ANTHROPOLOGICAL SCIENCE Vol. 114, 99–115, 2006

Ethnogenesis and craniofacial change in Japan from the perspective of nonmetric traits NANCY SUZANNE OSSENBERG1†*,YUKIO DODO2,TOMOKO MAEDA2, YOSHINORI KAWAKUBO2

1Department of Anatomy and Cell Biology, Queen’s University, Kingston, Ontario, K7L 3Y9 Canada 2Department of Anatomy and Anthropology, Tohoku University School of Medicine

Received 25 July 2003; accepted 19 December 2003

Abstract To examine population affinities in light of the ‘dual structure model’, frequencies of 21 nonmetric cranial traits were analyzed in 17 prehistoric to recent samples from Japan and five from continental northeast Asia. Eight bivariate plots, each representing a different bone or region of the skull, as well as cluster analysis of 21-trait mean measures of divergence using multidimensional scal- ing and additive tree techniques, revealed good discrimination between the Jomon-Ainu indigenous lineage and that of the immigrants who arrived from continental Asia after 300 BC. In Hokkaido, in agreement with historical records, Ainu villages of Hidaka province were least, and those close to the Japan Sea coast were most, hybridized with Wajin. In the central islands, clines were identified among Wajin skeletal samples whereby those from Kyushu most resembled continental northeast Asians, while those from the northernmost prefectures of Tohoku apparently retained the strongest indigenous heritage. In the more southerly prefectures of Tohoku, stronger traces of Jomon ancestry prevailed in the cohort born during the latest Edo period than in the one born after 1870. Thus, it seems that increased inter-regional mobility and gene flow following the Meiji Restoration initiated the most recent episode in the long process of demic diffusion that has helped to shape craniofacial change in Japan.

Key words: ethnogenesis in Japan, Tohoku, Ainu, cranial nonmetric traits

Introduction Kyushu and westernmost Honshu. Here, not only were the climate and terrain hospitable to rice paddy agriculture but The hybridization model of Japanese population history also, coincidentally, the indigenous population was sparser now widely accepted was proposed almost eight decades than in the northeast where the ecosystem was more produc- ago by Kiyono (1926, cited by Yamaguchi, 1982) and subse- tive for hunting and foraging. The precise nature of prehis- quently revitalized by the research of Yamaguchi (1982) and toric contact with the native people has yet to be determined. his students. With additional supporting data and a detailed In any case, contact between the two groups was inevitable outline, it is now known as the ‘dual structure model’ as the successive generations descended from continental (Hanihara, 1991). According to this model, Japan’s people immigrants expanded northeastward along the archipelago. today embody two ancestral lineages. First came the Jomon: During the subsequent Kofun period, 300–700 AD, the the Neolithic foraging, hunting, and fishing groups who population in western and central Japan grew. Reinforced by inhabited the archipelago from about 10,000 BC. The new immigrants and influences from the Asian continent, second lineage is that of the Yayoi, who arrived from conti- these communities became increasingly integrated by a nental northeast Asia around 300 BC, bringing rice agri- social and political system centered on the imperial court in culture and metal-working technologies, along with a the Kinki district. By the early 8th century under a policy different language and culture. Derived from widely diver- enacted by the imperial court, Wajin influence had spread gent branches of the great Mongoloid racial tree, these two northeast to the Kanto and neighboring regions. However, ancestral populations have contributed profoundly and during the proto-historic Kofun and early historic periods, indelibly to the unique cultural and genetic characteristics of Hokkaido and the Tohoku region of Honshu were still the the Japanese. territory of the so-called Ezo, Emishi or ‘barbarians’ The Yayoi people established the first immigrant commu- (Hanihara, 1990); i.e. the non-Wajin descendents of the nities close to their ports of entry in the region of northern Jomon who were still following a lifestyle based on hunting, fishing, and foraging. Eventually, by stages, these territories † Present address: 203 Victoria St, Kingston, ON K7L 3Y9, Canada too were subjugated and colonized. * Corresponding author. e-mail: [email protected] Studies of skeletal remains have documented large differ- phone: +1-613-542-5620 ences between the two founding groups in stature, limb bone Published online 30 September 2004 proportions, and morphological features of the skull and in J-STAGE (www.jstage.jst.go.jp) DOI: 10.1537/ase.00090 teeth. Regional variations, in a somewhat mosaic pattern

© 2004 The Anthropological Society of Nippon 99 100 N.S. OSSENBERG ET AL. ANTHROPOLOGICAL SCIENCE with respect to degree of similarity to one or the other found- ern portion of the main islands (present-day Kyushu, ing group, were manifest during the Yayoi period. Subse- Shikoku, Chugoku and Kinki districts) provided the West quently, the population of Japan seems to have become Jomon skeletal sample (n = 102 individuals). The Central increasingly homogeneous over time. In particular, nonmet- Jomon sample (n = 69) is from sites in present-day Tokai ric cranial studies (Dodo and Ishida, 1990; Dodo et al., and Kanto districts, while North Jomon (n = 60) sample 1992) emphasize the contrast between the regional and tem- derives from Tohoku district. The other two samples are poral homogeneity over the past 2000 years among popula- Hokkaido Jomon (n = 34) and Epi-Jomon (n = 25). Jomon tions of the central islands of Japan, and the highly remains are curated at several institutions. Those reported significant differences between the latter and the Jomon. here were studied in the Department of Anatomy, Sapporo Nevertheless, regional variations among populations of the Medical University; Department of Anatomy and Anthro- central islands are recognizable throughout all post-Jomon pology, Tohoku University School of Medicine; The Univer- periods, and persist even to the present day with respect to sity Museum, The University of Tokyo; National Science genetic polymorphisms in the blood and other characteris- Museum, Tokyo; and the Laboratory of Physical Anthropol- tics. ogy, Faculty of Science, Kyoto University. Significantly, these traits appear to vary regionally in a Ainu skeletons examined for this study were those exca- clinal pattern similar to those shown in other parts of the vated by Koganei (1893, 1894) from Edo period cemeteries world to have been generated by differential gene flow due associated with abandoned villages throughout Hokkaido. to demic diffusion (Weiss, 1988), occurring when an immi- The series was studied at The University Museum, The Uni- grant people expands into an already occupied territory. In versity of Tokyo. Most investigators have used this series as the case of Japan, a gradient of affinities would have resulted a single sample to represent relatively pure Ainu, i.e. mini- from differential gene flow as descendants of the immigrant mally affected by Wajin admixture. Here, however, because population, the Wajin Japanese, expanded from western of our focus on northern Japan and in light of historical doc- Japan northeastward along the archipelago, assimilating or, uments (Kodama, 1970) we partitioned the Koganei series perhaps in some cases being assimilated by, the aboriginal three ways: Southeast Ainu (Hidaka and Tokachi provinces: inhabitants. The genetic contribution from continental Asian n = 36), Northeast Ainu (Kushiro, Nemuro, Abashiri and ancestors is strongest in people of northern Kyushu and Soya provinces as well as Kunashiri Island: n = 50), and western Honshu and diminishes to the northeast, so that West Ainu (Rumoi, Ishikari, and Shiribeshi provinces: among the present-day people of Japan it is the Ainu of n = 31). Hokkaido who preserve the strongest genetic heritage from Wajin Japanese samples used in this study were defined the Jomon. both regionally and temporally. The boundaries of West, Clines generated by ancient demic expansion are demon- Central, and North regions are identical to those as defined strated to be remarkably stable (Cavalli-Sforza, 1986). The above for the Jomon. However, because of our focus on dual structure model also assumes that the dual ancestry of northern Japan we wanted a more fine-grained analysis of the Japanese people is evident even today, and that intermix- Tohoku than that attempted in previous investigations. ture is still going on. Accordingly, Tohoku was partitioned into South In view of the history of events related to the expansion of (Fukushima, Yamagata, and Miyagi prefectures) and North Wajin influence, we expected that the most northerly regions (Akita, Iwate, and Aomori). of Japan would reveal particularly strong evidence in sup- From about 1880 and until about 1960 universities in port of the dual structure model. Therefore, this investiga- Japan assembled (as did those in Europe and North Amer- tion focused on Tohoku and Hokkaido with the primary ica) collections of skeletons from the dissecting-room objective of determining whether or not the pattern of tem- subjects used to teach medical students. When carefully pre- poral and regional population affinities within northern pared and curated along with records giving for each indi- Japan revealed by nonmetric cranial traits fits the model. vidual the place and date of birth as well as date and cause of Secondly, we aimed to examine the data for evidence of a death, these collections are a priceless resource for many southwest to northeast gradient of affinities as predicted by kinds of clinical and basic osteological research. In general, the model. Thirdly, within a broader comparative frame- anthropological research uses such dissecting-room series to work, we hoped to corroborate the findings of other studies represent simply ‘modern’ or ‘recent’ Japanese. However, concerning relationships of Japan’s two ancestral lineages to considering that with the increased mobility of the popula- other northeast Asians. tion following the Meiji Restoration in 1868 microevolu- tionary change is ongoing in Japan even today, and that Materials during all periods Tohoku was impacted by Wajin gene flow later than were other regions of the central islands, we Japan deemed it worthwhile to separate the Tohoku University dis- Jomon samples were derived from many sites throughout secting-room series into two cohorts according to birth date. the central islands of the archipelago as well as Hokkaido. Accordingly, the South Tohoku Edo sample (n = 79) repre- These are middle to final Jomon period sites with dates sented individuals with birth dates between 1820 and 1870, ranging from about 3500 BC to 300 BC. Epi-Jomon sites in i.e. towards the very end of the Edo period. The South Hokkaido are later, roughly contemporaneous with Yayoi Tohoku Recent sample (n = 73) comprised those born 1871 and Kofun periods, 300 BC to 700 AD. For this study the or after, i.e. at least three years after the beginning of the sites were aggregated regionally. Those located in the west- Meiji Restoration. Most of these persons were born in Miy- Vol. 114, 2006 ETHNOGENESIS IN JAPAN 101 agi prefecture. (Yamaguchi, 1967, 1977, 1982, 1985; Yamaguchi et al., The North Tohoku Edo sample (n = 42), curated in the 1973; Dodo, 1974, 1975, 1986a, b, 1987; Okada and Department of Anatomy and Anthropology at the Tohoku Yamaguchi, 1975; Mouri, 1976, 1988, 1996; Yamano and University School of Medicine, was comprised of skeletons Yamaguchi, 1976; Dodo and Ishida, 1987, 1988, 1990, excavated from Edo period graves at various sites in the 1992; Katayama, 1988; Ishida, 1990, 1993, 1995, 1996; three northern prefectures. Birth dates of these individuals Ishida and Dodo, 1990, 1992, 1993; Ishida and Kida, 1991; are unrecorded, but would perhaps fall in the range 1700– Dodo et al., 1992, 1998; Hanihara et al., 1998a, b; Ishida and 1800 AD, i.e. several generations earlier than the very-late- Kondo, 1999; Hanihara and Ishida, 2001a, b, c, d; Dodo and Edo dissecting-room subjects. Kawakubo, 2002), and those by foreign scholars Central Wajin: The Medieval period sample (n = 70) was (Pietrusewsky, 1984; Ossenberg, 1986, 1992a, b; Kozintsev, from two 14th-century sites in the present-day city of 1990, 1992a, b), the validity of minor present-or-absent Kamakura believed to represent the mass burials of warriors cranial features for addressing problems concerning ethno- killed in battle. The Kamakura sample is predominantly genesis in Japan within a circum-Pacific context is now young adults, 44 men and 26 women, and is curated at The firmly established. University Museum, The University of Tokyo. The Central The present investigation was based on a battery of traits Edo sample (n = 117) is comprised mainly of crania from the developed by the senior author over 40 years of research Joshinji Cemetery dating from approximately 1650 to 1850 focused mainly on the peopling of northwestern North AD (Mizoguchi, 1997), and is curated at the National Sci- America. This trait-list differs from those used by Japanese ence Museum. The dissecting-room series of skeletons at workers; for example, of 23 features routinely scored by Chiba University was partitioned according to recorded date Ishida and Dodo (1992, Table 1) only nine are common to of birth in the same way as for Tohoku, and data for the very- both trait-lists. Note that Ossenberg does not collect data for late-Edo cohort of 29 individuals were combined with those four of the 10 traits shown by Dodo and Ishida (1990) to be for the Joshinji Cemetery. The Central Recent sample most highly discriminatory between the Jomon-Ainu and (n = 88) was comprised of dissecting-room subjects born Wajin lineages: ossicle at lambda, biasterionic suture trace, after 1870. foramen of Vesalius, and medial palatine canal. On the other West Wajin: These three samples were studied at the hand, her battery includes a number of informative features museum of the Faculty of Medicine, Kyushu University. The not routinely scored by other osteologists. Therefore, it was Yayoi sample (n = 31), dating from about 100 AD, was hoped that the findings reported here would complement the derived mainly from the Doigahama site in westernmost many previous studies of ethnogenesis in Japan based on Honshu, but with a few crania also from the northern this category of morphological evidence. With exceptions Kyushu Kanenokuma site. The West Edo sample (n = 45) noted in footnotes to Table 1, all observations were made by was retrieved from the 18th-century Tenpukuji cemetery. N.S.O. during the period September 2001–March 2002. The West Recent sample (n = 58) of dissecting-room skele- Table 1 lists the 21 features according to eight regions of tons was not partitioned by birth cohort and so encompasses the skull, and gives their frequencies in 22 cranial samples as a slightly greater temporal depth than do the other two recent well as in Jomon, Ainu, and Wajin aggregated samples. Data period samples. for adults of both sexes were pooled and for certain features juvenile data also were included. Descriptions of the features Continental northeast Asia and criteria for selecting and scoring them have been docu- Included in our study were five samples representing con- mented in earlier reports (Ossenberg, 1970, 1976, 1994, tinental northeast Asians. Mongols (n = 40) from Urga 2004); however, brief explanatory notes concerning certain (Ulaanbaatar) were studied at the United States National features may be useful here. Museum. Chinese (n = 72) from northeastern China (for- Occipito-mastoid bone ratio is the number of sides (left merly Manchuria) are curated at The University Museum, plus right) with a supernumerary ossicle in the occipito-mas- The University of Tokyo. The Tungus, a sample comprised toid suture divided by the total of: the number of sides with of individuals from the Ulchi, Negidal, and Evenki tribes an occipito-mastoid bone, plus the number of sides with an (n = 83) and the Chukchi (n = 44) were studied at the Insti- asterionic bone, plus the number of sides with a parietal tute of Ethnography of St Petersburg. These four series are notch bone. from the 17th century or later. An Okhotsk sample (n = 36), Of the three variants involving facial sutures, transverso- representing a prehistoric sea-mammal hunting people zygomatic suture trace is the most familiar, long recognized whose origins lay perhaps in the Amur region but who lived as one of the most powerful traits for circum-Pacific popula- for a time on the north coast of Hokkaido (Ishida, 1996), tion studies. Note that frequencies in Table 1 are higher and were studied at the Department of Anatomy, Sapporo Medi- show greater dispersion than those of Dodo and Ishida cal College. because N.S.O. counts as ‘trait present’ any discernible trace, whereas her colleagues count only cases 5 mm or Methods more in length. Infraorbital suture variant (Figure 2) and orbital suture variant are described by Kozintsev (1992b) as Cranial nonmetric traits two of the five traits he has found most valuable for ethnoge- Since the 19th century, Japanese scholars have been at netic research on a world-wide scale. In the dissecting-room the forefront of research on nonmetric skeletal traits. Owing series the number of observations of the former variant were to their investigations, especially since about 1970 limited by age-regressive obliteration of the infraorbital 102 N.S. OSSENBERG ET AL. ANTHROPOLOGICAL SCIENCE

Table 1. Side incidence of nonmetric traits in 22 cranial series from Japan and continental northeast Asia Jomon Ainu Westa Central North Hokkaido Epi-Jomon South- North- West east east n % n % n % n % n % n % n % n % Sutural bones Occipito-mastoid bone 125 17.6 55 32.7 70 31.4 34 20.6 23 13.0 67 17.9 97 18.6 62 17.7 Occipito-mastoid bone ratio 11722.26043.38332.63826.31822.22646.25036.04126.8 Facial sutures Transverso-zygomatic suture trace 141 85.8 64 71.9 84 85.7 33 84.9 20 70.0 58 70.7 78 70.5 56 50.0 Infraorbital suture variant 27 0 32 6.3 44 4.6 22 0 18 0 53 11.3 47 21.3 51 25.5 Orbital suture variant 5 40.0 7 0 9 11.1 7 14.3 9 0 62 12.9 79 12.7 55 14.6 Frontal bone Supraorbital foramen 165 13.9 81 12.4 110 10.0 52 13.5 35 2.9 71 26.8 98 18.4 59 20.3 Frontal grooves 155 20.7 65 44.6 87 31.0 53 14.2 31 16.1 67 17.9 92 17.4 58 15.5 Trochlear spur 126 0.8 64 4.7 91 2.2 48 0 33 3.0 71 9.9 97 4.1 60 5.0 Occipital bone Postcondylar canal absent 102 19.6 42 4.8 61 14.8 34 17.8 29 20.7 70 17.1 95 11.6 64 15.6 Lateral condylar canal 60 10.0 32 12.5 37 18.9 25 48.0 26 38.5 53 30.2 91 30.8 61 29.5 Hypoglossal canal bridged 142 13.4 73 15.1 82 18.3 45 8.9 38 5.3 70 18.6 98 25.5 64 25.0 Pharyngeal fossa 68 23.5 36 19.4 43 14.0 23 13.0 17 11.8 34 20.6 50 10.0 32 25.0 Temporal bone Marginal foramen of tympanic plate 139 5.0 93 3.2 89 1.1 50 2.0 42 4.8 69 4.4 96 12.5 63 9.5 Tympanic dehiscence 178 35.4 100 23.0 110 20.9 54 20.4 43 9.3 68 14.7 96 13.5 63 27.0 Sphenoid bone Pterygobasal spur or bridge 142 19.0 75 2.7 90 7.8 49 8.2 31 16.1 72 23.6 95 14.7 64 37.5 Clinoid bridge 69 7.3 30 6.7 30 10.0 18 5.6 22 4.6 67 17.9 90 7.8 58 5.2 Accessory optic canal 62 1.6 19 0 10 0 11 0 17 0 65 0 88 3.4 57 1.8 Mandible Accessory mental foramen 177 17.5 104 15.4 102 22.6 58 12.1 40 22.5 52 26.9 78 12.8 48 18.8 Mylohyoid bridge 164 11.0 96 9.4 94 17.0 50 34.0 40 22.5 44 13.6 78 9.0 44 6.8 Teeth Upper third molar suppressed 112 9.8 50 16.0 62 21.0 36 2.8 34 8.2 56 21.4 61 27.9 46 41.3 Three-rooted lower first molar 153 3.3 91 0 93 8.6 49 0 41 14.6 48 4.2 58 3.5 37 8.1 All observations were recorded by the senior author in 2001 and 2002 except as indicated in the following footnotes : a Data recorded for the Tsukumo site in 1981 were combined with those from the 2001/2 survey. suture; while in archaeologically retrieved remains, espe- varies from 1.5 (zero frequency) to −1.5 (100%), but in prac- cially Jomon, few observations of the orbital suture variant tice falls short of these extremes owing to a factor correcting were possible owing to post-mortem damage to the fragile for sample size. orbital walls. Multivariate biological distances between samples were Figures for supraorbital foramen in Table 1 are frequen- assessed by the MMD statistic, which is essentially a mean cies weighted by counting twice a left or right side with two sum of squares, over all traits, of the θ differences between or more foramina, and hence tend to be higher and show two samples. An MMD is regarded as larger than zero at the greater dispersion than those reported by other workers. 5% level of statistical significance when it exceeds twice its Although not customary in cranial nonmetric studies, the standard deviation (Sjøvold, 1977). In each of the bivariate list includes two relatively easy-to-score dental variants: plots (Figure 1, Figure 2) the θ scale for each trait was tai- upper third molar suppressed and three-rooted lower first lored to its own range. However, in contrast to Mahalanobis’ molar. These are among the key features of the Sinodont D employed as a distance statistic in craniometric analysis, dental complex (Turner, 1990). which equalizes the range of a large feature such as maxi- mum cranial breadth and a small feature such as nasal Analytic methods breadth in terms of its contribution to D, MMD does not For each of eight skull regions a bivariate plot was con- equalize the effective range from one trait to another. This structed in order to examine the consistency among the dif- means that a common feature such as supraorbital foramen ferent traits and skull regions with respect to pattern of with a frequency range of 3–88% (Table 1) contributes pro- population relationships. The value plotted for each trait was portionately more to the MMD than does a rare trait such as not the raw frequency but rather the arcsine value, θ, an trochlear spur with a frequency range of 0–15%. This might angular transformation which serves to stabilize the variance be seen as a weakness or a strength of MMD according to of the frequencies. As defined in the modified formula for one’s theoretical position as to how features should be Smith’s mean measure of divergence (MMD) statistic weighted taxonomically. Another difference between D and (Sjøvold, 1977), θ theoretically (i.e. in very large samples) MMD is that the latter incorporates no correction factor for Vol. 114, 2006 ETHNOGENESIS IN JAPAN 103

Table 1. (continued) Japanese: North Japanese: Central Japanese: West North South South Kamakura Edo Recent Yayoi Edo Recent Tohoku Tohoku Tohoku Edo Edo Recent n % n % n % n % n % n % n % n % n % Sutural bones Occipito-mastoid bone 41 14.6 151 11.3 135 7.4 99 9.1 219 17.8 165 15.8 55 23.6 81 4.9 113 10.6 Occipito-mastoid bone ratio 29 27.6 68 25.0 53 18.9 54 18.5 141 27.7 66 39.4 49 26.5 23 17.4 54 22.2 Facial sutures Transverso-zygomatic suture trace 27 48.2 139 39.6 125 44.0 97 32.0 220 32.3 166 34.9 50 42.0 62 17.7 101 36.6 Infraorbital suture variant 29 20.7 122 29.5 120 25.8 99 28.3 168 34.5 124 21.8 43 37.2 52 44.2 94 34.0 Orbital suture variant 31 32.3 148 21.0 140 27.9 61 19.7 214 38.3 166 41.6 47 27.7 66 25.8 110 39.1 Frontal bone Supraorbital foramen 64 45.5 158 39.9 125 52.8 116 45.7 229 43.2 174 44.3 61 50.8 88 47.7 107 39.3 Frontal grooves 36 22.2 146 28.1 123 29.3 109 22.0 232 32.8 172 27.3 55 20.0 66 34.9 112 26.8 Trochlear spur 52 7.7 158 7.6 138 15.2 105 7.6 231 9.5 174 7.5 53 15.1 82 6.1 113 11.5 Occipital bone Postcondylar canal absent 46 30.4 158 18.4 146 21.2 81 42.0 229 29.3 174 27.0 56 33.9 83 28.9 116 33.6 Lateral condylar canal 39 18.0 157 18.5 141 25.5 59 25.4 215 27.4 172 24.4 52 23.1 84 10.7 104 21.2 Hypoglossal canal bridged 54 11.1 158 12.0 146 11.6 86 9.3 231 7.8 172 12.2 58 8.6 86 12.8 116 15.5 Pharyngeal fossa 27 11.1 79 6.3 71 9.9 54 7.4 117 10.3 88 4.6 31 3.2 44 15.9 58 13.8 Temporal bone Marginal foramen of tympanic plate 62 8.1 151 17.9 135 18.5 127 10.2 229 14.0 176 13.6 43 9.3 78 18.0 114 14.9 Tympanic dehiscence 67 26.9 149 26.2 140 31.4 131 21.4 234 30.3 176 21.6 55 10.9 87 28.7 111 32.4 Sphenoid bone Pterygobasal spur or bridge 68 19.1 158 12.0 143 9.1 117 8.6 231 14.7 176 19.3 62 19.4 88 8.0 116 18.1 Clinoid bridge 45 6.7 154 2.6 140 5.0 51 7.8 223 6.7 176 2.8 40 2.5 81 2.5 114 7.0 Accessory optic canal 22 4.6 146 4.8 129 7.8 41 0 211 4.3 171 2.3 35 0 65 6.2 107 6.5 Mandible Accessory mental foramen 58 8.6 136 11.8 124 11.3 74 13.5 222 17.6 152 14.5 58 12.1 78 12.8 97 6.2 Mylohyoid bridge 58 3.5 135 4.4 100 5.0 72 5.6 222 2.3 148 2.7 56 3.6 77 13.0 90 5.6 Teeth Upper third molar suppressed 44 34.1 82 48.8 107 39.3 93 34.4 171 39.8 141 39.0 50 68.0 46 45.7 97 41.2 Three-rooted lower first molar 44 15.9 64 12.5 89 16.9 67 14.9 159 25.2 98 24.5 47 19.2 48 25.0 75 21.3

intertrait correlation. However, with the exception of super- ration of Jomon from Wajin owing to the lower frequencies numerary ossicle traits (wormian bones in various vault of occipito-mastoid bone in the latter. Ainu tend to be inter- sutures), intertrait correlation among nonmetric traits is gen- mediate. Regional variation among Jomon seems to be as erally shown to be either absent or so low as to have little great as among historic period Japanese. In the central effect on the MMDs (Saunders, 1989). region of the plot there is considerable overlap involving The triangular matrix containing MMDs for all pairwise Epi-Jomon, West Ainu and Central Edo, with North Tohoku population comparisons was analyzed by two different clus- Edo also close to this region. Yayoi seems misplaced, sitting tering techniques in order to produce two-dimensional conspicuously in the middle of the Jomon cluster. arrays of the samples. The clustering programs cannot han- The facial sutures (Figure 1b, Figure 3) give excellent dle negative numbers, so to prepare the data for entry into separation of Jomon from Wajin, with Ainu clearly interme- the multidimensional scaling program ALSCAL (SPSS for diate. Among the Ainu the Southeast group is closest to Windows, version 9.0.0), negative MMD values were elimi- Jomon, while West Ainu is closest to the Wajin cluster. nated by raising all the MMDs by the amount of the largest Again, as with the sutural bones the Wajin samples closest to negative value in the matrix. For analysis with the program the Jomon-Ainu lineage are those from the Tohoku region, ADDITIVE TREE (SYSTAT for Windows, version 7.0, especially North Tohoku Edo. A striking feature in this dia- SPSS), all negative MMDs were simply converted to zero. gram is the small cluster of the three Recent Wajin samples within the larger one containing all Wajin. Results Frontal bone traits (Figure 1c, Figure 4) provide excellent separation between Jomon and Wajin, with Ainu intermedi- Bivariate plots ate and not overlapping. The dispersion of samples within The bivariate plots are shown in Figure 1 and Figure 2. each cluster is considerable. In these respects the plots in Table 2 is the triangular matrix of MMDs, and Figure 5, Fig- Figure 1a, Figure 1b and Figure 1c are consistent. However, ure 6, Figure 7 and Figure 8 are the cluster diagrams based closer inspection shows that each pair of traits produces a on those MMDs. somewhat different arrangement of samples within the clus- The sutural bones plot (Figure 1a) show fairly good sepa- ters; for example, the facial sutures place Southeast Ainu 104 N.S. OSSENBERG ET AL. ANTHROPOLOGICAL SCIENCE

Table 1. (continued) Japan: aggregated samples Northeast Asia Jomon Ainub Japanese Mongolsc Chinese Tungusd Okhotsk Chukchid n % n % n % n % n % n % n % n % Sutural bones Occipito-mastoid bone 284 24.3 226 18.1 1083 12.7 73 15.1 131 15.3 137 14.6 37 24.3 73 24.7 Occipito-mastoid bone ratio 298 29.9 117 35.0 544 25.9 39 28.2 74 27.0 74 29.7 31 45.2 53 34.0 Facial sutures Transverso-zygomatic suture trace 322 82.9 192 64.6 1009 36.2 74 31.1 136 30.9 153 30.7 41 58.5 59 15.3 Infraorbital suture variant 125 3.2 151 19.2 868 31.0 62 40.3 119 55.5 145 60.0e 23 69.6 83 33.7e Orbital suture variant 28 14.3 196 13.3 1007 32.6 76 46.1 140 35.7 180 13.9e 23 34.8 74 23.0e Frontal bone Supraorbital foramen 408 12.5 228 21.5 1146 44.9 77 66.2 143 62.9 160 88.1 61 49.2 84 82.1 Frontal grooves 360 26.4 217 17.1 1073 28.0 78 29.5 138 36.2 157 14.7 59 18.6 82 14.6 Trochlear spur 329 1.8 228 6.1 1130 9.6 77 6.5 144 5.6 158 7.6 49 0 74 1.4 Occipital bone Postcondylar canal absent 239 15.5 229 14.4 1113 27.9 77 29.9 144 27.1 160 36.9 48 22.9 84 21.4 Lateral condylar canal 154 18.8 205 30.2 1047 22.1 78 26.9 141 22.7 144 30.6 42 21.4 60 23.3 Hypoglossal canal bridged 342 14.3 232 23.3 1131 11.0 78 9.0 143 11.1 169 11.3 49 18.4 85 14.1 Pharyngeal fossa 170 18.8 116 17.2 581 9.1 39 15.4 72 4.2 81 16.1 25 16.0 43 11.6 Temporal bone Marginal foramen of tympanic plate 371 3.2 228 9.2 1139 14.1 78 19.2 150 10.0 158 13.3 63 9.5 77 3.9 Tympanic dehiscence 442 27.2 227 17.6 1174 26.5 78 19.2 134 26.9 158 16.5 67 22.4 78 34.6 Sphenoid bone Pterygobasal spur or bridge 356 11.2 231 23.8 1183 14.1 78 10.3 144 10.4 166 3.0 49 2.0 86 2.3 Clinoid bridge 147 7.5 215 10.2 1047 4.8 78 7.7 144 6.9 164 0.6 31 12.9 86 10.5 Accessory optic canal 102 1.0 210 1.9 947 4.4 78 6.4 135 7.4 162 10.5 25 4.0 87 9.2 Mandible Accessory mental foramen 441 17.5 178 18.5 1023 12.8 78 15.4 110 16.4 124 15.3 44 18.2 23 13.0 Mylohyoid bridge 404 14.9 166 9.6 982 4.5 78 1.3 106 1.0 119 8.4 42 11.9 22 18.2 Teeth Upper third molar suppressed 260 12.7 163 29.5 851 41.6 73 34.4 116 21.6 133 48.1 41 24.4 61 14.8 Three-rooted lower first molar 386 3.4 143 4.9 710 19.9 67 28.4 100 27.0 97 21.7 46 26.1 17 41.2 All observations were recorded by the senior author in 2001 and 2002 except as indicated in the following footnotes : b The Koganei Ainu series was recorded in 1981 (Ossenberg, 1986), and again in 2001/2 for analysis of intra-observer replicability in scoring. The latter data were used. The MMD between the two scoring sessions was 0.0005 (standard deviation 0.003), statistically insignificant. c Recorded at the United States National Museum in 1995. d Recorded at the Institute of Ethnography, Leningrad in 1981. e These data were recorded by A. Kozintsev and kindly sent by him (pers. comm. 1992) to Ossenberg. closest to Jomon, whereas frontal bone traits place West Jomon and Ainu, but separates the Jomon-Ainu lineage Ainu in that position. fairly well from the Wajin. The three Recent Wajin samples It is interesting that trochlear spur with its small range of form a cluster within the larger one encompassing Wajin of frequencies, 0–15%, is as good a Jomon-Wajin discriminator all time periods. In Japan, postcondylar canal absent as is supraorbital foramen with its large range, 3–53%. appears to be a better discriminator than pharyngeal fossa. However, as noted above the MMD formula does not take The configuration of samples produced by the sphenoid this into account, but automatically gives the latter feature bone (Figure 2b) looks peculiar. This is mainly because greater weight than the former. accessory optic canal with frequency ranging from 0 to 8% In the temporal bone plot (Figure 1d) marginal foramen, is such a rare trait. In seven of the 17 samples (Table 1) not which is found significantly less frequently in the Jomon, one individual exhibited this anomaly. If these samples defines the three clusters nicely along the y-axis. Among the (Yayoi, Kamakura, Southeast Ainu, and four Jomon) had Ainu, the Southeast group is once again closest to Jomon. been very large, yet still produced no case of this trait, then West Ainu is closest to Wajin and, in fact, this sample, along all seven would have been located at θ=1.5 (0%) at the ori- with North Tohoku Edo and Kamakura, is in the region of gin of the y-axis. The actual dispersion of these samples overlap between Ainu and Wajin clusters. The three Recent along the y-axis simply reflects the correction for sample Wajin samples form a smaller cluster within the one contain- size in the calculation of θ (Sjøvold, 1977). Another pecu- ing all Wajin. The small Epi-Jomon and Yayoi samples are liarity is the placement of Southeast Ainu in the lower right outliers in this plot. Note that tympanic dehiscence, though a corner of the diagram, apart from all the other samples, well-known feature and included in most workers’ trait-lists, because of its zero frequency of the former trait together seems to provide, at least in this analysis, no coherent picture with the highest frequency, 18%, of clinoid bridge. As in the of population affinities in Japan. occipital bone plot, Jomon and Ainu overlap; but Jomon- The occipital bone diagram (Figure 2a) overlaps for Ainu are somewhat distinguished from Wajin, the latter gen- Vol. 114, 2006 ETHNOGENESIS IN JAPAN 105

Figure 1. Bivariate distribution of population samples according to θ values of nonmetric trait frequencies (see Analytic methods). (a) Sutural bones; (b) facial sutures; (c) frontal bone; (d) temporal bone. Jomon: open circles; Ainu: black circles; Yayoi, Kamakura, Edo: grey triangles; Recent Wajin: black triangles. N: South Tohoku; NN: North Tohoku. erally having higher frequencies of accessory optic canal Ainu the Southeast is closest to Jomon, the West closest to and lower frequencies of clinoid bridge than does the indig- Wajin. Among Wajin the Tohoku samples as well as enous lineage. Kamakura are the ones closest to Jomon-Ainu; while the The traits of the mandible (Figure 2c) produced yet three Recent samples representing west, central and north another configuration which distinguishes Jomon-Ainu from regions of the main islands of Japan, tend to form a subclus- Wajin. Again note the smaller Recent Wajin cluster within ter within the larger Wajin cluster. the main one. Among the Jomon samples with their gener- Each of the bivariate plots gave its own unique version of ally higher frequencies of both accessory mental foramen population affinities. At the same time, some common and mylohyoid bridge, Hokkaido stands out with its very themes emerged: high frequency of the latter, 34%. This peculiarity has 1. All eight skull regions provide good discrimination already been noted by Yamano and Yamaguchi (1976). between the Jomon and Wajin lineages. Both dental traits included in this study are key Sinodont 2. In four of the diagrams Ainu is clearly intermediate; in features and thus show, as expected, higher frequencies in four, Ainu overlaps with Jomon. Wajin Japanese than in Ainu or Jomon (Figure 2d). Certain 3. Among Ainu the Southeast group tends to lie closest to patterns seen in previous plots are reiterated here; among the Jomon; the West is closest to Wajin. 106 N.S. OSSENBERG ET AL. ANTHROPOLOGICAL SCIENCE

Figure 2. Bivariate distribution of population samples according to θ values of nonmetric trait frequencies (see Analytic methods). (a) Occipi- tal bone; (b) sphenoid bone; (c) mandible; (d) teeth. Symbols for Jomon, Ainu, and Wajin samples are the same as in Figure 1.

4. Yayoi tends to be an outlier. closest to Wajin. Within the large cluster containing nine 5. Among Wajin the Tohoku samples tend to be closest to samples representing Wajin Japanese from all periods lies Jomon-Ainu; this tendency is strongest in North Tohoku the tight subcluster containing Recent West, Central, and Edo. Tohoku regional populations. From left to right along the x- 6. Among Wajin there is a tendency to increasing regional axis there is the hint of a regional gradient of affinities homogeneity over time, i.e. the three Recent samples among Wajin; the three Tohoku samples are generally the form a subcluster within the bigger one containing Wajin ones closest to the Jomon-Ainu lineage, followed by the from all periods. Central samples, except that Kamakura seems slightly dis- placed towards Jomon-Ainu. Furthest to the right are West Cluster analysis of 21-trait MMDs: Japan Edo (Tenpukuji) and Yayoi. Among Wajin, Yayoi seems Multidimensional scaling of MMDs for the 17 samples somewhat aberrant. from Japan (Table 2) produced the configuration shown in The advantage of the additive tree clustering technique is Figure 5. Each of the six themes hinted at in the bivariate that in joining populations in stepwise fashion it can reveal plots is consolidated here. Ainu lie in the center of the plot, the intermediate position of a population with respect to oth- clearly separated from Jomon on the left and Wajin on the ers. Therefore it is well suited to examine the data for gradi- right, with Southeast Ainu closest to Jomon and West Ainu ents and for evidence of hybridization. For this additive tree Vol. 114, 2006 ETHNOGENESIS IN JAPAN 107 chi Chuk- — 71.3 hotsk — 91.2 76.9 — 75.6 42.8 62.1 Chinese Tungus Ok- — 3.2* 56.9 46.8 65.2 gols Mon- — 12.7 29.8 93.3 72.2 102.4 nese ates Northeast Asians — 83.1 142.1 158.3 225.3 111.2 216.0 2* 53.4 194.4 290.2 302.1 391.8 259.6 316.5 — 48.1 178.4 261.5 267.7 383.4 199.2 302.9 — 188.6 88.3 — 14.9 37.3 110.5 73.5 113.3 16.5* 106.8 37.9 — 8.6* 27.8 100.8 55.5 88.4 − 9.52.2 11.4* 156.1 3.9* 63.1 173.3 — 85.2 — 38.2 49.8 19.4 100.0 39.3 85.8 87.6 129.1 89.4 105.1 — 13.6* 262.1 149.9 — 27.5 41.0 73.2 85.1 85.6 nese West Japan: Aggreg 101.7 50.8 90.0 — 56.8 114.1 136.5 196.1 113.2 200.5 — 60.1 31.0 236.9 102.9 — 40.3 67.2 82.3 102.5 158.5 33.3 197.308.4 126.0 157.3 34.0 96.2 — 43.2 114.5 — 168.6 87.7 190.5 149.5 260.3 157.9 129.1 225.8 224.6 100.1 225.9 ic cranial traits 8.0* 12.5* 30.7 — 19.0 50.1 14.4 181.4 87.4 — 17.3 38.5 124.7 86.6 112.0 − 8 247.4 314.5 350.2 259.2 — 82.6 247.6 330.3 342.6 432.7 254.1 331.4 4.1* — 5.3* 25.8 27.8 3.3* 201.3 102.2 — 2.2* 19.2 104.0 67.8 99.6 − Edo Recent Yayoi Edo Recent Jomon Ainu Japa- 181.3 163.3 200.8 247.2 176.8 — 34.7 163.3 244.7 247.7 361.3 161.7 283.0 .5 222.3 210.8 251.6 264.9 225.1 — 62.4 198.3 287.1 276.7 377.4 214.5 318.0 3.8* — 16.8 24.3 22.0 24.8 15.0 164.7 78.0 — 29.3 40.7 84.6 81.9 93.3 88.6 206.7 182.3 266.9 275.7 180.1 — 83.2 185.9 257.5 282.1 435.3 230.0 334.4 − kura ce based on 21 nonmetr Kama- 7.4* − 93.2 149.4 217.3 193.3 237.7 270.2 213.9 25. South Recent 4.4* − Edo South Edo Table 2. Mean measures of divergen Table SE NE West North twice its standard deviation. Epi- Jomon kaido Jomon Ainu Japanese Tohoku Japanese Central Japa . 3 gnificant value; i.e. less than West Central North Hok- Mongols Chinese North Tungus Okhotsk is multiplied by 10 Each MMD Ainu Aggregate Japanese Aggregate Jomon Aggregate West: Recent West: Tohoku South: EdoTohoku South: RecentTohoku Central: Kamakura Central: Edo Central: Recent Yayoi West: Edo West: — 0.8* — 13.9 16.9 14.2 16.6 14.0 19.6 24.5 1 39.7 2 Jomon WestJomon CentralJomon NorthJomon Hokkaido —Epi-Jomon 86.6Ainu Southeast 22.5*Ainu Northeast — 35.2*Ainu West 78.7 3.5* North: EdoTohoku 70.7 72.6 87.7 69.2 — 111.8 51.8 101.7 20.3* 47.7 15.4* 179.5 — 181.5 114.7 18.7* 26.2 1 142.9 2.4* 153.1 57.8 79.8 189.6 69.5 187 97.5 149.0 141.6 164.2 — 167.8 231.2 154.9 233.1 36.8 — 208.5 52.6 276. 92.6 9.2* 138.0 24.5 173.0 — 1 56.9 105.5 15.4* 116.7 45.4 97.2 58.0 130.3 85.1 110.6 — 1 — 85.3 22.4* 112.4 41.5 92.0 67.8 1 64.5 69.3 66.3 74.3 An asterisk indicates a non-si 108 N.S. OSSENBERG ET AL. ANTHROPOLOGICAL SCIENCE

Figure 3. Facial suture variants in an Okhotsk male skull from the Omisaki site. Usually, the facial portion of the infraorbital suture runs from the foramen up to meet the orbital margin at some point medial to the zygomaxillary suture. Infraorbital suture variant is the case, as here, in which the suture intersects the zygomaxillary before reaching the orbital margin. Figure 5. Configuration of 17 population samples from Japan according to multidimensional scaling analysis of their 21-trait MMDs. Symbols for Jomon, Ainu, and Wajin are the same as in Figure 1.

Figure 6. Additive tree diagram based on 21-trait MMDs for the Figure 4. Frontal bone variants. Trochlear spur represents anoma- Jomon and Recent Wajin aggregates with ten other population samples lous ossification at the attachment of the tiny fibrocartilaginous pulley from Japan. for the tendon of the superior oblique muscle of the eyeball.

samples, those from Japan maintain the same relationships analysis (Figure 6) four Jomon samples were aggregated, as to each other as when analyzed by themselves. Thus, from were the three Recent Wajin. The Jomon-Ainu lineage forms left to right along the x-axis in increasing distance from a cluster distinct from that containing the Wajin Japanese. Jomon are: Southeast, Northeast and West Ainu; premodern Within the former there is a clear four-step gradient towards Wajin series from North Tohoku, South Tohoku, Central Wajin consistent with the pattern noted in previous dia- region, and West. Moreover, the Recent Wajin aggregate grams: Jomon, Epi-Jomon, Southeast Ainu, Northeast Ainu, maintains the same more or less central position relative to West Ainu. Within the Wajin cluster there is a three-stage Edo and earlier samples as did the Recent North, Central, gradient of affinities away from Jomon-Ainu: North Tohoku and West subcluster in Figure 5. Edo, Kamakura; and five other populations in a subcluster Wajin Japanese as a whole are more closely related to with no readily interpretable pattern of relationships. continental northeast Asians than to Jomon, being affiliated particularly with Mongols and north Chinese, a finding Cluster analysis of 21-trait MMDs: Japan and continen- wholly in agreement with that of other investigations. The tal northeast Asia Chukchi sample, which in the present study is the only rep- Figure 7 is a multidimensional scaling analysis of rela- resentative of the ‘Arctic Mongoloid’ (Alexseev, 1979) is an tionships of Japan’s two lineages within a broader geograph- outlier in this plot. Although Chukchi and Okhotsk (also an ical context. Perhaps the most striking feature of this plot is outlier) are closer to each other than either is to any other that, with the addition of five continental northeast Asian sample, nevertheless the position of Okhotsk is slightly Vol. 114, 2006 ETHNOGENESIS IN JAPAN 109

Discussion Jomon Our data concerning the strong differentiation between Japan’s two founding groups and the relationships of these to other northeast Asians were in agreement with virtually all previous studies based on a variety of evidence (Dodo, 1975; Yamaguchi, 1982, 1992; Hanihara, 1985; Howells, 1986; Mizoguchi, 1986, 1988a; Turner, 1990; Ishida and Dodo, 1992; Omoto, 1992; Ishida, 1993, 1995; Matsumura, 1995; Jin et al., 1996; Liu, 1996; Pietrusewsky, 1996; Horai and Omoto, 1998; and others). Notably, patterns of relation- ships involving Jomon apparently did not reflect varying degrees of affiliation between Jomon and any particular people of continental northeast Asia. Jomon was closer to all Japanese than to any other group of northeast Asians (Table 2, Figure 7, Figure 8) or native North Americans (Ossenberg, 1994). Jomon’s closest affiliations, at least in the limited context of this study, appeared to be entirely Figure 7. Configuration of population samples from Japan and within the Japanese archipelago; and, as emphasized by continental northeast Asia according to multidimensional scaling of other researchers (Yamaguchi, 1982, 1992; Hanihara, 1992; their 21-trait MMDs. Symbols for Jomon, Ainu, and Wajin are the Ishida and Dodo, 1992; Pietrusewsky, 1996), included not same as in Figure 1. just the Ainu but, indeed, through successive episodes of population movement and admixture since Yayoi times, all present-day Japanese. Other skeletal and dental researchers have described local differences among the Jomonese. Although the patterns of inter-regional relationships vary somewhat from one study to another (Yamaguchi, 1980, 1982, 1992; Howells, 1966, 1986; Dodo, 1982; Mizoguchi, 1988b; Mouri, 1988; Matsumura, 1989, 1995; Maeda, 2002; and others) a common pattern seems to be a west to northeast gradient of affinities, possibly consistent with the isolation by distance model or with adaptation to different environ- ments. However, in spite of marked differences with respect to ecozone, subsistence strategies, and population density from western Honshu to Hokkaido (Akazawa, 1996) and regardless of the great temporal depth of the Jomon, the Figure 8. Additive tree diagram based on 21-trait MMDs for pop- range of physical variation is generally found to be narrower ulation samples from Japan and continental northeast Asia. than that among modern Japanese (Dodo, 1982; Yamaguchi, 1982; Hanihara, 1991). shifted towards the other northeast Asians, including Wajin In Figure 5 the five Jomon samples spread out along the y- Japanese and Ainu. Ishida (1996) finds the affinities of axis appear to display a degree of regional differentiation Okhotsk complicated, owing perhaps to admixture with comparable to that among Wajin. However, Table 2 contra- Ainu and other neighbors. dicts this: in contrast to 64% (of 36) within-Wajin MMDs, The additive tree shown in Figure 8 summarizes and high- only 40% (of 10) within-Jomon MMDs were larger than lights relationships of Japan’s two lineages to other northeast zero at a statistically significant level. The size of the Jomon Asians (excluding Chukchi). For this analysis regional sam- cluster reflected mainly the large MMDs involving the West ples of Wajin across all time periods were aggregated, i.e. and Central samples. In fact, the largest of 49 within-group Wajin West is an aggregate of Yayoi, Edo, and Recent sam- distances in Japan (36 Wajin, 10 Jomon, 3 Ainu) was the ples. Again, away from continental northeast Asians and MMD, 0.087, for Central versus West Jomon. In contrast, towards Jomon-Ainu, the biological distances of Wajin Jap- none of the four within-Jomon MMDs for the North sample anese samples form a geographical gradient: West, Central, were statistically significant and, with one exception, a geo- South Tohoku, and North Tohoku. The overall relationships graphical gradient of affinities was not observed. The excep- within and between the two ancestral lineages remain the tion was Epi-Jomon for which MMDs were ranked: same. Hokkaido, 0.002; North, 0.015; Central, 0.069; West, 0.079. Ainu Craniometric and nonmetric analyses by several workers (reviewed by Dodo and Kawakubo, 2002) suggest that the 110 N.S. OSSENBERG ET AL. ANTHROPOLOGICAL SCIENCE

Epi-Jomon inhabitants of Hokkaido, with little or no genetic three Ainu regional samples versus North Jomon were influence from their Yayoi or Kofun contemporaries (ca. 300 smaller than those versus either Epi-Jomon or Hokkaido BC to 700 AD) in central Japan, were transitional in micro- Jomon (Table 2). evolutionary change from Jomon to Ainu. Cranial MMDs Skeletal radioisotope data provide dietary evidence that from the present study agreed with those of Dodo and prehistoric hunter-fisher-foragers in Hokkaido throughout Kawakubo (2002) in providing evidence for this reconstruc- the last 6000 years shared with the Ainu a primary depen- tion, i.e. in both studies MMDs for comparisons within the dence on marine resources, in contrast to the more mixed Jomon-Ainu lineage are much smaller than those between woodland/marine resource base in northeastern Honshu the Jomon-Ainu and Wajin lineages (Table 3). (Akazawa, 1996). Therefore, the closer affinity of Hokkaido But the question remains: which Jomon population was Ainu to North Jomon than to Hokkaido Jomon could hardly ancestral to Epi-Jomon and hence to Ainu? Comparisons in be attributed to some hypothetical shift in Ainu subsistence the Dodo and Kawakubo study do not include separate strategies from the Hokkaido marine-based to the northeast- regional samples. One by Matsumura (1995) based on 21 ern Honshu mixed pattern. Even if such a shift were to have dental crown and root traits, though it does not include Epi- occurred, it probably would not have distorted the cranial Jomon, does include regional Jomon samples for which nonmetric trait evidence for Ainu ancestry; other studies ranked MMDs versus Ainu (Hokkaido, 0.019; Central, (cited in Ossenberg, 1994) have underscored the remarkable 0.035; North, 0.068; and West Jomon, 0.120) suggest that conservatism of these features in the face of major environ- the Jomon to Ainu microevolutionary change could have mental changes. A more plausible explanation for the occurred within Hokkaido with little genetic influence from observed pattern of affinities (assuming that the MMDs truly Honshu. On the other hand, according to Yamaguchi (1982), reflect genetic relationships and not error due to unrepresen- Epi-Jomon and Ainu do show a close affinity to Jomon tative and undersized samples) is that the microevolutionary remains from Honshu, including the Tohoku region. Find- transition in Hokkaido from Jomon to Ainu was impacted ings in the present study agreed with those of Yamaguchi. strongly by aboriginal gene flow from northeastern Honshu. From Table 2 the pertinent MMDs for Jomon and Southeast To account for the Ainu’s apparent divergence away from Ainu (representing the least Wajin-mixed sample) were Hokkaido Jomon and convergence towards North Jomon we ranked as follows: offer the following reconstruction. It is based on the crucial premise that in the most northeastern region of Honshu some Epi-Jomon to Hokkaido Jomon, 0.002; communities preserved a predominantly Jomon genetic her- Epi-Jomon to North Jomon, 0.015; itage well into the medieval age. These would have been the Ainu SE to North Jomon, 0.019; so-called Emishi, first described in 7th-century Kofun Hokkaido Jomon to North Jomon, 0.020; period historical accounts. It is recorded that people belong- Epi-Jomon to Ainu, 0.037; ing to the same tribe as the one in Hokkaido, Watarishima Ainu SE to Central Jomon, 0.052; Emishi, inhabited the northernmost peninsulas of Honshu Ainu SE to Hokkaido Jomon, 0.058; from the 7th to 12th centuries (Kodama, 1970; Hanihara, Ainu SE to West Jomon, 0.071. 1990). Migration/interaction may have occurred in both Although in the above summary the Ainu values were directions across the Tsugaru Strait during Jomon times, i.e. those of the Southeast sample, in fact MMDs for each of the we noted that the Hokkaido-North Jomon MMD of 0.020 was not significantly greater than zero. However, based on the pattern of MMDs, we hypothesized that there was a sig- nificant interval from Epi-Jomon through Satsumon periods, Table 3. MMDs from Dodo and Kawakubo (2002) and from the present study, contrasting affinities within the Jomon-Ainu roughly 100 BC to 1100 AD, during which aboriginal gene lineage and between the Jomon-Ainu and Wajin lineages flow was predominantly from Honshu to Hokkaido. This interval would have coincided with the ancient through early Dodo and Kawakubo Present studyb (2002)a medieval ages in central Japan when warfare, colonizing ini- tiatives, and social dislocations would have caused popula- Within Jomon-Ainu tion pressure generally tending in a northeasterly direction. Jomon—Epi-Jomon 0.019* 0.015* Kazuro Hanihara (1990), citing the works of historians, Epi-Jomon—Ainu 0.016* 0.053 Jomon—Ainu 0.045 0.035 notes that the influence of the imperial court reached Miyagi prefecture in south Tohoku by the 7th century and expanded Jomon-Ainu to Wajin north to Iwate prefecture by the 10th century. Soldiers sent Ainu—Wajin 0.063–0.084 0.078–0.103 Epi-Jomon—Wajin 0.072–0.079 0.149–0.238 to these posts were accompanied by their families. At the Jomon—Wajin 0.084–0.104 0.155–0.201 same time local Wajin chiefs began to collect a great amount of gold and to build large Buddhist temples. Hence, Tohoku Values marked with an asterisk are statistically insignificant at the 5% level. became a mixed-residence quarter of Emishi and settlers a Jomon sample is from sites mainly in eastern Honshu, and the who had moved from west Japan. By the 12th century the Ainu is from Hokkaido. Wajin samples for which data are cited in powerful northern Fujiwara family had established a polity this table: Doigahama Yayoi, Kamakura, early modern Edo from over the Tohoku region which was to some extent indepen- Tokyo, modern mainland Japanese from eastern Honshu. b dent of the imperial court, and may have developed signifi- Samples from the present study for which data are cited in this table: Jomon North, Ainu aggregate, Yayoi, Kamakura, Central cant trade contacts instrumental in the change from Edo, South Tohoku Recent. Satsumon to Formative Ainu culture in Hokkaido (Hudson, Vol. 114, 2006 ETHNOGENESIS IN JAPAN 111

1999). Under these circumstances some aboriginal families, have become even more vulnerable to dilution through perhaps entire communities, may have become increasingly Wajin admixture. There is also a possibility that, in addition inclined from the 7th to 12th centuries to move across the to gene flow from the Japanese, the western Ainu were influ- Tsugaru Strait and join their relatives in Hokkaido. Thus, enced directly via invasion by a continental tribe, the Ashi- through recurrent migratory episodes from Epi-Jomon to hase, originally from north China, who had inhabited the medieval times and beyond, people of predominantly Jomon region near the mouth of the Ishikari River prior to their rout ancestry from northeastern Honshu would have reinforced in 660 AD by an expedition sent by the Japanese emperor the aboriginal gene pool in Hokkaido. We suggest, therefore, (Kodama, 1970, pp. 6–11). that the ancestry of Hokkaido Ainu included not only its own In contrast, communities in the southcentral/southeastern local Jomon people, but those of Tohoku as well. provinces of Hidaka and Tokachi retained the strongest The 19th-century series of skeletons excavated by Jomon heritage. Hidaka was the most densely populated Koganei (1893, 1894) from cemeteries in various Hokkaido province and, because the villages were dispersed along the sites usually has been employed as a single sample to repre- upper reaches of the salmon-fishing rivers and were not con- sent relatively unmixed Ainu, though Dodo (1974) splits the veniently accessible to Japanese, up to 1945 the south cen- series into ‘Ainu a’ and ‘Ainu b’ for purposes of a paper on tral and southeastern Ainu retained their traditional customs nonmetric trait theory and methodology. Some workers and habits and were the least racially mixed (Kodama, 1970, name the Koganei series ‘Ainu 1’ and use it to represent cen- pp. 40–46). tral and northeastern Hokkaido in distance analyses under- taken for various purposes, while ‘Ainu 2’ from Yakumo Wajin village in Oshima province (Watanabe, 1938) represents the Compared to other populations sampled from the central southwestern peninsula region. Tables and diagrams in these islands of Japan, the Edo period population of North Tohoku reports generally show ‘Ainu 1’ closer to Jomon than is retained the strongest genetic heritage from the Jomon-Ainu ‘Ainu 2’ (Dodo, 1982; Hanihara, 1992; Turner, 1992). In lineage as the following MMDs from Table 2 show. Hanihara et al. (1993), where the Yakumo series is ‘Ainu 1’ Ainu aggregate to: and a sample from Hidaka in southeastern Hokkaido (Oba, North Tohoku Edo, 0.038; 1973) is ‘Ainu 2’, the figures presented show that the latter eight other Wajin samples, 0.063–0.150. was the one more closely related to Jomon. Based on a more Jomon aggregate to: fine-grained analysis of regional variation in Ainu cranio- North Tohoku Edo, 0.107; metric characteristics (not of the Koganei series but of eight other Wajin samples, 0.156–0.262. remains from 20th-century archaeological fieldwork on Hokkaido, Kunashiri, and Sakhalin Islands), a northeastern This finding is consistent with our reconstruction in the Hokkaido sample is reported to be relatively distinct from preceding section where we hypothesized that migration the other Hokkaido sites (Hanihara, 1998). from Jomon and its descendent Emishi populations on the Our findings agreed with those reports cited above, which northern tip of Honshu, possibly well into medieval times, suggest that the Hidaka Ainu were least racially mixed with played a key role in the ethnogenesis of Hokkaido Ainu. To Japanese. A strikingly consistent pattern of affinity for the some extent microevolutionary transition in northern three Ainu samples to other populations was observed Tohoku may have paralleled that in Hokkaido. In the 15th (Table 2, Figure 5, Figure 6, Figure 7, Figure 8) in which the century, Ainu-speakers lived in what are now Aomori, Southeast sample was the one closest to Jomon, West was Iwate, and Akita prefectures (Inoue, 1999). Moreover, the one closest to the Japanese aggregate—closer, in fact, to historical records indicate that after the 16th century some Japanese than to Epi-Jomon or Jomon—while Northeast was Hokkaido Ainu migrated to north Tohoku (Hanihara, 1990). intermediate. In summary: According to archaeological and skeletal findings it appears that Ainu lived here even up to recent times (Kodama, 1970; Jomon aggregate to: Hanihara, 1990). To the present day the communities here Southeast, 0.034; Northeast, 0.043; West, 0.090; retain cultural and linguistic (Inoue, 1999; Wang and Ogura, Epi-Jomon to: 1996), somatometric (Kouchi, 1983), and molecular Southeast, 0.037; Northeast, 0.053; West, 0.093; genetic traits (Misawa and Hayashida, 1968; Omoto, 1992; Japanese aggregate to: Takeshita et al., 2001), as well as varieties of the house Southeast, 0.115; Northeast, 0.088; West, 0.057. mouse and domestic dog (Tanabe, 1992) which attest to the This pattern was interpreted to represent local differences unique history of Tohoku as the most recent part of central with respect to the timing and strength of gene flow such that Japan to come under Wajin influence. communities on the west side were most hybridized with West to northeast clines quantified in many independent Wajin Japanese. This is consistent with historical records studies on stature, craniofacial characteristics, and genetic for Hokkaido (Kodama, 1970). Since earliest times the factors in the blood in regional Japanese populations of the Tsushima Current in the Sea of Japan has aided travel by central islands constitute one of the cornerstones of support ship from Honshu northward along the west coast. More- for the dual structure model (Hanihara, 1991). MMDs in the over, between 1822 and 1855 the western population present study also tended to follow a west to northeast gradi- declined drastically due to infectious diseases brought by the ent (Table 2, Figure 5, Figure 6, Figure 7, Figure 8). This Japanese (Kodama, 1970, pp. 38–39); hence, the aboriginal pattern is most striking in the ranking of MMDs for the gene pool of the remnant west Ainu communities would Jomon aggregate sample versus Edo period Japanese and 112 N.S. OSSENBERG ET AL. ANTHROPOLOGICAL SCIENCE north Chinese, the latter sample representing a group possi- represented in the dissecting-room series, MMDs suggest bly closely related to the continental ancestor: North Tohoku that there may have been sufficient migration from central Edo, 0.107; South Tohoku Edo, 0.156; Central Edo, 0.201; and west Japan over that interval to dilute the aboriginal rep- West Edo 0.262; North Chinese, 0.268. resentation in the Tohoku gene pool. Clinal patterns were less apparent in rankings of Wajin Similarly, regional analysis of somatometric data shows versus Wajin MMDs, perhaps because of the confounding that the characteristics of western Honshu diffuse gradually effects of temporal differences and small samples, e.g. to the northeast, and diachronic analysis of these data shows Yayoi. And, as already noted, 36% of these values were sta- that the secular change in local populations is quite evident tistically insignificant. In any case, we should probably not even between birth cohorts separated, on average, by as little look too narrowly for geographical gradients. Archaeologi- as 30 years (Kouchi, 1983). This would reflect increased cal and historical records provide invaluable information mobility of the post-Edo population and is wholly consistent about early travel routes, settlement patterns, and specific with the dual structure model. episodes and events, according to which affinities in certain cases might better fit a mosaic than a cline model. Summary and Conclusions In the case of Kamakura, a relatively closer affinity of Japan’s aboriginal lineage to this medieval sample than 1. The findings of this study corroborated those of other to other post-Jomon samples from the central islands was investigations in support of the dual structure model of noted in the present study. Similar findings in previous stud- ethnogenesis in Japan. ies based on other categories of data (Howells, 1966; 2. The unique battery of nonmetric traits used in this study Yamaguchi, 1982; Brace et al., 1989; Pietrusewsky, 1996) powerfully differentiated between Japan’s indigenous have provoked controversy. Brace and colleagues, appar- Jomon-Ainu and immigrant northeast Asian ancestral lin- ently assuming that warriors killed at a medieval massacre eages. site must have been samurai, concluded from craniometric 3. The continental roots of the Wajin Japanese were closer evidence of Kamakura’s affinity to Ainu that samurai must to those of Mongols and north Chinese than to those of therefore have been Ainu. However, others (Hanihara, 1991; the Tungus, Okhotsk, or Chukchi. The Jomon had no rel- Pietrusewsky, 1996) have pointed out that the warriors mas- atives in northeast Asia other than its descendants in sacred at this site were more likely to have been retainers or Japan. servants of the Kamakura Shogunate. These lower-class 4. Gene flow between immigrant and aboriginal descendent warriors may very well have been drawn from communities communities initiated during the Yayoi period resulted in whose ancestry was preponderantly aboriginal. Samurai, on west to northeast gradients of morphological characteris- the other hand, belonged to a noble warrior caste more likely tics and affinities which seemed to have become stabi- descended from continental immigrants. Thus, persistence lized during the Edo period. in Japan of caste, class, or occupational identity differenti- 5. Following the Meiji Restoration of 1867, increased ated along racial lines would, according to the dual structure mobility of the population resulted in renewed gene flow model, be expected to present a mosaic pattern of affinities whereby the population of Japan became more homoge- among cemeteries. neous. Perhaps the most important finding in the present study 6. Affinities of Edo and recent period populations of was the evidence of increased inter-regional population Tohoku and Hokkaido are consistent with historical homogeneity throughout the central islands since the end of records and with other cultural and biological evidence the Edo period. Recent Tohoku, Central, and West samples indicating, that these regions were the latest to come form a tight subcluster within the larger configuration con- under the influence of Wajin Japanese. Among regional taining Wajin samples from all periods (Figure 5). In Miyagi populations of the Ainu, those from western Hokkaido and the other southern Tohoku prefectures people born at the were most, while those from the southeast (Hidaka and very end of the Edo period retained a slightly stronger trace Tokachi districts) were least mixed. of Jomon ancestry than those born after 1870, as suggested 7. In Hokkaido the microevolutionary transition from by analysis of two subsamples of the Tohoku University dis- Jomon through Epi-Jomon to Ainu may have occurred in secting-room cranial series. The MMDs were: close parallel to population changes in the most north- eastern region of Honshu, and may have been influenced Ainu aggregate to: during the Kofun and medieval periods by immigration South Tohoku Edo, 0.063; of Jomon descendants, the Emishi, across the Tsugaru South Tohoku Recent, 0.085. Strait. More archaeological and skeletal remains repre- Jomon aggregate to: senting the poorly documented interval from Final Jomon South Tohoku Edo, 0.156; to Edo are needed to piece together the ethnohistorical South Tohoku Recent, 0.173. picture for northeastern Honshu. The Edo period, 1603–1867, is known to have been a time 8. Craniofacial changes in Japan effected through micro- of stability and social control when people tended to spend evolutionary adaptation or through plastic response to their lives in or close to their birth communities. With the factors such as diet would have been superimposed on a Meiji Restoration in 1867, however, the mobility of the Jap- genetic substrate which was continually being reshaped anese people increased. Even though on average only two by gene flow between descendants of the two founding generations separated the pre- from post-Meiji birth cohorts lineages. Vol. 114, 2006 ETHNOGENESIS IN JAPAN 113

Acknowledgments (eds.), Prehistoric Hunter-Gatherers in Japan. The University Museum, The University of Tokyo, Bulletin, No. 27, pp. 137– The senior author would like to express her sincere grati- 161. tude for the opportunity to work in Japan for seven months Dodo Y. (1986b) A population study of the jugular foramen bridg- from September 2001 to March 2002 as visiting research ing of the human cranium. American Journal of Physical professor in the Tohoku University Museum through the Anthropology, 69: 15–19. invitation of Dr Hiroyuki Abe (President of Tohoku Univer- Dodo Y. (1987) Supraorbital foramen and hypoglossal canal bridg- ing: the two most suggestive nonmetric cranial traits in dis- sity), Dr Kei Mori (Director of the Tohoku University criminating major racial groupings of man. The Journal of the Museum), and Professor Yukio Dodo (Department of Anat- Anthropological Society of Nippon, 95: 19–35. omy and Anthropology, Tohoku University School of Medi- Dodo Y. and Ishida H. (1987) Incidences of nonmetric cranial vari- cine). The following persons gave permission to examine the ants in several population samples from East Asia and North skeletal collections under their care: Professor Gen America. The Journal of the Anthropological Society of Nip- Murakami (Sapporo Medical University), Professor Gen pon, 95: 161–177. Dodo Y. and Ishida H. (1988) Nonmetric analyses of the Doiga- Suwa (The University Museum, The University of Tokyo), hama crania of the Aeneolithic Yayoi period in western Japan. Dr Hisao Baba (National Science Museum, Tokyo), Profes- In: Nippon Minzoku-Bunka no Seisei (Establishment of Japa- sor Tanemichi Chiba (Chiba University School of Medicine, nese People and Culture) I. Rokko Shuppan, Tokyo, pp. 127– Chiba City), and Professor Takahiro Nakahashi (Graduate 142 (in Japanese). School of Social and Cultural Studies, Kyushu University, Dodo Y. and Ishida H. (1990) Population history of Japan as Fukuoka). The courtesy and assistance of these individuals, viewed from cranial nonmetric variation. The Journal of the their staff, graduate students, and colleagues are gratefully Anthropological Society of Nippon, 98: 269–287. Dodo Y. and Ishida H. (1992) Consistency of nonmetric cranial acknowledged. Support for her 1981 survey of collections in trait expression during the last 2,000 years in the habitants of Japan and the former Soviet Union was acknowledged ear- the central islands of Japan. The Journal of the Anthropologi- lier (Ossenberg, 1986). The opportunity to meet with Profes- cal Society of Nippon, 100: 417–423. sors Tasuku Kimura, Osamu Kondo, and graduate students Dodo Y. and Kawakubo Y. (2002) Cranial affinities of the Epi- (The University of Tokyo), Drs Bin Yamaguchi and Yuji Jomon inhabitants in Hokkaido, Japan. Anthropological Sci- Mizoguchi (National Science Museum) and Dr Koga (Anat- ence, 110: 1–32. Dodo Y., Ishida H., and Saitou N. (1992) Population history of omy Museum of Kyushu University) greatly enriched this Japan: a cranial nonmetric approach. In: Akazawa T., Aoki most recent research experience. Above all, the senior K., and Kimura T. (eds.), The Evolution and Dispersal of author is most deeply grateful to Professor Yukio Dodo, his Modern Humans in Asia. Hokusen-sha, Tokyo, pp. 479–492. colleagues, graduate students, and staff in the Department of Dodo Y., Doi N., and Kondo O. (1998) Ainu and Ryukyuan cranial Anatomy and Anthropology of Tohoku University for their nonmetric variation: evidence which disputes the Ainu- generous hospitality and assistance in facilitating her work Ryukyu common origin theory. Anthropological Science, in Japan. 106: 99–120. Hanihara K. (1985) Geographic variation of modern Japanese cra- nia and its relationship to the origin of Japanese. Homo, 36: 1–10. References Hanihara K. (1990) Emishi, Ezo and Ainu: an anthropological per- Akazawa T. (1996) Regional variation in Jomon hunting-fishing- spective. Japan Review, 1: 35–48. gathering societies. In: Omoto K. (ed.), Interdisciplinary Per- Hanihara K. (1991) Dual structure model for the population his- spectives on the Origins of the Japanese, International Sym- tory of the Japanese. Japan Review, 2: 1–33. posium 11-B. International Research Center for Japanese Hanihara K. (1998) Reanalysis of local variations in the Ainu cra- Studies, Kyoto, pp. 223–231. nia. Anthropological Science, 106 (supplement): 1–15. Alexseev V.P. (1979) Anthropometry of Siberian peoples. In: Hanihara K., Hanihara T., and Koizumi K. (1993) Biological rela- Laughlin W.S. and Harper A.B. (eds.), The First Americans: tionship between the Jomon-Ainu and Pacific population Origins, Affinities, and Adaptations. Gustav Fischer, New groups. Japan Review, 4: 7–25. York, pp. 57–90. Hanihara T. (1992) Dental and cranial evidence on the affinities of Brace C.L., Brace M.L., and Leonard W.R. (1989) Reflections on the East Asian and Pacific populations. In: Hanihara K. (ed.), the face of Japan: a multivariate craniofacial and odontomet- Japanese as a Member of the East Asian and Pacific Popula- ric perspective. American Journal of Physical Anthropology, tions, International Symposium 4, 1992. International 78: 93–113. Research Center for Japanese Studies, Kyoto, pp. 119–137. Cavalli-Sforza L.L. (1986) Population structure. In: Gershowitz Hanihara T. and Ishida H. (2001a) Frequency variations of discrete H., Rucknagel D.L., and Tashian R.E. (eds.), Evolutionary cranial traits in major human populations: I. supernumerary Perspectives and the New Genetics. Alan R. Liss, New York, ossicle variations. Journal of Anatomy, 198: 689–706 pp. 13–30. Hanihara T. and Ishida H. (2001b) Frequency variations of discrete Dodo Y. (1974) Non-metrical cranial traits in the Hokkaido Ainu cranial traits in major human populations: II. hypostotic vari- and the northern Japanese of recent times. The Journal of the ations. Journal of Anatomy, 198: 707–725. Anthropological Society of Nippon, 82: 31–51. Hanihara T. and Ishida H. (2001c) Frequency variations of discrete Dodo Y. (1975) Non-metric traits in the Japanese crania of the Edo cranial traits in major human populations: III. hyperostotic period. Bulletin of the National Science Museum, Series D, 1: variations. Journal of Anatomy, 199: 251–272. 41–54. Hanihara T. and Ishida H. (2001d) Frequency variations of discrete Dodo Y. (1982) A metrical analysis of Jomon crania from the cranial traits in major human populations: IV. vessel and Tohoku district. The Journal of the Anthropological Society nerve related variations. Journal of Anatomy, 199: 273–287. of Nippon, 90 (supplement): 119–128 (in Japanese). Hanihara T., Ishida H., and Dodo Y. (1998a) Place of the Hokkaido Dodo Y. (1986a) Metrical and non-metrical analyses of Jomon cra- Ainu (northern Japan) among circumpolar and other peoples nia from eastern Japan. In: Akazawa T. and Aikens C.M. of the world—a comparison of the frequency variations of 114 N.S. OSSENBERG ET AL. ANTHROPOLOGICAL SCIENCE

discrete cranial traits. International Journal of Circumpolar sität, Band 2, Kaiserlichen Universität, Tokyo, pp. 1–249. Health, 57: 257–275. Koganei Y. (1894) Beiträge zur physischen Anthropologie der Hanihara T., Ishida H., and Dodo Y. (1998b) Os zygomaticum Aino: II. Untersuchungen am Lebenden. Mittheilungen aus bipartitum: frequency distribution in major human popula- der Medizinischen Facultät der Kaiserlich-Japanischen Uni- tions. Journal of Anatomy, 192: 539–555. versität, Band 2, Kaiserlichen Universität, Tokyo, pp. 251– Horai S. and Omoto K. (1998) Peopling of Japan as inferred from 403. mitochondrial DNA polymorphisms in East Asians. In: Kouchi M. (1983) Geographic variation in modern Japanese Omoto K. and Tobias P.V. (eds.), The Origins and Past of somatometric data and its interpretation. The University Modern Humans: Towards Reconciliation. World Scientific, Museum, The University of Tokyo, Bulletin, No. 22, pp. 1– Singapore, pp. 54–87. 102. Howells W.W. (1966) The Jomon population of Japan: a study by Kozintsev A. (1990) Ainu, Japanese, their ancestors and neigh- discriminant analysis of Japanese and Ainu crania. Papers of bours: cranioscopic data. The Journal of the Anthropological the Peabody Museum of Archaeology and Ethnology, Volume Society of Nippon, 98: 247–267. 57, Harvard University, Cambridge, pp. 1–43. Kozintsev A. (1992a) Prehistoric and recent populations of Japan: Howells W.W. (1986) Physical anthropology of the prehistoric multivariate analysis of cranioscopic data. Arctic Anthropol- Japanese. In: Person R.J. (ed.), Windows on Japanese Past: ogy, 29: 104–111. Studies in Archaeology and Prehistory. Center for Japanese Kozintsev A. (1992b) Ethnic epigenetics: a new approach. Homo, Studies, The University of Michigan, Ann Arbor, pp. 85–99. 43: 213–244. Hudson M. (1999) Ainu ethnogenesis and the Northern Fujiwara. Liu W. (1996) The dental morphology and continuity of prehistoric Arctic Anthropology, 36: 73–83. and historic humans of China. In: Omoto K. (ed.), Interdisci- Inoue F. (1999) Isolated dialectical forms and language substra- plinary Perspectives on the Origins of the Japanese, Interna- tum. In: Omoto K. (ed.), Interdisciplinary Perspectives on the tional Symposium 11-B. International Research Center for Origins of the Japanese, International Symposium 11-B. Japanese Studies, Kyoto, pp. 43–64. International Research Center for Japanese Studies, Kyoto, Maeda T. (2002) Mandibular ramus morphology of the Jomon peo- pp. 293–308. ple from Hokkaido. Anthropological Science, 110: 27–40 (in Ishida H. (1990) Cranial morphology of several ethnic groups Japanese). from the Amur basin and Sakhalin. The Journal of the Matsumura H. (1989) Geographic variation of dental measure- Anthropological Society of Nippon, 98: 137–148. ments in the Jomon population. The Journal of the Anthropo- Ishida H. (1993) Cranial nonmetric variation of Circum-Pacific logical Society of Nippon, 97: 493–512. populations with special reference to the Pacific peoples. Matsumura H. (1995) A microevolutional history of the Japanese Japan Review, 4: 27–43. people as viewed from dental morphology. National Science Ishida H. (1995) Nonmetric cranial variation of Northeast Asians Museum Monographs, No. 9, National Science Museum, and their population affinities. Anthropological Science, 103: Tokyo. 385–401. Misawa S. and Hayashida Y. (1968) On the blood groups among Ishida H. (1996) Metric and nonmetric cranial variation of the pre- the Ainu in Shizunai, Hokkaido. Proceedings of the Japanese historic Okhotsk people. Anthropological Science, 104: 233– Academy, 44: 83–88. 258. Mizoguchi Y. (1986) Contributions of prehistoric Far East popula- Ishida H. and Dodo Y. (1990) Interobserver error in scoring non- tions to the population of modern Japan: a Q-mode path anal- metric cranial traits. The Journal of the Anthropological Soci- ysis based on cranial measurements. In: Akazawa T. and ety of Nippon, 98: 403–409. Aikens C.M. (eds.), Prehistoric Hunter-Gatherers in Japan. Ishida H. and Dodo Y. (1992) Differentiation of the northern Mon- The University Museum, The University of Tokyo, Bulletin, goloid: the evidence of cranial nonmetric traits. In: Hanihara No. 27, pp. 107–136. K. (ed.), Japanese as a Member of the Asian and Pacific Pop- Mizoguchi Y. (1988a) Affinities of the protohistoric Kofun people ulations, International Symposium 4, 1992. International of Japan with prehistoric Asian populations. The Journal of Research Center for Japanese Studies, Kyoto, pp. 79–94. the Anthropological Society of Nippon, 96: 71–109. Ishida H. and Dodo Y. (1993) Nonmetric cranial variation and the Mizoguchi Y. (1988b) Tooth crown diameters of the permanent population affinities of the Pacific peoples. American Journal teeth of the Epi-Jomon people from the Usu-10 and other sites of Physical Anthropology, 90: 49–57. in the seaside region of Funkawan Bay, Hokkaido, Japan. Ishida H. and Kida M. (1991) An anthropological investigation of Memoirs of the National Science Museum, 21: 211–220 (in the Sakhalin Ainu with special reference to nonmetric cranial Japanese). traits. The Journal of the Anthropological Society of Nippon, Mizoguchi Y. (1997) Human bones from Ikenohata-shichikencho 99: 23–32. relics. In: Ikenohata-Shichikencho Iseki (Keianji-Ato). Taito- Ishida H. and Kondo O. (1999) Nonmetric cranial variation of the ku Ikenohata-Shichikencho Iseki Chosakai, Tokyo, pp. 289– Ainu and neighbouring human populations. Perspectives in 303 (in Japanese). Human Biology, 4: 127–138. Mouri T. (1976) A study of non-metrical cranial variants of the Jin F., Saitou N., and Omoto K. (1996) Population genetic studies modern Japanese in the Kinki district. The Journal of the on the Chinese. In: Omoto K. (ed.), Interdisciplinary Perspec- Anthropological Society of Nippon, 84: 191–203. tives on the Origins of the Japanese, International Symposium Mouri T. (1988) Incidences of cranial nonmetric characters in five 11-B. International Research Center for Japanese Studies, Jomon populations from West Japan. The Journal of the Kyoto, pp. 15–24. Anthropological Society of Nippon, 96: 319–337. Katayama K. (1988) A comparison of the incidence of non-metric Mouri T. (1996) Nonmetric cranial variants in a medieval Japanese cranial variants in several Polynesian populations. The Jour- sample from Ichikishima-jinja site. Anthropological Science, nal of the Anthropological Society of Nippon, 96: 357–369. 104: 89–98. Kodama S. (1970) Ainu: historical and anthropological studies. Oba T. (1973) Craniology of the Hidaka-Ainu. Bulletin of the Hokkaido University Medical Library Series, Volume 3, Institute of North Eurasian Culture, Volume 7, Hokkaido Uni- Hokkaido University School of Medicine, Sapporo. versity, pp. 65–220 (in Japanese). Koganei Y. (1893) Beiträge zur physischen Anthropologie der Okada H. and Yamaguchi B. (1975) Human skeletal remains exca- Aino: I. Untersuchungen am Skelet. Mittheilungen aus der vated at the Hot Springs village site, Port Moller, Alaska Pen- Medizinischen Facultät der Kaiserlich-Japanischen Univer- insula, in 1972. Bulletin of the National Science Museum, Vol. 114, 2006 ETHNOGENESIS IN JAPAN 115

Series D, 1: 25–40. emphasis on migration routes of the dogs. In: Hanihara K. Omoto K. (1992) Some aspects of the genetic composition of the (ed.), Japanese as a Member of the Asian and Pacific Popula- Japanese. In: Hanihara K. (ed.), Japanese as a Member of the tions, International Symposium 4, 1992. International Asian and Pacific Populations, International Symposium 4, Research Center for Japanese Studies, Kyoto, pp. 160–173. 1992. International Research Center for Japanese Studies, Turner C.G. II (1990) Major features of Sundadonty and Sin- Kyoto, pp. 137–145. odonty, including suggestions about East Asian microevolu- Ossenberg N.S. (1970) The influence of artificial cranial deforma- tion, population history, and later Pleistocene relationships of tion on discontinuous morphological traits. American Journal Australian aboriginals. American Journal of Physical Anthro- of Physical Anthropology, 33: 357–372. pology, 82: 295–317. Ossenberg N.S. (1976) Within and between race distances in popu- Turner C.G. II (1992) Sundadonty and Sinodonty in Japan: the den- lation studies based on discrete traits of the human skull. tal basis for a dual origin hypothesis for the peopling of the American Journal of Physical Anthropology, 45: 701–716. Japanese Islands. In: Hanihara K. (ed.), Japanese as Member Ossenberg N.S. (1986) Isolate conservatism and hybridization in of the Asian and Pacific Populations, International Sympo- the population history of Japan: the evidence of cranial non- sium 4, 1992. International Research Center for Japanese metric traits. In: Akazawa T. and Aikens C.M. (eds.), Prehis- Studies, Kyoto, pp. 96–112. toric Hunter-Gatherers in Japan. The University Museum, Wang W.S-Y. and Ogura M. (1996) Explorations in the origins of The University of Tokyo, Bulletin, No. 27, pp. 199–215. the Japanese language. In: Omoto K. (ed.), Interdisciplinary Ossenberg N.S. (1992a) Microevolutionary parallels in the popula- Perspectives on the Origins of the Japanese, International tion history of Japan and aboriginal North America: the Symposium 11-B. International Research Center for Japanese evidence of cranial nonmetric traits. In: Hanihara K. (ed.), Studies, Kyoto, pp. 309–334. Japanese as a Member of the Asian and Pacific Populations, Watanabe S. (1938) Kraniologie der Yakumo-Aino. In: Kodama S. International Symposium 4, 1992. International Research (ed.), Crania Ainoica. Sapporo, pp. 1–174. Center for Japanese Studies, Kyoto, pp. 64–77. Weiss K.M. (1988) In search of times past: gene flow and invasion Ossenberg N.S. (1992b) Native people of the American North- in the generation of human diversity. In: Mascie-Taylor west: population history from the perspective of skull mor- C.G.N. and Lasker G.W. (eds.), Biological Aspects of Human phology. In: Akazawa T., Aoki K., and Kimura T. (eds.), The Migration. Cambridge University Press, Cambridge, pp. 130– Evolution and Dispersal of Modern Humans in Asia. 166. Hokusen-sha, Tokyo, pp. 493–530. Yamaguchi B. (1967) A comparative osteological study of the Ossenberg N.S. (1994) Origins and affinities of the native peoples Ainu and the Australian Aborigines. Australian Aboriginal of northwestern North America: the evidence of cranial Studies, No. 10, Australian Institute of Aboriginal Studies, nonmetric traits. In: Bonnichsen R. and Steele D.G. (eds.), Canberra. Methods and Theory for Investigating the Peopling of the Yamaguchi B. (1977) A comparative study of the skulls of Ontario Americas. Center for the Study of the First Americans, Cor- Iroquoians and of Asiatic Populations. Bulletin of the vallis, Oregon, pp. 79–115. National Science Museum, Series D, 3: 23–35. Ossenberg N.S. (2004) Ethnogenesis in the Central and Eastern Yamaguchi B. (1980) A study on the facial flatness of the Jomon Arctic: a reconstruction based on cranial nonmetric traits. In: crania. Bulletin of the National Science Museum, Series D, 6: Sutherland P. (ed.), Perspectives on the Paleoeskimo Fifty 21–28. Years After Jenness. Canadian Museum of Civilization, Hull, Yamaguchi B. (1982) A review of the osteological characteristics Quebec (in press). of the Jomon population n prehistoric Japan. The Journal of Pietrusewsky M. (1984) Metric and Non-Metric Cranial Variation the Anthropological Society of Nippon, 90 (supplement): 77– in Australian Aboriginal Populations Compared with Popula- 90. tions of the Pacific and Asia. Occasional Papers in Human Yamaguchi B. (1985) The incidence of minor non-metric cranial Biology, No. 3, Australian Institute of Aboriginal Studies, variants in the protohistoric human remains from eastern Canberra. Japan. Bulletin of the National Science Museum, Series D, Pietrusewsky M. (1996) Multivariate craniometric study of Japa- 11: 13–24. nese, Asians, and Pacific Islanders. In: Omoto K. (ed.), Inter- Yamaguchi B. (1992) Skeletal morphology of the Jomon people. disciplinary Perspectives on the Origins of the Japanese, In: Hanihara K. (ed.), Japanese as a Member of the Asian and International Symposium 11-B. International Research Center Pacific Populations, International Symposium 4, 1992. Inter- for Japanese Studies, Kyoto, pp. 65–104. national Research Center for Japanese Studies, Kyoto, pp. Saunders S.R. (1989) Nonmetric skeletal variation. In: İşcan, M.Y. 53–63. and Kennedy K.A.R. (eds.), Reconstruction of Life from the Yamaguchi B, Sato I., and Dodo Y. (1973) A brief note on the Skeleton. Alan R. Liss, New York, pp. 95–108. supra-orbital nerve groove on the frontal surface of the human Sjøvold T. (1977) Non-Metrical Divergence Between Skeletal cranium. Bulletin of the National Science Museum, Series D, Populations: the Theoretical Foundation and Biological 16: 571–579. Importance of C.A.B. Smith’s Mean Measure of Divergence. Yamano S. and Yamaguchi B. (1976) On the mylohyoid canal in OSSA International Journal of Skeletal Research, Volume 4 the human mandible. Bulletin of the National Science (supplement No. 1). University of Stockholm Osteological Museum, Series D, 2: 37–44. Research Laboratory, Solna. Takeshita H., Yasuda T., Nakashima Y., Mogi K., Kishi K., Shiona H., Sagisaka K., Yuasa I., Nishimukai H., and Kimura H. This paper was a part of the symposium, “Causes for Chronological (2001) Geographical north-south decline in DNASE1*2 in Changes in Craniofacial Morphology” (Anthropological Science, Japanese populations. Human Biology, 73: 129–134. Volume 112, Number 1) Tanabe Y. (1992) Phylogenetic studies on the Japanese dogs, with