1973年 (初版1956年). 速水融「近世奥羽地方人口の史的研究序論」三田学会雑誌75巷9号, 1982年, pp.70–92. 成松佐葸子『近世東北農村の人びと奥州安積郡下守屋村』ミネルヴァ書房, 1985年. 成松佐葸子『江户時代の東北農村二本松藩仁井田村』同文館, 1992年.

131

12:27:16:11:09 Page 131 Page 132

First published in Japan in Transition from Tokugawa to Meiji (eds. M. Jansen and G. Rozman), pp. 280–317, Princeton, 1986.

9 Japan in Transition from Tokugawa to Meiji: Population Changes *

careful study of Japan’s population on the eve of industrialization will help A resolve the longstanding debate over which came first—population growth or modern economic development. It should shed light on the crucial problem of how the demographic indices of a premodern society, such as general population trends, birth and death rates, average age at marriage, and average life expectancy, can affect that society’s subsequent industrialization. Finally, it should even allow us to determine which indices contributed to industrialization, and to what extent, so that we can posit population as an independent variable against economic development. One of the essential steps toward these goals is to decide whether population increase stems from a rising birth rate or a falling death rate, which indicates whether or not the demographic transition theory holds true for a given society. Yet even this basic question has not been answered for Japan. Des- pite an obvious need to clarify the nature of Japan’s demography from bakumatsu 1 to mid-Meiji, on the eve of industrialization, we have surprisingly few such demographic studies. In this chapter I seek to show that by re-examining the data of the 1820–1850 and 1870–1910 periods, together with other evidence on the transition decades themselves, it is possible to clear up many uncertainties about population change in Japan. Japan is blessed with an unusually rich store of source materials for the historical study of its population: bakufu and han population surveys based on village or town religious-investigation registers (shumon aratame cho)2 allow us to obtain precise figures for such demographic indices as long-range population trends and changing patterns of birth, death, and marriage. Furthermore, the establishment of the local administrative system in the 1890s and the appearance of fairly reliable demographic statistics for the first decade of the twentieth century mean that good data, on both the regional and the national level, are available for part of the Meiji period. But we lack national population data for the period from 1846, when bakufu population surveys ended, to 1872, when Meiji surveys began, and there is confusion about demographic rates over the following decades. We do have a national population figure for 1872 (33.1 million) based on data obtained from the domicile registration system (koseki seido) established by the Meiji government. The early Meiji population, however, has been the subject of much controversy. In 1930 the 1872 figure was revised upward to 34.8 million by the Cabinet Statistics Bureau (Naikaku To¯keikyoku), which added the estimated number of persons missed in the registration process.3 Yet, doubts persisted and

132

12:27:16:11:09 Page 132 Page 133

JAPAN IN TRANSITION FROM TOKUGAWA TO MEIJI

re-estimates were made. The nation’s population in 1875 was estimated variously as 35.3 million by the Cabinet Statistics Bureau, 34.8 million by Akasaka Keiko, and 37.2 million by Okazaki Yo¯ichi.4 The differences are not very great in absolute numbers, but as Umemura Mataji has shown, they assume critical significance when we attempt to determine the rate of population growth during the era.5 This problem is central to the mid-1960s Okazaki-Umemura debate over early Meiji population growth rates, a debate well known among Japanese economic historians. Okazaki’s position was that since the early Meiji population was relatively large to begin with, its subsequent rate of increase should be interpreted as low. Umemura, on the other hand, while noting that his own view was still a “hypothesis,” opposed Okazaki’s assertion from the standpoint that population increase preceded modern economic growth. Because an understanding of this debate is crucial to the problem at hand, we must examine the positions adopted by these two scholars in some detail. First, Okazaki dismissed the birth and death rates published in the early Meiji period as being incredibly low. Instead, he used the first modern census of 1920 to project life tables backward and estimate birth, death, and growth rates for each time frame within the Meiji period. Using these tables, he estimated that Japan’s population was 36.3 million in 1870 and 37.2 million in 1875. His results indicated that the Meiji birth rate remained at the level of thirty-five to thirty-seven per thousand, that it did not decline until after the turn of the century, and that the death rate fell almost continuously from the figure of thirty-one per thousand recorded at the start of the period. In short, Okazaki found that Meiji population increases stemmed solely from a decline in the death rate. Akasaka’s estimates, used by Umemura, were also derived from the 1920 census. Akasaka had projected these figures backward to obtain vital statistics, and her results were similar to the figures obtained in 1930 by the Cabinet Statistics Bureau—34.8 million in 1875. This figure was 2.4 million, or about 7 percent, lower than Okazaki’s. But a more important difference was that, according to Akasaka, the birth rate jumped greatly, from twenty-six per thousand to thirty- five per thousand. These figures argued that Meiji population increases should be attributed solely to a rise in the birth rate. Umemura noted that Okazaki’s findings conformed to the demographic transition theory, and he suspected that this conformity was more than coincidental. According to the demographic transition theory, a nation’s population changes during modernization by stages: high birth rates and high death rates, then high birth rates and low death rates, and finally low birth rates and low death rates. Umemura argued that this theory was unten- able, citing the example of the Scandinavian countries as well as Akasaka’s figures for Meiji Japan. The astounding increase in Meiji Japan’s birth rate noted by Akasaka, he asserted, could not be left unaccounted for. The more important and relevant of Umemura’s findings are as follows. First, he held that the decline in the death rate in an industrializing society often reflects an income level that rises owing to endogenous factors—the increased scope and wider distribution of that society’s aggregate production. Thus, rising income levels lead to increased personal expenditures and then to a declining death rate. Population growth, when it is achieved in this manner, conforms to the theory

133

12:27:16:11:09 Page 133 Page 134

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

that economic growth produces population increase. But—and this is of immense importance—a declining death rate does not always stem from endogenous economic betterment. In much of the Third World today, the death rate has been lowered by importing advanced technology and hygiene already developed in the industrialized nations, and this exogenously induced decline in the death rate actually obstructs economic growth and improvement of living standards. In regard to the birth rate, Umemura wrote that so long as the death rate remains high, a desire to ensure the survival of offspring will prevent people from taking steps to reduce the birth rate to a level below that of the death rate. But after the death rate has fallen and people no longer feel a compelling need to ensure the survival of offspring, they will lower the number of births out of a desire to raise their standard of living. Moreover, when income levels are on the rise and seem likely to continue to do so indefinitely, forces that once held down the birth rate become relaxed, and an increasing birth rate may accompany or even precede modern economic growth. Of course, Umemura’s views presuppose that the increase in modern Japan’s birth rate began in the bakumatsu and Restoration era, prior to large-scale industrialization or economic growth. Some form of the theory that population increase produces economic growth, he hypothesized, may hold true for Japan, where factors during the Tokugawa period probably had worked to raise the birth rate. Umemura’s hypothesis has been neither confirmed nor refuted by later Japanese scholars studying the historical demography of Japan. For one thing, intensive research on the Tokugawa period can be conducted only at the village-survey level. Second, vital statistics for the early Meiji period are unreliable. And third, there is a dearth of economic statistics in contrast to the wealth of population statistics available. Even so, we who specialize in Tokugawa economic and demographic history look with envy upon the national population data available for the early Meiji era, since the bakufu surveys upon which we must rely are less reliable as quantitative sources. For example: population figures for samurai households are omitted; statistics for certain domains or provinces do not contain child populations; and domain surveys were not conducted in a uniform, consistent manner. Bakufu surveys yield figures that are significantly lower than actual population levels, but we do not know how much lower;6 nor do we have firm evidence to determine whether or not underreporting continued at the same rate and in the same numbers throughout the Tokugawa. Hence, these imperfect source materials pose many pitfalls. Regardless of such pitfalls, we continue to use bakufu survey data to plot Tokugawa demographic trends.7 Indeed, these surveys are the only quantitative source materials available that indicate national population levels and break them down province by province.8 Without them, our understanding of Tokugawa society certainly would be much more limited than it is now. Thus, the early Meiji population data that have been virtually ignored (particularly by demographers) for being statistically flawed appear, to Tokugawa specialists, as a hitherto undiscovered Treasure Island awaiting exploitation. Their value in clarifying the nature of that momentous transition from Tokugawa to Meiji must be immense.

134

12:27:16:11:09 Page 134 Page 135

JAPAN IN TRANSITION FROM TOKUGAWA TO MEIJI

In the late Tokugawa and early Meiji, modernization had yet to be pursued in earnest; accordingly, population changes that accompanied it were still of minimal importance. This is not to ignore the rapid development of port cities, the appearance of export-oriented industrial areas centered on raw silk production, the opening of Japan to foreign trade and diplomatic intercourse, the changes in politics and military organization that accompanied the Meiji Restoration, and the modern means of transportation imported soon after that event. There have been many demographic changes produced by these developments, but prior to the Matsukata deflation policy of 1881 and the establishment of railroad networks and new political institutions in the early 1890s, Japan’s population composition and distribution were not drastically altered. True, the early Meiji government did introduce vaccination and other public health measures but many of these policies had already been initiated in the bakumatsu era, and their effects on Japan’s population were therefore gradual rather than sudden and disruptive. Thus, early Meiji population data, although they do not reflect late Tokugawa demographic conditions as accurately as modern statistics might, nonetheless present a picture that is remarkable clear—there was no radical disjunction between bakumatsu and early Meiji population structure and composition. Many of the demographic statistics published in the early Meiji period are given in units of the old Tokugawa-period “provinces” (kuni) as well as in the newly established “prefectures” (fu or ken).9 A problem with the prefectures of this period is that their boundaries changed frequently. On the other hand, although the old Tokugawa provinces officially ceased to exist as administrative units, they were clearly defined geographic entities. Even if statistics using these provinces as units were not published, we can still produce figures for them by first finding data for the subadministrative units of gun (counties) and ku (large cities or their wards) that constituted the provinces, and then adding these sub- totals together. This is possible because the gun units themselves did not change (except to be appended to, or detached from, one or another prefecture). It would be wonderful if we could treat early Meiji “province” figures as simple extensions of Tokugawa province figures, but so uncritical a use of source materials is not permissible. Tokugawa province statistics were compiled from han-gathered data, and the gathering process differed from domain to domain. For example, the 1846 figure for Satsuma, which contained a high proportion of unenumerated samurai, stood at 242,000. But the total for Meiji-period Satsuma “province” came to 549,000—a jump of 227 percent, which is utterly implausible even in light of the 23 percent increase registered for the nation as a whole during that same period.10 Yet, despite this shortcoming, a comparison of Tokugawa and Meiji general population trends is permitted: if the historian is not overly rigid in his attitude toward the limitations and imperfections found in early Meiji demographic data, they can be employed with great profit.11 Before discussing the results of my research, I should explain my basic assumption about the nature of the “population” that emerges from Tokugawa and Meiji statistics. As discussed above, bakufu statistics suffer from a variety of imperfections—omission of samurai households, non-standardized methods of census taking by the various domains, occasional double counting of individuals or groups, and so on. Absolute figures obtained from Tokugawa population surveys, then,

135

12:27:16:11:09 Page 135 Page 136

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

are not always reliable. To skirt the pitfalls involved, I adopted the following strategy. I have considered the population given for each province to be an index, and from the annual provincial or regional population, I sought to determine changes for every era under consideration. Assuming that there were no variations in survey methods within each domain throughout the Tokugawa period—and there is a good possibility that there were none—the figures derived should then be amenable to statistical treatment. As far as Meiji populations are concerned, the registered population after the 1872 registration (jinshin koseki) became the basis of my calculations, and I added or deleted changes in the register from that year until 1885. The allowances made for births, deaths, and entries (by marriage and adoption) into and deletions from the register are known as “legally domiciled population” (honseki jinko¯). Those who left their place of legally registered domicile, resided in one place for ninety days or more, and followed legal procedures to declare themselves to be living in a temporary domicile are known as “temporary-resident population” (kiryu¯ jinko¯). Finally, the “resident population” (genju¯ jinko¯) is calculated from the “legally domiciled” population: registered temporary incoming residents are added, and temporary outgoing residents subtracted, to obtain the “resident population.” This “resident population,” then, is a purely theoretical calculation and is different from the resident population counted in Japan’s truly modern censuses, which begin in 1920. In those censuses, the census takers actually met the people they counted.12 Yet the Meiji population counts derived by this method are still much more accurate than those available for the Tokugawa period. One noteworthy fact is that surveys conducted from domicile registers produce cumulative distortions in population statistics; in other words, population figures based on domicile registers become progressively more divorced from reality. In the early Meiji period, there may have been omissions in the registers, but the inaccuracies they produce are not blatant. Instead, our chief problem is how to determine the true size of the migrating population. The number of out-migrants should theoretically equal that of in-migrants—at least in the domestic sphere. But in almost all cases, the recorded population of the latter group was greater than that of the former. Many migrants terminated their domiciled residence in one area and re-migrated to another without duly reporting the fact. Because of this, and given the natural increase in the proportion of in-migrants that accompanied urbanization and rising population-mobility rates, it eventually becomes impossible to determine the resident population. In 1908, the difference between in- and out-migrating populations reached 2,420,000—about 4 percent of Japan’s total population. Earlier in the Meiji period the discrepancy was less serious: the first statistics available on a nationwide scale for migrating populations (in 1884) show that there were 230,000 more out-migrants than in-migrants, and this accounted for 0.7 percent of the national population. I have noted conspicuous flaws that skew statistical data derived from early Meiji domicile registers. Nevertheless, we must recognize that these and other Meiji data are amenable to quantitative analysis, and we should make full use of them to clarify conditions in nineteenth-century Japan.

136

12:27:16:11:09 Page 136 Page 137

JAPAN IN TRANSITION FROM TOKUGAWA TO MEIJI

THE DEBATE AMONG AMERICAN SCHOLARS National population statistics derived from bakufu population surveys, which are extant for the years 1721–1846, seem to provide quantitative evidence of population stagnation. Some have even described the situation as a Malthusian trap. In 1721, the population was 26.1 million; in 1846, 26.9 million. The highest figure recorded in these surveys is 27.2 million; the lowest, 24.9 million. We get a mean figure of 26.0 million for the period, plus or minus 1.2 million (or 4.6 percent). Scholars who emphasize the dark side of Tokugawa society, particularly those who consider population to be a dependent variable of the economy, are quick to interpret such population “stagnation” as resulting from economic stagnation. Then again, those who attribute a lack of Tokugawa population growth to abortion and infanticide resorted to by commoners find in these figures just what they are looking for—proof of the poverty and despair that plagued Japanese society in that “feudalistic” era. But are such perceptions accurate? Recently, certain American specialists on Tokugawa demographic and economic history have published important studies that shed light on these issues. In one such work, by Susan B. Hanley and Kozo Yamamura, it is argued that 1) because of relatively low birth rates in Tokugawa Japan, the concept of “demographic transition” is not applicable and 2) in most cases, death rates approximately equaled birth rates, resulting in a stable population. The authors found a slight increase in Japan’s population after the Tempo¯ mortality crisis of 1837–1838, an increase that continued throughout the Meiji period but did not begin with it. In addition, they saw “a remarkable similarity with pre- and early- industrial population trends in Europe and no similarity at all between Tokugawa Japan and the other nations of Asia today.”13 In another study, Thomas Smith put forth the following conclusions about one village located in central Japan: Both mortality and fertility . . . were low to moderate in Nakahara as compared to eighteenth-century European parishes. . . . [T]here can be little doubt that one of the reasons for low registered fertility was the practice of infanticide. . . . What is surprising is that the practice does not appear to have been primarily a response to poverty. . . . In short, it gives the impression of a kind of family planning.14 From the overall tone of Smith’s book, we may infer his position: 1) at least during the eighteenth century, steps taken by peasants to maintain families at a certain size held down population increase; 2) this presents a marked contrast to Ch’ing China, where the population is believed to have doubled between 1749 and 1819; and 3) this contrast may have much to do with Japan’s earlier and more rapid economic development. Smith notes: If population was held in check by deliberate controls over fertility, we should have in some respects a functional equivalent of the European marriage pattern [late marriage and high celibracy]. But if population was checked mainly by famine, as some historians believe, we should have reason to doubt the economic expansion and consequent rise in per capita income after 1721.15

137

12:27:16:11:09 Page 137 Page 138

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

James Nakamura and Matao Miyamoto also believe that in contrast to Ch’ing China, which experienced a tremendous population increase during roughly the same period, Tokugawa Japan’s basically “stagnant” population created relatively higher per capita income, paving the way for Japan to industrialize successfully and achieve high economic growth. Nakamura and Miyamoto stress differences in social and political institutions between Tokugawa Japan and Ch’ing China. For example, the Japanese ie (household) system “that tied the Japanese to a fixed vertical structure of mutual obligations and responsibilities and impartible inheritance” was in direct opposition to the family system of Ch’ing China. “In Japan,” they write, “this took the form of population control and long-run investment, both of which tended to cause per capita output to rise.”16 Of course these scholars do have differences of opinion. But they all agree that 1) the Tokugawa population did not increase; 2) this nonincrease resulted from the “rational” behavior of Japanese peasants who sought to maintain or improve their standard of living; and 3) these factors help explain Meiji Japan’s rapid and “successful” modernization, in contrast to that of late Ch’ing China. One American scholar, however, takes exception to these views. Carl Mosk, in a recent article, wrote:

They [Hanley and Smith] argue that peasant families in Tokugawa Japan deliberately kept family size small through infanticide to maintain a relatively comfortable standard of living. To the contrary, I suggest that fecundity was low and the chances of infant survival poor, so parents were forced to adopt strategies about the sex of the births they would permit to survive in addition to eliminating many weak and sickly oﬀspring soon after birth.17

Mosk estimated statistical data on food consumption and physical characteristics and discovered that, between 1874 and 1877, daily food consumption per individual was no more than 1,530 calories and 47.4 g` rams of protein, levels that subsequently rose along with per capita income. From these ﬁndings, he argued that “during the Tokugawa period desired fertility exceeded actual for the bulk of the population and this negative gap closed during the Meiji era.” Mosk’s statement contradicts the ﬁndings of Hanley and Yamamura, Smith, and Nakamura and Miyamoto. “The concept of a demographic transition is applicable to Japan,” Mosk suggested:

If we think of demographic transition theory as explaining why desired fertility falls from relatively high levels prior to and in the early period of modernization to low levels as modernization spreads, it is my contention that Japanese demographic history does not deviate from the main lines of demographic transition theory. . . . [T]he Japanese demographic transition experience conforms in its essential features to the experiences of Europe, North America, Australia, and New Zealand.18

Mosk’s challenge is clearly relevant to the Okazaki-Umemura debate on population trends in Japan. According to Mosk, the demographic transition theory once refuted by Umemura may hold true for the Tokugawa period after all—if we can certify that physical reasons made fertility lower than desired fertility and that this

138

12:27:16:11:09 Page 138 Page 139

JAPAN IN TRANSITION FROM TOKUGAWA TO MEIJI

low fertility stemmed from inadequate food consumption and malnutrition. (Of course Mosk did not deny that abortion and infanticide were practiced.) It now appears certain that new debates will emerge over why fertility was low in the Tokugawa period and what this fact means to the historian. Before those debates can be resolved, we must accumulate many regional studies—each based on reliable source materials that have been subjected to rigorous demographic and economic analysis, performed with the aim of determining the general standard of living and the actual steps taken to alter it, if any.19 Mind-boggling amounts of time, money, and energy will doubtless be required to complete this task, but I believe it can be done.

THE LATTER HALF OF TOKUGAWA Assuming that a useful purpose can be served now by viewing overall population trends before seeking out detailed demographic indices, I will look at provincial and regional data, derived mainly from bakufu and domain surveys in the Tokugawa period, and will attempt to link these figures to early Meiji population statistics for corresponding geographic areas.20 Bakufu population surveys, which are extant from 1721 to 1846, contain various noteworthy characteristics and problems. First, although absolute national population figures show stagnation, regional variation was pronounced: in some areas the population rose continuously, while in others it fell. These increases and decreases offset each other to give the overall appearance of “stagnation,” or a lack of aggregate population growth. Generally speaking, in the Kanto¯ and To¯hoku areas, population decline was the rule; in Kyu¯ shu¯ , Shikoku, and Chu¯ goku, increases predominated; and in central Japan, there was a slight population decrease in the Kinki region and an increase in Hokuriku. Map 9.1 shows that Tokugawa Japan’s “demographic center of gravity” moved across Lake Biwa from east to west, as eastern areas decreased in population and western areas rose. In 1721, the three northern Kanto¯ provinces of Ko¯zuke, Shimotsuke, and Hitachi as well as the six Sanyo¯ provinces of Bizen, Bitchu¯ , Bingo, Aki, Suo¯, and Nagato both had populations of 1.8 million. But by 1846, the Kanto¯ region had lost 0.5 million while the Sanyo¯ area had gained the same number, so that the total difference in population was roughly 1 million. Thus, the aggregate population count for these two areas shows an undeniable, yet purely accidental, form of “stagnation” that disguises the sharply contrasting regional population changes which actually took place. Tokugawa peasants were not tied to the land, notwithstanding what we are told in most high school textbooks: they were free to migrate as individuals, and they moved from village to city in surprisingly large numbers to look for work.21 How- ever, their geographical mobility was not limitless. Such movement was generally restricted to a hundred-mile radius of towns and cities within a specific region, such as Edo in the Kanto¯, or Kyoto and Osaka in the Kinki.22 Population increase in western Japan and population decline in eastern Japan just happened: these contrasting trends did not result from economic rationality—the migration of peasants in search of work—as asserted by Yamamura in his two-region model of Tokugawa Japan.23

139

12:27:16:11:09 Page 139 Page 140

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

Map 9.1 The Shifting Demographic Center of Gravity, 1720–1893 Redrawn from Nihon retto¯ retto¯ ni okeru jinko¯ bumpu no cho¯ki jikeiritsu bunseki (Tokyo: Shakai ko¯gaku kenkyu¯ sho, 1974), p. 125.

Instead, I propose the following explanation. In the eighteenth century, eastern Japan suﬀered from chronic bad weather which produced a long-term drop in agricultural production, but western Japan was largely unaﬀected.24 Owing to the high death rate in cities, which teemed with workers who had migrated from the depressed countryside, the Kanto¯ and Kinki regions (which included Edo, Kyoto, and Osaka) were subject to the negative-feedback function and their populations stagnated.

140

12:27:16:11:09 Page 140 Page 141

JAPAN IN TRANSITION FROM TOKUGAWA TO MEIJI

The long-term decline of northeastern Japan’s population in the eighteenth century can be understood better by examining the annual domain-population surveys of Yonezawa and Aizu han. As shown in Figure 9.1, these two domains enjoyed population growth throughout the seventeenth century; but from the start of the eighteenth, decline set in, and owing to the calamitous Ho¯reki and Temmei famines of 1756 and the 1780s, their populations fell precipitously to a level two-thirds that of their seventeenth-century peak. In Yonezawa han, the population had not quite recovered to its earlier peak level even by bakumatsu times, which means that this han contained more people in the late seventeenth century than in the mid-nineteenth. These facts suggest that the eighteenth-century decline in eastern Japan’s population was not a result of short-term famines—however devastating—during the 1750s and 1780s. Instead, the decrease was a long-term trend that was probably caused by chronic bad weather, which lowered agricultural output throughout the century. Although the population increase in western Japan during this same era was but 0.2 to 0.3 percent annually, it could not have occurred apart from the development of peasant by-employments, particularly in handicraft industries. One conspicuous example is Cho¯shu¯ han, where about half of the peasant’s income was earned from by-employments and nonagricultural activities, a fact that oﬀers persuasive evidence of economic development having produced population growth in this period and area.25 In sum, by discussing Japan’s population in the latter half of the Tokugawa period solely in terms of aggregate numbers, and then summarily labeling it “stagnant,” we preclude an appreciation of its complexities and regional variations. Only through more detailed studies that account for geographic diﬀer- ences and changes over time can we attain a meaningful historical understanding of “Tokugawa Japan” as a whole. A second problem has to do with the lack of bakufu survey statistics after 1846.26 If the bakufu had continued conducting national surveys as usual up to

Figure 9.1 Population Trends of Aizu Han and Yonezawa Han, 1680–1850 Data for Aizu han from Takahashi Bonsen, Nihon jinko¯ shi no kenkyu¯ (Tokyo: Sanyu¯ sha, 1941), pp. 208–217; for Yonezawa han, Yoshida Yoshinobu, Oitama minshu¯ seikatsu shi (Tokyo: Kokusho kanko¯kai, 1973), pp. 113–118.

141

12:27:16:11:09 Page 141 Page 142

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

the Restoration, population statistics for 1852, 1858, and 1864 would be available; but domestic and foreign crises left bakufu leaders with little time or inclination to conduct censuses. Since no further national population figures are available until 1872, when the new Meiji government instituted its registration system (jinshin koseki), we have a statistical blank space of twenty-six years—from 1846 to 1872. This mid-nineteenth-century era cannot be interpreted either as one of “stagnation” or as a simple continuation of eighteenth-century demographic trends. To begin with, as shown in Map 9.1, the nation’s “demographic center of gravity” during the period under consideration was furthest west in 1846, but by 1873 it had moved northeastward at considerable speed. Of course, Tokugawa and Meiji population statistics are not of uniform quality or accuracy, so the speed of this shift cannot be plotted with precision; but general trends can be determined. Clearly, Japan’s “demographic center of gravity” had been moving westward at a rapid pace prior to 1846; this speed slackened considerably in bakumatsu times, and by the beginning of Meiji, the center of gravity was moving eastward swiftly. We may conclude, then, that the eastward shift in Japan’s population distribution had already begun by the bakumatsu and Restoration era. Next, as shown in Figure 9.1, the population of Yonezawa han (in present-day Yamagata prefecture) began to rise after the start of the nineteenth century, and by bakumatsu times it had just about recovered the population losses incurred during the eighteenth century. Hence we need to reconsider whether or not the general trends of rising population in the west and falling population in the east hold true for the nineteenth century. Fortunately, national population statistics, though flawed in some respects, do exist for every sixth year during the twenty- four-year period from 1822 to 1846; by using these data in a discriminating manner, we gain valuable information with which to solve this problem. Finally, it should be noted that Japan suffered a disastrous mortality crisis triggered by nationwide epidemics in 1837 and 1838. Osaka, for example, lost 11 percent of its population during these two years. The disease in question remains unknown, but in the Owase area (of present day Mie prefecture) it evidently was accompanied by high fever and diarrhea.27 According to the death registers (kakocho¯) of a farm village near Hida-Takayama, seventy-three persons starved to death in 1837 alone; in addition, forty-nine children and twenty-three elderly persons died from the illness, and thirty-seven others died from acute infectious diseases. Infectious diseases took their greatest toll in lives during the sixth and seventh months of that year.28 The Tempo¯ mortality crisis, then, stemmed not just from famine-induced starvation, but also from infectious diseases—probably ailments of the intestinal tract and/or measles. Here again, a cause-and-effect relationship in history is hard to determine. Were the deaths from infectious disease caused by weakened resistance, the result of poor harvests that reduced food consumption and brought on malnutrition? Or did the poor harvests stem from widespread disease that was produced by some new, virulent strain of virus that sapped peasants of the strength to farm? The resolution of this problem will have to await identification of the culprit disease. In any case, the immensity of this Tempo¯ mortality crisis is beyond doubt. According to bakufu statistics, a 4.2 percent decline in Japan’s total population

142

12:27:16:11:09 Page 142 Page 143

JAPAN IN TRANSITION FROM TOKUGAWA TO MEIJI

occurred between 1834 and 1840. This was the second largest rate of decline for any six-year interval on record from 1721 to 1846; it was exceeded only during the 1780s. Between 1834 and 1840, only eight of Tokugawa Japan’s sixty-eight provinces registered a population increase, while twenty-seven showed a decrease of 5 percent or more. The area of most precipitous decline stretched from To¯hoku to Hokuriku, and traces of the decimation can be found in Meiji records. Statistics compiled in 1884 are the first for which quinquennial age structure is known.29 From these data we find that, in fifteen of the nation’s forty-three prefectures, the number of 45-to-49-year-olds30 (born between 1836 and 1840, the five-year interval in which the Tempo¯ mortality crisis occurred) was lower than the number of 50-to-54- year-olds (born in the preceding five-year interval, from 1831 to 1835). Further- more, those fifteen prefectures were all on the Japan Sea coast stretching from To¯hoku to Hokuriku. More detailed analysis reveals the following. Statistics in 1886, the first year for which the national one-year age structure is known,31 show that the age-49 population (born in the crisis year 1837) was 16 percent less than those age 50 (born in 1836). In theory, the higher up in the age structure, the smaller the population should be, since it normally is reduced to a greater degree through death by natural causes—barring large-scale out-migration from a certain area, of course. The above facts, then, mean either that there were few births in the crisis year of 1837 or that the infant death rate was extremely high. These Meiji data are many years removed from the crisis year; moreover, we must take into account regional variations in the adult death rate. Even so, regional statistics derived from bakufu surveys and Meiji source materials match beautifully. One more short-term population change remains to be discussed—the cholera epidemic that struck Japan right after the opening of the treaty ports. During the nineteenth century, there were several worldwide cholera pandemics; the third just happened to coincide with kaikoku (the opening of the country).32 The disease raged throughout Japan, starting at Nagasaki in 1858. Though there is no accurate record of how many people died or suffered from it, its imprint, like that of the Tempo¯ mortality crisis, was left indelibly on Meiji records. For example, there were relatively few births in 1861, when the pestilence hit its peak. The 1884 records giving quinquennial age structure show that, in forty-one out of forty- three prefectures, the number of 20-to-24-year-olds born in the five-year interval including 1861 was lower than the number of 25-to-29-year-olds born in the previous five-year interval (a national average of 93.2 percent). The one-year age- structure figures for 1886 reveal that the national population born in 1861 was 12 percent less than that born in 1860. Statistics show that the epidemic raged most furiously in the To¯kai, Kinki, Chu¯ goku, and Shikoku areas, and that the greatest loss of life occurred in the five central prefectures of Aichi, Gifu, Mie, Nara, and Wakayama. In those prefectures, the number of persons born in 1861 fell to less than 80 percent of those born in 1860. Thus, the mortality crises of the late 1830s and early 1860s were nationwide in scope and had brutal effects on Japan’s population. Figure 9.2 shows one-year age structures for males and females at the end of 1886, plots their age backward, and shows changes in the population for each structure. This graph indicates that the Tokugawa population declined sharply on four separate occasions: the Tempo¯

143

12:27:16:11:09 Page 143 Page 144

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

Figure 9.2 Age Composition in Japan, December 31, 1886

mortality crisis of 1837–1838; in 1846;33 in 1851; and from 1859 to 1862. It is quite conceivable that either the birth rate was extremely low, or the infant death rate was extremely high for these years. What we must keep in mind, however, is that such abnormal conditions were of short duration and were caused mainly by communicable diseases having little connection with national economic conditions. In sum, from the beginning of the nineteenth century down to the Restoration, exogenous short-term factors worked to hold down population increase on a few occasions; what is more, there took place pronounced regional variations in population change that cannot be ignored. The Tempo¯ mortality crisis and bakumatsu cholera epidemics were most devastating in areas where population was on the rise (except for To¯hoku); the bakumatsu epidemics, in particular, wrought their greatest havoc in Japan’s economically advanced regions. Before we can sort out the cause-and-eﬀect relationship between population growth and economic development in Tokugawa Japan, we must discount such short-term, exogenously induced negative eﬀects on population change. The lack of population surveys broken down by province for the bakumatsu era prevents

144

12:27:16:11:09 Page 144 Page 145

JAPAN IN TRANSITION FROM TOKUGAWA TO MEIJI

us from measuring the precise impact that the cholera epidemics had on Japan’s population, but existing statistics do shed some light on the Tempo¯ crisis. The twenty-four years from 1822 to 1846, for which surveys were taken every sixth year, can be divided up into four intervals of six years each. When we disregard the 1834–1840 interval (in which the Tempo¯ mortality crisis occurred) we obtain a picture of long-term, overall national population trends on a province-by-province basis. During the other three six-year intervals, the population increased 5.2 percent, and of the sixty-seven provinces for which reliable data exist,34 all but eleven showed an increase. Without a doubt, this means that Japan’s early-nineteenth-century population was growing. More detailed analysis of provincial figures for those three six-year intervals reveals that in thirty-five provinces the population showed some growth in all three intervals. Furthermore, among these thirty-five provinces, Echigo, Ho¯ki, Izumo, Bingo, Aki, and Sanuki enjoyed a population increase of more than 2 percent in all three intervals. Thus, in the first half of the nineteenth century, most of Japan’s provinces were enjoying steady population growth. Chu¯ goku, Shikoku, and northern Kyu¯ shu¯ , areas that enclose the Inland Sea, displayed continuous growth, as did certain provinces in the To¯hoku, Hokuriku, and To¯san regions. By way of contrast, Kanto¯ and Kinai—the core areas of Tokugawa Japan and the economically most advanced—failed to grow. (See Maps 9.2 and 9.3.) We may summarize the significance of these observations on late Tokugawa population as follows. In the first half of the nineteenth century, western Japan (particularly areas surrounding the Inland Sea) enjoyed steady population growth. Such growth, though to a lesser extent, also took place in other parts of the country—even in To¯hoku. Japan’s population was on the upswing, and this increase continued throughout the bakumatsu era; although temporarily checked by the Tempo¯ mortality crisis, late Tokugawa population growth soon resumed. Why was it that the areas around Edo, Kyoto, and Osaka—the nation’s heartland—enjoyed no large population increases, while western Japan and To¯hoku, whose population had been decimated in the eighteenth century, did? One reason may lie in the high death rate found in all preindustrial cities; but the real answer, I think, should be sought outside of population structure, in economic development. In the Inland Sea area, village industries and peasant by-employments were developing during this period, and sericulture and the silk- reeling industry were beginning to emerge in To¯hoku. But the Kanto¯ and Kinai, in contrast to these “developing” areas, were already “developed.” The divergent forms of population change that occurred in these two types of regions conform precisely to phenomena that characterize periods of proto-industrialization. Yet until we obtain detailed regional and, if possible, village-level studies, the question of whether population growth engendered economic development or vice versa will remain unsolved.

THE EARLY MEIJI The ﬁrst national population statistics available for post-Restoration Japan date from 1872, and are derived from the new Meiji government’s domicile registration system. From this time on, Japan’s population continued to rise each and

145

12:27:16:11:09 Page 145 Page 146

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

Map 9.2 Population Change in Japan, 1822–1846 (Except 1834–1840)

every year. The average annual growth rate was, it is true, less than 1.0 percent during the period under consideration, but it is nevertheless important to emphasize that national statistics show sustained growth in the Meiji population. Modern medicine and sanitation were implemented to prevent epidemics of typhoid fever and cholera, which had taken a great toll in lives during the Tokugawa

146

12:27:16:11:09 Page 146 Page 147

JAPAN IN TRANSITION FROM TOKUGAWA TO MEIJI

Map 9.3 Population Change in Japan, 1872–1885

period. Consequently, acute short-term fluctuations in Japan’s population were eliminated. (See Figure 9.3.) But when we examine regional population growth trends and the reasons for population change in early Meiji Japan, we find clear differences from the Tokugawa period. Map 9.3 indicates that southern To¯hoku, northern Kanto¯, Kai

147

12:27:16:11:09 Page 147 Page 148

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

Figure 9.3 Age Composition in Japan, December 31, 1908 Data from Naikaku-to¯keikyoku, Nihon zenkoku jinko¯ seitai to¯kei (1911).

and Shinano, Tokyo, and Osaka enjoyed the greatest population growth in early Meiji; yet most of these areas were not growing in the first half of the nineteenth century or in the latter half of the Tokugawa period. (See Map 9.2 for Tokugawa trends.) Thus, the nation’s population was definitely increasing during the early Meiji era, and the regions that spurted ahead to take the lead in this increase were those that had been “also-rans” during the late Tokugawa. Figure 9.4 shows the relationships between various types of population change from 1822 to 1846 and from 1872 to 1885 in fourteen regions. (Note, however, that I have omitted changes from 1834 to 1840 in order to minimize the effects of the Tempo¯ mortality crisis.) In late Tokugawa, we see that the population was rising in all regions except northern Kanto¯. Hokuriku and the side of the To¯hoku region along the Japan Sea showed the highest growth; To¯kai, Kinai, and the areas

148

12:27:16:11:09 Page 148 Page 149

JAPAN IN TRANSITION FROM TOKUGAWA TO MEIJI

Figure 9.4 Comparison of Population Change, by Region, 1822–1846 and 1872–1885 Key: A. East To¯hoku: Mutsu (in 1868 divided into Mutsu, Rikuchu¯ , Rikuzen, Iwaki, and Iwashiro) B. West To¯hoku: Dewa (in 1868 divided into Uzen and Ugo) C. North Kanto¯: Ko¯zuke, Shimotsuke, and Hitachi D. South Kanto¯: Musashi, Sagami, Kazusa, Shimoosa, and Awa E. Hokuriku: Sado, Echigo, Etchu¯ , Noto, Kaga, Echizen, and Wakasa F. To¯san: Kai, Shinano, and Hida G. To¯kai: Izu, Suruga, To¯to¯mi, Mikawa, Owari, and Mino H. Kinai: Yamashiro, Yamato, Kawachi, Izumi, and Settsu I. Peripheral Kinai: O¯¯ mi, Iga, Ise, Shin, Kii, Awaji, Harima, and Tanba J. San’in: Tango, Tajima, Inaba, Ho¯ki, Oki, Izumo, and Iwami K. Sanyo¯: Mimasaka, Bizen, Bitchu¯ , Bingo, Aki, Suwo¯, and Nagato L. Shikoku: Awa, Sanuki, Iyo, and Tosa M. North Kyu¯ shu¯ : Chikuzen, Chikugo, Hizen, Iki, Tsushima, Buzen, and Bungo N. South Kyu¯ shu¯ : Higo, Hyu¯ ga, Osumi, and Satsuma

surrounding Kinai had little growth. The figure for southern Kyu¯ shu¯ is surprisingly low, probably because of the dubious, misleading counting methods employed by Satsuma han: population growth in this region was most likely much greater. In early Meiji, eastern Japan (from Hokuriku and To¯kai eastward) recorded a higher growth rate than western Japan; and, surprisingly enough, the region of greatest increase was northern Kanto¯ which had been declining at the most precipitous rate in the nation during the latter half of the Tokugawa period. Next came southern Kanto¯ and the Pacific coast of To¯hoku. From 1872 to 1885, Japan’s greatest population growth occurred precisely in those areas where it had failed to increase significantly from 1822 to 1846. During the early Meiji era, the regions in eastern Japan that enjoyed the fastest growth in population were those in which sericulture, silk reeling, and textile industries were the most developed; the regions that showed the least growth were those in which rice was the most important single agricultural product. These two facts lead us to believe, as independent evidence confirms, that the terms of trade

149

12:27:16:11:09 Page 149 Page 150

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

shifted in favor of silk and against rice. In fact, the eight provinces that displayed the greatest population increase between 1872 and 1885 (over 18 percent) were Shimotsuke, Ko¯zuke, Musashi, Iwaki, Sagami, Settsu, Iwashiro, and Mino, in that order. Except for Settsu, which includes the metropolis of Osaka, all were located in eastern Japan; moreover, all were centers of sericulture. Kai and Shinano, the nation’s most important sericulture centers, both enjoyed more than a 15 percent increase. (See Map 9.3.) Hokuriku and the Japan Sea coast of To¯hoku became the areas of least population growth in eastern Japan. As for western Japan, population increase did continue in early Meiji, but at a low rate. Map 9.3 shows that except for Settsu (Osaka), no western province is included in Cluster I, which denotes a 17.5 percent rate of increase or higher. Furthermore, Cluster II contains only the four western provinces of Yamato, Kawachi, and Izumi (which surround Osaka) and Sanuki (in Shikoku). The low figure for southern Kyu¯ shu¯ may be the result of casualties suffered in the Satsuma Rebellion of 1877. In western Japan as well, Kinai, which had a rather low rate of population increase from 1822 to 1846, enjoyed the highest growth rate in early Meiji. This turnabout probably can be explained by improved urban living conditions in Kinai, one of the nation’s two highly urbanized areas. Improved hygiene no doubt lowered the high mortality rate that had afflicted this urban region in Tokugawa days. All of this means that Japan’s “demographic center of gravity” moved eastward during the Tokugawa-Meiji transition (see Map 9.1). Figure 9.4 shows “eastern” Japan as everything eastward from Hokuriku and To¯kai and “western” Japan as everything from Kinki westward. After the beginning of Meiji, eastern Japan’s rate of population increase outstripped that of western Japan. The relationship between rates of population change in the east for these two periods shows that regression is negative and that the correlation coefficient is quite high (r =−.745). By way of contrast, the dotting is scattered in the west, and the correlation coefficient is not at all significant. Our problem is to determine precisely when this “turnabout” in rates of increase between east and west took place and to discover what caused it. Japan’s national population began to rise after the start of the nineteenth century. As previously noted, if we disregard the short-term Tempo¯ mortality crisis, we find that the population in about 80 percent of the Tokugawa provinces was growing during the first half of that century. Later, in the bakumatsu era, there is strong evidence to indicate that an onslaught of cholera lowered the population for a short time, especially in central Japan. But pestilence did not halt the general trend of national population increase. The emergence of export-oriented industries such as raw silk, textiles, and tea meant increased opportunities for employment in the regions where those industries flourished, and this new economic development did much to alter demographic conditions in Japan. By emphasizing the trend of population growth in nineteenth-century Japan, we should not lose sight of the short-term negative factors which lessened that growth. We have already mentioned one of those negative factors—the cholera epidemics of the 1860s. But Figure 9.3 shows that another fairly sharp drop in the number of births occurred from 1869 to 1872. This latter decrease can be accounted for partly by undercounting, but a more important cause lies in the

150

12:27:16:11:09 Page 150 Page 151

JAPAN IN TRANSITION FROM TOKUGAWA TO MEIJI

warfare and social disorder that occurred during and immediately after the Restoration. Combined with the cholera epidemics of the same decade, then, two significant decreases in the number of births took place during this ten-year bakumatsu-Restoration era, which may mean that the national population fell for a short time in this period. By 1872 or 1873, the causes of this short-term decrease clearly were eliminated, and Japan’s sustained population growth began. Yet we must remember that the latent potential for this continued growth had existed from the 1820s. The Satsuma Rebellion may have affected regional population change in southern Kyu¯ shu¯ , but the number of casualties was relatively small when viewed on the level of national population statistics. Of greater consequence were the cholera epidemics of 1879 and 1886, for on both occasions more than 100,000 people died. By the late 1870s, however, the Meiji government was moving to eliminate cholera, infectious diseases of the intestinal tract, typhoid fever, and venereal disease, all of which had hitherto taken a great toll in lives. This stepped-up government program to eliminate disease and improve hygiene is overshadowed by its larger goal—to achieve “national wealth and strength”—and, indeed, some scholars interpret modern medicine and hygiene as having been sacrificed to that end.35 This is clearly untrue. Modern medicine and hygiene were essential means to attain that end, and the Meiji government actively sought to disseminate them throughout society. The police were mobilized to combat cholera, and vaccination against smallpox became obligatory, though not everywhere in Japan. Between 1874 and 1882, the construction of hospitals increased ten- fold.36 Medical schools were established at a fast clip. Of course, these measures did not prove effective overnight: it took time to lower the death rate for normal years, but infectious diseases and typhoid, which had decimated Tokugawa populations, virtually disappeared. Not counting the two cholera epidemics already mentioned, no plague resulted in 100,000 or more deaths. Such improvements in hygiene and medicine raised the reproductive capacities of urban populations, which had suffered from high death rates during the Tokugawa period: the negative feedback function between urbanization and population increase was definitely eliminated at some point in the early Meiji. Thus far, we have outlined the factors behind post-Restoration national population growth. We should now discuss the problem of regional variation, particularly with respect to geographic areas in which Meiji population trends represented turnabouts in relation to Tokugawa trends. In the first half of the Meiji period, the two areas that displayed the highest rate of population increase were 1) southern To¯hoku, northern Kanto¯, and To¯san, where the sericulture and silk-reeling industries flourished; and 2) the cities of Tokyo, Yokohama, Osaka, and Kobe, where modern industry and the military establishment produced population growth. The former regions had the highest growth rate for rural areas in Japan, and the latter regions, the highest for urban areas. By province, Shimotsuke, Ko¯zuke, Musashi, Iwaki, and Sagami led the nation in that order; and all of these provinces lay in the east. Moreover, the major centers of Japan’s export-oriented industries—sericulture, silk reeling, and tea—were also located in the east. Statistics for the twelve provinces of Iwaki, Iwashiro, Uzen, Ko¯zuke, Musashi, Sagami, Kaga, Kai, Shinano, Hida, Mino, and

151

12:27:16:11:09 Page 151 Page 152

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

Tajima show that in 1882 the per-villager production of raw silk and silk thread accounted for 20 percent of the total value of agricultural output. Except for Tajima, all twelve lay in the east.37 By comparing these eleven eastern silk-producing provinces and the eleven provinces that lay in the Sanyo¯ and Shikoku regions, we get markedly different pictures of population change. In nine of the eleven Sanyo¯ and Shikoku provinces, the value of ordinary agricultural products (rice and other foodstuffs) per person accounted for more than 75 percent of total agricultural output. Table 9.1 shows the rates of population change for the two regions between 1872 and 1885 as well as their respective per capita values for silk-related production and ordinary agricultural production (cereals and foodstuffs). We should keep in mind that the statistics for agricultural production used here are by no means highly reliable and that they do not include the villagers’ earnings from non-agricultural by- employments. Nevertheless, it is clear that in the silk-producing provinces, not only did the value of silk production account for a much greater percentage of

Table 9.1 Population Growth and Per Capita Agricultural Production, 1872–1885

Province Growth Rate Food Production Silk Production

% yen % yen

-  Iwaki 22.0 66.6 8.80 26.5 3.51 Iwashiro 18.3 54.7 9.63 41.0 7.22 Uzen 12.6 74.8 8.82 25.0 2.95 Ko¯zuke 24.8 31.7 6.83 65.4 14.08 Musashi 23.9 68.7 12.05 21.9 3.85 Sagami 19.3 70.5 8.04 23.0 2.63 Kaga 12.8 67.4 10.86 22.1 3.57 Kai 17.1 61.4 9.48 33.9 5.23 Shinano 15.0 53.4 9.30 42.6 7.43 Hida 14.6 66.3 19.44 32.6 9.56 Mino 18.0 67.0 6.67 23.0 2.29  18.0 62.0 32.5 (Average income per capita) (16.47) (9.99) (5.67)

O¯¯   Mimasaka 11.9 82.9 6.36 1.3 0.10 Bizen 7.5 78.5 10.95 0.3 0.05 Bitchu¯ 12.8 83.9 6.83 0.9 0.08 Bingo 14.2 77.8 6.61 0.2 0.02 Aki 13.9 76.5 6.59 0.1 0.01 Suo¯ 9.9 76.2 6.98 0.3 0.03 Nagato 8.5 93.0 8.52 0.2 0.02 Awa 11.9 65.1 7.49 3.0 0.35 Sanuki 15.3 74.3 9.12 –– Iyo 14.5 84.5 5.95 0.7 0.05 Tosa 4.8 81.0 7.57 1.2 0.12  11.4 79.5 0.7 (Average income per capita) (9.57) (7.54) (0.07)

152

12:27:16:11:09 Page 152 Page 153

JAPAN IN TRANSITION FROM TOKUGAWA TO MEIJI

total agricultural output than in Sanyo¯ and Shikoku, but the per capita value of food production was also higher than in Sanyo¯ and Shikoku. To a certain extent this contrast stems from the spread in food prices between these two regions, since interregional diﬀerences in rice prices were great during the 1880s. In the silk-producing provinces as a whole, the average price of rice per koku was 7.4 yen; in the six Kanto¯ and To¯san provinces, it was as high as 8.4 yen. But in Sanyo¯ and Shikoku, the price averaged 6.4 yen. In the Japan Sea coastal provinces, the price was roughly 4.2 yen per koku; in consuming areas of the Paciﬁc coast, it was 7 to 8 yen. Given these regional price variations, it seems that the higher cost of food products in the silk-producing regions was related not only to proximity to the Tokyo market, but also to increased demand caused by rapid population growth and by growing purchasing power. These investigations allow us to draw the following tentative conclusions. First, sometime between 1846 and 1872, the rate of population increase for eastern Japan surpassed that for western Japan. Second, this spurt in population growth rates was most pronounced in eastern silk-producing provinces.

A REGIONAL COMPARISON In this section we will compare the early Meiji demographic development of two selected areas that had displayed significantly different patterns of population change during the Tokugawa period, particularly in the eighteenth century. Of course, comparative analyses of this type must be conducted for many other areas as well, but limitations of time and space force us to focus on these two regions of marked contrast. One area, Ko¯zuke and Shimotsuke provinces, lies in northern Kanto¯ and con- stitutes the modern prefectures of Gumma and Tochigi; the other area, Bingo and Aki provinces, is located in the Sanyo¯ region and forms present-day Hiroshima prefecture.38 I chose these two areas because 1) both had roughly the same size population at the beginning of the nineteenth century—852,000 for Ko¯zuke- Shimotsuke and 889,000 for Bingo-Aki—according to the bakufu survey of 1804; 2) these eastern and western areas displayed contrasting patterns of population change in the Tokugawa and Meiji periods, as shown on Maps 9.2 and 9.3; and 3) provincial and prefectural boundaries coincide in both areas, which allows us to link their respective Tokugawa and Meiji statistics. Between 1721 and 1846, the population of Ko¯zuke-Shimotsuke fell a full 29 percent, from 1,130,000 to 807,000—an annual decrease of 0.2 percent. During the same period, in contrast, the population of Bingo-Aki rose 34 percent, from 682,000 to 915,000—an annual increase of more than 0.2 percent.39 As noted earlier, the population loss in Ko¯zuke-Shimotsuke, the area of greatest population decline in the nation, was caused by bad weather in the eighteenth century and by heavy out-migration to Edo. Even within the limited interval of 1822–1846, Ko¯zuke-Shimotsuke suffered a 5.3 percent population loss, while Bingo-Aki enjoyed an 8.9 percent gain. As revealed in Figure 9.5, the population of the BA Area was definitely on the upswing and, if not for the Tempo¯ mortality crisis of 1837–1838, would have grown 0.4 percent annually during this interval, which is a fairly high sustained growth rate for a preindustrial society.40

153

12:27:16:11:09 Page 153 Page 154

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

Figure 9.5 Population Trends in KS (Ko¯zuke-Shimotsuke) and BA (Bingo-Aki) Areas, 1822–1885

But from 1872 on, the wide spread in population figures for these two regions rapidly began to close.41 Between 1872 and 1885, Bingo-Aki registered a population increase of 14 percent, or 0.9 percent annually—which is by no means low, even when early Meiji underregistration is taken into consideration. On the other hand, Ko¯zuke-Shimotsuke recorded an astounding 25.6 percent increase, or 1.7 percent annually, during the same period. According to bakufu surveys, the Bingo-Aki population had overtaken that of Ko¯zuke-Shimotsuke sometime between 1804 and 1822; yet we now see that by 1885 the two areas had roughly the same number of inhabitants. And if figures for migrating populations (discussed below) are added, Ko¯zuke-Shimotsuke has clearly overtaken Bingo-Aki once again. Precisely when did this unusual population growth in Ko¯zuke-Shimotsuke begin? Statistics for 1885, the first year for which province-by-province quinquennial age structures are known, provide a clue, for they permit us to compare population age-structures in the two areas. (See Figure 9.6.) The decrease in 46-to-50-year-olds and 21-to-25-year-olds in Bingo-Aki indicates the damage inflicted by mortality crises in 1837–1838 and 1861. Although there are no pronounced drops in the Ko¯zuke-Shimotsuke age structures, the KS Area has higher (by 1.3 percent) component ratios than Bingo-Aki for 6-to-10-year-olds and 1-to-5-year-olds. If we assume that there was no great difference in child mortality rates between the two areas, this last discrepancy points to the emergence of a high birth rate in Ko¯zuke-Shimotsuke, beginning from about 1875. As mentioned earlier, provincial and prefectural vital statistics are available from 1877, but their reliability is by no means great. Figure 9.7 drawn up from those vital statistics, shows that although there clearly was a difference in crude birth rates between the two areas, their crude death rates were not far apart. But

154

12:27:16:11:09 Page 154 Page 155

JAPAN IN TRANSITION FROM TOKUGAWA TO MEIJI

Figure 9.6 Age Components in KS (Ko¯zuke-Shimotsuke) and BA (Bingo-Aki) Areas, 1885

these figures are inconceivably low, to be sure. To obtain more precise statistics on births, I compiled ratios of the number of females of childbearing age to the number of births for the respective areas, which also permitted me to address general fertility. Unfortunately, age differentials for 1880–1883 are different from those for 1884–1885, so I could not obtain consecutive statistics. Therefore, I sought to determine ratios of the number of females aged 20–49 to the number of births for the period 1880–1883, and of the number of females aged 20–39 to the number of births for 1884–1885. The averages for 1880–1883 were 0.158 for Ko¯zuke- Shimotsuke, 0.124 for Bingo-Aki, and 0.125 for the nation as a whole. Averages for 1884–1885 were 0.216 for Ko¯zuke-Shimotsuke, 0.177 for Bingo-Aki, and 0.178 for the nation as a whole. Thus we see that Bingo-Aki averages were virtually the same as national averages, but that those for Ko¯zuke-Shimotsuke were 26 and 21 percent higher, which shows that fertility clearly was high in the KS Area. One further factor behind population change—a social factor—should be discussed. This has to do with in-migrating and out-migrating populations as they appear in domicile registers. Needless to say, when one area has a greater number of register entries and in-migrants than it has register deletions and out-migrants, that area in theory should be taking people away from other areas. But in actuality, the number of out-migrants never equals that of in-migrants as they should when national totals are made from register counts. For example, on January 1, 1884, there were 520,000 in-migrants and 290,000 out-migrants. But there was little difference in the register entries and register deletions—both were about 110,000. We should not lose sight of this statistical shortcoming when assessing the social changes in population shown in Table 9.2. Note that there were more register entries and legally recognized in-migrants coming into

155

12:27:16:11:09 Page 155 Page 156

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

Figure 9.7 Crude Birth and Death Rates in KS (Ko¯zuke-Shimotsuke) and BA (Bingo-Aki) Areas, 1877–1885

Table 9.2 Population Indices for Ko¯zuke-Shimotsuke (KS) and Bingo-Aki (BA) Areas, 1884

KS Area BA Area

Legally domiciled population 1,246,151 1,262,561 Births 39,043 32,142 In-registered 5,338 2,325 Deaths 23,088 21,271 Out-registered 4,411 3,634 Total entries 45,417 36,227 Total deletions 27,830 25,326  + 17,587 + 10,901 In-migration 27,338 2,745 Out-migration 9,966 6,159  + 17,372 − 3,774 Resident population 1,263,523 1,258,787

N: All changes occurred in 1883. “Legally domiciled population” = honseki jinko¯; “resident population” = genju¯ jinko¯.

156

12:27:16:11:09 Page 156 Page 157

JAPAN IN TRANSITION FROM TOKUGAWA TO MEIJI

Ko¯zuke-Shimotsuke than there were persons leaving that area. For Bingo-Aki, on the other hand, out-migration was greater than in-migration. But we should note that the population turnover thus produced was not of significant size: it was less than the turnover produced by natural causes. Thus, there was not much difference between the “legally domiciled population” (honseki jinko¯) and the “resident population” (genju¯ jinko¯) in this period. Yet, as Table 9.2 shows, when we compare “resident population” figures for the two areas, we find that Ko¯zuke-Shimotsuke had already surpassed Bingo-Aki by the beginning of 1884. The reader no doubt has already guessed the causes of this rapid population increase in the Ko¯zuke-Shimotsuke area. According to agricultural output statistics for both areas in 1882, the per capita value of agricultural output in Ko¯zuke was 21.5 yen; in Shimotsuke, 10.6 yen.42 When we consider that the national average for this year was 11.9 yen (s.d. + 3.8 yen), we see that, in contrast, Ko¯zuke’s figure is extremely high. Moreover, the figure for Ko¯zuke, when broken down, shows that the value of ordinary agricultural production (mainly cereals and other foodstuffs) was 6.8 yen, or 32 percent of the total, while that for specialized agricultural production (raw materials for industry) was 14.7 yen, or 68 percent of the total. Of these specialized products, the value of silk thread and raw silk output was 14.1 yen, accounting for 65 percent of total agricultural output. No similar example can be found elsewhere in Japan. By way of contrast, in Bingo-Aki the average per capita value of agricultural output was 8.6 yen. Ordinary agricultural production accounted for 77 percent of total output, specialized agricultural production, 23 percent. Rice made up approximately 60 percent of the former production figure and comprised 49 percent of total agricultural production in the area. In Ko¯zuke, the price of rice was 8.6 yen per koku; in Aki, 6.1 yen. In Ko¯zuke, the amount of rice produced per farmer was 0.53 koku; in Aki, 0.69 koku; and on the national average, 0.99 koku. The value of rice production per farmer, however, was another matter, with Ko¯zuke showing 4.6 yen to Aki’s 4.3 yen. Figure 9.8 shows trends in sex ratios for the two areas.43 In 1750, the first year

Figure 9.8 Sex Ratio Trends in KS (Ko¯zuke-Shimotsuke) and BA (Bingo-Aki) Areas, 1750–1885

157

12:27:16:11:09 Page 157 Page 158

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

for which figures are available, the sex ratio for Ko¯zuke-Shimotsuke was 123.5 (females = 100), but by 1846 it had dropped sharply, to 106.5. On the other hand, the sex ratio for Bingo-Aki increased slightly, from 104.8 to 106.5, during the same period. In Ko¯zuke-Shimotsuke, population decline was combined with a lowering of the sex ratio: the male population fell to approximately 67 percent of its original size during this period; the decline in the female population was limited to about 20 percent. At present it is impossible to ascertain just what caused this decline. In general, the sex ratio for central Japan during the late Tokugawa period remained at a normal level, while in the Kanto¯ and To¯hoku regions, and in Shikoku and Kyu¯ shu¯ , it fell from a high to a normal level.44 No single factor can explain the changes that occurred in the Tokugawa sex ratio as well as the diverse population changes that took place simultaneously thoughout Japan. Rather, the answer lies in the changing position of women in society and in the family, in the rising value of women’s labor, and in the declining mortality rate for females during pregnancy and childbirth. These changes, moreover, took place according to regional time lags during the Tokugawa period—beginning and taking hold earliest in central Japan, and extending to remoter areas by the early eighteenth century. These changes then spread to all parts of the nation more or less evenly after the period when population data became available. Sex ratios in the early Meiji era for our two areas are as follows. In 1872, Ko¯zuke-Shimotsuke had a ratio of 101.9 and Bingo-Aki 104.1; the national average was 103.0. In 1885, the Ko¯zuke-Shimotsuke ratio was 100.0; Bingo-Aki’s was 104.9; and the national average was 102.4. During the twenty-six year statistical gap from 1846 to 1872, then, there was a turnabout in the sex ratios for the two areas. Figure 9.8 shows that in 1750 the sex ratios for the two areas were far apart—one far above and the other far below the national average. But like the original difference in population size between these two areas, this wide divergence in sex ratios gradually narrowed. By 1830, both were about the same as the national average, and by the early Meiji era, both were slightly below it. All of this indicates a turnabout in the relative standings held by the two areas with respect to sex ratios. Perhaps most impressive is the rapid decline of the sex ratio in Ko¯zuke-Shimotsuke after 1828. Most likely the sex ratio of surviving babies had become balanced a few decades before that year; or there may have been a lowering of the death rate in women (which had been high until 1828). In any case, something important must have occurred to lower the sex ratio. At this point, I can offer only the following observations. The sex ratio in Ko¯zuke-Shimotsuke became moderate completely apart from any change in population size. If we are to interpret this fact as resulting from the “rational” behavior of peasants, who made up the vast majority of Japan’s population, we must first come up with an explanation as to why the balancing of the sex ratio was a “rational” occurrence.

CONCLUSION We have come to see that Japan’s national population began increasing from the start of the nineteenth century and that, after the 1820s, this trend of stable growth held true for virtually all localities. The mortality crises of 1837–1838 and

158

12:27:16:11:09 Page 158 Page 159

JAPAN IN TRANSITION FROM TOKUGAWA TO MEIJI

1861 cannot be overlooked; yet, despite these temporary setbacks, the population grew at an annual rate of 0.3 to 0.4 percent. Regional variations did exist; but on the whole Japan’s population continued to rise during the statistical gap from 1846 to 1872. Toward the end of this quarter- century, those regions where the sericulture and silk-reeling industries had developed recorded an annual population growth on the order of 1 percent. This growth probably stemmed mainly from an increased birth rate; but a lowered sex ratio or, in other words, a higher percentage of females to males (beginning in the 1830s) may also have been a contributory factor. What do these findings tell us about the relationship between population and the economy, a problem introduced at the beginning of this chapter? No firm conclusions can be arrived at until many, many more empirical studies have been conducted; at this time, I can submit the following very tentative answer. During the Tokugawa period, there was moderate population growth in the nation’s most economically advanced region—the west—particularly in areas surrounding the Inland Sea, where opportunities were plentiful for nonagricultural by-employments in cottage industries. The Hokuriku area, where rice was cultivated successfully as a cash crop, also enjoyed moderate growth. The apparent “stagnation” in aggregate national population figures was a purely chance occurrence caused by exogenously induced population decreases in To¯hoku and northern Kanto¯ that offset gains in western Japan and Hokuriku. The nation’s most highly urbanized areas—southern Kanto¯ and the Kinai—failed to grow, or even declined slightly in population, owing to the negative feedback function. Considered together, these facts suggest that economic development and population increase in Tokugawa farming villages went hand in hand: population stagnancy was definitely not a precondition of economic development. I wish to reserve judgment as to whether or not the demographic transition theory applies to the period under examination, since the early Meiji birth rate did in fact rise in those regions where export-oriented industries developed. Even so, no proper conclusion can be drawn until we take into consideration the period after 1890, when industrialization began in earnest—and such an inquiry lies beyond the scope of this study. Japan’s population increase, which began at the start of the nineteenth century and picked up speed after the 1850s, stemmed mainly from a rising birth rate. Yet this does not necessarily mean that the death rate did not fall. In many of the villages that I have surveyed, mortality did in fact decline after the 1840s.45 However, empirical studies are too few to be able to draw any definite conclusion on this matter, and in any case, the short-term death rate jumped abnormally during the mortality crises of 1837–1838 and 1861. Japan was free from the horrors of pestilence only after the 1870s, when it imported modern science and hygiene from the West. Prior to that, Japan’s population increase was interrupted by exogenously induced crises on a few occasions, but the latent potential for increase surfaced undeniably in the nineteenth century. What phrase should we use to describe the pattern of population increase that took place in nineteenth-century Japan—an increase whose causes have yet to be identified fully? Western Japan’s population had been growing at approximately 0.2 percent annually since 1822; it did not begin increasing suddenly in the

159

12:27:16:11:09 Page 159 Page 160

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

nineteenth century. So it was for the nation as a whole. Japan’s national population was on the upswing during the late Tokugawa period; this increase only accelerated in the nineteenth century. I believe that the emergence of the stem family as the unit of agricultural production in Tokugawa Japan provided Japan with the potential for population growth. Therefore, growth was to be expected in the Tokugawa population; indeed, the absence of population increase would have required explanation. When we consider the low ratio of urban population and the impressive rural economic expansion that took place in areas where population increased, we cannot but conclude that moderate population growth in farming villages went hand in hand with economic development in this preindustrial society. During the statistically desolate quarter-century from 1846 to 1872, this combination of moderate population increase accompanied by economic development spread throughout Japan, spurred on by irreversible domestic commercial advances plus the appearance of export-oriented industries after the opening of the treaty ports. After 1872, this trend became the sustained population growth that continues to our day.

NOTES

* The author gratefully acknowledges the contribution made by Bob Tadashi Wakabayashi, not only for his patient translation work, but also for many helpful comments and suggestions made during the preparation of this chapter. 1. We shall here consider the bakumatsu era to be 1843–1868. 2. The bakufu conducted twenty-two national population surveys from 1721 to 1846, at six-year intervals. Of these surveys, twelve break down aggregate national figures province by province. 3. Naikaku to¯kei kyoku, ed., Meiji gonen iko¯ waga kuni no jinko¯ (Tokyo: Tokyo to¯kei kyo¯kai, 1930). 4. See Umemura Mataji, “Meijiki no jinko¯ seicho¯,” in Shakai keizai shi gakkai, ed., Keizai shi ni okeru jinko¯ (Tokyo: Keio¯ tsu¯ shin, 1969), pp. 118–141. 5. Ibid. See also Okazaki Yo¯ichi, “Meiji jidai no jinko¯: toku ni shussho¯ritsu to shibo¯ritsu ni tsuite,” Keizai kenkyu¯ , vol. 16, no. 3 (1965), pp. 207–213; Umemura Mataji, “Meiji jidai no jinko¯ ni tsuite: kommento” and Okazaki Yo¯ichi, “Umemura Mataji-shi no komento ni taisuru kaito¯,” both in Keizai kenkyu¯ , vol. 16, no. 4 (1965), pp. 356–359. 6. Sekiyama Naotaro¯ has gone over these materials with the utmost care, and probably for that reason they are known to foreign researchers as “the Sekiyama data.” His estimates, though not precise, place the unreported population count at between 4.5 million and 5 million. See Sekiyama Naotaro¯, Kinsei Nihon no jinko¯ ko¯zo¯ (Tokyo: Yoshikawa ko¯bunkan, 1969), p. 117. However, we should note that Sekiyama’s estimate was made by “logically linking” a national population figure based on the 1846 bakufu survey with early Meiji government statistics. These early Meiji statistics are themselves by no means reliable. Moreover, given the fluid social and political conditions that did so much to reduce bakufu and daimyo authority in the late Tokugawa period, Sekiyama’s estimate should not be accepted as an absolute of the highest accuracy. We must take into consideration regional variations in the source materials that yielded his figures as well as periodic fluctuations in them. 7. For example: Hayami Akira, “Tokugawa ko¯ki jinko¯ hendo¯ no chiikiteki tokusei,” Mita gakkai zasshi, vol. 64, no. 3 (1971); Umemura Mataji, “Tokugawa jidai no jinko¯ to keizai,” in Umemura et al., eds., Su¯ ryo¯ keizai shi ronshu¯ : 1, Nihon no keizai hatten (Tokyo: Nihon keizai shimbunsha, 1976), pp. 3–18; Susan B. Hanley and Kozo Yamamura, Economic and Demographic Change in Preindustrial Japan, 1600–1868 (Princeton: Princeton University Press, 1977), pp. 38–68. 8. Surveys by the bakufu and a few of the domains, often listing little more than population figures by sex, constitute the exception. 9. Published demographic statistics on the early Meiji are to be found in Naimusho¯, ed., Nihon zenkoku kosekihyo¯ (1872–1876) and Nihon zenkoku koko¯hyo¯ (1877–1878). These two works are available in reprint editions: Nihon to¯kei kyo¯kai, ed., To¯kei kosho shiriizu, vol. 4 (Tokyo: Nihon To¯kei Kyo¯kai, 1965); Naimusho¯ kosekikyoku, ed., Nihon zenkoku gun-ku bun jinko¯ hyo¯ (Tokyo: Naimusho¯ kosekikyoku, 1879) and Nihon zenkoku jinko¯ hyo¯ (Tokyo: Naimusho¯ kosekikyoku, 1880–1885).

160

12:27:16:11:09 Page 160 Page 161

JAPAN IN TRANSITION FROM TOKUGAWA TO MEIJI

We are able to use these early Meiji statistics with confidence owing to the splendid annotated bibliography compiled by Hosoya Shinji, Meiji zenki Nihon keizai to¯kei kaidai shoshi (Tokyo: Hitotsub- ashi daigaku keizai kenkyu¯ jo, 1974–1980). Professor Hosoya, of the Nihon Keizai To¯kei Bunken Center at Hitotsubashi University, has listed the legal basis, statistical genealogies, and contents of individual records. In addition, the So¯rifu To¯kei-kyoku has published (Tokyo: 1976) So¯rifu To¯keikyoku hyakunen shi shiryo¯ shu¯ sei; vol. 2, pt. 1, “Population,” lists laws, regulations, and other legal documents that bear on population studies. Thanks to these two works, we can approach early Meiji statistics in an orderly fashion and can utilize a particular set of data with secure knowledge of its relationship to other kinds of data. 10. See the “Kunibetsu jinko¯hyo¯,” in Sekiyama, Kinsei Nihon no jinko¯ ko¯zo¯, pp. 137–138. Between 1846 and 1872, there were extremely high jumps—150 percent or more—in the provinces of Kaga, Etchu¯ , Hizen, Hyu¯ ga, Osumi, Satsuma, and Tsushima. In these areas, populations were systematically under-reported in the Tokugawa period. 11. One promising avenue of future research is to use the gun or ku units (of which there were over 700) as a framework to set up manageable or appropriate-size populations for investigation in conjunction with other statistics, say of prices or production (though these are even less reliable than demographic data). Detailed studies of early Meiji demographic and economic history can be undertaken in this manner. 12. This is discussed by Osamu Saito¯, “Migration and the Labour Market in Japan, 1872–1920: A Regional Study,” Keio¯ Economic Studies, vol. 10, no. 2 (1973), p. 48. 13. Hanley and Yamamura, Economic and Demographic Change, p. 318. 14. Thomas C. Smith, Nakahara: Family Farming and Population in a Japanese Village, 1717–1830 (Stanford: Stanford University Press, 1977), p. 147. 15. Ibid., p. 5. 16. James I. Nakamura and Matao Miyamoto, “Social Structure and Population Change: A Compara- tive Study of Tokugawa Japan and Ch’ing China,” Economic Development and Cultural Change, vol. 30, no. 2 (1982), p. 262. 17. Carl Mosk, “The Decline of Marital Fertility in Japan,” Population Studies, vol. 33, no. 1 (1979), p. 37. 18. Carl Mosk, “Fecundity, Infanticide, and Food Consumption in Japan,” Explorations in Economic History, vol. 15, no. 1 (1978), pp. 269–289. 19. I have published a book-length regional study of Tokugawa population in Suwa county, Shinano province: Kinsei no¯son no rekishi jinko¯ gakuteki kenkyu¯ (Tokyo: To¯yo¯ keizai shuppansha, 1973). A similar study on Owari and Mino provinces is in preparation. 20. Province-by-province Tokugawa population figures are available for 1721, 1750, 1786, 1792, 1798, 1804, 1822, 1828, 1834, 1840, and 1846. The figures for 1792 are published in Minami Kazuo, “Kansei yonen no shokoku jinko¯ ni tsuite,” in Nihon rekishi, no. 432 (May 1984), pp. 42–47, and those for 1840 are published in Minami Kazuo, Bakumatsu Edo shakai no kenkyu¯ (Tokyo: Yoshikawa ko¯bunkan, 1978), pp. 164–185. Figures for the other years can be found in Sekiyama, Kinsei Nihon no jinko¯ ko¯zo¯, pp. 137–139. 21. Akira Hayami, “Labor Migration in a Pre-Industrial Society: A Study Tracing the Life Histories of the Inhabitants of a Village,” Keio¯ Economic Studies, vol. 10, no. 2 (1973), pp. 1–18. In the same issue, see also Susan B. Hanley, “Migration and Economic Change in Okayama during the Tokugawa Period,” pp. 19–36, and W. Mark Fruin, “Farm Family Migration: The Case of Echizen in the Nineteenth Century,” pp. 37–46. 22. For Edo, see Minami, Bakumatsu Edo shakai no kenkyu¯ ; for Kyoto, Hayami Akira, “Kyo¯to machikata no shumon aratamecho¯,” in Kenkyu¯ kiyo¯ Sho¯wa 55-nendo (Tokyo: Tokugawa rinsei shi kenkyu¯ sho, 1981). 23. Hanley and Yamamura, Economic and Demographic Change, pp. 28–37. 24. From 1732 to 1733, however, western Japan did suffer from famines caused by pests, which bred in massive numbers and inflicted damage on rice plants. 25. Thomas C. Smith, “Farm Family By-employments in Pre-Industrial Japan,” Journal of Economic History, vol. 29, no. 4 (December 1969), pp. 397–423; Shunsaku Nishikawa, “Productivity, Subsist- ence, and By-employment in Mid-Nineteenth-Century Cho¯shu¯ ,” in Explorations in Economic History, vol. 15, no. 1 (1978), pp. 69–83. 26. Prior to the first bakufu survey of 1721, Japan’s early Tokugawa population increase had already tapered off, but a discussion of that era lies beyond the scope of this study. 27. Hayami Akira, “Kishu¯ Owase-gumi no jinko¯ su¯ sei,” in Kenkyu¯ kiyo¯ Sho¯wa 43-nendo (Tokyo: Tokugawa rinsei shi kenkyu¯ sho, 1969). 28. Suda Keizo¯, Hida “O” jiin kakocho¯ no kenkyu¯ (private printing, 1973), pp. 154–155.

161

12:27:16:11:09 Page 161 Page 162

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

29. Naimusho¯ kosekikyoku, ed., Nihon zenkoku jinko¯hyo¯ (surveyed January 1, 1884; date of publication unknown). 30. Throughout this chapter “year-old” will refer to sai in the Japanese method of reckoning ages. 31. Naikaku to¯keikyoku, ed., Nihon teikoku daishichi to¯kei nenkan (surveyed December 31, 1886; published in 1888). 32. Tachikawa Sho¯zo¯, Byo¯ki no shakai shi (Tokyo: Nihon ho¯so¯ shuppan kyo¯kai, 1971), pp. 180–223. 33. This was the year hinoe-uma, considered to be an inauspicious year for females to be born. There- fore, the “population decline” observed in statistics for that year actually stems from nonregistration of female births and, presumably, infanticide. 34. This excludes Shimo¯sa, where records are poor. 35. Tachikawa, Byo¯ki no shakai shi. 36. Ibid. 37. On the value of agricultural production, see No¯musho¯ no¯mukyoku, ed., Meiji ju¯ gonen no¯sanhyo¯ (Tokyo: Meiji bunken shiryo¯ kanko¯kai, Meiji zenki sangyo¯ hattatsu shi shiryo¯, 1965 reprint of 1882 ed.), bessatsu 5. For village population, I have used Sambo¯ Hombu, ed., Kyo¯buseihyo¯ Meiji ju¯ ninen, 2 vols. (Kyoto: Yanagihara shoten, 1978 reprint of 1879 ed.). I considered every administrative unit of 5,000 persons or more to be an “urban area” or “city” (toshi) and subtracted these urban populations in my estimates for each province. 38. These areas are abbreviated “KS Area” and “BA Area,” respectively, in Figures 9.5–9.8 and Table 9.2. 39. On source materials for the Tokugawa period, see fn. 20, above. 40. Any direct linkage of the figures for 1846 with those for 1872 (see Figure 9.5) is purely speculative. To do this, I added the estimated unreported population to the 1846 figure, and arbitrarily set the death rate from the 1861 cholera epidemic at 5 percent. The estimates of unreported population for 1846 were derived from statistics compiled by status (mibun) in 1881; these amount to 4 and 5 percent, respectively, for the Ko¯zuke-Shimotsuke and Bingo-Aki areas. I also assumed that there was no change in their rates of population increase between 1846 and 1872 except for the year 1861. Of course, these presuppositions must be verified by future town- and village-level studies, but population age-structures (to be discussed below) lend credence to such assumptions. 41. On source materials for the Meiji period, see fn. 9, above. 42. See fn. 38, above. 43. Statistics for sex ratios derive from source materials listed in fn. 20, above. 44. See Hayami Akira, “Tokugawa ko¯ki jinko¯ no chiikiteki tokusei,” Mita gakkai zasshi, vol. 64, no. 8 (1971), pp. 67–80. Recently, Thomas C. Smith, Susan B. Hanley and Kozo Yamamura, and Carl C. Mosk have entered the debate on changes in the Tokugawa sex ratio. 45. In Yokouchi village, Shinano province, the crude death rate was 19.5 per thousand from 1826 to 1850, and 17.5 per thousand from 1851 to 1871. In Nishijo village, Mino province, the death rate fell from an average of 21 per thousand before 1830 to 16 per thousand after 1845–1850. See Akira Hayami, “Demographic Aspects of a Village in Tokugawa Japan,” in Paul Deprez, ed., Population and Economics (Winnipeg: University of Manitoba Press, 1970), pp. 109–125, as well as Hayami Akira, “No¯shu¯ Nishijo¯ mura no jinko¯ shiryo¯,” in Kenkyu¯ kiyo¯ Sho¯wa 47–nendo (Tokyo: Tokugawa rinsei shi kenkyu¯ sho, 1973). From just two village surveys, of course, nothing can be said about Tokugawa Japan as a whole. Scores of similar studies must be accumulated.

162

12:27:16:11:09 Page 162 Page 163

PART III DEMOGRAPHY THROUGH THE MICROSCOPE

12:27:16:11:09 Page 163 Page 164

12:27:16:11:09 Page 164 Page 165

Co-authored with L.L. Cornell and ﬁrst published in Journal of Family History, vol. 11, no. 4, pp. 311–328, 1986.

10 The Shu¯mon Aratame Cho¯: Japan’s Population Registers

POPULATION REGISTRATION IN JAPAN he history of local population documents in Japan can be divided into three Tperiods: prior to 1600, 1600 to 1870, and 1870 to the present. This paper discusses the character, quality, and availability of the shu¯ mon aratame cho¯, the local population registers of the middle period, but it is worthwhile to mention the others as well, since they indicate how rich the sources of demographic history are in Japan. The Japanese government first compiled population records in 702 A.D. Using a contemporary Chinese model, the government surveyed land and population in each administrative unit in order to redistribute land and requisition corvée labor. These registers contained such information as the area of distribution of arable land and the composition of the population differentiated by sex, age, and status. However, their coverage was limited to areas under the control of the central government, and only a few—forty cases scattered over a three-century period—survive. Almost all of them have been printed and published, and hence are available to all interested persons. They are poor sources of population data, because large numbers of men in the prime of life were omitted, resulting in skewed sex ratios and age structures. Despite this, these early population registers are a priceless source for understanding Japan’s agricultural families in the eighth to tenth centuries, and many scholars have made use of them (Kishi, 1973; Farris, 1985; Taeuber, 1958:5–11). After 1004 A.D., population documents—with the exception of a few upper-class genealogies—were completely nonexistent for the next 600 years. In the late sixteenth century, following the end of more than a century of civil war, both central and local authorities turned once again to land and population surveys for their administrative needs (Hayami, 1963), producing the ninbetsu aratame cho¯ (literally, “census investigation registers”). The earliest ones measured the amount of labor power a lord could commandeer from his domains, and consequently their contents included the number of households in a given com- mune, and the name, status, and age of each household’s members. Women and children were grossly underregistered. With the passage of time these records improved, and by the middle of the seventeenth century they included details such as the number of domestic animals, the construction of houses, and assess- ments of landholdings. While these registers can offer us invaluable information

165

12:27:16:11:09 Page 165 Page 166

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

about village life in seventeenth century Japan, they were compiled irregularly, and not throughout the entire country. In addition, by the beginning of the eighteenth century other kinds of documents superseded them and their compilation was abandoned. Turning to the modern period, we find that the new civil government after the Meiji Restoration of 1868 instituted a new population registration system as part of its general creation of a nationwide legal system. This, the koseki system, like its shu¯ mon cho¯ predecessor, used the household as the fundamental unit of registration. However, the requirement that an event be recorded in the koseki before it was recognized legally was a new departure in a society which previously had relied on customary law. It was many years before the system was completely implemented (Taeuber, 1958: passim). It remains the foundation of civil membership today. The koseki are closed to researchers because their earliest years include designation of outcaste status, information that continues to prejudice social relations. However, Brown (1978) possesses a fine data base which covers approximately 25,000 individual farmers, merchants, and samurai in a town in northeastern Japan for the period from 1872 to 1979. These data were collected before the ban on access was instituted. Modern population statistics date from the first national census in 1920.

DEVELOPMENT OF THE POPULATION REGISTRATION SYSTEM The European powers, advancing into Asia, arrived in Japan in the second half of the sixteenth century, just as domestic unification was beginning. The first to establish relations with Japan was Portugal. It did so just when the Society of Jesus, itself newly established, was beginning to take charge of the missionary efforts which became a principal motive for overseas expansion. The lords of western Japan anticipated profits from trade with Portugal, and the Jesuit missionaries—hearing that the people, exhausted after long-continued disturbances, were seeking spiritual foundations—expected a receptive population. Within a comparatively short time, a large number of people were converted to Christianity, and Japan’s “Christian Century” began (Boxer, 1951). By the end of the sixteenth century, however, the position of Christianity had become tenuous. The evangelical competition between the Franciscans, head- quartered in Manila and backed by the Spanish ruling family, and the Domini- cans alerted Japan’s rulers to the threat Christianity posed to national unification. The unifiers of the country had to choose to accept Christianity as the unifying national ideology, or to reject it. It took half a century to reach the final decision, but ultimately they chose to reject Christianity in its entirely, banish the missionaries, convert believers to Buddhism, and establish a system which severely pun- ished those who did not submit. The Tokugawa government, justifying its actions by reference to the desperate Christian rebellion in 1637, ended commercial and political relations with all European countries—except the Netherlands, whose objectives were thought to be purely commercial—and prohibited the return of all Japanese nationals and ships abroad. The proof of its thoroughness lies in the laws of the next year, 1638, establishing the investigations of religious sects, the shu¯ mon aratame. These laws made Christianity illegal and required each Japanese

166

12:27:16:11:09 Page 166 Page 167

THE SHU¯ MON ARATAME-CHO¯

to present to the local magistrate proof of membership in a Buddhist temple. Among the samurai, these investigations were carried out by rank, and among the peasants and townspeople, by the village or block in which they resided. These actions created inimitable sources for Japanese demographic history. At first, compilation of the shu¯ mon aratame cho¯ (literally “religious investigation registers”; hereafter, “population registers”) was confined to approximately the one quarter of the country under the direct control of the Tokugawa government. In 1665, however, the Tokugawa government ordered all lords to compile population registers and, in 1671, to compile them annually. While organized Christian resistance had disappeared thirty years earlier, there were still reports of Christians who only pretended to be Buddhists, and of Christians who rejected conversion entirely. The central government also perceived that it could strengthen its own control over the country by enforcing national religious investigations. Consequently, from 1671, all those living in Japan—whether samurai, farmers, or townspeople, whether resident in directly administered areas or not—were registered under this system every year. There were, however, numerous important exceptions. Tokugawa Japan was neither a centralized state nor an entirely decentralized one. The larger lords did not always accede to the orders of the central government and did not necessarily construct population registers according to the forms prescribed in domains under direct control. For example, population registers covering the masses of the people have never been found in either Satsuma (Sakai et. al., 1975) or Tosa. In Cho¯shu¯ , the oldest extant registers date from the nineteenth century. In Kishu¯ they were complied only once every six years—in the rat and horse years of the solar-lunar calendar—and registered only those eight sai (Japanese years) old and older. These domains were large and powerful enough to maintain their own special forms of administration. In addition, while the Tokugawa government ordered all lords to institute the system of religious investigation, it gave them no guidance on the method of constructing the population registers. Consequently the documents were not standardized and the contents varied according to the lord who created them. Town and village officials generally created the population registers once a year, on a fixed date, under the supervision of central government or domainal officials who made official visits on that day, and following locally prescribed investigation procedures. Hence, each village’s set of registers can be thought of as a panel study whose waves occurred annually. Two copies were prepared: one, the original, was presented to the lord; the other, a duplicate, remained in the locality. Occassionally there was also a third copy, destined for an intermediate administrative unit. Compiling these documents was one of the major routine jobs of village and town administrators, and many records describing preparation and attendant expenses have survived. What order the compilers used in recording residents is not known. The order of a listing tended to remain constant from year to year. It appears to represent the location of households on a fixed path around the village rather than a ranking of households by status or some other principle. But there were exceptions. For example, when a person’s status changed, say, from farmer to country samurai, he was deleted from one series of registers and entered in another.

167

12:27:16:11:09 Page 167 Page 168

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

When the results for a single administrative unit would not fit into a single volume they were divided among several. Sometimes the volumes were differentiated by sect, so that when members of a household belonged to different sects, a sort of family reconstitution is necessary to determine family size or household composition. The original copies forwarded to the lords have by now almost entirely disappeared. We do not know to what use they may have been put. It is unlikely that lords examined each of the entries to determine whether Christians were recorded in their domains. Some of the lords who were deeply concerned about the populations of their domains did use the aggregate totals found at the end of each village’s or ward’s registers. In 1721 and 1726, and every sixth year thereafter, the Tokugawa government imposed an obligation on each lord or chief magistrate to report on the population of the lands under his supervision. The population registers may have been used for that purpose as well.1 However, they were not vital for that kind of enumeration. Consequently, population registers of no particular use to the lord, from the hundreds and thousands of villages and towns within his domain, piled up in his storehouses year after year. Sometimes they were even sold as scrap paper during the Tokugawa period. Some domains regularly preserved them for only three years (e.g. Tsushima domain). In others the records disappeared later. For example, the population registers from the Suwa area in Shinano province, which have served as the base of much recent demographic analysis (e.g. Hayami, 1973a; Cornell, 1981), were preserved from 1665 to 1871, but were saved from being cut up and burned during the disturbances of the Meiji Restoration only through the efforts of a few dedicated volunteers. The population registers that exist today are almost entirely those duplicates that remained in the hands of town and village officials. Luckily, this turns out to be a boon for historical scholarship, because these officials in preparing new registers for a given year, commonly noted changes that had occurred during the previous year by marking them in the existing registers with red ink or with paper slips pasted over the entries. These emendations enable us to trace changes in the composition of the population. Many of these documents, however, have not survived. The population registers, like other town and village administrative documents of the Tokugawa period, were regarded as the personal property of the officials who created them. The official’s office was usually also his home, and the volume of administrative documents stored in it was continually expanding. In the absence of any legal protection, documents were disposed of, consciously or unconsciously, as their owner saw fit. Many were simply neglected and became nests for rats or food for insects. The registers could also become scrap paper. For example, few documents have survived in tea-producing areas, because they were used to prepare and package tea. They could also be used for undercoating in sliding paper walls, and, in poor households, to make paper screens. Many reasons explain the survival of these sources. When a village office was passed down through many generations of a single family, the chances for survival were improved. On the other hand, when many households in the village shared an office, some retained documents and others did not, and quite frequently

168

12:27:16:11:09 Page 168 Page 169

THE SHU¯ MON ARATAME-CHO¯

documents appear sporadically and continue for only a short time. Long, perfect series of shu¯ mon aratame cho¯ are rare. Still, the population registers used thus far are only a small part of those probably extant. Further research will undoubtedly reveal additional sources. Japan is yet in an age of discovery of sources pertaining to the Tokugawa period, and many population registers still rest undisturbed in old houses in the country.

THE CONTENT OF THE POPULATION REGISTERS The earliest population registers contain only information related directly to their objective: statements that the listed persons were Buddhist, the sect and temple to which they belonged, and their position within the household. Figures 10.1 and 10.2 show one entry from a population register dating from 1638 from an agricultural village in Mino province in central Japan.2 This is one of the oldest complete copies of a shu¯ mon aratame cho¯ extant. Household members are listed in columns in order from right to left, with the temple aﬃliation of each preceding his or her name. The seal of the head of household, Hachizaemon, is impressed over his name, as well as the names of the rest of the members of his household. Five or six of these listings, covering the membership of a single group of

Figure 10.1 Population Register from Niremata Village, Mino Province, 1638.

169

12:27:16:11:09 Page 169 Page 170

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

Figure 10.2 Transliteration of Population Register from Niremata Village, Mino Province, 1638.

administratively related households (gonin gumi), were gathered together and were stamped with the temple seal. At the end of such a volume, there was a postscript appended by the scribe, a village oﬃcial, attesting that there was not a single Christian in the village. The document was addressed to the oﬃce of the shogunate administrator who managed the Tokugawa domains in the Mino area at the time. Individual ages were not recorded, in accord with the limited requirements of the original mandate.

170

12:27:16:11:09 Page 170 Page 171

THE SHU¯ MON ARATAME-CHO¯

After 1670, the information in the population registers became more detailed. They could include a household’s holdings of arable land, the kinds and quality of domestic animals, and each person’s age and reasons for movement and destinations of those who moved. Figures 10.3 and 10.4 show a single entry from the same village, but dating from 1804. Considerably more information is included, and this enhances the document’s value for historical demographic research. In many cases the title also included both elements, and such population registers were often termed shu¯ mon ninbetsu aratame cho¯.

Principles of Compilation There were two general principles of compilation. The ﬁrst is the “permanent residence” principle, according to which persons were registered in their permanent residence, often the place where they were born or where their family had lived for generations. Persons who had moved to other towns or villages and those whose whereabouts were unknown continued to be listed in this type of register. If a separate register of those who had disappeared was maintained, identifying

Figure 10.3 Population Register from Niremata Village, Mino Province, 1804.

171

12:27:16:11:09 Page 171 Page 172

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

Figure 10.4 Transliteration of Population Register from Niremata Village, Mino Province, 1804.

absentees was not difficult. But sometimes the registers included very old people who had actually left the locality and had died elsewhere. Including such individuals in calculations could result in erroneous conclusions about the size of the elderly population and the structure of households. In younger cohorts, there were perverse fluctuations in figures for age structure and proportion married because temporary departure from the village for work (dekasegi) was not recorded. One must bear in mind the inherent limitations of registers of this kind when using them as the single source for historical demography.3 The second kind of register was compiled using the principle of “current

172

12:27:16:11:09 Page 172 Page 173

THE SHU¯ MON ARATAME-CHO¯

domicile.” These registers resemble those currently being compiled by local governments both in Europe and in Japan. Figures 10.3 and 10.4 show a document of this type. Registers compiled in this way give far more exact demographic indices than those employing the principle of permanent residence. So far there has been no systematic study of which lords used which method, but it appears that the principle of permanent residence was employed far more often.

Household Units It is clear from Figures 10.1 to 10.4 that the unit of record in the population registers is the household rather than either a smaller unit such as the individual or the family, or a larger unit such as the interrelated and economically interdependent households termed a do¯zoku. Each household is clearly delineated from its neighbors, often explicitly by the notation “one house” (ikken) at the beginning of the entry. In addition, the first person listed is usually designated as the “household head” (koshu). Each household has only one household head. Single-person households did exist, and the single person was described as its head, but examination of the household’s history usually indicated that this person was the remnant of a previously functioning reproductive unit. The unit of organization in the population registers is also not just a family: it can include servants and married children of the household head. Demonstrating that the unit of registration is not something larger than the household is more difficult, because we do find in the population registers what we do not expect to find in Japan’s stem family households: coresident married siblings (cf. Nakane, 1967). However, further examination of these households suggests that these married siblings are temporary coresidents who leave at an appropriate point in the household developmental cycle. Occasionally, the problem arises of a sudden increase in the number of households in a village. For example, the number of households in Yokouchi rose from 57 in 1732 to 72 in 1734, a 25 percent increase in a population that grew by only four percent. It is hard to attribute this sudden change entirely to the gradual maturation of households in the developmental cycle, though it could result from the fission of formerly dependent families into independent households. This sudden increase is probably due either to a redistribution of stakes in the village polity or to administrative fiat. Such changes cannot be disaggregated without compromising the integrity of the sources, but luckily they are fairly uncommon.

Name Peasants were not legally permitted to have surnames until after the Tokugawa period, so each individual was listed by his or her personal name. The name indicates a person’s sex, and may also, as in the cases of girls named Tome (stop) and Sue (end) indicate birth order, though not necessarily accurately. Boys regularly changed their childhood names to adult names at about the age of ﬁfteen; women lost their names at marriage and were thereafter designated simply “bride” ( yome), “wife” (tsuma, nyo¯bo¯), or, as appropriate, “widow” (goke). Once retired, household heads and their wives tended to lose their names and be recorded simply as “father” or “mother.” A few servants and distant relatives also have tags rather than names: gejo (female servant), for example.

173

12:27:16:11:09 Page 173 Page 174

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

Arbitrary changes of name, often for auspicious purposes, were fairly common. Hayami (1979: 74) cites a man who had five different names during his lifetime. The characters used in writing a name can also change, without the pronunciation changing. Despite such instability and the duplication of common names, the existence of age and household membership data means that individuals are not difficult to identify as they move through the registers. On the other hand, names are a significant problem in the linking of the population registers to other sources of data. Sometimes only fifty to sixty percent of the individuals in one register can be matched accurately with those in another source of data.

Relationship The relationship of an individual to the head of the household was expressed in a small set of simple terms, which have been common to the Japanese language since antiquity (R.J. Smith, 1962), and are still in use today. They are not strictly kinship terms but include designation of status within the household, such as “household head.” A few occupational terms such as “servant” (genin) also appear. They are unambiguous, and confusion arises in only a few cases. One occurs when a retired head of household is designated simply as “father”—his status in relation to the current head of household—when he could be the natural father, adopting father, father-in-law, or simply the predecessor (e.g. a considerably older brother or an uncle) of the current household head. Another instance occurs when parents of a person designated “grandchild” cannot be identiﬁed because the household contains more than one married couple of the appropriate age. Most such problems can be resolved by consulting the history of the household as it is revealed in registers of preceding or subsequent years. This system of relationships is simple; very few persons lack designation of relationship, and households as units rarely moved between villages in this period, so a population register is a gold mine for examining the genealogical composition of a village. Indeed, creating genealogical charts is the purpose for which many local historians and antiquarians have used them.

Age The traditional method of calculating age in Japan, as in the rest of East Asia, is that a child is considered one year old at birth and becomes one year older at each subsequent New Year’s Day. In an exceptional case, a child could be two years old on its second day of life if it were born on the last day of the year. Japanese age designations are meant to distinguish cohorts rather than to provide exact ages. This method affects calculation of statistics on fertility, age at marriage, age distribution, death rate, and life expectancy; and the numbers need to be adjusted to take account of it. Age tends to be recorded systematically, more likely by the inertia of recopying the registers every year than by annual investigation. Hence, the problem of age heaping around certain digits, such as zero and five is absent. Although ill-omened ages do exist in the traditional chronology, Hanley has found only avoidance of ill-omened days (1974: 129). In addition, she finds that the most common age discrepancy is only one year. Scholars can use either Japanese age or chronological age, but they must make clear which is

174

12:27:16:11:09 Page 174 Page 175

THE SHU¯ MON ARATAME-CHO¯

being employed. In this article we have used Japanese ages, designated by sai (“years old”) rather than the Western form. Various domains used different rules for the age at which individuals were first registered, and this hampers demographic analysis. However, in the majority of domains and, with the exception of the large cities, in areas under central government control, persons who were born—providing they survived—were registered at the subsequent compilation of the population registers, often in the first half of the next year. Consequently most newborns first appear in the registers at the age of two sai.

Movement of the Population When there were changes in the composition of the population—births, deaths, movements due to marriage or work, disappearance and other unknown causes, and erasure from the register as punishment—it was usual to record the circumstances (reason, location, date) surrounding them. However, since there was no standard form of entry for these events, they were described in many diﬀerent ways. Frequently, a separate volume dealing only with population movements was prepared, bearing the title Ninzuzo¯gencho¯, “Register of Increases and Decreases in the Population” (see T.C. Smith, 1977). This was also not standardized. More common was the method of writing the change on a small piece of paper and pasting it into the registers. However, if some of these emendations became unstuck and lost over the course of time, the result was an incomplete picture of population change. The most durable corrections were written between the lines in red or black ink. Many such examples survive. It is fairly common for changes to be registered in complete detail, but how fully were dates recorded? Usually, only the year and the month were entered. In a few cases the day was recorded as well, and the frequency of this practice tends to increase with the passage of time. If the exact date was recorded, it can be converted to the Gregorian calendar, enabling seasonal analysis of the causes of population change.

Other Data If information about arable land and ownership of domestic animals was entered along with information about the circumstances of the family, the value of the documents increases markedly. Arable land was measured by a system peculiar to Tokugawa Japan, the kokudaka system. It assessed the value of land in terms of the volume of rice it produced. For example, a farm household’s holdings of arable land—both paddy and dry field—might be expressed as Taka: five koku three to eight sho¯ six go¯, a decimal system representing 5.386 koku of rice. How- ever, what was recorded in the registers were only holdings within the village, those outside being ignored.4 The major deficiency in the registers is that occupation of household members is not listed, except in a handful of cases from the later registers.

175

12:27:16:11:09 Page 175 Page 176

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

THE POPULATION REGISTERS AS DEMOGRAPHIC SOURCES Information about the constituent parts of population change—birth, death, marriage, and migration—is readily available in the registers, and since the members of a household are listed together, family reconstitution is rarely required. Nonetheless, there are a number of problems of which a researcher must be aware.

Births and Deaths Not all births and deaths were noted in the registers. Only those persons who survived from birth to the subsequent compilation were listed. Those who died in the interval appear neither in the registers themselves nor in the registers of increase and decrease. This is the principal problem in using these materials for population statistics. Ignoring this point results, in Taeuber’s words, in “birth and death rates . . . so low as to be improbable” (1958: 29). Her perplexity arose from calculations that showed birth and death rates in late Tokugawa Japan fluctuating between twenty and thirty per thousand. These calculations, however, were made when the character of the registers was not yet well understood. There are a number of methods for estimating the number of hidden births. One is to use a Coale-Demeny model life table to compensate for the number of deaths which occur in the first year of life. Another possibility is to calculate infant mortality from other documents of the same period and to use it to adjust the figures. One source for this are the documents compiled by many lords in northeastern Japan for the purpose of giving subsidies for pregnancies or births, in response to the severe declines in population in the eighteenth century. When these investigations were carried out, they listed whether a stillbirth occurred and whether the child who was born grew up or not. So far, only a few such villages, where usable runs of both documents survive for the same years, have been discovered. Kito’s research (1976) on the 1808–1826 cohort of births in a single village in northeastern Japan found a stillbirth rate of eighty per thousand and infant mortality of 180 per thousand. It is still too early to consider these figures representative of contemporary Japan, but it does appear that the number of births that cannot be observed in the population registers—not counting stillbirths—amounts to about twenty percent of the number of infants who first appear in the registers at the age of one sai or two sai. Future examination of temple death registers may also assist us in resolving this problem (Ginsberg and Jannetta, 1985). This under-representation of the number of births naturally results in an underestimation of deaths, and especially of the significance of infant mortality. Careful consideration of the data leads us to estimate the actual birth and death rates from the births and deaths recorded in the population registers at the levels indicated in Table 10.1. Even when listed persons are not noted as births, we can treat those who appear at one or two sai as such. This is not true of deaths. No problem arises if the reason for the disappearance of a name is stated clearly, but death is only one of a number of possible reasons for disappearance from the registers, and the actual death rate will be underestimated in proportion to the number of persons whose

176

12:27:16:11:09 Page 176 Page 177

THE SHU¯ MON ARATAME-CHO¯

Table 10.1 Correction of Population Register Figures Through Estimation

Population: 1000

Births Deaths

Number of births recorded in the 30 Number of deaths reported 25 registers Number of births added by 6 Number of deaths added by 6 correction correction Total births 36 Total deaths 31 Birth rate before correction 0.030 Death rate before correction 0.025 Birth rate after correction 0.036 Death rate after correction 0.031

reason for disappearance is unknown. Thus, in calculating statistics of population movement, it is imperative that the entries used be exact. When dealing with births, there are sometimes problems determining who a child’s parents are. In rare cases, a woman clearly beyond the possible age of child-bearing is recorded as the mother of a baby. One explanation is that “in almost every case where this occurred, there was a grown daughter living at home and, since it was presumed that the real relationship of the child was grandchild, he was so classiﬁed” (Hanley and Yamamura, 1977: 205). It is also possible that the husband had fathered an illegitimate child by a woman other than his wife. Tokugawa Japan was a monogamous society, but persons of the upper classes did have concubines. However, the category of relationship “concubine” is not used in the population registers, and legitimate and illegitimate children cannot be diﬀerentiated. (Strictly speaking, it is possible that a child registered to a wife who had not completed her childbearing years could in reality be illegitimate.) The number of children who were born with either a mother or a father absent number seven percent of the births listed in the registers of one village in central Japan from 1773 to 1871. This may not be representative of the entire country, but it is still considerably higher than the statistics on illegitimacy compiled by the government in the 1880s (Hayami, 1980a). The problem of identifying the parents of children when there is more than one appropriate couple in the household is mentioned above.

Marriage Marriage can always be recognized as a specific reason for migration, whether or not it is listed as such. This is because in many registers the date of marriage and the origin of the bride or groom are listed, not only in the year of marriage but every year thereafter until the person dies or leaves. Hence, age at marriage and place of origin can be determined even when some years of the registers are missing. However, in accordance with peasant customs of the period, some marriages were not recorded until after a pregnancy or even after a birth. Con- sequently the interval between marriage and first birth is usually short. Age at marriage is readily observable, but when a person enters from another village there is no information as to whether it is the person’s first or a later marriage. When a person migrated for marriage, his or her registration shifted from the

177

12:27:16:11:09 Page 177 Page 178

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

original place of residence to a new one. The formalities of migration were concluded by having certiﬁcates from the original temple and village oﬃcial sent to the new place of residence. As a consequence, registrations of marriage were more precise than those of birth or death.

Migration As mentioned above, population registers compiled on the current domicile principle recorded the particulars of migration in detail. The formalities were the same as in the case of marriage: the migrant was required to carry a copy of his certificate to the village officials at his new place of residence. Had the registers in this locality been compiled using the principle of permanent residence, his name would not be recorded in them but, in most cases, in a separate volume such as a “register of service.” Where the register is one of current domicile, information about migration is complete and enables us to carry out detailed analysis of migration with confidence (Hayami, 1973b). This system of transferring documents worked well in rural areas but not as well in the cities. Urban migration was an important political problem to the domain lords, because the exodus of the rural population appeared to them to mean a decline in agricultural yields and hence a decline in their own income. Efforts were made to curb such migration with strict controls on who could leave or by prohibiting it entirely, but these measures were largely ineffectual. The ruse migrants commonly employed was leaving for an urban area on dekasegi— temporary labor migration—and simply never returning to the village. The numbers of individuals using this strategy were always far greater than those who left with official approval and the numbers increased with the passage of time. These individuals then appeared in urban registers as “vagrants” or “homeless persons.”

AVAILABILITY TO SCHOLARS Having discussed the construction, distribution, and survival of the registers, we should also consider how available they are to scholars. Because they are hand- transcribed, privately-owned documents written in the Japanese language, they are necessarily less accessible than is desirable, but there is some evidence that this may be changing. The vast majority of the extant population registers are privately owned manuscripts and are scattered across the country. Hence, they become available even to Japanese scholars only after careful and time-consuming searches. When discovered, they may be deficient in quality and coverage. A few larger collections do exist, however. The Archives of the Ministry of Education (Mombusho¯ shiryo¯kan) contains a large number of original and microfilmed documents covering a wide variety of areas. They continue to be collected there. Keio University maintains an extensive collection of microfilms, and some scholars have small collections from their regions of interest. The Genealogical Library of the Church of Latter-day Saints in Salt Lake City, Utah has also begun microfilming them. Unfortunately, even though there is an abundance of manuscript documents, so far there have been no efforts to catalog them and to describe the quality and extent of their coverage. This remains a top priority for future research.

178

12:27:16:11:09 Page 178 Page 179

THE SHU¯ MON ARATAME-CHO¯

Unlike the eighth- to tenth-century population registers mentioned earlier, few later registers have been transcribed and published. Further work along this line would be an important contribution to scholarly research, because the ability to read handwritten documents from the Tokugawa period requires skills much less easily learned than for reading European documents of the same period. Since the information is simple and repetitive, printing it would make decoding relatively easy even for those who do not know Japanese. Fruin (1973) employed printed registers from the province of Echizen, and Sakamoto and Miyamoto (1971–77) have prepared some for a ward in Osaka. Most scholars have worked from handwritten registers transcribed directly to coded data sheets. Much of this transcription has been carried out by Hayami’s research group at Keio University. Hayami employs three principal coding forms modeled after those designed by Louis Henry: the basic data sheet, the family reconstitution form, and the individual tracing sheet (see Hayami, 1979). The data sheet lists, on a single page, the names and ages of the members of a household for a 25-year period; the reconstitution links married couples and their offspring, indicating age of mother at each parity; and the tracing sheet summarizes individual histories. The population registers become most accessible to scholars when they take the form of computer-readable data accompanied by a detailed coding book. A number of scholars have used such data sets in their own work. Sasaki (1978), for example, has coded data covering 40,275 individuals in the city of Takayama from 1773 to 1871. Hanley’s work on the Okayama area (Hanley and Tamamura, 1977) and T.C. Smith’s work on the village of Nakahara (1977) also employ computer-generated data bases, and Cornell (1983) has also prepared such a base for the village of Yokouchi. While a start has been made, these efforts are scattered and researchers each have employed their own idiosyncratic methods to create the data base. There is a need for consultation with other scholars so that data preparation can use the most efficient and informed methods.

RESEARCH USING POPULATION REGISTERS The population registers have shown themselves to be one of the best sources for understanding the lives of the common people in Tokugawa Japan. Naturally it was Japanese scholars who initiated this research. Hayami has wrestled with these documents since the 1960s, achieving notable results (Hayami, 1967, 1968, 1969, 1971, 1973a, 1973c, 1981, 1982a, 1982b; Hayami and Uchida, 1972). Sasaki (1978) has assembled and computerized a magnificent set of documents dealing with the city of Hida Takayama, and Saito¯ (1983) has used province-level documents to examine proto-industrialization in Japan. In the United States, among the first to use the population registers was R.J. Smith (1972), who examined mobility in urban areas. In the 1970s Fruin (1973), Hanley (1977), and T.C. Smith (1977) analyzed demographic trends in a number of villages and Cornell (1981) used the registers for research on peasant succession and inheritance patterns. The usefulness of the registers goes far beyond their value as a source of aggregate demographic statistics. By virtue of their size and comprehensiveness, they also offer us a unique glimpse into household processes that would otherwise

179

12:27:16:11:09 Page 179 Page 180

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

be invisible. And they help to understand “traditional Japan”: the baseline against which change is measured in many sociological, anthropological, and political studies. One research area for which the registers are likely to be of exceptional value is life-course analysis. Ethnographic and folkloric accounts have already provided abundant detail about traditional life-cycle events, but it is difficult to generalize from these accounts. This is true not only for events in the individual life cycle, but also for those in the developmental cycle of the household, which are less well articulated in the ethnographic record. Historical demographers have already produced a great deal of information about the average timing of such important demographic events as marriage. However, the wealth of data available in the population registers allows us to examine variability and changes in the timing of a single event, the timing of sequences of events, and the timing of non-demographic events. We can examine variance in timing—the extent to which individuals conform or fail to conform to a postulated schedule—by sex, class, region, or other variables. Hayami (1980b), for example, shows that upper- class peasant women entered into childbearing significantly earlier than did lower-class ones, presumably because the former had little experience of service outside the village. This applies to household events as well: Cornell (1983) examined the order of three events—father’s retirement, father’s death, and heir’s marriage—and how fluctuations in their timing reveal intergenerational conflict. It is also possible to examine changes in timing. Cornell (1981) has shown that over a three-hundred year period, sons left the household increasingly earlier, which led directly to smaller and less complex households as time passed. One unique feature of the Japanese registers is that the growth and decline of a household can be followed over a number of generations. In theory, this can be done in European records as well, but there a succession of households is frequently only a mechanical process—one nuclear family happening to replace another at the same address—while in Japan there is continual residence of the same family group. Familial ideology in Japan emphasized the continuity of the household over generations, admonishing a household head to revere his own ancestors and to provide a successor. Hence departures from the household, especially departures of entire families, and arrivals into it, especially arrivals of adults, reflect deliberate decisions made according to social rules. Even in the absence of economic data, a record of these movements can give us important information on patterns of succession and inheritance and on social responses to demographic difficulties such as childlessness. How many generations a household endures can be a useful statistic. As Hayami (1983) has pointed out, there are significant regional differences in Japan in household survival that reflect variation in adherence to traditional rules of household formation and probably arise from different economic opportunities in rural and urban areas. The economic interdependence of households cannot be measured, except where land records exist and can be linked to the population registers. Social and political interdependence, however, can be measured by examining such questions as whether spouses are recruited from related households and in which direction the exchange takes place, and whether households are able to

180

12:27:16:11:09 Page 180 Page 181

THE SHU¯ MON ARATAME-CHO¯

turn to dependent households to resolve heirlessness and other problems of membership. Research using data from only one village raises doubts about whether the findings are representative of the larger area. While it is desirable to amass documents from a number of villages in the same area, doing so may be difficult, since most such documents are privately owned and the owner’s permission is necessary for access. Hayami, however, has so far not found this a great hindrance. He has already made regional collections of documents from a number of villages in a mountain valley in central Japan (1973a) and from the urban plains near Nagoya. He is now collecting documents from a locality in northeastern Japan. Tokugawa Japan was a society with enormous local differences in demographic patterns, social customs, and economic development, so that carrying out a large number of local studies in various areas is crucial to further understand this important period in Japanese history. The population registers are not the only documents that can be used to investigate population questions in this period. As mentioned earlier, a number of lords also took what amounts to a census (ninbetsu). For example, one set of ninbetsu aratame cho¯ covers a domain in northeastern Japan for every year from the late seventeenth century onward and contains more detailed information than even the best population registers. The census even lists individuals who were ill. There are also the temple death registers, which list the deceased person’s name, age, date and location of death, and sometimes its cause. These have served as a source for studies of mortality due to epidemic diseases (Jannetta, 1983). Although entries in the population registers contain relatively little information about any single individual, merging the details of many individual lives over long periods of time enables us to discern patterns of social and demographic behavior that would otherwise be invisible. When the population registers come to be combined with other sources of data such as temple death registers, land and tax registers, and local folklore, they will undoubtedly prove a lode of incom- parably detailed information about the lives of the common people in the Japanese past.

REFERENCES Boxer, C.R. 1951 The Christian Century in Japan: 1549–1650. Berkeley: University of California Press. Brown, Keith 1978 “Family history in a Japanese town: 1872–1971.” Paper presented to the Conference on Historical Demography and Family History in East Asia. Oxford, England, August 20–25. Cornell, L.L. 1981 “Peasant family and inheritance in a Japanese community, 1671–1980: an anthropological analysis of local population registers.” Unpublished Ph.D. dissertation, Johns Hopkins University. 1983 “Retirement, inheritance, and intergenerational conﬂict in preindustrial Japan.” Journal of Family History 8:55–69.

181

12:27:16:11:09 Page 181 Page 182

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

Farris, William Wayne 1985 Population, Disease, and Land in Early Japan. Cambridge: Harvard Uni- versity Press. Fruin, William M. 1973 “Labor migration in nineteenth century Japan: a study based on Echizen domain.” Unpublished Ph.D. dissertation, Stanford University. Ginsberg, Caren A., and Ann B. Jannetta 1985 “Changing age-specific mortality in the Japanese Alps: 1775–1934.” Paper presented to the Annual Meeting of the Population Association of America, Boston, Massachusetts, March 28–30. Hanley, Susan B. 1974 “Fertility, mortality, and life expectancy in pre-modern Japan.” Population Studies 28:127–142. Hanley, Susan B. and Kozo Yamamura 1977 Economic and Demographic Change in Preindustrial Japan, 1600–1868. Princeton, N.J.: Princeton University Press. Hayami, Akira 1963 “Epanouissement du ‘nouveau régime seigneurial’ aux 16e et 17e siécles.” Keio Economic Studies 1:21–51. 1967 “The population at the beginning of the Tokugawa period.” Keio Economic Studies 4:1–28. 1968 “The demographic analysis of a village in Tokugawa Japan: Kando-shinden of Owari Province, 1778–1871.” Keio Economic Studies 5:50–88. 1969 “Aspects demographiques d’un village Japonais, 1671–1871.” Annales E.S.C. 3:617–639. 1970 “Demographic aspects of a village in Tokugawa Japan.” Pp. 109–125 in P. Deprez (ed.), Population and Economics. Manitoba: University of Manitoba Press. 1971 “Mouvements de longue durée et structures Japonaises de la population a l’epoque de Tokugawa.” Annales de Démographie Historique 1971:247–263. 1973a Kinsei Noson no Rekishijinko¯gakuteki Kenkyu¯ (The Historical Dem- ography of Tokugawa Villages). Tokyo: To¯yo¯ Keizai Shinpo¯sha. 1973b “Labor migration in a pre-industrial society.” Keio Economic Studies 10:1–18. 1973c “Demografia e economia no Japão preindustrial.” Anais de Historia (Brasil) 4:5–22. 1979 “Thank you Francisco Xavier: an essay in the use of micro-data for historical demography of Tokugawa Japan.” Keio Economic Studies 15:65–81. 1980a “Illegitimacy in Japan.” Pp. 397–402 in Peter Laslett et al. (eds.), Bastardy and Its Comparative History. London: Cambridge University Press. 1980b “Class differences in marriage and fertility among Tokugawa Villagers in Mino Province.” Keio Economic Studies 17:1–16. 1981 “Kyoto Machikata no Shu¯ mon Aratame Cho¯ (The Population Registers of a Street in Kyoto).” Tokugawa Rinseishi Kenkyu¯ jo, Kenkyu¯ Kiyo¯ 1980:502–541. 1982a “Kinsei ou Chiho Jinko¯ no Shitekikenkyu¯ Joron (Introduction to the historical study of population in northeastern Japan).” Mita Gakkai Zasshi 75:70–92. 1982b “Population changes in the transition period: 1820–1890.” Paper presented

182

12:27:16:11:09 Page 182 Page 183

THE SHU¯ MON ARATAME-CHO¯

to the Tokugawa-Meiji Transition Conference, White Sulphur Springs. West Virginia, U.S.A., August, 1982. 1983 “The myth of primogeniture and impartible inheritance in Tokugawa Japan.” Journal of Family History 8:3–29. Hayami, Akira, and Nobuko Uchida 1972 “Size of household in a Japanese country throughout the Tokugawa era.” Pp. 473–515 in Peter Laslett (ed.), Household and Family in Past Time. Cambridge: Cambridge University Press. Jannetta, Ann B. 1983 “Epidemics and mortality in Tokugawa Japan: 1600–1868.” Unpublished Ph.D. dissertation. University of Pittsburgh. Kishi, Toshio 1973 Nihon Kodai Sekicho¯ no Kenkyu¯ (Study of the Population Registers of Ancient Japan). Tokyo: Hanawa Shobo. Kito¯, Hiroshi 1976 “Tokugawa Jidai no Nyu¯ ji Shibo¯—Kainin Kakiagecho¯ no Tokeiteki Kenkyu¯ (Infant mortality in Tokugawa Japan—a statistical study of the ‘register of pregnancies’).” Mita Gakkai Zassjo 69:88–95. Nakane, Chie 1967 Kinship and Economic Organization in Rural Japan. London: Athlone Press. Saito¯, Osamu 1983 “Population and the peasant family in proto-industrial Japan.” Journal of Family History 8:30–54. Sakai, Robert K. et al. (eds.) 1975 The Status System and Social Organization of Satsuma. Tokyo: University of Tokyo Press. Sakamoto, Heiichiro, and Miyamoto, Mataji (eds.) 1971–1977Osaka Kikuyamachi Shushi Ninbetsucho¯ (The Population Registers of Kikuyamachi in Osaka), 7 Vols. Tokyo: Yoshikawa Kobunkan. Sasaki, Yoichiro¯ 1978 “A computer procedure for research in historical demography: a program for family reconstitution.” Paper presented to the conference on Historical Demography and Family History in East Asia, Oxford, England, August 20–25. Smith, Robert J. 1962 “Japanese kinship terminology.” Ethnology 1:349–359. 1972 “Small families, small households, and residential instability: town and city in pre-modern Japan.” Pp. 429–471 in Peter Laslett (ed.), Household and Family in Past Time. Cambridge: Cambridge University Press. Smith, Thomas C. 1977 Nakahara: Family Farming and Population in a Japanese Village, 1717– 1830. Stanford: Standford University Press. Taeuber, Irene B. 1958 The Population of Japan. Princeton, N.J.: Princeton University Press. Uchida Masao (comp.) 1975 Nihon Rekijutsu Genten (Collected Table of Japanese and Western Calendars). Tokyo: Yuzenkaku.

183

12:27:16:11:09 Page 183 Page 184

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

NOTES

1. Hayami (1982a) uses statistical documents on each domain’s population created by the lords and chief magistrates of northeastern Japan, who were visited by the great famines of the eighteenth century. Many lords kept such statistics, using various formats. 2. In Japanese chronology, the year is Kan’ei fourteen. Before the adoption of the Gregorian calendar on 9 September 1872 by the Meiji government, Japan employed the Chinese solar-lunar calendar. Maintenance of the calender was the responsibility of a given noble house, who chose the name of the year by divination. The name of the year could be changed at any time. For example, the document shown in Figures 10.1 and 10.2 is dated the fifteenth year of Kan’ei. The series of years designated Kan’ei began with the thirtieth day of the second month of the tenth year of the previous series, Genna (1624), and continued until the sixteenth day of the twelfth month in the twenty-first year of Kan’ei, when it was succeeded by the year series Sho¯ho¯ (1644). In this calendar, a month corresponded to one phase of the moon, so it comprised 29 or 30 days. An intercalary month was added once every two or three years to make these phases coincide with the four seasons. The Japanese calendar was usually about one or two months behind the Gregorian one. For example, the Gregorian year 1638 began on the sixteenth day of the eleventh month of Kan’ei fourteen and ended on the twenty-sixth day of the eleventh month of Kan’ei fifteen. Conversely, Kan’ei fifteen lasted from 14 February 1638 through 2 February 1639. Consequently, there is likely to be a disparity of about one year in the Western calendar when one converts the beginning and ending dates of a given year in the Japanese calendar. In this chapter, we have simply designated Kan’ei fifteen as 1638. For exact comparisons of dates, see Uchida (1975). 3. T.C. Smith (1977) employs the population registers of a single village to establish the existence of parity-specific infanticide. Those population registers are of the “permanent residence” type, and consequently statistics on mean household size and average life expectancy are higher than those derived from villages in the same area whose registers were compiled on the principle of current domicile. The villages were not in themselves different but were governed under different administrative systems. 4. The value of a given parcel of land was assessed by a land survey. Such surveys were begun at the end of the sixteenth century by Toyotomi Hideyoshi and were continued by the Tokugawa government. On the occasion of a survey, each parcel of land was assigned a value, expressed in a quantity of rice, which represented its average annual production. The size of any amount of land, not only farms but also villages, provinces, and lord’s holdings, could be expressed by volume in koku.

184

12:27:16:11:09 Page 184 Page 185

First published in Population and Economics (ed. P. Deprez), pp. 109–125, Manitoba, Canada, 1970.

11 Demographic Aspects of a Village in Tokugawa Japan

INTRODUCTION okugawa Japan1 bequeathed to us an abundant supply of materials for Tpopulation study. Although these materials cannot be compared to modern scientific demographic data, they are excellent for their day. Japan in the Tokugawa period was a rare example of a well documented pre-modern state. There are two reasons for this fortunate situation. First, when Japan was finally unified in the late sixteenth century, the shogun and the provincial feudal lords became very much interested in the quantitative aspects of the regions they governed and hence conducted surveys on the size and population of their domains. These early surveys were the first attempt of its sort since the eighth and ninth centuries when a type of census had been carried out. Second, as a result of the strict ban put on Christianity by the Tokugawa government, a registration of the religious faith of every individual was carried out every year. The people were ordered to report what religious body they belonged to and to avow themselves Buddhists. In general, all the people, both urban and rural, were ordered to enter the name of their family temple in the register of their town or village, the basic unit for this national survey. This religious registration system was put into effect on a nationwide level about 1670. At first, the system had nothing to do with census-taking, but later it also came to be utilized for that purpose. Besides the names and the family temples of individuals, their sex, age, and relationship to their household head, and such matters as the reason and the date for any change in residence were entered. The form of these registrations differed in various parts of the country. The fact, however, that census-like registrations existed in many provinces, and especially the fact that they were conducted annually, should be considered a significant event facilitating the demographic study of the Tokugawa period. As a result of the existence of the two types of data described above, Japan is fairly well endowed with demographic materials. This report is the result of a demographic study I made using the example of Yokouchi village in Suwa county in the province of Shinano. The materials for this village are, from the point of view of a continuous study over a period of years, the best which I have discovered to date. I have made a study of various aspects of the demographic history of this village by combining the traditional method of

185

12:27:16:11:09 Page 185 Page 186

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

population study with a tracing of the demographic behavior of families through the method of family reconstitution. In Suwa han, the domain to which this village belonged, shumon aratame cho or religious registrations were drawn up for every village for 201 years from 1671 to 1871. The form of the survey continued to be almost the same throughout the period. There are shumon aratame cho extant for the village of Yokouchi for 144 years out of this 201 year period. In general, there is no one method for determining to what extent these materials can be trusted, but for Yokouchi village the materials for the ﬁrst half of the period can be relied upon to a greater extent than those of the latter half. This generalization is based on the extent to which the materials are continuous, on the extent to which births are reported, and on the extent to which the reasons for any changes are recorded. Thus, if this span of 201 years is divided into 25-year periods, the reliability of Period II (1701–1725) is the highest in every respect. There is the least amount of continuity in the materials from Period V (1776–1800) on. Also, the recording of births is incomplete in Periods VII and VIII (1826–1871). The method used for studying these materials was ﬁrst, at the most basic level, to record on cards—one card for every family for every year—the name, sex, age, and relationship to the head of the household of every person in the family and to note the reasons for any changes in the household. Using these cards it was possible to make horizontal studies on the village for any year for which there were materials. And because there is no break in the continuity of the materials of over four years, it was possible to make vertical times studies on each family. The next step was to record on one card the demographic behavior of each couple in order to obtain from the basic cards more detailed material concerning marriage and births. For this it was possible to adopt the method developed in the 1950s by Henry Fleury. This method of dealing with the basic materials, that is of using the family reconstitution method carried out in the studies on the historical demography of Western Europe, was extremely valuable in tracing long-term demographic changes. And because the shumon aratame cho for this village record all family members in one place, family reconstitution was extremely easy to carry out, in comparison with the work European demographers have in reconstructing families from birth, marriage, and death records. The village under study, Yokouchi, is situated on the edge of the Suwa basin in Nagano prefecture in central Japan. (Fig. 11.1) At the beginning of the Tokugawa period, this village was estimated at 350 koku on the lord’s cadastre, which meant that the cultivated land produced about 1,800 bushels of grain. The altitude, about 2,600 feet above sea level, is so high that agriculture cannot be carried out in the winter. In the second half of the nineteenth century, sericulture and the manufacture of spinning silk were developed in the Suwa basin, but it is not clear how these developments were related to population growth.

186

12:27:16:11:09 Page 186 Page 187

DEMOGRAPHIC ASPECTS OF A VILLAGE IN TOKUGAWA JAPAN

Figure 11.1 Map

DEMOGRAPHIC OBSERVATIONS OF THE VILLAGE General Trend The demographic history of this village as it appears in the records started with 189 persons and 27 households in 1671. The population grew steadily in the ﬁrst hundred years, and had grown to 524 persons by 1771. These ﬁgures are the registered population which is based on the number registered in the shumon aratame cho. If we obtain the actual population by subtracting from the registered population those who went out to work and by adding those who entered the

187

12:27:16:11:09 Page 187 Page 188

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

village to work, the increase is from 184 persons to 493. In any event, the increase in population was rapid, and moreover, it continued over a 100 year period. The average growth rate per year was as high as 0.9%. (Fig. 11.2) Compared to the growth in population, the growth in the number of households was even greater. During the same period the number of households increased 3.6 times, from 27 to 98. Naturally this meant a decrease in the average number of family members, namely, a change toward smaller families. In general it can be said that in the sixteenth and seventeenth centuries there was a movement away from the large households of the pre-Tokugawa period which contained subordinate labor toward the creation of the nuclear family in the farming villages. In this village also, according to the materials of the beginning of the period, more than 15% of the households had more than ten members. Com- pared to this, by the middle of the eighteenth century, the ﬁgure had become less than 9% and the average number of persons per household had fallen from 7.0 persons to 5.9. (Table 11.1) This phenomenon was not based on a fall in the birth rate. In the early period there remained to some extent households which included unmarried blood

Figure 11.2 Population Trends, Number and Mean Size of Households

188

12:27:16:11:09 Page 188 Page 189

DEMOGRAPHIC ASPECTS OF A VILLAGE IN TOKUGAWA JAPAN

Table 11.1 Distribution of Households by the Number of Married Couples and by the Household Size

Year 1671 1722 1771 1823 1871

Married Couples per Household 0 2 3141532 1 1328576569 2 9 15 21 20 17 3 354 11 4 ––1 –– Average 1.48 1.43 1.17 1.07 0.88 Household Size 1 ––159 2 126 99 3 1 6141420 4 3 6182220 5 4 4231732 6 2 10 10 15 15 7 459 88 8 254 54 9 543 1– 10445 1– 11133 – 1 12 –––3 – 13 ––11– 14 ––– –1 15 – 2 ––– Average 7.00 6.67 5.36 4.87 4.45

relatives and unrelated members such as genin and ge jo, subordinate male and female workers. These gradually disappeared and small families composed of only one couple and their children became the general trend. Consequently, the percentages of the married went from 47.2% for males and 55.9% for females aged 16–50 in the fourth quarter of the seventeenth century to 54.2% for males and 76.1% for females in the third quarter of the eighteenth century. (Table 11.2) It is thought that this trend toward small families was already relatively well developed in the period for which we have materials but it is necessary to take note of this because it is one of the most important factors in the rapid population growth in the ﬁrst half of the Tokugawa period. The above population trends underwent a complete change in the fourth quarter of the eighteenth century. First, the total population of the village was stabilized and up until the new population increases in the latter half of the nineteenth century, there were some short-term changes but the population as a whole remained almost completely stationary. The growth rate for the second 100 year period averaged 0.05% per year. Families became nearly as small as was possible, but even so, the number of households increased out of proportion with the increases in population, and the average number of persons per household became still smaller, decreasing to between 4.7 and 5 persons. The percentage of

189

12:27:16:11:09 Page 189 Page 190 1871 1871 – – 1850 1851 1850 1851 – – (per cent) (per cent) 1825 1826 1825 1826 – – 2.7 1.2 0.4 1800 1801 1800 1801 – – 1775 1776 1775 1776 – – Proportions Married 1750 1751 1750 1751 – – ––– Table 11.2 Table 1725 1726 1725 1726 – – 2.2 1700 1701 1700 1701 – – – 25303540455050 39.8 Total 54.4 65.6 80.6 33.6 47.2 78.7 74.1 89.7 92.7 91.0 18.0 64.5 89.0 58.7 94.6 88.9 82.2 13.5 50.1 88.4 51.8 97.6 73.0 90.7 54.2 95.9 9.0 35.4 98.2 58.8 90.2 48.7 86.6 22.9 53.4 89.4 74.8 84.0 62.6 93.6 10.5 64.2 90.0 80.0 83.3 50.0 97.7 8.6 36.2 85.7 68.6 76.6 46.4 89.1 79.6 20 2025303540455050 42.5 53.3 Total 82.1 75.4 54.2 61.2 72.5 55.9 47.8 85.6 44.6 92.8 78.4 45.8 73.9 74.4 74.6 70.8 78.3 100.0 45.5 100.0 81.5 74.7 82.2 81.8 83.5 73.9 77.3 30.3 57.4 76.1 89.7 66.3 88.0 93.0 94.3 32.8 82.9 68.9 87.0 89.1 73.9 83.6 92.3 19.7 67.9 78.9 89.6 97.6 81.0 82.6 18.8 98.7 56.4 64.6 72.4 84.6 82.5 66.7 84.4 68.3 86.3 71.9 – – – – – – – – – – – – – – – – 21 26 31 36 41 46 16 (1) Male Age16 Periods 1671 (2) Female Age16 21 26 Periods31 36 41 1671 46 16

12:27:16:11:09 Page 190 Page 191

DEMOGRAPHIC ASPECTS OF A VILLAGE IN TOKUGAWA JAPAN

the married, after reaching a peak in the ﬁrst half of the period, leveled oﬀ, and in the long run perhaps tended to decrease.

Age Structure Next let’s look at the age structure by period.2 In the ﬁrst half of the 200 year period, the half in which there was a high population growth rate, and particularly up until the seventeenth century, the typical form of the recorded population was pyramid shaped. However, in the second half of the period, from the second half of the eighteenth century, the typical form was closer to the bell-shaped pattern. This may have been due in part to a decline in the percentage of infant and child mortality, but it can better be understood as a contrast between the ﬁrst 100 year period in which there was a high birth rate and the latter 100 year period in which the high birth rate had ended. (Fig. 11.3) From the above it is clear that there was a rapid increase in population from 1671 to 1775, and that after that, up until the 1850’s, the population remained stationary. The reasons for this population increase were a rise in the percentage of married persons, and resulting from this, a change toward smaller families and a continued high birth rate.

Birth and Death Rates During the period in which shumon aratame cho were carried out, only the babies who survived were registered, and thus it is not known what the true birth rate was. However, we can consider the percentage of persons recorded at the age of one or two compared to the current population an index of the birth rate. For the ﬁfty year period from 1671–1725 this rate was 3.5%–4.0%, and during the next

Figure 11.3 Age Composition (1) 1676–85 Average

191

12:27:16:11:09 Page 191 Page 192

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

(2) 1751–60 Average

(3) 1826–35 Average

fifty years it was 2.5%–3.0%. During the final 100 years it was 2.0%–2.5%. The figures for this last period are too low, however looked at, but this is thought to be due to the unreliability of the historical materials. However, it is certain that during the first 100 years there was a continuously high birth rate. (Table 11.3) On the other hand, if we look at changes in the death rate calculated from the number of deaths of persons clearly shown in the records to have died, we see a rate for the first fifty year period of 2.5%, and after this period a rate of around

192

12:27:16:11:09 Page 192 Page 193

DEMOGRAPHIC ASPECTS OF A VILLAGE IN TOKUGAWA JAPAN

Table 11.3 Birth and Death Numbers and Rates (per thousand)

Births Deaths

Periods Number Rate Number Rate

1671–1700 235 36.2 170 26.2 1701–1725 316 40.9 202 26.2 1726–1750 287 29.7 209 21.6 1751–1775 318 27.5 198 17.1 1776–1800 243 21.0 233 20.2 1801–1825 283 24.9 236 20.8 1826–1850 253 22.8 216 19.5 1851–1871 224 22.5 174 17.5 Total 2159 28.2 1639 19.6

2.0%. These figures also are too low to be trusted. This is due to the fact that the recorded deaths are more imperfect than the recorded births. The above figures are low due to the fact that infants who died before being recorded in the shumon aratame cho are not included, and some adjustment should be made to include these figures. However, because the life expectancy was extended and infant mortality decreased to a certain extent, the death rate in the second hundred year period was probably lower than in the first hundred year period. It is not known whether or not the material concerning infant mortality became more unreliable over time, but in the first hundred year period, on the average, out of 1,116 persons whose births were recorded, 26% died by age five, and 31% by age 10. If we added to this the infant and child deaths which were not recorded, it would probably be seen that nearly 40% of the babies born died before reaching age 10. However, the infant death rate was improved in the first hundred year period, and while the death rate was 39% up to age 10 from 1671–1725, during the next fifty year period it fell to 26%. (Table 11.4)

Life Table If we study the probability of deaths based on the age of death, the curve gradually shifts upward, and along with this, the life expectancy at age two is extended. In the period 1671–1700, the life expectancy for males was 24.8 years and that for females 29.8, but in the period 1726–1750, it rose to 34.8 years for males and 35.2 for females. If we look at changes in the average life expectancy for each age group, we see that there was a remarkable lengthening of life expectancy for males under age 20 and for females under age 40. These changes correspond to changes in the composition of age groups and an increase in the proportion of the productive population in the total population. (Fig. 11.4)

Marriage Next, if we look at the age of marriage, we can see a distinct trend for females. The average age of marriage for females in the seventeenth century was

193

12:27:16:11:09 Page 193 Page 194

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

Table 11.4 Child Deaths

Periods of Sex No. of Number of deaths by age Survival birth births rate at 10 1 2 3456–10

m. 119 – 18 6 15 4 6 0.59 1671–1700 f. 108 – 14 4 9 5 5 0.66 t. 227 – 32 10 24 9 11 0.62 m. 157 – 19 12 14 6 4 0.67 1701–1725 f. 145 2 18 19 12 8 7 0.55 t. 302 2 37 31 26 14 11 0.60 m. 155 – 9 8 7 7 13 0.72 1726–1750 f. 123 – 10 8 5 5 10 0.69 t. 278 – 19 16 12 12 23 0.70 m. 160 1 9 9 4 5 9 0.77 1751–1775 f. 149 – 5 10 5 3 6 0.81 t. 309 1 14 19 9 8 15 0.79 m. 591 1 55 35 40 22 32 0.69 Total f. 525 2 47 41 31 21 28 0.68 t. 1116 3 102 76 71 43 60 0.69

Figure 11.4 (1). Life Table-Survival Rates

194

12:27:16:11:09 Page 194 Page 195

DEMOGRAPHIC ASPECTS OF A VILLAGE IN TOKUGAWA JAPAN

Figure 11.4 (2). Probability of Death for Age Speciﬁc Groups

18.8 years, but in the ﬁrst quarter of the eighteenth century, it dropped to 16.9 years. Gradually rising, it reached 21.7 years in the latter half of the nineteenth century. For males there was a stable trend of between 25.6 and 28.5 years. (Fig. 11.5)

195

12:27:16:11:09 Page 195 Page 196

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

Figure 11.5 Mean Age at Marriage

Flowing Out of Labor Another interesting aspect is the population exodus, especially that of the male labor force. In the beginning, the numbers were not very large, and those who left were generally employed in agriculture in neighboring areas. Naturally, a certain number of persons from other villages came into Yokouchi, but very soon the destination of persons leaving their villages changed from neighboring villages to Edo (now Tokyo) where they became employed in the commercial or service

196

12:27:16:11:09 Page 196 Page 197

DEMOGRAPHIC ASPECTS OF A VILLAGE IN TOKUGAWA JAPAN

sectors. The numbers of those who left Yokouchi village began to increase from the second quarter of the eighteenth century, and at the peak, about 30% of the male labor force worked in Edo or other places. (Fig. 11.6) What we know about the demographic history of this village from the basic ﬁgures is as follows: At the beginning of the 200 year period, the type of large farms which depended on subordinate labor was still in existence. Consequently, there were many unmarried persons. The marriage rate and consequently the birth rate were limited to a low level. But at the next stage, small farming became predominant, and the family became the unit for farming. Unmarried persons rapidly disappeared and the marriage and birth rates soared to their maximums. But when the exodus of labor began, the rates went down and the population was stabilized. This population stagnation in the latter half of the Tokugawa period has been frequently spoken of as a Malthusian equilibrium. Perhaps the economic conditions became so unfavorable that the inhabitants of the village had to choose between bringing down their standard of living or controlling the number of births. There was very little room for bringing down the living standard, and this led them to choose the second method. In the ﬁrst place, they began to delay the age of marriage, and then to control the number of births, usually through abortion or infanticide. From 1671 to 1725, married women who survived until age 45 with their husbands had 6.5 births if the women married between ages 15–20, 5.9 births if married at 21–25, and 1.5 if at over 26. In contrast to this, a hundred years later, from 1776 on, women averaged 3.6–3.8 births whenever they married. I think this change was due to the practice of birth control. (Table 11.5)

Figure 11.6 The Proportion of the Population Working Outsides of the Village

197

12:27:16:11:09 Page 197 Page 198

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

Table 11.5 Average Number of Births by Age at Marriage

Age at marriage Average number of births Cases

Females married 1671–1725 under 20 6.5 39 21–25 5.9 7 26–30 1.5 2 over 31 2.3 3 Females married 1726–1775 under 20 4.2 67 21–25 3.4 19 26–30 3.0 6 over 31 1.0 1 Females married 1776–1821 under 20 3.7 50 21–25 3.6 39 26–30 3.8 6 over 31 1.5 2

FRF ANALYSIS Next I want to report the results of tracing the demographic behavior of married couples in the first 100 year period, the period for which the materials are most reliable. In this village, there was a total of 168 married couples in which the wife was born before 1700. Among these, 39 were already registered as married in the first year for which we have materials, 1671, and consequently it is not known at what age they began their married life. There were 56 women for whom the period of married life is known, and who continued to be married until the end of the possible childbearing age at 50. And there were 73 women whose married life ended before the end of the childbearing age. For those whose married life ended before the child-bearing age was over, in 33 cases the husband died, in 28 cases the woman herself died, 11 marriages ended in divorce, and the reasons for the remaining 23 are unclear. After a wife died or was divorced, 80% of the husbands remarried. In contrast to this, the remarriage rate for women after their husband died or divorced them was a very low 20%. If we look at the number of births by age groups of married women, we see that the annual number of births for women aged 16–20 was 0.166, for women aged 21–30 it was just over 0.3 births, for women aged 31–40 just over 0.2 births, and practically none for women over 40. The peak in births is between ages 21–30, and the number falls in a staircase pattern for each succeeding age. The number of births in the under 20 age group is surprising small. The age of marriage and the number of births are naturally correlated, as are the number of years the marriage continued and the number of births, but there is a higher correlation between age and the number of births. Consequently, there is a greater effect from postponing the age of marriage past 20 than by reducing the age of marriage to less than 20. Under the age of 20, as the age of marriage is raised one year the number of births varies by only 0.2. In contrast to this, the number of births varies by 0.3 per year for every year that the age of marriage is raised in the 21–30 age group. There is evidence for this in the fact that as a whole the birth rate fell from

198

12:27:16:11:09 Page 198 Page 199

DEMOGRAPHIC ASPECTS OF A VILLAGE IN TOKUGAWA JAPAN

the second half of the eighteenth century due to the fact that the average age of marriage was raised to over 20. (Fig. 11.7, 11.8 and 11.9) Next, if we look at the interval between births of mothers born before 1700, we see that the average interval from marriage to the birth of the first child is 3.2 years, and the interval between the first and the second child is 3.1 years. The interval between the second and third child was 3.3 years. The average interval between the fifth and sixth child rose to 3.9 years, but after this the gap narrowed to around 3 years. The fact that the order seems to have almost nothing to do with the average interval between births means that women who bore many children continued to bear them at about the same intervals to a fairly advanced age. The fact that the number of births falls as a whole with an increase in age is

Figure 11.7 Age-Speciﬁc Fertility

199

12:27:16:11:09 Page 199 Page 200

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

Figure 11.8 Mean of Births According to the Wives’ Ages at Marriage

due to an increase in the number of women who completely stopped having children. Because there are few examples, it cannot be definitely said that there is a difference in the sex ratio according to the order of birth. Even so, the following extremely interesting trend can be observed. Namely, of first born children, 77 were males and 88 females, a sex ratio of 87.5, while of last born children 74 were males against 55 females, or a sex ratio of 134.5. Gradually the percentage of males increased. It was not possible to prove whether this was due to natural or artificial influences.

A CONCLUSION The example of this village is only one out of tens of thousands in the Tokugawa period, but the population changes for Japan as a whole are almost the same as for this village. It has been considered up to now that the total population of Japan in the beginning of the seventeenth century was 18 million, but this is a completely

200

12:27:16:11:09 Page 200 Page 201

DEMOGRAPHIC ASPECTS OF A VILLAGE IN TOKUGAWA JAPAN

Figure 11.9 Fertility According to the Age at Marriage unscientific estimate. I have shown that this figure, viewed in the light of local historical materials of the first half of the seventeenth century, is an overestimate, and that the population was probably less than 10 million. In 1721, the first national population survey was carried out by the Tokugawa government, and this survey reported a population of 26 million. In any case, from the seventeenth century to the beginning of the eighteenth century, the population of Japan increased rapidly. This increase, to some extent, was due to the increase in the urban population due to the rapid establishment of castle towns. However, the increase in the farm village population was even greater. As can be seen in the case of Yokouchi village, it was possible for the population to triple in a hundred year period. However, after 1721, there was almost no change in the national population according to the Tokugawa government surveys. The survey became increasingly inaccurate, but up until the final survey in 1846 the recorded population fluctuated at around 26–28 million. In 1872, the new household registration system came into existence. Reliable national population figures based on this registration showed a population of 35 million, and thus the population in the latter part of the Tokugawa period was not at all stationary. In certain areas there was continuous growth. However, the growth rate in the latter half of the Tokugawa period was in any case lower than that of the first half. The changes in the national trends and in the trends of Yokouchi village resemble each other. Hasty conclusions should be avoided, but it is my opinion that the various figures drawn from this village can be taken as a national model. We do not

201

12:27:16:11:09 Page 201 Page 202

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

know when the rapid population growth in Yokouchi village began to take place, but it is probable that it began at the same time that the change to small families began to take place, following changes in agriculture. At the same time there was probably an improvement in housing and food, and a rise in the standard of living. Thus the infant and child mortality rate fell and the average life expectancy was lengthened. As a result, the increase in population was greater than the birth rate. No problems arose as long as this increased population was combined with an increase in other production requisites, in the case of this village, with an increase in cultivated land. However, in the second half of the eighteenth century, the area which could be reclaimed into cultivated land reached its limit. Further increases in the population resulted in an exodus of the population to go to work in the cities. Furthermore, the excess in population caused great changes in the demographic history of this village from this time on. The people in Yokouchi were forced to somehow cope with more people than the village could support. Leaving the village to work elsewhere formed one solution. In addition, the age of marriage for females was raised, and to some extent there was a decrease in the percentage of the married. However, we cannot say that the population stagnation in the latter half of this 200 year period resulted from the above factors. In explaining the sharp contrast between the ﬁrst and second hundred years, it is surmised that population control was being carried out. Conversely, the demographic history of this village in the ﬁrst hundred year period is an example of a transitional period in which population control was not necessary. An average yearly growth rate of slightly under 1% is not at all exceptional; rather if other factors remain in balance, this is a normal growth rate. This situation can be seen in other villages which I have analyzed. In these villages also, it was possible to reclaim land for cultivation. In short, the above example shows that when it was not necessary to control the size of the population, and when there were no famines or epidemics, it was possible for an increase of this extent to take place in the pre-industrial society of Japan. I began this type of research into the demographic history of the Tokugawa period using series of shumon aratame cho or religious registrations only a few years ago organizing an informal study group. Collecting and organizing the materials and then translating them into statistics takes a tremendous amount of time, and thus up to now we have only be able to report on the results of a very few analyses. However, if work progresses as expected, it will be possible to obtain a number of studies similar to this one. Because there are numerous good materials, it should be possible for me and the group to report on a number of similar studies in the very near future.3

NOTES

1. Japan from 1600 to 1870 2. It should be noted that both the calendar year and the calculation of age in the Tokugawa period diﬀered from that of the West. The diﬀerences in the calendar have been largely ignored in this paper, but all ages mentioned have been calculated according to the Japanese method. This method is to count a person as being one year of age at birth, and to add one year at the beginning of every

202

12:27:16:11:09 Page 202 Page 203

DEMOGRAPHIC ASPECTS OF A VILLAGE IN TOKUGAWA JAPAN

new calendar year. Thus it is necessary to subtract approximately one year from every age mentioned in order to obtain the age according to the Western method of calculation. 3. These are our selected list of articles written in English. A. HAYAMI, “The Population at the Beginning of the Tokugawa Period.” Keio Economic Studies. Vol. 4, 1966–67. A. HAYAMI, “The Demographic Analysis of a village in Tokugawa Japan: Kando-Shinden of Owari Province, 1778–1871.” Keio Economic Studies. Vol. 5, 1968. A. HAYAMI, “Population History of a village in Pre-industrial Japan.” (to be published in the “DAEDALUS Historical Demography Special Issue.”)

203

12:27:16:11:09 Page 203 Page 204

First published in Keio Economic Studies, vol. 17, no. 1, pp. 1–16, 1980.

12 Class Differences in Marriage and Fertility Among Tokugawa Villagers in Mino Province*

INTRODUCTION rogress in historical demographic analysis of the lives of Japanese villagers in Pthe Tokugawa period (1603–1868) has shown that there are clear differences in their demographic behavior and indices by economic class.1 Professor Thomas C. Smith, for example, shows in his research on the village of “Nakahara,” that there are distinct class differences in age at marriage, life expectancy, family size, and so on (Smith, 1977). According to Susan B. Hanley and Kozo Yamamura’s research on villages in Okayama, there are definite differences in, for example, average age at first and last birth by class.2 The author has demonstrated, in his research on the village of Nishijo in the province of Mino that there is a clear differential by class in migration.3 Villagers by no means acted homogenously in demographic matters. Given this, it is necessary, if at all possible, to investigate the demographic behavior of the villagers by class. We must investigate the extent of this discrepancy—whether it is minimal or whether it is so large as to affect the reproductive rate—quantitatively, through statistical analysis. This paper uses family reconstitution to derive indices of marriage and birth rates from observation of class differences in six villages on the Nobi plain (near Nagoya), site of the author’s present research. Its object is fact-finding rather than analysis, so it explains the amount of differential through various charts and graphs. A more complete analysis, and comparison with previous results, particularly those of life expectancy, must be left until later.

PREPARATION OF DOCUMENTS AND STATISTICS The six villages which form the basis for this study, all in the province of Mino, are the villages of I¯numa, Ena county; Higashikaiden, Katagata county; Kano, O¯¯ no county; Nishijo, Anpachi county; and Arioshinden and Nekojishinden, both in Tagi county. Aside from ¯Inuma, which is located to the east, in a hilly area, the villages are situated on a plain. The last three villages are waju¯ villages, built on land reclaimed from the surrounding fens and protected by dikes. Table 12.1 shows the period covered by the documents, the number of usable years of data,

204

12:27:16:11:09 Page 204 Page 205

CLASS DIFFERENCES IN MARRIAGE AND FERTILITY

Table 12.1 Coverage of the Shu¯ mon aratame cho¯

Years Village Population Covered Usable (beginning–end)

I¯numa 1713–1868 (111)* 331–369 Higashikaiden 1665–1869 (71) 233–311 Kano 1753–1868 (65) 134–168 Nishijo 1773–1869 (97) 366–381 Arioshinden 1685–1800 (80) 76–227 Nekojishinden 1709–1865 (53) 371–579

* Figures in parentheses are number of years the documents are usable.

and the beginning and final year populations for each of the six villages. The earliest usable records are from the end of the seventeenth century and, with one exception, they continue up until about the time of the Meiji Restoration in 1868. For the purposes of family reconstitution, however, large gaps in a series of documents limits their usable span of time. Deficient recording and large numbers of mistakes also restrict their use. The earliest registers of Higashikaiden, Arioshinden, and Nekojishinden are unusable for family reconstitution due to such deficiencies. As one can see from the table of marriage cohorts, Table 12.2, the family reconstitution forms (FRFs) do not include material from the seventeenth century. When, as in Nishijo, the date of marriage is included in the documents, one can construct and classify FRFs for marriage cohorts considerably in advance of the time the records are actually available. Table 12.2 shows numbers of completed and uncompleted families, and cohort of wife’s birth and marriage for each of the six villages. The timespan of cohorts is twenty-five years, but this figure has no intrinsic significance. A total of 1,803 FRFs are used in this paper, and of these the number of completed families is 438, or 24 percent. From a demographic standpoint this number is inadequate. But this amount of material, all from a single area, and all providing information for observing class differences in demographic indices, is the largest we have ever had. One cannot do detailed examination of each period or each village, so this paper will examine the evidence by aggregating the FRFs which cover these six villages over a period of two centuries. In a family reconstitution study, the term “completed family” means a marriage which continues unbroken to the end of the wife’s childbearing period, here defined as 50 sai.4 This gives us detailed information on the measurement of fertility and the number of children born during the period. However, since the length of time covered by the sources is limited, and, for various reasons, there is a large number of incomplete marriages, the number of completed families, as shown in Table 12.2 is small. There is no need to describe in detail the procedure for compiling FRFs from shu¯ mon aratame cho¯ here. I have described the method of assembling the shu¯ mon aratame cho¯ before, elsewhere. Once the Basic Data Sheets have been made, creating FRFs is rather simple. The most usual form of household in the shu¯ mon

205

12:27:16:11:09 Page 205 Page 206

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

Table 12.2 Number of Families Reconstituted, by Village, for Completed and Uncompleted Families, Cohort of Wife, and Marriage Cohort

I¯numa Higashi- Kano Nishijo Ario- Nekoji- Total Kaiden Shinden Shinden

Total of FRF 588 269 62 322 214 348 1803 Completed 170 73 8 63 57 67 438 family Uncompleted 418 196 54 259 157 281 1365 family Cohort of 1626–1650 2 3 5 wife 1651–1675 34 24 58 1676–1700 74 3 3 40 120 1701–1725 72 26 14 20 52 184 1726–1750 66 37 16 49 50 13 231 1751–1775 84 55 20 70 37 55 321 1776–1800 69 54 9 67 8 101 308 1801–1825 95 54 58 111 318 1826–1850 86 40 56 68 250 1851–1875 3 2 5 unknown 3 3 Marriage 1701–1725 47 1 38 86 cohort 1726–1750 72 29 11 21 45 178 1751–1775 54 37 17 52 50 210 1776–1800 83 53 22 70 41 269 1801–1825 75 42 3 61 103 284 1826–1850 95 51 58 79 283 1851–1875 73 52 66 235 unknown 89 13 9 7 40 100 258

aratame cho¯ from these villages is a small family, so conﬁrming who is the child of which couple presents no great problem. The FRFs naturally take married couples as their units of observation, so children of men and women who are not married do not fall into our sample. If their number were large, there would be serious errors in the measurement of fertility. But in Nishijo, one of the six villages, they make up only 69 of the total of 993 births over the entire period. Finally, in dividing the FRFs by class, we have used the class position of the family at the time of marriage. The classes are as follows, by kokudaka: Class I, 10 koku and above; Class II, from 2 koku up to 10 koku; Class III, under two koku. Class I is a class of families which cannot cultivate their own land entirely by their own labor power, but are dependent on hired labor or on leasing their land to tenants. The families in Class III have little or no land of their own, and support themselves by borrowing land from others or by working as hired laborers on others’ land. Class II, the intermediate one, is composed of families which, by and large, cultivate land which they own themselves. Of course this is a rather rough division, and because the range of actual cases is continuous, it is diﬃcult to draw a clear line of demarcation between classes. Another problem is that there is movement between classes. If there were a great

206

12:27:16:11:09 Page 206 Page 207

CLASS DIFFERENCES IN MARRIAGE AND FERTILITY

deal of this, basing our demographic statistics on the class level at the beginning of marriage as listed in the FRFs, would produce erroneous results belying the title of this paper. When we examine the degree of change in class level between the beginning and the endpoint of marriage for the 1,389 couples recorded in the FRFs, we see that, as Table 12.3 shows, there was no change in 76 percent of the cases. Thus, using class of ownership at the beginning of marriage provides a ﬁrm foundation for our examine of class diﬀerences. The number of cases of movement to a higher class is greater than the reverse. This is because there are many cases in this area where households that were established as branches of others and, at their establishment, were mizunomi—that is, households whose own land was nil—later received a parcel of land (or just a house and garden) from their main households.

AGE AT MARRIAGE One can readily imagine that there might be class differences in the age at which men and women marry. For example, more prosperous classes might have the opportunity to marry earlier, and lower classes be forced to marry later. The author has noticed significant difference in average age at marriage for women of different classes in his work on the village of Nishijo. Tables 12.4 and 12.5 show these differences for men and women in the six villages. The cases used in these figures are those where it is clear that the husband’s marriage is his first. In most cases women come from other villages, so the number of samples where one is sure that it is the wife’s first marriage is very small. Here we have used the larger number of cases available to us, and therefore, in order to determine whether there is a difference in quality by class, or how the age at marriage is distributed, our investigation of the wife’s age is not limited to first marriages. (When we look at the relationship between age at marriage and number of births we will examine their class difference by wife’s age at first marriage.)

Table 12.3 Change in Class Standing, Beginning and End of Marriage

Class Number Percentage

I–I 112 73.2% I–II 29 19.0 I–III 12 7.8 II–I 20 5.2 II–II 274 71.5 II–III 89 23.2 III–I 9 1.1 III–II 173 20.2 III–III 673 78.7 Total 1389 Upwards 202 14.5 Downwards 130 9.3 No change 1059 76.1

207

12:27:16:11:09 Page 207 Page 208

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

Table 12.4 Age at Marriage (male)

Number Age Total Class I II III

14 1 1 2 15 1 1 16 3 3 17 1 4 4 9 18 1 5 12 18 19 3 7 13 23 20 1 9 18 28 21 4 11 24 39 22 5 12 27 44 23 7 12 23 42 24 5 14 29 48 25 8 12 30 50 26 10 13 28 51 27 9 24 38 71 28 11 15 34 60 29 9 18 21 48 30 5 14 25 44 31 8 13 24 45 32 3 12 20 35 33 3 11 17 31 34 4 13 19 36 35 2 8 12 22 36 2 4 20 26 37 4 2 12 18 38 1 3 8 12 39 1 4 2 7 40 1 2 3 6 Over 40 2 8 22 32 Total 111 251 489 851 Mean 27.9 28.1 28.1 28.1 S.D. 4.99 5.35 5.48

Table 12.4 shows that there is no significant difference by class in age at marriage for men. The modal age is 28 sai for men in class I, 27 sai for Classes II and III. Men’s behavior is homogenous. However, in Table 12.5, which shows age at marriage for women, differences by class are evident. There is a three-year difference in age between Class I and Class III, and the value for Class II is situated directly in between these. The mode, with Class I at 17 sai, Class II at 18 sai, and Class III at 20 sai, parallels the mean completely. There are significant differences in age at marriage for women by class, and it is clear that the higher class marries earlier and the lower class later. Analysis of labor migration (dekasegi) in Nishijo explains this by showing that migration to cities by women of the lower classes is prevalent, and the practice

208

12:27:16:11:09 Page 208 Page 209

CLASS DIFFERENCES IN MARRIAGE AND FERTILITY

Table 12.5 Age at Marriage (female)

Number Age Total Class I II III

11 1 1 12 3 3 13 2 3 5 14 4 7 9 20 15 6 10 11 27 16 8 14 13 35 17 20 23 27 70 18 14 33 36 83 19 14 29 39 82 20 6 28 50 84 21 7 18 44 69 22 9 22 47 78 23 8 14 34 56 24 6 7 36 49 25 1 8 18 27 26 3 4 23 30 27 1 6 20 27 28 6 13 19 29 3 20 23 30 6 7 13 31 2 11 13 32 2 3 5 33 1 3 4 34 1 3 4 8 35 3 4 7 36 1 3 4 37 2 2 39 2 40 1 Over 40 1 1 2 Total 111 251 487 849 Mean 19.4 21.0 22.4 21.6 S.D. 3.99 4.78 5.06

of marrying only after one returns to the village is more common than among women of higher classes.5 This serves as a kind of population limitation among lower-class families in the village. Differences in the relative age of husband and wife arise from these class differences. Table 12.6 shows the difference, by class, between wife’s and husband’s age at his first marriage. As one can see, the mean disparity varies by class. The husband is most often eight years older in Class I, nine years older in Class II, and four years older in Class III, so that the age difference between spouses is wider for the higher class and narrower for the lower classes. In Class I there are only two cases, 1.8 percent of the total, where the wife is older than her husband. In Class II there are 18, 7.7 percent, and in Class III, 49 examples, coming to 10.4

209

12:27:16:11:09 Page 209 Page 210

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

Table 12.6 Age Diﬀerence of Husband and Wife, by Class

Diﬀerence (years) Class I II III Total

−12 1 1 −10 1 1 − 933 − 733 − 611 − 522 − 4257 − 34812 − 2 7 11 18 − 1251421 0162229 + 1282939 + 2 6 11 27 44 + 3492942 + 4 7 12 49 68 + 5 7 19 32 58 + 6 8 18 43 69 + 7 9 21 29 59 + 8 11182655 + 9 8 22 28 58 +109162146 +117141233 +12 4 9 16 29 +13 4 4 17 25 +14 7 7 13 27 +15 4 4 6 14 +161359 +171135 +18 1 7 7 15 +19 1 2 3 +204127 Over +20 1 3 7 11 Total 109 233 472 814 Mean +8.8 +7.3 +5.9 +6.7 S.D. 4.79 5.03 5.30

percent of the total number of cases. In the highest class, a wife being older than her husband is exceptional; in the lower class where more than a tenth of the women are older than their husbands, this nicety is unimportant.

DURATION OF MARRIAGE Table 12.7 shows the distribution of the duration of marriage by class for the 609 couples whose beginning and ending dates of marriage can be ascertained. Since the records of the villages of ¯Inuma and Kano include couples who have left the village permanently, these two villages are ignored and the results are based on data from the other four villages.6 There is a signiﬁcant diﬀerence of about four

210

12:27:16:11:09 Page 210 Page 211

CLASS DIFFERENCES IN MARRIAGE AND FERTILITY

Table 12.7 Duration of Marriage, by Class (number/cumulative total)

Years Class I Class II Class III Total

1 5/97 4/160 29/314 38/571 2 4/93 5/155 19/295 28/543 3 5/88 7/148 6/289 18/525 4 2/86 6/142 14/275 22/503 5 2/84 5/137 14/261 21/482 6 /84 4/133 7/254 11/471 7 4/80 2/131 12/242 18/453 8 4/76 4/127 8/234 16/437 9 1/75 4/123 13/221 18/419 10 3/72 3/120 9/212 15/404 11 5/67 3/117 8/204 16/388 12 2/65 8/109 10/194 20/368 13 3/62 2/107 9/185 14/354 14 3/59 3/104 10/175 16/338 15 /59 3/101 5/170 8/330 16 3/56 5/ 96 6/164 14/316 17 1/55 3/ 93 5/159 9/307 18 1/54 3/ 90 7/152 11/296 19 1/53 1/ 89 9/143 11/285 20 1/52 / 89 3/140 4/281 21 /52 4/ 85 4/136 8/273 22 2/50 1/ 84 4/132 7/266 23 4/46 8/ 76 5/127 17/249 24 3/43 3/ 73 12/115 18/231 25 2/41 3/ 70 6/109 11/220 26 2/39 3/ 67 9/100 14/206 27 5/34 3/ 64 6/ 94 14/192 28 1/33 5/ 59 6/ 88 12/180 29 1/32 3/ 56 3/ 85 7/173 30 3/29 8/ 48 4/ 81 15/158 31 3/26 4/ 44 2/ 79 9/149 32 3/23 4/ 40 8/ 71 15/134 33 3/20 2/ 38 7/ 64 12/122 34 1/19 2/ 36 2/ 62 5/117 35 /19 1/ 35 6/ 56 7/110 36 3/16 / 35 6/ 50 9/101 37 2/14 2/ 33 5/ 45 9/ 92 38 2/12 2/ 31 5/ 40 9/ 83 39 1/11 8/ 23 6/ 34 15/ 68 40 1/10 1/ 22 6/ 28 8/ 60 41 /10 2/ 20 3/ 25 5/ 55 42 1/9 3/17 4/21 8/47 43 1/8 3/14 3/18 7/40 44 1/7 2/12 2/16 5/35 45 1/6 1/11 2/14 4/31 46 3/ 3 2/ 9 1/ 13 6/ 25 47 /3 3/6 2/11 5/20 48 / 3 1/ 5 5/ 6 6/ 14 49 1/ 2 1/ 4 / 6 2/ 12 50 / 2 2/ 2 / 6 2/ 10

211

12:27:16:11:09 Page 211 Page 212

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

Table 12.7 (cont.)

Years Class I Class II Class III Total

51 /2 /2 1/5 3/9 52 / 2 1/ 1 / 5 3/ 8 54 1/ 1 / 1 2/ 3 2/ 5 55 /1 /1 1/2 1/4 57 / 1 1/ 0 / 2 2/ 3 58 / 1 1/ 1 1/ 2 65 / 1 1/ 0 1/ 1 68 1/ 0 1/ 0 Total 102 164 343 609 Mean 21.2 22.3 18.6 20.1 S.D. 14.55 14.38 14.35

Figure 12.1 Duration of Marriage, by Class.

years between Class II and Class III (but not signiﬁcant between Class I and Class II). What can be termed the “curve of survival rate” for duration of marriage is shown in Figure 12.1. We can discover several important facts by inspecting this graph. First, Class II’s “marriage survival rate” is generally greater than Class I’s, but they change at about the same rate. In contrast to this, the level of Class III is rather low, and, particularly at a duration of twenty years, becomes 14 percent lower than Class I. About 60 percent of marriages in Class III end in less than twenty years, and the conditions of marriage are less stable than in the other classes. Couples whose marriages endure for 25 years—that is, until their silver wedding anniversary—make up 40.2 percent of Class I, 42.7 percent of Class II, and

212

12:27:16:11:09 Page 212 Page 213

CLASS DIFFERENCES IN MARRIAGE AND FERTILITY

31.8 percent of Class III. Those who celebrate their golden wedding anniversary marking fifty years of marriage number only about 2 percent in each class. One can explain part of this difference in duration of marriage by class by the clear differences in the wife’s age at marriage. Wives of Class III marry, on the average, four years later than those of Class II. But the difference cannot be due solely to this. I think that the reason that most marriages in Class III end within twenty years is that poverty promotes their instability. Table 12.8 shows the class status of each of the partners in a marriage. In assigning men and women to classes it was possible to include only those examples where it was clear into which class each person was born—that is, only those where both spouses married within the same village. For both the men and women, cases where the spouse was originally from the same economic class are most common. The number of most divergent cases, where one of the spouses is from Class I and the other from Class III, is only a small part of the total. One can observe some marriages between the neighboring Classes I and II, but marriage within the same class accounts for about 80 percent of all marriages in Class III. One must say that the opportunity to rise out of Class III by marriage was very limited.

BIRTHS AND FERTILITY Family reconstruction is, to begin with, a technique developed in order to measure fertility, and its greatest importance lies in the possibility of creating the crucial index measure of age-speciﬁc fertility. However, though it is usually so treated, fertility as reﬂected in the shu¯ mon aratame cho¯ is in fact not actual fertility. Stillbirths are not recorded in the registers, of course. Also, since the documents were complied only once a year, persons born after the date when registers were compiled in one year, and dying before the next year’s registration date, are not counted either. It is reasonable to assume that infant mortality was rather high, and consequently, the number of births that could have been counted in the registers is somewhat lower than the actual number of births. We do not have very much evidence about how great this underestimation was, but we can obtain an estimate of the number of concealed births and deaths based on Hiroshi Kito’s research using a kainin kakiagecho¯—a document listing pregnant wives to whom

Table 12.8 Comparative Class Status of Husband and Wife

Class of Husband Total I II III

Class of wife I 18 13 3 34 II 17 31 33 81 III 1 26 141 168 Total 36 70 177 283

213

12:27:16:11:09 Page 213 Page 214

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

a sort of subsidy should be given—from a village in northeastern Japan in the early nineteenth century (Kito, 1976). According to Kito, there were 271 births in this village from 1808 through 1826, of which 21 were stillbirths. Of the remaining 250 births, 241 were unquestionably live births, and of these 41 infants died within the first year of life. For the 32 of these whose month and day of death is known, 16 died within four weeks of birth, 9 died between four weeks and three months, 4 between three and six months, 2 between six and nine months, and one between nine months and a year of birth. The total calculation shows that 29 infants died within their first six months of birth, 12 percent of the total number of births. If we include those whose date of death is unknown, we can presume that about twenty percent of all live-born infants died within six months. If we use these figures, acknowledging that it requires a certain amount of boldness to vault over the spacial and temporal distance and apply these to a different location in Tokugawa Japan, and recognizing that, at this point, these are the only figures calculated directly from such sources that we have, the following inferences are possible. Suppose there are 1000 actual births. Of these, 200 die within the first six months of birth. Eight hundred persons remain. If we assume that there is no seasonal bias in births, these 800 persons will be the figures that we will capture as “births” in the shu¯ mon aratame cho¯ from its compilation in one year to its compilation in the next. In comparison to the 800 “births” listed in the records, there are 200, that is 25 percent, births and deaths which go unrecorded. For this reason it is necessary to adjust the number of births upwards by about 25 percent in calculating fertility. Needless to say, one must also make adjustments in the mortality figures. It is incumbent on a researcher to make such adjustments when comparing these figures with fertility figures from a modern census, but such an adjustment is not necessary when comparing the results with other calculations based on the shu¯ mon aratame cho¯. In this paper no adjustment has been made in either numerical values or fertility. Figure 12.2 is a graphic representation of Table 12.9. It shows directly that, with the exception of the youngest ages, the fertility of Class I is considerably higher than that of Classes II or III. Was the fertility of Class I in fact actually higher than that of the other two? High infant mortality rates, which varied by class, could result in such a difference. Population control may have been greater in Classes II and III. At the present time the author cannot give a definite answer as to which of these two possibilities is the more likely. But it is clear that there are obvious class-based differences in “fertility” as it is expressed in the shu¯ mon aratame cho¯. These differences in fertility are reflected in the distribution of number of births to couples in each class. The 441 couples with completed families are divided into three classes and the distribution of number of births per class is shown in Table 12.10. On the average, couples of Class I have 1.7 to 1.9 more children than those in the other two classes. Table 12.11 gives, for each class, an estimate of the number of children born to completed families where the wife marries between 16 sai and 30 sai, based on the age-specific fertility given in Table 12.9. At the same age at marriage, there is no conspicuous difference between classes in the number of births. However, based on his own previous research, the author believes that 4.5 births or more

214

12:27:16:11:09 Page 214 Page 215

CLASS DIFFERENCES IN MARRIAGE AND FERTILITY

Figure 12.2 Age-speciﬁc Fertility by Class.

Table 12.9 Age-speciﬁc Fertility by Class

Age Group

Class Under 15 16–20 21–25 26–30 31–35 36–40 41–45 46–50 Total

I 0 0.218 0.324 0.291 0.242 0.174 0.088 0.025 0.215 (0) (60) (162) (146) (103) (68) (28) (6) (573) II 0.095 0.224 0.302 0.233 0.199 0.157 0.073 0.010 0.176 (2) (97) (315) (287) (240) (171) (68) (8) (1188) III 0.154 0.228 0.282 0.251 0.201 0.151 0.080 0.020 0.176 (8) (133) (492) (613) (510) (336) (147) (30) (2269) Total 0.111 0.224 0.295 0.250 0.205 0.155 0.079 0.018 0.181 (10) (290) (969) (1046) (853) (575) (243) (44) (4030)

per family are necessary to maintain the reproduction of the family unit in agricultural villages at this period. Hence, if the woman does not marry at 24 sai or earlier in Class I, or at 22 sai or earlier in Classes II and III, the completed family will not be large enough to maintain itself as a unit. However, as we have shown above, marriage is earliest in Class I, and latest in Class III. In Classes II and III, as shown in Table 12.5, the average age at marriage is close to the limit at which the family can be maintained. On the other hand, since the average age at marriage is much lower in Class I, 19.4 sai, one family can have six or more births. On the average, the number of children exceeds the number necessary to reproduce the family, so a household can establish branch households or supply husbands for daughters of households without their own

215

12:27:16:11:09 Page 215 Page 216

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

Table 12.10 Number of Births by Class

Class Number of Births Total I II III

0 1 15 38 54 1292940 2 2 13 29 44 3 2 25 31 58 4 7 13 27 47 5 7 28 39 74 6 4 17 32 53 7 8 11 21 40 8 42814 9 2248 103148 110011 Total 42 136 263 441 Mean 5.6 3.9 3.7 3.9 S.D. 2.46 2.28 2.59

Table 12.11 Number of Births to Completed Families by Wife’s Age at Marriage

Class

Age at marriage I II III

16 6.8 6.0 6.1 17 6.6 5.8 5.8 18 6.4 5.5 5.6 19 6.2 5.3 5.4 20 5.9 5.1 5.2 21 5.7 4.9 4.9 22 5.4 4.6 4.6 23 5.1 4.3 4.4 24 4.7 4.0 4.1 25 4.4 3.7 3.8 26 4.1 3.4 3.5 27 3.8 3.1 3.3 28 3.5 2.9 3.0 29 3.2 2.7 2.8 30 2.9 2.4 2.5

sons. The combination of differing fertility and differences in age at marriage results in considerable variety in the distribution of number of births by class, as Table 12.10 shows. In determining the total number of births one problematic factor is how long a wife continues to bear children. This is expressed as the wife’s age at her final childbirth. If there were clear differences in the average age by class, it would

216

12:27:16:11:09 Page 216 Page 217

CLASS DIFFERENCES IN MARRIAGE AND FERTILITY

be incumbent on us to include this in our consideration of the factors which influence fertility. One would expect that the age would be higher in the upper class and lower in the lower one. But in actual calculations it is difficult to discern any significant differences among the three classes. Our figures for the completed families in our sample show 39.1 sai for Class I, 38.1 sai for Class II, and 39.0 sai for Class III. At least for completed families, there is no process which equal- izes the number of births by class through changes in the age at final birth. However, since there is a difference in duration of marriage by class, uncompleted families must also be included in our consideration of this question. Excluding the cases where the wife was under 50 sai and in a continuing marriage in the final year of the documents, there is a total of 872 couples in this sample. Age at last birth is 33.8 sai in Class I, 34.3 sai in Class II, and 34.9 sai in Class III. The age at last birth is younger than the previous results, but one still cannot perceive any significant difference due to class. We can conclude that population was not limited by altering the wife’s age at last birth, or, if it was, it did not vary by class.

CONCLUSION The scope of this paper is limited, but its aim has been to examine whether there are class differences in demographic indices by examining the FRFs. The only measure in which there is no differential is age at marriage for men. We have detected significant class differences in both women’s fertility and their age at marriage. In addition, there are clearly qualitative differences which result in a greater number of births in the higher classes and a fewer number in the lower ones. This shows that there was no limitation on the number of children in the economically prosperous classes; conversely, limitation did occur in the poorer ones. Limiting the number of births by delaying marriage is what Malthus terms a preventative check. The low fertility involved here also suggests that the positive checks of abortion and infanticide were also operating. Let us consider which of these checks was more often employed by the poorer classes. It is clear from Table 12.5 that there was a difference of exactly three years in age at marriage, in both the mode and the mean, between Class I and Class III. Turning to Table 12.11, in Class I, when age at marriage is 19 sai, the number of births is 6.2. In Class III, when it is 22 sai, there are 4.6 births. Three years difference in age at marriage is accompanied by a difference of 1.6 in the number of births. If one holds age constant, there is a difference of 0.8 births between the two classes. A positive check of some sort causes this difference. Within a single class, as well, three years difference in age at marriage is accompanied by a difference of 0.8 in the number of births. That is to say, the proportions of the decrease in number of births attributable to the preventative check is exactly equivalent to the proportion attributable to the positive check. One cannot imagine that there was any factor which deliberately caused these two values to be consistent with each other. However, one thing remains to be said. This is that the poorer villagers in these six villages were consciously limiting the number of children they had. They did it through these two checks, positive and preventative. Villagers of the lower class, Class III, held the number of births down to the very limit at which the family unit

217

12:27:16:11:09 Page 217 Page 218

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

could be reproduced, to the extent that there is a fair number of households which failed to produce an heir and became extinct. In the prosperous class, Class I, there was no need for population limitation, and, at least in the village of Nishijo, there is not a single case of a family of this class becoming extinct.7 Our examples come from only a few of the fifty or sixty thousand villages which existed in Japan during the 270 years of the Tokugawa period. The problem of whether these examples are representative of the entire country remains. But we can now take the statement that the villagers exhibited different demographic behavior dependent on the class to which they belonged as an established fact. Further research must take full account of this and be concerned with defining its limits

NOTES

* Translated by Laurel L. Cornell, Department of Social Relations, Johns Hopkins University. 1. Records of kokudaka, listings of the value of housing and arable land owned by households within a village, and expressed in terms of measures of rice, are generally used as indicators of the economic status of Tokugawa-era villagers. In most cases these values reflect land surveys carried out in the last decade of the sixteenth or the earlier part of the seventeenth century. 2. Hanley and Yamamura (1977), Table 9.5, page 236. 3. Hayami (1973a), Tables I and II, page 8. 4. The author uses the Japanese method of calculating age, expressed as sai, throughout. A child is considered one year old at its birth; at its first New Year’s Day, and at each subsequent New Year’s, it gains one year in age. Hence there may be a discrepancy of up to twelve months between a person’s age counted by the Japanese and by the Western method. Subtracting a year from the Japanese age will suffice for general purposes. 5. Hayami (1973b), Table 1, page 180. 6. Two different methods were used in constructing the shu¯ mon aratame cho¯; the principle of permanent residence and the principle of actual residence. In the first method, persons born in a village continued to be registered in the shu¯ mon aratame cho¯ there, even if they moved to another village. There were probably not very many cases where villagers in the Tokugawa period left their place of permanent residence, but in villages where the shu¯ mon aratame cho¯ are constructed on the permanent residence principle, one sometimes finds listings of men and women, or couples who are extremely elderly and who are actually not living in the village, or even alive. Registers complied on the latter principle do not have this defect and show the population which actually lives in the village. The four villages used here are of this type. Since these documents are especially useful for the purpose of creating population statistics, in future I hope to search actively for them. 7. For class differences in the formation of branch households and the extinction of households in the village of Nishijo, see Hayami (1973a), Table IX, page 17.

REFERENCES Hanley, Susan B., and Yamamura, Kozo (1977). Economic and Demographic Change in Preindustrial Japan, 1600–1868, Princeton: Princeton University Press. Hayami, Akira (1973a). “Labor Migration in a Pre-industrial Society: A Study Tracing the Life Histories of the Inhabitants of a Village,” Keio Economic Studies, Vol. X, No. 2, 1–17. —— (1973b). “Jinko¯gakuteki shihyo¯ ni okeru kaisokan no kakusa,” Tokugawa Rinseishi Kenkyu¯ jo, Kenkyu¯ kiyo¯, 178–193. Kito, Hiroshi (1976). “Tokugawa jidai no¯son no nyu¯ ji shibo¯,” Mita Gakkai Zasshi, Vol. LXIX, No. 8, 88–95. Smith, Thomas C. (1977). Nakahara: Family Farming and Population in a Japanese Village, 1717–1830, Stanford: Stanford University Press.

218

12:27:16:11:09 Page 218 Page 219

First published in Keio Economic Studies, vol. 10, no. 2, pp. 1–17, 1973.

13 Labor Migration in a Pre-industrial Society: A Study Tracing the Life Histories of the Inhabitants of a Village *

INTRODUCTION or demographic studies on Tokugawa Japan the shu¯ mon-aratame-cho¯ F(宗門改帳)1 is the most fundamental document for use as evidence. These registers are the product of the strict ban on Christianity enforced by the Tokugawa Government, originating with the policy that family temples certify that all persons living in Japan were of the Buddhist faith. As this registration was conducted with regard to all residents, every year in policy, the resulting records are in the nature of a family register, which serves us as the basic data for research in historical demography. A large number of the shu¯ mon-aratame-cho¯ give the names, ages, and positions of individuals in their households and note in detail any changes in the households such as births, deaths, marriages, or moves. Occasionally, but only very rarely, the assessed amounts of land (valued in terms of rice) and the number of livestock are also recorded and this is considered the most important material for determining the status of the people at that time. If a long-run time series of these registers exists for a village or town, its value is especially high. Such materials can be used for a basic demographic analysis and also for more sophistcated research, such as family reconstitution. We can also use them for the analysis undertaken in this article.2 In a good time series of shu¯ mon-aratame-cho¯, we can observe the entire process from entry to exit of all individuals registered and various behavior on their part recorded together with their familial environment. This information we can gather in suﬃcient quantity to classify and then test statistically. Family reconstitution is a type of tracing survey, and the demographic data drawn from this analysis are indeed abundant and high in factual reliability, but it concentrates on marriage and birth. Migration, another important aspect of demographic behavior, is scarcely touched upon by family reconstitution. If we trace an individual through every recorded experience in his lifetime from birth to death, however, migration is naturally placed within the sphere of observation. In this article we will examine migration in Nishijo¯ Village, Ampachi County, in the Province of Mino. This will be combined with a conventional static demographic analysis of the village and family reconstitution. Our object is to show the signiﬁcance of migration as found in this village. The population treated here is limited to this one village, and though this limitation may substantially restrict the conclusions to be drawn, the study of the years for which national statistics do

219

12:27:16:11:09 Page 219 Page 220

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

not exist would be impossible without gathering individual analyses from diﬀer- ent areas.

NISHIJO¯¯ VILLAGE Nishijo¯ Village is a farming village located on the No¯bi Plain in central Japan. During the latter period of the Tokugawa regime this village was ruled by the

220

12:27:16:11:09 Page 220 Page 221

LABOR MIGRATION IN A PRE-INDUSTRIAL SOCIETY

O¯¯ gaki han (藩, domain) from the castle town of O¯¯ gaki, located about 4 miles northwest of the village. This area is in the delta of three rivers, the Kiso, Nagara, and Ibi, and is called Waju¯ (輪中, literally, “in the ring”) from its unique topography of hamlets and ﬁelds encircled by high embankments. The soil here is fertile and in addition to rice, the chief crop, cotton, rapeseed, and other cash-crops were cultivated, but no data on output exist. Fortunately for us, the shu¯ mon-aratame-cho¯ of this village are in a complete series of 97 years without any break whatever, between 1773 and 1869.3 In

221

12:27:16:11:09 Page 221 Page 222

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

addition, the contents of the registers are recorded in minute detail and with the greatest quantity of information we could possibly expect from historical material of this kind. Any change in residence, reasons for such, and destinations are all set down and even post-migration movements are described. In using these historical materials which exist in long time-series, we have created a basic form for recording all of the facts known about one individual. This form we have called the Personal History Tracking (PHT) and a copy is included here. The items recorded on the PHT are each individual’s code number, the reasons for his or her entry and exit from the documents, his name, the family he belongs to, any change in his position in the family, marriages and divorces, births and deaths in the family, adopted children, and any other information we have on his demographic behavior. Every item above is added with the year when change occurred and the age of the individual at the time. The back of the form is used mostly for organizing the information, and so that we might ﬁnd at a glance the year and duration of an event which occurred in the life history. We used various colors, enabling us to make a visual classiﬁcation. Gathering a large number of PHTs made in this way, we appropriately classify and statistically arrange them in order to observe and analyze the behavior of ordinary people living in a pre-modern society. This is what we, the Keio Group, have as our basic objective. Through the usual methods of analyzing the shu¯ mon-aratame-cho¯, we obtained the following information on Nishijo¯ Village during the Tokugawa period. The total residential population in 1772 was 366 (181 men and 185 women); in 1869, it was 381 (200 men and 181 women). The number of households decreased from 93 to 78. Though there was no drastic change in population at either end of the series, (as seen in Fig. 13.1) a big decrease occurred at the end of the 1780s and again in the latter half of the 1830s. These two drops in the population correspond to the second and third of the three great natural disasters of Tokugawa Japan. The population decrease in the 1780s is called the Tenmei

Figure 13.1 Trends of Population and Household Size.

222

12:27:16:11:09 Page 222 Page 223

LABOR MIGRATION IN A PRE-INDUSTRIAL SOCIETY

Famine which occurred when poor crops continued for several years. During the 5 years from 1785 to 1789, 51 persons died in Nishijo¯ and the village population dropped from 361 to 299, a decrease of about 17%. A decrease in the 1830s, called the Tempo¯ Famine, was due to a poor harvest and an epidemic in the years 1837–1838. It is recorded that 28 persons died in the village in 1837, with the crude death rate rising as high as 90‰. This is the highest rate for a single year in the period covered here. Except for these two big drops, the movement of the population was rather slow. However, after the drop at the end of the 1830s, the population recovered at a high rate of increase with an annual average of 1.2%. Despite the hardly discernible trends in the total population figures for the village, the birth rate was much higher than the death rate.4 The total number of births during this period was 993, while there were only 722 deaths. This difference of 271 would be a natural increase. The average crude birth rate was 31.9‰, while the average death rate was 23.2‰. Despite a large natural increase, however, the total population of this village did not show any significant increase because much migration took place.

MIGRATION IN THE TOKUGAWA PERIOD In the descriptive histories written to date, the peasants in the Tokugawa period are often reported to have been bound to their land and legally prohibited from migrating. Indeed, there were some institutional restrictions on their movement and the geographical scope of their lives must have been much smaller than that of the present. But this does not mean that peasants spent their lives in the village where they were born or even in the very limited neighborhood of their native village. On the contrary, the analysis of the basic historical material we are now undertaking shows that far more migration took place than would be expected. Though some institutional limitations existed, it is even doubtful how strong these were in reality. Moreover, peasants sometimes migrated without following any of the required procedures, and once they entered into a big city, they seemed never to have been caught. We might think that the limitations on migration would have been due to an underdeveloped system of transportation. There were no means of public transportation on land, and thus ordinary people had to walk. As a consequence, long distance migration took many days, much money, and was always very fatiguing. However, migration within a ten days scope of time was not infrequent, and if the migration was to a big city the scope was somewhat enlarged. Here we will be discussing the migration of labor, not migration for marriage or adoption. Both kinds of labor migration to be dealt with below are expressed in the term dekasegi (出稼, literally, “leaving home to work”). This term covers both permanent and temporary migration for two reasons: ﬁrst, the meaning of the word is not permanent but temporary migration and peasants were expected to return home in the future. One of the reasons why such an expression was adopted was due to the institutional limitation on making permanent moves to some other domain. The second reason is that Japanese labor has tended to work at certain places for a certain period and then return to their native places, a trend

223

12:27:16:11:09 Page 223 Page 224

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

which has existed up to very recent years. All migration of labor was witten as dekasegi or working temporarily at so and so’s house as servants. In truth, however, there was a mixture of permanent and temporary migrants categorized under this one term, but without looking at the long-term records of individuals, we cannot distinguish permanent from temporary migration. The causes of labor migration can be explained largely by economic reasons. First, most of the consuming expenditures of the Tokugawa Shogunate and the daimyo (大名) took place in the cities and towns and a large number of merchants and manual laborers gathered here. As the living conditions in cities and towns were, on the whole, bad, especially the ability to withstand epidemics and famine, urban populations could not maintain themselves without a constant inflow from the rural areas. Though the historical demography relating to cities is almost completely undeveloped, from the survey conducted by the Tokugawa Shogunate from 1843 to 1867 we know that 25% to 33% of the common people of Edo were born outside the city.5 Consequently, cities by their very existence caused migration to increase, and if economic activities increased and the city population grew larger, this increase was drawn from rural areas.6 On the other hand, two phases can be discerned for inter-village mobility. The first is the type widely seen up to the eighteenth century, that of the genin (下人) or servant, who migrated to take up long-term employment—in some case lifelong service. The existence of this type of labor is thought to be traced back to ancient times, but early in the Tokugawa Period it gradually began to disappear. This was due to the change in the type of farm management, which could be seen on a nationwide scale.7 This conversion to small-scale management which depended chiefly on the labor of the immediate family members was an important change which characterized the seventeenth century. After the conversion to small scale management was completed, the second phase of migration set in. This change was due to the development of rural handicraft industries which needed a large labor force and to the cultivation of cash crops as side work which also induced migration. Labor was supplied by contract, the term of which was usually a year, but it could often be extended by several years by revising the contract. The Japanese term for these persons who worked under yearly (or term) contracts is nenki-ho¯ko¯nin (年季奉公入). In consequence, labor migration of this type continued pari passu with industrial development, and labor supply and labor demand areas were both created. Of course, in a well-to-do farmer’s household, there were always men and women who were employed as domestic servants on long-term contracts, but because in one village there were no more than two or three of these households, in terms of numbers these were not important. Migration for non-economic reasons included services in samurai (侍) families. This meant a demand for domestic servants, but such service was sometimes compulsory. This article deals mainly with the dekasegi of men and women from this village, but this does not mean that no men and women came to work in this village. We will also treat this subject to the extent necessary.

224

12:27:16:11:09 Page 224 Page 225

LABOR MIGRATION IN A PRE-INDUSTRIAL SOCIETY

THE TRACING OF LIFE HISTORIES The Dekasegi rate The percentage of persons who experienced going away from home to work from among all persons born and brought up in the village can be called the dekasegi rate. Obtaining this rate presents certain difficulties. In order to prove that a villager died without ever leaving his village to work, we must be able to trace his activities from birth to death. The period which the documents cover is only 97 years and thus there are only a few cohorts whom we can trace to their deaths. However, as very few people left home to work in their old age, the cohort group born between 1773 and 1825 can be used for our examination as the villagers born in 1825 would be 45 years old in the last document.8 The population included in this cohort group consists of 244 men and 250 women, of whom 62 boys and 60 girls died before reaching age ten. If these deaths are subtracted, 182 men and 190 women remain. Of these, 87 men, or 48% of the 182 men, and 117 women, or 62% of the 190 women, left on dekasegi. Though there is some difference between the figures for men and women, the dekasegi rate can be said to have been far higher than we expected, and we are ignoring here people who went to work within the village. Among the reasons why so many villagers went out of the village to work were the facts that this village had many cities in the surrounding areas, the rural handicraft industry had developed near here, and the village was located on a plain so that it was comparatively easy for villagers to move. Also, the village happened to have a high birth rate. When we examine the households by class, how do the dekasegi rates vary? Fortunately, the land held by every household is recorded in the shu¯ mon-aratame- cho¯ of this village in terms of the assessed rice output. However, because we do not know how much land households held outside their own village, landholding by household cannot be accurately classified. But if we use the information

Figure 13.2 Rural Dekasegi by Direction and Distance.

225

12:27:16:11:09 Page 225 Page 226

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

we have, households can be classified into (a) landowners, (b) owner cultivators, (c) owner-tenant cultivators, and (d) tenants.9 Using this classification for each individual at birth, the dekasegi rate of each class can be shown as in Table 13.1. For the men, the dekasegi rate of each the tenant group is far higher than in any other class, but among the other three classes there exist no significant differences. For women, the rate is highest for the tenant group, and the rate gradually decreases with each group, with the lowest for the landowner class. The rate among the tenants is 63% for men and 74% for women, and thus we know that the principal source of dekasegi labor was this class. Table 13.2 shows the proportion of households in each classification which have members who went out on dekasegi at three points in time: 1780, 1820, and 1860. This table also shows that the highest dekasegi rate is found in the tenant households.

The Age of Leaving on Dekasegi At what age did the villagers start going on dekasegi? In order to observe the conditions on the labor supply side and on the labor demand side, we shall divide all the households into tenants and non-tenants. Then, we shall divide the initial dekasegi destinations into cities (including towns) and villages. The ages at first dekasegi are shown in this classification in Table 13.3. Generally speaking, the mean age in tenant households was rather low; some left home for work as young as age six. Slightly fewer went to rural areas than to cities. The lowest age at first dekasegi is found among tenants who went to rural areas, while the highest was for non-tenants who went to the cities.

Dekasegi Destinations If the initial destinations for dekasegi are divided into cities and rural areas, we can see in Table 13.3 that about an equal number of men went to each, but a larger number of women went to rural areas than to cities. The villagers from the non-tenant classes went primarily to big cities.

Table 13.1 Diﬀerential Dekasegi Rates by Landholding Status of Household

Status Male, % Female, %

Landowner 39.4 32.5 Owner cultivator 29.6 59.1 Owner-tenant cultivator 27.8 64.7 Tenant 63.1 74.0

Table 13.2 Percentage of Households Whose Members went out on Dekasegi

Status 1780, % 1820, % 1860, % Total, %

Landowner 33 33 60 39 Owner cultivator 30 29 40 32 Owner tenant cultivator 50 50 45 48 Tenant 46 63 56 54

226

12:27:16:11:09 Page 226 Page 227

LABOR MIGRATION IN A PRE-INDUSTRIAL SOCIETY

Table 13.3 Age at the First Dekasegi by Destination

Tenant Non-tenant

Destination Urban Rural Urban Rural

1) Male Range (age) 8–47 6–29 8–33 13–18 Mean 15.9 13.6 16.4 15.8 Mode 11 11 –– Sample size 48 46 20 4 2) Female Range (age) 6–24 7–25 9–21 10–18 Mean 14.7 13.4 14.7 14.8 Mode 14 13 –– Sample size 42 73 36 5

Table 13.4 Distribution of Dekasegi by Destination

Destination (man-years) Distribution (percent)

Rural Town City Total Rural Town City

(1) Male 1773–1800 309 25 540 874 35.4 2.9 61.8 1801–1825 325 87 501 913 35.6 9.5 54.9 1826–1850 170 115 496 781 21.8 14.7 63.5 1851–1868 138 112 354 604 22.9 18.5 58.6 Total 942 339 1891 3172 29.7 10.7 59.6 (2) Female 1773–1800 405 69 427 901 45.0 7.7 47.4 1801–1825 382 86 577 1045 36.6 8.2 55.2 1826–1850 377 125 445 947 39.8 13.2 47.0 1851–1868 151 228 203 582 26.0 39.2 34.9 Total 1315 508 1652 3475 37.8 14.6 47.5

In order to ensure accurate figures on dekasegi destinations, in Table 13.4 we measured their compositions in man-years. We divided the years between 1773 and 1868 into four periods and dekasegi destinations into three classifications:10 cities, towns, and rural areas. The total amount of dekasegi reached 6,647 man- years, with an annual average of 68. However, we can see that dekasegi destinations changed a good deal over time. Fewer men and women went to rural areas on dekasegi, as increasingly people began to go to cities and towns. It is significant that the women going on dekasegi to towns had the highest rate toward the end of the period. It is quite clear from their destinations that they were employed in the area of the developing textile industry which spread over 5 to 7 mile area to the east of this village.11 We showed the amount of dekasegi for each year in Fig. 13.3. If we compare men with women, we find the swing for women is much larger than for men. For

227

12:27:16:11:09 Page 227 Page 228

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

Figure 13.3a Dekasegi by Sex.

Figure 13.3b Dekasegi by Destination.

women, the ratio of the peak exodus and minimum exodus was 58:19, i.e., nearly 3:1. This ratio was 47:18 in the case of men. There were two clear peaks of dekasegi for women, in 1781–1795 and 1815–1826, while for men there was no clear peak, only a leveling oﬀ of dekasegi. It is clear that the second peak for women was due to the increase of dekasegi to cities and towns. This change cannot be explained by population changes within the village. The total population, as seen in Fig. 13.1, was low and showed a stagnant trend in the middle of the period. If we look at the sex ratio and age composition, we can see no conspicuous change, and after 1840 when dekasegi was at its minimum, we see an increase in the population aged 16–50. Given this, the swing in dekasegi for women can be

228

12:27:16:11:09 Page 228 Page 229

LABOR MIGRATION IN A PRE-INDUSTRIAL SOCIETY

thought to have resulted from conditions of employers or the difference in income obtainable in this village and the dekasegi destinations. The trend for dekasegi to rural areas was declining, but if we combine dekasegi to this village with dekasegi from it, we see labor migration among villages as characteristic of this period. On the whole, the number of people coming on dekasegi to this village is far smaller than that of people going out from this village, with a ratio in terms of man-year of 447:2030. If we examine migration in terms of direction, we observe the interesting characteristics seen in Fig. 13.3. As previously mentioned, a larger number of women than men went out on dekasegi. Men mostly went south for dekasegi and next in terms of frequency to the north, east, and west, while dekasegi to this village was from north, east, south, and west. For women, dekasegi from the village was to the south, east, north, and west, in that order, while dekasegi to this village from the north, west, south, and east. Women most frequently migrated in from north but migrated out for the south and east. Thus there seems to have been a directional trend for labor migration involving this village. To the southeast of this village lay the city of Nagoya, and in the area between Nagoya and the village the textile industry was well developed. Characteristic of labor migration in a pre-modern society, labor migration here was not to a distant place in a single journey, but occurred gradually through a series of short moves, first to places nearby. Here it is clear that 80% of the migration among villages took place within a 2.5 mile radius and flowed in a definite direction. The statistics on the dekasegi destination by city calculated in man-years for each periods is shown in Table 13.5. Except for the very few who went to Edo, people mostly went to Nagoya, 15 miles to the southeast of Nishijo¯, to O¯¯ saka, 75 miles to the southwest, or to the area between these two major cities. Before 1800, the largest number of both sexes went to Kyo¯to for dekasegi. In the following quarter of a century, men continued to go mainly to Kyo¯to, but women went mostly to Nagoya. After 1826, men mostly went to Nagoya, too. In addition to these three cities, other important dekasegi destinations were O¯¯ gaki (both men and women), Sakai (men), and Tsu (women). In terms of dekasegi to towns, the destinations are listed in descending order of importance: for men, Yokkaichi and Hamada in the Province of Ise (100 man- years); Kasamatsu (81), Takehana (65), and Imao (39) all in the Province of Mino; and for women, Takehana (173); Noma in the Province of Owari (108); Kurigasa in the Province of Mino (43); Hamada (42); Hagiwara in the Province of Owari (37); and Kasamatsu (31). These were small cities or towns located on the No¯bi Plain and the surrounding area and which formed the administrative, transporting, and industrial center of the locality.

Duration of Dekasegi There are only a few persons for whom the length of time they stayed on dekasegi can be accurately measured. In the 1773–1825 cohort group 43 men and 92 women ended their dekasegi for reasons other than death. The average duration of dekasegi for these men and women analyzed by classes and destinations is as follows: for men, 13.3 years for tenants and 8.6 years for non-tenants; 8.7 years for those who went to cities and towns and 12.3 years for rural areas only; and for

229

12:27:16:11:09 Page 229 Page 230

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

Table 13.5 The Destination of Dekasegi to Urban Areas (man-years)

1773–1800 1801–1825 1826–1850 1851–1868 Total percent

(1) Male Edo 45 19 64 3.4 Nagoya 134 75 199 182 590 31.2 Kuwana 18 18 1.0 Tsu 9 17 2 28 1.5 O¯¯ gaki 13 33 48 19 113 6.0 Hikone 10 17 27 1.4 Kyo¯to 340 230 34 27 631 33.4 O¯¯ saka 8 87 148 71 314 16.6 Sakai 38 50 18 106 5.6

Total 540 501 496 354 1891 100.1

(2) Female Nagoya 117 229 290 91 727 44.2 Kuwana 12 12 0.7 Tsu 16 39 5 60 3.6 O¯¯ gaki 22 95 17 43 177 10.8 Hikone 9 1 1 11 0.7 Kyo¯to 223 170 44 18 455 27.7 O¯¯ saka 27 38 87 51 203 12.3

Total 426 572 444 203 1645 100.0

women, 14.0 years for tenants and 15.2 years for non-tenants; 13.7 years for those who went to cities and towns, and 12.5 years for rural areas only. As far as we can tell from these figures, the pattern of distribution for men and women is in sharp contrast. For people who went on dekasegi to cities and towns, the records contain only the names of the places and it is not possible to estimate how long these people continued to work under the same employers. But for people who went to rural areas, their employers’ names are recorded and so we can measure the length of dekasegi contracts. According to this, we know that from the standpoint of distribution the largest number of people had one-year contracts. In a sample of 261 men, 115, or 45%, had one-year contracts, while 70 women from a sample of 219, or 32%, had the same. The longer the contract, the smaller is the percentage of the people who held them. Ninety percent of the men and 68% of the women had contracts of less than five years in the length. The total average employment period was 2.9 years for men and 5.2 years for women. The reason why women served longer than men is that many of them served as domestic servants. Also, we find that the later the time during the period studied, the longer was the contract. It seems that employment for a year or two as a farmhand decreased, while the proportion of domestic service increased.

230

12:27:16:11:09 Page 230 Page 231

LABOR MIGRATION IN A PRE-INDUSTRIAL SOCIETY

Termination of Dekasegi Table 13.6 contains the statistics on various reasons why dekasegi came to an end. “Continued” indicates that in the final year of the documents the dekasegi was still continuing. With the exception of these people, among 329 men and women, 126 (38%) died on dekasegi, 113 left dekasegi for marriage and other reasons but did not return to their village, and only 87, or 27%, ever returned home. Though their employment is called “dekasegi,” the percentage who returned home is very low, especially for women. However, moves to other villages for marriage can be balanced by moves to Nishijo from other villages. Specifically, 34% of the men returned home while only 15% of the women did. For men, the dekasegi rate was 48%, and 32% of those who survived to eleven years of age eventually went away to another area as part of that labor force. Twenty-five percent of the women left the village. The average age for the termination of dekasegi was 30.3 years for men and 27.8 years for women.

THE RELATIONSHIP OF MIGRATION DATA TO OTHER DEMOGRAPHIC INDICES In the indices relating to dekasegi, differences were seen among the villagers’ classes. In outline, the dekasegi rates were higher and the starting age for dekasegi were younger for the lower classes. Consequently, we see an increased risk of death in the cities, delay in the age at marriage, and differences in the reproduction rate among different classes, all caused by dekasegi. However, more research is required to confirm these observations. First, it is necessary to determine if there are differences in death-rates by age groups due to dekasegi. Figure 13.4 shows the differences in the death-rate in several age groups by those who had been on dekasegi in cities and those who had not. We must take into consideration that those who had no dekasegi experience in cities included men and women who were too weak or physically handicapped to go on dekasegi, but at least as far as men are concerned, those in almost all age groups who experienced dekasegi in cities had a higher death-rate. For women, there is evidence of a reverse relationship. As seen in the following, women in the village got married at an earlier age and a large number of them died from pregnancy and childbirth.

Table 13.6 Reasons for Terminating Dekasegi (Persons)

Reason Male Female

Continued 34 31 Death 62 64 Returned home 48 18 Marriage 5 103 Adoption 8 Miscellaneous 16 2 Unknown 3 Total 176 218

231

12:27:16:11:09 Page 231 Page 232

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

Figure 13.4 Age Speciﬁc Mortality by Dekasegi Experience.

Table 13.7 Age Speciﬁc Fertility

Age group Number of births Fertility

16–20 56 0.257 21–25 201 0.319 26–30 225 0.285 31–35 203 0.252 36–40 147 0.201 41–45 65 0.097 46–50 15 0.026

Next let us look at the effect on age at marriage. In the cohort group born between 1773 and 1825, the average age at marriage for men who had been on dekasegi was 29.2, while for those who had no dekasegi experience, the average was 28.2. The average ages at marriage for women were 26.3 and 20.7 respectively. For both men and women, those who had dekasegi experience had higher ages at marriage. For women especially, the difference can be said to be fairly large. Age specific fertility (Table 13.7) obtained through family reconstitution gives a decrease of two in the number of children born due to this delay in age at marriage. The effect of dekasegi on the demographic indices is thus seen in the death rate for men and in the decrease of births for women. Dekasegi can be said to have had a negative effect on Nishijo¯’s population. But the effect of dekasegi on the demographic indices differed according to class. We can see this most clearly with regard to fertility. The number of children in completed families analyzed by the age at marriage of women is shown in Table 13.8. These figures show no great differences by class. But if all women who had the experience of dekasegi married at age 26.3 while those with no experience

232

12:27:16:11:09 Page 232 Page 233

LABOR MIGRATION IN A PRE-INDUSTRIAL SOCIETY

Table 13.8 Average Number of Births in Completed Families by Age at Marriage*

Age at marriage Number of births

16 7.19 17 6.93 18 6.67 19 6.41 20 6.16 21 5.90 22 5.58 23 5.26 24 4.94 25 4.62 26 4.31 27 4.02 28 3.74 29 3.45 30 3.17

* calculated from Table VII.

of dekasegi married at age 20.7, as was shown in the previous section, and if the differencial dekasegi rates, presented in Table 13.1, apply to them, we would see differential reproduction rates by class. If we calculate the number of births in completed families12 the results would be as follows: the average number of births would be 5.4 among land owners, 4.9 among owner cultivators, 4.8 among owner tenant-cultivators, and 4.7 among tenants. These are all figures for completed families, but actually married life must have ended earlier in many cases. Furthermore, we have found that a gross reproduction rate of 2.2 was the minimum level for maintaining population in a Tokugawa rural society.13 The reproduction rate in Nishijo¯, with the exception of the landowners, was, therefore, very close to this line in completed families. The rate must have been below this line for the tenants who had been on dekasegi. If so, the tenants would have found it impossible to maintain their population and the number of tenant households would have decreased without interclass moves. In fact, there were many tenant households which became extinct due to the lack of heirs or which had to adopt sons from others. We obtain the facts seen in Table 13.9 when we classify changes of household head. During the period studied 66 households became extinct, of which 60 were in the tenant class. This represents about 32% of all changes of household heads in the tenant class, and is very high compared with 6% in the owner classes. In a balance sheet of the numbers between newly established branch households and extinct households, the landowners and the owner cultivators would receive a plus account, while the tenants would receive a minus account. How- ever, the proportion of each class in this village changed scarcely at all during the period studied, thus providing evidence that an interclass movement of households from upper to lower took place. When a branch household was established, it was usually of a lower class than the main household. It is a phenomenon

233

12:27:16:11:09 Page 233 Page 234

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

Table 13.9 The Presence of Heirs by Class

Class Inherited No Heirs

Landowner 34 0 Owner cultivator & owner-tenant 61 6 cultivator Tenant 183 60 Total 278 66

worthy of close attention that moves among classes had a downward trend in this labor supplying village.

CONCLUSION The population of Japan during the period 1721–1846, generally speaking showed decreases in the northeastern area, stayed level in the central area, and increased in the southwestern area.14 The decrease in the northern area is thought to be attributable primarily to deteriorating natural conditions (climate growing colder). In the eastern and central part of Japan where the economic development was greatest and the proportion of urban population was highest, nobody has ever spoken of any reason for stagnation of the local populations. What has been observed in this article is that if economic development raises the proportion of the urban population, much migration of labor to the city from rural areas will take place. It is thought that the higher death rate in cities and the ﬂourishing dekasegi resulted in lowering the population in the labor supplying areas. Accord- ing to the thesis which Dr. E. A. Wrigley advanced, this is the “negative feedback” which is observed between population and economic development in pre- industrial societies.15

NOTES

* The author is extremely indebted to Dr. Susan B. Hanley for revising his English manuscript. 1. These may be literally translated “faith investigation registers.” They are introduced in great detail in the following: Robert J. S, “Small families, small households and residential instability: town and city in ‘pre-modern’ Japan,” in Household and Family in Past Time, ed. by Peter L, Cambridge, 1972. pp. 431–436. 2. In order to apply family reconstitution or a tracing study in the same manner as attempted in this article, it is necessary to use the document over a period of at least one hundred years without any great break. 3. These documents are located in St. Paul’s University Library, Tokyo. The author is very thankful to Professor Hideo H for having kindly given him the opportunity of using them. 4. Special attention must be given to the following points regarding the births and deaths obtained from the sho¯mon-aratame-cho¯: these records were compiled every year, the people living at the time being the subject of the investigation. In consequence, those who died before new documents were compiled are not included in the data. Therefore, the birth rate and the death rate used here are somewhat lower compared with the reality. It would be necessary to multiply them by 1.25. 5. M Kazuo, Edo no shakai ko¯zo¯, Tokyo, 1969, p. 196. 6. Akira H, “Demograﬁa e economia no Japa¯o pré-industrial,” Anais de História, No. 4, 1972, Assis (Brasil).

234

12:27:16:11:09 Page 234 Page 235

LABOR MIGRATION IN A PRE-INDUSTRIAL SOCIETY

7. For observations concerning Suwa County of Shinano Province, see Chap. 2 of H, Kinsei no¯son no rekishi jinko¯gakuteki kenkyu¯ , Tokyo, 1973. 8. Due to the Japanese traditional method of counting ages, a person becomes one year at his birth. When the new year arrives the child will have another year added, irrespective of the age of the child counted by the Western way. In this article age counting follows the traditional Japanese method. 9. Households of the landowner class comprised 5–15% of the total number of households, the owner cultivator 5–20%, the owner tenant-cultivator 0–20%, and the tenants 60–70%. The fluctuations were due to institutional changes in the evaluation in 1810. 10. This classification is for the sake of convenience and is only nominal. Cities here have populations above 10,000, while towns have the suffix (-machi and -shuku) and a population below 10,000. 11. Takehana of the Province of Mino, and Ichinomiya and Okoshi of the Province of Owari are its centers. H Hideo, Kinsei no¯son ko¯gyo¯ no kiso katei, Tokyo, 1960. 12. The families where wives were continually married to age 50 and beyond. 13. H, op. cit., p. 163. 14. Akira H, “Mouvements de longue durée et structures japonaises de la population a l’époque de Tokugawa”, Annales de démographie historique, 1971, pp. 247–263. 15. E. A. W, Population and History, London, 1969. Chaps. 3 and 4.

235

12:27:16:11:09 Page 235 Page 236

12:27:16:11:09 Page 236 Page 237

PART IV FAMILY AND HOUSEHOLD

12:27:16:11:09 Page 237 Page 238

12:27:16:11:09 Page 238 Page 239

Co-authored with Nobuko Uchida and ﬁrst published in Household and Family in Past Time (ed. Peter Laslett), pp. 473–516, Cambridge, 1972.

14 Size of Household in a Japanese County Throughout the Tokugawa Era

INTRODUCTORY e have three main purposes in this chapter. First, to examine the structure Wand size of households in pre-industrial Japan. Second, to present the results of the historical demographic research which is being carried out by the study group at Keio University on the demographic history of changes in household size during the Tokugawa era (1603–1868). Finally, to present material which can be used in international cross comparison with the results presented by Peter Laslett for England.1 Fortunately the myth that the evolution of the small family was a product of industrialisation does not exist in Japan. The most common explanation given today is that small families or households consisting of one married couple and several children became common throughout the nation between the sixteenth and eighteenth centuries, and took the place of the larger households of earlier periods, many or most of which had more than one married couple apiece. There are household registers of a much earlier period in the eighth and ninth centuries and the fragments which survive indicate a very large household size, perhaps as big as 20 to 30 persons per household on average. After a long break in the records, suitable material for comparison becomes available again in the seventeenth century. Notwithstanding considerable variation from region to region, the great majority of Japanese households consisted of from 5 to 10 members during that period. Although families of from 20 to 30 are not rare in the seventeenth-century records, the average household size was almost certainly less than 10 in the areas near cities, where market economy and transportation facilities were well-developed. A persistent and uninterrupted decline in average household size continued throughout the Tokugawa era, so that by the middle of the eighteenth century, the usual family size reached that common today, 4 or 5 members. There has been little research until recently on the decline in the size of households during the pre-industrial stage of Japanese history. The study group at Keio University is engaged in the systematic analysis of the shu¯ mon-aratame-cho¯ as one listing source, and has adopted the method known as family reconstitution to test a number of indicators useful in demographic studies.

239

12:27:16:11:09 Page 239 Page 240

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

SOURCES: THE SHU¯ MON-ARATAME-CHO¯¯ 2 The first question to be raised with regard to the shu¯ mon-aratame-cho¯ concerns their reliability. At first glance, one is taken aback by the number of inaccuracies. For example, the births which occurred after the registration had been taken in one year should appear in the following year’s register, but some children were not entered until they were three or four years old. If birth rates are calculated without regard for these omissions, they are ‘so low as to be improbable’.3 This is true of death rates also. Another important issue in determining the reliability of the shu¯ mon-aratame- cho¯ is deciding who among the registered family members was actually resident. The rule was that every resident recorded should be one who was indeed situated in a family group in the village at the time. In many cases however, it is not clear whether ‘family member’ refers to a person actually living in the family, or merely a registered member who was in fact living elsewhere. In actual practice absent members were often registered, and persons who were in fact situated in the family group were omitted from the register. With the coming of the nineteenth century and greater mobility in the population, the reliability of the registers certainly declined. In general, the number of omissions increased in cities, where the demand for workers had grown as a result of commercial and manufacturing development. On the other hand in agricultural villages, the actual number of residents was often less than the number recorded in the shu¯ mon-aratame-cho¯. There are also problems throughout the Tokugawa era arising from the unit upon which the shu¯ mon-aratame-cho¯ were based. The village was the basic unit used, but some villages were under the control of more than one daimyo¯, and there exist examples of the register for one village being divided into a number of parts, each under the jurisdiction of a different daimyo¯. In large villages shu¯ mon- aratame-cho¯ were often made up separately for each of the religious sects to which the villagers respectively belonged. Therefore it was possible for members of the same family to be listed in different registers, since there were no restrictions against marriage between different religious sects. Thus the demographic information obtained from various registers often turns out to be of little value unless all of the various parts of the register can be located. Another problem in using the shu¯ mon-aratame-cho¯ as a source for demographic data has to do with the calculation of age. It was the custom in Japan to count a person as being one year old at birth, and to add one year at the beginning of every new calendar year. Thus it is necessary to subtract approximately one year from every age mentioned in order to obtain the age according to the Western method of calculation. Also babies born in a specific year who died before registration was undertaken were not recorded. It is therefore very difficult to estimate the birth rate using shu¯ mon-aratame-cho¯. Nevertheless there can be no doubt of the demographic value of the shu¯ mon- aratame-cho¯ in spite of these defects, especially in the case of communities which have preserved successive registration recordings over long periods of time without break.

240

12:27:16:11:09 Page 240 Page 241

SIZE OF HOUSEHOLD IN A JAPANESE COUNTY

HANDLING AND ANALYSIS OF DATA In total there are 800 volumes of the shu¯ mon-aratame-cho¯ available for 120 communities in Suwa County. The 38 communities where data were most ample were selected and their records analysed. Data items for each household were copied on to a basic data card, recording age and position within the family for each household member. The data for one particular village – the village of Yokouchi – where documents survive for no less than 140 years between 1671 and 1871, had already been worked on in an earlier study,4 and we drew upon those workings as necessary. The map in Fig. 14.1 shows the position of the county, the lake and the village. There are three distinguishable classes of settlement among the 38 communities chosen. First come the machi or shuku, which may be translated as ‘towns’, since they displayed some of the urban characteristics typical of the Tokugawa era. Secondly there were the villages which had existed before the coming of the Tokugawa regime, and finally the shinden, or newly settled villages, which were founded under the Tokugawa themselves. It will be noted that there are only two towns (machi) included in this study, namely Tomono-machi and Kanazawa. Furthermore neither of these two was a castle town, but rather a post town (shuku) along a highway, or perhaps a local centre. Since these two towns mostly shared the characteristics of agricultural villages, we have not set them apart in this study. We have divided the 38 communities analysed into four geographical districts which are also to be seen in Fig. 14.1. These are: (1) District W, west of Lake Suwa, and (2) District E on the east bank of the lake. Both the W and E Districts are located on the flat land area near Lake Suwa and developed early. Castle towns were located there, highways were built and a market economy developed to a high degree. It should be particularly borne in mind that the silk industry was introduced into the W District after 1830. (3) In general, District C consists of villages located in the valleys of a mountainous area and a road going to Edo by way of Ko¯fu runs through this district. (4) The fourth district, designated as District Y, lies on the west side of Mt Yatsugatake and is a gently sloping area. This district was developed and settled in the seventeenth century and was well advanced after that time. Shinden, that is ‘newly settled villages’, were numerous in this region, and the eight villages which we have analysed were all founded after 1600. Fig. 14.2 traces the course of indices of population for each district in the County. These were obtained in previous studies of this region.5 There are area differences in the indices, but in general the population grew during the problematic early period, was stagnant during the middle period, and then tended to grow again in the last part of the Tokugawa era. Unfortunately the shu¯ mon-aratame-cho¯ of these 38 communities have not in every case been kept in a manner which is most convenient to our analysis. In order to average these out and still take advantage of the largest possible number of cases, the following procedure was adopted. First the 200-year period from 1671 to 1871 has been divided into twenty sub-periods of ten years each, omitting 1871. Then from among the body of data for each community, figures have been

241

12:27:16:11:09 Page 241 Page 242

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

Figure 14.1 Map of Tokugawa Japan showing Suwa County and district divisions

chosen from as near to the midpoint of each ten-year sub-period as possible for the purpose of analysis. For example in the case of a community where ﬁgures for the sub-period 1671 to 1680 are available for 1673, 1679 and 1680, we have chosen those for 1673, i.e. the year closest to the middle of the sub-period. In this way we have avoided the error of choosing data in such a manner that observations would be concentrated around a particular year. The list of communities whose ﬁgures have been used is given in Table 14.1. This table also presents the

242

12:27:16:11:09 Page 242 Page 243

SIZE OF HOUSEHOLD IN A JAPANESE COUNTY

Figure 14.2 Population trends, by district (1671–80 = 100)

Table 14.1 38 Communities in Suwa County, Shinano Province, Japan, 1671–1871 in date order, giving population, households and mean size of household

No. Community Date Population Households MHS

W-11 Ayuzawa 1673 84 12 7.00 1684 83 12 6.92 1699 103 13 7.92 1703 107 12 8.92 1717 94 14 6.71 1724 113 16 7.06 1736 105 18 5.83 1741 100 19 5.26 1757 115 17 5.53 1766 134 28 4.79 1777 134 25 5.36 1788 132 28 4.71 1795 132 27 4.89 1821 134 29 4.62 W-12 Misawa 1688 246 19 12.95 1699 293 22 13.32 1710 239 35 6.83 1717 280 35 8.00 1724 298 35 8.51 1739 295 54 5.46 1757 290 55 5.27 1766 280 56 5.00 (Continued Overleaf) 243

12:27:16:11:09 Page 243 Page 244

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

Table 14.1 Continued

No. Community Date Population Households MHS

1788 243 50 4.86 1796 239 55 4.35 1810 260 56 4.64 1840 290 64 4.53 1860 351 86 4.08 W-13 Hashihara 1684 143 18 7.94 1703 149 17 8.77 1730 150 26 5.77 1736 159 24 6.63 1741 150 23 6.52 1766 170 33 5.15 1777 163 35 4.66 1788 172 38 4.53 1796 162 37 4.38 1819 165 33 5.00 1866 201 51 3.94 W-14 Koguchi 1678 194 31 6.26 1690 207 35 5.91 1701 243 39 6.23 1750 255 53 4.81 1767 245 49 5.00 1782 209 51 4.10 1796 212 53 4.00 1807 220 56 3.93 1823 221 57 3.88 1840 199 59 3.37 1866 268 62 4.32 W-15 Imai 1757 442 80 5.53 1766 451 79 5.71 1799 413 108 3.82 W-16 Takei 1693 95 22 4.32 1717 134 30 4.47 1750 114 32 3.56 1756 103 31 3.32 1767 104 33 3.15 1795 79 20 3.95 1822 69 18 3.83 1856 99 25 3.96 1862 100 23 4.34 W-10 Tomo-no-Machi 1794 380 94 4.04 1814 418 102 4.10 1830 481 116 4.26 W-21 Hagikura-Shinden 1714 42 10 4.20 1729 58 10 5.80 1747 69 13 5.31 1765 91 12 7.58 1828 168 34 4.94 1844 134 30 4.47 1855 165 28 5.89 244

12:27:16:11:09 Page 244 Page 245

SIZE OF HOUSEHOLD IN A JAPANESE COUNTY

E-31 Kosaka 1689 242 37 6.54 1703 292 42 6.95 1719 364 62 5.87 1728 382 80 4.78 1766 395 96 4.11 1787 406 97 4.19 1799 395 95 4.16 1839 481 114 4.22 1866 617 137 4.50 E-32 Aruga 1684 801 124 6.46 1700 848 143 5.93 1716 825 161 5.12 1724 829 180 4.61 1742 760 184 4.13 1751 743 187 3.97 1788 591 146 4.05 1795 627 154 4.07 1849 697 161 4.33 E-41 Ueno-Shinden 1684 26 5 5.20 1700 51 7 7.29 1716 71 13 5.46 1724 59 12 4.92 1742 69 15 4.60 1751 75 16 4.69 1788 109 22 4.95 1795 119 22 5.41 1849 140 26 5.38 E-42 Ushiroyama-Shinden 1687 53 9 5.89 1695 74 10 7.40 1703 86 12 7.17 1716 100 13 7.69 1726 97 13 7.46 1749 135 23 5.87 1755 131 23 5.70 1766 136 28 4.86 1782 148 25 5.92 1799 128 25 5.12 1811 113 29 3.90 1858 128 30 4.27 E-33 Okuma 1687 353 65 5.43 1695 429 69 6.22 1707 425 85 5.00 1719 430 98 4.39 1725 399 95 4.20 1739 406 100 4.06 1750 459 109 4.12 1759 405 111 3.65 1766 415 114 3.64 1783 347 92 3.77 1796 290 90 3.22 1809 305 89 3.43 1854 308 81 3.80 (Continued Overleaf)

245

12:27:16:11:09 Page 245 Page 246

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

Table 14.1 Continued

No. Community Date Population Households MHS

E-34 Shimokaneko 1673 438 63 6.95 1688 409 64 6.39 1718 449 57 7.88 1760 389 82 4.74 1771 371 81 4.58 1782 337 72 4.68 1796 337 74 4.55 1811 387 82 4.72 1854 364 90 4.04 1866 359 89 4.03 E-35 Fukushima 1689 251 53 4.74 1696 254 53 4.79 1705 230 45 5.11 1717 182 38 4.79 1757 248 57 4.35 1767 271 65 4.17 1775 274 70 3.91 1782 291 74 3.93 1796 306 74 4.14 1821 337 75 4.49 1855 359 79 4.54 1866 347 83 4.18 E-36 Ijima 1672 256 44 5.82 1684 285 43 6.63 1700 315 50 6.30 1705 322 48 6.71 1716 272 53 5.13 1725 307 61 5.03 1750 294 62 4.74 1754 280 65 4.31 1767 292 72 4.06 1774 279 73 3.82 1788 280 64 4.38 1795 279 62 4.50 1802 303 70 4.33 1852 294 67 4.39 1861 296 68 4.35 E-37 Arai 1673 197 24 8.21 1686 198 22 9.00 1695 185 25 7.40 1705 173 30 5.77 1716 176 29 6.07 1726 169 38 4.45 1745 163 36 4.53 1753 173 41 4.22 1766 177 40 4.43 1776 182 43 4.23 1788 169 44 3.84

246

12:27:16:11:09 Page 246 Page 247

SIZE OF HOUSEHOLD IN A JAPANESE COUNTY

1800 185 46 4.02 1815 214 50 4.28 1821 221 55 4.02 1839 233 60 3.88 1858 245 54 4.54 1869 290 60 4.83 E-38 Uehara 1680 673 104 6.47 1684 707 99 7.14 1694 814 107 7.61 1707 770 104 7.40 1711 768 112 6.86 1724 738 123 6.00 1738 700 124 5.65 1742 681 123 5.54 1754 615 118 5.21 1773 579 118 4.91 1795 538 127 4.24 1811 569 135 4.21 1838 515 141 3.65 1851 544 147 3.70 1865 559 145 3.86 E-39 Yokouchi 1675 201 29 6.93 1685 224 30 7.47 1695 263 41 6.41 1705 283 44 6.43 1715 320 50 6.40 1725 356 50 7.12 1734 388 72 5.39 1745 430 77 5.58 1755 462 77 6.00 1765 483 89 5.43 1775 511 102 5.01 1786 470 99 4.75 1796 479 98 4.89 1805 462 100 4.62 1815 500 96 5.21 1825 498 97 5.13 1836 477 104 4.59 1846 471 107 4.40 1855 443 114 3.89 1866 496 112 4.43 E-51 Chino 1674 325 51 6.37 1681 372 54 6.89 1706 592 85 6.96 1724 581 102 5.70 1769 571 127 4.50 1783 544 124 4.39 1791 480 117 4.10 1814 490 122 4.02 1841 537 131 4.10 1852 583 136 4.29 1864 613 146 4.20 (Continued Overleaf)

247

12:27:16:11:09 Page 247 Page 248

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

Table 14.1 Continued

No. Community Date Population Households MHS

E-52 Nakagawara 1676 77 12 6.42 1682 72 10 7.20 1694 84 11 7.64 1720 136 22 6.18 1754 153 33 4.64 1766 174 38 4.58 1776 163 40 4.08 1783 152 39 3.90 1795 145 38 3.82 1805 185 42 4.40 E-43 Sakamuro-Shinden 1674 68 10 6.80 1681 60 12 5.00 1706 54 8 6.75 1724 85 10 8.50 1769 133 26 5.12 1783 157 28 5.61 1791 152 33 4.61 1814 165 33 5.00 1841 159 45 3.53 1852 177 43 4.12 1864 194 45 4.31 C-61 Kifune-Shinden 1725 148 21 7.05 1747 134 34 3.94 1756 153 37 4.14 1775 166 35 4.74 1796 179 38 4.71 1815 200 40 5.00 C-70 Kanazawa-Machi 1725 637 99 6.43 1747 588 106 5.54 1756 610 124 4.92 1775 630 132 4.77 1796 582 124 4.69 1815 597 133 4.49 C-71 Sezawa 1689 302 27 11.19 1697 347 29 11.97 1707 337 29 11.62 1724 405 37 10.95 1744 367 37 9.92 1834 309 82 3.77 1858 294 78 3.77 1864 301 78 3.86 C-72 Tsukue 1684 148 18 8.22 1699 155 17 9.12 1729 255 27 9.44 1736 233 29 8.03 1788 249 30 8.30 1802 254 30 8.47 1857 280 64 4.38

248

12:27:16:11:09 Page 248 Page 249

SIZE OF HOUSEHOLD IN A JAPANESE COUNTY

C-73 Shimotsutagi 1678 134 20 6.70 1684 126 21 6.00 1724 160 32 5.00 1756 153 36 4.25 1765 143 39 3.67 1788 110 31 3.55 1792 107 34 3.15 1809 122 31 3.94 1837 150 34 4.41 1864 164 38 4.32 C-74 Tabata 1688 92 12 7.67 1724 126 20 6.30 1746 130 26 5.00 1756 138 26 5.31 1768 139 29 4.79 1775 139 29 4.79 1788 133 30 4.43 1795 150 30 5.00 1839 123 33 3.73 1855 142 37 3.84 1864 160 40 4.00 C-75 Kuzukubo 1687 205 23 8.91 1766 263 33 7.97 1796 300 42 7.14 1843 335 81 4.14 1857 362 85 4.26 1864 419 96 4.36 C-76 Okkoto 1747 769 103 7.47 1768 885 125 7.08 1812 960 217 4.42 1823 947 218 4.34 1833 985 218 4.52 1843 958 218 4.39 1864 1,019 234 4.35 C-62 Sezawa-Shinden 1671 209 31 6.74 1703 400 29 13.79 1719 474 49 9.67 1750 500 74 6.76 1815 413 91 4.54 Y-81 Aragami-Shinden 1717 195 21 9.29 1725 212 24 8.83 1748 181 30 6.03 1755 207 30 6.90 1766 202 34 5.94 1773 214 40 5.35 1788 227 44 5.16 1799 245 53 4.62 1846 212 58 3.65 Y-82 Kamisugasawa-Shinden 1673 45 8 5.63 1688 54 10 5.40 1694 60 10 6.00 (Continued Overleaf)

249

12:27:16:11:09 Page 249 Page 250

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

Table 14.1 Continued

No. Community Date Population Households MHS

1724 101 11 9.18 1737 101 11 9.18 1747 115 12 9.58 1754 121 13 9.31 1766 127 16 7.94 1795 116 26 4.46 1816 125 27 4.63 Y-83 Kikuzawa-Shinden 1675 88 15 5.87 1685 114 16 7.13 1699 135 15 9.00 1855 306 69 4.43 Y-84 Anayama-Shinden 1688 197 22 8.95 1706 272 22 12.36 1718 310 23 13.48 1724 326 33 9.88 1747 339 37 9.16 1755 370 38 9.74 1766 412 46 8.96 1782 434 51 8.51 1795 476 53 8.98 1807 477 59 8.08 1815 465 77 6.04 1848 458 95 4.82 1855 501 98 5.11 1865 514 104 4.94 Y-85 Yamada-Shinden 1686 218 21 10.38 1695 258 23 11.22 1703 279 29 9.62 1712 277 45 6.16 1724 280 39 7.18 1789 337 76 4.43 1800 360 79 4.56 1810 345 82 4.21 1834 327 86 3.80 1867 283 73 3.88 Y-86 Tsukinoki-Shinden 1688 184 21 8.76 1695 224 22 10.18 1707 255 22 11.59 1716 295 24 12.29 1725 322 25 12.88 1735 383 36 10.64 1776 540 81 6.67 1789 526 97 5.42 1795 517 97 5.33 1803 469 96 4.89 1838 573 134 4.28 1857 637 157 4.06 1870 704 156 4.51

250

12:27:16:11:09 Page 250 Page 251

SIZE OF HOUSEHOLD IN A JAPANESE COUNTY

Y-87 Nakamichi-Shinden 1680 117 18 6.50 1688 117 14 8.35 1695 156 15 10.40 1706 190 17 11.18 1718 229 17 13.47 1724 244 18 13.56 1755 333 23 14.48 1768 375 29 12.93 1789 430 43 10.00 1795 438 41 10.68 1815 496 112 4.43 1834 523 119 4.39

population, number of households and mean size of household (hereafter referred to as MHS) for each community. Thus for the 38 communities we have a total of 382 cases available for analysis. In the analysis itself, we have occasionally divided the whole period of 200 years into ﬁve other sub-periods, as follows:

Period Years

1 1671–1700 2 1701–1750 3 1751–1800 4 1801–1850 5 1851–1870

As Laslett has stated, the words ‘family’ or ‘household’ are difficult to define. It is clearly inappropriate to look for evidence to satisfy our preconceptions, especially in historical documents. In this study the following rule is observed. A unit which is treated as a group within the context of the shu¯ mon-aratame-cho¯ is regarded as a household for the purpose of analysis. There is some difference of opinion about the nature of the unit used in the religious registrations. There are undoubtedly a few cases of households which appear unlikely to have existed in fact, for example, a household consisting of one child only two years of age. Moreover it must always be borne in mind that some of the household units may not have been economically or socially independent, and that this may not be apparent from the data appearing in the document.6 Nevertheless in this chapter we have decided to confine our attention to those points which are apparent from the data in their original form, ignoring the vagueness and shortcomings mentioned above. The reason why the word ‘household’ has been used in this chapter rather than ‘family’ is simply because ‘household’ appears to be the more general term.

251

12:27:16:11:09 Page 251 Page 252

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

DISTRIBUTION OF THE MEAN SIZE OF HOUSEHOLD In general, family size in Suwa County from 1671 to 1870 showed a tendency to decline. The figures for totals in the final section of Table 14.2 show that during the sub-period 1671 to 1700 MHS over all areas was 7.04; during the next three fifty-year periods the ratio declined as follows, 6.34, 4.90, 4.42. By 1851–1870 the average size was 4.25, only 60% of the size at the beginning. This decrease was by no means uniform. If we examine changes which went on in districts during the rough time divisions of Table 14.2, it is clear that there were differences in the rate of decline. The trend in MHS for each region is shown in Fig. 14.3 in ten-year sub-periods, and here the differences are even more apparent. In the plain area near Lake Suwa, decline in household size is already well advanced by the end of the seventeenth century, and by the end of the eighteenth century the ratio falls as low as 4.5 to 4.0. Over the next 100 years it appears that household size was almost completely stable. In Districts C and Y, on the other hand, reduction in household size occurs more slowly, especially in District Y where stabilisation in the 4.5 to 4.0 region does not take place until 1830. Furthermore MHS was higher at the beginning of the eighteenth century as compared to other regions, the highest figures of all being 12.7 for District C and 11.2 for District Y. When undertaking MHS analysis under these conditions, it is impossible to proceed directly to the use of correlation coefficients as Laslett has done, using the mean of the MHS over all regions as the dependent variable. This is because the MHS is not here homogeneous. In order to examine the widely differing causes and conditions for the composition of MHS we have been obliged first to examine its distribution and then to use analysis by correlation coefficient. The variables selected were those believed to have a strong relationship with MHS.

Figure 14.3 Trends in MHS, by district

252

12:27:16:11:09 Page 252 Page 253

SIZE OF HOUSEHOLD IN A JAPANESE COUNTY

When using the figures of Table 14.2, it must not be forgotten that sample sizes differ from district to district and period to period. Nevertheless this table makes it quite evident that variation in household size grew decidedly less everywhere as the years went by. Variance and standard deviation of household size, the usual statistical measures of dispersion, can be seen to be higher at the beginning than at the end, and the range between maximum and minimum size goes down as well. All this holds good both for districts and for individual communities as well as for the whole collection. If the values of the various indicators are drawn graphically, the tendency towards convergence in the variance of household size which accompanied the general decline in size itself is seen to be very marked indeed. But space is not available to display the full effect and all that we can do is to lay strong emphasis

Table 14.2 Preliminary ﬁgures and ratios in each district, by period

Periods 1671– 1701– 1751– 1801– 1851– Totals 1700 1750 1800 1850 1870

(1) W District Communities 9 20 24 12 6 71 Population 1,448 3,164 5,095 2,759 1,184 13,650 Households 184 515 1,094 654 275 2,722 MHS (overall) 7.87 6.14 4.66 4.22 4.31 5.01 MHS (mean of means) 8.06 6.28 4.79 4.29 4.42 5.51 MHS (median) 7.00 6.03 4.75 4.31 4.20 6.26 Minimum size of households 4.32 3.56 3.15 3.37 3.94 3.15 Maximum size of households 13.32 8.92 7.58 5.00 5.89 13.32 Variance 8.43 1.98 0.94 0.22 0.46 3.54 Standard deviation 2.90 1.41 0.97 0.47 0.68 1.88 (2) E District Communities 32 44 55 23 19 173 Population 9,605 15,807 17,542 8,459 7,216 58,629 Households 1,480 2,901 4,013 1,964 1,726 12,084 MHS (overall) 6.49 5.45 4.37 4.31 4.18 4.85 MHS (mean of means) 6.59 5.76 4.47 4.32 4.22 5.14 MHS (median) 6.53 5.61 4.35 4.28 4.27 4.74 Minimum size of households 4.74 4.06 3.22 3.43 3.70 3.22 Maximum size of households 9.00 8.50 6.00 5.38 4.83 9.00 Variance 0.89 1.29 0.34 0.25 0.08 1.45 Standard deviation 0.95 1.13 0.59 0.50 0.28 1.21 (3) C District Communities 9 16 19 13 9 66 Population 1,718 5,663 5,229 6,353 3,141 22,104 Households 198 752 1,004 1,426 750 4,130 MHS (overall) 8.68 7.53 5.21 4.46 4.19 5.35 MHS (mean of means) 8.50 7.93 5.13 4.63 4.13 6.03 MHS (median) 8.22 7.21 4.77 4.53 4.26 4.86 Minimum size of households 6.00 3.94 3.15 3.73 3.77 3.15 Maximum size of households 11.97 13.79 8.30 8.47 4.38 13.79 Variance 3.68 7.03 1.99 1.34 0.06 5.87 Standard deviation 1.92 2.65 1.41 1.16 0.24 2.42 (Continued Overleaf) 253

12:27:16:11:09 Page 253 Page 254

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

Table 14.2 Continued

Periods 1671– 1701– 1751– 1801– 1851– Totals 1700 1750 1800 1850 1870

(4) Y District Communities 14 20 21 11 6 72 Population 1,967 4,926 7,007 4,470 2,945 21,915 Households 230 496 1,007 945 657 3,338 MHS (overall) 8.55 9.93 6.94 4.73 4.48 6.39 MHS (mean of means) 8.13 10.31 7.64 4.84 4.49 7.78 MHS (median) 8.56 9.75 6.90 4.43 4.47 7.56 Minimum size of households 5.40 6.24 4.43 3.66 3.88 3.66 Maximum size of households 11.22 13.56 14.48 8.08 5.11 14.48 Variance 3.78 4.81 7.91 1.42 0.19 8.64 Standard deviation 1.94 2.19 2.81 1.19 0.44 2.96 All Districts Communities 64 100 110 59 40 382 Population 14,738 29,560 34,873 22,041 14,486 115,698 Households 2,092 4,664 7,121 4,989 3,408 22,274 MHS (overall) 7.04 6.34 4.90 4.42 4.25 5.19 MHS (mean of means) 7.40 7.12 5.20 4.48 4.27 5.86 MHS (median) 8.22 8.64 5.35 4.62 4.41 5.00 Minimum size of households 4.32 3.56 3.15 3.37 3.70 3.15 Maximum size of households 13.32 13.79 14.48 8.47 5.89 14.48 Variance 3.65 6.15 3.39 0.75 0.17 4.92 Standard deviation 1.91 2.48 1.84 0.87 0.41 2.22

on the following point. By the end of the Tokugawa era differences in the distribution of households by size which are referable to locality, to the various districts into which we have divided the material, vanish entirely. The distribution of households by size is identical in every region by the 1870s. We regard this as a conclusion of considerable importance for the study of the size and structure of the household, in Japan and everywhere else. Figures 14.4 and 14.5 have been prepared with international comparison in mind. The first contrasts distribution of households by size in Districts E, W and C (where the pattern of distribution stabilises early) for the years 1751 to 1800, with the distribution worked out by Laslett for 91 communities in the 1564–1821 period for England. The Japanese percentage figures for MHS refer to 98 communities, consisting of 6,111 households with a total population of 27,866, and a general MHS equal to 4.56. In the case of Japan the distribution of households by size is concentrated in the range from 3 to 5, which appears to be somewhat higher than that in England. In Japan indeed 55.8% of households consist of 3 to 5 members, as against 47.1% in England. Households with 1 to 2 members and those with more than 7 members were commoner in England than Japan. Figure 14.5 shows the distribution of communities by mean household size with a unit interval of 0.5 persons. It will be seen that in the period 1701 to 1750 MHS is highly dispersed. The largest group, that containing 5 to 5.5 persons,

254

12:27:16:11:09 Page 254 Page 255

SIZE OF HOUSEHOLD IN A JAPANESE COUNTY

Figure 14.4 Percentage distribution of households by size, Japan and England Japan: 98 communities, 1751–1800 (MHS 4.56) England: 91 communities, 1564–1821 (MHS 4.72)

Figure 14.5 Percentage distribution of communities by mean household size, Japan and England

occupies only 12% of the total area of the distribution. However by 1801 to 1850 the interval from 3.5 to 5.0 persons occupies 80% of the distribution area, whilst in England 80% of households are concentrated in the 4 to 5.5 member area. The resemblance of the shape of the English graph to that of Japan in the ﬁnal stage of

255

12:27:16:11:09 Page 255 Page 256

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

household development there suggests to us the possibility that this distribution may be characteristic of societies where the process of reduction in household size and of variation in size between households is well advanced. It is a notable fact in comparing Japan with England that the values of MHS and variance of MHS in the final years of the Tokugawa era were actually lower than those found in England between the sixteenth and the nineteenth centuries. When the rate of shrinkage in household size is looked at geographically some sort of pattern emerges. It is in the flat area around Lake Suwa that MHS converges most rapidly and distinctly. In the Y and C Districts, especially in the Y region, which is in a more remote mountain area, MHS remains quite high until late in the Tokugawa era. All this suggests to us that the trend in household size can be thought of as one accompanying the spread of a monetary economy. In the final period the MHS of all communities converges in the 3.0 to 5.0 region.

CORRELATION COEFFICIENCY ANALYSIS OF THE MHS What variables were important in determining the MHS? If we suppose that size of household reflects all types of economic and social forces affecting human life, the number of such variables increases without limit. Our present object is to examine the correlation of MHS with a number of the variables which can be recovered from historical data, and then use these correlations as a guide to the understanding of the composition of the Japanese household in Tokugawa times. The variables we have chosen are more demographic in nature than those used in the studies of England.7 This is due to the limitations of the data, since it is not possible to determine occupation or social position from the shu¯ mon-aratame-cho¯. The following fifteen possible explanatory variables for the MHS have been selected, all of which are measurable. (1) The ratio of the number of households with two or more married couples to the number of households with at least one couple. It has been said that during the Tokugawa period the fragmentation of large agricultural households into small households was very common, as was the dissolution of multiple family groups. To confirm this statistically we must examine the correlation between this variable and MHS. (2) The average number of married couples per household, where the average is taken only over households with at least one such couple. (3) The ratio of households with a married head to the total number of households. (4) The ratio of the number of households with three or more generations to the total number of households. (5) The ratio of households without resident kin to the total number of households. If variable (5) is correlated with MHS, the sign of the coefficient should be negative. If we define resident kin more precisely, they are the members of the household which are left after exclusion of the head of the household, his wife, his lineal ascendants and descendants, and his servants. It is impossible to determine from the data itself whether resident kin are relatives of the husband or of the wife.

256

12:27:16:11:09 Page 256 Page 257

SIZE OF HOUSEHOLD IN A JAPANESE COUNTY

(6) The ratio of the number of households with one or more of several types of servants to the total number of households. (7) The average number of servants in households with one or more servants. (8) The number of servants in the total population. Since decline in household size may have been due to the dissolution of agricultural units dependent upon servant labour, explanatory variables (6), (7) and (8) serve to test this theory. (9) The number of unmarried children of the head of the household. (10) Fertility. Since fertility cannot be measured in a direct way from the data, we have used the ratio of the number of children under ten to the number of wives between sixteen and fifty as a function of fertility, and as a substitute variable. Whether this is a valid measure of fertility or not can be confirmed from Yokouchi, where fertility can be measured more directly. (11) The proportion married of males between sixteen and sixty. (12) The proportion married of females between twenty-one and forty, the age range of highest fertility. (13) Age at marriage for women. Since this cannot be computed directly from the data, we have taken the age when the proportion married exceeds 50% and used it as a function of the age at marriage. This proportion has been arrived at graphically, by joining together the levels showing proportion married in successive five-year female age groups and observing the point at which this line crosses the 50% level. For example, if in one age group in one village the proportion married for females in the age group sixteen to twenty is 40%, while that for twenty-one to twenty-five-year-olds is 60%, drawing a straight line between these two levels and locating the point where the line crosses the 50% level gives an estimate of the average marital age, in this case 20.5 years. As in the case of variable (10), we can compare the results of this approximation with the age at marriage computed for Yokouchi, where data permit age at marriage to be exactly calculated from family reconstitution. (14) The sex ratio. (15) The ratio to the total population of persons leaving the household temporarily for outside work. The above fifteen variables were employed in correlation coefficiency analysis, where MHS was taken as the dependent variable. The correlation coefficients (r) using all 382 cases are given in Table 14.3. The correlation of variables numbers (11: proportion of males married), (13: female age at marriage), (14: sex ratio) and (15: absent workers) with the MHS is low, and may be regarded as insignificant. For the remaining variables the coefficients are all significant at the 1% level. (For 400 degrees of freedom at the 1% significance level the value of r must be 0.128 or higher.) The correlations obtained for variables numbers (1), (2), (3), (4), (5), and (9) are particularly high.8 Variables (1) and (2) relate to the number of married couples per household and variable (4) measures the number of generations per household which is virtually the same as the number of married couples per household. It is clear that the size of the household is strongly affected by the number of married couples which it contains. The high negative value for variable (5) implies that the number of resident kin has likewise a marked influence on the size of households. Variable (9) is the number of unmarried children of the head

257

12:27:16:11:09 Page 257 Page 258

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

Table 14.3 Variables correlated with mean household size (382 communities)

Coeﬃcient when household size is correlated with: (1) Proportion of households with two or more couples in the households 0.835 with at least one married couple (2) Mean of the number of couples in households with at least one 0.899 married couple (3) Proportion of households headed by married couples 0.600 (4) Proportion of households with three or more generations 0.529 (5) Proportion of households without resident kin −0.684 (6) Proportion of households with servants 0.444 (7) Mean number of servants in households with servants 0.248 (8) Mean number of servants over all households 0.402 (9) Mean number of unmarried sons and daughters of household heads 0.529 (10) Fertilitya 0.319 (11) Proportion, males married, age 16–60 0.032 (12) Proportion, females married, age 21–40 −0.167 (13) Mean age at marriage, femalesb −0.045 (14) Sex ratio 0.021 (15) Proportion of those working temporarily away from home 0.086

a See pp. 263–4, 271–4. b See pp. 263–4, 271–2.

of the household, and, like variable (10), may be regarded as an indication of fertility. No comment appears to be necessary on variable (3). The set of variables with the next highest correlation coefficients are nos (6), (7), (8), (10) and (12) covering servants, fertility and proportion married. Variable (8) indicates the number of servants in the total population. It will be seen that the earlier the period, the larger the number of servants per household and we shall discuss this relationship below. As for variable (10) – children per married woman – it may be remarked that a marked fluctuation in the death rate amongst children, which is certainly possible at that time, would have given rise to a higher correlation of this variable with mean household size than is found in our results. Variable (12) – proportion of women married – yields a negative correlation with MHS, but a high proportion married has both a positive and a negative effect on the size of households. The positive effect occurs because the birth rate increases, and the negative effect because marriages give rise to independent households of smaller size. The above analysis relates to all regions and all time periods. It has been emphasized that MHS itself varies widely from region to region, and the same can be said about variables (1) to (15). For this reason, it is necessary to calculate correlation coefficients for regions and by time period; and the results of this are given in Table 14.4. The variable most closely correlated with the MHS over the whole period is the number of married couples per household. As we shall emphasise later in the discussion, it is interesting to notice how other factors become more important after this variable stabilises. For example in District E, the correlation coefficient for married couples gradually declines, and other variables, such as the number of generations per household, or the number of children per household rise to

258

12:27:16:11:09 Page 258 Page 259

SIZE OF HOUSEHOLD IN A JAPANESE COUNTY

Table 14.4 Variables correlated with mean household size, by district

(1) W District 1671–1700 1701–1750 1751–1800 1801–1850 1851–1870 Total (n = 9) (n = 20) (n = 24) (n = 12) (n = 6) (n = 71)

Variable Correlation coeﬃcients 1 0.606 0.834** 0.873** 0.540 0.913* 0.785** 2 0.474 0.763** 0.891** 0.502 0.941** 0.715** 3 0.437 0.505* 0.492* 0.685* 0.240 0.627** 4 0.705* 0.260 0.704** 0.684* 0.693 0.511** 5 −0.926** −0.626** −0.852** −0.418 −0.832* −0.696** 6 −0.059 0.385 0.491* –– 0.467** 7 −0.025 0.810** −0.002 –– 0.010 8 −0.083 0.448* 0.426* –– 0.410** 9 0.899** 0.723** 0.690** 0.936** 0.884* 0.874** 10 0.059 0.500* 0.038 0.209 −0.088 0.361** 11 −0.617 −0.200 0.210 0.528 0.082 −0.036 12 0.047 −0.289 0.506* 0.501 0.581 −0.192 13 0.950** −0.719** 0.486* 0.499 – 0.171 14 −0.027 −0.199 0.454* 0.004 0.688 −0.042 15 −0.102 0.204 −0.542** 0.473 −0.556 0.365** (2) E District 1671–1700 1701–1750 1751–1800 1801–1850 1851–1870 Total (n = 32) (n = 44) (n = 55) (n = 23) (n = 19) (n = 173)

Variable Correlation coeﬃcients 1 0.620** 0.864** 0.728** 0.478* −0.013 0.734** 2 0.652** 0.869** 0.733** 0.581** 0.094 0.792** 3 0.429* 0.596** 0.581** 0.589** 0.509 0.703** 4 0.179 0.643** 0.641** 0.734** 0.558* 0.279** 5 −0.378* −0.610** −0.597** −0.234 −0.401 −0.472** 6 0.473** 0.240 0.391** –– 0.640** 7 0.436* −0.100 −0.015 –– 0.287** 8 0.607** 0.210 0.235 –– 0.632** 9 0.307 0.349* 0.157 0.194 0.575** 0.647** 10 0.121 0.049 0.125 0.120 0.272 0.328** 11 −0.438* 0.365* 0.145 0.182 0.048 −0.034 12 −0.340 0.315* −0.074 0.279 0.030 −0.205** 13 −0.146 −0.119 −0.088 0.170 −0.008 −0.034 14 0.294 −0.056 0.024 0.254 0.047 0.087 15 0.096 −0.065 −0.156 0.040 −0.362 0.306** (3) C District 1671–1700 1701–1750 1751–1800 1801–1850 1851–1870 Total (n = 9) (n = 16) (n = 19) (n = 13) (n = 9) (n = 66)

Variable Correlation coeﬃcients 1 0.918** 0.789** 0.922** 0.833** −0.041 0.877** 2 0.974** 0.906** 0.942** 0.943** −0.044 0.929** 3 0.003 0.581* −0.166 0.196 −0.167 0.528** 4 0.840** 0.814** 0.678** 0.256 0.500 0.635** 5 −0.460 −0.694** 0.868** −0.687** −0.766* −0.678** 6 0.210 0.708** 0.059 −0.067 – 0.705** (Continued Overleaf)

259

12:27:16:11:09 Page 259 Page 260

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

(3) C District Continued 1671–1700 1701–1750 1751–1800 1801–1850 1851–1870 Total (n = 9) (n = 16) (n = 19) (n = 13) (n = 9) (n = 66)

7 0.181 0.746** 0.287 ––0.573** 8 0.300 0.837** 0.117 −0.067 – 0.707** 9 0.146 0.692** 0.283 0.305 0.564 0.690** 10 0.011 0.750** 0.240 0.229 0.345 0.585** 11 −0.180 −0.356 −0.299 −0.207 −0.641 −0.260* 12 0.415 −0.460 0.156 0.040 −0.550 −0.381** 13 0.438 0.379 −0.359 −0.623* −0.518 0.171 14 0.438 −0.333 0.214 0.340 −0.455 −0.028 15 −0.598 0.347 0.108 0.115 0.335 0.159 (4) Y District 1671–1700 1701–1750 1751–1800 1801–1850 1851–1870 Total (n = 14) (n = 20) (n = 21) (n = 11) (n = 6) (n = 72)

Variable Correlation coeﬃcients 1 0.717** 0.842** 0.874** 0.748** 0.162 0.811** 2 0.846** 0.960** 0.980** 0.977** 0.333 0.946** 3 0.483 0.248 0.523* 0.564 0.215 0.593** 4 0.570* 0.622** 0.403 0.653* 0.869* 0.499** 5 −0.285 −0.368 −0.828** −0.880** −0.767 −0.654** 6 0.578* −0.238 –––0.273 7 0.456 0.539* –––0.511** 8 0.473 −0.033 –––0.258 9 0.539 −0.125 0.568** 0.233 0.477 0.422** 10 0.183 0.016 0.347 0.178 0.476 0.299* 11 0.163 0.457* −0.273 0.138 −0.152 0.018 12 −0.003 0.355 0.021 0.165 −0.038 −0.195 13 0.572* −0.258 −0.433* −0,598 −0.094 −0.081 14 −0.082 −0.318 0.486* −0.183 −0.789 0.010 15 0.044 −0.215 0.007 0.205 – 0.236

* Signiﬁcant at 0.05. ** Signiﬁcant at 0.01.

prominence. In Districts C and Y we find that MHS remained high until late in the Tokugawa period and that the number of resident kin was significant to MHS over the whole period studied. Servants do not seem to have had a uniform effect on MHS. The correlation coefficient in District E is high from 1671 to 1700, but for District W it is negative and very low. District C shows high values for 1701 to 1750, but after 1801 the relationship disappears, because of the absolute drop in the number of servants. The most interesting result apparent from Table 14.4 is that there are several coefficients which are high in the Total column even though they are low for each sub-period. For example in District E, variable (10) – children per married woman – is low in every period, never significant even at the 5% level. However, in total the value of r is 0.328, which is above the critical value required at the 1% level. When trying to decide on determinants of MHS, therefore, it is necessary to

260

12:27:16:11:09 Page 260 Page 261

SIZE OF HOUSEHOLD IN A JAPANESE COUNTY

distinguish between those having an influence in the long run and those having an influence in the short run. Variable (10), which indicates fertility, is always significantly related to MHS over the whole period of analysis, that is in the Total column, at the 5% level or better, but in the sub-periods the relationship barely reaches the level required for 5% significance. Other variables which perform very differently in the long and short run are nos (3), (6), (7), (8) and (9). Variables nos (1), (2), (4) and (5), on the other hand, appear to be effective in both the short and long run.

CHANGES IN THE VARIABLES As was mentioned in the previous section, some of the variables which inﬂuence the MHS underwent substantial change during the period studied. In this section we shall consider changes in the major variables.

Number of Married Couples Per Household In each region and in every period except the last, the strongest relationship is that between MHS and the number of married couples. The trends in the average number of married couples per household over time for each region are given in Fig. 14.6. Values for Districts E and W, east and west of Lake Suwa, are very similar and they decline together slowly over the years. For District Y – the new village area – before the 1810s, and especially up to the 1760s, the ﬁgures remain at a very high level. District C – the mountains – is in between the two mentioned above, so far as these ﬁgures are concerned. Data on the ratio of the number of households with only one married couple, to all households, for the four regions over time are shown in Figure 14.7. If we adopt the principle that the small family

Figure 14.6 Numbers of married couples per household, by district

261

12:27:16:11:09 Page 261 Page 262

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

is dominant when this ratio reaches 50%, the period of dominance arrives in 1730 in District W, in 1720 in District E, in 1800 in District C and in 1810 in District Y. The time lag between regions is relatively small. In Districts W and E, this ratio stabilises after 1780 at about 65% to 70%, except for a short period. In 1830 District C joins Districts W and E in this range. When this ratio stabilises, its correlation with MHS goes down. A high correlation between the number of couples per household and MHS is characteristic of periods when shrinkage in the household size is not taking place. Table 14.5 shows the average value by region and period for variable (1), that is to say the ratio of households with two or more couples to all households with at least one married couple. Table 14.6 shows the mean of means for variable (2) i.e. the average number of couples in households with at least one couple.

Resident Kin Per Household Shrinkage in household size was also a function of the drop in the number of resident kin per household, and it will be seen that the number of resident kin was large in areas like Districts Y and C where the MHS was also large. Figure 14.8 shows the trend in the average number of resident kin per household. Up to the

Figure 14.7 Ratio of households with one couple to total number of households, by district

Table 14.5 Variable 1: Ratio of households with two or more couples to all households with at least one married couple, by district

W E C Y Total

1671–1700 0.527 0.342 0.505 0.481 0.421 1701–1750 0.333 0.303 0.511 0.658 0.413 1751–1800 0.262 0.225 0.316 0.533 0.302 1801–1850 0.244 0.215 0.259 0.369 0.259 1851–1870 0.188 0.188 0.161 0.278 0.195 Total 0.306 0.261 0.357 0.511 0.333

262

12:27:16:11:09 Page 262 Page 263

SIZE OF HOUSEHOLD IN A JAPANESE COUNTY

Table 14.6 Variable 2: Mean number of couples in households with at least one married couple, by district

W E C Y Total

1671–1700 1.97 1.45 1.85 1.83 1.66 1701–1750 1.41 1.40 1.79 2.34 1.65 1751–1800 1.32 1.25 1.43 2.05 1.43 1801–1850 1.25 1.24 1.29 1.37 1.28 1851–1870 1.20 1.18 1.16 1.30 1.20 Total 1.41 1.31 1.51 1.92 1.48

Figure 14.8 Number of resident kin per household, by district

year 1820 we ﬁnd once more a trend which is unique to District Y, the new villages, but after this point the average converges to 0.5 for all regions.

Birth Rate and Fertility Both birth rate and fertility might be considered as determinants of the size of households, but in the data we have used in this study and with the methods we have employed, it is impossible to observe these two variables directly. As has already been stated, all that could be done was to compute a fertility index from the ratio of children to wives. These are two ways of reckoning this index. The first is by age groups, and the second is by relationship to the head of household. In the first case the ratio of children under ten in the community to the number of married women between twenty-one and forty is computed, and in the second the number of resident unmarried children of the head of the household. There are several points which must be borne in mind when these figures are used as functions of the birth rate or of the degree of fertility. The first is the death rate among children. If there are large changes in this rate, then it will be impossible to assume that there is a stable relationship between the substitute ratio series and the actual birth and fertility rates. This point can be examined and explained once again from the detailed study of the village of Yokouchi, where we have undertaken family reconstitution.

263

12:27:16:11:09 Page 263 Page 264

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

The next issue is that of mobility, which will create problems if it affects children under ten years of age. Geographic movement of children in that age group independent of their parents is difficult to imagine, but it must be kept in mind that the practice of adopting both male and female children was very common in Japan during this period. In some cases it is possible to distinguish adopted children from other children, but not in every case. The data for Yokouchi provides a check here as well. The third question about the use of the substitute series for fertility concerns the extent to which information on the birth and fertility rates is consistent with information on the population married and on age at marriage. If we suppose that the proportion married and the age at marriage vary widely, then these substitute variables will not be good indications of the birth rate, which makes it essential that the proportion married and age at marriage be closely examined where possible. There is yet a further consideration which arises because the number of unmarried children depends on the average age at marriage. If the age at marriage falls, the number of unmarried children declines, but because the birth rate rises the figures for MHS will move in the opposite direction. In actual fact, however, it is hard to imagine that age at marriage would vary sufficiently over a short period to affect the birth rate. Taking all these things into consideration, it seems perfectly possible that the correlation of our substitute fertility figures with MHS could be low over short periods. We find that the correlation results, with the exception of those for District W, are in fact low and unstable for sub-periods, but higher in the long term. Table 14.7 shows the trend in the number of unmarried children per household, and Table 14.8 shows the ratio of children under ten to the total number of wives between 16 and 50 for each community. As we can see from

Table 14.7 Variable 9: Mean number of unmarried children of household head, by district

W E C Y Total

1671–1700 2.80 2.53 2.61 2.49 2.57 1701–1750 1.91 1.95 1.84 1.97 1.93 1751–1800 1.36 1.45 1.23 1.37 1.38 1801–1850 1.43 1.34 1.19 1.25 1.31 1851–1870 1.47 1.45 1.29 1.35 1.40 Total 1.72 1.76 1.57 1.73 1.72

Table 14.8 Variable 10: Fertility:a ratio of children aged 10 or under to wives aged 16–50, by district

W E C Y Total

1671–1700 2.25 2.24 2.33 2.35 2.28 1701–1750 2.04 1.82 1.90 1.99 1.91 1751–1800 1.73 1.62 1.48 1.49 1.59 1801–1850 1.84 1.65 1.59 1.56 1.66 1851–1870 1.64 1.87 1.58 1.63 1.73 Total 1.89 1.82 1.74 1.82 1.81

a See pp. 271–4.

264

12:27:16:11:09 Page 264 Page 265

SIZE OF HOUSEHOLD IN A JAPANESE COUNTY

Table 14.8, differences in the ratio from district to district are not large for simultaneous periods, and changes over time seem to be larger than the variation across communities for any given interval. In all districts the figures are highest for 1671 to 1700, lowest for 1751 to 1800 (except only in District W), and rise slightly thereafter. The figures given in Table 14.8 show virtually the same trends as those in Table 14.7. We can state quite confidently that the birth rate and degree of fertility in all districts of Suwa County showed a virtually identical pattern over time.

Proportion Married The marriage rate for males aged sixteen to sixty is given in Table 14.9, although it is not directly related to fertility. Except for the fact that District Y – the area of new settlements – appears to be slightly higher than the rest, there seem to be no meaningful trends in the series, and as we have seen, correlation with MHS was found to be low. The proportion of married females aged twenty-one to forty is a diﬀerent matter and is given in Table 14.10. In every district it is apparent that between the two sub-periods 1671 to 1700 and 1701 to 1750, and the rest of the period of analysis, there is a considerable diﬀerence in the value of this index. The rate is ten per cent higher after 1750 than before, and it is also clear that the rate is slightly higher in District Y than elsewhere.

Age at Marriage Since the average age at marriage cannot be measured directly from the data, we have used, as has been said, a substitute series. Table 14.11 gives the trend for females for all our villages over all sub-periods, and the ﬁgures vary little across

Table 14.9 Variable 11: Proportion of married males, age 16–60, by district

W E C Y Total

1671–1700 0.614 0.553 0.577 0.663 0.589 1701–1750 0.552 0.549 0.558 0.567 0.555 1751–1800 0.579 0.561 0.602 0.639 0.585 1801–1850 0.557 0.592 0.616 0.584 0.588 1851–1870 0.638 0.556 0.578 0.581 0.577 Total 0.577 0.560 0.587 0.611 0.577

Table 14.10 Variable 12: Proportion of married females, age 21–40, by district

W E C Y Total

1671–1700 0.699 0.749 0.652 0.801 0.739 1701–1750 0.672 0.740 0.725 0.731 0.722 1751–1800 0.812 0.811 0.812 0.838 0.816 1801–1850 0.768 0.824 0.817 0.825 0.811 1851–1870 0.804 0.808 0.794 0.826 0.807 Total 0.750 0.783 0.768 0.798 0.777

265

12:27:16:11:09 Page 265 Page 266

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

Table 14.11 Variable 13: Mean age at time of marriage, females,a by district

W E C Y Total

1671–1700 20.0 20.3 19.9 18.5 19.9 1701–1750 22.0 21.2 21.6 21.1 21.3 1751–1800 19.9 20.6 19.5 18.3 19.9 1801–1850 19.7 20.2 20.0 20.0 20.0 1851–1870 21.2 20.7 21.3 20.3 20.7 Total 20.4 20.6 20.6 19.7 20.4

a See pp. 507–508.

districts, but do change significantly over time. The highest marriage ages occur in 1701–50 in every district, while the periods when the average age at marriage is lowest are 1751–1800 and 1801–50. This result appears to contradict the data in Tables 14.7 and 14.8, and an explanation is required. First it is clear that the rate of child mortality fell over the whole period, and that this must have led to a lengthening of the average life span. The proportion of children who lived at least to the age of ten in Yokouchi was 62% in 1671–1700 for a sample of 227 births, 60% in 1701–25 for 302 births, but 70% in 1726–50 for 278 births and 79% in 1751–75 for 309 births. Thus the survival rate for the 1–10 age group increased over the years, and an average life expectancy at two years of age based on estimates of the age of death, went as follows; in 1671–1700 it was 24.8 years for males, 29.8 years for females; during the period 1751–75 it increased to 35.9 for males and 38.3 for females. This increase in expectation of life is also clear from the ratio of births to the number of married females able to bear children, and from the ratio of the number of children under ten to the number of females able to bear children. The first ratio represents degree of fertility, and the second ratio represents fertility in combination with the child mortality rate. If both rates are converted to index form and compared, we find that in the first period the two indices gradually approach each other, while in the later period the index representing both fertility and the effects of changes in the child mortality rate becomes higher. Since this index is inversely related to the child mortality rate this gives us strong grounds for supposing that the change can be explained solely in terms of the decrease in the child mortality rate. If this is in fact the case, and if there were no large changes in the proportion of females married or in the age at marriage, the average number of children per household should have increased. However as Tables 14.7 and 14.8 show, the number of children per household in fact decreased. How can this be explained? In the analysis of Yokouchi, the relationship between the movements in the average age at marriage of females and the number of births was carefully examined and it was found that in the latter half of the period the consistent relationship between the two sets of figures disappeared. From this we drew the tentative conclusion that some kind of artificial constraint was operating on the number of births. Family reconstitution carried out for this village led us to the same conclusion. The fact that child mortality was declining over the whole period means that

266

12:27:16:11:09 Page 266 Page 267

SIZE OF HOUSEHOLD IN A JAPANESE COUNTY

the imposition of birth control which we have assumed must have kept down the number of births to a very low level. It is this we believe which explains why the number of children per household was lower during the second half of the period.

Servants9 The number of servants per household varies widely. The category ‘servants’ includes (i) so-called ‘subordinate servants’, who were engaged in agricultural labour and who remained with the household on a hereditary basis, and (ii) domestic servants. It would seem that the work-system of the agricultural household changed over the period of analysis, finally coming to depend primarily on family labour. The number of servants in category (i) accordingly decreased. Servants of the second type persisted longer in some agricultural households, but eventually their numbers also declined. There were cases, as we have seen, where the number of servants dropped to zero and it was impossible to compute a correlation coefficient. The trend in the variable representing the number of servants per household, variable (8), is given in Fig. 14.9. It is clear from the graphs that District C – the mountainous area – had a larger number of servants than elsewhere. In other districts the average number of servants per household had already fallen to 0.1 as early as 1730. The figure for District C finally fell below 0.1 in 1770. The correlation between MHS and mean number of servants per household for the whole period studied can therefore only be measured for District C and for the first two periods at most in the case of the remaining districts. In Tables 14.12 and 14.13 will be found the time series for variable (6), the ratio of the number of households with servants to the total number of households, and for variable (7) the average number of servants per household. At the peak, in the series for District C, 36.1% of the households had one or more servants. The number of servants

Figure 14.9 Number of servants by household, by district

267

12:27:16:11:09 Page 267 Page 268

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

Table 14.12 Variable 6: Proportion of households with servants, by district

W E C Y Total

1671–1700 0.212 0.132 0.361 0.091 0.167 1701–1750 0.060 0.042 0.166 0.063 0.070 1751–1800 0.012 0.005 0.043 – 0.011 1801–1850 ––0.002 – 0.000 1851–1870 ––––– Total 0.048 0.037 0.102 0.035 0.050

Table 14.13 Variable 7: Mean of the number of servants in households with servants, by district

W E C Y Total

1671–1700 2.63 2.16 3.01 1.20 2.24 1701–1750 2.57 1.82 3.24 1.65 2.17 1751–1800 3.58 1.00 2.19 – 2.03 1801–1850 ––2.00 – 2.00 1851–1870 ––––– Total 2.88 1.83 2.76 1.48 2.17

did not simply decline progressively, as perhaps might be expected, for households having servants suddenly ceased to employ them.

EFFECT OF EXPLANATORY VARIABLES IN COMBINATION Let us now turn to the examination of the combined effects of various explanatory factors which we have been discussing. Figure 14.10 traces the course of what might be called residual mean household size after resident kin and servants have been subtracted, and this residual is taken to be what is called elsewhere in this volume the primary, elementary or nuclear family of the household head, his wife and children. It will be noticed that the large differences in MHS across various districts fall markedly when resident kin and servants are excluded in this way. It appears, therefore, that the reason for the higher MHS in Districts Y and C was chiefly the larger number of resident kin per household. It is clear from the data in this figure that convergence to a smaller household size had already begun in 1740.

Temporal and Spatial Factors Two types of variable can be distinguished amongst those which we have been analysing: (a) those that vary over time but not across districts and (b) those that vary across districts. The variables presented in Tables 14.4, 14.7, 14.8, 14.10 and 14.11 show variation over time but not across districts; and those presented in Tables 14.5, 14.6, 14.12, 14.13 and 14.14 show variation across districts but not over time. The variables showing variation over time include the number of married couples per household, the number of servants and the generational depth of households, whilst those showing variation across districts comprise the

268

12:27:16:11:09 Page 268 Page 269

SIZE OF HOUSEHOLD IN A JAPANESE COUNTY

Figure 14.10 MHS after subtracting resident kin and servants, by district

Table 14.14 Proportion of households with three or more generations, by district

W E C Y Total

1671–1700 0.375 0.267 0.330 0.331 0.305 1701–1750 0.266 0.226 0.313 0.447 0.292 1751–1800 0.270 0.230 0.221 0.303 0.250 1801–1850 0.290 0.296 0.307 0.336 0.304 1851–1870 0.283 0.297 0.270 0.391 0.303 Total 0.287 0.252 0.281 0.361 0.284

proportion married, the age at marriage and the number of children. We shall call the variables of type (a) temporal factors and those of type (b) spatial factors. When variables exhibiting spatial or temporal variation are used in correlation analysis with MHS, it will be expected that there should be diﬀerences in the types of correlation obtained. In the case of temporal factors, the correlation with MHS is strong, in whatever area or in whatever period it is examined. The same may be said of the Total column for each of these temporal variables. But in the case of the spatial variables, the relationship is weak, especially in the analysis by sub-period for each district. Nonetheless in the Total column correlation values for a few of the spatial variables are fairly high.

Population Change and MHS Finally we would like to examine the changes in the population and in MHS and the relationship between the two. In a study of another area, the province of Bungo, in the 1620s, published in 1968,10 the author concluded that the large size of the households there was due to some kind of constraint upon the increase in

269

12:27:16:11:09 Page 269 Page 270

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

population. Conditions were such that the number of agricultural serfs living in the household went up, but so few of them married that the birth rate was held down. Thus a balance seemed to be maintained between births and deaths. If this conclusion was in fact correct, decline in MHS due to a decline in the number of resident kin and servants meant that one of the factors keeping the growth of population in check was very gradually being eroded in Suwa County during the Tokugawa era. For this reason it would be natural to expect some increase in the population. Figure 14.2 shows that there was an upward trend which was especially strong in District Y, and it should be noted that the rise in the population coincides with the change in the MHS.11 In almost all cases decrease in MHS was accompanied by increase in the population, except for the 1671 to 1720 period in District Y, where although the population was increasing very rapidly, MHS was in fact increasing.12 We undertook the following exercise in order to test this observation. The amount of increase in population in one district in one ten-year interval was correlated with the amount of decrease in MHS in the same district during the same sub-period. Only about half the possible cases could be analysed in this way and the results were somewhat complicated for presentation here, and rather variable in quality. It can be stated, however, that there was a significant inverse relationship between household size and population growth in all districts except District C, the mountain area. There were exceptions in some periods and regions and in District E, east of the lake, correlation was low in the earlier period. Nevertheless, it appears that we can conclude from the above analysis that the decrease in MHS or the transition to the small family was generally accompanied by an increase in the population. Moreover, it can usually be said that the faster the fall in the size of household, the faster the rate of population growth. In Suwa County it seems that the increase of population which accompanied evolution towards the small family rose to a peak in the earlier period of the Tokugawa era. After the middle of the eighteenth century when, except in District Y, the nuclear family stabilised in size, population restrictions were introduced. The large household system, which acted as a restraint on the growth of population, gradually relaxed its influence and population began to rise rapidly, making new restraints necessary. The method chosen to restrain the growth of population was not a return to the large household system, but cruder artificial restraints on population growth.

ANALYSIS OF THE VILLAGE OF YOKOUCHI So far we have analysed 38 communities in Suwa County, divided into four districts. One village of the 38, Yokouchi, in District E on the plain east of Lake Suwa (see map in Figure 14.1), has the most continuous collection of the shu¯ mon-aratame-cho¯ for the 201 years from 1671 to 1871. No less than 144 individual years are covered by the listings which have survived. The object of this ﬁnal section is to make use of this quite exceptional source of information in order to try to resolve some of the problems left outstanding in our analysis so far. We have two general purposes in mind. The ﬁrst is to test one or two of the hypotheses which have been suggested in the foregoing discussion of the 38 communities

270

12:27:16:11:09 Page 270 Page 271

SIZE OF HOUSEHOLD IN A JAPANESE COUNTY

in the Suwa area. The second is to exploit some special features of the data for Yokouchi, in order to attempt a more thorough analysis of the major issues posed in the earlier portion of this chapter. The first problem is to decide how reliably age at marriage can be measured indirectly by the method described above, that is by making use of the trend revealed by the changing proportion of married persons in successive age groups. The results of this exercise can be checked at Yokouchi since both age at marriage and proportion married at various ages can be measured directly from the data over practically the whole Tokugawa era. Figure 14.11 presents the outcome of this test. It is clear that the indirect measure adopted is a good one. Indirect computation using the proportion married as described above should, therefore, usually give a close approximation to the average age at marriage. The next problem is to pronounce on the efficiency of our indirect measures of fertility. Fertility is calculated as the number of births to every woman of mar- riageable age, so the problem of measuring it will be solved if we can compute the ratio of children born to the number of wives present at a given point in time. The age of all persons and the number of births is available for Yokouchi, so that it is possible to compose the two following ratios: first the ratio of all births to the number of wives in the most fertile age group, 21 to 40, and second the ratio of the number of children under ten to the number of wives in the age group 16 to 50. This comparison is given in Figure 14.12, and its noteworthy features are as follows: (i) Both the actual and the substitute series have roughly the same trend. (ii) There is a time lag between the two series: if we examine the peaks we find that the second lags behind the first by about seven years. (iii) There is change over time. In the early period the first ratio is higher than the second; in the middle period the two ratios are approximately equal, but in the later period the first ratio

Figure 14.11 Age at marriage at Yokouchi

271

12:27:16:11:09 Page 271 Page 272

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

Figure 14.12 Fertility measures for Yokouchi (nine-year moving averages)

is lower than the second. (iv) Both ratios reach a low point in the period 1770–90. (v) Both series have a cycle of almost twenty years. This set of results raises some detailed interpretative issues, but we must pass them over on this occasion. There are one or two other questions which can be referred to research results from Yokouchi. The first is the extent of the adoption of children under ten years of age. Adoption played an important role in Japanese life since it was an expedi- ent against the extinction of family lines, and the practice was fairly common. This raises the possibility of error in using figures for the number of children under age ten as a measure of fertility. Table 14.15 shows the prevalence of the practice at Yokouchi, where only ten per cent of the total number of the children in question were adopted under the age of ten. This is not large enough to bias any computation. As for child mortality, Table 14.16 has been drawn up from previous research results so as to provide finer detail. It shows that the death rate of children decreased substantially over the Tokugawa era. Those living to age eleven or more rose from 62% early in the period to 79% at the end. Finally the results for age specific fertility obtained from the analysis of family reconstitution forms may be mentioned. These are given in Figure 14.13 and reveal clear differences in level at different periods. We believe that this was due to some restraint on population increase, rather than to a decrease in fecundity. The distribution of the number of births by age at marriage is set out in Figure 14.14 as evidence for this hypothesis. During the earlier period, when we believe there were no restrictions on population there was an obvious relationship between the number of births and the age at marriage, and it is in our view a highly significant

272

12:27:16:11:09 Page 272 Page 273

SIZE OF HOUSEHOLD IN A JAPANESE COUNTY

Table 14.15 Total adopted children and adopted children under ten at Yokouchi

Recorded as adopted

Periods Total Under age ten

1671–1700 11 1 1701–1750 13 0 1751–1800 38 0 1801–1850 35 2 1851–1870 32 4 Total 129 7 (5.4%)

Table 14.16 Child mortality and survival rate at Yokouchi

Year of birth 1671–1700 1701–1725 1726–1750 1751–1775 Total

Number of births 227 302 278 309 1,116 Number of deaths at 1 – 2 – 1 3 age: 2 32 37 19 14 102 31031161914 42426129 71 5 9 14 12 8 43 Sub-total 75 110 59 51 295 63411826 7355114 821429 911125 10 2 – 22 6 Sub-total 11 11 23 15 60 Grand total 86 121 82 66 355 Number of survivals to 141 181 196 243 761 age over 11 Rate of deaths age 0.33 0.36 0.21 0.17 0.26 1 to 5 Rate of deaths age 0.38 0.40 0.30 0.21 0.31 1 to 10 Survival rate of age 11 0.62 0.60 0.70 0.79 0.69

fact that this relationship becomes less marked during the later part of the period. The main feature of the sources for the village of Yokouchi is that enough has been preserved to allow us to follow for no less than 200 years the demographic activity of each individual in each household. This provides the opportunity for analysis by family reconstitution, the results of which have already been reported,13 and we should like to record here some of its implications for the problems raised above. Figure 14.15 shows trends in population, in number of households and in MHS over the 200-year period of analysis. Comparison with

273

12:27:16:11:09 Page 273 Page 274

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

Figure 14.13 Age-speciﬁc fertility at Yokouchi

Figure 14.14 Number of births by age at marriage (wives in complete families only) at Yokouchi

274

12:27:16:11:09 Page 274 Page 275

SIZE OF HOUSEHOLD IN A JAPANESE COUNTY

Figure 14.15 Trends in population, number of households and mean size of household at Yokouchi

Figures 14.2 and 14.3 suggests that the shape of these curves for Yokouchi is virtually the same as those exhibited by District E, the area east of Lake Suwa in which Yokouchi itself is located. Population growth does not quite conform, since the population of Yokouchi went on growing until the 1770s. Figure 14.16 sets MHS alongside the following: number of married couples per household; number of resident kin, and the number of unmarried children of the head of the household. The values of MHS and the trend in the MHS time series are virtually the same as those for District E as a whole. The feature of this evidence which best illustrates the decline of MHS is the change from households containing several married couples to households with one only. Where decline in MHS was accompanied by an increase in population it seems likely that this was usually the result of the break-up of households. In analysing this phenomenon we will ignore the sociological factors making for the preservation of the household line and concentrate on the demographic aspects of the fragmentation process. Table 14.17 shows the rate of household fission over approximate twenty-five year intervals. Each household usually divided into two, but there were also cases where one household broke up into three or four others. If the number of households which were split up in this way is divided by the total number of households, we find that the ratio is highest in the years around 1700. In general during the earlier half of the period the ratio is high, low from 1776 to 1850, and that there is a slight increase after 1851. The behaviour of this ratio corresponds well with the movement of the indices in Figure 14.15. The evidence from Yokouchi allows us to trace the fission process for a single household over the full 200 years. It is possible to work out from the successive listings the lineage of the households which had their origin in household number

275

12:27:16:11:09 Page 275 Page 276

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

Figure 14.16 Mean size of household: married couples, resident children and resident kin per household at Yokouchi

Table 14.17 Division of households at Yokouchi, by period

Dividing to Mean no. dividing Ratio of divided to Periods 234Total per year whole households

1671–1700 25 1 – 26 0.86 2.57 1701–1725 10 1 – 11 0.44 0.90 1726–1750 21 6 – 27 1.08 1.53 1751–1775 16 4 1 21 0.84 0.96 1776–1800 7 2 – 9 0.36 0.37 1801–1825 14 –– 14 0.56 0.56 1826–1850 14 –– 14 0.56 0.53 1851–1871 17 3 – 20 0.95 0.84 Total 124 17 1 142 0.71 0.93

12 of the ﬁrst year, 1671. This household was originally composed of three couples and a total of eleven members. Table 14.18 has been prepared from this evidence, giving for each sub-period the number of members and the number of couples in each household for the whole complex which originated in household number 12 of 1671. From this table it is clear that the tempo of change in MHS is about the same as that of the rest of the village and even of District E itself. The original household eventually gave rise to twenty-seven further households during the course of the 200-year period of analysis, that is there was on average a division once every 7.4

276

12:27:16:11:09 Page 276 Page 277

SIZE OF HOUSEHOLD IN A JAPANESE COUNTY

Table 14.18 A Yokouchi household of 1676 and its successor households: households, household members and married couples (For years when division actually took place)

Years Households Members Married couples

1676 1 12 3 1678 3 13 3 1689 4 18 5 1699 5 26 7 1706 6 32 7 1732 7 39 11 1734 9 47 13 1746 10 62 13 1763 12 62 16 1767 14 68 18 1769 16 70 19 1801 14 55 13 1802 14 49 12 1832 13 64 14 1836 14 72 15 1840 15 75 14 1844 16 82 16 1854 19 84 18 1859 19 77 18 1863 19 78 15 1870 19 77 19

years. Nine households died out, so that by 1871 the total number of households in the lineage was nineteen. The total number of members in all the related households at the end of the period was 6.4 times the number at the beginning, and the number of couples multiplied by 6.3. This increase in households and in married couples was only one-third of the increase in the number of households in the village over the period. There is no reason to believe that household number 12 of 1671 was an exception in the history of Yokouchi.14 Table 14.19 gives the change in the number of households for each lineage by twenty-ﬁve-year intervals. Original household number 20 gave rise to the largest number of successor households: twenty-ﬁve separate households arose from it over the 200 years. Household number 22, on the other hand, did not divide once over the entire period. Eleven of the original twenty-seven households died out completely, thus giving a rate of attrition of about 40%. On average the number of households increased 6.7 times in 200 years, when the number which actually survived is taken as the divisor. When the original number of households is taken as the divisor, the number of households increased four times. Reprinted from Household and Family in Past Time. Comparative studies in the size and structure of the domestic group over time. Edited with an analytic introduction on the history of the family, by Peter Laslett, with the assistance of Richard Wall, both of the Cambridge Group for the History of Population and Social Structure.

277

12:27:16:11:09 Page 277 Page 278

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

Table 14.19 Number of households in each lineage at Yokouchi

No. of households 1671 1700 1725 1751 1775 1801 1825 1850 1871

1 1335 656 67 2 1311 1–– –– 3 1112 32––– 41––– ––– –– 5 1112 333 33 6 1221 22––– 7 1122 233 44 81––– ––– –– 9 1123 334 45 10 1222 222 1– 11 1235 456 53 12 1 5 6 10 16 13 12 16 19 13 1448 999118 14 1 1 –– ––– –– 15 1 ––– ––– –– 16 1122 233 44 17 1124 544 46 18 1224 532 22 19 1223 544 53 20 14471111121625 21 1211 21––– 22 1111 111 11 23 1113 553 22 24 1222 354 44 25 1122 2–– –– 26 1 ––– ––– –– 27 1236 89111111 Othersa – 111 1281013 Total 27 46 50 77 101 95 97 109 120

a Including households which moved from elsewhere and households whose origin is unclear.

NOTES

1. See above, Chapter 4, pp. 125–158. 2. See above, Smith, Chapter 17, pp. 431–436, for an account of the origin and content of these registers. See also, Hayami, Population at the beginning of the Tokugawa period (1967). 3. Taeuber, Population of Japan (1958): 29. 4. Hayami, Aspects démographiques d’un village japonais (1969). 5. Hayami (1968 ii). 6. On this point see Nomura, Cultural conditions affecting population trends (1953). 7. See above, Laslett, Chapter 4, p. 155. 8. These differ sharply from the variables having a significant effect on mean household size in pre- industrial England, see above, Laslett, Chapter 4, pp. 155–156. 9. See also below, Nakane, Chapter 19, pp. 519–523. Compare the proportion of households with servants in pre-industrial England, above, Laslett, Chapter 4, pp. 151–152, and eighteenth-century Netherlands, above, van der Woude, Chapter 12, pp. 307–308, 314–315.

278

12:27:16:11:09 Page 278 Page 279

SIZE OF HOUSEHOLD IN A JAPANESE COUNTY

10. For example the following statistics are available for Hayami County for the year 1622. Married Unmarried Proportion married

Blood relatives 991 31 97.0% Servants 34 358 8.7% See Hayami (1968 i): 92. 11. In England in the late eighteenth century the tendency was for MHS and population to increase together; see Richard Wall, “Mean household size in England from printed sources”, in Household and Family in Past Time (ed. Peter Laslett), pp. 199–203, Cambridge, 1972. Philipp J. Greven Jr (“The average sizes of families and households in the province of Massachusetts in 1764 and in the United States in 1790: an overview”, in Household and Family in Past Time (ed. Peter Laslett), p. 555, Cambridge, 1972) also equates a rise in MHS with a rapid growth of population in two counties in Massachusetts. 12. This exception can be explained in the following way. The increase in the MHS for District Y during this period is not due to the appearance of a large household system having a large number of servants with a low marriage rate. For some reason the branching oﬀ of households after the marriage of the oﬀspring of the head of the household was restricted. Thus the number of direct relatives of the head of the household and resident kin increased above what it would have been with the normal amount of branching. For the high marriage rate in District Y, see Tables 14.11 and 14.12. 13. Hayami (1969). 14. Smith, see above, Chapter 17, pp. 438–440, 470–471, also assesses the durability of households in Tokugawa Japan. Compare Blayo’s analysis of the development of 184 conjugal households in the French village of Grisy-Suisnes, above, Chapter 9, pp. 261–264. See also Hammel’s account of population turnover in sixteenth-century Serbia, above, Chapter 14, pp. 343–344, 348.

279

12:27:16:11:09 Page 279 Page 280

Co-authored with Kurosu Satomi and ﬁrst published in Journal of Japanese Studies, vol. 27, no. 2, pp. 295–321, 2001.

15 Regional Diversity in Demographic and Family Patterns in Pre-industrial Japan *

ntil recently the population of pre-industrial Japan has been studied at the Unational level using bakufu estimates dating from 1721 to 1846 and at the village level using individual village and temple records extant from the mid- to late Tokugawa period. This has left us with an inadequate understanding of the population dynamics in the period. Now, from recent studies in historical demography and the history of family patterns, we have data that indicate clear regional differences in demographic patterns and family structure. However, these studies have yielded several findings that are seemingly paradoxical or at least require further study. Demographic patterns and family structure are two very significant societal phenomena that are interrelated. For example, when a certain family structure is observed, certain corresponding demographic patterns are usually found. The opposite also holds: when certain demographic patterns are found, a corresponding family structure is expected. But in three regions of Japan, studies do not reveal what we would expect. In the northeastern or Tohoku region where most families were stem families1 and a majority were multigenerational, the number of persons per family tended to be large and the age at marriage for women low, but the number of births was also low and population in the region did not increase.2 What explains this paradox? Ordinarily, the lower the age at marriage for women, the more children they have, and one would expect the population to increase. However, this region had a problem maintaining the population level due to its low birth rate. A second paradox comes from Ogaki domain in the central region where the age at marriage for women was high, but the fertility rate in villages was not as low as one would expect from the age at marriage. Yet despite the high fertility rate, population did not increase in many parts of this region. Third, an ongoing study of southwestern Kyushu indicates another paradoxical pattern that differs from those found in the northeastern and central regions. In the southwest, although the age at marriage for women was high, marital fertility was relatively high and the population increased. There were also numerous premarital births.3 From the accumulation of studies made over the past several decades plus the more recent analyses, we not only find these paradoxes, but we now have sufficient data from various regions in Japan to begin to draw some conclusions about population and family in the pre-industrial period, defined here as dating from the Tokugawa period when demographic data first became available until

280

12:27:16:11:09 Page 280 Page 281

REGIONAL DIVERSITY IN DEMOGRAPHIC AND FAMILY PATTERNS

the mid-Meiji period when modernization, industrialization, and urbanization began to transform society. There is a voluminous number of scholarly works on the regional characteristics of family structure in Japan.4 While these works have made significant contributions, none is an empirical study made using either an extensive amount of historical sources from the Tokugawa period or the statistical evidence that became available in the Meiji period. Furthermore, the northeast versus southwest dichotomy implicit in the studies made to date on Japanese family structure may require further refinement.5 Based on recent demographic studies, we have learned there were at least three diverse demographic patterns and patterns of family structure in pre-industrial Japan. Speaking broadly, there is no doubt that these three patterns noted above existed until the late Meiji period when the pace of modernization, industrialization, and urbanization visibly quickened and civil codes based on European legal concepts were adopted in 1898. This article surveys these patterns, analyzing the demographic data vis-à-vis the distinctive characteristics of family structure in three regions to show how and speculate why demographic patterns and family structure differed by region. The essay starts with a discussion of approaches to family structure, followed by an introduction of historical sources and the findings of regional diversity at the macro and micro levels. It then applies a simulation analysis in order to contrast regional differences as well as to see the relationship between demographic patterns and family structure.6 Finally, we attempt to explain the causes of regional diversity. Our aim is not only to fill the lacunae left by the earlier studies but also to offer an overview of the demographic and family patterns of the Tokugawa and early Meiji periods, and by doing so to sketch a new image of pre-industrial Japanese society.

PATTERNS OF FAMILY STRUCTURE: CULTURAL AND STRUCTURAL PERSPECTIVES Before discussing family structure based on empirical evidence, let us describe the following two analytic perspectives that can be adopted in researching family structure. The first views family structure as fundamental in defining the character of a society, and thus it remains unchanged over a long period. Apart from what is readily apparent on the surface, there exist substantive characteristics of family structure that stay unaltered despite changes in the environment or any external shocks. For example, those who adhere to this perspective argue that England is a society of nuclear families; they are observing a long-established historical characteristic and not an outcome of urbanization and industrialization. From this perspective, Japan is a stem family society, albeit different from that found on the European continent, and the “nuclearization” of the family structure is only a superficial and transient development that has the potential of being reversed to cause Japan’s family structure to revert back to its fundamental character: a society of stem families. This is the perspective that views family structure as a manifestation of “culture” and can be called a cultural perspective. The second perspective holds that family structure is not constant; it changes due to changes in the environment, such as urbanization, changes in economic

281

12:27:16:11:09 Page 281 Page 282

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

conditions caused by industrialization, or for other reasons (e.g. promulgation of a new legal system, fewer children after the demographic transition, and late or no marriage). In this perspective, it is possible for a society to change from one based on joint families to one based on stem families and then to one consisting mostly of nuclear families, and it is equally possible for a society to follow a path that is the reverse of the one just described. This perspective does not hold family structure to be invariable but rather sufficiently malleable to change over time. Thus family structure is a manifestation of “civilization” and this can be called a structural perspective. What we have learned from historical evidence has led us to primarily adopt the second perspective. It is important to recognize that family structure is sufficiently adaptable to change in the socioeconomic environment, which differs by location within a society as well as by nation. However, the first perspective is not without merit. As we discuss in the latter part of this article, social norms specific to subregions were significant factors affecting family structure in pre-industrial Japan. What has to be emphasized here in taking either perspective is that discussing the demographic and family patterns of Japan as a whole is not justified. At a minimum, it is necessary to analyze the patterns by region and by period. This necessity is clearly demonstrated, for example, in Hayami’s study of the Suwa district of Shinano Province, which was conducted using a large number of shu¯ mon-aratame-cho¯ (SAC) from the early Tokugawa period.7 In the 1670s, the decade of the first extant SACs, this district had a fair number of joint families made up of stem families and nuclear families and one cannot conclude from the data that the district had a single predominant family structure. However, the situation changed during the seventeenth century: joint families steadily meta- morphosed into stem families and nuclear families. This occurred as families in the district increasingly produced for the market because stem and nuclear families were found to be best suited to responding to changing economic realities.8 Prior to the Tokugawa period, the lack of urban population resulted in a non- dynamic rural society not oriented toward producing for the market. But just after the establishment of the Tokugawa bakufu, castle towns and post towns were created: in order to meet growing urban demand, rural residents had to increase the efficiency of agriculture. The older type of cultivation dependent on subordinate labor used on large-scale farms rapidly diminished. Instead, small-scale agriculture based on family labor—independent small farmers—became common all over Japan. During the seventeenth century, a large number of unmarried men and women who had occupied serf-like, near-indigent positions within the joint family structure became servants on short-term contracts or married and maintained households of their own. This was the origin of the kind of peasant society that characterized the Tokugawa period and formed the societal basis of Japan’s agricultural villages. The rise of peasant society meant an increasing marriage rate, a higher birth rate, and a growing population. According to one estimate, the total population of the Suwa district at the end of the seventeenth century was 2.5 times that at the beginning of the century.9 It is clear that both the demographic pattern and the basic family structure of the Suwa district changed during the Tokugawa period in response to changing socioeconomic conditions. It is possible to say that the most suitable “family

282

12:27:16:11:09 Page 282 Page 283

REGIONAL DIVERSITY IN DEMOGRAPHIC AND FAMILY PATTERNS

structure” was adopted in the course of change in the socioeconomic environment. The following analysis is conﬁned to observations from the eighteenth and nineteenth centuries. Suwa, at this point, is one of the few areas where we can observe longitudinal changes starting from the seventeenth century. Although the time span of the observations is limited, we adopt the structural perspective to show spatial dynamics.

DEMOGRAPHIC AND FAMILY PATTERNS BY REGION Sources at the macro level for studies of pre-industrial Japan are limited. National- level bakufu surveys provide provincial totals for only 12 years between 1721 and 1846. Thus we must make use of the national-level aggregated statistics, some of which fortunately are usable population statistics collected by the Meiji government, which begin in the 1880s.10 At the micro level, there are so few detailed sources for towns and villages from the Meiji period that the small number that do exist are of little value. What makes possible the micro-level analysis of population and family is the compilation of a sufficient number of village and other local studies. The EurAsia Project on Population and Family History (EAP)11 carried out under Hayami Akira’s direction has collected shu¯ mon-aratame-cho¯ and ninbetsu-aratame-cho¯ (NAC) from several hundred towns and villages across Japan, and a significant number—about 25—have been coded and incorporated into the project’s database.12 The regional distribution of the data gathered to date, however, is uneven in that we have no data from northern parts of the Tohoku region or from Shikoku, and the data we have for the Tokai, San’in, and northern Kyushu regions are scarce except for Nobi in the Tokai region. Because this study draws on recent works that have dealt primarily with three regions of Japan—the northeast, central, and southwest—where the necessary historical sources are available, the discussion here too follows this tripartite division. However, the data are not regionwide but samples from within the larger region. Thus, when we refer to the Tohoku or northeastern region, we really mean data from Nihonmatsu domain of Mutsu Province and the mountainous areas of Aizu (both of which are in today’s Fukushima Prefecture) for which a relatively large amount of micro data sources have been gathered and processed. Similarly, the data for the central region come from the Mino area (in today’s Gifu Pre- fecture). And, in the southwestern region, we have data from coastal villages facing the East China Sea (in Nagasaki and Kumamoto Prefectures) for which we have just begun to collect and code micro historical sources. Thus, our knowledge of the southwest comes primarily from studies of Nomo village in Hizen Province near Nagasaki and Takahama village in the district of Amakusa in Higo Province. Here we should insert the following caveat. For those wishing to analyze demographic patterns and family structure in the Tokugawa period, the SAC and NAC, compiled by town or village, provide excellent information. But, while they provide detailed information for the specific towns and villages they cover, we have to assume that the observed patterns represent regions as a whole. In short, we need to make use of macro and micro sources that differ significantly in nature.

283

12:27:16:11:09 Page 283 Page 284

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

Macro Studies The extant bakufu surveys from 1721–1846 provide a picture of regional diversity in pre-industrial Japan. While the Tokugawa population was once called “stagnant,” there were major regional variations within the national figure: until the beginning of the nineteenth century, the population of the northeastern region was on the decline while that of the central region stagnated and the southwest increased.13 We need to rely on micro-level evidence to find how these diverse population trends emerged. To understand the demographic and family patterns associated with these population trends, we examine the demographic snapshots gathered by the Meiji government in 1886.14 We confine our discussion here to three maps created from the Meiji data which suggest prominent demographic and family patterns in the region; they are based on the first nationwide statistics on average family size, the number of couples per household, and the estimated average age at marriage by prefecture. We hope to demonstrate some regional patterns that coincided with the above population trends during the late Tokugawa to mid-Meiji period. Map 15.1 presents the average number of persons per household. The average was clearly highest in the northeastern region and lowest in the central region. In the southwestern region (coastal region facing the East China Sea), the average was a little higher than in the central region. Household size varied for a number of reasons, including the existence of extended (vertical or horizontal) family members and the number of children or live-in servants. As an attempt to show the variation in family structures, Map 15.2 presents the number of couples per household. This number was distinctively higher in the northeast than in the other two regions, suggesting that, in the northeast, the major family structure was stem families consisting of parents and a married offspring and spouse. In contrast, in the central region, especially in the Kinki area, the average was less than one, indicating that a large proportion of families were nuclear families with one couple per household. The average of the southwestern region fell in between those of the northeast and central regions. Maps 15.3a and 15.3b present the estimated median age at marriage which was assumed to be the age when the ratio of married persons exceeded 50 per cent of the total of “married and unmarried” persons per prefecture found in the 1886 compilation.15 Although estimating the average age at marriage in this manner is less than ideal, we believe the estimated averages can be fruitfully used in comparing the averages of age at marriage within Japan. As Hayami has written elsewhere,16 these maps show unmistakably that, excluding cities, age at marriage was low in the eastern region and high in the western (central and southwestern) region. The line dividing the two regions runs through Shizuoka, Nagano, and Toyama and tends to follow the geological Fossa Magna—a fault line that crosses Japan from Toyama Prefecture on the Japan Sea to Shizuoka Prefecture on the Pacific.17 The Fossa Magna may serve as the line dividing these two marriage patterns. Although not shown here, the percentage of female household heads also followed a similar distribution: higher in the western regions.18 Taken altogether, this suggests a higher status of women in the central and southwestern regions than in the east. Derivation of any conclusive regional patterns would require far more precise

284

12:27:16:11:09 Page 284 Page 285

REGIONAL DIVERSITY IN DEMOGRAPHIC AND FAMILY PATTERNS

Map 15.1 Number of Persons per Household, 1886

data on the macro level. For example, while the distribution of age at marriage indicated two major patterns, further speciﬁcations may become necessary when we measure more precisely the timing of marriage, or when we consider the system of marriage as a whole including divorce and remarriage.19 Because the macro data do not provide such precision, this is where we turn to the micro studies. Our intention here is not to draw any clear boundaries for the three regions but to investigate the details of the relationships among them.

Micro Studies The Northeast. The most notable demographic fact of the samples from this region was an extremely low age at marriage. In two villages in Nihonmatsu domain, the average age at first marriage for men was around 19 and for women around 13 years of age during the first half of the eighteenth century. The age at marriage rose for women to around 17 as the Tokugawa period neared its end.20 From the 1680s to the Meiji Restoration (1868), the average age at first marriage

285

12:27:16:11:09 Page 285 Page 286

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

Map 15.2 Number of Couples per Household, 1886

in four mountainous villages in Aizu was 22.3 for men and 17.1 for women.21 They did not marry as early as in Nihonmatsu domain, but the age at ﬁrst marriage was still quite low. Despite such early marriages, the number of births per couple was extremely low. In Nihonmatsu domain, the average number of children per “completed family”22 was fewer than 4 and many couples had fewer than 3. The mountainous villages in Aizu also had a similarly low number of births.23 The number of births was too low to assure the next generation, and sons were often adopted or sons-in- law were recruited to maintain households over time.24 Thus the stem family structure was maintained in this region, among landholders in particular.25 The region maintained its population through a large inﬂow of people, most of whom came in as spouses and servants from regions where population was increasing, such as Echigo Province which lay across the O¯¯ u mountain range from the Aizu villages.26 For example, in Niita, 25.4 per cent of the male immigrants and 12.7 per cent of the women came from neighboring Echigo Province for adoption, marriage, and other reasons.27

286

12:27:16:11:09 Page 286 Page 287

REGIONAL DIVERSITY IN DEMOGRAPHIC AND FAMILY PATTERNS

Map 15.3a Estimated Median Age at Marriage: Males 1886

We also discovered that women in this region stopped having children at a relatively young age. The average age at last observed birth for women in completed families in Shimomoriya and Niita was around 33 years.28 This is considerably younger than the ages at which women ceased childbearing in the central region discussed below. Yet another distinctive practice was that after marriage the people of this northeastern region went on service (dekasegi), that is, they worked away from home for long periods.29 For example, in Niita, the majority of those who went to work outside the village began to work at an average age of 27.4 for males and 29.0 for females during the 1720–50 and 1826–50 periods. Nagata found that the majority (90 per cent) of emigrants from Niita and Shimomoriya later returned to the villages in stark contrast to Nishijo¯ (one half of men and 15 per cent of women), a village in the central region discussed below.30 She suggests that marriage was a way to keep men and women bound to households and villages. Thus, migration

287

12:27:16:11:09 Page 287 Page 288

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

Map 15.3b Estimated Median Age at Marriage: Female 1886

for service (often indentured service) of either husband or wife, or both, probably lessened the period of exposure to the risk of pregnancy and thus prolonged the interval between births, and, in turn, reduced the overall number of births. However, postmarital service does not seem to have been the sole reason for the long interval between births and low birth rate in the northeast. Noriko O. Tsuya and Satomi Kurosu found a clear sign of sex-selective infanticide based on the sex ratios of births by sex composition of surviving children.31 For example, when couples had no surviving children, their next child was much more likely to be a girl, but when couples without a son but with two or more daughters had another child, they were twice as likely to have a boy as a girl. Although we cannot generalize this ﬁnding to other villages, their study clearly suggests family planning among couples in the region. The Central Region. Data from six villages in Mino Province show that average age at ﬁrst marriage during the late eighteenth century and the mid-nineteenth

288

12:27:16:11:09 Page 288 Page 289

REGIONAL DIVERSITY IN DEMOGRAPHIC AND FAMILY PATTERNS

was as high as 28.0 for males and 20.5 for females.32 While this was considerably higher than in the northeast, the number of children per couple was much larger than that in the northeast. For instance, in the villages of Kandoshinden and Nishijo¯, the total marital fertility rate was in the range of 5.9 to 6.5.33 Women in completed first marriage had births recorded, on average, every three to three and a half years.34 Not only was the fertility level higher than in the northeastern region, there was no sex-selective fertility control either in these sample villages. It should be noted, however, that some studies based on other villages of the same region found evidence of family planning in the gender distribution of children.35 Hayami maintains that the sex preferences found in these studies may be an artifact of the small sample size. But some cases from our six villages in Mino Province did show a relatively low age for women at the birth of their last child. These cases suggest an early end to childbearing and thus hint at the possibility of the preference for a particular gender distribution of children. Still, the largest proportion of women bearing their last child (complete first marriage) was between ages 41 and 45.36 Further, in this region, in sharp contrast to the northeast, virtually all of those who went on service did so prior to marriage. In the case of Nishijo¯, for which particularly good SACs have been preserved for over 97 years (1773–1869), 50 per cent of males and 62 per cent of females (of those who survived beyond age 11) had service experience either in nearby villages or in cities. The average age when service began was 14 for males and 13 for females, and for both males and females service lasted an average of 13 to 14 years. Service was undertaken mostly in cities (70 per cent of male service and 62 per cent of female service), and when we calculate person-years of service in only three distant large cities— Kyoto, Osaka, and Nagoya—we find that the total for males was 1,535 person- years (48.4 per cent of the total number of person-years of all male service) and 1,385 person-years for females (40.0 per cent). More than one-third of the people who went to these largest cities from Nishijo¯ and other nearby villages did not return to their villages but spent the rest of their lives in the cities.37 The data here indicate that in this region, service did not mean what the word usually implies—going to work elsewhere for a period of time—but rather going to cities to work on a permanent basis. This strongly suggests that migrants enabled the pre-industrial cities of Japan to maintain their population levels despite the cities’ low birth and high death rates, the oft-noted “urban graveyard effect.”38 We must, of course, be cautious in accepting this conclusion because much more study is needed to analyze both birth and death rates in cities rather than uncritically accepting the “urban graveyard effect.” We have to learn why the rural population migrated to cities to become poorly paid servants and who exactly were the large majority—the lower economic strata—of city dwellers, that is, what their occupations were, how large a proportion of them were married, and whether they migrated within a city as well as from city to city. We can at present make two important observations regarding service. First, the service discussed in this section had direct and indirect consequences. The population in the rural areas was directly prevented from rising because only slightly more than half of those who went to cities returned to their villages. The indirect consequence is that the number of births in villages fell below the level

289

12:27:16:11:09 Page 289 Page 290

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

necessary to maintain the population because those women who did return married later, at age 25 or older, that is, after their peak period of fecundity. Second, the proportion of people who experienced service was low in the upper stratum and high in the lower stratum of society. In the case of Nishijo¯, the average age at marriage for all females was 24. However, the average age at marriage for women who had service experience was 25.9, in contrast to 21.5 for those who did not. It is thus evident that the age at marriage differed by socioeconomic stratum because while only a small number of females of the upper (landowner) stratum had service experience, a large number from the lower (tenant) stratum had this experience.39 As a result, many families in the lower stratum were unable to have enough children to avert termination of the lineage (zekke). In contrast, the upper stratum, of which only a small proportion were sent to service, married at an earlier age and thus tended to have more children than were needed to maintain the family line. These “excess children” were adopted by others, migrated elsewhere, or, most frequently, started a branch family (bunke). These new households invariably joined the ranks of the lower stratum, in effect taking the places of those families that had suffered zekke.40 Thus the basic family structure of this region was the stem family, but household size was small with a relatively high frequency of nuclear families in the family cycle. The Southwest. Because study of the southwestern region began only in the 1990s, we are able to draw few definitive conclusions about it. Nevertheless, there is little doubt that this region had distinctive demographic characteristics: both high birth rates and high rates of population increase. The bakufu’s national census shows that Hokuriku, San’in, San’yo, and Shikoku had the highest rates of population increase in Japan. The total population in these areas rose by 50 per cent between 1721 and 1846 at an annual average rate of 0.4 per cent, a rate that is high for a pre-industrial society.41 This increase was the result of natural increase (the difference between the birth and death rates) and was not due to substantial immigration. The southwest, unlike the central region, had no large cities that could significantly affect the rate of change in the total population of the rural areas. In the Kinki area, 32.7 per cent of the population was urban,42 but in the region of San’in and San’yo and the southwest, the urban population did not exceed 10 per cent in any area. In sharp contrast to the central region, in the southwest only a very small number of persons went to cities on service. As a result, the southwest was only minimally subjected to the “urban graveyard effect.” As in the central region, age at marriage in the southwestern region was high. However, so too were the marital fertility rate and the number of premarital births. Divorce and remarriage were common, and births between marriages were far from rare. Among women born in Nomo between 1802 and 1821, Tsuya found that the average age at first marriage was around 25; pregnant brides and premarital (or out-of-wedlock) childbearing were common, and marital fertility was high with little indication of deliberate fertility control.43 Also, women whose first marriage was completed had on average 4.6 children. The average number of children ever born to Nomo women was thus similar to or even higher than the average number for central Japan. Late marriage, high fertility, and more liberal attitudes toward premarital sexual activities in Nomo imply that marriage meant

290

12:27:16:11:09 Page 290 Page 291

REGIONAL DIVERSITY IN DEMOGRAPHIC AND FAMILY PATTERNS

neither a safety valve for population losses nor economic adjustment by the household.44 It appears that this region had a social norm that differed from the one prevailing in the northeast and central regions. Table 15.1 summarizes empirical findings relating to inheritance and succession, family structure, and demographic patterns. Note that the expressions used—low, high, small, and large—only indicate age at marriage and number of births in one region relative to other regions. Since the purpose of this table is to give a general idea about the regional differentials of demographic and family patterns for the Tokugawa to mid-Meiji periods, class differentials and trends within regions are not included.

SIMULATION OF DEMOGRAPHIC AND FAMILY PATTERNS BY REGION We now perform a simple simulation for each region in order to contrast regional diﬀerences and to derive the relationship between demographic patterns and family structure. That is, we trace how these patterns changed over time for a “representative” family in each region.

Assumptions and Method In order to model a representative family, parameters are chosen to be as close as possible to the empirical observations discussed above. The assumptions that go into the simulation are as follows. 1. Succession by the eldest (surviving) son was practiced. Inheritance and succession by the youngest child of either sex and temporary succession by females did occur, but infrequently. 2. All married persons lived to age 70 and stayed married. This may appear as a strong assumption. However, death at age 70 was not uncommon among those who lived to maturity.45 Also, the frequency and timing (i.e. quick) of remarriage that followed marriage dissolution46 meant a majority maintained the status of “being married.” For the purpose of this simulation, we believe this assumption suffices. 3. The sex of children born alternated between male and female. This is a strong assumption but does not deviate too far from what is seen for a large sample of the population. We assumed the sex of the first-born child alternated by generation in the same family and also that the sex of each child born to a couple alternated. Inclusion of sex selection would not alter the results of this simulation. 4. All households consisted only of family members and had no servants. Since most coresiding servants were usually concentrated in the tiny proportion of households with the largest landholdings, this assumption holds true for the majority of households. 5. We made specific assumptions for each region for age at marriage, wife’s fertility history, number of surviving children, and the timing of departure of nonheir children. 5–1. In the northeast, males married at age 20 and females at age 17, and beginning in the second year of marriage the wife had a child every three years for a total of four children. The wife had her last child at age 27. Of the four children,

291

12:27:16:11:09 Page 291 Page 292 Family and Demographic in Pre-industrial Patterns Family Japan Table 15.1 Table Northeast Central Southwest rst child Low High Middle rst marriage Low High High ﬁ ﬁ Number of births birthAge at of last childChildren born out of wedlock of womenStatus FewService experience LowTiming of service FewUrbanization limitationFamily trendPopulation Low Less After marriage Few High High Low Many Decrease Before marriage High More Stable Low High Many Before marriage High Many High Less Increase Low Low Major family patternInheritance patternSuccession pattern household sizeAverage age at Average Stem family birthAge at of Primogeniture Large Eldest son (eldest child) Eldest son (last child) Stem and nuclear families Primogeniture/nonegalitarian Primogeniture/egalitarian joint families nuclear, Stem, Eldest son (last child) Small Large

12:27:16:11:09 Page 292 Page 293

REGIONAL DIVERSITY IN DEMOGRAPHIC AND FAMILY PATTERNS

one died at age 1 and a second at age 5; only two children survived to adulthood. Of the two children, one remained in the household while another left for marriage at the age following the assumption. 5–2. For the central region, the assumptions were that males married at age 27 and females at 22, the wife had six children beginning in the second year of marriage with one born every two years, and the last child was born when the wife was 33 years old. Of the six children, one died at age 1 and another at age 5. Of the remaining four children, two remained in the household until age 16 when they went elsewhere, one succeeded his father, and the last became a successor in his or her spouse’s family. The assumptions relating to children reflect the fact that service (in effect, moving to cities) was common in this area. 5–3. For the southwest, the assumptions regarding age at marriage were the same as for the central region, and the wife had a child every two years beginning in the second year of marriage, bearing a total of six children within marriage. In contrast to the central region, however, this region did not have nearby cities that offered employment opportunities. Also, premarital conception was common. Thus, we assumed that all females had a premarital child at age 20 and bore a total of seven children, their last child at age 33. Two of the children died at age 1 and one at age 5. One male child who did not succeed his father lived with his parents until age 35, but at age 30 he married a woman aged 25 who had four children: one before her marriage and three born between the ages of 26 and 30. One female child lived with her parents until age 24 and had a premarital child at age 20. The simulation made under these assumptions traces how family compositions change over time. To be more specific, we generate hypothetical lifecourses of couples following the assumptions precisely and repeatedly for several generations, and observe the change and development of the household structure. We identify the beginning of the family cycle in the year after the death of the wife (both husband and wife die at age 70; see assumption 2) and mark it as the initial point (one for each of the three regions) of observation. That is, the observation of the changes in family structure starts 55 years after the year of marriage of the first couple in the northeast, and 50 years after that in both the central and southwestern regions. We follow the family cycle for 100 years from the respective initial points of the three regions.

Results We contrasted the households in the family cycles that developed among these three hypothetical scenarios and found that at least ﬁve characteristics emerge from this simulation. We describe these and then discuss some salient ﬁgures. 1. Number of couples per household: In the northeast, two married couples per household was the most frequently observed pattern. Over the 50 years of the family cycle, three couples per household were observed in only three years, and one couple per household also in only three years. For the central region, the dominant pattern was one couple per household during 28 years and two per household in the remaining 22. The southwestern region yielded the most diverse patterns: three couples for 6 years, two couples for 19 years, and one couple for 25 years. When we observe three couples in one household in this region, they consisted of a couple and two married children, that is, a joint family.

293

12:27:16:11:09 Page 293 Page 294

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

2. Household size: The size of households in the northeastern region ranged from four to seven persons and the mode—five persons—accounted for 42 per cent of total households; thus, the distribution was statistically normal. In contrast, in the central region, household size shifted more widely from three to eight persons, even though a household size of eight persons was observed in only four years. Also, in contrast to the northeast, the change in the mean household size occurred in a readily identifiable pattern of rising (three to eight) and falling (eight to three), alternating in approximately 16–17 years. In the southwest, the amplitudes in household size were largest, ranging from 3 to 14. The size was as high as 14 members because children who were not designated successors continued to live in the household, married, and had children of their own. If this exceptional situation is eliminated, the pattern of change resembles that seen for the central region. 3. Family form: In the northeast, the most prevalent form was the three- generation household which was found in 45 years, in contrast to four-generation households seen in three years and two-generation households observed in two years. In central Japan, two-generation households were seen in 24 years and three-generation households in 26. In the southwest, three-generation households were seen in 32 years and two-generation households in 18. 4. Number of persons of working age: In each region, the number of working- age persons (aged 16 to 60) in each household varied within a narrow range of two and five. 5. Number of dependent persons: In the northeast, the number of dependent persons (under 16 or over 60) within a household ranged between four and zero (zero, however, was found in only two years out of 50 years of observation). In the central region, the range was wider—from zero to six—and zero was seen for 12 years. In contrast to the northeast, the amplitude of fluctuation is wider and had discernible trends, alternating between rising and declining with each trend lasting about 30 years. In the southwest, the amplitude of fluctuation was even wider—from zero to ten—also with observable, alternating rising and declining trends, with each trend also lasting about 30 years. 6. Ratio of working persons to all persons: Figure 15.1 presents how the ratio of working persons to the total number of persons in the household changed over time in each of the three regions. In the northeast, the ratio of working persons to the total number of persons in each household (the ratio of items 4/2) fluctuated between 33.3 and 100 per cent. But the standard deviation of the ratio was relatively small (13.5) because we observed less than 50 per cent in only nine years and 100 per cent in only two years. In contrast, in the central region, this ratio fluctuated between 25.0 per cent and 100 per cent. The standard deviation of the ratio was large (25.3) because a ratio of 100 per cent was seen in 12 years while 50 per cent or less was found in 18 years. In this region, the pace of change was very rapid: it took only seven years for the ratio to fall from a period of 100 per cent to 25 per cent, and it rose again to 100 per cent in 12 years. The pattern was similar in the southwest which also had a high standard deviation for the ratio (25.0). The lowest ratio seen was 28.6 per cent, and it took only ten years to fall to this low ratio from 100 per cent. Thus, the results of the simulation are consistent with what we observed in the

294

12:27:16:11:09 Page 294 Page 295

REGIONAL DIVERSITY IN DEMOGRAPHIC AND FAMILY PATTERNS

Figure 15.1 A Simulation of the Proportion of Productive Population per Household in Three Areas

macro and micro studies. In the northeastern region, there were many multigeneration households and virtually no two-generation households. Household size was stable and the ratio of working persons to household size also changed little. In the central region, however, two- and three-generation households were equally prevalent. Both the size of household and the ratio of working persons to total number of persons in households fluctuated widely. In the southwest, the number of years in which three-generation households were most prevalent was almost twice as large as the number of years when two-generation households were dominant (64 per cent versus 36 per cent). It is especially noteworthy that in the northeast, the ratio of working persons to total number of persons in the household remained stable—never falling below 33.3 per cent—and had a smaller standard deviation. Why was this so when in the other two regions this ratio fluctuated widely between 100 per cent and as low as 25 per cent? This ratio, we believe, is crucial in understanding the economic life of households because when more working persons in a household were gainfully employed, the household enjoyed a large income, while a household with fewer working persons had an increased risk of economic crisis. This being the case, why do we observe such a significant interregional variation in this ratio that is so important in determining the economic well-being of households?

CAUSES OF REGIONAL DIVERSITY By synthesizing the empirical ﬁndings at the macro and micro levels, we hypoth- esize that the level of necessity in keeping working-age persons in the household varied with diﬀerences in the natural and socioeconomic environments. The natural conditions of the northeastern region are wretched. Winters are very cold,

295

12:27:16:11:09 Page 295 Page 296

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

and people are often snow-bound and cannot work outdoors. Thus, the presence of only a few working persons in a household could not assure an income above subsistence level. Consciously or not, the people in this region behaved in ways to avert economic crisis by maintaining the highest possible ratio of working-age persons to the total number of people in the household. Marriage, rather than temporary service, appears to have been used to assure the recruitment of working-age population. Women who married early and entered another household had all their children early and became part of the labor force in their husbands’ households. Going on service while married both brought in income and reduced the number of births. It is not hard to suspect that peasants limited births deliberately to reduce the number of economically dependent persons in the household, thus raising the ratio of working persons in a household. However, early marriage increased the number of multigeneration households, thus making the northeastern region a prototypical stem family society. Already shorter than in other regions, the intergenerational age difference was made even shorter because of a practice in this region known as ane-katoku (succession by the head’s eldest child regardless of sex).47 It appears that those living in this region had few difficulties in maintaining multigenerational households. On the other hand, in the central region (except the areas facing the Japan Sea), it was possible to engage in agricultural work even in winter. The region was more urbanized and offered more employment opportunities. Thus many people went on service before marrying, married relatively late, and had no reason to limit births. The interval between generations grew longer over time, and most households had only one couple even though households consisting of nuclear families from two generations were sometimes observed. This did not mean that this region was dominated by nuclear families but rather that nuclear families were a transitory phenomenon in the family cycle. This also meant the people of this region lived in nuclear families at some point during their lives. In the southwestern region, natural conditions were even better than in the central region, but urbanization lagged behind that in the central region. This meant the southwest lacked the “safety valve” of the “urban graveyard effect”; that is, there were few cities with their high mortality rates to absorb population from the surrounding rural areas. Population in this region continued to increase steadily because the unmarried had no place to go and remained in their parents’ homes. The size of households in the region thus became larger. As noted earlier, the proportion of the working-age population within a household is an important economic indicator. It seems safe to posit that in the northeastern region there existed a social mechanism that minimized the variation in this proportion. An important question then is: how did the people of the central and southwestern regions cope with wide and rapid fluctuations in this proportion, i.e. with the economic difficulties caused by such fluctuations? One possible answer is frequent changes in ownership of land (via sales and pawning), and institutions of mutual financial assistance (e.g. tanomoshi-ko¯ 48 and miyaza) maintained by farming and fishing villages in these two regions. Indeed, we find a large number of documents on land transactions and pawning among the historical records of the Tokugawa period preserved in the villages of these regions.49 Given such evidence, it seems reasonable to assume that many

296

12:27:16:11:09 Page 296 Page 297

REGIONAL DIVERSITY IN DEMOGRAPHIC AND FAMILY PATTERNS

households in these regions managed to survive through the economic crises resulting from a decline in the proportion of working persons in their households by selling or pawning land. They then bought the land back or redeemed the pawned land when the proportion of working persons in the households returned to 100 per cent for a suﬃciently long period, that is, after everyone in the household worked for a number of years to reestablish the economic basis of the household. Because studies linking these economic activities with the observed changes in demographic patterns are yet to be undertaken, we present the preceding only as a hypothesis to be tested. Further studies are also needed to support the view that institutions such as tanomoshi-ko and miyaza centering in the Kinki area that had become prevalent by the sixteenth century were not merely local investment groups or religious institutions but also provided mutual ﬁnancial aid so that those members who faced economic hardships could weather the crises.

DIVERSITY OF NORMS Until now, we have attributed the regional diversity in demographic and family patterns to the diversity of socioeconomic environments—a structural perspective. For the final part of our discussion, we would like to consider the origin of regional patterns deriving from diversity of societal norms. Emmanuel Todd created a map of Europe consisting of 483 geographic units (the counterparts of Japan’s prefectures, that is, fu and ken).50 He discovered that the absolute nuclear family was found in regions that faced the North Sea where Anglo- Saxons, Jutes, and Frisians—for whom the absolute nuclear family was the social norm—came to settle. Similarly, he found that the egalitarian nuclear family was most frequently found in regions where Roman cultural legacies were especially prominent.51 Put simply, what Todd sought and found were factors determining the regional norm of family structure in ethnology as well as in ancient cultural and institutional legacies. This is highly instructive in pursuing further analyses of the differences in demographic patterns and family structure in Japan. The existence of varying norms of family structure can be explained by the diverse origins of Japanese society. Although it is widely accepted that Japan had an indigenous population and that other peoples came later from across the sea and intermingled with the indigenous population, there is no accepted theory as to when and whence these latecomers came to Japan. Our hypothesis, based on the existence of three distinct regional demographic and family patterns, is that the differences in these patterns are due to differences in the societal norms (cultures) of the people who came to live in each of the three regions. To be specific, in the northeast where the indigenous population was the hunting and gathering Jo¯mon culture,52 population density was crucial: it had to be sufficiently high to enable the people to maintain the society, but if it exceeded a specific “carrying capacity,” its very existence would be threatened. Thus, in time, this region evolved a culture in which early marriage (accompanied by control over births) and stabilizing the proportion of working-age population became embedded as a fundamental part of the societal norm. Even after the region adopted agriculture as the basis of its economy, the societal norm of early

297

12:27:16:11:09 Page 297 Page 298

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

marriage and limiting the number of dependent (i.e. nonproductive) family members continued as an essential means to eliminate or reduce the risks of threats to the continued existence of families from the unfavorable climate. In the central region, the Yayoi culture evolved because the indigenous population of this region was subjugated by people who came to this region from the Korean peninsula and China (around 300 ..), bringing agriculture and the technology for producing iron. Since the “carrying capacity” could be increased by cultivating more land (and uncultivated land was abundant for a long period), there was little need to limit population (or, if the need existed, it was much less acute than in the northeast). Furthermore, this was the region where the Yamato court, which unified the country, was based and where ancient Chinese institutions—the ritsuryo¯ system—had the strongest impact. The effects of the ritsuryo¯ system spread across the nation, and the region-specific social and institutional characteristics of the northeastern and southwestern regions grew weaker over time. The distinctive demographic and family patterns observed for the southwest are, we believe, due to the fact that this region had a large inflow of Southeast Asians with their own social norms who moved northward (via Taiwan and the Ryukyu Islands) to settle in this region. (In Korea too, Cheju Island, where peoples of Southeast Asia settled, has a “culture” similar to that found in the southwestern region of Japan and different from that of the peninsula.53) We speculate that there may have been a “culture” of the East China Sea coastal regions just as there was a “culture” of the North Sea coastal regions in Western Europe. The distinctive characteristics of the “culture” of the southwestern region spread elsewhere by means of waterborne transportation, a much more effective means of moving both persons and goods rapidly and economically than premodern surface transportation. Thus those who arrived in the region moved eastward using waterborne transport and disseminated their culture to areas surrounding the Seto Inland Sea and the coastal areas of the Japan Sea and the Pacific Ocean. However, the current state of research does not tell us how far inland their cultural influence extended.

CONCLUSION In this essay, we have taken several important steps toward better understanding why demographic and family patterns differ by region. However, much remains to be done. We still rely on a very limited number of case studies in analyzing the interregional differences in the patterns. To be more certain of the interregional differences, as well as to explore the possibility of identifying more than three regional patterns, we need more case studies. For example, areas where population increased during the eighteenth century (Hokuriku, San’in, San’yo, and Shikoku) require further study. Also, simulation study can be improved with more complex assumptions (e.g. inclusion of mortality differentials, possibility of sex preferences) and techniques. We still have a number of puzzles, and to solve these we must be willing to make use of findings made by linguists, geologists, and other scholars. For example,

298

12:27:16:11:09 Page 298 Page 299

REGIONAL DIVERSITY IN DEMOGRAPHIC AND FAMILY PATTERNS

linguists have established that dialects differ east and west of the Fossa Magna, as did average age at marriage.54 (An example among the most commonly used words that differ east and west of Fossa Magna is aru [“it is”] in the east and oru in the west.)55 However, in analyzing the currently available studies to try to answer the questions posed at the beginning of this article, we realize we have to recognize not two but at least three distinguishable patterns: northeastern, central, and southwestern. And the reason for the paradoxes or anomalies can be explained on the basis of existing evidence by the diversity of the ethnological origins of the people who settled in the three regions combined with later adaptations to the physical environment and economic conditions. The original settlement occurred thousands of years ago and, in the intervening periods, people mixed with one another and the peculiar characteristics of each region were weakened. However, if we look at the Meiji statistics which reflect Japan prior to modern urbanization, industrialization, and the establishment of the modern civil code, we can see the obvious “scars” of the original diversity. Japan was by no means a society with one social pattern even in the Tokugawa or Meiji period.

NOTES

* This article draws on material from Akira Hayami, The Historical Demography of Pre-Modern Japan (Tokyo: University of Tokyo Press, 2001). We gratefully acknowledge the permission of the University of Tokyo Press. 1. A stem family is one in which there is no more than one married couple in each generation. 2. For informative studies of Nihonmatsu domain based on micro sources, see Narimatsu Saeko’s Kinsei To¯hoku no¯son no hitobito—O¯ shu, Asaka-gun, Shimomoriya-mura (Kyoto: Minerva Shobo¯, 1985) and Edo jidai no To¯hoku no¯son—Nihonmatsu han, Niita-mura (Tokyo: Do¯bunkan, 1992). For Aizu domain, the most useful studies include Okada Aoi, “Kinsei no¯min shakai ni okeru setai to kakei no keisho¯—Aizu sankanbu no shu¯ mon-aratame-cho¯ o chu¯ shin toshite,” EurAsia Project on Population and Family History Monograph Series No. 1 (1996); and Kawaguchi Hiroshi’s “Minamiyama okurairiryo¯ ni okeru sonraku jinko¯ no tokushoku—1758,” Tokyo Kasei Gakuin Tsukuba Tanki Daigaku Kiyo¯, Vol. 1 (1991), and “Minamiyama okurairiyo¯ ni okeru sonraku jinko¯ no tokushoku—1871,” Tokyo Kasei Gakuin Tsububa Tanki Daigaku Kiyo¯, Vol. 2 (1992); “19-seiki shoto¯ no Aizu: Minamiyama okurairiryo¯ ni okeru jinko¯ hendo¯ to shussei seigen,” Rekishi chirigaku, Vol. 40, No. 5 (1998). 3. These are based on data obtained from Hayami Akira, ed., Kokusei cho¯sa izen Nihon jinko¯ to¯kei shu¯ sei, 22 vols. (Tokyo: To¯yo¯ Shorin Publishing Co., 1992–93). Also, for a micro-level study, see Noriko O. Tsuya, “Patterns of Nuptiality and Fertility in Southwestern Tokugawa Japan: The Case of the Village of Nomo,” in T. J. Liu et al., eds., Asian Population History (Oxford: Oxford University Press, forthcoming). 4. For a bibliographic survey of these studies, see Shimizu Hiroaki, Jinko¯ to kazoku no shakaigaku (Tokyo: Saishobo¯, 1986), pp. 134–53. 5. For criticism of such studies, see two studies by Kumagai Fumie, “Nihon no kazoku to chiikisei— Higashi Nihon kazoku o chu¯ shin to shite” and “Nihon no kazoku to chiikisei—Nishi Nihon kazoku o chu¯ shin toshite,” found respectively in Vols. 1 and 2 of Kumagai Fumie, ed., Nihon no kazoku to chiikisei (Kyoto: Minerva Shobo¯, 1997). In contrast, Osamu Saito¯ maintains the stem-family system is the basis of the Japanese traditional family regardless of variation due to demographic conditions. He reached this conclusion after contrasting the regional diﬀerences of family structure based on the 1920 national census samples with those of Europe (“Two Kinds of Stem-Family System? Traditional Japan and Europe Compared,” Continuity and Change, Vol. 13, Part 1 [1998], pp. 167–86). 6. A similar attempt was made by Akira Hayami, The Historical Demography of Pre-Modern Japan (Tokyo: University of Tokyo Press, 2001), chap. 6; Akira Hayami and Emiko Ochiai, “Family Patterns and Demographic Factors in Pre-Industrial Japan,” in Liu et al., eds., Asian Population History. However, these studies dealt only with two regions. The approach and interpretation are further developed in this article.

299

12:27:16:11:09 Page 299 Page 300

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

7. Hayami Akira, Kinsei no¯son no rekishi jinko¯gaku-teki kenkyu¯ (Tokyo: To¯yo¯ Keizai Shinpo¯sha, 1973), Table 3–10, p. 95. 8. Ibid., p. 101. 9. Ibid., p. 24. 10. Hayami, ed., Kokusei cho¯sa izen Nihon jinko¯ to¯kei shu¯ sei. 11. The EurAsia Project was organized by Hayami Akira and sponsored by the Japanese Ministry of Education, Science, Sports and Culture between April 1995 and March 2000 for the purposes of (1) making international demographic comparisons of Japan, China, Italy, Belgium, and Sweden, all of which have continuously compiled population registers prior to 1900; and (2) studying the history of population and family within Japan through the population registers. 12. Shu¯ mon-aratame-cho¯ are population registers from religious investigations; ninbetsuaratame-cho¯ are secular population registers. The history of population registers, their reliability, their compilation, and computerization of the data sets are described in detail elsewhere: Akira Hayami, “Thank You Francisco Xavier: An Essay in the Use of Micro-data for Historical Demography of Tokugawa Japan,” Keio Economic Studies, Vol. XVI, Nos. 1–2 (1979), pp. 65–81; Laurel L. Cornell and Akira Hayami, “The Shumon Aratame Cho: Japan’s Population Registers,” Journal of Family History, Vol. 11, No. 4 (November 1986), pp. 311–28; Ono Yoshihiko, “Bunkakei no keisanki-riyo¯ II: de¯ta nyu¯ ryoku no yu¯ za¯ intafeisu (Rekishijinko¯gaku no baai)” Nihon kenkyu¯ , No. 8 (1993), pp. 165–82. 13. Hayami, The Historical Demography of Pre-Modern Japan, chap. 3. 14. Naimusho¯ Koseki Kyoku, ed., “Meiji 19-nen 1-gatsu 1-nichi shirabe Nihon Teikoku minseki koko¯hyo¯,” compiled in Hayami, ed., Kokusei cho¯sa izen Nihon jinko¯ to¯kei shu¯ sei, Vol. 2. The data used for the calculations are from “Dai 9 kakuchiho¯ seinen oyobi umuhaigu¯ hyo¯” included in the compilation. 15. The estimated median is used instead of the mean because of the limitations of available records; this is not a standard technique but is believed to give a reasonable proxy for the average age at marriage. For details, see Akira Hayami, “Another Fossa Magna: Proportion Marrying and Age at Marriage in Late Nineteenth Century Japan,” Journal of Family History, Vol. 12, Nos. 1–2 (1987), pp. 57–72. 16. Ibid. 17. The Fossa Magna boundary is located between the Sea of Okhotsk Plate and the Amurian Plate, dividing Japan clearly into two, east and west. See Map 1. Richard Bowring and Peter Kornicki, eds., Cambridge Encyclopedia of Japan (Cambridge: Cambridge University Press, 1993), p. 3. 18. It is interesting to note that data at the micro level confirm this finding. The proportion of women who became heads upon the death of the previous heads was twice as large in a central village as in northeastern villages. Aoi Okada and Satomi Kurosu, “Succession and the Death of the Household Head in Early Modern Japan: A Case from a Northeastern Village, 1720–1870,” Continuity and Change, Vol. 13, No. 1 (1998), pp. 143–66. 19. Kurosu, Tsuya, and Hamano contrasted the consequences of the first marriages in the four villages from the three regions. See Satomi Kurosu, Noriko O. Tsuya, and Kiyoshi Hamano, “Regional Differentials in the Patterns of First Marriage in the Latter Half of Tokugawa Japan,” Keio Economic Studies, Vol. XXXVI, No. 1 (1999). 20. Noriko O. Tsuya and Satomi Kurosu, “Economic and Household Factors of First Marriage in Early Modern Japan: Evidence from the Two Northeastern Villages, 1716–1870,” in Muriel Neven and Catherine Capron, eds., Family Structures, Demography and Population. A Comparison of Societies in Asia and Europe (Liège: Laboratoire de Démographie de l’Université de Liège, 2000), pp. 131–57. 21. Okada, “Kinsei no¯min shakai.” 22. Completed families refer to marriages lasting until the end of the childbearing age of wives. Here, the sample was comprised of women who married before age 30. 23. In Shimomoriya, the number of births by wives (more accurately, the number of births recorded) did not exceed four regardless of the woman’s age at marriage. Narimatsu, Kinsei To¯hoku no¯son no hitobito, p. 231. In Niita too, the number exceeded four only for women married at age 20 or below and only in the late Tokugawa period (1776–1859) (Narimatsu, Edo jidai no To¯hoku no¯son, p. 123). And in Aizu, the average number of births in completed families was around three. Kawaguchi Hiroshi, “17–19 seiki ni okeru Aizu Minamiyama okurairiryo¯ no jinko¯ hendo¯ to sono chiikiteki tokucho¯” (Ph.D. diss., Tsukuba Daigaku, 1993). 24. Satomi Kurosu, “Long Way to Headship, Short Way to Retirement: Adopted Sons in a Northeast- ern Village in Pre-Industrial Japan,” The History of the Family: An International Quarterly, Vol. 3, No. 4 (1998), pp. 393–410; Okada and Kurosu, “Succession and the Death of the Household Head.” 25. Okada Aoi, “Kinsei no¯min shakai ni okeru setaikosei no life cycle,” Shakaigakuhyo¯ron, Vol. 15, No. 1 (2000), pp. 136–52.

300

12:27:16:11:09 Page 300 Page 301

REGIONAL DIVERSITY IN DEMOGRAPHIC AND FAMILY PATTERNS

26. Along with the southwestern region, Echigo Province was another area where population continued to rise during the Tokugawa period. According to the bakufu survey, the total population increased by 30 per cent from 1721 to 1834 (932,000 to 1,225,000). 27. Narimatsu, Edo jidai no To¯hoku no¯son, pp. 78 and 98. 28. Here the sample included only first marriages. Noriko O. Tsuya and Satomi Kurosu, “Reproduc- tion and Family Building Strategies in 18th and 19th Century Rural Japan: Evidence from Two Northeastern Villages,” paper presented at the annual meeting of the Population Association of America, New York, 25–27 March 1999. 29. Service (dekasegi) migration was a temporary migration although it included labor migration as well. 30. Mary Louise Nagata, “Labor Migration, Family and Community in Early Modern Japan,” in Pamela Sharpe, ed., Women, Gender and Labor Migration: Historical and Global Perspectives (London: Routledge, 2001). 31. Tsuya and Kurosu, “Reproduction and Family Building.” 32. Hayami, Kinsei No¯bi chiho¯, p. 85. 33. The total marital fertility rate is the sum of age-specific birth rates for married women expressed as rates “per woman.” The nondemographer can think of these as average number of children per married woman. 34. Ibid., pp. 149–50, 222–23. 35. William G. Skinner, “Conjugal Power in Tokugawa Japanese Families: A Matter of Life or Death,” in Barbara Diane Miller, ed., Sex and Gender Hierarchies (New York: Cambridge University Press, 1993), pp. 236–70; Thomas C. Smith, Nakahara: Family Farming and Population in a Japanese Village, 1717–1830 (Stanford: Stanford University Press, 1977). 36. Hayami, Kinsei No¯bi chiho¯ no jinko¯, keizai, shakai (Tokyo: Sobuncha, 1992), pp. 220–23. 37. Ibid., pp. 268 and 273. Calculated from Tables 10–7 and 10–11. 38. This phenomenon was first discussed by Allan Sharlin in his “Natural Decrease in Early Modern Cities: A Reconsideration,” Past and Present, No. 79 (1981), pp. 126–38. In Japan, Saito¯ Osamu advanced the same view in Sho¯ka no sekai, uradana no sekai: Edo to O¯ saka no hikaku toshi-shi (Tokyo: Riburopoto, 1987). Also, Hayami’s study of Osaka’s Kikuya-c.ho¯ showed that, during the entire 110 years from the 1760s to the 1860s, the town’s crude death rate exceeded the crude birth rate and thus it could not have maintained its level of population without immigrants. Hayami Akira, “Kinsei ko¯ki O¯¯ saka kikuya-cho¯ no jinko¯ to nyu¯ yo¯ji-shibo¯,” Chiba Daigaku keizai kenkyu¯ , Vol. 13, No. 3 (1998). The situation was undoubtedly the same for other towns in Osaka, and thus we can assume that Osaka continued to have a large number of inmigrants from rural areas. 39. Hayami, The Historical Demography of Pre-Modern Japan, chap. 5; Kinsei No¯bi chiho¯, p. 230. 40. Hayami, Kinsei No¯bi chiho¯, pp. 303–7. 41. Hayami Akira, “Kinsei ko¯ki chiikibetsu jinko¯ hendo¯ to toshi jinko¯ hiritsu no kanren,” Tokugawa renseishi kenkyu¯ kiyo¯ (1975). 42. The percentage of people living in cities is the number of persons within an area who were living in cities with populations over 5,000 divided by the total population of the area. The data are taken from Kyo¯bu seihyo¯ compiled in 1875. 43. Tsuya, “Patterns of Nuptiality.” 44. Kurosu, Tsuya, and Hamano, “Regional Differentials in the Patterns of First Marriage in the Latter Half of Tokugawa Japan.” 45. Distribution of death rates by age in the northeast is in Narimatsu, Kinsei To¯hoku no¯son, p. 76 (the case of Shimomoriya in Mutsu Province); distribution for the central region is in Hayami, Kinsei No¯bi chiho¯, p. 242 (the case of Nishijo¯ in Mino Province). In both, the highest rates are clustered for those in their 70s. At present, no similar data can be found for the southwest. 46. Kurosu et al., “Regional Differentials.” 47. Maeda Takashi, Ane-katoku (Osaka: Kansai Daigaku Shuppanbu, 1976). 48. There were many types of tanomoshi-ko¯. However, the most prevalent was a mutual aid organization in which a loan taken out by one person was provided by all other members, and this pattern was repeated over the years. The principal intent of such a tanomoshi-ko¯ was to aid households encountering temporary financial predicaments. 49. For example, in his study of Nakahara, Thomas C. Smith, using village tax records, found that landownership changed very frequently. Based only on those sources containing pertinent information, he found that there were 415 instances of recorded changes in the amounts of land held by villagers from 1717 to 1830 (Smith, Nakahara). 50. Emmanuel Todd, L’invention de l’Europe (Paris: Ed. du Seuil, 1990). 51. Ibid., p. 47.

301

12:27:16:11:09 Page 301 Page 302

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

52. Hayami considers this indigenous ethnic group to have gone through experiences (e.g. being driven away or assimilated into the immigrants’ culture) similar to those of the Celts. (E. Todd and Hayami Akira, “Kazokuko¯zo¯ kara mita atarashii ‘Nippon’ zo¯” taidan, in KAN: History, Environment, Civilization, Vol. 4 [Tokyo: Fujiwara Shoten, 2001], pp. 28–48.) 53. On family structure, see an anthropological survey made during the latter half of the 1930s: Izumi Seiichi, Saishuto¯ (Tokyo: To¯kyo¯ Daigaku Shuppankai, 1966). See especially section 7 of chapter 3, pp. 131–40. 54. Kokuritsu Kokugo Kenkyu¯ sho, ed., Nihon gengo chizu, Vol. 2, Map 53 (Tokyo: Okurasho¯ Shuppankyoku, 1982). 55. See several maps included in Tokugawa Munemasa, Nihon no ho¯gen chizu (Tokyo: Chu¯ o¯ko¯ronsha, 1979).

302

12:27:16:11:09 Page 302 Page 303

Co-authored with Aoi Okada and ﬁrst published in The History of Family, vol. 10, pp. 195–229, 2005.

16 Population and Households Dynamics: A Mountainous District in Northern Japan in the Shu¯mon Aratame Cho¯ of Aizu, 1750–1850

1. OBJECTIVE AND HISTORICAL DATA: LAUNCHING A COMPARATIVE STEM-FAMILY PROJECT he objective of this study is to clarify the characteristics of the population Tand stem family households in the Aizu mountainous district located in northeastern Japan using as the historical source material, Shu¯ mon Aratame Cho¯ (religious faith investigation register, hereafter SAC). This article will also consider indices useful for further comparative research between the European stem family system – particularly in the French Pyrenees – and the Japanese family system. The stem family system is based on the principle of a single child independently inheriting the family estate. In Japan, the typical family system is called ie. This ie system is a subordinate concept of the stem family system. In our view the term ie connotes a more rigid and extreme system than the stem family system. The ie system emphasized the parent–child relationship and followed the rules of family persistence and single–child independent inheritance by the eldest son. The relationships of the compositional elements of the household and ideolo- gies that integrate these relationships are thought to establish family structure. Thus, in this article we attempt to clarify the relationships between elements of a family based on the composition of the household as well as household cycles. Through observation of heirship succession, we also attempt to clarify the ideals that integrate family structure. Accordingly, we will present information on the cycles of household composition and on the succession of household heads and heirs as analytical indices of the family system. We also hope to clarify the ways in which demographic factors aﬀected the thought process of these people. We use Shu¯ mon Aratame Cho¯ (SAC) from the Aizu mountainous district as our primary source (Hayami, 2001a, pp. 26–37). The Aizu mountainous district was chosen for the study because of its similar geographic conditions to the French Pyrenean region, with which we draw comparisons of stem family systems, and because research could be conducted there. For the Aizu mountainous district, historical materials are available that cover a comparatively long period of time for

303

12:27:16:11:09 Page 303 Page 304

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

several villages; all these materials are first rate and well suited to the analysis of populations and families. The exact area studied was located in contemporary western Fukushima Prefecture. The Aizu district is a series of steep mountains, with a flat plain in the north (see Kadokawa Dictionary of Japanese Place Names, 1991, p. 53). In the time period of the SAC surveys, the mountainous region was home to numerous settlements dotting the narrow strips of flatlands along the valleys with terraces clinging to the slopes of the mountains. Villages were found even in the most remote mountain areas, but generally the scale of each settlement was small. In many cases, mountains or deep valleys blocked access to them, requiring upland people to take a long way around when traveling even to the next village (Fig. 16.1). The present research was conducted using SACs from four communities in the Aizu upland district. Two of them, Kanaizawa mura (village or hamlet) and To¯nosu mura, were farming villages in relatively flat areas. The other two, Kuwanohara mura and Ishibushi mura, were located in completely mountainous terrain. By including both flatland and mountainous villages in this study, we hope to create a better picture of the Aizu district as a whole. Since each of these villages had its own individual set of characteristics, the area could hardly be considered homogeneous. To complicate matters further, the kokudaka (land assets) of each village was comparatively small in size. If we had taken a single village as the subject for quantitative analysis of population and families, we would have had difficulty determining whether the results of our observations were representative of the entire Aizu area or specific to the village under study. Therefore, we are considering the totals from these four villages, thus creating a

Figure 16.1 Map of Japan.

304

12:27:16:11:09 Page 304 Page 305

POPULATION AND HOUSEHOLD DYNAMICS

relatively large sample for inferring the characteristics of the mountainous region. Before studying family forms, we shall identify demographic trends for the entire Aizu district and compare these ﬁgures with the rest of Japan.

2. DEMOGRAPHY IN THE AIZU REGION 2.1. The Sources: Nationwide Population Survey By the Tokugawa Government The Tokugawa government began a nationwide population survey in 1721, and then conducted it once every 6 years between 1726 and 1846 (Hayami, 2001b, pp. 56–58). The year 1846 is now known to be the final year the survey was conducted. The daimyo (feudal lord) and local magistrates were required to report population counts for the areas under their jurisdiction. As such, the survey was conducted 22 times. For twelve of these surveys, population counts can be obtained for each province, and for nine of those twelve, the population counts are broken down by gender (Hayami, 1982). First, let us use these figures to examine the characteristics of population trends in the O¯¯ u region (the northernmost area of Honshu¯ ), which includes the Aizu Lord territory. The O¯¯ u region consisted of two provinces, Mutsu (east) and Dewa (west). During the first half of the period studied, the populations of both provinces decreased considerably. In particular, the population of Mutsu province decreased 20% between 1721 (the first year of the census) and 1786. Dewa Province also experienced a decrease of more than 8%. As will be discussed later, the population decrease in the O¯¯ u region began at the beginning of the 18th century. Particularly large decreases in population occurred later as a result of the Ho¯reki famine (1755), Tenmei famines (1783, 1785, 1786), and a series of bad harvests. There was a remarkable decrease in the population of Mutsu province. During this period, the poor harvests that led to the decrease were the result of no summers and poor sunlight. Another cause of famine was the continuous cold air streams that pushed down from the Sea of Ohotsuku and resulted in a great deal of damage along the Pacific coast. The population of the region bottomed out in 1786, followed by a gradual upturn. Another large decrease in population occurred during the latter half of the 1830s. This decrease was the result of poor harvests and an epidemic that struck the entire country in 1837 and 1838. The exact cause of the epidemic is unknown, although reports from that time indicate victims suffered from a digestive disease. During the period of study, the population of Mutsu Province fell to its lowest level in 1840, when it represented 77% of the population in 1721. The area had a rapid recovery over the next 6 years, however, and it appears that the region then entered a period of long-term population recovery. To summarize, the population for the O¯¯ u region declined from 1721 until the end of the 18th century due to a succession of poor harvests, but began to turn around at the beginning of the 19th century. There was a temporary decline resulting from disasters in the 1830s, but it could be said that overall, the area was experiencing an increase in population during the 19th century. The decrease in population for the O¯¯ u region in the 18th century was particular to this region as well as to northern Kanto¯, but not to the rest of Japan. The population in southwestern Japan actually increased during this time. We have

305

12:27:16:11:09 Page 305 Page 306

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

compared O¯¯ u with the Chu¯ goku region (located at the western end of Honshu¯ ), which in 1721 had a population that was close to it in size (O¯¯ u region population: 2.48 million, Chu¯ goku region population: 2.73 million), and the subsequent transitions in the populations of each region. With the exception of the time of the Tempo¯ disasters, the population of the Chu¯ goku region increased during all periods and the area was on the track of growth. During the 113 years from 1721 to 1834, the population of the Chu¯ goku region increased by 25%, an annual growth rate of 0.2%. For the same period, the population of the O¯¯ u region decreased by 16% during the first 65 years, an annual decrease of 0.2%. During the next 48 years until 1834, the population increased by 11% (an annual growth rate of approximately 0.2%). By the last year of the period, the population in the O¯¯ u region was 2.52 million compared to 3.31 million for the Chu¯ goku region, a difference of approximately 800,000 inhabitants. The survey conducted by the Tokugawa government did not encompass the Japanese population as a whole, however. Samurai (warriors) and their families were not registered; and some local lords excluded children and youth from the counts. Hence, inconsistencies interfere with the reliability of the population statistics. Nevertheless, these data may be adequate for examining and comparing trends within each of the regions. Keeping this in mind, we observe that O¯¯ u and Chu¯ goku regions experienced opposite population trends during the first half of the period studied and similar trends during the latter half of the period. Temporary population decrease during the 1830s occurred in a parallel fashion in the two regions, and the population increase in the latter half of the period may have been caused by the same factors. Since opposite trends in population change undoubtedly took place during the first half of the period, we can only conclude that independent factors leading to population fluctuation were at work in each of the regions.

2.2. Understanding Sex Ratio Changes in the O¯ u Region An important index obtained from the census results of the Tokugawa government was sex ratio. In the O¯¯ u region, the sex ratio was initially extremely unbalanced with a higher proportion of males. Over time it balanced out to a more normal statistic. Looking at these statistics in combination with the changes in population, we can see between 1750 and 1786 a large decrease in the male population and a smaller decrease in the female population. Comparing the number of men and women in Mutsu and Dewa provinces, we see that the decrease occurred exclusively within the male population. In Mutsu province, the difference was most significant at the beginning of the period with the gap closing somewhat by 1786, and the sex ratio reaching a normal level by the end of the period. We compared the sex ratio for the O¯¯ u region between 1750 and 1850 with figures for the five provinces in central Japan that had the most balanced sex ratios (the smallest differences in the numbers of men and women). Clearly, the sex ratio was abnormally high in the O¯¯ u region for the initial period, approached the national average in the 1780s, and then gradually decreased to a nearly parallel pattern. How should we interpret these changes in sex ratio; specifically, how did this fluctuation occur during the period of the O¯¯ u region’s population decrease in the

306

12:27:16:11:09 Page 306 Page 307

POPULATION AND HOUSEHOLD DYNAMICS

18th century? According to some authors, the people of this area had become somewhat accustomed to performing abortions and infanticide as a means of population control made necessary by poverty during the 18th century (as Kawaguchi, 1998). Population control itself may have existed in the region, and the authors of the present study are not attempting to dispute this. This raises several questions, however. If these people had truly wanted to implement population control because of the hardships of life, then control measures taken after birth, specifically infanticide, would have been applied only to female children. Reducing the number of female children would have resulted in a reduction in the number of future births in the family and fewer mouths to feed. Furthermore, boys would have been more suitable than girls for farm labor, which would contribute more to the family income. Therefore, if infanticide was to be implemented and a choice had to be made between killing a boy and a girl, barring exceptional circumstances, the girl would have been the one chosen. From the information available, we cannot determine if the abnormally high ratio of men in the O¯¯ u region in 1750 was the result of infanticide or of women having a shorter average lifespan. In either case, we are left with the question of how to explain the sudden normalization of the sex ratio during a period with famines in both the 1750s and 1780s. Of course, many factors other than infanticide contribute to setting the sex ratio of a resident population, such as the sex ratio of children born, differences in mortality rates between males and females at each age, and differences in the ratio of relocation between males and females. For the details of these factors, we must wait for the results of the historical demographic analysis based on the many population registers from the O¯¯ u region, both Shu¯ mon Aratame Cho¯ (SAC) and Nimbetsu Aratame Cho¯ (NAC) (Hayami, 2001a). The apparently contradictory fact of sex-ratio normalization during a period of population reduction, however, cannot be explained as the result of population control due to poverty. Based on what is currently known, we might consider external factors, such as extended periods of bad weather, and that faced with the threat of a declining population, these people attempted to curtail it by ceasing to practice female infanticide. In any event, specific information about fluctuations in death patterns must be ascertained to answer this question.

2.3. Population Trends of the Aizu Domain Let us now look at populations by clan domains. Surveys of the population of the Aizu domain (territory of the Lord of Aizu) were made from a comparatively early time; they were kept annually in the Aizu Domain Administration Records, beginning in 1675. The Aizu domain did not consist solely of mountains, but also included farming regions in ﬂatlands located near the provincial castle. While the population trends in such areas cannot be taken as representative for the entire region, they are the only population series data that could be obtained for the immediate area. The population trends (index) for 1675–1846 are shown in Fig. 16.2 (Hayami, 2001a, pp. 75–78). The population of the Aizu domain was approximately 50,000 inhabitants at the start of the period studied (1675) and grew over the next 20 years at an annual rate of approximately 0.5%, which was an extremely high rate of increase for the

307

12:27:16:11:09 Page 307 Page 308

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

Figure 16.2 Population trends of Mutsu province (white circles), Aizu Lord territory (black line), and Aizu mountainous district (black diamonds) (1675–1846). Source: Hayami (2001a).

time. Although the population continued to increase for an additional 25 years, the rate slowed, dropping to an annual average of approximately 0.04%. Drawing a logistic curve, we see that this census began during the final phase of a period of population growth. After that time, the population fluctuation trends of the Aizu domain were nearly the same as those of Mutsu province, as Fig. 16.2 reveals as well as the trend for the mountainous district currently under study. The population of the Aizu domain increased from the 17th century to the start of the 18th century before dropping below 40,000 in the 1780s, 1830s, and 1840s to reach a level of approximately two-thirds of the highest recorded population in 1713. The Aizu region entered a phase of new population increase in 1843. Although not shown in Fig. 16.2, the population exceeded 50,000 in 1890, but even this figure does not reach the population size observed in the late 17th and early 18th centuries. Although these trends in population shift were common in many areas of Japan, they were extremely pronounced in the area studied.

2.4. Population Trends in the Aizu Mountainous District Do the population trends in the Aizu mountainous district correlate with those for Mutsu Province, as well as the Aizu domain as a whole?1 Fig. 16.3 shows population trends of the four villages studied. A population increase occurred from 1691 to around 1776. The population began to decrease around 1777, reaching its lowest point in the middle of the 1820s, but then rebounding and starting to increase again in 1823. From this graph, we can see that the famine of the 1780s had a fairly strong impact on the Aizu mountainous district. At this point, we will categorize the Aizu population into three periods: (1) a period of demographic increase from 1691 to 1776, (2) a period of decrease around 1777 to 1822, and (3) a period of recovery starting in 1823 (Period 3). These trends will be linked to the general household trend shown below. The population trends of the Aizu mountainous district correspond almost

308

12:27:16:11:09 Page 308 Page 309

POPULATION AND HOUSEHOLD DYNAMICS

Figure 16.3 Population trends in four villages of the Aizu mountainous district (1691–1871). From top to bottom: the villages of Kanaizawa (black squares), Ishibushi (white triangles), To¯nosu (white circles), and Kuwanohara (black circles). Sources: see note 2.

exactly with data from Mutsu Province and the Aizu domain as a whole. Regard- less of whether these trends are representative of the area, no heterogeneity was seen.

2.5. Population Characteristics: Sex Ratios and Age Structure We will now take a look at population characteristics, looking first at sex ratio and then at age structure. Fig. 16.4 shows the sex ratio of the Aizu mountainous district. The mean sex ratio of 120.06 for 1691–1859 is considerably unbalanced with an overabundance of males, even though it fluctuated between 120 and 140 during the decades. The particularly large imbalance in the sex ratio during the 18th century lessened as the population began to recover in the 19th century. Although Kawaguchi proposed that this imbalance was the result of selective birth control by sex (1998, p. 13), the present authors have several doubts about this theory. To observe the age structure, the populations of the four villages have been totaled for each of the three periods, resulting in the population pyramids presented as Fig. 16.5. The population on the left, for Period 2 (1777–1822), is barrel shaped, indicating a decrease in the population of youth. By contrast, the figure on the right is a clean and distinct pyramidal shape with a large base for Period 3 (1823–1871), a time of population recovery.

2.6. Marriage, Birth, and Death: Implications on Family Forms Marriage, birth, and death can be seen as demographic indices that determine the changes and structure in a population organized in family households and the

309

12:27:16:11:09 Page 309 Page 310

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

Figure 16.4 Mean sex ratio of the Aizu mountainous district, four villages combined (1691–1859).

Figure 16.5 Age structure by sex in the Aizu mountainous district for two periods: 1777–1822 and 1823–1871.

Japanese ie. Previous historical demographic research indicated that the low age of ﬁrst marriage in the O¯¯ u region was a demographic characteristic of this area of Japan (Hayami, 2001a). Although the death rate at that time was much higher than it is today, stem families in the O¯¯ u region often consisted of three or even four generations because of the early age at ﬁrst marriage (Hayami, 1992, pp. 227–241). For the observation of marriage, birth, and death, we have used data from 1750 onward as a comparatively continuous set of historical source material from the Aizu mountainous district available for this time period.2 To analyze marriage status, three factors have been chosen: (1) the ratio of marriage

310

12:27:16:11:09 Page 310 Page 311

POPULATION AND HOUSEHOLD DYNAMICS

to ascertain major trends, (2) the age at ﬁrst marriage to understand when men and women married and (3) the ratio of individuals with a living spouse. Actual population survey reports tabulate legally registered marriage to yield the number of marriages. In the past, concept and customs of marriage varied with the region, but this potentially interesting research topic will not be considered in the current article. Instead, we shall proceed to take the ﬁrst year when the annual SAC listed the terms “husband,” “wife,” “bride,” or “groom” as the year in which the wedding took place (Hayami, 1987).

2.6.1. Marriages First, the total number of marriages will be used as the numerator for calculating the marriage ratio from 1758 to 1859 without distinguishing between first marriages and remarriages. Since the population under observation was small, we calculated marriage ratios on the basis of 5-year averages, but there are still extreme fluctuations. The ratio fell below 10‰ for the periods 1765–1769 and 1800–1804. Otherwise it fluctuated between 10‰ and 20‰ and rose above 20‰ in 1835 and thereafter. On the whole, the average marriage ratio for the whole period was 14.9‰. By comparison, the average marriage ratio during 1760–1870 for Yambe mura (today, Tendo City in Yamagata Prefecture, located in the O¯¯ u district about 100 km north of Aizu) was 18.2‰ (Kinoshita, 2002, p. 52). The marriage ratio in the Aizu mountainous district never reached this level, but it should not be considered low when compared, for example, with our data for Nishijo mura in Mino Province, located in central Japan, where the marriage ratio exceeded 10‰ only three times during the same period of study (Hayami, 1992, p. 226).

2.6.2. Age at First Marriage Next, let us look at the age at first marriage in the Aizu mountainous district toward the end of the Tokugawa regime. In previous research for this period, we found that age at first marriage was higher in western Japan and lower in eastern Japan. We therefore expected that the Aizu mountainous district would follow the norm of early first marriage. Several preliminary problems needed to be resolved, however. If a person stayed in his or her village of birth until marrying, it was possible to judge whether a marriage was the person’s first or a remarriage. Such judgment, however, was not possible in the case of people (1) who were born before the period we are studying but married within that period, or (2) who were born within that period but left their villages temporarily to work elsewhere, perhaps as apprentices or servants, or (3) who moved to a village for the purpose of getting married. Therefore, to secure an accurate result, we limited the study of first marriage to brides and grooms who did not leave their respective villages of birth during the period under study. Unfortunately, this decision reduced our sample. Also since limiting the data to these conditions could introduce some type of bias into the data collected, we had to apply certain conditions in calculating the age at first marriage (shown in Table 16.1). Data were divided into three categories: data (a) for all marriages (including remarriages), (b) for marriages that were clearly first marriages, and (c) for marriages that appeared to be first marriages, but for which some uncertainty still existed. This third category covers

311

12:27:16:11:09 Page 311 Page 312 1859) – NSD rst marriage fi (including unclear cases) (c) NSDAge at at NSDAge rst fi rst marriage in four villages of Aizu mountainous district (1758 fi marriage (b) Standard deviation. = Sources: see note 2. Sources: SD NSDAge at at NSDAge Age at marriage at Age (including remarriages and unclear marriages) Age at marriageAge at (including remarriages and unclear marriages) and age at nosu 25.15nosu 20.52 238 6.99 22.56 260 112 5.58 4.47 17.94 23.41 98 2.64 147 19.53 5.29 181 4.60 ¯ ¯ Table 16.1 Table nameVillage (a) Males KanaizawaTo IshibushiKuwanohara for 4 villagesTotal 26.39Females 26.03Kanaizawa 24.77To 27.45IshibushiKuwanohara for 4 villagesTotal 19.80 20.13 19.66 20.86 310 780 9.47 117 115 9.32 22.15 12.57 7.60 22.42 22.84 22.31 379 159 910 406 7.67 138 5.15 133 76 7.48 59 5.10 16.62 23.23 10.53 5.12 17.25 5.90 4.96 23.51 15.74 18.89 23.61 24.25 147 335 2.87 39 51 2.95 18.60 194 2.15 3.16 18.93 506 6.29 18.85 18.90 6.51 82 83 7.58 7.67 296 697 6.68 6.15 115 105 7.85 4.48

12:27:16:11:09 Page 312 Page 313

POPULATION AND HOUSEHOLD DYNAMICS

the total number of marriages after subtracting those that were clearly remarriages. In all these categories, ages calculated for men show no large variations. On the contrary, for women of this Aizu mountainous district, there were large age variations in the second category (marriages that were clearly first marriages) between Ishibushi mura and Kuwanohara mura, but there was little variation in men’s ages, which may simply be due to the small sample size, however. Comparing the second and third categories (marriages that were clearly first marriages and marriages that appear to be first marriages but which include some uncertainty), we found a difference of approximately 1–2 years for men and a maximum difference of 3 years for women. We therefore consider the age at first marriage for men was 22–24, and for women, 16–19. One particularly notable characteristic of this table is the standard deviation for women in the second category (marriages that were clearly first marriages), at the same time there was very little fluctuation for men. This variation in the age at first marriage for women suggests that there was an age considered to be the proper marrying age (although it varied among the villages), and that all girls tended to marry as soon as they reached that age.

2.6.3. Ratio of People with Living Spouses This proper age at marrying becomes even clearer when we look at the ratio of people with spouses alive by age groups (Fig. 16.6). Compared with the values shown for men, the ratio of women with spouses increased more rapidly around the average age at ﬁrst marriage. From this graph, we can also see that for all villages, the proportion of married women in their early 30s exceeded 90%, while the ratio of married men of the same age did not reach 80%. This may be an

Figure 16.6 Ratio of married people by sex and age, Aizu mountainous district (1691–1859).

313

12:27:16:11:09 Page 313 Page 314

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

Table 16.2 Total fertility rate (TFR) and total marital fertility rate (TMFR) in four villages of Aizu mountainous district (1758–1859)

TFR TMFR

Kanaizawa mura 2.58 3.13 To¯nosu mura 3.23 4.14 Ishibushi mura 2.60 3.04 Kuwanohara mura 2.90 3.73 Total for 4 villages 2.79 3.42

mura = village. Sources: see note 2.

eﬀect of the high sex ratio of men to women, as discussed above and shown by Fig. 16.4.

2.7. Fertility Based on the early age at first marriage for women, we might have expected that the women of the mountainous district would have given birth to many children. Like the marriage rate, fertility and death rates were also calculated for 5-year periods. There is some fluctuation in the crude fertility rate,3 but also an increase from just below 20‰ at the beginning of the period studied to nearly 30‰ in its latter half. The total fertility rate (TFR) was very low: 2.58 for Kanaizawa mura, 3.23 for To¯nosu mura, 2.60 for Ishibushi mura, and 2.90 for Kuwanohara mura (Table 16.2). In the case of Yambe mura (Tendo¯ City in Yamagata prefecture), we divided the period from 1760 to 1870 into thirds, and the period from 1760 to 1799 had the lowest TFR of 4.61. This further emphasizes the very low fertility rate of Aizu. Even looking at the total marital fertility rate (TMFR), the only village to exceed a value of 4 was To¯nosu mura. Therefore, although the age at first marriage was low, the fertility rate was also low. Of course, the actual fertility rate would have been higher than that recorded since it would have included children who died before being registered in SACs. Even taking this into account, however, the TFR for this Aizu area was still exceptionally low.

2.8. Mortality Lastly, let us look at mortality rates. Compared with the fertility rate, the crude death rate for the period studied shows a greater degree of fluctuation. This may be due to the effects of disasters such as famine and disease. When looking at the age-specific death rates, the death rate for children up to the fourth year following birth was as high as 40, regardless of gender. After that age, the rate drops below 10. For the age group 20–50, however, the rate for women was much higher than for men, probably as a result of pregnancy and parturition. After age 50, the rate increased for both men and women. How long did the residents of the Aizu mountainous district live during the period studied? Because it is difficult to create mortality tables for each village due to the small sample sizes, we instead created mortality tables for each sex using

314

12:27:16:11:09 Page 314 Page 315

POPULATION AND HOUSEHOLD DYNAMICS

the totals for the Aizu mountainous district. Two difficulties arose due to the characteristics of SACs being used as historical material: (1) since children were not listed until the second calendar year following birth, the infant mortality rate is impossible to calculate accurately; and (2) SAC entries for some people were discontinued for unknown reasons. As Kinoshita indicated, “there are too many cases to ignore of record entries being discontinued for unknown reasons, that is, cases of entry discontinuance that may or may not have been due to death” (1995, p. 163). We solved the first difficulty, considering as a working hypothesis that SAC entries for “children in their second year” indicated children who had reached the first birthday; we evaluated the death rate for children age 0 as 200‰, taking the midpoint for “life years” as 70% (Hayami, 1992, pp. 248–250). As for the second difficulty, the number of cases of records being discontinued for unknown reasons was too large to be ignored (304, 130 men and 174 women). Therefore we adopted Kinoshita’s proposed correction: we attributed death as the cause of discontinuance for unknown reasons to all people in their 65th year following birth and to one-half of those younger than that (Kinoshita, 1995, pp. 163–171). We then drafted a mortality table based on these hypotheses. This table indicates that the average duration of life at the time of birth was 39.7 for men and 36.0 for women, and that the point at which the survival rate for all births dropped below 50% was age 47 for men and age 30 for women. Comparing their survival rate curves, we see a high mortality rate for both men and women before age 5, after which the mortality rate stabilizes and little difference is seen in their death rates until approximately age 20. A difference between the sexes can be seen between ages 20 and 60, however, when the survival rate curve of men is higher than that of women. We remember that the average age at first marriage for women in the Aizu mountainous district was 17.25 (Table 16.1). Considering the life course of these women, their lower survival rate in adulthood may have been the result of deaths due to pregnancy and childbirth, hence the aforementioned unusual sex ratio. With more knowledge about all these demographic characteristics—especially population depression, early marriage, high female mortality, and unbalanced sex ratios (with more men than women), we shall try to clarify how demographic characteristics played their role in the household structure and in the development of a stem-family system.

3. HOUSEHOLDS DYNAMICS IN AIZU MOUNTAINOUS DISTRICT 3.1. Trends in the Number and Size of Households Keeping in mind speciﬁc regional trends in population change over time, we can now study household composition and family transmission behavior in the same mountainous district of Aizu domain. Number, size, and composition of households have been studied all over Japan for the Tokugawa period, both at the micro and macro levels, and important regional variations have been identiﬁed (Hayami, 2001a; Hayami & Kurosu, 2001; Hayami, Saito, & Toby, 2004). Here we intend to focus on the household reproduction cycle over time and models of

315

12:27:16:11:09 Page 315 Page 316

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

reproduction of the stem family, which was the common family system of this district (Suzuki, 1968; Koyama, 1959; Berkner, 1972). To clarify the composition of households, it is important to analyze how household structure changed with time and if a cycle may be identiﬁed in its development. The conjugal family unit (CFU) is connected to the reproduction process, particularly in stem families (Laslett, 1972). Therefore, analysis of the household cycle must be limited to periods for which a continuous series of historical materials is available. As a consequence, we have limited the use of historical materials in the analyses of households and family estates to periods for which a comparatively continuous series of data for the Aizu mountainous district were available (sources are listed in note 2). As Fig. 16.7 shows, the number of households in each village was maintained at nearly a constant level throughout the period studied (1752–1859). In Fig. 16.8, when the average distribution of household size in the four villages was combined, the maximum size of the household was 13. The mean household size was 4.6, and the distribution (in percent) was concentrated around four-, ﬁve- and six-person households. Contrary to what one might expect, no extremely large households existed in the Aizu mountainous district. This can also be considered one characteristic of the local rural household (ie).

Figure 16.7 Trends in the number of households in four villages of the Aizu mountainous district (1752–1859). From top to bottom: Kanaizawa, To¯nosu, Ishibushi, and Kuwanohara.

Figure 16.8 Distribution (in percent) of household size when four villages of the Aizu mountainous district are combined (1680–1860).

316

12:27:16:11:09 Page 316 Page 317

POPULATION AND HOUSEHOLD DYNAMICS

3.2. Types of Household Structure: Cambridge Categories and Classes Historical demographers and family historians have frequently used the Ham- mel & Laslett model, which we will also refer to here as the “Cambridge model” to classify household structure according to its categories and classes (Hammel & Laslett, 1974; Laslett, 1972). The Cambridge model classifies households into those that include non-family members, such as apprentices, servants, and boarders (tenants), and those that include only family members. Next households are divided into two levels with classes under each of six categories (Table 16.3). The organization of classes in each category is based on how the family may be extended from the conjugal family unit (CFU) of the household head: the family might extend upward to an older generation, horizontally to members of the same generation, or downward to a younger generation. These extensions could include individuals as well as couples forming their own CFUs. It is not an exaggeration to say that this classification model is used overwhelm- ingly in the analysis of household composition. Many studies published in Europe and elsewhere have used this classification model, which provides a unique frame for international comparisons of household structures. Such a model, however, is not effective for classifying households in societies such as Japan—where the stem family household is typical, or China—where the joint family household is typical. This is a major inadequacy of this model. When attempting to categorize a household, the largest classification units available in the Cambridge model are categories 5 and 6. Therefore, both stem family households, peculiar to

Table 16.3 Cambridge classiﬁcation system of household structures

Category Class

Source: Laslett (1972, p. 31, Table 1.1).

317

12:27:16:11:09 Page 317 Page 318

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

traditional Japanese society, and joint family households, which include CFUs of collateral relatives, must be classified under either category 5 (Multiple family household) or category 6 (Indeterminate). Such categories do not clarify the ratio of stem family households within a society. If the stem family household could be classified using the “classes” under each category, there would be no problem. In previous historical studies, this model has actually been used to investigate cycles of Japanese stem family households (Kinoshita, 1990; Kito¯, 1981). Yet problems remain. For example, class 5a includes families that extend upward to a CFU from a generation older than the household head’s CFU; this could include cases in which there is a direct secondary unit up and cases in which there is a collateral secondary unit up. Therefore, in the strictest sense, the use of classes still does not precisely identify stem family households. Also class ranges are so narrow that many complex cases do not fit into any class. Fundamentally, the significance of classification is lost when too many classification types become available. Actually, the detection of household cycles, one of the themes of this article, requires that the variety of classification types be kept to a minimum. Therefore, we have made some revisions in the Cambridge model.

3.3. Creating Subcategories and Subclasses for the Modified Cambridge Model For our results to be applied inter-regionally (in Japan) and for international comparisons, we have attempted to create in a form that is easy to understand and allows us to revisit the original Cambridge model. One situation that we faced in this article was unique to the traditional society of Japan: the stem family household in which older parents lived with one household heir. Therefore, within this modified model, we needed to find a method for extracting only stem family households. To clarify the number of stem family households independently, first, subcategories had to be established as an auxiliary classification unit under the original categories (see lower part of Table 16.4). Two subcategories were added to category 5 (Multiple family households): 5i (Stem family households) and 5ii (Joint family households). Thus, stem family households can be separated from other household compositions in a way that simply ignores the class. Since existing historical demographic analyses make use of the class unit, our subcategories cannot be readily used for comparisons with the results of other research. Hence we also established subclasses under the class unit (see upper part of Table 16.4). These subclasses only affect classes 5a, 5b, 5c, and 5e of category 5. Establishing subclasses makes it possible to separate these classes between stem family household formats and other household formats (not stem). By placing subclasses under the two subcategories “stem family households” and “joint family households,” stem family household data can be extracted even when the Cambridge model is used (Table 16.4). This may seem a bit complicated, but minimizing the modifications to the original Cambridge model enables us to compare our results with research from other countries as well as with the results of previous research. Also we are more easily able to understand the ratio of stem family households, a format peculiar to Japan that was manifest during the Tokugawa Period. In this article, we will

318

12:27:16:11:09 Page 318 Page 319

POPULATION AND HOUSEHOLD DYNAMICS

Table 16.4 Modiﬁed Cambridge Model classifying household structures

Category Class Subclass

1. Solitaries 1a Widower (Widow) 1b Single or Marital Status Unknown 2. No family 2a Cohabitation with Siblings 2b Cohabitation with Other Relatives 2c Cohabitation between Persons of Unknown Relationship 3. Simple family 3a Married Couple households 3b Married Couple and Children 3c Widower and Children 3d Widow and Children 4. Extended 4a Extension Upward family 4b Extension Downward households 4c Extension Horizontally 4d Combination of 4a, 4b and 4c 5. Multiple 5a Includes Secondary Units up 5a 1 Stem Secondary Units up, with family Only Conjugal Family Unit per households Generation 5a 2 Collateral Secondary Units up 5b Includes Secondary Units 5b 1 Stem Secondary Units down, down with Only Conjugal Family Unit per Generation 5b 2 Collateral Secondary Units down 5c Includes Secondary Units Horizontal 5e Others 5e 1 5a1+5b1 5e 2 5a2+5b1, 5a1+5b2 or 5a2+5b2 5e 3 5a1 (or 5a2)+5c 5e 4 5b1(or 5b2)+5c 5e 5 5a1or 5a2)+ 5b1(or 5b2)+5c

Category Subcategory Class Subclass

5. Multiple family households 5i Stem family households 5a 1 5b 1 5e 1 5ii Joint family households 5a 2 5b 2 5c 5e 2 5e 3 5e 4 5e 5

319

12:27:16:11:09 Page 319 Page 320

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

refer to this as the modified Cambridge model, and apply it to the analysis of household composition in the Aizu mountain district. A serious difficulty also arises when attempting to assess the morphology of a household: what percentage of households must correspond to a specific type for that household composition to be considered typical? A standard percentage value must be determined. In his analysis of marital formats, Murdock suggested a standard percentage value of 20% to determine if a marital format was a general family format (Murdock, 1949 (Murdock, 1965 paperback edition), p. 28). On the other hand, Saito¯ (1999) used the results of analysis by Toda to set the standard percentage value for stem family households at 30% as a way to clarify the characteristics of household compositions in the late Tokugawa Period. In his analysis, Toda (1937) used data from the first national census conducted in 1920. At that time, the family system seemed to be firmly entrenched in Japanese society, and people of Japan led a lifestyle based on family system ideals. We shall thus use this ratio (30%) as one of the standard percentage values. Taking into consideration the developmental cycle of family composition, Todd claimed that when a computer used as conditions the typical death rates and fertility rates under an ancient regime to create a simulation, the results showed that three-generation households never exceeded one-third of the total at any given time (Todd, 1990, p. 44). Previous research has thus already discussed this standard percentage value. In the light of those arguments, it seems appropriate to set the standard value for the stem family household cycle at around 30% for this article.

3.4. Household Structure Using the modified Cambridge model, we classified households for each village after compensating for years in which household composition data were missing and then calculated totals for the entire Aizu mountainous district. The total was 14,145 household-years after compensating for missing years (before that, the number of samples was 13,477) (Table 16.5). Looking at the highest category used in the model was category 5 (multiple family households), which made up the largest ratio at 39.6%. The next largest unit level was category 3 (simple family households), 28.4%; followed by category 4 (extended family households), 22.6%. In the subcategories under category 5, the proportion of stem family households in subcategory 5i accounted for 35.7% of all households. Stem family households made up approximately 90% of all multiple family households, and this household composition made up the largest ratio of any household composition for any of the categories (except category 5—Multiple family households) or subcategories. Now let us examine whether differences can be detected according to social classes.4 More multiple family households are observed as the level of social class increases (Table 16.5). Comparing the ratios between classes, we see a difference of more than 20% between the upper and lower classes. A difference of approximately 15% is also seen for subcategory 5i (Stem family households). There was approximately the same possibility of a joint family household forming in the upper class (9.6%) as of a solitary family household forming in the middle and lower classes. Conversely, a joint family household was likely to form in the

320

12:27:16:11:09 Page 320 Page 321

POPULATION AND HOUSEHOLD DYNAMICS

Table 16.5 Household structure by social classes, in the Aizu mountainous district (1752–1859), using the modiﬁed Cambridge model

Category Sub category Total Upper class Middle class Lower class

%N%N%N%N

1. 8.37 1184 1.98 70 9.55 736 13.00 378 2. 0.83 117 0.82 29 0.56 43 1.55 45 3. 28.46 4025 19.94 704 30.42 2345 33.57 976 4. 22.69 3209 24.53 866 22.24 1714 21.64 629 5. 39.66 5610 52.72 1861 37.23 2870 30.24 879 All 100.00 14145 100.00 3530 100.00 7708 100.00 2907 5. 39.66 5610 52.72 1861 37.23 2870 30.24 879 5i. Stem family 35.77 5059 43.09 1521 35.08 2704 28.69 834 households 5ii. Joint family 3.90 551 9.63 340 2.15 166 1.55 45 households

Unit: household-year. Sources: see note 2.

middle or lower classes, as well as a solitary family household forming in the upper class (1.9%).

3.5. Transference between Household Categories The first step in detecting household composition cycles is to clarify for each household the degree of transference among the household composition categories, looking at this transference during 2-year periods from the first year observed until the last year of the study (for example, transference between the first and second year, between the second and third year, between the third and fourth year, and so on) (Koyama, 1959; Kito¯, 1981). For our study, we applied an elaborate matrix by Kinoshita (1990) and used class units. Our main purpose was to determine whether it was possible to detect cycles related to stem family households. To avoid the necessity of complex observations, for the present study we eliminated category 5 of the modified Cambridge model and used only the two subcategories 5i (Stem family household) and 5ii (Joint family household). The transference of a household from one category to another is shown in Table 16.6, a matrix that could be read as follows: the row heading indicates the category/subcategory from which the household transfers and the column heading indicates the category/subcategory to which the household transfers. This table uses the four well-known categories 1, 2, 3, and 4, and the two subcategories (5i and 5ii) we added to the modified Cambridge model, as described earlier. For convenience, we also added category 0. The row Category 0 indicates the appearance of a new household, possibly as a result of the formation of a branch household, the restoration of a family line that had been extinct, or a relocation. The column category 0 indicates a household that ceased to exist due to the family line becoming extinct or relocating. With the addition of category 0, household composition can be quickly comprehended at the point where a branch household was created, as well as composition before a family line

321

12:27:16:11:09 Page 321 Page 322

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

Table 16.6 Transference of households from one category to another during 2-year periods, categories/subcategories (including not transferred)a in the Aizu mountainous district (1752–1859) using the modiﬁed Cambridge model

a 0b 12345i5iiTotal

0b 02512954165 0 38.46 1.54 44.62 7.69 6.15 1.54 100.00 0 0.18 0.01 0.21 0.04 0.03 0.01 0.46 1 Solitaries 53 1051 1 50 12 0 0 1167 4.54 90.06 0.09 4.28 1.03 0 0 100.00 0.38 7.50 0.01 0.36 0.09 0 0 8.32 2 No family 2 17 82 0 15 0 0 116 1.72 14.66 70.69 0 12.93 0 0 100.00 0.01 0.12 0.58 0 0.11 0 0 0.83 3 Simple family 8 68 9 3610 74 207 1 3977 households 0.20 1.71 0.23 90.77 1.86 5.20 0.03 100.00 0.06 0.48 0.06 25.75 0.53 1.48 0.01 28.36 4 Extended family 1 5 18 197 2806 136 7 3169 households 0.03 0.16 0.57 6.22 88.55 4.29 0.22 100.00 0.01 0.04 0.13 1.41 20.01 0.97 0.05 22.60 5i Stem family 2 0 3 70 237 4622 43 4977 households 0.04 0 0.06 1.41 4.76 92.87 0.86 100.00 0.01 0 0.02 0.50 1.69 32.96 0.31 35.50 5ii Joint family 00161739486549 households 0 0 0.18 1.09 3.10 7.10 88.52 100.00 0 0 0.01 0.04 0.12 0.28 3.47 3.92 N 66 1166 115 3962 3165 5009 538 14021 % 0.47 8.32 0.82 28.26 22.57 35.72 3.84 100.00

See Kinoshita (1990). Sources: see note 2. a The row heading indicates the category/subcategory from which the household transferred and the column heading indicates the category/subcategory to which the household transferred. b Category 0=Collateral family formation, family line restoration, relocation.

became extinct. Although category 0 allows us to classify cases of household appearance and disappearance due to relocation, absolutely no cases of transference into or out of the area for relocation were seen for our district. There were, however, cases where the head of a solitary household went to another village for an apprenticeship with no other household members left behind. In such cases, the household was counted as moving to category 0, if the number of household members was zero for ﬁve years or more. Accordingly, transference to category 0 indicated household extinction, disappearance, and which the household members transferred to another area for ﬁve years or more, with no one remaining in the Aizu mountainous district household. Transference from category 0 indicated

322

12:27:16:11:09 Page 322 Page 323

POPULATION AND HOUSEHOLD DYNAMICS

cases of branch household formation, the restoration of an extinct family, or the return of household members to a household that had no members for 5 years or more. The seven category and subcategory classifications will be used as we proceed with our study of household dynamics. In each cell of Table 16.6, the top number indicates the actual count of cases, the second number indicates the ratio of this number to the total number of households moving from the corresponding row category, and the bottom number indicates the ratio of this number to the total of all household transference. Kinoshita (1995) extracted and analyzed only cases of transference from one household composition category to another, but this does not tell us what percentage of the total underwent a change in household composition. Therefore, we first created a matrix that included cases of no transference between categories. This matrix covered 14,021 cases. Next, cases in which there was no change in household composition were eliminated from the matrix. This left a new matrix that included only cases in which there was a change from one household composition category/subcategory to another (1,364 cases). Since the total number of cases studied, including those for which there was no transference as seen previously, was 14,021, the ratio of transference between household composition categories was 9.7%. Accordingly, the study we make of transference between household composition categories encompasses 9.7% of the total number of households in the Aizu mountainous district. Our observations of the transference of household compositions between categories indicate a high level of stem family household transference related to the whole. Although this can be seen for each of the social classes, transference is particularly notable in the upper class in both directions between joint family households and stem family households. No major differences between the middle and upper class were seen, but in the middle class, transference related to the simple and extended family household compositions was more prevalent than in the upper class. In the lower class, transference related to solitaries and extinct households occurred more frequently than in the upper and middle classes (Okada, 2001).

3.6. Household Cycles and Patterns To analyze household composition cycles, we extracted all cells in the category transference matrix for which the ratio of transference was 1% or more of the total. Then we examined the category or subcategory to which each household composition transferred and followed the subsequent transference (passage) of each household. These results are shown in a cyclical diagram (Fig. 16.9). In this diagram, numbers in the square outlines indicate the corresponding categories or subcategories from our modiﬁed Cambridge model. The arrows indicate transference to the corresponding category or subcategory, and arrows express all transference occurring at a ratio of 1% or more. Transference ratios of 10% or more are shown by the thickest arrows, transference ratios of 5% but less than 10% are shown by medium arrows, and transference ratios of 1% but less than 5% are shown by thin arrows. Added ﬁgures near each arrow read as follows:

323

12:27:16:11:09 Page 323 Page 324

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

P(a→b) = c/d. The “P(a→b)” next to each arrow indicates households transferring (passing) from category a to category b. In the c/d fraction on the right side, the numerator c shows the actual number of households moving from category a to b. The denominator d shows the total number of transference cases. Under some of the P(a→b) items are P e→a (a→b) items, which indicate a breakdown of the transference before P (a→b) transference. For example, in the case of P e→a (a→b), the household ﬁrst transferred from e to a, and then from a to b. In the g/h fraction on the right, the numerator g shows the actual number of cases in which there was a transference from category e to a and then to b. The denominator h shows the total number of cases in which there was transference from category e to a. The same applies to any additional breakdown of an item. Fig. 16.9 shows the household composition cycles for the entire mountainous district, combining the four villages (1752–1859). This general diagram shows clearly a cycle concentrating around stem family households (5i) and simple family households (3), also going through extended family households (4). Looking at each of the social classes, we notice that, for the upper class, the arrows indicate transference concentrated around the stem family household (5i) (Okada, 2001). Another characteristic of the upper class is transference related to joint family households. Transference between joint and extended family households occurred only in the upper class, and transference between stem and joint family households was then at the ratio 5% but less than 10% ratio, represented by medium arrows. For the middle class, transference occurred primarily among the stem, extended and simple family households—similar to the trend in the upper class. Transference between the stem and joint family household categories in the middle class was at the ratio of 1% but less than 5%, indicated by thin arrows. Also, the transference between joint and extended family household categories observed for the upper class did not occur in the middle class. For the lower class, transference was considerably more complex than for either middle or upper classes. As with the middle class, transference in the lower class occurred primarily among stem (5i), extended (4) and simple family households (3). While most of the transference in the middle class occurred among these three household compositions, in the lower class, more transference appears among the other categories as well, with a more obvious diversity in the types of transference. Particularly noteworthy were transferences from simple family to solitary households, and from solitary to extinct households (category 0). It reﬂects a pattern by which the household composition changes from a simple family household to a person living alone (solitary), after which the household becomes extinct.

3.7. Frequency of Cycles Looking at the Aizu mountainous district as a whole and considering the ﬁgures attached to the arrows in Fig. 16.9, we see that the ratio of patterns returning to stem family households (5i) was 33.5% while the ratio of patterns returning to simple family households (3) was 9.9%, a fairly large diﬀerence between the two

324

12:27:16:11:09 Page 324 Page 325

Figure 16.9 Household structure transferences (passage) and cycling for all social classes, using modiﬁed Cambridge model in four villages of the Aizu mountainous district combined (1752–1859). Notes: 1. Numbers in the square outlines indicate the corresponding household structure categories or subcategories. The arrows indicate transference to the corresponding category or subcategory, and arrows express all transference occurring at a ratio of 1% or more. Main categories: (1) Solitaries (2) No Family (3) Simple Family Households (4) Extended Family Households (5i) Stem Family Households (5ii) Joint Family Households 2. Passage: P(5i→5ii) means passage from subcategory 5i (stem family household) to subcategory 5ii (joint family household). P(5i→5ii) = 44/1364 means 44 cases of passage from subcategory 5i (stem family household) to subcategory 5ii (joint family household), on a total of 1364 transferences observed. P(5ii→5i) = 39/1364 means 39 cases of passage from subcategory 5ii (joint family household) to subcategory 5i (stem family household) on a total of 1364 transferences observed. Cycle: P 5i→5ii (5ii→5i) = 29/44 means that on a total of 44 cases of passage from subcategory 5i (stem family household) to subcategory 5ii (joint family household), 29 are returning back to subcategory 5i (stem family household).

12:27:16:11:09 Page 325 Page 326

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

types. Of the four cycles that were patterns returning to stem family households, the most common was circulation between stem (5i) and extended family households (4). Since the pattern of circulation among stem, extended and simple family households is generally considered to be typical among patterns returning to stem family households, it was surprising to find a high ratio of circulation between stem (5i) and extended family households (4). Looking at these cycles by social class, we see that the rate of the pattern returning to stem family households was 41.95% in the upper class, 31.6% in the middle class, and 28.9% in the lower class. The ratio of patterns returning to stem family households in the upper class was approximately 10% higher than that of the middle class and approximately 13% higher than that of the lower class, fairly large differences. The rate was higher in the upper class than in the other classes due to the higher ratio of the cycle of circulation between stem and joint family households. Furthermore, although no large differences existed between social classes in the ratios of the cycle of circulation between stem and extended family households, there was a difference of 3–4% between the upper and middle class for the ratio of the cycle of circulation among stem, extended and simple family household categories, the cycle type usually considered typical for stem family households. For the upper class, this cycle was truly the most common of the patterns returning to stem family households, but for the middle and lower classes, the cycle of circulation between stem and extended family households was the most common. For the cycle of circulation between stem and extended family household categories, there was little difference among the social classes, but we can surmise that this cycle was also a pattern that would have been resistant to failure in all social classes. The lower ratios of the pattern of circulation among the stem, extended and simple family household categories in the middle and lower classes may have been due to the existence of household composition transitions that deviated from the normal patterns in these classes. By contrast, the highest ratio of the pattern returning to simple family households was seen in the middle class at 11.6%, followed by the lower class at 10.3%, and the upper class having the lowest ratio at 5.4%. The high ratio seen for the middle class is a result of the high ratio of the cycle of circulation between simple and extended family household categories, another pattern returning to simple family households. This type of circulation may also include cases in which a transition could not be made to a stem family household. Of course, we can expect that this included a fair number of cases in which divorced siblings returned to their original homes and later remarried. From these observations, we can infer that the pattern returning to stem family households was established and stable in the upper class. Although the pattern returning to stem family households was the cycle type that could have been established most readily in the middle and lower classes, the ratio of the pattern returning to simple family households was higher in these classes than in the upper class. The environments in the middle and lower classes were not ones in which the pattern of returning to stem family households could be as readily and stably established as in the upper class. The cycles of each pattern of return were drawn from historical material, and the corresponding rates were calculated, but this question remained: how should

326

12:27:16:11:09 Page 326 Page 327

POPULATION AND HOUSEHOLD DYNAMICS

this information be interpreted? Based on these ratios, was the stem family household the typical household composition of the Aizu mountainous district? This is a persistent problem when taking a quantitative approach and an important matter of consideration for this article. As mentioned earlier, we studied the reference values from previous research and set the reference at a rate of 30%. Thus, let us compare the results of the analysis of household composition cycles with these reference values. The ratio of the pattern returning to stem family households for the Aizu mountainous district as a whole was 33.5%, which slightly exceeded this threshold limit by only a small amount. When looking at the district as a whole, it would be appropriate to conclude that the stem family household was the typical household composition. Now, looking at each of the social classes, we see that the upper class well exceeds this threshold value; clearly, the stem family household was the typical household composition for the upper class. For the middle class, the ratio is almost equal to the threshold value, making the stem family household typical for that class as well. The ratio in the middle class, however, was approximately 10% lower than that of the upper class, and the ratio of patterns returning to simple family households was higher for the middle class than for the other classes, making the rules of stem family household composition weaker than with the upper class. Even if the rules of stem family household composition were as strong in the middle class as in the upper class, the middle class was not as blessed as the upper class with an environment for realizing stem family households. Thus, in some cases the cycle may have been unable to form. In the lower class, the ratio was slightly below the reference value. From the analysis of household composition cycles, the stem family household was not the typical household composition for this class.

3.8. Duration and Timing of Cycles Lastly, we would like to clarify how many years one pattern of returning to stem family households lasted. For the cycle of circulation among the stem, extended and simple family household categories—the cycle considered to be the typical pattern returning to stem family households—the average number of years for completing the cycle was 32.6. By comparison, in two villages located in Nihonmatsu domain in the neighboring lowlands (an area analyzed using the same methods as applied here to the Aizu mountainous district), the average number of years for this typical cycle was 35.7 (Okada, 2000). The cycle was slightly shorter in the Aizu mountainous district, but the difference was not large. Accordingly, we can consider the period for the typical pattern returning to stem family households to be roughly between 30 and 35 years. The differences between classes in the two villages in the territory of Nihon- matsu domain could be studies because the length of the period studied and the large number of households. In the case of our Aizu mountainous district, since we had only 1364 cases, even if totals for all of the villages were used, it would be difficult to study class differences. Thus, as an alternative solution, we calculated the number of continuing years for each category and subcategory to study social class differences (Table 16.7). One surprise was that although we expected the number of continuing years would be highest in the upper class for all categories

327

12:27:16:11:09 Page 327 Page 328

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

Table 16.7 Average continuity of household structure (in years, by social classes in four villages of Aizu mountainous district (1752–1859)

All Upper Class Middle Class Lower Class

Average N Average N Average N Average N (years) (years) (years) (years)

1. Solitaries 9.63 98 6.30 10 10.38 58 9.30 30 2. No family 3.47 32 3.63 8 2.85 13 4.09 11 3. Simple family households 9.16 301 7.65 65 9.32 176 10.35 60 4. Extended family households 7.80 321 8.67 82 7.43 176 7.70 63 5. Multiple family households 9.83 351 9.81 114 9.39 179 11.22 58 5i. Stem family households 10.19 302 9.97 88 9.90 159 11.38 55 5ii. Joint family households 7.59 49 9.27 26 5.30 20 8.33 3

For categories, see Table 16.4. Sources: see note 2.

and subcategories that was not the case. The number of continuing years for solitaries and simple family households was lower for the upper class than for the middle and lower classes. Comparing the upper and lower classes, the number of continuing years for simple family households was 2.7 years shorter for the upper class than for the lower class, and the number of continuing years for stem family households was 1.4 years shorter. It appears that family composition development occurred more quickly among the upper class. Even if some accident befell a household in the upper class, it would have had superior capabilities for dealing with the accident, and the reproduction of upper-class families would have occurred in a somewhat planned manner. From these observations, six cycles in the household compositions of farming families in the Aizu mountainous district were found. Four of the cycles return to the stem family household and two return to the simple family household. Among these cycles, the pattern of circulation between the stem and extended family households was the most common that occurred. Basically, the most commonly occurring type of family cycle in the Aizu mountainous district was as follows: one of the parents of the married adult head of the stem family household would die, resulting in the formation of an extended family household; while the remaining parent was alive, one of the children of the household head would marry and live together with the family. In contrast with central and southwestern Japan, the generation interval in northeastern Japan was shorter, and our study provides further evidence of the validity of this observation (Hayami & Kurosu, 2001). Totaling the number of patterns returning to stem family households yields a total transference ratio of 33.5%. From this, we can conclude that the typical household composition in the Aizu mountainous district was indeed the stem family household. We also discovered, however, that household composition circulation patterns varied greatly depending on social class. In the upper class, the ratio of patterns returning to stem family households exceeded 40%, clearly indicating that in the upper class more than the others the stem family household composition was a

328

12:27:16:11:09 Page 328 Page 329

POPULATION AND HOUSEHOLD DYNAMICS

stable family condition functioning under extremely strong and binding rules. One of the patterns worthy of attention is the cycle of circulation between the stem and the joint family households. When considering the stem family cycle, the joint family household is a composition that seems contradictory to this cycle. The fact that there was a high ratio of this household composition in the upper class can be used to elucidate the family strategies of households belonging to the upper class. A typical example of this circulation between joint and stem family household compositions went as follows: multiple children married and then lived together in the parents’ household; later on, one child and his or her spouse remained in the household while the other siblings and their spouse left. This type of circulation, of course, signifies the creation of a branch household. Accordingly, with this cycle, the creation of the branch families did not occur at the point when the siblings other than the household heir married, but rather at a later time. This was one of the strategies applied to traditional branch family creation in Japan. In the upper class, family reproduction took place within stable patterns returning to stem family households, but the cycles were not as stable in the middle and lower classes. The lower the social class, the lower the ratio of patterns returning to stem family households. It was 31.6% in the middle class, nearly the same value as the reference value defined in this article. Thus, the pattern returning to stem family households was also the representative cycle of the middle class. The ratio of patterns returning to simple family households was also higher in the middle class than in the other social classes. In the middle class, there was a mixture of patterns returning to stem and returning to simple family households, but not the same level of stable cycle formation for patterns returning to stem family households as was seen for the upper class. While a strong binding force was applied to the rules of stem family household formation in the upper class, this binding force may not have been as strong in the middle class. Even if the binding force was equivalent, the social environment of the middle class may not have led to the realization of stem family household formation as easily as it did in the upper class. In the lower class, the ratio of patterns returning to stem family households was 28.9%, just slightly below the threshold value. From our observations of household composition cycles, we are unable to verify that the stem family household was representative of the household composition in the lower class. The ratio of patterns returning to simple family households was also lower in the lower class than in the middle class, and there were many diverse types of household composition transitions that could not be classified into a specific pattern. Of the various household composition in the lower class, the ratio of patterns returning to stem family households was indeed the highest, but this ratio is still low when compared to the same ratios for other social classes. No other specific cycles exist to replace the pattern returning to stem family households. Even if the members of the lower class had intended to achieve some specific pattern, various difficulties may have made it impossible for them to realize the pattern, forcing each household to select a household composition that corresponded to the conditions at a given time.

329

12:27:16:11:09 Page 329 Page 330

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

4. SUCCESSION AND INHERITANCE From the analysis of household cycles, we discovered cycles in the patterns returning to stem family households in the Aizu mountainous district during the late Tokugawa period. Deep differences existed, however, depending on the social class. Since the ratio of stem family households in the lower class was below the threshold limit set for this article, the real representativeness of the stem family household could not be verified. Therefore, we would next like to study further the transmission system in the stem families. If the ideals of ie existed in the farming society of the late Tokugawa period, then such ideals would have determined who would inherit the family estate. The most common case would undoubtedly be a succession of the family estate by the eldest son. We can also expect to see some types of rules related to the succession of family estates. Previous research has shown that succession by the eldest son did occur in the farming societies of the Tokugawa period, but various other types of people were able to succeed. Historians emphasize succession by the eldest son, while historical demographers emphasize heir diversification (Hayami, 1983; Cornell, 1981; Narimatsu, 1985, 1992; Okada and Kurosu, 1998; Okada, 2002). Keeping both of these concepts in mind, let us examine whether succession by the eldest son also occurred in the farming societies of the Aizu mountainous district as we attempt to elucidate family strategies. During the period studied (1752–1859), there were 597 opportunities for family estate transfer in the Aizu mountainous district (Table 16.8). The household estate was transferred from one household head to a new household head in 513 of these cases (85.9%); branch households were formed in 22 cases, resulting in a collateral family (3.6%); arrangements could not be made for an heir (or, preparations were intentionally left undone), in 62 cases (10.3%) resulting in a discontinuance of the household (becoming extinct). For every 10 family estate transfers, 8.5 were from the old household head to a new household head (heir). Since transfer of family estates took place under fairly stable conditions in the Aizu mountainous district, apparently a process applied to the decision about the person who would become the estate heir. In this process, we can expect to read family strategies.

Table 16.8 Opportunity for transfer of household structure by reason and social classes in four villages of Aizu mountainous district (1752–1859)

Transfer Collateral familya Extinct family Total

%N%N%N% N

Upper class 24.12 144 1.84 11 0.34 2 100.00 157 Middle class 45.90 274 1.51 9 5.86 35 100.00 318 Lower class 15.91 95 0.34 2 4.19 25 100.00 122 All 85.93 513 3.69 22 10.39 62 100.00 597

Sources: see note 2. a The result of branching out.

330

12:27:16:11:09 Page 330 Page 331

POPULATION AND HOUSEHOLD DYNAMICS

4.1. Reasons for Headship Change Let us now take a further look at the 513 cases of transfer patterns in which the family estate was transferred to a new household head. We shall refer to the household head who transfers the family estate to a new household head as the household head and to the new household head as the heir. The types of estate transfer can roughly be classified into three categories: (1) those that happened during the lifetime of the household head (hereafter referred to as ante mortem transfers), (2) those after the death of the household head (postmortem transfers), and (3) transfers that occurred for an unknown reason (Table 16.9). The ratio of antemortem transfers was 53.2% and of postmortem transfers, 40.3%. A difference of more than 10% existed between the two types of transfers. Apparently it was a custom to transfer family estates while the household head was still alive, although the prototype of an antemortem transfer was the household head handing over the family estate to the heir and then retiring from active life. Our observations of historical materials, however, reveal antemortem transfers occurring for various other reasons such as marriage, divorce, disappearance of the household head, or the formation of a branch household. In this article, the expression ‘retire from active life’ does not refer to any official system of retirement; it only indicates that the household head transferred the family estate to the heir but then continued living together in the same household. The term “disappearance” indicates that the household head was actually missing. People disappeared rather frequently in those days; records show, for example, they were called back by the Tokugawa government, and those who had abandoned their homes later made requests to return home.5 Therefore, antemortem transfer occurred for three reasons: (1) the household head retired from active life but continued to live in the same household, (2) the household head married or left home (for marriage, divorce or branching), and (3) the household head disappeared. While marriage and leaving home accounted for 5.1% and disappearance for 2.2% of the antemortem transfers, the household head’s retirement accounted for 92.6% of them, but for only 49.3% of all family estate transfer patterns (Table 16.9). The ratio of postmortem transfers was

Table 16.9 Reason for transfer of household structure in four villages of Aizu mountainous district (1752–1859)

Reason All Upper class Middle class Lower class

%N%N%N%N

Antemortem transfer 53.22 273 54.17 78 54.74 150 47.37 45 (a) Predecessor retirement 49.32 253 51.39 74 50.36 138 43.16 41 (b) Marriage/Leaving home 2.73 14 1.39 2 3.28 9 3.16 3 (c) Disappearance 1.17 6 1.39 2 1.09 3 0 1 Postmortem transfer 40.35 207 42.36 61 38.69 106 42.11 40 Unknown reason 6.43 33 3.47 5 6.57 18 10.53 10 Total 100.00 513 100.00 144 100.00 274 100.00 95

Excluding branching out and extinction. Sources: see note 2.

331

12:27:16:11:09 Page 331 Page 332

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

40.3%—not a great difference compared with transfers upon the retirement of the household head. Since postmortem transfers would also have included cases in which the head died suddenly at a young age, postmortem transfers were not limited to cases in which the head served in that position until dying at an old age. In the O¯¯ u region, there was a concentration of folkways in which no retirement system existed. “Succession following the death of the previous head” was prevalent (O¯¯ take, 1982, p. 330). From our studies, however, we have not found this to be the customary practice among the farming families in the Aizu mountainous district; rather an antemortem transfer was highly probable. Now, let us look at the reasons for estate transfer by social class (Table 16.9). In all of the social classes, the frequency of antemortem transfer was higher than that of postmortem transfer. Although antemortem transfer was more frequent for the middle class, little difference existed in the ratios between the middle and upper classes. Looking for specific reasons for antemortem transfer, we see that retirement was frequent in the upper class but decreased with the level of social class. Between the upper and lower classes, the difference in the ratios of transfer to retirement was approximately 8%. In the middle class, since the ratio of transfer for marriage or leaving home was higher than for the upper class, the ratio of antemortem transfers in the middle class was also higher. The ratio of postmortem transfer was highest in the upper class and lowest in the middle class, apparently because the ratio of transfers for unknown reasons was lower for the upper class than for the other groups. When comparing only the ratios of transfer to retirement and postmortem transfer in each social class, the ratio of postmortem transfer for the lower class was higher than the other classes, however. From this, we may be able to discover one of the family strategies applied to the transfer of family estates: using retirement from active life to transfer the family estate to a prepared heir in a planned manner. The higher the social class is, the higher is the ratio of planned transfer. When carefully planned, strategic family estate transfers were implemented, the planning had included the selection of the heir and the timing of the transfer. At this stage, however, it is impossible to judge whether the fact that the ratio of planned transfer decreased along with the social class level was the result of a lack of planning or if some other type of family strategy was being used in the lower classes. In the middle and lower classes, antemortem transfers also included a few cases for reasons other than retirement from active life. For example, some in the middle class may have given preference to marriage or leaving home over the maintenance of the family estate, and in the lower class, postmortem transfer may have been the strategically preferable choice.

4.2. Who Was the Heir Table 16.10 shows the position of the heir in relationship to the household head. Some explanation is needed concerning the “legitimate son” category used in this table. Here categories such as “only son,” “eldest son,” and “second or later son” do not necessarily indicate birth order, but the order of male siblings at the time of the succession of the family estate.6 Thus, the ranking of “eldest son” in this table does not necessarily indicate the ﬁrst born male, but the eldest legitimate son at the time of succession. This also applies to the category “only son,” which refers

332

12:27:16:11:09 Page 332 Page 333

POPULATION AND HOUSEHOLD DYNAMICS

Table 16.10 Relationship of successor to predecessora in four villages of Aizu mountainous district (1752–1859)

(1) Male 89.47 459 Legitimate son Only son 43.66 224 Eldest son 13.65 70 Second or Later son 0.78 4 Subtotal 58.09 298 Husband 0.58 3 Father 2.53 13 Grandchild 2.73 14 Subtotal 5.85 30 Son-in-law 12.48 64 Adopted son 9.75 50 Subtotal 22.22 114 Older brother 0.39 2 Younger brother 2.14 11 Nephew 0.78 4 Subtotal 3.31 17 (2) Female 10.53 54 Wifeb 7.80 40 Daughter 1.17 6 Mother 1.56 8 Subtotal 10.53 54 Total 100.00 513

Sources: see note 2. a See Hayami (1992, Table 11–2). b Widow in the case of postmortem transfer. Female siblings are ignored here.

not only to male heirs who had no siblings from the time of birth until the time of succession, but also to male heirs who may have had a male sibling earlier in life but were the only son at the time of succession. Accordingly, there is adequate possibility that people ranked in the “only son” category may have actually had male siblings, but prior to the time of family estate succession, those male siblings left the family for reasons of adoption, marriage, etc. In this table, female siblings are ignored. From this table, we can understand that the heir possessed certain characteristics. First of all, nearly all of the heirs were male, approximately 90%. Secondly, the ratio of heirs who were legitimate sons was not as high as one might expect (58%). The third characteristic was a fairly unexpected ﬁnding: among legitimate sons, the ratio of eldest sons with a younger brother living in the same household at the time of succession was not as high as expected, while only sons with no male siblings at the time of succession had the highest ratio. Of the 298 cases of succession by a legitimate son, he was the only son at the time of heirship succession in 224 of the cases (75.1%), and there were only 70 cases of succession by the eldest son (23.4%). Furthermore, there were only four cases (0.7%) of succession by a

333

12:27:16:11:09 Page 333 Page 334

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

second or younger son who had an older sibling at the time of heirship succession. In all but one of these four cases, this older male sibling remained single for his entire life. Perhaps there were some circumstances that made it impossible for the older brother to succeed to the headship.7 The fourth characteristic is the diversity in the relationship of the male heirs to the household head. While female heirs were chosen from a limited range of relationships, males could succeed even if the relationship was more distant, even as a collateral (O¯¯ to, 1996). These observations have clarified that the ratio of succession by the eldest son was unexpectedly low, the ratio of succession by an only son was higher, and that even women could succeed, although they were the exception. Although male heirs came from a diverse range of relationships to the household head, female heirs came from a limited range. Was there any correlation between the reason for succession and the gender and relationship of the heir? Since there are few cases of succession due to marriage, leaving home, or disappearing, it is difficult to use them for comparisons. Therefore, we compared transfers resulting from the retirement of the household head with headship succession when the head died (Table 16.11). The most prominent characteristic for headship succession by females was largely the death of the head. In only one exceptional case (0.4%) of succession did a female replace a retiring household head: in that case, a man had married into the family and transferred the headship to his mother-in-law. Two years later, the mother-in-law transferred the headship back to her son-in-law. In the case of succession when the head died, however, the ratio of female heirs was 23.7%, a high ratio indeed. Approximately ten times more women succeeded when heads died than when they retired. When a household head was preparing for retirement and considering an heir, women would not have been included among the possible candidates. A study of Niita mura, Nihonmatsu domain, yielded the same results (Okada, 1998). In the case of succession when the head died, women might also be candidates. Still the table shows that women were not chosen from a wide range of relationships to the household head but rather were limited to the widow, daughter, or mother of the head. In the case of males, no differences were seen concerning the relationship to the household head when succession was due to death or retirement. In both cases, the heir was selected from a fairly broad range of relatives. The ratio of transfers to a collateral relative or a non-kin was high in the case of succession due to retirement. In the case of a legitimate son, comparing retirement-transfer and headship succession at death reveals a difference of approximately 3%, which is extremely low. Transfers to a daughter’s spouse or adopted son, however, occurred more than twice as often in the case of retirement than it did when the head died. The ratio of transfers to a collateral relative—a younger brother or nephew—was also higher in the case of retirement-transfer. When headship was transferred from the household head following his retirement, the candidates included a wide range of male relatives. In the case of headship succession when the head died, women were also included as candidates, as we saw. To make an heir of somebody who was not a stem kin, a family strategy of household head retirement had to be used. In case of transfer to a son-in-law or adopted son, approval from the kinship organization (do¯zokudan) was necessary;

334

12:27:16:11:09 Page 334 Page 335 Unknown (%) Unknown 1859) – Transfer. All (%) Transfer. Postmortem Disappearance (%)Disappearance Postmortem in four villages of Aizu mountainous district (1752 a 2). Sources: see note 2. Sources: 2). – Marriage/Leaving Home (%) Widow in the case of postmortemWidow transfer. b See Hayami (1992, Table 11 Table (1992, See Hayami a Predecessor Retirement (%) 273 253 14 6 207 33 1.10 0 0 50 17.39 3.03 Antemortem All (%) Transfer. Relationship of successor to predecessor by reason for transfer Relationship Table 16.11 Table b DaughterMotherSubtotalTotalN 0 0.73 1.83 100.00 0.40 0.40 0 100.00 100.00 7.14 7.14 0 100.00 0 50 0 100.00 22.71 2.90 100.00 2.42 6.06 0 3.03 Transfer(1) MaleLegitimate son Antemortem Only sonEldest son sonSecond or Later SubtotalHusband 98.17 1.47FatherGrandchild 42.86 13.92SubtotalSon-in-lawAdopted Son 58.24 99.60 1.58Subtotal 1.10Older Brother 44.66 2.56 13.44 BrotherYounger 2.93Nephew 17.95 10.99 6.59Subtotal 59.68 92.86(2) Female 2.93 0.37Wife 1.19 28.94 28.57 0 28.57 2.77 1.58 19.37 10.67 5.53 57.14 1.10 50.00 4.40 1.83 3.16 30.04 0 0 14.29 14.29 0 0 0 0 14.29 0 1.19 4.35 0.40 14.29 77.29 0 0 7.14 33.33 16.67 33.33 43.96 0 12.56 0 93.94 7.14 7.14 0 16.67 0 0 56.52 48.48 0 0 18.18 1.45 8.21 4.83 66.67 3.38 0 14.49 50 0 0 0 6.28 6.06 9.09 0.97 6.06 15.15 0 6.06 0 0 22.71 1.45 0.48 3.03 3.03 6.06 6.06 0

12:27:16:11:09 Page 335 Page 336

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

if the household was a branch household, approval was needed from the head of the main family household (honke). For someone who was not a stem kin to become the heir of the household, approval was needed from the people of the village. The new heir had to be approved so that he could join the ranks of the family group structure and village organization. Therefore, a strategy of household head retirement may have been used so that the household head could provide the necessary backup and support for the potential heir.

4.3. Socio-differentiated Strategies Did the relationship of the heir to the household head differ according to social class? Since our sample size is very small, detailed relational information like that in Tables 10 and 11 cannot be produced for this type of study when examining individual social classes. In Table 16.12, the male heirs divided into the categories “legitimate sons,” “stem kin,” “collateral relatives,” and “sons-in-law/adopted sons”. Because of the small number of female successors, they were not divided into smaller categories. This table shows no gender-based differences according to social classes. At any social level, female heirs accounted for approximately 10% of all heirs. Some socio-differentiation appears concerning male successors, however. Succession by a legitimate son occurred at a ratio of 62.5% in the upper class, 58.0% in the middle class, and 51.8% in the lower class. The difference between the upper and middle classes was not great, but the difference between upper and lower classes exceeded 10%. In the lower class, the ratios of succession by sons-in-law/adopted sons and by collateral relatives were higher than those ratios in the upper and middle classes. Now let us look at these differences according to the various reasons for transfer (Table 16.13). No case of headship succession to a female heir while the head was alive—antemortem transfer—appeared in the middle class. In the upper class, although we do see cases of headship succession to a female heir while the head was alive, the ratios of succession to stem kin, collateral relatives, and sons- in-law/adopted sons were lower than those ratios in the other classes. In the lower class, the ratio of succession of a legitimate son was 5–6% lower than this ratio in the upper and middle classes, and the ratio of succession to a collateral relative was also lower than that ratio in the other two classes. Conversely, the ratio of

Table 16.12 Relationship of successor to predecessor by social class for four villages of Aizu mountainous district (1752–1859)

All (%) Upper class (%) Middle class (%) Lower class (%)

1. Male successor 89.47 88.89 89.78 89.47 Legitimate son 58.09 62.50 58.03 51.58 Stem relatives 5.85 5.56 5.84 6.32 Collateral relatives 3.31 2.08 3.65 4.21 Son-in-law/Adopted son 22.22 18.75 22.26 27.37 2. Female successor 10.53 11.11 10.22 10.53 Total 100.00 100.00 100.00 100.00 N 513 144 274 95

Sources: see note 2.

336

12:27:16:11:09 Page 336 Page 337 1859) – (1752 Postmortem transfer Postmortem Predecessor retirement All(%) Upper (%) Middle (%) Lower All (%) Upper (%) Middle (%) Lower All (%) Upper (%) Middle (%) Lower (%) (%) (%) 273 78 150 45 253 74 138 41 207 61 106 40 Relationship of successor to predecessor by reason for transfer in social class four villages Aizu mountainous districtRelationship Table 16.13 Table Class Male successor1. Legitimate sonStem relatives relativesCollateral Antemortem transfer 98.17 sonSon-in-law/Adopted 28.57 96.15 successor Female 2. 58.24 (%)Total 28.21 100.00 4.40 58.97N 6.59 95.56 1.83 28.67 3.85 59.33 99.60 5.13 31.11 3.85 53.33 100.00 30.04 4.67 100.00 59.68 7.33 97.56 100 28.38 4.44 62.16 6.67 77.29 0 30.43 100 4.35 58.70 78.69 31.71 4.44 5.53 58.54 75.47 14.49 4.05 56.52 0.40 5.41 80.00 6.56 100 65.57 5.07 100 5.80 16.04 54.72 2.44 0 4.88 22.50 47.50 100 1.45 4.83 0.00 0 100 6.56 2.44 0.94 3.77 100 22.71 5.00 100 21.31 5.00 24.53 100 20.00 100 100 100

12:27:16:11:09 Page 337 Page 338

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

succession of sons-in-law/adopted sons was higher in the lower class than in the other two classes. If we analyze headship succession while the head was alive, focusing only on those successions that occurred due to household head retirement, we see that succession by a female heir occurred only in the lower class. In such cases there were no female heirs in the other two classes. The proportion of transfers to legitimate sons was the highest in the upper class with a clear decrease in this ratio as the level of social class decreased. In the case of headship succession when the head died—postmortem transfer— the ratio of female heirs was just above 20% in all classes. At its zenith in the middle class, it reached approximately 25%. There are clear differences among the classes in the types of male heirs. In the upper class, the succession of a legitimate son accounted for more than 80% of all transfers to males, while the ratio of transfers to collateral relatives was zero. The ratio of transfers to a legitimate son decreased with the social class level down to 47.5% for the lower class, much lower than in the upper class (Table 16.13). In a trend opposite to that seen for legitimate sons, the ratio of transfers to sons-in-law/adopted sons increased as the social class level decreased. This ratio was 6.5% for the upper class and 22.5% for the lower class. From this, it can be clearly seen that in both cases of headship succession— ante- and postmortem—there was a stable degree of heirship transfer to legitimate sons in the upper class. As the social class level decreased, the possibility increased for sons-in-law/adopted sons (and in the case of headship succession when the head died, females) to be included as candidates for heir. A particularly large difference is seen between the upper and lower classes. The ratio of transfer to a legitimate son was lower in the lower class, but transfers to sons-in-law/ adopted sons made up this difference. There may have been fewer legitimate sons in the lower class, and even if a legitimate son existed, he might have left home to marry or be adopted, leaving no legitimate son in the household at the time of succession. In some cases, the headship was taken by a son-in-law or a male adopted from another household. There may also have been cases in which the heirship was given to a son-in-law or adopted son even though a legitimate son was present. Clearly, the possibility for heirship succession of a legitimate son increased with the social class. We cannot claim that these observations reveal solid characteristics related to heirship succession, but at least they reveal that a single rule did not exist for heirship succession in the Aizu mountainous district. An heir did not have to be chosen from a rigid range of candidates; instead there was a set of extremely loose, flexible rules that allowed for considerable leeway in heir selection.

4.4. Flexible Models of Family Succession We will now attempt to summarize the conditions of family succession in the Aizu mountainous district. Family succession while the head was alive and when he died. Succession while the head was alive can be further divided into succession resulting from (1) the household head’s retirement or (2) the marriage, departure or disappearance of the household head. Reasons for succession when the head retired and when he left home can be interpreted from the age of the household

338

12:27:16:11:09 Page 338 Page 339

POPULATION AND HOUSEHOLD DYNAMICS

head at the time of transfer, as well as the presence or absence of females in the range of heir candidates. When a household head retired, females were not included among the range of heir candidates, but when he married out, left home or disappeared, females were included as heir candidates. The diﬀerences between transfers due to the household head’s retirement and his death included the likelihood of a female acting as heir and the degree of possibility for succession by a range of male relatives that included collateral relatives. Succession was not governed by a single rule dictating that the heirship must go to the eldest son. In the upper class, there was a more stable level of heirship succession by a legitimate son than occurred in the middle and lower classes. However, this “legitimate son” was not necessarily the eldest son in male birth order in the family, but was the eldest son (or only son) remaining in the household at the time of succession. Also, compared to a legitimate son, a son-in-law or adopted son was generally older at the time of succession and served as household head for a shorter time. Accordingly, the characteristics of family succession depended on whether the heir was a legitimate son or someone else. In the upper class, there was a fairly stable application of the orthodox rule of heirship succession by the eldest son. Still, families were not strictly bound to this rule, which, in fact, was applied rather loosely. Among the farming families of the Aizu mountainous district, strategies for succession took into account conditions needed to maintain the household. This tendency grew stronger as the social class level decreased. The fact that rules related to heir selection had little binding force, even if they did exist, was the result of demographic obstacles and hardship that made it diﬃcult for rural families to plan for an heir.

5. CONCLUSION We first examined population trends to better understand household characteristics and family reproduction system in a highland area of northeastern Japan, the Aizu district. In comparison with other regions of Japan, marriage was early and fertility was low. The demographic specificities of the mountainous region under study will be taken into consideration when discussing further the timing of household reproduction. We have identified types, frequencies and durations of household cycles over time during the late Tokugawa period. Heirship succession by eldest sons did occur, forming lineal families. Reproduction was regulated by ideals of the Japanese house system, the ie. Families were therefore incorporated into ie systems. Diversity has also been found in the cycle of household composition and in the possibilities of heirship. The family system was not strong or deeply rooted but rather a flexible one that allowed for a fair amount of leniency and versatility. The results of our observations of family succession according to socioeconomic class have shown a fairly large differentiation, despite the small sample size. We can conclude that for the upper and middle classes of the Aizu mountainous district in the late Tokugawa regime families possessed ideals of ie, and ie systems were created. Still, we also saw considerable diversity among these ideals of ie and ie systems with a broader scope of diversity in the middle rather than in the upper class. This scope of diversity broadens even further in the lower class,

339

12:27:16:11:09 Page 339 Page 340

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

and we can conclude that the lower class families were not regulated by ideals of ie, nor were they bound to ie systems.

NOTES

1. Demographic trends for Ishibushi mura, To¯nosu mura, and Kuwanohara mura have been analyzed by Kawaguchi. These data were used to clarify the demographic trends of the Aizu mountainous district for this article. For detailed analyses of births and systems of birth control not dealt with in this article, see Kawaguchi (1998). 2. Sources used: Kanaizawa mura, Aizu gun, Mutsu Province, Shu¯ mon Aratame Nimbetsu Iebetsu Cho¯, 1755–1837 (2 years missing); To¯nosu mura, Aizu gun, Mutsu Province, Shu¯ mon Aratame Nimbetsu Iebetsu Cho¯, 1790–1859 (3 years missing); Ishibushi mura, Aizu gun, Mutsu Province, Shu¯ mon Aratame Nimbetsu Iebetsu Cho¯, 1752–1812 (5 years missing); Kuwanohara mura, Aizu gun, Mutsu Province, Shu¯ mon Aratame Nimbetsu Iebetsu Cho¯, 1752–1834 (7 years missing). 3. When looking at fertility rates, the characteristics of the Shu¯ mon Aratame Cho¯ (SAC) lists make it necessary to multiply the number of births shown there by a factor of 1.20–1.25 to obtain the actual number of births. For this article, this kind of adjustment will be made when calculating the total fertility rate (TFR). We believe that it is sufficient to make a comparison within the Aizu mountainous district. 4. Social class was determined by calculating the average kokudaka of each household during the period studied (Hayami et al., 2004). The three levels were upper class, middle class, and lower class. The average kokudaka of the upper class was 5 koku or higher, of the middle class, at least 2 koku but less than 5 koku, and the lower class 0 to less than 2 koku. 5. See Fukushima Ken Shi Edition (1971, p. 830) and Nihonmatsu Shi Edition (1982, p. 922). Narimatsu has analyzed the Nimbetsu Aratame Cho¯ of the Niita mura of the Nihonmatsu domain and concluded that disappearance was not a rare event in the farming village life of the corresponding time (1992, pp. 201–203). 6. Opinions differ about the concept of the eldest son or only son at the time of succession (for example, Tsubouchi, 2001, p. 4). Family sociology and family history emphasize the importance of birth order and oppose this concept. In the farming society of the Tokugawa period, however, the death rate was fairly high up to age 10. As such, we believe that birth order need not be emphasized. A child with a long life span would have most likely been considered to be more important than a first-born child who died at an early age. In cases in which there were both an older and younger brother, the older brother became the adopted son of another household and the younger brother became the heir, the reasons for this must, of course, be pursued, and we are in no way attempting to negate the necessity of looking back as far as the time of birth to examine the presence of siblings. 7. In one case, the historical materials had ended, and we do not know what happened after that time. The household head had a spouse and children, however, and his older brother was still single as of the final year of the period studied.

REFERENCES Berkner, L. K. (1972). The stem family and the developmental cycle of the peasant household: An eighteenth-century Austrian example. American Historical Review, 77, 398–418. Cornell, L. L. (1981). Peasant Family and Inheritance in a Japanese Community, 1671–1980: An Anthropological Analysis of Local Population Registers. Unpublished Ph. Thesis, Johns Hopkins University. Fukushima Ken Shi Edition. (1971). Fukushima Ken Shi. Kinsei, vol. 2 (p. 1). Hammel, E. A., & Laslett, P. (1974). Comparing household structure over time and between cultures. Comparative Studies in Society and History, 16(1), 73–109. Hayami, A. (1982). Kinsei O¯¯ u Chiho¯ Jinko¯ no Shiteki Kenkyu¯ Joron. Mita Gakkai Zasshi [Mita Journal of Economics], 75(3), 70–92. Hayami, A. (1983). The myth of primogeniture and impartible inheritance in Tokugawa Japan. Journal of Family History, 8, 3–29.

340

12:27:16:11:09 Page 340 Page 341

POPULATION AND HOUSEHOLD DYNAMICS

Hayami, A. (1987). Another Fossa Magna: Proportion marrying and age at marriage in late nineteenth century Japan. Journal of Family History, 12, 57–72. Hayami, A. (1992). Kinsei No¯bichiho¯ no Jinko¯, Keizai, Shakai [Population, economy and society early modern Japan: A study of the No¯bi region]. Tokyo: So¯bunsha. Hayami, A. (2001a). The historical demography of pre-modern Japan. Tokyo: University of Tokyo Press. Hayami, A. (2001b). Rekishijinko¯gaku de mita Nippon: Tokyo: Bungeishunju¯ . Hayami, A., & Kurosu, S. (2001). Regional diversity in demographic and family patterns in preindustrial Japan. Journal of Japanese Studies, 27(2), 295–321. Hayami, A., Saito, O., & Toby, R. P. (Eds.) (2004). The economic history of Japan: 1600–1990. Emergence of economic society in Japan, 1600–1859, vol. 1. Oxford: Oxford University Press. Kadokawa Dictionary of Japanese Place Names. (1991). Kawaguchi, H. (1998). 17–19 Seiki no Aizu Minamiyama Okurairi-ryo¯ niokeru Jinko¯hendo¯ to Shussho¯seigen [Population trend and control in a rural area in north-eastern Japan from the 17th to the 19th c.]. Rekishichiri Gaku, 191, 5–25. Kinoshita, F. (1990). To¯hoku Chiho¯ Ichi No¯son niokeru Setai no Hensen 1760–1870 [Changes in household size and composition in a Japanese village, 1460–1870]. Minzokugaku Kenkyu, 55, 1–20. Kinoshita, F. (1995). Rekishijinko¯gaku no Seimeihyo. In Yamaguchi K., et al., (Eds.), Seimeihyo¯ Kenkyu¯ (pp. 152–178). Tokyo: Kokinshoin. Kinoshita, F. (2002). Kindaika Izen no Nippon no Jinko¯ to Kazoku-Ushinawareta Sekai karano Tegami. Kyoto: Minerva Shobo¯. Kito¯, H. (1981). Kinsei No¯son niokeru Kazoku Keitai no Shu¯ kiteki Henka. Jo¯chi Keizairon Shu¯ , 27(2 & 3), 7–22. Koyama, T. (1959). Kazokukeitai no Shu¯ kiteki Henka. In S. Kitano, & Y. Okada (Eds.), Ie¯ Sono Ko¯zo¯ Bunseki (pp. 67–83). Tokyo: So¯bunsha. Laslett, P. (1972). Introduction: The History of Family. In P. Laslett, & R. Wall (Eds.), Household and family in past time (pp. 1–89). Cambridge: Cambridge University Press. Murdock, G. P. (1949). Social structure. New York: Macmillan (Japanese translation: 1986). Narimatsu, S. (1985). Kinsei Tohoku No¯son no Hitobito: O¯ shu Asaka-gun Shimomoriya- mura [People in a northeastern agricultural village in early modern Japan: The village of Shimomoriya, Asaka Country, O¯¯ u Region]. Kyoto: Minerva Schobo¯. Narimatsu, S. (1992). Edojidai no Tohoku No¯son: Nihonmatu-han Niitamura. A Northeastern Village in Edo period: Niita, the Nihonmatsu Domain. Tokyo: Do¯bunkan. Nihonmatsu Shi Edition. (1982). Nihonmatsu Shi, vol. 4. Okada, A. (1998). Jo¯togata Koshu no Tokucho¯ [Characteristics of headship succession: The case of Niita village]. Teikyo¯ Shakaigaku [Teikyo¯ Journal of Sociology], 11, 109–135. Okada, A. (2000). Kinsei No¯min Shakai niokeru Setaiko¯sei no ‘Cycle’: Nihonmatu Nikason no Shiryo¯ wo mochiite [The cycle of household structure in early modern peasant society]. Shakaigaku Hyo¯ron [Japanese Sociological Review], 51(1), 136–152. Okada, A. (2001) Tokugawa Ko¯ki Sonrakushakai niokeru Kakei no Keisho¯ to Setai: Minamiyama Okurairi-ryo no Shumon-aratamecho¯ wo to¯shite. Unpublished doctoral dissertation, Japan. Okada, A. (2002). Kinsei Nominshakai niokeru Katoku no Keisho¯ to Sono Senryaku: Mutunokuni Asaka-gun Shimomoriyamura Ninbetu-aratamecho¯ wo Chu¯ shin

341

12:27:16:11:09 Page 341 Page 342

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

toshite. In A. Hayami (Ed.), Kinsei Iko¯ki no Kazoku to Rekishi (pp. 101–126). Kyoto: Minerva Shobo¯. Okada, A., & Kurosu, S. (1998). Succession and the death of the household head in early modern Japan: a case study of a northeastern village, 1720–1870. Continuity and Change, 13(1), 143–166. O¯¯ take, H. (1982). Nihon Shakai no No¯min kazoku [Peasant Families in Japanese Society] (1st ed.). Tokyo: So¯bunsha (1962). O¯¯ to, O. (1996). Kinsei No¯min to le-Mura-Kokka [Peasants in early Modern Japan and Family-Village-State]. Tokyo: Yoshikawa Ko¯bunkan. Saito¯, O. (1999, September). The developmental cycle of the stem-family: A Japanese perspective. Paper presented at the Majorca Conference, Majorca, Spain. Suzuki, E. (1968). Nihon No¯son Shakaigaku Genri. Suzuki Eitaro¯ Chosakushu¯ , vols. 1 & 2. Tokyo: Miraisha. Toda, T. (1937). Kazoku ko¯sei [Family Structure] (rev. ed.). Tokyo: Shinsensha (1982). Todd, E. (1990). L’invention de l’Europe. Paris: Seuil. Tsubouchi, R. (2001). Keisho no Jino¯shakaigaku. Kyoto: Minerva Shobo¯.

342

12:27:16:11:09 Page 342 Page 343

PART V EPILOGUE

12:27:16:11:09 Page 343 Page 344

12:27:16:11:09 Page 344 Page 345

First published in Journal of Family History, vol. 12, no. 1–3, pp. 57–72, 1987.

17 Another Fossa Magna: Proportion Marrying and Age at Marriage in Late Nineteenth-century Japan

INTRODUCTION apan has been regarded as a country consisting of a single race, single language Jand homogeneous culture. Compared with China, India or with the U.S. and the USSR, this may be true. In the premodern period, however, Japan exhibited many local differences in its social life. These local variations are particularly important for an understanding of Japan’s history and modernization, and even contemporary Japan, since these local differences were utilized as a resource in the modernization of Japan. Economically, local differences resulted from local geographic conditions and they intensified local comparative advantages by maximizing local specialization in production. Culturally, during the Tokugawa period (1603–1868), when peace was maintained for more than two centuries nationwide, local differences stimulated people’s curiosity, and many local gazet- teers and studies were compiled. Travel became an industry. Among those regional variations, differences in social structure and in family life still remain unexplored. The present study investigates regional differences in age at marriage and proportions marrying, using the earliest national statistics available from the early Meiji period on.

AGE AT MARRIAGE Age at marriage is an important variable in both historical demography and family sociology. Age at marriage, particularly age at first marriage for women, determines population size through the number of births. Of course, birth control can affect the relationship between age at first marriage and number of offspring, and, regardless of the presence or absence of birth control, late marriage will result in fewer births. It is an important issue in social anthropology as well. Age at marriage interacts with issues of inheritance and family succession. Early marriage may speed up the family life cycle and increase the number of families consisting of three or more generations. In the case of Japan, there have not been any noteworthy historical studies of age at marriage. As a result, the relationship between age at marriage and population

345

12:27:16:11:09 Page 345 Page 346

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

change, inheritance and family succession, have yet to be analyzed. The reasons for this seem to be twofold. First, since birth control appears to have been rather commonly practiced, early marriage did not always result in an increase in population. Second, parents and children usually lived together until the middle of the twentieth century. This naturally provided for family succession without any particular need to consider age at marriage. Before the first Japanese census of 1920, statistics on age at marriage were inadequate. This shortage in historical materials frustrated efforts to study the subject. Nevertheless, population records from the Tokugawa period can be used to study age at marriage and other demographic measures for small populations at the village and town level. The Tokugawa records are problematic in a number of ways: it is impossible to gather data on more than a few hundred individuals at a time, and it is difficult to estimate how representative the places studied were of the wider region or of Japan as a whole. In the present study, the author will not use the aforementioned Tokugawa documents and will rely, instead, on records gathered by the Meiji government in the latter half of the nineteenth century. During the Meiji period (1868–1912) many of the social practices of the Tokugawa period were still present. We can, therefore, make inferences concerning the country as a whole in the Tokugawa period from Meiji-period documents. By using such documents, we can explore historical and spatial variation in age at marriage among many scattered communities and prepare more detailed local studies based on the registers of the Tokugawa period. We know that Tokugawa Japan was a society which had significant spatial and diachronic differences in age at marriage. By using Meiji-period documents, the author has been able to expand his investigation of several Tokugawa villages to a nationwide picture which may be multicolored. The following three cases (two of them from the author’s research) illustrate well these variations in the age at marriage. In Yokouchi village, Shinano Province, for example, the age at first marriage for females at the beginning of the eighteenth century was seventeen sai.1 (Traditionally, age was counted in Japan in the following manner: one sai at birth and another sai each new calendar year. A person born in the last month of a year was considered two years old [sai] by the next month, i.e. the first month of the next year.) In Yokouchi, age at first marriage for females gradually increased over time. By the middle of the nineteenth century, it had reached 22 sai. Until 1770, the population size of the village as a whole had been increasing, but it levelled off thereafter. It is clear that there was some relationship between age at marriage and population size in Yokouchi, even though it is not clear exactly what the nature of that relationship was. Comparing the number of births between wives in the two birth cohorts of 1651–1700 and 1801–1825, we find that number to be 6.4 and 3.8 respectively.2 In another case—Nishijo village in Mino Province—age at first marriage for the wives of the 1773–1835 cohort differed conspicuously according to landholding. In the upper class of peasants, age at first marriage was 21.5 sai while in the lower class of peasants, it was 24.5 sai. This age difference appears to come principally from the presence or absence of the experience of working away from the home (dekasegi). In the upper class, only 33 percent of female children who survived

346

12:27:16:11:09 Page 346 Page 347

ANOTHER FOSSA MAGNA

beyond eleven sai experienced dekasegi. In the lower class, 74 percent of the same group experienced dekasegi.3 Dekasegi usually began at age twelve or thirteen sai and continued for twelve to thirteen years in urban areas. Since women returned to their native village at age 24 or 25 sai for marriage, their age was naturally delayed relative to the age of those women who did not leave on dekasegi. The difference of three years in age at first marriage for upper- and lower-class peasant women was crucial. It is known through the historical demographic study of Nishijo that in order for the net reproduction rate (NRR) to exceed 1.0, women had to have more than 4.4 births per marriage. For this to happen, they almost had to marry before age 24 sai. Thus, in those households where women married later than 24 sai, a successor could not be easily guaranteed. Indeed, in Nishijo, there were no cases among the upper class of peasants where a successor could not be found during the period under study (1773–1869). But among lower-class peasants, when the head of the household died, a successor was not present 35 percent of the time and the family died out.4 In addition to such a demographic explanation, it should be pointed out that upper-class peasants had property and accordingly sought to find successors to the family estate. This resulted in the adoption of sons and daughters to be successors in the family. The difference of three years in the age at marriage was significant for another reason: women are most fertile between the ages of 21 and 25. Whether or not one marries during this period will have a major impact on the birth rate. The third case comes from a village study using a very good series of registers in Tohoku.5 The Tohoku district of northern Japan where this village is located experienced a sharp decline in population from the early eighteenth century on. Quite contrary to our expectations, however, the age at first marriage for females was surprisingly low. It was as low as 11.2 sai in the early eighteenth century. By the middle of the nineteenth century it had risen to 18.7 sai, still four years lower than in the other two cases from central Japan. Age at marriage was thus significantly different depending upon the place and time of investigation. The lack of comprehensive data for the Tokugawa period makes it impossible to analyze the variation in age at marriage for the country as a whole. But the documents compiled by the Meiji government after its establishment in the last decades of the nineteenth century make it possible to study age at marriage systematically.

DATA AND ESTIMATES Before analyzing them, it might be useful to make some observations about the nature of the data. The ﬁrst census in Japan was compiled in 1920—a rather late date among the developed countries of the world. Because the quality of population data before the 1920 census was so poor, it has been said that demographers have generally ignored issues concerning the Japanese population before 1920. However, it is an exaggeration to think that population data from before 1920 is without any value whatever. In 1872 the Meiji government initiated a household registration system which

347

12:27:16:11:09 Page 347 Page 348

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

required that all inhabitants register their household circumstances at an appropriate level of local administration. Thereafter, for reasons of marriage or migration, the register was amended and the government was able to count the population on the basis of this “Registered Population” (honseki jinko). The estimation problem in this “Registered Population” was created by that portion of the population which moved for a temporary period, the so-called kiryu jinko. According to the law, if persons moved for a period of longer than ninety days from their registered location, they were required to register as part of the kiryu population. This temporary migrated population was added to or subtracted from the registered population of each administrative unit, producing the genju jinko or residential population. However, many persons avoided the temporary registration requirement and, as a consequence, the figures for the residential population are not very reliable. After 1890, when the urbanization and industrialization of Japan began in earnest, the numbers of persons migrating locally in Japan increased dramatically and the reliability of the residential population figures slipped accordingly. Because of this, the government was unable to rely upon such demographic measures calculated at the local administration offices by the beginning of the twentieth century, and it used police population records created by the police independently to augment and revise the residential population figures, particularly in urban areas.6 Nevertheless, before 1890 or before Japan’s industrialization and urbanization had become noteworthy, the kiryu jinko or temporarily displaced population was not so large, and the honseki jinko was rather close to the actually residential population. As a result, the reliability of the pre-1890 population data may be rather high. Recently, bibliographic sources concerning the pre-1890 statistics have become more available, making it possible to cross-reference these materials with population records. This has generally increased the ease of use and reliability of such materials.7 The present study uses as its principal source the invaluable Table of House- holds and Population of Imperial Japan (Nihon Teikoku Minseki Kokohyo: hereafter NTMK), which was compiled first in 1886. It has been almost entirely overlooked as a source for the demographic study of Japan.8 Among its many virtues, the NTMK lists the population at each age for each prefecture in Japan and how many of those persons at each age were marrying, using as the basis the honseki jinko. Thus, the December 31, 1886, compilation provides the first countrywide statistics about proportions marrying by age for the prefectures of Japan. Since 1920, the published census does not count the population at each age, but rather aggregates the data in five-year age groups. The value of the NTMK, which records each age if it was counted by traditional sai, is quite significant. It is astonishing, therefore, that this valuable data source of honseki jinko (NTMK) has been entirely overlooked in demographic research so far. This may be a result of the absence of continuous uniform records from the termination of the honseki jinko register in 1897 and the commencement of the National Census in 1920. The intervening registers compiled every five years from 1898 to 1918, the Population Statistics of Imperial Japan (Nihon Teikoku Jinko Tokei: hereafter

348

12:27:16:11:09 Page 348 Page 349

ANOTHER FOSSA MAGNA

NTJT), contained diﬀerent information from that in the NTMK.9 It was almost impossible, therefore, to construct a systematic time series. In addition to the NTMK, which forms the major data source for this study, the Table of Households and Population for Towns (Tofu Meiyu Kokohyo: hereafter TMK), will be used as well.10 The TMK counted the urban population in 1884, and it is believed to provide highly reliable information. Finally, along with the NTJT of 1898, the 1899 Vital Population Statistics of Imperial Japan (Nihon Teikoku Jinko Dotai Tokei: hereafter NTJD), which connected age at marriage and proportion marrying, will also be used.11

PROPORTION MARRYING AND AGE AT MARRIAGE Even if we do not have information on age at marriage, by using the method published by John Hajnal in 1953, it is possible to estimate the age at marriage from the proportion of persons marrying.12 In his essay, Hajnal calculates the age at marriage from two different kinds of modern census data for the proportion of unmarried persons. Because the census data was calculated in five-year age groups and because Hajnal wanted to estimate changing age at marriage over time, the calculations involved in his effort were no easy matter. Unfortunately, Hajnal’s sophisticated technique is inappropriate for analyzing the single-year data from Meiji Japan. In an earlier study on Japan the author used a simple estimate which calculates the age at marriage when the proportion marrying reaches 50 percent.13 In this calculation I assumed that 100 percent of the population would marry, and that age at marriage was distributed evenly within a given five-year age group. But these assumptions are not always realistic. In Japan, although the proportion marrying was higher than in the West, it did not reach 100 percent at any age. Further, the age at marriage, particularly for women, was not distributed evenly within any five-year age group. An accurate distribution of ages at marriage for women would be skewed to the particular age with a long tail toward the right. By contrast, in the present study I have used data which provide information on both age and proportion marrying in each age, and after calculating the relationship between these figures, I have applied the results to the NTMK. Unfortunately, there is no single source for obtaining these two figures of age at marriage and proportion marrying in one calculation. But by using the 1898 NTJT, one can get results for persons married and unmarried in 1898 and combine this information with the distribution of age at marriage obtained from the 1899 NTJD. Both these sources aggregate data for the country as a whole. Here, however, because I want to find the relation between proportion marrying and age at marriage as a whole, I will ignore the local differences that might exist (see Figure 17.1). By combining the data from these two records, we can reach the following conclusions. First, the proportion marrying did not reach eighty percent at the highest. For men, it reached 78.8 percent at age 44, and for women, it reached 79.4 percent at age 37. Second, the distribution of age at marriage reached its peak at age 23 for men and age twenty for women. Third, the shape of the distribution of age at marriage for women is more peaked and sharper than that for men. About 38 percent of men marry between the ages of 21 and 25. About

349

12:27:16:11:09 Page 349 Page 350 Distribution Marriage of Age at (1899) Figure 17.1

12:27:16:11:09 Page 350 Page 351

ANOTHER FOSSA MAGNA

48 percent of women marry between 18 and 22. Because these ages at marriage include figures for remarriages, the distribution of ages at first marriage would be more concentrated in reality. Of course, the average age at marriage will depend upon the length of time that one uses to calculate age at marriage. But we know from prior analyses of Tokugawa population records that almost all first marriages for both sexes occurred before age 35 sai.14 By studying the NTJD and looking at persons who reach age 35, we calculate an average age at marriage for men of 25.7 and of 22.4 for women. Using the NTJT for obtaining figures for the proportion marrying at these ages, we find that 45 percent of men aged 26 were married, and 44 percent of women aged 22 are married. Although these figures are slightly higher than 40 percent, we can say that they resemble a half of the highest proportion marrying at any age for men and women. According to the results above, even if the average age at marriage cannot be calculated directly from the data on hand, if the proportion marrying at various ages is known, the age that corresponds to half of the highest proportion marrying is approximately equal to the average age at marriage.

NTMK OBSERVATIONS The NTMK records population data by prefecture for the number marrying and not marrying at each age on the basis of honeseki jinko. First, the proportion marrying in each prefecture for men aged 28 sai and for women aged 23 sai was calculated and placed in Table 17.1. The results can be summarized as follows. For both sexes, there is more than a 30-point difference between the highest and lowest proportions. In particular, the difference for women is much greater than for men. The higher proportions can be found in the northern and eastern parts of Japan (except Tokyo and Kanagawa where urbanization was significant), and the regions east of and including Toyama, Nagano, and Shizuoka.15 In these areas, except for Toyama for men and Yamanashi for women, the proportion of persons marrying was well over 55 percent. By comparison, to the west of these regions, except for a few prefectures where the proportions were slightly higher than the countrywide average, the proportion marrying was well under 55 percent.

Table 17.1 Proportion Marrying by Prefecture (1886)

DISTRICT MALE (age at 28 sai) FEMALE (age at 23 sai) Prefecture population marrying proportion population marrying proportion

HOKKAIDO 1604 715 44.6% 1932 856 44.3% TOHOKU Aomori 4299 2985 69.4 4396 3217 73.2 Iwate 4763 3394 71.3 5159 3759 72.9 Miyagi 5533 3493 63.1 5820 3903 67.1 Akita 5889 3967 69.6 5886 4096 69.6 Yamagata 5333 3326 62.4 6130 3857 62.9 Fukushima 6191 4187 67.6 7664 5248 68.5 (Continued Overleaf )

351

12:27:16:11:09 Page 351 Page 352

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

Table 17.1 Continued

DISTRICT MALE (age at 28 sai) FEMALE (age at 23 sai)

Prefecture population marrying proportion population marrying proportion

KANTO Ibaragi 7355 4566 62.1 8196 4781 58.3 Tochigi 4562 3018 66.2 5742 3349 58.3 Gunma 4413 3028 68.6 5544 3715 67.0 Saitama 7531 5137 68.2 9034 5317 58.9 Chiba 8550 5788 67.7 9367 6175 65.9 Tokyo 8624 3881 45.0 9882 4474 45.3 Kanagawa 5746 3197 55.6 7585 3700 48.8 HOKURIKU Niigata 11788 6826 57.9 12242 6935 56.6 Toyama 4520 2400 53.1 5953 3923 65.9 Ishikawa 5057 2260 44.7 6076 3204 52.7 Fukui 4093 1899 46.4 4882 2578 52.8 CHUBU Yamanashi 2993 1778 59.4 4044 2104 52.0 Nagano 7685 4656 60.6 9558 5816 60.8 Gifu 6552 2956 45.1 7215 3709 51.4 Shizuoka 7209 4600 63.8 8421 5531 65.7 Aichi 10294 5193 50.4 11117 6026 54.2 Mie 6493 3675 56.6 7507 3707 49.4 KINKI Shiga 4579 2130 46.5 5217 2262 43.4 Kyoto 6616 3046 46.1 6978 3316 47.5 Osaka 12220 5454 44.6 15376 5593 36.4 Hyogo 11807 5308 45.0 12375 5526 44.7 Wakayama 4897 1993 40.7 5415 1877 34.7 CHUGOKU Tottori 3190 1681 52.7 3152 1684 53.4 Shimane 5067 2410 47.6 5433 2818 51.9 Okayama 8564 4051 47.3 8654 4186 48.4 Hiroshima 10201 4958 48.6 10322 5027 48.7 Yamaguchi 6924 2964 42.8 8138 3749 46.1 SHIKOKU Tokushima 5530 2737 49.5 5660 2516 44.5 Ehime 12652 5777 45.7 13341 5925 44.4 Kochi 4508 2570 57.0 4573 3062 67.0 KYUSHU Fukuoka 8427 4042 48.0 11198 5127 45.8 Saga 4304 1842 42.8 4278 2001 46.8 Nagasaki 5328 2494 46.8 6617 2721 41.1 Kumamoto 7846 3973 50.6 9022 4169 46.2 Oita 5756 3110 54.0 6748 3605 53.4 Miyazaki 2737 1519 55.5 3592 1901 52.9 Kagoshima 6446 3088 47.9 8934 3563 39.9 Okinawa 3069 1798 58.6 3378 1488 44.0 TOTAL 287745 153870 53.5 327753 172096 52.5

352

12:27:16:11:09 Page 352 Page 353

ANOTHER FOSSA MAGNA

The area with lower proportions marrying was, for both sexes, Kinki, where Osaka and Kyoto are located, and west Hokuriku and Chugoku Districts. Outside of that region, there are also Tokyo and Kanagawa, one of the most urbanized areas, and Hokkaido, which was growing rapidly in the late nineteenth century. Thus the proportion marrying at certain ages has a pronounced regional bias. A similar sort of bias can be found displayed on Maps 17.1 and 17.2, which show age at marriage by prefecture, estimated from the proportion marrying of each age obtained in the above-mentioned way. The pattern of distribution of age at marriage corresponds closely to the pattern of distribution of proportion marrying shown on Table 17.1. For men, the age at marriage was well over 25 sai in the areas to the west of Toyama, Nagano, and Shizuoka, except for Okinawa. East of this line, except for Tokyo, age at marriage was well under 25 sai.16 For women, the age at marriage was well over 21 sai, except for Kochi, in the areas to the west of Toyama-Nagano-Shizuoka line. East of this line, except for Tokyo, the age at marriage for women was well under 21 sai. For both sexes in Tohoku, age at marriage was signiﬁcantly early. In the six prefectures that comprise this area, for males age at marriage was under 23 sai and for women under 19 sai. On the other hand, in the Kinki area and most of western Japan, late marriage was the rule. In the 21 prefectures making up this region, age at marriage for men was over 26 sai in thirteen prefectures, and age at marriage for women was over 22 sai in fourteen prefectures. From these ﬁndings it is possible to suggest that there were two patterns of marriage in Japan in the late nineteenth century—a pattern of early marriage in eastern Japan and one of late marriage in the western Japan. By looking more closely at representative prefectures from each region—Iwate Prefecture in the east and Wakayama Prefecture in the west—it is possible to reach the following conclusions. As we can see from Figure 17.2, the pattern of early marriage in Iwate in the east and the pattern of late marriage in Wakayama in the west for both sexes holds true until age 40 sai. In Iwate, for example, the proportion marrying for females age 20 sai was as high as 70 percent, but in Wakayama, it was a low 35 percent. The proportion of men marrying at age 25 sai was 50 percent in Iwate but 25 percent in Wakayama. Thus Iwate shows proportions marrying at the ages twice as high as Wakayama. Another important consideration is the rural-urban comparison of the proportion marrying. We chose three prefectures: Tokyo, the most urbanized one where nearly three-quarters of the population were dwelling within the urban boundary; Fukushima, an eastern rural prefecture; and Saga, a western rural prefecture. Figures 17.3a and 17.3b shows the proportion marrying from age 15 sai to 50 sai in these three prefectures. Tokyo was slightly higher than Saga in the younger age group, but after age 28 sai for men and age 25 sai for women, it became lower than that of Saga. This must be called an “urban type.” By contrast, Fukushima started and reached the peak earlier than Saga, but the proportion was almost same as Saga, and much higher than that of Tokyo. This must be called a “rural type.”

353

12:27:16:11:09 Page 353 Page 354

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

Map 17.1 Estimated Age at Marriage (1886): Men

354

12:27:16:11:09 Page 354 Page 355

ANOTHER FOSSA MAGNA

Map 17.2 Estimated Age at Marriage (1886): Women

355

12:27:16:11:09 Page 355 Page 356 Proportion Marryingand Wakayama. Iwate (1986): Figure 17.2

12:27:16:11:09 Page 356 Page 357 Proportion Marrying (1886) Men: Tokyo, Fukushima and Saga Proportion Marrying Tokyo, (1886) Men: Figure 17.3a

12:27:16:11:09 Page 357 Page 358 Proportion Marrying (1886) Women: Tokyo, Fukushima and Saga Tokyo, Proportion Marrying (1886) Women: Figure 17.3b

12:27:16:11:09 Page 358 Page 359

ANOTHER FOSSA MAGNA

CONCLUSION How can we explain these conspicuous regional differences in the proportion marrying and age at marriage? Geographical location—east or west—and degree of urbanization seem to be the decisive factors. The division line, incidentally, itself does fall on the Fossa Magna, a geological trough line which divides Japan into east and west. What we are interested in is the relationship between age at marriage and economic development. If age at marriage is dependent on economic circumstances, as is thought generally to be the case in the premodern world, such regional differences in the proportion marrying and age at marriage in Japan cannot be explained at all. This is because western Japan until the end of the nineteenth century was more prosperous and economically developed than eastern Japan. Although detailed information on the input and output of each prefecture cannot be obtained for this period, the degree of urbanization and the proportion of urbanized population can be computed from the TMK of 1884. In Tokyo Prefecture where 73 percent of the population was urban, the proportion marrying was relatively low and age at marriage relatively high in the more urbanized areas. But, unexpectedly, the Tohoku region is found to have a higher proportion of urban population than the rest of Japan. Except for large urban concentrations like Tokyo, Osaka, and Kyoto, small and medium sized cities did not appear to have the expected attributes.17 The differential age at marriage and proportion marrying between eastern and western Japan may be partly affected by marriage customs. In the eastern part of Japan, primogeniture was more common. People married earlier and sons and their wives worked with their parents in the fields. Intensive labor in the relatively larger fields was needed especially in the shorter working season. A more com- pressed family cycle (earlier marriage and earlier child-bearing) was, therefore, more suitable. By contrast, the western part of Japan contained a variety of inheritance customs, including ultimogeniture (inheritance by youngest son), and non-sex- selective inheritance. Consequently, people did not need to marry quickly. Heads of households wished to hold on to their authority as long as possible, because their landholdings were more valuable than in the eastern part. We suspect, therefore, that differences in age at marriage between eastern and western Japan and rural and urban areas were not strictly a result of geography and of degree of urbanization. Rather, they were related to differences in the family cycle and in inheritance customs as well as to differences in the type of agricultural production and size of farm. The exact nature of this relationship, especially where cause and effect are concerned, still requires systematic examin- ations. One is rather hard-pressed to find a consistent thread of explanation. In order to fully understand the context one should employ knowledge about agricultural technology as well as social-anthropological and cultural evidence for an explanation of these regional variations. The relationship between age at marriage and economic circumstances was one of mutual interactions, rather than a one-way process. On the one hand, age at marriage was determined by demand for labor in family in the countryside; on the

359

12:27:16:11:09 Page 359 Page 360

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

other hand, because of its regulatory power over population size, age at marriage also shaped economic conditions. My future research in this area will test this hypothesis.

ACKNOWLEDGMENTS The author gratefully acknowledges the contributions made by Professor W. Mark Fruin, California State University, Hayward, and Professor Tamara K. Hareven, Clark University, to the translation and revision of this article. The latter also gave me many suggestions and comments. Without their patient assistance, this article would not have appeared. Of course, I am totally responsible for remaining mistakes.

NOTES

1. See Hayami, 1973a), Figure 11–2, p. 188. 2. Calculated from Hayami, 1973a), Table 12–16, p. 220. 3. See Hayami, 1973b, Table 1, p. 8. 4. See Hayami, 1983, Table 7, p. 27. 5. See Narimatsu, 1985, pp. 83–84. 6. For example, Nihon Teikoku Jinko Seitai Tokei (Population Statistics of Imperial Japan) of 1908, edited by Bureau of Statistics, Cabinet Office, inserted “residential population surveyed by police” as an appendix. In this survey, the population of Tokyo Metropolitan Area was given as 2,328,791, but this was evidently an undercount because the survey of the Statistics Bureau estimated the population at 3,053,402, thirty percent higher than the police survey. The main reason for this gap appears to have been the double-counting of kiryu jinko, the population that had temporarily moved to Tokyo. 7. Hosoya, 1978, and Sorifu Tokeikyoku (Bureau of Statistics, Cabinet Office), 1976. 8. Naimusho (Ministry of Home Affairs), 1886. 9. Naikaku Tokeikyoku, 1898. Naikaku Tokeikyoku (Bureau of Statistics, Cabinet Office) compiled the statistics from 1898 for every five years until the year 1920 when the first census was carried out. 10. Naimusho (Ministry of Home Affairs), 1884. 11. Naikaku Tokeikyoku, 1899. 12. Hajnal, 1953. 13. Hayami, 1973a, pp. 123–127. 14. For Yokokuchi village, Shinano Province, see Hayami, 1973a, pp. 212–213. 15. The proportion marrying: for men age 26–30 sai, Tokyo 45.2, total 53.4; for women age 21–25 sai, Tokyo 33.9, total 50.5 percent. 16. The average age at marriage: for men, Tokyo age 26.0 total age 25.3; for women, Tokyo age 21.4, total age 21.3 sai. 17. For the variations of inheritance customs in Tokugawa Japan, see Homusho, 1877 and 1880.

REFERENCES Hajnal, John. 1953. “Age at Marriage and Proportion Marrying.” Population Studies 7(2): 111–136. Hayami, Akira. 1973a. Kinsei noson no rekishi jinkogakuteki kenkyu (Historical Demography of a Rural Society in Tokugawa Japan). Tokyo: Toyokeizai Shinposha. —— . 1973b. “Labor Migration in a Pre-Industrial Society: A Study of Tracing the Life Histories of the Inhabitants of a Village.” Keio Economic Studies 10: No. 2. —— . 1983. “The Myth of Primogeniture and Impartible Inheritance in Tokugawa Japan.” Journal of Family History 8: 3–29. Homusho (Ministry of Justice). 1877 (also 1880). Zenkoku minji kanrei ruishu (Surveys

360

12:27:16:11:09 Page 360 Page 361

ANOTHER FOSSA MAGNA

on the Customs of Civil Affairs) in Meiji bunka zenshu. (Collection of Meiji Civil- ization) Vol. 8: Law. Tokyo: Nihon Hyoronsha. (Reprinted in 1929.) Hosoya, Shinji, ed. 1978. Meiji zenki nihon keizaitokei kaidai shoshi: fukoku kyohei hen, jo-2 (Bibliography of Economic Statistics of Early Meiji Japan: Volume of Enrichment and Armament A-2). Tokyo: Institute of Economic Studies, Hitotsu- bashi University. Naikaku Tokeikyoku (Bureau of Statistics, Cabinet Office), ed. 1898. Nihon Teikoku Jinko Tokei (Population Statistics of Imperial Japan), surveyed December 31, 1898. Tokyo: Naikaku Tokeikyoku. —— . 1899. Nihon Teikoku Jinko Dotai Tokei (Vital Statistics of Population of Imperial Japan), surveyed December 31, 1899. Tokyo: Naikaku Tokeikyoku. Naimusho (Ministry of Home Affairs); ed. 1884. Tofu Meiyu Kokohyo (Table of the Population of Cities and Towns), Bureau of Geography, surveyed December 31, 1884. Tokyo: Naimusho. —— . 1886. Nihon Teikoku Minseki Kokohyo (Table of Civil Registration of Imperial Japan), Somukyoku Kosekika (Registration Section, Department of General Affairs), surveyed December 31, 1886. Tokyo: Naimusho. Narimatsu, Saeko. 1985. Kinsei tohoku noson no hitobito (Peasants Life of a Village in the Northern Tokugawa Japan). Kyoto: Minerva shobo. Sorifu Tokeikyoku (Bureau of Statistics, Cabinet Office), ed. 1976. Sorifu tokeikyoku hayakunen shiryo shusei (100 Years Collection of Documents of the Bureau of Statistics, Cabinet Office).

361

12:27:16:11:09 Page 361 Page 362

First published in Bastardy and its Comparative History (eds. P. Laslett et al.), pp. 397–402, London, 1980.

18 Illegitimacy in Japan

llegitimacy in Japan has not, until recently, been the subject of academic study. INational statistics were not collected before the beginning of the twentieth century, and only fragmentary and disorganized data are available for the period prior to the late nineteenth century. Before the Meiji Restoration of 1868 there was no clear distinction between legitimate and illegitimate children. Concubinage was a recognized practice among the wealthy and upper classes and in farming communities sexual relationship before marriage was an accepted custom. The imperial house was no exception. The Taisho Emperor, who succeeded in 1912 and founded the present dynasty, was a child of the last Meiji Emperor by a concubine. Thus it is fair to say that Japanese society has traditionally been tolerant of illegitimacy. There is not even an appropriate word in the traditional Japanese language for an illegitimate child. It is possible to indicate that someone is the child of a concubine, but there is no Japanese equivalent of the English word ‘bastard’. The distinction between kosei (legitimate) and shisei (illegitimate) in the statistics of the Meiji period is the result of the introduction, after the Restoration, of European philosophy and systems of registration, which required the creation of new terms to translate Western concepts. Prior to the Restoration, therefore, public documents do not distinguish between these two categories: in the shu¯ mon aratamecho¯ (registers of religious faith) there is neither a categorical nor a ter- minological distinction between legitimate and illegitimate children in any household. It is, however, possible to count the number of children born to unmarried parents, or to widows or widowers, though, strictly speaking, one should not conclude that these children were illegitimate: quite frequently a couple married after they had had a child. In the present study those children who appear in the shu¯ mon aratamecho¯ as having only one parent are classed as illegitimate. Taking these children as a percentage of the total number of births, calculated from the best available material1 (an unusually detailed set of registers for a single village, complete for a continuous period of one hundred years), we discover that of 503 males born in the period 1773–1869, 34 or 6.8 per cent, and of 490 females, 35 or 7.1 per cent fall into this category. These ﬁgures are the only statistics available to date and represent, in my opinion, the lower limit of the illegitimacy ratio in Tokugawa Japan. The ﬁrst illegitimacy statistics taken in the post-Restoration period were prepared by the government of Metropolitan Tokyo in 1878.2 According to these

362

12:27:16:11:09 Page 362 Page 363

ILLEGITIMACY IN JAPAN

Figure 18.1 Main districts and cities in Japan

ﬁgures, of 29,755 live births in that year, 791 were illegitimate. The proportion of illegitimate births (the illegitimacy ratio) in this 1878 sample is 2.7 per cent. This percentage must be on the low side because the illegitimacy ratio rises over the succeeding years. The ratio climbs sharply to 8.9 per cent in 1887, then reaches, after minor ﬂuctuations, a high of 10.0 per cent in 1897. Such an increase is almost certainly due to the gradual improvement in the technique of

363

12:27:16:11:09 Page 363 Page 364

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

gathering information. The same observation applies to metropolitan Osaka, for which we have the next earliest set of statistics. Taking later changes into consideration, the illegitimacy ratio of 7.2 per cent in 1881 is too low. The ratios of 17.5 per cent in 1888, and 18.6 per cent in 1897, are far more plausible. Thus the ratios for the early 1880s are unreliable and should not be used as a basis for research. Table 18.1 shows the illegitimacy ratios in 1887 and 1897 for which years the statistics may be accepted as accurate, for those fu (metropolitan cities) and ken (prefectures) where survey data were taken. Out of 47 fu and ken, data exist for 19, or about 40 per cent for 1887 and 12, about 25 per cent for 1897. The average national illegitimacy ratio indicated by these is 8.1 per cent for 1897, a ratio quite compatible with the ratio of 8.8 per cent in the more reliable registration system of 1900.3 Table 18.1 shows that the ratio rose over the ten-year period in all the prefectures studied. At the same time there are clear regional variations: illegitimacy ratios are highest in Osaka, Tottori and Tokyo and lowest in the prefectures north of Tokyo, or west and south of Tottori. Urban statistics for this period present us with a diﬀerent picture. In Table 18.2 the 54 cities with populations of more than 25,000 in 1897 are arranged in order of illegitimacy ratios: the very wide range of ratios is immediately obvious. Hakodate, the highest, has an illegitimacy ratio 14 times that of Kagoshima, the

Table 18.1 Illegitimacy ratio by ken or fu, Japan, 1887 and 1897

Ken or Fu 1887 1897

%% Aomori-ken 2.7 3.9 Akita-ken 3.4 – Iwate-ken 2.4 – Miyagi-ken 3.4 – Fukushima-ken 3.9 – Tochigi-ken 3.2 – Saitama-ken 2.3 3.4 To¯kyo¯-fu 8.9 9.9 Kanagawa-ken 3.6 5.9 Toyama-ken 2.1 3.4 Fukui-ken 3.9 – Yamanashi-ken – 2.7 Shizuoka-ken 2.5 – O¯¯ saka-fu 14.9 19.6 Wakayama-ken 2.6 – Tottori-ken 9.7 10.1 Ko¯chi-ken 5.4 6.8 Saga-ken 1.8 3.4 Nagasaki-ken 4.5 – Kumamoto-ken 1.7 3.9 Kagoshima-ken – 5.4 Total 6.7 8.1

Note: Ken and fu are local administrative units created after the Restoration. The three most urbanized prefectures, Tokyo, Kyoto and Osaka, were called fu, usually rendered ‘metropolis’, all others, ken.

364

12:27:16:11:09 Page 364 Page 365

ILLEGITIMACY IN JAPAN

Table 18.2 Illegitimacy ratio in cities over 25000 inhabitants, Japan, 1897

1 Hakodate (Hokkaido) 43.2 29 Maebashi (Gumma) 8.6 2 Sakai (Osaka) 25.2 30 Niigata (Niigata) 8.6 3 Shimonoseki (Yamaguchi) 24.7 4 Nara (Nara) 23.7 31 Fukuoka (Fukuoka) 8.5 5 Okayama (Okayama) 23.6 32 Tsu (Mie) 8.4 6 Osaka (Osaka) 22.5 33 Aomori (Aomori) 8.3 7 Sapporo (Hokkaido) 21.7 34 Hiroshima (Hiroshima) 8.0 8 Kobe (Hyo¯go) 21.3 35 Sendai (Miyagi) 7.8 9 Takamatsu (Kagawa) 19.0 36 Toyama (Toyama) 7.4 10 Wakamatsu (Fukushima) 18.3 37 Yamagata (Yamagata) 7.3 38 Fukui (Fukui) 7.1 11 Matsue (Shimane) 18.0 39 Chiba (Chiba) 6.9 12 Tottori (Tottori) 17.8 40 Naha (Okinawa) 6.8 13 O¯¯ tsu (Shiga) 17.2 14 Takasaki (Gumma) 17.2 41 Ko¯fu (Yamanashi) 6.5 15 Kyoto (Kyoto) 16.7 42 Akita (Akita) 6.4 16 Nagoya (Aichi) 16.6 43 Ujiyamada (Mie) 6.2 17 Nagasaki (Nagasaki) 15.7 44 Kumamoto (Kumamoto) 5.8 18 Yokohama (Kanagawa) 15.1 45 Utsunomiya (Tochigi) 5.8 19 Ko¯chi (Ko¯chi) 14.1 46 Mito (Ibaraki) 5.6 20 Tokyo (Tokyo) 13.1 47 Shizuoka (Shizuoka) 5.3 48 Morioka (Iwate) 5.3 21 Gifu (Gifu) 12.7 49 Kurume (Fukuoka) 4.2 22 Wakayama (Wakayama) 12.7 50 Takaoka (Toyama) 4.0 23 Himeji (Hyo¯go) 11.0 24 Matsumoto (Nagano) 10.8 51 Yonezawa (Yamagata) 3.7 25 Tokushima (Tokushima) 10.7 52 Saga (Saga) 3.6 26 Matsuyama (Ehime) 9.4 53 Hirosaki (Aomori) 3.2 27 Kanazawa (Ishikawa) 9.3 54 Kagoshima (Kagoshima) 3.1 28 Nagano (Nagano) 8.7

lowest. Twelve of the top 20 cities are in Kinki, Chu¯ goku and Shikoku districts. The other 8 in the top 20 are all, except Wakamatsu and Takasaki, major cities of over 100,000 inhabitants. The ratios for Takasaki and Wakamatsu are unusually high in 1897, but if one were to substitute the average ratios for the ﬁve years 1892–6 these two places would fall below the top 20, so that it would appear that their 1897 ratios and rankings are the result of survey errors. The more sparsely populated districts of Tohoku in the north and Kyu¯ shu¯ in the southwest, on the other hand, show low ratios. Except for Nagasaki and Wakamatsu the highest ratios are 8.3 per cent in Aomori in the north and 8.5 per cent in Fukuoka in the south. It should be noted that eight of the lowest ranking twenty cities are in these two districts. The following observations may therefore be made on the data in Table 18.2: 1 The ratio is high in the Osaka-dominated Kinki district, as far east as the Nagoya region, and in western Honshu¯ and Shikoku. 2 Tokyo, the capital and largest city, shows a high ratio.

365

12:27:16:11:09 Page 365 Page 366

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

3 Cities in the newly colonized island of Hokkaido are also very high. 4 Port towns like Yokohama, Shimonoseki and Nagasaki have quite high ratios. To amplify this picture a calculation has been made of the coefficient of correlation between the ratio and the distance from Osaka for all cities in Table 18.2, except those in Hokkaido. A correlation of r = 0.54 was obtained. Confining the calculation to the cities under observation (1), the correlation is r = 0.70, which is within the significant limit. A city’s illegitimacy ratio is therefore generally inversely proportional to its distance from Osaka. In contrasting the illegitimacy ratios of urban and rural areas we find the urban ratios to be generally higher. Material which provides such data is scarce, but Table 18.3 shows the ratios in urban and rural areas for 1887 and 1897 where they are available. In those prefectures which have towns to which observations (2) to (4) apply, the ratios are much higher in the towns than in the rural areas:

1887 1897 Tokyo 1.9 × 2.6 × Kanagawa 4.1 × 3.1 × Nagasaki 6.2 × –

But in the other prefectures the differences in illegitimacy ratios in urban and rural areas are much smaller, averaging only 1.5 times in 1887. This pattern is confirmed by the high coefficient of correlation between urban and rural illegitimacy ratios in these prefectures: r = 0.94 in 1887 and r = 0.92 in 1897.

Table 18.3 Illegitimacy ratio by urban and rural areas, Japan, 1887 and 1897

1887 1897

Urban Rural Urban Rural

%% % % Aomori-ken 1.3 2.8 3.2 3.9 Akita-ken 5.0 3.4 Iwate-ken 1.9 2.4 Miyagi-ken 6.2 3.1 To¯kyo¯-fu 10.3 5.3 13.1 5.1 Kanagawa-ken 11.9 2.8 15.1 4.8 Toyama-ken 2.6 2.1 7.4 3.1 Fukui-ken 7.3 4.1 7.1 5.0 Shizuoka-ken 3.7 2.5 O¯¯ saka-fu 20.2 12.7 22.7 17.3 Wakayama-ken 7.2 2.2 Tottori-ken 15.1 9.4 17.7 9.6 Ko¯chi-ken 5.0 5.5 14.1 6.5 Saga-ken 1.5 1.8 3.6 3.4 Nagasaki-ken 22.9 3.7 Kumamoto-ken 2.2 1.7 5.8 3.9 Total 10.3 3.5 13.2 3.9

366

12:27:16:11:09 Page 366 Page 367

ILLEGITIMACY IN JAPAN

These prefectures have an extremely high positive correlation between illegitimacy ratios in urban and rural areas. However, in Tokyo, Kanagawa and Nagasaki, which have towns to which observations (2) to (4) are applicable, though the rural ratios are low, the urban ratios are very high. The ratios in these cities are clearly determined independently from the ratios in the surrounding rural districts. The regional variations in rural illegitimacy ratios in Table 18.3 can be explained in the same manner as was applied in observation (1) for Table 18.2. That is, for the prefectures considered in Table 18.3, the illegitimacy ratios are inversely proportional to the distance of the seats of the prefectural government from Osaka with a coefficient of correlation of r = 0.68. Certain conclusions may be drawn; but the question is, first of all, how to interpret the correlation of the illegitimacy ratio with distance from Osaka. Osaka was the centre of commerce and finance in Tokugawa Japan and the operations of the market economy were focused on that city.4 The markets for rice, the staple commodity, and for other products were there. Highly sophisticated commercial, financial and credit institutions were all based and developed in Osaka. This was a city of half a million commercially minded individuals, in complete contrast to the political city of Edo, dominated by samurai-bureaucrats. (The population in Edo after 1700 is estimated at around 500,000 samurais and 500,000 commoners. The samurai population of Osaka, on the other hand seems to have been about 10 per cent.) Compared with Edo, Osaka was less steeped in Confucian ethics, more secular and more realistic. These factors seem to have contributed to a higher illegitimacy ratio in Osaka and the surrounding areas. After the Meiji Restoration Osaka rapidly lost its unique position. For, whereas in the Tokugawa period rice as a means of paying taxes had flowed through Osaka from all parts of Japan, the land tax was now paid in cash and Osaka’s storehouses no longer bulged with rice. The introduction of steam power made possible direct, nationwide transport links between producing and consuming districts and this affected adversely the position of Osaka as a supplier of goods to Edo, now renamed Tokyo. The late nineteenth century witnessed the rapid disappearence of the pattern and mode of life of Tokugawa Japan because of modernization, industrialization and urbanization. In Tokyo, the capital, in port cities like Yokohama, Kobe, Shimonoseki and Nagasaki, in the two newly developed cities of Hokkaido and Kobe, which were all centres respectively of modern transport and industry, the army and the administration, higher illegitimacy ratios kept pace with these modern developments, and produced the regional patterns described above. After 1900 the illegitimacy ratio began to decline and it is, therefore, of interest to discover the peak years of illegitimacy. The data for the years 1887–97 give interesting regional differences in the peak years. Complete data for the entire period are available for only eleven prefectures. In Osaka, Tottori and Tokyo, where the ratios are high, the peak came early in the years 1890–4. By contrast, in the other prefectures, with lower ratios, the peak ratios are in the later years, 1895–7. The same relationship appears in the cities. In eight Kinki cities the average peak year is 1892. In ten cities in Tohoku and Hyu¯ shu¯ the average peak year is 1895, three years later. In cities, especially those Kinki cities with the highest ratios, illegitimacy had reached its peak and begun to decline at the end of the nineteenth century. For other cities and rural districts there is only very scant data for

367

12:27:16:11:09 Page 367 Page 368

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

the whole eleven years, but there is evidence that in these areas the illegitimacy ratio did not peak until the beginning of the twentieth century. It is commonly assumed that a rising illegitimacy ratio is the usual concomitant of industrialization and urbanization in almost all countries. In Japan, on the contrary, the ratio clearly decreased as industrialization and urbanization progressed. It is not possible to predict how long this decrease will continue, but as the ratio is now less than one per cent, an asymptotic eﬀect and probably a future rebound may be anticipated. However, the declining illegitimacy ratio as industrialization progressed in Japan must also be set against the high illegitimacy ratio in pre-industrial Japan. To explain this decrease is a task for the future, requiring more detailed analysis. It is hoped that closer examination of shu¯ mon aratamecho¯ and other registers of residents in the cities will help to explain the high illegitimacy ratios of pre-industrial Japan.

NOTES

1. Shu¯ mon aratamecho¯ of Nishijo-mura, Anpachi-gun, Mino-nokuni, 1773–1869. See Akira Hayami, ‘Labor migration in a pre-industrial society: a study tracing the life histories of the inhabitants of a village’ (1973). 2. Naikaku To¯kei Kyoku (Bureau of Statistics), I shin igo teikoku to¯kei zairyo¯ isan (Imperial statistical data since the Restoration) IV (1913). 3. Shirley F. Hartley, ‘The decline of illegitimacy in Japan’ 1970, Table 1. 4. William B. Hauser, Economic institutional change in Tokugawa (1974), chapter 2.

368

12:27:16:11:09 Page 368 Page 369

First published in Reitaku Journal of Interdisciplinary Studies, vol. 9, no. 2, pp. 1–10, 2001.

19 Koji Sugi and the Emergence of Modern Population Statistics in Japan: the Inﬂuence of German Statistics 1

INTRODUCTION n 1868, the start of the Meiji Restoration brought about the end of the long Iperiod of Tokugawa rule in Japan. The newly established Meiji Government began performing various types of statistical surveys in an attempt to grasp the overall picture of the country’s situation. At that time, the government learned many things from German statistics. In this short note, I would like to examine the details of what Japan learned from the German statistics and clarify the effects German statistics had on demographic statistics in Japan. When looking at why it was German statistics that were chosen as a model, if we trace Japanese history back many centuries, we see that during Japan’s period of national isolation (1639–1854), the only Western country that was permitted to have relations with Japan was the Netherlands, more precisely the Dutch East India Company. Through trade with the Netherlands, medical and natural science documents were brought into Japan, and these books became the basis for the foundation of European-style sciences known as “Dutch Studies.” In Nagasaki, the location of the Company at Deshima, lived families that worked as interpreters for many generations, so there were several people in the area that could understand Dutch. After the U.S. and Japan formed a treaty of amity in 1854, Japan began establishing diplomatic relationships with the U.S. and many countries of Europe, leading to increased opportunities for the Japanese people to come in contact with Westerners. An important thing to remember here is that the only European books that could be imported until that time were written in Dutch and limited to natural science and medicine. After Japan opened its doors, however, the range of available documents included English, French and German and was no longer limited to natural science and medical related documents. The Tokugawa Government had established an organization called “Office of Examination on Books by Dutch,” but this was changed to the “Office of Examination on Foreign Books” in 1856. In 1862, the first students sent overseas by the Tokugawa Government left for the Netherlands. Considering the foreign language skills of the Japanese people at that time, there was really no other choice in destination. The Tokugawa Government worked enthusiastically to introduce modern technology and

369

12:27:16:11:09 Page 369 Page 370

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

knowledge from the West, and sent study groups abroad to learn as much as they could. On this point, the daimyo, territorial lords and members of the Imperial Court, the primary force behind the Meiji Restoration, formed a conservative coalition, and for this reason, the dissolution of the Tokugawa Government and the formation of a new government became known as the Meiji Restoration rather than the Meiji Revolution. The relationship between Japan and Europe prior to the Meiji Restoration is a very important premise for the theme covered in this paper.

KOJI SUGI AND EARLY POPULATION STATISTICS “Koji Sugi” is one name that cannot be left out of a discussion of statisticians working in both the pre- and post-Meiji Restoration eras. Sugi was born in Nagasaki in 1828. In 1848, he entered the famous Dutch Studies School (Teki-juku) in Osaka. In 1850, he moved to Edo (now Tokyo), and in 1855, his outstanding ability in Dutch Studies led to his selection as a retainer for Masahiro Abe, one of the most passionately active supporters of foreign diplomacy in the Tokugawa Government at that time. Sugi worked at the “Oﬃce of Examination on Foreign Books,” which had been re-established under its new name by Abe, and was also a teacher at the School for Western Studies built by the Tokugawa Government (1864). Sugi regularly lectured on Dutch newspapers. One day, a newspaper article about Bavarian educational statistics sparked in him a deep interest. In his age seventy years commemorative volume,2 Sugi states, “In 1855 or 1856, Bavarian educational statistical surveys included a study of literacy, a study which I felt needed to be implemented soon in Japan as well. Later, in 1860/1861, books on statistics arrived from the Netherlands. These books included information about population surveys, covering such statistics as births, deaths, marriages, divorces, population movement and crime statistics, and I carefully read these books with great interest. As students studying abroad returned to Japan and shared what they had learned about statistics, my interest in statistics grew even stronger.” (Paraphrased by this author.) This was right at the time when two students sent abroad by the Tokugawa Government to study (Amane Nishi (1829–1897) and Mamichi Tsuda (1829–1903)) had returned to Japan after studying statistics under Simon Visseling (1818–1888) at Leiden University. These two students returned to Japan in 1865 and began teaching at the same school as Sugi. Sugi borrowed Tsuda’s notebook from the Visseling lectures and translated the notes into Japanese (the translation was published in 1874). The next work on statistics to be translated into Japanese after the Visseling notes was “Elements de statistiques ...” (Paris, 1856) by Alexandre C. Moreau de Jonnes (1778–1870) of France. This translation was published from 1874 to 1878. The translator was Gensho Mitsukuri (1846–1897), a scholar who had studied in France. After 1875, various other statistics-related books from England were translated and published, helping to instill Japan’s scholastic world with knowledge about statistics. With the Meiji Restoration in 1868, the Tokugawa Government, of which Sugi was a retainer, was given rule over Shizuoka, approximately 200 kilometers

370

12:28:16:11:09 Page 370 Page 371

KO¯¯ JI SUGI AND THE EMERGENCE OF MODERN POPULATION STATISTICS

west of Tokyo. Sugi also relocated temporarily to Shizuoka, where he made a great achievement. While temporarily located in Shizuoka, Sugi attempted demographic surveys (censuses) based on his own concepts. The results of demographic surveys done for two towns are still in existence,3 and while the results do not contain data for individual returns, this was the first demographic survey in Japan to show population totals by age and occupation as well as marriage status and statistics for people leaving and entering the area for work. This was truly the first attempt of modern census type survey compiled from hundreds of population. Back to 1870, the Meiji Government recognized the ability of Sugi’s methods for compiling statistics, and made an offer to Sugi. The Meiji Government wanted to implement a nationwide population survey that it had been planning. However, this survey planned by the Meiji Government differed from Sugi’s method in that it was based on the family register, a registration system also implemented by the government. The family register was implemented by a daimyo, which had achieved the greatest standing in the Meiji Restoration, and therefore, the family register system was strongly influenced by Confucianism. With Confucianism, the most important aspect of a person’s identity is the social position (Stand) to which the person belongs. The first thing clearly marked on a family register is the class—nobility, samurai, commoner—to which the family belongs. Furthermore, the population was recorded as a population de jure. (Japan still uses this type of family register that defines a person’s legal identity and is required for any type of legal activity.) Sugi believed that demographic surveys and this family register had completely different principles and that the family register could not be used to develop national demographic statistics. He wrote an advise stating his opinion that demographic statistics had to be based on an impartial statistical survey of all people. He then resigned his position immediately and returned to Shizuoka. The following year, however, Sugi received official notification from the Meiji Government that the ideas expressed in his advise had been recognized, and under a strong request by the government, he returned to Tokyo and, as part of the Cabinet Bureau, undertook the drafting of general statistics for Japan. During this time, he became a charter member of Meirokusha; a civil organization based on a “campaign for enlightenment” in the 1870s, and was actively involved in the movement through lectures and writings. At the same time, he also submitted many essays related to his own concepts about an ideal demographic survey, an idea he could not easily abandon. At 12:00 p.m., December 31, 1880, Sugi’s dream became a reality when a demographic survey was carried out on the Kai Province (what is now Yamanashi Prefecture) as a test for a nationwide survey. The results of this survey were published in a 452-page document entitled “The Survey of the de Facto Popula- tion of Kai Province.”4 Recent censuses in villages in Yamanashi Prefecture have also led to the discovery of original census returns in several villages. This publication and the census returns tell us that the 1880 census was a modern national census conducted by trained personnel who went from house to house to complete the census. According to Sugi, 110,000 household-census returns were produced and distributed to the town and village offices, then the completed returns collected.

371

12:28:16:11:09 Page 371 Page 372

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

One aspect of this census receiving considerable attention is the extreme detail with which occupations were surveyed. The occupations of each family member were recorded, and side jobs were also surveyed and recorded. Considering that, at the time of this census, most occupational surveys were limited to the occupation of the head of the household, we can see that this surveying of the occupations of each household member was the result of Sugi’s thorough understanding of the modern national census methods used in the Western countries. Although literacy, one of Sugi’s initial primary concerns, was not studied, the census did survey whether or not people went to elementary school. There were no major differences in the results of Sugi’s census and the census (January 1, 1880) done by the government using the family registers (looking at the total population for all prefecture, Sugi’s census indicated 397,412 and the government survey 395,696, only a slight difference). There was, however, a rather large difference in the totals for Nishi-Yamanashi-gun, which included the city of Kofu, where the prefecture capital was located. Sugi’s census, using de facto population, had a much higher total than the government survey, which was based on de jure population. This difference was probably due to the influx of people into the city. (The Sugi census showed a population of 35,743 for Nishi-Yamanashi-gun, approximately 10% higher than the total of 32,383 indicated by government survey based on family registers.) Sugi also made comparison calculations of the cost of performing a census using Austria at 2.6 sen and Prussia at 1.5 sen per capita. In Japan, the cost, including preparatory expenses, was 1.449 sen. Therefore; a census of the entire country (population: 35,925,000) would total 52,557 yen (100 sen = 1 yen) and would involve 243,000 census workers. Sugi strongly emphasized the fact that at this cost, a fundamental statistical survey could be conducted for the entire country of Japan. The budget for the central government for that year was 63 million yen, so a census could indeed be done for a mere 0.8% of the total budget. When compiling the census for the Kai Province, Sugi also drafted approximately 400,000 data sheets for individuals based on the household data sheets that were collected, and then extracted various statistics based on these individual data sheets. In this aspect, the Sugi census was organized in much the same way as a modern census. His concept for the individual data sheets came from a book on statistics by M. Bloch. With Sugi’s success in the census of the Kai Province, he was confident that he could organize the national census, but a pitfall waited. In 1881, the government adopted various retrenchment policies in an attempt to curtail the ongoing inflation that had been plaguing the country. Furthermore, with the political changes that were also taking place at the time, the high officials that had supported Sugi were removed from government office, and in 1885, Sugi himself had to leave government employ. One other reason for it might be that he was not appointed to the organization that handled statistics within the newly formed government. Still, Sugi continued his work in statistics, compiling and publishing the results of his Kai Province census, and in 1883, he founded the School of Statistics, where he also served as a teacher. With his many years of experience in compiling statistics, Sugi began to devote himself to the training of exceptional human

372

12:28:16:11:09 Page 372 Page 373

KO¯¯ JI SUGI AND THE EMERGENCE OF MODERN POPULATION STATISTICS

resources. He organized a scholastic society for statistics, published statistics- related magazines, and never tired of conducting lectures on statistics for young government officials. The school founded by Sugi, unfortunately, had a weak financial foundation, and closed in just three short years. However, looking at the list of students who trained at the school, we find a large number of administrative officials who later were active in all aspects of statistical work in Japan. Through the curriculum and textbooks used at that school, we could learn about how statistical education was conducted in Japan at that time.

THE EFFECTS OF GERMAN STATISTICS ON STATISTICAL EDUCATION IN JAPAN In 1874 or 1875, Sugi obtained a copy of “Lehr- und Handbuch der Statistik in ihrer neuesten wissenschaftlichen Entwicklung” (Max Haushofer, Wien, 1872) from a friend who had been sent to work in Vienna, Austria. It is said that this book had a strong effect on Sugi. Haushofer was not highly regarded as a statistician in Germany, a fact that was known even in Japan. Iwasaburo Takano (1871–1949), who made a thorough study of statistics under Gustav von Mayr and was the first Japanese professor of statistics at the Imperial University of Tokyo (established 1903), went so far as to say that if Sugi had been familiar with von Mayr’s writings, which were published five years after the Haushofer’s book, the study of statistics in Japan would have most likely taken a different direction. In addition to books on statistics, Haushofer (1840–1907) also wrote novels and collections of poems, and was a person involved in a wide range of intellectual activities. As such, his work was considered to be a suitable introductory basis for use in the formation of scholastic studies in Japan during its “age of enlightenment.” According to Sugi’s commemorative volume, about the same time he was reading Haushofer, he also had an opportunity to read “Die Moralstatistik und die christliche Sittenlehre. Versuch einer Sozialethik auf empirischer Grundlage” (Alexander von Oettingen, 1868–1873). One could say that Sugi’s foundation for statistical studies was solidified by these works. Sugi was indeed captivated by Haushofer’s work, and used his work as text for 33 lectures given at statistical society meetings from January 1879 to December of the following year. According to people who attended his lectures, Sugi also incorporated concepts from Seussmilch in his enthusiastic lectures on Western statistics, German statistics in particular, given to approximately 20 people each time. Many people who attended his lectures later became active in the academic world and government organizations. Haushofer’s work strongly affected Sugi’s Kai Province census as well as the many essays he drafted related to his opinions on demographic surveys. Sugi also became a teacher at the School of Statistics, which he founded, and based his curriculum on Haushofer’s work. The text he was using at that time was the second edition of Haushofer’s book, published in 1882. According to Sugi’s own lecture notes of the time,5 he gave lecture on the line of the Haushofer’s textbook. There also remained notebooks from a student who had attended his

373

12:28:16:11:09 Page 373 Page 374

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

lectures.6 Of the five volumes of Haushofer’s work, Sugi faithfully adhered to “Geschichte der Statistik” (Erstes Buch), “Bevölkerungsstatistik” (Tweites Buch), “Das gesellschaftliche und politische Leben” (Viertes Buch), and “Moralstatistik” (Fünftes Buch), in his lectures. “Drittes Buch” was used in a course handled by another instructor (K. Okamatsu), who was apparently not as faithful as Sugi to the original text, and who frequently incorporated the theories of statisticians such as G. von Mayr and W. Roscher in his lectures. Although Sugi had many opportunities to use Haushofer’s work, no complete translation of Haushofer’s work was ever published in Japanese. His work was translated into Japanese not a volume at a time, but a few chapters at a time, by the people studying under Sugi, and the translations appeared numerous times between 1882 and 1907 in scholastic journals related to statistics. As a result, Haushofer’s work was well known in Japan at that time. Of course, teachings at the School of Statistics were not limited solely to Haushofer’s work. Other texts that we now know were used include “Traité théorétique et pratique de statistique” (M. Block, Paris, 1878) and “Allgemeine Bev- ölkerungsstatistik” (J.F. Wappaeus, Leipzig, 1859–1861). Still, the school focused on the work of Haushofer, and it can be said that, for better or worse, the establishment of statistical studies in Japan was greatly affected by German statistics through Haushofer’s work. The first Japanese professor of statistics at the influen- tial Imperial University of Tokyo was Iwasaburo Takano (1871–1949), who studied under von Mayr from 1895 in Munich, pursuing household budget surveys and social surveys. Takano left the Imperial University of Tokyo in 1919 to work on the resolution of social problems and become active in labor movements. In any event, it can be said that German statistics, more than those of any other country, affected the actual compilation of statistics by the statistical academia of Japan in the early days of the formation of modern Japan.

THE MODERN CENSUS IN JAPAN Sugi often asserted the necessity of implementing a census, but the Meiji Govern- ment had many other things that had to be dealt with, and there was no one in the government particularly interested in proceeding with a census. However, there was a worldwide movement to perform simultaneous censuses in all countries around the world to commemorate the year 1900. In 1895, the International Statistical Institute queried Japan about whether or not the country would participate in the census, which led to the Japanese government’s decision to proceed with the census. Census executing committees were even formed within the House of Peers and House of Representatives. Unfortunately, the time for the census fell just before the impending Russo-Japanese War (1904–1905). The government diverted as much of its budget as possible to military armament, and eventually, plans for the census were dropped. The next opportunity for a census came in 1910. However, Naosaburo Hanabusa, director of the Cabinet Bureau of Statistics formed in 1898, began compiling and publishing demographic statistics for static populations (état de la population) and dynamic populations (mouvement de la population) based on family registers. This may have led to increased hesitation about performing a census.

374

12:28:16:11:09 Page 374 Page 375

KO¯¯ JI SUGI AND THE EMERGENCE OF MODERN POPULATION STATISTICS

Such statistical documents were written in both Japanese and French since the survey of 1899. As such, Japan did not perform a modern census until 1920, making it the last advanced industrial nation, and even one of the last Asian countries, to begin performing modern censuses. However, thorough preparations were made for the census, so the census results are highly reliable. Sugi died in 1916, just a few years before the first modern census was performed in Japan. However, his earnest wish for the implementation of the national census was realized, although not where he had expected. As a result of Japan’s victory in the Sino-Japanese War (1894–1895), Taiwan became a colony of Japan. A call for a census in Taiwan started in 1903. Census preparations got underway during the Russo-Japanese War, under the direction of Shimpei Goto, Director of the Civil Affairs of Taiwan, and the census was implemented immediately after the conclusion of the war, on October 1, 1905 (total population: 3,039,751). The survey was performed in the form of a modern census. Training sessions were held for the 6,000 census takers, household data forms were drafted, and the takers went door-to-door collecting the census forms. In a society where the literacy rate was less than 10%, the reliability of this type of survey can be questionable. Still, there was a surprising lack of age heaping, and this was indeed the first modern census performed for a Chinese population. The results of the census were published in “A Provisional Survey of Households and Population in Taiwan,”7 but as the text was written in Japanese, these documents are not well known on a global level, and very little research focusing on these documents has actually been done. An English synopsis has been published,8 and only one work that does make use of the census is published.9 It is interesting to note that the person in charge of this important census was a student who had attended the School of Statistics and studied under Sugi. The person at the center of the implementation of this census, who promoted and organized the census, was Shichisaburo Mizushina, one of the first graduates of the School of Statistics. Twenty years after graduating, Mizushina made Sugi’s dream a reality in Taiwan. In addition to Mizushina, various other members of the School of Statistics’ first graduating class were also involved in the census. Kaichi Nagayama was in charge of occupational surveys, Isao Takemura was in charge of surveys related to the physically disabled, households and residents, and Yuishiro Takeda was in charge of surveying populations, tribes, physical characteristics, age and lineage. Of the seven people overseeing the census, four had attended the School of Statistics and studied under Sugi, and played leading roles in the implementation of the census. The survey of occupations was done in great detail, covering primary occupation, secondary occupation, and the employment status of each family member. The survey covered much more detail than the family register-based census in Japan, which only surveyed the occupation of the head of the household. This is one point in which the fundamental concepts of Sugi statistical education were reflected in the census. It was 30 years earlier that Sugi first learned about German statistics through the “Handbuch” of Haushofer, realized the need for implementing a modern census, and studied methods of implementing such a census. One gets a sense of

375

12:28:16:11:09 Page 375 Page 376

POPULATION, FAMILY AND SOCIETY IN PRE-MODERN JAPAN

astonishment when considering the fact that his students were the ones to actually implement this type of census, but in colonial Taiwan rather than in Japan.

AN EPILOGUE ON SUGI In 1885, Sugi left government work and never again returned to official work. His School of Statistics closed in 1886 due to financial difficulties, after just a short time in operation. At the age of 59, Sugi completely retired from regular work. After that, he began lecturing at the Japan Academy and at the Association of Statistics, societies of which he had been a member since 1879, contributed short works to magazines and newspapers, and was awarded numerous honors. He suffered weak eyesight due to the intensive reading he did under poor conditions in his youth, but still outlived eight of his eleven children, and his wife, to reach the ripe old age of 90. Curiously, the budget for the first national census in Japan was approved at the Parliament just one year before his death, December 4, 1917. Although Sugi never once traveled abroad, he knew more about the “Modern West” than perhaps anyone else in Japan, and put this knowledge to good use through the compilation of statistics and the training of a great number of successors. Unfortunately, the foundation of the newly formed Meiji Government was weak, and those belonging to the factions responsible for the Meiji Restoration eventually occupied the important positions in government. Individuals such as Sugi found themselves alienated and unable to advance to higher positions within the government. In recent times, however, we are at last taking a new look at the achievements of Sugi and giving him the credit and honor he deserves.

NOTES

1. This note was originally presented at the Workshop “Lexis in Context,” held in Max Planck Institute for Demographic Research, Rostock, Germany, on August 28/29 2000. I would express my appreciation to Director Professor James W. Vaupel and Organizer Dr. Jochen Fleischhacker of the Institute for inviting me to the meeting, Reitaku University to support my attending and the staﬀs of Kiko International, Kyoto, Japan, to translate at the earlier stage. 2. “Sugi sensei koenshu.” (Lectures of Mentor Sugi) 1902 (private edition). 3. In op. cit. Appendix. 4. Tokei-in (Bureau of Statistics) ed., Kai no kuni genzai ninbetsu shirabe. 1882 (reprinted in 1968). 5. “Statistiks” reprinted and edited by Shinji Hosoya, Nihon Tokei Kyokai, 1983. 6. Manuscripts by Masao Yokoyama. A collection of the library of Institute of Economic Studies, Hitotsubashi University. 7. Rinji Taiwan Koko Chosa. 7 vols. 1907–08. 8. The Special Population Census of Formosa, 1905: Report of the Committee of the Formosan Special Census Investigation. 1909, Tokyo, Imperial Printing Bureau. 9. G.W. Barclay, Colonial Development and Population in Taiwan. Princeton, 1954.

376

12:28:16:11:09 Page 376 Page 377

Index

age at marriage Akasaka Keiko, 133 Aizu mountainous district, 311–12, 313 Aki province, 153–8 estimates, 1886, 354–5 ‘amusement quarters’, 38–9 late 19th century, 349–59 ane-katoku, 296 Mino province, six villages, 207–10 Arioshinden village, 204–205 Nishijo¯ village, 346–7 Asama, Mount, 54, 120 Northeast Japan, 285–6 Ashikaga Shogunate, 15 regional differences, 284, 287–8, 345–7, 354, 359 bakufu surveys, 134–5, 137, 141–2 Southwest Japan, 290–1 Beppu village, age distribution, Suwa Country, 265–7 80–1 Tohoku village, 347 Bingo province, 153–8 Yokouchi village, 271, 274, 346 brewing industry, 37 agriculture 18th century, 116 cadastral surveys, 49 developments during Tokugawa period, calendar, 184 115–17 Cambridge Model, 317–18 fertilizer, 60 castle towns, 47, 106 ‘great transformation’, 115–16 census data horse-power to man-power change, 60, defects, 103 65–7 late 19th century, 347–9 ‘industrious revolution’, 116 Central region of Japan, 288–90 livestock changes, 67–9 Dekasegi rate, 289–90 vertical land use, 116 Ch’ing China, 3, 117 Ainu tribe, 40 ‘Chinese World Order’, 11, 15, 17 Aizu mountainous district cholera, 109–10, 143, 151 age at first marriage, 311–12, 313 Christianity age structure, 309, 310 hostility towards, 6–7, 16, 17–18, 75, fertility, 312, 314 166–7 household dynamics, 315–29 Portuguese influence, 15–16 Ishibushi village, 304 rebellion, 1637, 166 Kanaizawa village, 304 class differences Kuwanohara village, 304 allocation of families to Classes, marriage, 309–11 206–207 mortality, 314–15 marriage and fertility, 204–18 people with living spouses, 312 cotton industry, 27, 37 population trends, 308–309 currency debasement, 32–3, 58–9 sex ratio, 309, 310 total marital fertility rate, 312, 314 daimyo¯ To¯nosu village, 304 18th century changes, 34 Aizu region controlled through own families, 19–20 demography, 305–15 Edo visits, 47–8 population trends, 307–308 postings, 15

377

12:28:16:11:09 Page 377 Page 378

INDEX

demography ‘Great Land Reclamation Period’, 23, 30 Aizu region, 305–15 guns, introduction of, 13 ‘centre of gravity’, 139–42, 150 population enumeration, 1721, 76 Hajnal, John, 349 regional differences, 280–302 Hanley, Susan B, 106, 137 religious faith recording, 75–6 hansatsu, 32–3 Tokugawa period, beginning, 75–6 harvests, bad, 54, 105, 115, 119, 305 Yokouchi village, 185–203 Haushofer, Max, 373–4 desired fertility, 138–9 Hayami county Dewa province, 121 assumptions made in estimates, 93–5 population by sex and sex ratio, 124–5, population calculations, 80–98 128 population change, 1721–1846, ‘Dutch Studies’, 369 103–105 heirship succession principles, 303 economic development vs population land survey, 77 growth, 132, 145, 158–60 Higashikaiden village, 204–205 Edo Himeji Castle, 21 daimyo¯ visits, 47–8 Horeki famine, 105, 110, 122, 123, Mitsui store, 25, 26 305 Edo Shogunate, 33–4 household size ‘Edo sickness’, 36 Suwa Country, 239–79 Edo, Cotton wholesale dealer, 25 household structures, 69–70 epidemics, 108, 110, 110, 115, 142 household structures – Aizu mountainous Ester Boserup thesis, 113 district EurAsia Project on Population and Family Cambridge Model, 317–18 History, 283 conjugal family units, 316 Europeans, contact with, 5–7 cycle duration and timing, 327–9 Examination of the Sects, The, 6–7 cycle frequencies, 324–7 cycles and patterns, 323–4 family nuclearization, 101 differences by social class, 320–1 family patterns flexibility of heir choice, 338–9 regional differences, 280–302 headship change, reasons, 331–2 family structures heir choice, 332–6 diversity of norms, 297–8 heir choice by class, 336–8 European comparison, 297 household numbers, 315–17 Hayami county, 84–8 household sizes, 315–17 not constant, 281–3 Modified Cambridge Model, 318–20 female infanticide, 128 opportunity for estate transfer, 330–1 fertility stem family households, standard Aizu mountainous district, 312, percentage, 320 314 succession and inheritance, 330–9 Mino province, six villages, 213–17 transfers between categories, 321–3 Northeast Japan, 287 Suwa County, 263–5 ie system, 303 Yokouchi village, 197–201, 271–2, illegitimacy, 177 274 distance from Osaka, 367–8 fishing industry, 31–2 Osaka, 364 food processing industry, 28 overview, 362–8 forestry industry, 28–9 rates, 1887, 1897, 364–6 FRF (family reconstruction forms) statistics, 362–4, 367–8 analysis, 198–201, 205–206, 239 Tokyo, 362–4 urban and rural areas, 366–7 genin, 97, 224 industrial revolution – ‘industrious German statistics, 373–4 revolution’, 64–72

378

12:28:16:11:09 Page 378 Page 379

INDEX

‘industrious revolution’ life-course analysis, 180 industrial revolution, 64–72 ‘little ice age’, 116 Japan, 30, 45 meaning and use, 65 marketing system, 58 peasantry, 60–1 marriage infectious disease, 151 Aizu mountainous district, 309–11 I¯numa village, 204–205 population registers, 177–8 Ishibushi village, 304 Yokouchi village, 193–6, 198–201 Ishigaki village, age distribution, marriage and fertility 80–1 class differences, 204–18 ‘isolation’ policy, 15 marriage duration Mino province, six villages, 210–13 Japan marriage survival rate 16th century, 12–15, 42–51 Mino province, six villages, 212–13 17th century, 16–30, 42–51 married couples per household 18th century, 30–40, 52–63 Suwa Country, 261–2 18th century North-eastern, married proportions 119–31 Suwa Country, 265 ‘closed country’ policy, 50 von Mayr, Gustav, 373, 374 economy-minded society, 42–3 Meirokusha, 371 industrious revolution, 30, 45 mining industry, 29 map of main cities and districts, Mino province, 55–6, 65–9 363 Mino province, six villages social and economic change, 42–51 age at marriage, 207–10 Jinshin Census, 76–7 allocation of families to Classes, 206–207 kabu nakama, 25 births, 213–17 Kai province survey, 371–2, 373 class differences, 204–18 Kanaizawa village, 304 conclusions, 217–18 Kanazawa, 241 fertility, 213–17 Kano village, 204–205 infant mortality, 213–14 Kinai area, 45–6 labor migration and return, kokudaka, 77–80 208–209 population relationship, 88–92, marriage duration, 210–13 96–7 marriage survival rate, 212–13 kokudaka system, 14–15 Miura Baien, 59 economic tax system, 20, 22 Miyamoto Matao, 138 goodwill to neighbours, 20 miyaza, 296–7 Ko¯mi village Mizushina Shichisaburo, 375 age groups, 82–4 modern census, 374–6 Korea, Japanese invasion, 1592, 3 Modified Cambridge Model, 318–20 Koriyama-kamimachi, 122 ‘money-exchanging business’, 25 death rate, 129–31 Mosk, Carl, 138–9 population by sex, 125 Mutsu province, 121 koseki system, 166 population by sex and sex ratio, 124–5, Ko¯zuke province, 153–8 128 Kuwanohara village, 304 Kyoho famine, 110 Nagashino, Battle of, 13 Nakamura, James, 138 land available for use, 101 Nekojishinden village, 204–205 land tax, 49, 56–8 Nihonmatsu domain Laslett, Peter, 239 population by sex, 125 life expectancy population by sex and sex ratio, Edo era, 36–7 124–9

379

12:28:16:11:09 Page 379 Page 380

INDEX

Niremata village population register, changes, 1872–85, 145, 147–9 1638, 169–72 changes by province, 53 Nishi Amane, 370 legally domiciled, 136 Nishijo¯ village, 204–5 migration, 136 age at marriage, 346–7 regional changes, 151–8 age on leaving Dekasegi, 226 regional comparisons, 1822–85, background, 220–3 153–8 Dekasegi, 224 registered, 136 Dekasegi destinations, 226–9 resident, 136 Dekasegi duration, 229–30 ‘stagnation’, 103–105 Dekasegi rate, 225–6 temporary resident, 136 genin, 224 population growth vs economic labor migration, 219–35 development, 132, 145, 158–60 migration, 223–4 population increases and industry growth migration data and other demographic industry growth, 107 measures, 231–4 population registers personal history tracking form, 222 age calculations, 174–5 Tempo¯ famine, 223 availability, 178–9 Tenmei famine, 222–3 births and deaths records, 176–7 textile industry, 229 compilation principles, 171–3 tracing life histories, 225–31 content, 169–75 North, Douglas C, 113 demographic sources, 176–8 Northeast Japan, 285–8 development, 166–9 age at marriage, 285–6 historical background, 165–6 Dekasegi rate, 287–8 household units, 173 fertility, 287 marriages, 177–8 infanticide, 288 migration, 178 prototypical stem family society, 296 names, 173–4 succession by eldest child, 296 occupations, 175 population movement, 175 Oda Nobunaga, 13, 43, 46 problems, 176–8 Okawasegumi, 25 relationships, 174 Okazaki Yo¯ichi, 133, 138 research use, 179–81 Osaka, Dojima rice market, 24 survival, 168–9 Osaka, Kabuki Theatre, 39 system, 109 O¯¯ u region, 305–306 population relationship sex ratio changes, 306–307 kokudaka, 88–92, 96–7 Owari province, 65–9 population statistics, 369–76 17th century, 79–80 Pax Tokugawa, 34–5 Population Statistics of Imperial Japan peasantry (NTJT), 348–9 disarmed, 46 population surveys economically oriented, 48 1721–1846, 114, 119–20 ‘industrious revolution’, 60–1 population trends names, 173 17th century, 99–102 personal history tracking form 17th century urban population growth, Nishijo¯ village, 222 101–102 Philip II, King of Spain, 4 18th century, 52–3, 102–108 Ploughing in Early Spring, 31 19th century, 108–11 population Malthusian trap, 77 18th century changes, 35 Malthusian trap denied, 111 American debate, 137–9 peasant stem family society, 55 by age, 1886, 144 pre-industrial Japan, 113–18 changes, 1822–46, 145–6, 149 Tokugawa period, beginning, 75–98

380

12:28:16:11:09 Page 380 Page 381

INDEX

trends by region that add to zero growth, figures, 1622, 82 107–108 Nishijo¯ village, 219–22 urban decline, 53 overview, 165–84 ‘urban graveyard’ effect, 54–5 reliability, 240 urban population growth, 105–106 Suwa Country, 240 Portugal, first contacts with Japan, 15–16 unreliability in 18th century, principle of comparative advantage, 107 103 proportion marrying Yokouchi village, 186 late 19th century, 349–59, 356–8 silk industry, 27–8, 37, 151–3 regional differences, 354, 359 silver production, 5–6, 58 smallpox vaccination, 110 regional differences Smith, Thomas, 137 age at marriage, 284, 287–8, 345–7, so communes, 12 354, 359 Society of Jesus, 5 causes of diversity, 295–7 Southwest Japan, 290–1 Central region of Japan, 288–90 age at marriage, 290–1 conclusions and cautions, 298–9 ‘culture’ spread by waterborne couples per household, 284, 286 transport, 298 demography caveats, 283–4 Spain, relations with, 4 macro studies, 284–5 stem family system, basis, 303 Northeast Japan, 285–8 Sugi Koji, 370–3, 375–6 persons per household, 284, 285 superstitious prejudice against some proportion marrying, 354, 359 numbers, 80 proportion of working-age population in Suwa County household, 296 38 communities analysed, 241–2 simulation of demographic and family age at marriage, 265–7 patterns, methods, 291–3 birth rate, 263–5 simulation of demographic and family child mortality, 266 patterns, results, 293–5 data handling and analysis, 241–2 Southwest Japan, 290–1 data tables, 243–51 Registration of Men and Domestic Animals in family structure, 281–3 Kokura Domain, The, 79–80 fertility, 263–5 ‘religious inquisition system’, 19 four geographical districts, 241–2 Rest after the good harvest, 33 household, definition, 251 Rice Planting in Late Spring, 32 household size, 239–79 ritsuryo¯ system, 298 life expectancy, 266 ryotei, 38 map, 242 married couples per household, 261–2 Samurai married proportions, 265 control of, 20 mean household size (MHS), social power, 22 comparisons with England, 254–6 sankin-ko¯tai, 47 mean household size (MHS), Sasaki Yoichiro, 106 correlation coefficient analysis, Satsuma, 46 256–61 Satsuma Rebellion, 151 mean household size (MHS), School of Statistics, 372–3, 375, distribution, 252–6 376 mean household size (MHS), Sengoku daimyo¯, 12–13, 45–6 population change, 269–70 sex ratio, definition, 125–6 mean household size (MHS), definition, Shimotsuke province, 153–8 251 Shumon Aratame Cho¯, 6–7, 240, 303 population trends by district, shu¯ mon-aratame-cho¯ 1671–1860, 243 age calculations, 240 resident kin per household, 262–3 Aizu mountainous district, 303 servants, 267–8

381

12:28:16:11:09 Page 381 Page 382

INDEX

shu¯ mon-aratame-cho¯, 240 Umemura Mataji, 133–4, 138 variables over district, 268–9 ‘urban graveyard eﬀect’, 289 variables over time, 268–9 Yokouchi village analysis, 270–8 Vital Population Statistics of Imperial Japan Suwa district, 69–70 1898 (NTJD), 349

Table of Households and Population of Warring States period, 12 Imperial Japan (NTMK), 348 Wilkinson, R G, 113 Table of Households and Population for Towns Wrigley, E A, 106, 114, 234 (TMK), 349 tanomoshi-ko¯, 296–7 Xavier, Francisco, 5 temporary migration, 36, 56 Tempo¯ famine, 108, 110, 223 Yamamura, Kozo, 106, 137 Tempo¯ mortality crisis, 142–4 Yokouchi village Tenmei famine, 54, 105, 110, 120, 122, adoption of young children, 272–3 222–3, 305 age at marriage, 193–6, 271, Todd, Emmanuel, 297 274 Tohoku village, age at marriage, age groups, 82–4 347 age structure, 191–2 Tokugawa Ieyasu, 3, 6, 44, 46 analysis, 270–8 Tokugawa rule, 17–30 birth rates, 191–3 agricultural developments, 115–17 child mortality, 273 discrimination, 39–40 death rates, 191–3 economic change, 44–6 demography, 185–203 economic policy, 59–60 exodus of population, 196–8, four classes of society, 48–9 200 ‘Great Transformation’, 44–6 family size, 189 industries development, 27–30 fertility, 197–201, 271–2, 274 international negotiations, 40 FRF (family reconstruction forms) market economy, 26, 27 analysis, 198–201 market economy diagram, 57 general demographic trends, 187–91 mass culture, 44 households, 188–9 overview, 61–2 life expectancy, 193–5 population at the beginning, 75–98 map, 187 population increase, 23 marriage, 193–6, 198–201 population overview, 114–15 mean size of households (MSH), trends, Tokugawa Shogunate, 13–14 273, 275–6 Tomono-machi, 241 proportion married, 190 Tordesilhas, Treaty of, 5 shu¯ mon-aratame-cho¯, 186 Toyotomi Hideyoshi, 3–4, 6, 13, 43, successor households, 276–8 46 use as national model, 201–202 Transport, 18th century, 38 Yonezawa, 121 travel industry, 37–8 population by sex and sex ratio, 126, Tsuda Mamichi, 370 129 To¯nosu village, 304 Yoshida Togo, 77–8

382

12:28:16:11:09 Page 382 Page 383

12:28:16:11:09 Page 383 Page 384

12:28:16:11:09 Page 384