<<

J.Anthrop.Soc.Nippon 人 類 誌 98(3):247-267(1990)

Ainu, Japanese, Their Ancestors and Neighbours: Cranioscopic Data

Alexander KOZINTSEV

Institute of Ethnography, Leningrad, U.S.S.R.

Abstract Seven sutural and two foraminal non-metric characters were studied in ancient and recent skulls from and surrounding areas. Variation of three principal traits in most groups can be best explained by a mixture of two components, Jomon and Mongoloid, the latter one increasing in the following order: West Jomon people -East Jomon people - Ainu -Sakhalin Ainu -Ryukyuans - - people -West Japanese - Japanese . In the Japanese and their ancestors beginning from the , the Mongoloid component is very similar to the one represented in modern Chinese. Its introduction to Japan in Yayoi times was abrupt and obviously caused by massive immigration. The further increase of Mongoloid traits from Yayoi to modem times was slow. Traces of Jomon (or Ainu) substratum are seen in all modern Japanese groups which is probably due to an early shift of West Jomon populations to agriculture and their mixture with the immigrants. The Okhotsk group of Hokkaido shows an unusual combination of Tunguso-Manchurian and Ainu features.

Keywords Non-metric cranial traits, Ainu, Japanese, Jomon, Yayoi, Kofun

Kofun (4th to 7th centuries, A.D.) is widely Introduction debated. Views range from considering the Origins of various ethnic groups inhabiting Japanese to be direct descendants of the Jomon Japan are discussed by many physical anthro- people ("transformation theory") to virtually pologists in and outside Japan. The only point denying the presence of a local substratum in the which seems quite certain is the close genetic rela- Japanese race ("substitution theory"). tionship between the Jomon population Somewhere between these extremes are various and modern Ainu. However, the racial affilia- modifications of the "metisation theory" (for a tions of Ainu and their ancestors are still obscure. review of theories see SUZUKI, 1981; Far from solved are the issues of early MIZOGUCHI, 1986). Those who believe that Japanese ethnic history. The role of aboriginal immigrants played a major role differ as to the neolithic populations and immigrant groups source of migration (, , or Siberia). which entered Japan during the periods of Yayoi Mongoloids infiltrated Japan not only from (3rd century B.C. - 3rd century A.D.) and the southwest but also from the north. One of

Article No. 9003 Received May 8, 1990 248 A. KOZINTSEV the northern groups brought to Hokkaido the Natural History Institute, National Science Okhotsk culture (late A.D.) Museum, Tokyo: Jomon (75), and Japanese of attributed to Eskoaleuts by some authors and to the period, 18th-early 19th centuries (109). Nivkhs by others. Over a span of 2,000 Thus, following groups were studied: years, various populations of Japan were in- 1. Jomon and Epi-Jomon (470). Most skulls volved in a process of mixture. How did it in- belong to the late stages (3rd to 1st millennia fluence the biological features of modern B.C.). Due to a very poor preservation, Japanese and Ainu? Those are the main problems chronological sub-division is possible for a few of Japan's racial history. To solve them, physical traits only. Two geographical sub-groups were anthropologists must bring together information segregated: East Jomon and Epi-Jomon (212)- on numerous systems, one of them being cranio- Hokkaido, Tohoku, and Kanto; West Jomon scopy, the study of non-metric cranial traits. (258) -Tokai, Kinki, Chugoku, and Kyushu. In spring 1989 I was granted an opportunity 2. Yayoi of West Japan (158). About a half to undertake a cranioscopic study of several of the series comes from Doigahama site in collections of skulls stored at various institutions Yamaguchi Prefecture. of Japan. These collections provided the basis for 3. Kofun of West Japan (172). the present study. 4.Okhotsk culture of Hokkaido, mostly from Omisaki (30). Materials studied in Sapporo were Materials pooled with a skull from the Shikotan Island The description of minor anatomical details studied in Leningrad. is possible even with badly crushed skulls. The 5. Ainu: Hokkaido (199), and Sakhalin (112); following figures in parentheses refer to a in the latter group, skulls studied in Japan were minimal number of individuals (number of intact pooled with those in the U.S.S.R. (KOZINTSEV, crania plus maximal number of observations for 1988). fragments belonging to one side of the skull). 6. Japanese: Tokyo (114), Tokyo of the Edo Department of Anatomy, Sapporo Medical period (109), West Japan (92), and Ryukyu College: Jomon and Epi-Jomon (76), Okhotsk Islands (115). Due to the isolated position of the culture (29), Hokkaido Ainu (86), and Sakhalin Ryukyuan group, it is dealt with separately. Ainu (3). The pooled Chinese series includes skulls of Department of Anatomy, Kyushu University the late 19th-20th centuries A.D. studied in (Fukuoka City): Jomon (20), Yayoi (158), Kofun Japan and the U.S.S.R. (KOZINTSEV, 1988). (172), and modern Japanese (92). All are from They come from various localities mostly outside Kyushu and southwestern Honshu. mainland China (Pyongyang, Vladivostok, Ulan- Faculty of Science, University: Jomon Bator, , Indonesia). No data concerning (133), Hokkaido Ainu (13), Sakhalin Ainu (41), birth place could be found in the catalogues. and modern Ryukyu Islanders (sensu lato) from Information about other series is given Okinawa, Tokunoshima, and Kikai Islands (115). elsewhere (KOZINTSEV, 1988). The University Museum, the University of Tokyo: Jomon (166), Hokkaido Ainu (100), Methods Sakhalin Ainu (3), Tokyo Japanese of the late The branch of racial osteology which I termed 19th - 20th centuries (114), and modern Chinese "ethnic cranioscopy" is not just studying non- of various regions (69). metric cranial traits. Its main principle, based on Cranioscopy of Ainu, Japanese, Their Ancestors and Neighbours 249 a long-standing tradition of Soviet physical intersection with the palatal spine nor with the anthropology, is the rigorous selection of a few point where it touches the median palatine suture most informative traits. The heuristic nature of (There are some exceptions; see KOZINTSEV, this principle was shown in many studies, in- 1988). cluding those of Japanese somatology. When 7. Metopic suture (MS). traits are chosen at random, the classification of 8. Supraorbital foramen (SOF). Following Japanese populations is rather erratic and DODO (1987), any foramina on the supraorbital virtually no conclusions regarding ethnic history margin opening into the orbit were included. may be drawn (KOUCHI, 1983). However, when 9. Hypoglossal canal bridging (HGCB). Only informative traits are used, the classification of complete bridging was included. the same groups becomes far more suggestive and Sutural traits nos. 1 to 6, shown to be very obviously reflects population events of the past useful in racial studies, were described in detail (LEVIN, 1971). (KOZINTSEV,1988). No. 7, metopism, may also Until recently, cranioscopy was dominated by be informative in some cases (KOZINTSEV,1975, principles of the numerical taxonomy. Students 1988). The diagnostic value of two foraminal tried to use as many traits as possible without variants, nos. 8 and 9, was demonstrated by much concern about their value. Only during the DODO (1987) and TOMASHEVICH (1988). Only last years did some authors arrive at a conclu- crania of adults were studied. For all traits, except sion that a few good traits are better than many no. 2 (SMS), male and female skulls were pooled; bad ones (DODO, 1987; TOMASHEVICH, 1988; for SMS, unweighted male-female averages were KOZINTSEV, 1988). calculated. In the present study, nine traits are used: For further analysis, traits are combined. This 1. Occipital Index (01). "A" being the is usually done by means of multivariate number of sides with a simultaneous presence of methods. The most widely used method, cluster wormian bone(s) in the occipito-mastoid suture analysis, appears to be the least appropriate and absence of wormians in the inferior portion because it is unsuitable for studying racial of the lambdoid suture, and "B" denoting the mixture. The intermediate position of hybrid number of sides with a reverse combination, OI populations is not reflected by the dendrograms, =A/(A+B), sides with asterial bones not which results in a distortion of true relationships. included. When groups from Japan are studied, the Jomon 2. Sphenomaxillary suture (SMS). people cluster with Ainu while the Japanese 3. Transverse zygomatic suture posterior trace together with other Mongoloids form another (TZST), from 2 mm on. cluster (HANIHARA, 1985; MIZOGUCHI, 1986; 4. Infraorbital pattern type II (IOP II) - OSSENBERG, 1986; WANG, 1987; DODO, 1988; sagittal Infraorbital suture intersected by the BRACE et al., 1989). This confirms the obvious zygomatic bone. fact that Japanese, contrary to Ainu, are de- 5. Infraorbital pattern complexity index finitely Mongoloid. But this does not tell us (IPCI)-frequency of sides with accessory whether or not the Japanese have any common (transversal or other) Infraorbital sutures. racial traits with Ainu and the Jomon people. 6. Transverse palatine suture index (TPSI) - Factor analysis, principal coordinates, and usually, frequency of sides with the foremost other related methods are better since they give point of the suture coinciding with neither its information about the status of hybrid groups.

250 A. KOZINTSEV

Two-dimensional reduction of data show them as Mongoloids? To answer this question, definite gradients which are sometimes explainec we need some point of reference because there by admixture and sometimes by diachronic is no fixed set of Mongoloid traits. One reference changes. Typical examples are: Jomon-Ainu population are the Caucasoids. In each trait, the -Japanese; Ainu-Japanese-; extreme Caucasoid value is taken for zero point, Jomon-Yayoi-Kofun-Aapanese; Jomor the extreme Mongoloid value for 100% point, -Japanese-Chinese (or Koreans); Jomon- and the position of a group on this scale is Ainu-Baikal Neolithic; Jomon-Yayoi- averaged over several traits. Formerly, this Mongols (TAGAYA and IKEDA, 1976; SUZUKI, Mongolo-Caucasoid Index, MCI, was based on 1981; HANIHARA, 1985; DODO, 1986; five traits (KOZINTSEV, 1988) but now, due to MIZOGUCHI, 1986; DOI and TANAKA, 1987; the works of DODO (1987) and TOMASHEVICH WANG, 1987; MOURI, 1988; YAMAGUCHI, (1988), we may add one more independent 1988). The resulting pattern of relationship; variant, SOF. Taking that in Eurasia this trait depends on the choice of groups. Since no varies from 20% (0.927 radians) in Caucasoids preliminary model or hypothesis is required, the to 70% (1.982 radians) in Mongoloids, the range random element is almost inevitable. being 1.055 radians, we may add to MCI the sixth It appears more productive to begin with term, according to the formula hypothesis which must be tested. This approach which proved to be very helpful in the study of Japan's racial history (DoI and TANAKA, 1987; HANIHARA, 1987), will be employed in the More economically, the expanded MCI may be calculated thus: present paper. The main testing tool is provided by the inter-racial scaling method introduced by DEBETS (1968). In each trait, the position of a MCI(6)=5/6MCI(5)+15.798SOF-14.645. group is estimated on a scale whose limits are defined by two postulated racial components, and As can be seen from Table 1, there is little the estimate is averaged over several traits. The difference between values of MCI (5) and MCI characters used here were shown to be virtually (6) in Mongoloid groups. However, in Ainu and independent on an individual level (KOZINTSEV, especially in the Jomon people, MCI (6) is much 1988; TOMASHEVICH, 1988) which makes com- lower than MCI (5). According to the expanded plex statistical procedures unnecessary. index, only Sakhalin Ainu are close to Siberian Mongoloids while the Hokkaido Ainu and the Results and Discussion Jomon people are considerably less Mongoloid Per-side frequencies and indexes are given in (about the same as the Japanese, Chinese, and Table 1. The indexes were computed according Indonesians). Yet, the overall estimate conceals to published formulae (KOZINTSEV, 1988). enormous variation not present in the Before any calculations, frequencies (p) were Mongoloids. Examining the six terms of MCI transformed into angles measured in radians separately, we see that they range from 35% to (phi), so that phi =2 aresin *p, and BARTLETT's 65% in modern Japanese, from 35% to 73% in correction was applied when necessary. Chinese, and from 31% to 81% in Indonesians. The first question concerns racial affinities of However, in the Jomon series the range is from Ainu and their neolithic ancestors. Can we regard a negative value, -18% ("super-Caucasoid") Cranioscopy of Ainu, Japanese, Their Ancestors and Neighbours 251 for SOF to 125% ("super-Mongoloid") for is rare also in Caucasoids and African Blacks, TZST. A similar, though milder, contrast is seen and low frequencies of IOP II are typical also in the Ainu. Such fluctuations make the average for Australoids, Africans, and Amerinds. Yet, estimate senseless because, with only six traits, the zone of maximal frequencies of TZST is quite the confidence interval is too wide. close to Japan, spreading from lower Amur into Another reference point is given by the only East Siberia and Mongolia. Such distribution is Australoid series at our disposal, Papuo- hardly due to chance. Rather, it seems to reveal Melanesian. In the Mongolo-Australoid Index, ancient local roots of the Ainu-Jomon complex. MAI, based on this series, the opposite end of Turning to other groups from Japan, we see the scale is provided by average Mongoloid rather a clearer picture. Compared to the people of than extreme Mongoloid frequencies which ex- Northern and Central Asia, modern Japanese as plains the higher values and larger variance of as Chinese and Indonesians are considerably MAI as compared to MCI. Here, Ainu and less Mongoloid (Table 1). Contrary to a mosaic Jomon people look on the average as Mongoloid pattern observed in Ainu and the Jomon people, as Siberians, and much more Mongoloid than the reduction is seen in all 7 traits included in Japanese, Chinese, or Indonesians. But again, MCI and MAI. Thus, formally, we may speak terms for single traits fluctuate strongly, from of either a Caucasoid or an Australoid shift. super-Mongoloid (TZST) to super-Australoid Essentially, the first possibility is of course ruled (IOP II). out. As to the Australoid tendency, its presence Our results confirm what was already well in the Mongoloids of East and Southeast Asia known: the trait combination in Ainu and their is confirmed by other systems and may be due ancestors is absolutely unique. My earlier belief to early contacts of Mongoloid and Australoid that Ainu cranioscopic complex is Mongoloid groups on the Asian mainland. The Jomon (KOZINTSEV,1988) was based on the study of the population could not be the source of this Sakhalin series only, which was an unhappy tendency. choice. If we consider the Proto-Ainu type to be The Proto-Japanese of the Yayoi and Kofun best represented by the Jomon series (and there periods at first sight appear to be more are good reasons for that), Sakhalin Ainu appear Mongoloid than modern Japanese. According to to be the most mixed population because in 7 MCI (6) they are between the Japanese and traits out of 9 they are further from the Jomon northern Mongoloids, and according to MA!, group than are the Hokkaido Ainu (Table 1). their place is in the lower part of the northern This view is confirmed by other studies (LEVIN, Mongoloid range. Craniometrically, the Proto- 1958; YAMAGUCHI, 1973; MIZOGUCHI, 1986; Japanese also display a Siberian tendency MOURI, 1988). (HANIHARA, 1985; MIZOGUCHI, 1986, 1988; One trait, however, deserves special attention WANG, 1987; YAMAGUCHI, 1988). But - TZST . Its frequency in modern Ainu is the examining each trait separately we see that only world's highest, but in the Jomon people it is still in OI do the Yayoi and Kofun people deviate higher. The trait is well studied (KOZINTSEV, towards the Siberians, the number of observa- 1988), and it is certain that among other peoples, tions being very small here. In most other high frequencies are met only in northern diagnostic characters the shift is towards the Mongoloids. Two other characteristic features of Jomon group. It refers primarily to IOP II, the Jomon people and Ainu are less specific: SOF TPSI, and, in Yayoi people, also to MS and SOF. 252 A. KOZINTSEV

Table 1. Values of cranioscopic traits and indexes (%)

Note: Numbers in parentheses after names of groups are minimal numbers of individuals, those after trait value denote sample sizes, that is, numbers of two orbital plus two temporal sides for SMS, and left plus right sides for other lateral traits. See text for abbreviations. * Estimated by regression due to lack of data ** After DODO (1987) and TOMASHEVICH(1988)

Frequencies of SMS and TZST do not contradict use the Continental Mongoloid Index, CMI, the hypothesis that differences between the Proto- based on TZST and IOP II (Table 1). It equals Japanese and their descendants are caused mainly zero in Indonesians, 3 to 42010in Eskoaleuts and by a stronger Jomon component in the former. Chukchi, 43% in Chinese, and 100% in average This does not exclude the possibility of Siberian Mongoloids of Siberia and Mongolia (KOZINT- connections (see below). SEV, 1988). In groups from Japan, except Let us now turn to the Okhotsk series which Okhotsk, CMI ranges from 46% to 88%, and is quite peculiar. MCI and MAI may not be in the Okhotsk series it reaches 103%, a value estimated in this case because frequencies of some typical for the populations of Inner Siberia and traits are based on a very small number of obser- sharply different from what is seen in Eskimos vations. Values of 01, TZST, and TPSI are and Aleuts. within the Mongoloid range. In SOF, however, After this general racial assessment, we may the Okhotsk people are similar to Ainu. To try to specify the information using a local in- evaluate the racial status of this group, we may stead of global scale and a limited instead of full Cranioscopy of Ainu, Japanese, Their Ancestors and Neighbours 253

Table 1. Continued from the opposite page

set of characters. To select optimal traits, we shall frequencies as units of analysis. Statistically examine five amalgamated groups: 1) Jomon significant results are easier to obtain with the people, 2) Ainu, 3) Proto-Japanese of the Yayoi first method, but the second approach is more and Kofun periods, 4) Japanese of the 18th to Table 2. Discriminative power of single traits 20th centuries, except the Ryukyuans, and 5) Ryukyuans (the Okhotsk series is omitted because of incomplete data). Traits differentiating these group most efficiently will be considered optimal. Two methods will be employed. First, each of the 5 groups will be regarded as a whole, and for each side of the skull, a 2* 5 table will be analyzed by means of the chi-squared test. According to the second method, the original structure will be retained, Jomon, Ainu, and Proto-Japanese groups including two series each, modern Japanese, three, and Ryukyuan, one. Here, the Note: For chi-squared and between-group variance, there discriminating power of the traits will be assessed are 4 degrees of freedom; for within-group variance, 5. Chi-squared values refer to a side with a stronger by comparing between-group to within-group discriminating effect. For ANOVA, the sides are pooled. variation using the ANOVA with transformed * P<0 .05; ** P<0.01;* P<0.001. 254 A. KOZINTSEV

Fig. 1. Frequencies of transverse zygomatic suture posterior trace (TZST) and infraorbital pattern type II (IOP II). American Indians: 1-South America, 2-North America; Eskoaleuts and Chukchi: 3-Eskimos, 4-Aleuts, 5-Chukchi, 6-Welen, 7-Ekven; Tunguso-Manchurians: 8-Negidals, 9-Orochi, 10- Ulchi, 11- Nanays, 12-Evenki; Turks and Mongolians: 13-Tuvinians, 14-Kalmyks, 15-Mongols, 16-Telengits, 17-Cis-Baykal Buryats, 18-Trans-Baykal Buryats, 19-Yakuts, 20-Shorians, 21-Khakassians; Uralians: 22-Khants, 23-Mansi, 24-Nenets. informative (Table 2). Japanese but sharply distinct from their direct A really strong discriminating effect is shown descendants, Ainu. Thus, high incidence of by 4 traits, TZST, IOP II, SOF, and HGCB, the HGCB appears to be a specific and rather late first one being especially powerful. Yet, looking of Ainu, and a similarly high occurrence at Table 1 and Figs, l and 2, we notice that, con- in Nivkhs is probably due to chance. It is worth trary to the first 3 characters which are racially mentioning, however, that in Ryukyuans, HGCB meaningful, HGCB presents an erratic pattern. is a little more frequent than in other Japanese In this trait, Jomon people are close to the groups. Whether or not this points towards a Cranioscopy of Ainu,Japanese,Their Ancestors and Neighbours 255

Fig. 2. Frequencies of supraorbital foramen (SOF) and hypoglossal canal bridging (HGCB). Japanese: 1- Tokyo, 2-Kanto, 3-West Japan, 4-Tohoku, 5-Edo, 6-Kofun; Eskimos: 7-Siberian, 8-Alaskan, 9-Canadian. Groups studied by Y. DODO but not by me are marked by vertical bars. specific relationship of Ryukyuans with Ainu (see components. First of them is easily identified as below) is not clear because the value of this trait Jomon (Figs. 1, 2). In all 3 traits, West Jomon remains uncertain. people occupy an extreme position while the East So we are left with 3 traits, TZST, IOP II, and Jomon group shows a slight shift towards other SOF. Previously they were shown to be in- populations. dependent on the within-group level (KOZIN- The second component obviously belongs to TSEV, 1988; TOMASHEVICH, 1988). In the the Mongoloid race. A more detailed diagnosis modern Japanese series, as well, no correlation is difficult since we have very few comparable between them was found. However, on the inter- data from adjacent territories. We shall attempt group level, the situation is quite different. to specify the racial status of the hypothetical Among 36 coefficients of SPEARMAN'srank cor- Mongoloid component putting on its place relation between 9 traits in 10 groups from Japan various groups available to us. Transformed fre- (Okhotsk group not included), three are signifi- quency of each of the 3 traits in such a group cant at the 1% or even 0.1% level: will provide the 100% point, and the frequency TZST vs. IOP II: -0.95 in the West Jomon group will be taken for zero TZST vs. SOF: -0.87 point. On this scale, the position of each group IOP II vs. SOF: +0.93. from Japan (except Okhotsk) will be estimated, Such a high correlation between independent and the estimates will be averaged over 3 traits. traits is usually caused by a mixture of two racial The average score will be denoted Mongolo- 256 A. KOZINTSEV

Jomon Index, MJI. For obvious reasons, the si2 being the variance of MJI in i-th of the 9 higher the concordance between estimates based groups from Japan. In other words, the fitness on single traits, the more realistic would this two- of a certain two-component model will be component model appear. For each group from assessed by the stability of between-group correla- Japan, the standard deviation of MJI will be tions after the introduction of an outside group taken as a measure of discordance, and for the as one of the components. It would seem that, entire model, discordance will be measured by the with only 3 traits, it is almost impossible to ob- average standard deviation, s: tain significant results. However, the high con- cordance of these traits in Japan leaves room for optimism. Ten models will be tested.

Fig. 3a. Mongolo-Jomon Index (MJI), based on frequencies of TZST, lOP II, and SOF, in the populations of Japan: Evenkian model. For each group, minimal of the three estimates, average estimate (MJI, marked by a circle), and maximal estimate are shown.

Fig. 3b. Same as Fig. 3a: Ulchian model. Fig. 3c. Same as Fig. 3a: Orochian model. Cranioscopy of Ainu, Japanese, Their Ancestors and Neighbours 257

Fig. 3d. Same as Fig. 3a: Nanay model.

Fig. 3e. Same as Fig. 3a: Negidal model. Fig. 31. Same as Fig. 3a: Buryat model. The results are presented in Figs. 3a-3j. The order of MJI values is almost the same in all cases: West Jomon-East Jomon-Hokkaido Ainu-Sakhalin Ainu-Ryukyuans-Yayoi -Kofun-West Japanese-Edo Japanese- Tokyo Japanese (in only one model there are two permutations affecting adjacent positions). Similar ordering of groups is seen in other systems (TAGAYA and IKEDA, 1976) and pro- bably it reflects a true Mongoloid gradient in Japan. Our results confirm that Jomon traits in the Japanese decrease from Yayoi times to the

Fig. 3g. Same as Fig. 3a: Mongolian model. 258 A. KOZINTSEV

Fig. 3h. Same as Fig. 3a: Indonesian model.

Fig. 31. Same as Fig. 3a: Baikal Neolithic model. Fig. 3j. Same as Fig. 3a: Chinese model. modern period (SUZUKI, 1981; YAMAGUCHI, to be quite different in their ability to explain 1982). empirical data. In most cases, the hypothetical However, the shapes of the graphs are Mongoloid component was definitely very unlike strikingly dissimilar. Though for each group the one represented in modern Evenki, Tunguso- separately, the heterogeneity of variances in Manchurian peoples of Amur (Ulchi, Orochi, various models, as measured by BARTLETT's Nanays, Negidals), Buryats, Mongols, or (SNEDECORand COCHRAN, 1980) test, does not Indonesians, the average standard deviations reach significance level, ranking of variances ranging from 18% to 52%. A usually high within each group and comparing sums of ranks positive correlation between MJI and its variance for the ten models yields a highly significant value within a model is easy to interpret: the closer a of the FRIEDMAN'S (SIEGEL, 1956) test (x2= hybrid population is to one of the parental races, 39.8. d.f.=9. P<0.0011. showing the models the stronger effect is produced by the variation Cranioscopy of Ainu, Japanese, Their Ancestors and Neighbours 259 within this race upon the estimated proportion opportunity for racial analysis. Of course, the of the corresponding parental component, and satisfactory overall result does not imply that the the larger becomes the distortion caused by an Chinese model is equally fit in all cases. Presently incorrect choice of a specific parental group. Only we shall analyze specific situations, proceeding rarely do the empirical data fit into the above- in the increasing order of MJI. mentioned models. One example is provided by Jomon the Sakhalin Ainu who, theoretically, could very As previously said, in all 3 key traits the well be a mixture of Jomon people with Ulchi Jomon specifics is stronger in the western group or Orochi. while the eastern shows a slight Mongoloid A different pattern is seen when the Baikal tendency. The differences are not significant, and neolithic group is used, the average standard this refers also to other traits, except TPSI in deviation being only 11.5 %, with only little which East Jomon group is again a little more tendency of increasing towards the Mongoloid Mongoloid. Smallest variance of MJI for this end of the scale. However, the result is still better series is obtained by using trait frequencies not if we use the modern Chinese series. Here, the in the Chinese but in Evenki, Nanays, and graph looks like a narrow strip (s=8.6%), and Mongols. Yet, even in the latter cases, MJI is not there is no correlation between MJI and its significantly different from zero. Addition of variance. With the introduction of the Chinese TPSI does not alter the situation. as an "out-group", the between-group correla- Partitioning into local groups reveals maximal tion is not merely retained but becomes even deviations from the typical Jomon trait complex stronger (TZST vs. TOP II:-0.96; TZST vs. not in the north (Hokkaido) where they could be SOF:-0.90; IOP II vs. SOF:+0.95), an effect explained easier but rather in the center (Kanto). which is not present in any other model. So our Summing up, it should be said that the data are consistent with the hypothesis that the differences between the eastern and western series differences between most groups are caused may be random. Previous authors stated that the mainly by variations in the proportion of two Jomon population was homogeneous (YAMA- components, aboriginal (Jomon), and immigrant GUCHI, 1982; DODO,1986), and this conclusion (Mongoloid), the latter being very close to the one remains unchallenged. which is retained in modern Chinese. MJI values and their standard errors, based Hokkaido Ainu on the "Chinese model", are as follows (s.e.= Our data, as well as results obtained in most s/*3): East Jomon 7.6*4.1%; Hokkaido Ainu other studies, show quite definitely that 21.2*2.9%; Sakhalin Ainu 38.9*7.6%; Hokkaido Ainu are direct descendants of the Ryukyuans 57.4*8.6%; Yayoi people Jomon people mixed with the Mongoloids. How 59.1*3.7%; Kofun people 68.0*5.2%; West large is the Mongoloid component in the group Japanese 70.3*3.6%; Edo Japanese 80.4* that we have studied? In models with smallest 0.9%; Tokyo Japanese 82.4*3.2%. Such esti- variances (Baikal neolithic, Negidal, Chinese), mates should certainly be viewed differently from MJI equals 21-24% and is significantly different those based on gene frequencies. Yet, though the from zero. genetic nature of the anatomical variants Historical data may help us choose a more examined is unknown, the exceptionally high con- realistic model. Let us first suggest that the main cordance of three independent traits offers a good source of the Mongoloid component was the 260 A. KOZINTSEV population of the Okhotsk culture (this seems to the other two. Presently we can merely state plausible since our Hokkaido Ainu series dates that the proportion of the Mongoloid component from the period before the beginning of an in- in the Hokkaido Ainu series is likely to be tensive Ainu-Japanese intermarriage). But the somewhere between 1/7 and 1/5. MJI estimate based on the hypothesis of East Jomon people (or their descendants) being mixed Sakhalin Ainu with the Okhotsk people is quite unsatisfactory: As compared to the Hokkaido Ainu, this 30.9*16.7%. Due to a very large error, this group is definitely more mixed with the value, though obviously overestimated, is Mongoloids. The situation is more complicated insignificant. here because the Mongoloid admixture could On the other hand, if the Mongoloid admix- have been added not to the original Jomon gene ture came mainly from the Japanese, then, using pool but to that of Hokkaido Ainu already the pooled Japanese series as a 100% point and affected by mixture with other Mongoloid the West Jomon group as a zero point, we ob- populations. tain for Hokkaido Ainu an MJI of 27.3*4.4% In this case too, the Okhotsk model is the least (P<0.05). If the East Jomon and a pooled satisfactory, MJI being 64.8*33.1% if the East Jomon group are used as zero points, MJI equals Jomon group is taken for zero point, and 18.6*9.3% and 24.1*6.0%, respectively, both 67.9*51.4% if it is replaced by the Hokkaido estimates being insignificant. Ainu. Maybe the standard set of traits must not It is also possible that intermixture began early be used here? But the Okhotsk people differ from but occurred mainly in the central or southern the Jomon people mainly in the same 3 parts of Japan rather than on Hokkaido, and the characters. Probably, very large errors are due principal source of Mongoloid genes were the to a small size of the Okhotsk sample. Proto-Japanese. Then, if we replace the modern Substituting Okhotsk people by the Nivkhs makes Japanese by the Yayoi people in our model, MJI the errors smaller but the estimates remain in- in the Hokkaido Ainu would rise up to significant (50.3*20.3% and 33.7*10.2%, 36.8*7.3% with the West Jomon group as zero respectively). point, 26.2±13.4% with East Jomon group, and Much better results are obtained with the use 33.0*9.2% with the pooled Jomon group, only of two Tunguso-Manchurian groups from Amur. the first estimate being significant. If, for instance, the Sakhalin Ainu were a mixture Yet, it should be remembered that in the of the descendants of East Jomon people with Japanese population, a Jomon substratum is Ulchi, the proportion of the Ulchi component present, and it is still more pronounced in the would be 38.4*3.4%. Replacing Ulchi by Orochi Yayoi population (see below). Therefore, the true yields an MJI of 36.1*1.9% (both estimates are amount of the Mongoloid component in our significant at the 1 % level). If we put Hokkaido Hokkaido Ainu group is less than the estimates Ainu in place of the Jomon people, MJI would given above. Probably, it is close to MJI values become 23.4*6.1% or 22.2*6.6%, respectively. which are obtained with the Chinese model, And, considering that the mixture of Chinese and 21.2*2.9%, 14.1*7.0%, and 18.5*4.4%, for East Jomon components would produce an MJI West, East, and pooled Jomon options, of 14.1% for Hokkaido Ainu, the Mongoloid respectively. Here too, only the first value is component received by the Sakhalin Ainu from significant but there are no reasons to prefer it the Hokkaido Ainu would be close to (1-0.234) Cranioscopy of Ainu, Japanese, Their Ancestors and Neighbours 261

*0.141, that is, 10.8%. Ulchi are 100% features the Ryukyuans are similar to the Mongoloid, so from them the Sakhalin Ainu southern Mongoloids (TAGAYA and IKEDA, could receive 0.234*1.000, or 23.4%. The total 1976; WANG, 1987). amount is 34.2%, sufficiently close to 38.4%, an Whether or not the similarity of the Ryukyu MJI value which would result from a direct people and Ainu implies their specific relation- mixture of the ancestors of Ulchi with the ship or reflects only a common Jomon heritage, descendants of the East Jomon people. Sub- is difficult to say because most traits setting these stituting Ulchi by Orochi we have 33.2% which 2 groups apart from the Jomon group may be is also close to a "direct" estimate, 36.1% . explained by either a mixture with the In our calculations we disregarded the Mongoloids or usual diachronic changes. In our "Mongoloid" shift of the East Jomon group program, only HGCB apparently belongs to since it may be only a local peculiarity. But it neither of these categories (see above) but in this should be remembered that if frequencies in the trait the Ryukyuans are much closer to the Jomon West Jomon group were used, MJI would in- people (and to the Japanese) than to Ainu. Our crease by 5-6%. Anyway, the proportion of the data allow to say only that in the Ryukyu people, Mongoloid component in the Sakhalin Ainu is contrary to Ainu, the Mongoloid component probably no lower than 1/3. predominates over the Jomon component. If the According to historical and ethnological West Jomon group is taken for zero point, MJI sources, Ainu lived side by side with Ulchi and in Ryukyuans ranges from 57% to 70% in Orochi on Sakhalin and lower Amur which various models, only the Indonesian model resulted in both cultural exchange and inter- yielding 85%. marriage (SMOLYAK, 1975). The mixture with Despite craniometric evidence suggesting the Nivkhs probably also occurred, and it would southern connections, the Indonesian model is be premature to deny it on the basis of our again one of the worst, the same as with most results. The Nivkh cranial series is small and its other groups from Japan, while the smallest origin is not quite certain. variances are produced by using frequencies in the Negidal, Chinese, and Baikal neolithic Ryukyuans In most systems of traits this group is distinct groups. For geographical reasons, the Chinese model should be preferred, with an MJI of from other Japanese populations by a tendency 57.4*8.6%. In spite of a fairly large error, this which in a synchronic aspect aligns the estimate is significantly different from zero, from Ryukyuans with Ainu (BALZ, 1911; LEVIN, 100%, and from MJI in the Hokkaido Ainu. 1971; HANIHARA, 1983), and in a diachronic Though it is not significantly different from MJI aspect, with the Jomon people (TAGAYA and values in other Japanese populations, the Jomon IKEDA, 1976; YAMAGUCHI, 1982; MIZOGUCHI, 1986). Accordingly, this tendency may be inter- (or Ainu) tendency of the Ryukyuans is quite significant in single traits, TZST and IOP II. preted as either a fact supporting the idea of Ainu migration from the south (LEVIN, 1971) or Yayoi people merely a result of a less intensive influx of As indicated by the Japanese anthropologists, Mongoloid immigrants into isolated areas of the the Yayoi population was very heterogeneous, extreme south and extreme north of Japan some groups being indistinguishable from the (HANIHARA, 1983). In some craniometric Jomon people and others apparently belonging 262 A. KOZINTSEV to an immigrant Mongoloid type (KANASEKI, indicating that here as well, the presence of the 1966 quoted by SUZUKI, 1981). According to Jomon substratum is beyond doubt. craniometrical evidence, the immigrants con- Japanese centrated in northern Kyushu and westernmost If the Ryukyuans are excluded, differences Honshu, places whence most of our series comes. between the remaining 3 Japanese series do not However, in this group too, the Jomon reach significance level in any of the 3 traits. It substratum is quite appreciable. According to the should be mentioned, however, that in West best model, Chinese, the MJI for the Yayoi series Japanese, the increase of MJI over that of the is 59.1*3.7% if the Jomon component is re- is only 2-8% (disregarding the two presented by the western group. So we may state most aberrant models), while in two series from with a 95% probability that the proportion of this Tokyo which are very similar, the increase component is no lower than 1/4. amounts to 12-22%. According to the Chinese In "northern" models (Evenki, Tunguso- model which fits the empirical data even better Manchurians of Amur, Buryats, Mongols), MJI here than in the case of Proto-Japanese groups, is larger but its errors are 3 and more times as MJI estimates are as follows: West Japanese, large as in the Chinese model. With the Baikal 70.3*3.6%; Edo, 80.4*0.9%; Tokyo, neolithic series, the result is somewhat better 82.4*3.2% (all are significantly below 100%). (68.6*7.5%) but the error is still twice larger The ANOVA reveals with a 95% probability that than in the Chinese model, and the presence of West Japanese are less Mongoloid than the a Jomon component is established with only a other two populations. In the pooled Japanese 90% probability. sample, MJI equals 78.2*1.9%; replacing West Kofun people Jomon by the pooled Jomon group has virtually Recent calculations based on demographic and no effect (77.4*1.9%). craniometric data suggest that the proportion of On the second place, same as in the case of Mongoloid immigrants in West Japan reached Ryukyu and Yayoi groups, is the Baikal neolithic 80-90% during the Kofun period (HANIHARA, model with MJI of 81-96% (not significantly 1987). Our MJI estimates are a little lower, different from 100%). For the Kofun people, this 68-81% if we disregard the aberrant Indonesian model was on the 3rd place, and for Hokkaido model. The increase in the proportion of the Ainu it was even the best one. This deserves Mongoloid component is only 6-10%. In the best attention if we remember about the increased model (Chinese), MJI equals 68.0*5.2%. Yet in MCI and MAI in the Yayoi and Kofun groups this case, the Buryat model is almost as good (see above), about the craniometric similarity of (71.9*5.4%). Interestingly, all 8 northern these groups to northern Mongoloids (HANI- models produce better results with the Kofun HARA,1985; YAMAGUCHI, 1988; MIZOGUCHI, group than with the Yayoi group while for the 1986, 1988; WANG, 1987), about the parallels Chinese and Indonesian models the opposite is between the Japanese immunoglobulin complex true. Does this mean that Japan's ties with Siberia and that of modern Baikal populations (MATSU- or Mongolia became stronger during the Kofun MOTO, 1987), and about the relationship of the period? Japanese language with the Altaic stock. Anyway, both estimates, "Chinese" and However, the culture of wet rice obviously was "Buryat" , are significantly different from 100% not introduced to Japan from Siberia or Cranioscopy of Ainu, Japanese, Their Ancestors and Neighbours 263

Mongolia. Probably it is possible to avoid this in the Yayoi period to 77.4% in the modern contradiction by considering that no later than period, that is, the change of the immigrant to the neolithic period, northern Mongoloids pene- aboriginal elements ratio by a factor of 2.5 per trated into north-eastern China where sub- 2,000 years, would imply that their coefficients sequently they could have mixed with the Pacific of annual increase differed by only 0.05%. Mongoloids and adopted the wet rice agriculture Similar calculations for the period from Yayoi (GOKHMAN and RESHETOV, 1981). Until we to Kofun (1.5-fold increase per 500 years) yield obtain comparative materials from this territory a difference of 0.08%. or Korea, it remains to be repeated that, But how can we reconcile these modest according to our data, the Chinese model is the estimates with the conclusions of the Japanese best both for the Proto-Japanese and especially demographers stating that the spread of wet rice for the modern Japanese groups. agriculture in West Japan was paralleled by a So if the main Mongoloid component is the very intensive population growth? At the end of same in Japanese and Chinese, the "deficit" of the Jomon period, the population of Japan was this component in the former as compared to the close to 75.8 thousand, and by th end of Kofun latter is likely to be somewhere between 1/5 and period it reached 5.4 million, that is, a 70-fold 1/4 which is significant at the 10%level. There increase occurred (KOYAMA, 1979 quoted by are sufficient reasons to believe that this estimate HANIHARA, 1987). HANIHARA points out that represents the proportion of the Jomon (or Ainu) this could be due to either a natural increase substratum in the modern Japanese population, which appears unusually high for that epoch except the Ryukyuans where it is higher. (0.43% per annum) or an intensive immigration. Let us now return back to trace the dynamics Yet, for the aboriginal Jomon population the of Mongoloid traits in the Japanese and their same demographic study failed to reveal any ancestors. There are as yet no indications that tendency of growth. In fact, there are indications Mongoloids infiltrated into Japan before the that towards the end of the neolithic period a Yayoi period. If the Late/Final Jomon and Epi- decline has occurred. A prolonged co-existence Jomon skulls are taken as a separate group and of these strikingly different demographic the Chinese model is applied to them, MJI value tendencies would inevitably result in a complete is negative (-4.7%), showing that not a slightest disappearance of any traces of the Jomon sub- Mongoloid tendency is present. However, in the stratum in the anthropological type of the Yayoi period the Mongoloid component makes Japanese. Yet, this did not happen. The question a sharp rise and becomes predominant in West is, why? Japan which supports the idea of a massive The principal reason seems to be that already immigration from the continent (HANIHARA, during the Yayoi age, the major part of the 1987). aboriginal population of West Japan mixed with The further increase of Mongoloid trait com- the immigrants, shifted from hunting and plex was slow: from the Yayoi age to the modern gathering to agriculture and, having acquired a times, MJI grew by 20 units, so the average rate higher increase rate, began to progressively out- of advance was close to one unit per century. number those aboriginal groups which were not Regarding the MJI as a very rough approxima- involved in the process of mixture. It may be tion of the true proportion of immigrants (or speculated that the latter were the ancestors of their genes), the shift of MJI value from 57.8% Ainu. A similar economic and demographic 264 A. KOZINTSEV differentiation occurred, for instance, within the classification but has nothing to do with types of aboriginal population of Siberia after the advent economy. Thus, Mongoloid hunters of the Baikal of the Russians. neolithic are much further from the Jomon It was more than once stated that in West people (who are also neolithic hunters but not Japan where conditions for wet rice agriculture Mongoloids) than from Mongoloid farmers, are favourable, most Jomon populations adopted Chinese and Japanese. the farming economy (AIKENS and HIGUCHI, Okhotsk people 1982; AKAZAWA, 1986), and this idea is con- Both culturally and biologically, this group firmed by our results. It can probably also ex- appears to be an alien element in Japan. Its racial plain the stronger Jomon tendency that we status is difficult to assess because our sample is observed in modern western Japanese. very small. The overall impression is that of However, none of the present data support the Siberian Mongoloids (see above). The combina- "transformation theory" stating that under the tion of TZST and IOP II aligns the Okhotsk influence of agriculture, Jomon people changed people with the Tunguso-Manchurians of Amur into Yayoi people and then into modern (Fig, 1). Craniometrically, the group is also close Japanese, the role of immigrants being minimal to Ulchi and Nanays (YAMAGUCHI, 1975; (SUZUKI, 1981). This theory implies that, several ISHIDA,1988). Archaeological evidence indicates millennia after the emergence of the Mongoloid that the Okhotsk culture included a Proto- race on the continent, it spontaneously emerged Tunguso-Manchurian (Mo-ho) component from again in Japan, and the outcome was strikingly Amur (DYAKOVA, 1982). similar over the whole complex of independent On the other hand, frequencies of SOF and systems of traits. It certainly seems utterly im- HGCB link the Okhotsk series with Ainu (Fig. plausible that the biological similarity of Japanese 2). Craniometrical data also seem to point to Chinese and Koreans is due to a common towards a possible Ainu admixture type of agriculture. Farming economy emerged several times in various parts of the world but, (YAMAGUCHI, 1975 quoted by ISHIDA, 1988). However, in a mixed group, an intermediate as far as we know, it never resulted in racial con- rather than mosaic combination could be ex- vergence. To be sure, several diachronic bio- logical tendencies sometimes went parallel with pected. Similarity to Eskimos, Aleuts, or Nivkhs is not seen in any of the 3 key traits. Yet, no the spread of farming. This may be true for definite results may be obtained with such a small gracilization, increasing frequency of dental sample. caries, and perhaps some other disease (KOZINT- SEV, 1980). There are also somatic traits, such Conclusions as body length or head index, which are labile under environmental changes (KOUCHI, 1983). 1. Of the 9 cranioscopic traits examined, But several dozens independent racial features TZST, IOP II, and SOF turned out to be most linking the Japanese with other Mongoloid informative for the racial analysis of Japan's peoples were shown to be remarkably stable. populations. These independent characters show Cranioscopic traits are rather illustrative in a high between-group correlation indicating that this respect. Looking at Fig. 1, it is easy to see the main factor of race formation was the that the position of groups complies sometimes mixture of two components, Jomon (or Ainu), to a linguistical and sometimes to a racial and Mongoloid. Cranioscopy of Ainu, Japanese, Their Ancestors and Neighbours 265

2. The Jomon complex, represented with some habitants of West Japan, the Jomon element is modifications in modern Ainu, has no analogies stronger than in Tokyo Japanese. outside Japan. Its most striking feature, an ex- 6.Okhotsk people of northern Hokkaido are tremely high frequency of TZST, links Jomon close to Tunguso-Manchurians of Amur in some people and Ainu with northern Mongoloids but traits and to Ainu in other traits. on the whole neither Ainu nor their neolithic ancestors may be attributed to the Mongoloid (or Acknowledgments any other) major race. I am deeply obliged to my host scientist, Prof. 3. The East Jomon group, as compared to the Y. DODO of the Sapporo Medical College, who West Jomon group, shows a slight Mongoloid organized my work in Japan and allowed me to tendency which may be accidental. In the study collections stored in Sapporo. I wish to Hokkaido Ainu series, the Mongoloid admixture express my gratitude to Prof. Y. SHIBATA of the (possibly pre-Japanese) amounts to 1/7-1/5. Kyushu University, Prof. T. NISHIDA of the In Sakhalin Ainu, it is no lower than 1/3, the Kyoto University, Prof. T. AKAZAWA of the most likely source being Ulchi and Orochi. In the University Museum, the University of Tokyo, Ryukyuan series, the Mongoloid component is and Dr. B. YAMAGUCHI of the National Science predominant but does not exceed 3/5; southern Museum, Tokyo, for the permission to study Mongoloid ties of the Ryukyuans are not con- materials kept under their care. Finally, my firmed by our data. sincere thanks are due to persons without whose 4. In the Japanese and their ancestors generous and efficient help my work would be beginning from the Yayoi age, the main impossible, Dr. N. DOI, Dr. K. KATAYAMA, and Mongoloid element is very close to the one Dr. Y. MIZOGUCHI. I also thank Mr. B. represented in modern Chinese. It differs from KOZINTSEV who helped me to prepare the the northern Asiatic complex by a weaker expres- manuscript. The present study was financially sion of all Mongoloid traits which is probably due supported by a grant from the Japan Society for to early contacts of the Pacific Mongoloids with the Promotion of Science. Australoids on the continent. Proto-Japanese ties with Siberia, especially with the neolithic popula- 抄 録 tion of Baikal area, are also possible. 5. The abrupt change of the cranioscopic 頭骨 の観 察 項 目か らみ たア イ ヌ と日本 人 な ら び にそ の 祖 先 と近 隣 集 団 complex with the transition from Jomon to Yayoi was definitely caused by a large-scale immigra- Alexander KOZINTSEV tion from the continent. However, the further in- crease of the Mongoloid component was slow: 筆 者 は 日本 学術 振 興会 の招 きで,1989年 の3月 か ら West Japan Yayoi, 58-59%, West Japan Kofun, 5月 ま で,日 本 国 内 の大 学 と博 物館 で骨 人 類 学 的研 究 67-68%, modern Japanese (except Ryukyuans), を行 な う機 会 に恵 まれ た.調 査 資料 は,札 幌医 科 大 学, 70-82%. The presence of the Jomon anthropo- 九 州 大 学 医 学 部,京 都 大 学 理 学 部,東 京 大 学 総 合研 究 logical substratum is established with certainty in 資 料館 お よび 国立 科 学 博 物館 に保 管 さ れ て い る.東 日 all Japanese groups. This can be explained by a 本 お よび西 日本 の縄 文 人,土 井 ヶ浜 を中心 と した西 日 shift of most Jomon populations of West Japan 本 の渡 来系 弥 生 人,西 日本 古 墳 人,西 日本 現代 人,東 to agriculture and their mixture with the immi- 京在 住 の江 戸 な らび に現 代 人,南 西 諸 島人,北 海 道 ア grants during the Yayoi period. In modern in- イ ヌ,サ ハ リ ンア イ ヌ お よび宗 谷 大 岬 出土 の オ ホー ッ 266 A. KOZINTSEV

ク文化 人 の頭 骨 標 本 で あ る. DEBETS, G.F., 1968: An attempt for craniometric 9項 目の 頭骨 非計 測 的形 質 を調査 したが,そ の うち, decision of Mongoloid component in various ethnic groups in the USSR. In: Problems of Anthropology 日本 の諸集 団 の系 統 関係 を明 らか にす るの に特 に有効 and Historical Ethnography in Asia. Nauka, Moscow, な項 目 は,横 頬 骨 縫 合痕 跡(TZST),眼 窩 下縫 合 第2型 pp. 13-22. (In Russian) (IOP II)お よび眼 窩 上 孔(SOF)の3項 目で あ っ た.こ DODO, Y., 1986: Metrical and non-metrical analyses れ ら3項 目の高 い集 団 間 相 関 を考 慮 す る と,日 本 列 島 of Jomon crania from Eastern Japan. In: AKAZAWA,T, and C. M. AIKENS, eds., Prehistoric の 諸集 団 は二 つ の要 素 一縄 文 的 要 素 とモ ン ゴ ロイ ド的 Hunter-Gatherers in Japan. Univ. Mus., Univ. of 要 素-の 混 合 の度 合 い に よ って 特徴 づ け られ て い る こ Tokyo, Bull., 27, Univ, of Tokyo Press, Tokyo, pp. とが強 く示 唆 さ れ た. 137-161. そ こ で,こ の3項 目 の非 計 測 的 形 質 の 出現 頻 度 に基 DODO, Y., 1987: Supraorbital foramen and hypoglossal づ い て,各 集 団 に対 して モ ン ゴロ イ ド ・縄 文示 数(MJI) canal bridging: the two most suggestive nonmetric cranial traits in discriminating major racial groupings を算 出 したが,そ の結 果 か ら次 の よ うな事 が 明 らか に of man. J. Anthrop. Soc. Nippon, 95: 19-35. な った. DODO, Y., 1988: Population history in Japan as viewed 1.西 日本縄 文人 → 東 日本 縄 文 人→ 北 海 道 ア イ ヌ→ サ from cranial nonmetric variants. J. Anthrop. Soc. ハ リン アイ ヌ→ 南 西 諸 島人 → 弥生 人 → 古墳 人 → 西 日 Nippon, 96: 197. DOI, N, and Y. TANAKA, 1987: A geographical cline 本 現 代 人 →東 京 現 代 人 の順 にモ ン ゴロ イ ド的要 素 が in metrical characteristics of Kofun skulls from 増 大 す る. western Japan. J. Anthrop. Soc. Nippon, 95: 2.弥 生 時 代 以 降 の 日本人 のモ ン ゴロ イ ド的 要素 は現 325-343. DYAKOVA, O.B., 1982: The Mo-ho component in the 代 中 国人 の そ れ に きわ め て近 い. Okhotsk culture of Sakhalin and northern Hokkaido. 3.モ ン ゴ ロイ ド的要 素 の流 入 は弥 生 時代 に突 然 に始 In: Problems of Siberian Archaeology and まるが,そ れ は明 らか に大 陸 か らの 大量 の渡 来 に起 Ethnography. Irkutsk, pp. 115-117. (In Russian) 因 す る. GOKHMAN, 1.1. and A.M. RESHETOV, 1981: On the 4.弥 生 時 代 以 降現 代 まで のモ ン ゴロ イ ド的 要素 の増 southern borders of the prehistoric northern Asiatic Mongoloids' distribution. Soviet Ethnography, 6: 加 は僅 か で あ る. 78-88. (In Russian) 5.現 代 日本 人 に は明 らか に縄 文 的要 素 の痕 跡 が 認 め HANIHARA, K., 1983: Ainu and Ryukyus (Okinawa られ,し か もそ れ は西 日本 現代 人 に よ り明瞭 であ る. Islanders). In: KONDO, S., K. HANIHARA, J. IKEDA 6.北 海 道 の オ ホ ー ツ ク 文化 人 は,あ る 特 徴 で は ア and N. WATANABE, eds., Recent Progress of Natural Sciences in Japan, 8. Science Council of ム ール 河流 域 の ツ ング ース ・満 州 系 と近 いが,別 な特 Japan, Tokyo, pp. 25-30. 徴 で は ア イ ヌ に類 似 す る. HANIHARA, K., 1985: Origins and affinities of Japanese (百々幸 雄 抄 訳) as viewed from cranial measurements. In: KIRK, R. and E. SZATHMARY,eds., Out of Asia: Peopling the References Americas and the Pacific. Journal of Pacific History AIKENS, CM. and T. HIGUCHI, 1982: of Press, Canberra, pp. 105-112. Japan. Academic Press, New York, pp. 1-357. HANIHARA, K., 1987: Estimation of the number of AKAZAWA, T., 1986: Hunter-gatherer adaptations and early migrants to Japan: a simulative study. J. the transition to food production in Japan . In: Anthrop. Soc. Nippon, 95: 391-403. ZVELEVIL, M. ed., Hunters in Transition. ISHIDA, H., 1988: Morphological studies of Okhotsk Cambridge University Press, Cambridge, pp. 151-165. crania from Omisaki, Hokkaido. J. Anthrop. Soc. BALZ, E., 1911: Die Riu-Kiu-Insulaner, die Aino and Nippon, 96: 17-45. andere Kaukasierahnliche Reste in Ostasieri. KANASEKI, T., 1966: People of the Yayoi period. In: Korrespondenz-Blatt der Deutschen Gesellschaft fur WAJIMA, S. ed., Archaeology of Japan, 3. Kawade Anthropologie, Ethnologie and Urgeschichte, 42 , Tokyo, pp. 460-471. (In Japanese)

(8-12): 187-191. [金 関丈夫,1966:弥 生 時 代 人.日 本 の 考 古 学 III 弥 BRACE, CL., ML. BRACE and W.R. LEONARD, 1989: 生 時 代.河 出 書 房,東 京,pp.460-471.] Reflections on the face of Japan: a multivariate KOUCHI, M., 1983: Geographic variation in modern craniofacial and odontometric perspective. Am. J . Japanese somatometric data and its interpretation . Phys. Anthropol., 78: 93-113 . Univ. Mus., Univ, of Tokyo, Bull ., 22. Univ. of Cranioscopy of Ainu, Japanese, Their Ancestors and Neighbours 267

Tokyo Press, Tokyo, pp. 1-102. SMOLYAK, A.V., 1975: On the cultural exchange and KOYAMA, S., 1919: Jomon subsistence and population. ethnic history of the peoples of Sakhalin. In: Origins Senri Ethnological Studies, National Museum of and Ethnic History of the Northern Peoples. Nauka, Ethnology, 2: 1-65. Moscow, pp. 43-77. (In Russian) KOZINTSEV,A.G., 1975: Metopism and its distribution SNEDECOR,G.W. and W.G. COCHRAN, 1980: Statisti- in various human groups. Races and Peoples, 5: 55-68. cal Methods. 7th ed. The Iowa State Univ. Press, (In Russian) Iowa, pp. 252-253. KOZINTSEV, A.G., 1980: Transition to agriculture and SUZUKI, H., 1981: Racial history of the Japanese. In: human ecology. In: Early Farmers. Nauka, SCHWIDETZKY, I., ed., Rassengeschichte der Leningrad, pp. 6-33. (In Russian) Menschheit, 8. R. Oldenbourg, Munchen/Wien, pp. KOZINTSEV, A.G., 1988: Ethnic Cranioscopy: Racial 7-69. Variation of Cranial Sutures in Modern Man. Nauka, TAGAYA, A. and J. IKEDA, 1976: A multivariate Leningrad, pp. 1-167. (In Russian) analysis of the cranial measurements of the Ryukyu LEVIN, M.G., 1958: Ethnic Anthropology and Ethnic Islanders (males). J. Anthrop. Soc. Nippon, 84: History of the Far Eastern Peoples. Transactions of 204-220. the Institute of Ethnography, 36. Nauka, Moscow, TOMASHEVICH,TV., 1988: Regularities in the distribu- pp. 1-359. (In Russian) tion of supraorbital foramina in human populations. LEVIN, M.G., 1971: Ethnic Anthropology of Japan. Problems of Anthropology, 80; 119-128. (In Russian) Nauka, Moscow, pp. 1-235. (In Russian) WANG, L., 1987: Racial relationships between Chinese MATSUMOTO, H., 1987: Characteristics of the and Japanese: a statistical study of cranial Mongoloid and neighbouring populations on the basis measurements. Acta Anthropologica Sinica, 6: 10-18. of the genetic markers of immunoglobulins. J. (In Chinese with English summary) Anthrop. Soc. Nippon, 95: 291-304. (In Japanese with YAMAGUCHI, B., 1973: Facial flatness measurements English summary) of Ainu and Japanese crania. Bull. National Science MIZOGUCHI, Y., 1986: Contributions of prehistoric Museum, 16: 161-171. Far East populations to the population of modern YAMAGUCHI, B., 1975: Cranial features of the Okhotsk Japan: a Q-mode path analysis based on cranial culture people. In: KATO, S., K. OTUKA, K. measurements. In: AKAZAWA, T. and CM. SAKURAI and B. YAMAGUCHI, eds., Sea Mammal AIKENS, eds., Prehistoric Hunter-Gatherers in Hunters: Origin of the Okhotsk Culture. Dorumen, Japan. Univ. Mus., Univ. of Tokyo, Bull., 27. Univ. 6: 47-90. (In Japanese) of Tokyo Press, Tokyo, pp. 107-136. [山 口 敏,1975:オ ホ ー ツ ク 人 の 顔 つ き.加 藤 晋 平 MIZOGUCHI, Y., 1988: Affinities of the protohistoric ・大 塚 和 義 ・桜 井 清 彦 ・山 口 敏:海 獣 狩 猟 民:オ Kofun people of Japan with pre- and protohistoric ホ ー ツ ク 文 化 の 源 流.ど る め ん,6:47-90.] Asian populations. J. Anthrop. Soc. Nippon, 96: 71-109. YAMAGUCHI, B., 1982: A review of the osteological MOURI, T., 1988: Incidences of cranial nonmetric characteristics of the Jomon population in prehistoric characters in five Jomon populations from West Japan. J. Anthrop. Soc. Nippon, 90, Supplement: Japan. J. Anthrop. Soc. Nippon, 96: 319-337. 77-90. OSSENBERG, N.S., 1986: Isolate conservatism and YAMAGUCHI, B., 1988: The facial flatness measure- hybridization in the population : the ments of the protohistoric skeletal remains from evidence of nonmetric cranial traits. In: AKAZAWA, eastern Japan. In: YOKOYAMA, K., ed., Origins of T, and CM. AIKENS, eds., Prehistoric Hunter- the Japanese and Their Culture. Professor Masafumi Gatherers in Japan. Univ. Mus., Univ, of Tokyo, Nagai Memorial Volume, 1. Rokko Shuppan, Tokyo, Bull., 27. Univ, of Tokyo Press, Tokyo, pp. 199-215. pp. 35-46. (In Japanese) SIEGEL, S., 1956: Nonparametric Statistics for the [山口 敏,1988:東 日本 の古 墳・横 穴 墓 出 土 人骨 の Behavioral Sciences. McGraw Hill Book Co., New 顔 面 平 坦 度 計 測.日 本 民 族 ・文 化 の生 成1永 井 昌 文 York, pp. 166-173. 教 授 退 官 記 念 論 文 集.六 興 出版,東 京,pp.35-46.]

Alexander KOZINTSEV Institute of Ethnography Universitetskaya nab., 3, Leningrad-34, 199034, USSR