23© 2011 The Anthropological Society of Nippon

ANTHROPOLOGICAL SCIENCE

Vol. 120(1), 23–32, 2012

Facial characteristics of the prehistoric and early-modern inhabitants of the Okinawa islands in comparison to the contemporary people

of Honshu

Hitoshi FUKASE1*, Tetsuaki WAKEBE2, Toshiyuki TSURUMOTO

2, Kazunobu SAIKI2, Masaki FUJITA3, Hajime ISHIDA1

1Department of Human Biology and Anatomy, Faculty of Medicine, University of the Ryukyus, Nishihara 903-0215, Japan2Department of Developmental and Reconstructive Medicine, Course of Medical and Dental Sciences,

Graduate School of Biomedical Sciences, Nagasaki University, Nagasaki 852-8523, Japan3Okinawa Prefectural Museum and Art Museum, Naha 900-0006, Japan

Received 11 April 2011; accepted 26 April 2011

Abstract Modern inhabitants of the Okinawa islands have been supposed to represent a relativelyclose genealogical connection to the prehistoric Jomon and the indigenous Ainu, when compared withpeople living on Japan’s main island (Honshu). However, several previous studies have also clarifiedsome morphological facial traits discernible between the early-modern Okinawans and the Jomon ofHonshu. In the present study, to further evaluate the Jomon–Okinawan relationship, we comparedfacial forms between skeletal samples from the Okinawa islands and Honshu, both in the Jomon andearly-modern periods. Results of the comparisons showed that the Okinawa Jomon tended to possessa flatter interorbital region than the Honshu Jomon, but most measurements did not significantly differbetween the two Jomon groups. This confirms that people sharing both Jomon culture and fundamentalfacial features lived throughout almost the entire Japanese archipelago, including the Okinawa islands.Results also demonstrated that the early-modern Okinawans had a significantly lower and broader facewith transversely broader and flatter interorbital region than the early-modern Honshu Japanese. Suchfacial characteristics of the early-modern Okinawans qualitatively matched those of the OkinawaJomon. However, compared with the Okinawa Jomon, the early-modern Okinawans exhibited an ab-solutely taller/narrower face and a further flatter nasal root. This could stem in part from a certainamount of population flow from surrounding regions during historic periods, as suggested by manyearlier studies. Nevertheless, the results obtained in this study indicate that, comparatively, modernOkinawan people have retained physical traits including facial forms of the Jomon who survived inthe same southernmost islands of the Japanese archipelago. Geographical variations of modernJapanese phenotypes should also be accounted for partly by those of the ancestral Jomon characteris-tics that were manifested by the end of the Jomon period.

Key words: Jomon, facial flatness, Ryukyu, Japanese population history

Introduction

A number of studies have investigated geographical vari-ations in physical and genetic traits of the modern Japanese,and have clarified some attributes showing a geographicalcline that peripherally spreads from the western part ofJapan’s main island (Honshu) across the Japanese archipela-go (e.g. Ueda, 1935; Tagaya and Ikeda, 1976; Kanda, 1978;Ikeda and Tagaya, 1980; Watanabe et al., 2004; Hammer etal., 2006; Yamaguchi-Kabata et al., 2008; Matsukusa et al.,

2010). This is interpreted, in association with the populationhistory of the Japanese, as a reflection of regional differencesin the levels of admixture between hunter-gatherer-fishersin the Jomon period (c. 13000–2350 BP) and subsequentimmigrants thought to have come from the Asian continentin the Yayoi period (c. 2350–1650 BP). A dominant geneticinfluence of the Yayoi immigrants on the modern people liv-ing in Honshu is generally supposed (e.g. Dodo and Ishida,1990; Hanihara, 1991, 1993; Nakahashi, 1993; Omoto andSaitou, 1997; Pietrusewsky, 1999). In contrast, a lesser ex-tent of admixture has been presumed for the native inhabit-ants of the northernmost and southernmost regions of theJapanese archipelago, Hokkaido and the Ryukyu islandchain, respectively (Figure 1). Specifically, the indigenousAinu people in Hokkaido are currently inferred as being thephylogenetically closest to the Jomon people, from the bulkof anthropological evidence (e.g. Brace et al., 1989; Dodo

* Correspondence to: Hitoshi Fukase, Department of Human Biol-ogy and Anatomy, Faculty of Medicine, University of the Ryukyus,Uehara 207, Nishihara, Okinawa 903-0215, Japan. E-mail: fukase@med.u-ryukyu.ac.jp

Published online 25 June 2011 in J-STAGE (www.jstage.jst.go.jp) DOI: 10.1537/ase.110411

24 H. FUKASE ET AL. ANTHROPOLOGICAL SCIENCE

and Ishida, 1990; Turner, 1990; Hanihara, 1991; Ossenberget al., 2006).

It has also been suggested that the modern inhabitants ofthe Okinawa islands, the central islands of the Ryukyu islandchain, have a closer relationship with the Ainu and theJomon than the people of Honshu do (see reviews inHanihara, 1991; Fukumine et al., 2006; Matsukusa et al.,2010). Indeed, both the modern inhabitants of the Okinawaislands (hereinafter referred to as the modern ‘Okinawans’)and the Ainu are well known to frequently exhibit commonphenotypic characteristics such as a low and broad face,large eye openings with double eyelids, wet earwax, and anoverall hairy body (e.g. Ikeda, 1978; Hanihara, 1991;Yoshiura et al., 2006). The results of archeological studieshave demonstrated that the hunting-fishing-gathering sub-sistence strategy continued until the 12th century AD in theOkinawa islands (Okinawa Prefectural Culture PromotionFoundation, 2003), unlike in Honshu where a substantialtransition to rice agriculture occurred between the Jomonand Yayoi periods (Imamura, 1996). This indicates that geneflow derived from the Yayoi immigrants was muted in theOkinawa islands until the beginning of the medieval Gusukuperiod associated with the Ryukyu kingdom (Matsushita andOhta, 1993; Asato, 1996; Doi, 2003).

However, both morphological and genetic findings havealso emphasized differential traits of the Ryukyuans (includ-ing the Okinawans) from the Jomon/Ainu cluster, cautioningagainst a simple model that regards the modern Ryukyuansas direct descendants of the prehistoric Jomon, and suggest-ing considerable population influx from surrounding regionsinto the Ryukyu area during medieval times (Dodo et al.,1998, 2000; Asato and Doi, 1999; Hatta et al., 1999;Pietrusewsky, 1999, 2004, 2010; Higa et al., 2003;Fukumine et al., 2006; Manabe et al., 2008, 2011). Dodo etal. (2000) compared facial skeletons among many Japanese

groups from different periods and regions, and found that theearly-modern Okinawans did not exhibit some facial charac-teristics that are manifested in both Jomon and Ainu (e.g. ananteriorly prominent glabella and a strongly elevated nasalridge). The early-modern Okinawans indeed had a trans-versely flatter nasal root than even the modern Honshu-Japanese (Dodo et al., 2000).

The circumorbital morphology varies considerablyamong worldwide major human groups, and has also beenstudied to evaluate intergroup kinships (Woo and Morant,1934; Gill et al., 1988; Ishida, 1992; Ishida and Dodo, 1997;Hanihara, 2000). Recently, Smith (2009) compared the mor-phological and molecular genetic data of many modern hu-man groups, and demonstrated that the three-dimensionalcircumorbital form, as well as the cranial basal shape, hassignificant correlation with the genetic intergroup distances.Since the Yayoi immigrants possessed a remarkably ‘flat’face in contrast to the Jomon, temporal and geographicalvariations in the Japanese facial flatness have often been in-terpreted in association with levels of admixture between thetwo ancestral groups (e.g. Nakahashi and Iizuka, 1998;Kawakubo, 2007; Okazaki and Nakahashi, 2011). In thiscontext, the flat face of the early-modern Okinawans may beregarded as indicating genealogical remoteness to bothJomon and Ainu groups (cf. Dodo et al., 2000).

However, most previous studies that assessed the Jomon–Okinawan relationship employed Jomon samples excavatedfrom Honshu, and only a limited number of studies haveinvestigated the conditions of the Jomon from Okinawa(Matsushita and Ohta, 1993; Oyamada et al., 1995; Wakebeet al., 1996; Doi, 2003). It is therefore still unclear whetherthe southern Okinawa-Jomon shared facial features with theHonshu-Jomon, although the craniofacial form of the north-ern Hokkaido-Jomon is reported to generally fall within theHonshu-Jomon range of variation (Yamaguchi, 1981;Matsumura et al., 2001). Wakebe et al. (1996) examinedfacial forms of the prehistoric Okinawan individuals datedbetween the Late Jomon and Early Yayoi periods (c. 4000–2250 BP), and pointed out their significantly flat face incomparison to the Jomon of Honshu. In light of this, thetransversely flat face of the early-modern Okinawans(Matsushita et al., 1988; Dodo et al., 2000; Fukumine et al.,2001) may largely result from retention of a Jomon trait inOkinawa, rather than substantial reformation of the popu-lation.

Thus, facial forms of the prehistoric inhabitants of theOkinawa islands can be clues both to explore the southernlimit of the Jomon people with morphologically ‘homolo-gous’ faces in the Japanese archipelago (cf. Matsumura etal., 2001; Mizoguchi, 2010) and to re-evaluate the Jomon–Okinawan connection (cf. Dodo et al., 2000). In order to as-sess these issues, we quantitatively compared facial skele-tons from the Okinawa islands and Honshu in both Jomonand early-modern periods, with particular attention to theinterorbital form.

Materials and Methods

The skeletal specimens used in the present study are listedin Table 1. Adult male samples without apparent skeletal

Figure 1. Map of the Japanese archipelago.

FACIAL FEATURES OF THE JOMON PEOPLE IN OKINAWA 25Vol. 120, 2012

deformation and pathological traits were selected. The sex ofthe Jomon and early-modern Okinawan samples was evalu-ated from pelvic morphology, and from appendicular mor-phology in the cases that lacked sufficiently preservedassociated pelvic remains.

In the present study, to match the names of the prehistoricperiods between Honshu and Okinawa, we provisionallyemployed the names of ‘Jomon’ and ‘Yayoi,’ instead of‘Kaizuka,’ which is occasionally used specifically to describethe Okinawan prehistoric age (Okinawa Prefectural CulturePromotion Foundation, 2003). Okinawan prehistoric sam-ples spanning from the Late to Final Jomon period (c. 4000–2350 BP) were chosen for analysis. However, the specimensderived from the Mashiki-Azamabaru site are dated from theFinal Jomon to Early Yayoi period (c. 3000–2250 BP).Therefore, we refer to the used Okinawan collection as the‘Okinawa Jomon/Yayoi’ in this study. The Jomon materialsfrom Honshu, referred to as the ‘Honshu Jomon,’ span fromthe Middle to Final Jomon period (c. 5000–2350 BP).

The ‘early-modern’ Okinawan samples are thought tospan a chronological age of approximately 1700–1900 AD.The major part of this collection consists of individuals fromKume Island (the second largest island in the Okinawa is-lands), but earlier studies have shown little significant differ-ences in craniofacial and nonparametric characteristicsbetween individuals from Kume Island and Okinawa Island(Suda, 1940, 1950; Dodo et al., 1998; Pietrusewsky, 1999;Fukumine et al., 2001, 2006). The materials of the early-modern Honshu Japanese come from anatomy departmentcollections made by the University of Tokyo from approxi-mately 1890 to 1910 AD. The age at death of the HonshuJapanese specimens ranged from 20 to 40 years (mean =

28.0, SD = 5.9), according to archival records.Linear measurements were taken with a digital caliper,

principally following Martin’s definitions and methods(Knussmann, 1988; Baba, 1991), as illustrated in Figure 2aand b. Facial height index was defined as the ratio of upperfacial height (M48) to biorbital breadth (M43). The ratio ofinterorbital breadth (M49a) to the biorbital breadth was alsoderived. Nasal height was determined as the distance fromthe nasion (n) to the subspinale (ss) instead of the nasospina-le (ns). The nasal height index was defined as the ratio of thenasal height to nasal breadth (M54). The simotic chord(M57) was the minimum horizontal breadth of the nasalia.The simotic subtense was determined as the minimum dis-tance from the median crest of the nasal bones to the simoticchord, calculated using trigonometric functions (e.g.

Yamaguchi, 1973). The simotic index was derived as the ra-tio of the simotic subtense to the chord.

Circumorbital forms in profile were investigated by use ofthe three-dimensional digitizer Microscribe G2X (Immer-sion Co.) and the coordinate analysis software 3D-GADES(developed by Shinya Matsukawa). Essentially referring toMartin’s definitions (Knussmann, 1988; Baba, 1991), thefollowing anatomical landmarks were taken: glabella (g),nasion (n), subnasion (sbn), rhinion (rhi), subspinale (ss),staphylion (sta), maxillonasofrontale (mnf), and the mostlateral point of the nasal aperture. The subnasion and rhinionwere defined as the most strongly curved point and the mostanterior point on the crest of the nasal bones, respectively(see Figure 2c). The coordinates of all the landmarks wereprojected onto the midsagittal plane, which was defined aspassing through the nasion, subspinale, and staphylion. Afacial reference line was then set to pass the mnf and themost lateral point of the nasal aperture (the two landmarkswere each averaged between right and left sides). Promi-nence distances were defined as the minimum distancesfrom the reference line to glabellar and nasal landmarks onthe midsagittal plane (Figure 2c). Specifically, to evaluatethe relative prominence of the glabella, the glabella indexwas determined as the ratio between the prominence dis-tance of glabella and the interorbital breadth. For assessingthe shape of the profile, we also calculated angles betweenthe reference line and n–rhi [Angle (1) in Figure 2d], the ref-erence line and sbn–rhi [Angle (2)], g–n and sbn–rhi [Angle(3)], and n–sbn and sbn–rhi [Angle (4)].

Two-tailed Mann-Whitney’s U-test was conducted forcomparisons between the materials from Okinawa andHonshu in both Jomon and early-modern periods. In addition,to summarize the measured dataset and present an overviewof individual positions within the total variation, principalcomponent analysis (PCA) was performed using a total of15 variables of the linear and angular measurements (seeTable 2). Individuals with more than six missing values wereexcluded for the PCA, but missing values were, if present forthe employed specimens, complemented with values esti-mated by stepwise multiple regression using available vari-ables. To include the linear and angular variables in the sameanalysis, the PCA was conducted with a correlation matrixderived from standardized (Z-score) values. Although PCAassumes variables with normal distribution, we chose to usethe 15 variables including three variables (the simotic sub-tense, the prominence distance of nasion, and the angle be-tween g–n and sbn–rhi) whose normality was rejected by the

Table 1. Materials

Sample name n Sex Date Site Location

Okinawa Jomon/Yayoi 14 Male c. 4000–2250 BP Bugeido, Furuzamami, Gushikawajima, Mashiki-Azamabaru

NU, OPAC, OPMAM

Honshu Jomon 20 Male c. 5000–2350 BP Hobi, Kasori, Matsudo, Nakazawahama, Ohora, Sanganji, Tsugumo, Ubayama, Yosekura

UMUT

Early-modern Okinawans 30 Male c. 1700–1900 AD Kujichi, Yacchi-no-gama NU, OPACEarly-modern Honshu Japanese 28 Male Collection made

c. 1900 AD(The Kanto district) UMUT

n, sample size. NU, Nagasaki University; OPAC, Okinawa Prefectural Archaeological Center; OPMAM, Okinawa Prefectural Museum and ArtMuseum; UMUT, The University Museum, The University of Tokyo.

26 H. FUKASE ET AL. ANTHROPOLOGICAL SCIENCE

Kolmogorov–Smirnov test (P < 0.05), in order to obtain anoverall trend of facial forms from the all linear and angularmeasurements. This should not substantially affect results ofthe PCA, considering the robustness of general linear mod-els. These statistical analyses were carried out using the soft-ware SPSS 10 (SPSS Inc.). Statistical significance ofP < 0.05 is referred to in describing the results.

Results

Descriptive statistics of the measurements are listed inTable 2. Results of the indices examined are shown as boxplots in Figure 3. Comparisons of the prehistoric samplesshowed that the Okinawa Jomon/Yayoi and the HonshuJomon were not significantly different in most cases. Thismay be due in part to the small sample size, but suggests

their overall facial resemblance (see Figure 4). However, theinterorbital breadth, including the simotic chord, tended tobe both absolutely and relatively larger in the OkinawaJomon/Yayoi, whereas the simotic subtense was converselygreater in the Honshu Jomon. This resulted in the signifi-cantly greater simotic index in the Honshu Jomon. Theprominence distances of the glabellar and nasal landmarkswere generally shorter in the Okinawa Jomon/Yayoi (seeFigure 5). In fact, the glabella index value was significantlysmaller in the Okinawa Jomon/Yayoi, indicating their weakglabellar prominence.

Comparisons of the early-modern people from Okinawaand Honshu showed that the Okinawans had a relativelylow/broad face compared to the Honshu Japanese (Figure 4),as shown by the significantly lower values of the facial andnasal height indices in the Okinawans (Figure 3). This

Figure 2. Measurements used in this study. Facial dimensions were measured with a digital caliper (a, b). Facial forms in profile were exam-ined (c, d) using the coordinates of anatomical landmarks taken with a three-dimensional digitizer. The reference line was defined as a line passingthrough the mnf and the most lateral point of the nasal aperture. Prominence distances were defined as the minimum distances from the referenceline to glabellar and nasal landmarks (c). Some angles were also determined (d); Angle (1) is the angle between the reference line and n–rhi; Angle(2), the reference line and sbn–rhi; Angle (3), g–n and sbn–rhi; Angle (4), n–sbn and sbn–rhi.

FACIAL FEATURES OF THE JOMON PEOPLE IN OKINAWA 27Vol. 120, 2012

Table 2. Descriptive statistics of measurements

Jomon (Jomon/Yayoi for Okinawa) Eary-modern

Okinawa Honshu Okinawa Honshu

n Mean SD n Mean SD n Mean SD n Mean SD

Upper facial height (n–av, M48)a 10 67.3 4.2 17 69.1 4.6 23 68.2 3.3 27 70.7* 4.2Biorbital breadth (fmt–fmt, M43)a 14 108.6 4.7 18 109.6 4.9 30 105.2* 3.9 28 103.5 3.7Interorbital breadth (d–d, M49a)a 8 23.9 2.2 18 23.5 2.4 29 22.3** 2.4 28 20.2 2.3Nasal height (n–ss)a 9 52.9 3.4 18 52.6 4.9 28 54.0 2.9 28 55.6 3.3Nasal breadth (M54)a 10 26.7 1.5 18 27.0 2.0 28 26.2* 1.4 28 25.1 1.9Simotic chord (Minimum nasal breadth, M57)a 8 10.1 1.5 18 9.9 1.6 29 8.7** 1.5 27 7.1 1.8Simotic subtensea 8 3.4 1.3 18 4.4* 1.0 26 2.5 0.9 27 2.5 0.8Prominence distance of glabellaa 10 6.5 1.8 17 8.1 1.9 29 5.3 1.4 27 5.3 1.3

nasiona 10 3.4 1.4 18 3.8 0.6 28 2.3 0.8 27 2.1 0.7subnasiona 8 5.1 1.6 19 5.5 1.1 27 4.0 1.1 27 3.4 1.2rhiniona 6 12.2 2.6 17 12.7 2.3 28 10.7 1.8 27 11.9 2.4

Angle between the reference line and n–rhia 6 27.8 5.4 17 29.1 4.8 28 23.2 4.3 27 24.9 4.8the reference line and sbn–rhia 6 40.7 8.5 17 38.5 5.5 28 40.3* 6.2 27 36.2 5.3g–n and sbn–rhia 6 121.1 10.2 17 120.1 9.7 28 121.7 9.0 27 126.9* 6.8n–sbn–rhia 6 151.9 5.2 17 155.3 6.9 28 147.8 5.8 27 151.4* 4.9

Facial height index (upper facial height/bior-bital breadth)

10 0.63 0.03 16 0.63 0.04 23 0.64 0.03 27 0.68** 0.04

Nasal height index (nasal height/breadth) 9 2.01 0.13 17 1.96 0.15 27 2.07 0.13 28 2.23** 0.21The ratio of interorbital to biorbital breadth 8 0.22 0.01 18 0.21 0.02 29 0.21** 0.02 28 0.19 0.02The ratio of simotic chord to interorbital

breadth5 0.44 0.06 17 0.42 0.06 28 0.39* 0.07 27 0.35 0.07

Simotic index (simotic subtense/chord) 8 0.33 0.10 18 0.45** 0.10 26 0.28 0.08 27 0.35* 0.10Glabella index (Prominence distance to

glabella/interorbital breadth)6 0.27 0.05 16 0.35* 0.08 28 0.25 0.08 27 0.27 0.08

n, sample sizes. A value with one or two asterisks on the right shoulder is significantly larger than that of the other area. Significance level:*P < 0.05; **P < 0.01.

a Measurements used for principal component analysis (PCA).

Figure 3. Index measure-ments. Upper and lower mar-gins of box plots represent the75th and 25th percentiles,respectively. The median is theline bisecting the box. Asterisksrepresent significance levels:*P < 0.05; **P < 0.01.

28 H. FUKASE ET AL. ANTHROPOLOGICAL SCIENCE

stemmed mainly from a difference in the facial height ratherthan the facial breadth. Furthermore, it was found that the in-terorbital breadth was both absolutely and relatively greaterin the early-modern Okinawans. Associated with this hori-zontally broad interorbital region, their simotic index valuewas significantly less than that of the Honshu Japanese.Analyses of the profile showed that the prominence distancesdid not differ between the two groups in all the cases. Theangle between the reference line and sbn–rhi was larger inthe Okinawans, while the angles of both n–sbn–rhi and g–n/sbn–rhi were conversely greater in the Honshu Japanese.These results suggested that the crest of the nasal bones wassagittally more curved in the early-modern Okinawans, andtherefore their infraglabellar recess appeared relatively deep(Figure 5).

Results of the PCA are listed in Table 3 and displayed inFigure 6. The first four principal components representedgreater than 1.0 of eigenvalue, and in total contributed73.9% to the total variation. The first principal component(34.4% of the contribution) is chiefly related to the facialbreadths and the elevation of the nasal crest, leading togreater PC1 scores in the two Jomon groups (Figure 6A).Both second and third principal components (17.0% and14.1%, respectively) are associated with the facial heightsand the angles of the nasal root region. The early-modernHonshu Japanese had a comparatively high face with shal-low and open infraglabellar notch, resulting in their muchgreater scores for PC2. The fourth principal component(8.4%) appears related to the facial breadth and the flatnessof the glabellar region.

Figure 4. Frontal views of facial skeletons.

FACIAL FEATURES OF THE JOMON PEOPLE IN OKINAWA 29Vol. 120, 2012

Discussion

The facial comparisons between the Okinawa Jomon/Yayoi and the Honshu Jomon did not exhibit significant dif-ferences in most measurements, indicating general similarityof their facial frameworks (Figure 4). This is also illustratedby the results of the PCA (Figure 6), and is consistent withprevious descriptive remarks on the Jomon people of theOkinawa islands (Matsushita and Ohta, 1993; Doi, 2003,2004). The results therefore support that people who sharedboth Jomon culture and basic facial features did live

throughout the Japanese archipelago including the Ryukyuregion. However, compared with the Honshu Jomon, theOkinawa Jomon/Yayoi tended to show greater and lesservalues in the simotic chord and subtense, respectively, re-sulting in a significantly lower simotic index value. Further-more, the prominence of the glabellar and nasal root regionswas relatively weak in the Okinawa Jomon/Yayoi, as seen inthe profiles of Figure 5. These findings suggest that theJomon had an overall flatter face in Okinawa than in Honshu,which also accords with previous observations (Wakebe etal., 1996; Nagashima and Wakebe, 2003). Nevertheless, thisminor distinction would not refute the previous view that theextent of geographical diversity of the Jomon craniofacialform was modest relative to intra-group variations of otherEast Asian groups including the modern Japanese (Dodo,1982; Yamaguchi, 1982; Kondo, 1994; Matsumura et al.,2001; Matsumura, 2007; Hanihara and Ishida, 2009; Ishidaet al., 2009).

Suda (1940) undertook an anthropometric survey on thefaces of live inhabitants of Okinawa Prefecture, and assertedthat a low/broad facial proportion and a low nasal height arenoteworthy features of this population. This has been subse-quently corroborated by both somatometric and skeletalstudies (Miyake, 1940; Naito, 1976; Tagaya and Ikeda,1976; Doi, 2003). The results of the present study alsoshowed the significantly lower face of the early-modernOkinawans than that of the early-modern Honshu Japanese.With the facial flatness, the lower value of the simotic indexin the early-modern Okinawans was found to stem principallyfrom the broader interorbital region. Furthermore, the analy-sis of the profile showed the nasal bone of the early-modern

Figure 5. Lateral views of facial skeletons. The images are not exactly to scale. The rightmost image in the top row is a mirrored image.

Table 3. Principal component analysis of the facial measurements

PC1 PC2 PC3 PC4

Upper facial height 0.30 0.74 0.43 0.00Biorbital breadth 0.75 0.10 −0.13 0.30Interorbital breadth 0.54 −0.05 −0.39 0.61Nasal height 0.27 0.67 0.49 0.10Nasal breadth 0.44 0.16 −0.29 0.32Simotic chord 0.73 −0.23 −0.26 0.25Simotic subtense 0.86 −0.14 −0.03 −0.17Prominence distance of glabella 0.60 −0.26 −0.24 −0.41

nasion 0.75 −0.13 −0.31 −0.46subnasion 0.88 −0.06 0.09 −0.10rhinion 0.68 0.10 0.50 −0.29

Angle of the reference line and n–rhi 0.49 −0.27 0.66 0.13the reference line and sbn–rhi 0.07 −0.62 0.67 0.31g–n and sbn–rhi −0.16 0.71 −0.21 −0.03n–sbn–rhi 0.48 0.61 −0.12 0.04

Eigenvalue 5.2 2.5 2.1 1.3%VAR 34.4 17.0 14.1 8.4

30 H. FUKASE ET AL. ANTHROPOLOGICAL SCIENCE

Okinawans to be sagittally more curved, accompanying theforward projection of the lower nasal bone (see Figure 5).These results suggest that the Okinawan face should not nec-essarily appear ‘flat’ from a lateral view when comparedwith the Honshu Japanese. In addition, the remarkably lownasal height of the Okinawan living face (Suda, 1940) can beexplained by the combined effects of the overall low faceand the sagittally curved nasal bone, since facial and nasalheights in the living body are often measured using the sell-ion, the most receding midsagittal point on the nasal crest,instead of the skeletal landmark of the nasion (Miyake,1940; Knussmann, 1988; Baba, 1991).

Thus, facial forms appear to have considerably differedbetween the people of Okinawa and Honshu in the early-modern age. It should be noted, however, that the specimensexamined were early-modern ones, which is of importanceconsidering that faces further became relatively narrow and

high with increasing stature in both Okinawa and Honshuduring the last hundred years (Omori, 1976). This recentchange in the facial proportions can co-vary with the interor-bital flatness (Suzuki, 1969; Kouchi and Mochimaru, 2006;Kawakubo, 2007), and therefore it remains to be seenwhether the ‘present’ Okinawans still possess the same fa-cial characteristics as the individuals examined in this study.

The above comparisons in the two time periods indicatethat the Jomon and early-modern inhabitants of the Okinawaislands shared some facial traits: the low/broad face, thecomparatively broad interorbital region, and the concomi-tant small value of the simotic index (Figure 3). However,when compared with the Okinawa Jomon/Yayoi, the early-modern Okinawans tended to show a relatively high/narrowface with a further flat nasal root. This difference is qualita-tively comparable to the facial shift from the Jomon to themodern Japanese in Honshu, and may result from a south-ward population influx from the main islands of the Japa-nese archipelago during medieval times (e.g. Asato, 1996;Doi, 2003; Higa et al., 2003; Manabe et al., 2008, 2011).Nevertheless, the results indicate that the early-modernOkinawans had facial features seen in both Okinawa Jomon/Yayoi and early-modern Honshu Japanese, as plausiblyillustrated by the PCA result (Figure 6A). It is therefore po-tentially hypothesized that, even with a certain level of ad-mixture, the modern Okinawans should comparatively retainphysical traits including facial forms of the prehistoricJomon who survived in the same southernmost islands of theJapanese archipelago. In addition, phenotypic and geneticvariations seen among regional groups in Japan, perhaps in-cluding differences between the Okinawans and Ainu (cf.Dodo et al., 1998, 2000), should be explained not only bylevels of genetic retention but also partly by geographicalvariations of the ancestral Jomon features that had beenmanifested by the end of the Jomon period.

Acknowledgments

We would like to thank Drs. N. Doi and R. Kimura (Uni-versity of the Ryukyus) for helpful suggestions and warmencouragement throughout the various aspects of this re-search. Special thanks go to Dr. N. Seguchi (University ofMontana) for many inspiring suggestions at a very earlystage of this study, and to Dr. Y. Dodo (Tohoku University)for thoughtful comments that have greatly improved themanuscript. Thanks also go to the Physical AnthropologyLaboratory of the University Museum, The University ofTokyo, and the Okinawa Prefectural Archaeological Center,Okinawa, for access to the materials used in this study. Thisstudy was supported by a Grant-in-aid for Scientific Re-search (Nos. 22370087, 22870025) from the Japan Societyfor the Promotion of Science.

References

Asato S. (1996) Population history of modern Ryukyuans asviewed from archaeology. Journal of Geography, 105: 364–371.

Asato S. and Doi N. (1999) Where Did the Okinawan PeopleCome From? Borderink, Okinawa (in Japanese).

Baba H. (1991) Osteometry. In: Eto M. (ed.), Anthropology,

Figure 6. Principal components graph. The stars surrounded bylarge open symbols represent centroids for each group: box, the Oki-nawa Jomon/Yayoi; circle, the Honshu Jomon; triangle, the early-modern Okinawans; cross, the early-modern Honshu Japanese.

FACIAL FEATURES OF THE JOMON PEOPLE IN OKINAWA 31Vol. 120, 2012

Supplement 1. Yuzankaku, Tokyo (in Japanese).Brace C.L., Brace M.L., and Leonard W.R. (1989) Reflections on

the face of Japan: a multivariate craniofacial and odontomet-ric perspective. American Journal of Physical Anthropology,78: 93–113.

Dodo Y. (1982) A metrical analysis of Jomon crania from theTohoku district. Journal of the Anthropological Society ofNippon (Supplement), 90: 119–128 (in Japanese with Englishabstract).

Dodo Y. and Ishida H. (1990) Population history of Japan asviewed from cranial nonmetric variation. Journal of theAnthropological Society of Nippon, 98: 269–287.

Dodo Y., Doi N., and Kondo O. (1998) Ainu and Ryukyuan cra-nial nonmetric variation: evidence which disputes the Ainu-Ryukyu common origin theory. Anthropological Science,106: 99–120.

Dodo Y., Doi N., and Kondo O. (2000) Flatness of facial skeletonsof Ryukyuans. Anthropological Science, 108: 183–198.

Doi N. (2003) History of Okinawa viewed from human remains.In: Okinawa Prefectural Culture Promotion Foundation (ed.),Okinawa Prefecture History. Okinawa Prefectual Board ofEducation, Okinawa, pp. 573–610 (in Japanese).

Doi N. (2004) Prehistoric and Gusuku people in Okinawa asviewed from skeletal morphlogy. Anthropological Science,112: 291 (abstract).

Fukumine T., Doi N., Ishida H., Zukeran C., Sensui N., Saso A.,and Higa T. (2001) Human skeletal remains from the Yacchi-no-Gama and Kanjinbaru grave sites. In: Okinawa PrefecturalArchaeological Center (ed.), Yacchi-no-Gama and Kanjin-garu Grave Sites. Okinawa Prefectural Archaeological Cen-ter, Okinawa, pp. 345–385 (in Japanese).

Fukumine T., Hanihara T., Nishime A., and Ishida H. (2006) Non-metric cranial variation of early modern human skeletalremains from Kumejima, Okinawa and the peopling of theRyukyu Islands. Anthropological Science, 114: 141–151.

Gill G.W., Hughes S.S., Bennett S.M., and Gilbert B.M. (1988)Racial identification from the midfacial skeleton with specialreference to American Indians and whites. Journal of Foren-sic Sciences, 33: 92–99.

Hammer M.F., Karafet T.M., Park H., Omoto K., Harihara S.,Stoneking M., and Horai S. (2006) Dual origins of theJapanese: common ground for hunter-gatherer and farmer Ychromosomes. Journal of Human Genetics, 51: 47–58.

Hanihara K. (1991) Dual structure model for the population his-tory of the Japanese. Japan Review, 2: 1–33.

Hanihara K. (1993) The population history of the Japanese.Nippon Ronen Igakkai Zasshi, 30: 923–931 (in Japanese).

Hanihara T. (2000) Frontal and facial flatness of major humanpopulations. American Journal of Physical Anthropology,111: 105–134.

Hanihara T. and Ishida H. (2009) Regional differences in craniofa-cial diversity and the population history of Jomon Japan.American Journal of Physical Anthropology, 139: 311–322.

Hatta Y., Ohashi J., Imanishi T., Kamiyama H., Iha M.,Simabukuro T., Ogawa A., Tanaka H., Akaza T., Gojobori T.,Juji T., and Tokunaga K. (1999) HLA genes and haplotypesin Ryukyuans suggest recent gene flow to the OkinawaIslands. Human Biology, 71: 353–365.

Higa T., Hanihara T., Sunakawa H., and Ishida H. (2003) Dentalvariation of Ryukyu islanders: a comparative study amongRyukyu, Ainu, and other Asian populations. American Jour-nal of Human Biology, 15: 127–143.

Ikeda J. (1978) Anthropology 6: The Japanese II. Yuzankaku,Tokyo (in Japanese).

Ikeda J. and Tagaya A. (1980) Geographic variation of the anthro-pometric measurements in the Japanese islands. Anthropolog-ical Science, 88: 397–410.

Imamura K. (1996) Jomon and Yayoi: the transition to agriculturein Japanese prehistory. In: Harris D.R. (ed.), The Origins andSpread of Agriculture and Pastoralism in Eurasia. Smithso-

nian Institution Press, Washington DC, pp. 442–464.Ishida H. (1992) Flatness of facial skeletons in Siberian and other

circum-Pacific populations. Zeitschrift für Morphologie undAnthropologie, 79: 53–67.

Ishida H. and Dodo Y. (1997) Cranial variation in prehistorichuman skeletal remains from the Marianas. American Journalof Physical Anthropology, 104: 399–410.

Ishida H., Hanihara T., Kondo O., and Fukumine T. (2009) Cranio-metric divergence history of the Japanese populations.Anthropological Science, 117: 147–156.

Kanda S. (1978) Muscular and arterial systems of the Japanese. In:Ikeda J. (ed.), Anthropology 6: The Japanese II. Yuzankaku,Tokyo, pp. 13–61 (in Japanese).

Kawakubo Y. (2007) Geographical and temporal variation infacial flatness in the crania of eastern Japan. AnthropologicalScience, 115: 191–200.

Knussmann R. (1988) Anthropologie. G. Fischer Verlag, Stuttgart.Kondo O. (1994) The skulls of Ubayama shell-mounds II. An anal-

ysis of intra- and inter-regional variation of the Jomon popu-lation. Anthropological Science, 102: 59–74.

Kouchi M. and Mochimaru M. (2006) Inter-generation differencein the Japanese facial flatness. Anthropological Science (Jap-anese series), 114: 45–53 (in Japanese with English abstract).

Manabe Y., Kitagawa Y., Oyamada J., Igawa K., Kato K., KikuchiN., Maruo H., Kobayashi S., and Rokutanda A. (2008) Popu-lation history of the northern and central Nansei Islands(Ryukyu island arc) based on dental morphological varia-tions: gene flow from North Kyushu to Nansei Islands.Anthropological Science, 116: 49–65.

Manabe Y., Oyamada J., Kitagawa Y., Igawa K., Kato K.,Matsushita T., and Rokutanda A. (2011) Nonmetric dentalcharacteristics of the early modern population of OkinawaIsland in Nansei Islands, Japan. International Journal ofOsteoarchaeology, 21: 679–693.

Matsukusa H., Oota H., Haneji K., Toma T., Kawamura S., andIshida H. (2010) A genetic analysis of the Sakishima island-ers reveals no relationship with Taiwan aborigines but sharedancestry with Ainu and main-island Japanese. American Jour-nal of Physical Anthropology, 142: 211–223.

Matsumura H. (2007) Non-metric dental trait variation amonglocal sites and regional groups of the Neolithic Jomon period,Japan. Anthropological Science, 115: 25–33.

Matsumura H., Anezaki T., and Ishida H. (2001) A morphometricanalysis of Jomon skeletons from the Funadomari site onRebun Island, Hokkaido, Japan. Anthropological Science,109: 1–21.

Matsushita T. and Ohta J. (1993) Skeletal remains from Gushikawa-jima sites, Okinawa. In: Izena Village Board of Education(ed.), Gushikawajima sites. Izena Village Board of Education,Okinawa, pp. 215–244 (in Japanese).

Matsushita T., Saiki K., and Tsurumoto T. (1988) Early-modernskeletal remains excavated from the Kujichi grave in Ginozavillage, Okinawa. In: Ginoza Board of Education (ed.), Exca-vation Report of the Kujichi Grave and the Kujichi-baru Site.Ginoza Board of Education, Okinawa, pp. 107–140 (in Japa-nese).

Miyake M. (1940) Anthropometry of the Japanese: Anthropologi-cal and Prehistoric Courses. Yuzankaku, Tokyo, pp. 1–69 (inJapanese).

Mizoguchi Y. (2010) Difference in geographical variation patternsof cranial measurements between the Jomon and Kofun peri-ods of Japan. Bulletin of the National Museum of Natural Sci-ence, Series D, 36: 13–26.

Nagashima S. and Wakebe T. (2003) Morphological analysis ofthe Jomon-type Yayoi people. Science of Humanity, 42: 19–25 (in Japanese).

Naito Y. (1976) The Okinawa islanders as viewed from anthropo-lometric measurements. In: Kyugakkai-Rengo (ed.), Oki-nawa: Nature, Culture, and Society. Kobundo, Tokyo, pp. 77–86 (in Japanese).

32 H. FUKASE ET AL. ANTHROPOLOGICAL SCIENCE

Nakahashi T. (1993) Temporal craniometric changes from theJomon to the Modern period in western Japan. AmericanJournal of Physical Anthropology, 90: 409–425.

Nakahashi T. and Iizuka M. (1998) Anthropological study of thetransition from the Jomon to the Yayoi periods in northernKyushu using morphological and paleodemographical fea-tures. Anthropological Science (Japanese series), 106: 31–53(in Japanese with English abstract).

Okazaki K. and Nakahashi T. (2011) Developmental variation infacial forms in Japan. Anthropological Science, 119: 49–65.

Okinawa Prefectural Culture Promotion Foundation (2003) Oki-nawa Prefecture History, Archaeology. Okinawa PrefectualBoard of Education, Okinawa (in Japanese).

Omori A. (1976) Temporal changes of the Okinawan people. In:Kyugakkai-Rengo (ed.), Okinawa: Nature, Culture, and Soci-ety. Kobundo, Tokyo, pp. 137–143 (in Japanese).

Omoto K. and Saitou N. (1997) Genetic origins of the Japanese: apartial support for the dual structure hypothesis. AmericanJournal of Physical Anthropology, 102: 437–446.

Ossenberg N.S., Dodo Y., Maeda T., and Kawakubo Y. (2006)Ethnogenesis and craniofacial change in Japan from the per-spective of nonmetric traits. Anthropological Science, 114:99–115.

Oyamada J., Manabe Y., Kitagawa Y., Rokutanda A., andNagashima S. (1995) Tooth size of the protohistoric Kofunpeople in Southen Kyushu, Japan. Anthropological Science,103: 49–60.

Pietrusewsky M. (1999) A multivariate craniometric study of theinhabitants of the Ryukyu Islands and comparisons with cra-nial series from Japan, Asia, and the Pacific. AnthropologicalScience, 107: 255–281.

Pietrusewsky M. (2004) Multivariate comparisons of female cra-nial series from the Ryukyu Islands and Japan. Anthropologi-cal Science, 112: 199–211.

Pietrusewsky M. (2010) A multivariate analysis of measurementsrecorded in early and more modern crania from East Asia andSoutheast Asia. Quaternary International, 211: 42–54.

Smith H.F. (2009) Which cranial regions reflect molecular dis-tances reliably in humans? Evidence from three-dimensionalmorphology. American Journal of Human Biology, 21: 36–47.

Suda A. (1940) Anthropometric measurements of the RyukyuIslanders. Journal of the Anthropological Society of Nippon,55: 39–63 (in Japanese).

Suda A. (1950) The Ryukyu Islanders viewed from Anthropology.Japanese Journal of Ethnology, 15: 109–116 (in Japanese).

Suzuki H. (1969) Microevolutional changes in the Japanese popu-lation from the prehistoric age to the present-day. Journal ofthe Faculty of Science, University of Tokyo, Section V, Vol.III, Part 4: 249–309.

Tagaya A. and Ikeda J. (1976) A multivariate analysis of the cra-nial measurements of the Ryukyu islanders (males). Anthro-pological Science, 84: 204–220.

Turner C.G., II (1990) Major features of Sundadonty and Sino-donty, including suggestions about East Asian microevolu-tion, population history, and late Pleistocene relationshipswith Australian aboriginals. American Journal of PhysicalAnthropology, 82: 295–317.

Ueda T. (1935) Anthropometric comparison between the Koreansand the Japanese. In: The Anthropological Society of Tokyo(ed.), Nihon-Minzoku. Iwanami, Tokyo, pp. 112–163 (inJapanese).

Wakebe T., Saiki K., and Nagashima S. (1996) Facial flatness ofprehistoric peole in Okinawa, Japan. Acta AnatomicaNipponica, 71: 56 (abstract in Japanese).

Watanabe T., Saiki K., Okamoto K., and Wakebe T. (2004) Metri-cal and nonmetrical analyses of modern female crania in thenorthwestern Kyushu area. Anthropological Science, 112:147–159.

Woo T.L. and Morant G.M. (1934) A biometric study of the‘flatness’ of the facial skeleton in man. Biometrika, 26: 196–250.

Yamaguchi B. (1973) Facial flatness measurements of the Ainuand Japanese crania. Bulletin of the National ScienceMuseum, 16: 161–171.

Yamaguchi B. (1981) Human skeletal remains from Hokkaido. In:Ogata T. (ed.), Anthropology 5: The Japanese I. Yuzankaku,Tokyo, pp. 137–156 (in Japanese).

Yamaguchi B. (1982) A review of the osteological characteristicsof the Jomon population in prehistoric Japan. Anthropologi-cal Science (Supplement), 90: 77–90.

Yamaguchi-Kabata Y., Nakazono K., Takahashi A., Saito S.,Hosono N., Kubo M., Nakamura Y., and Kamatani N. (2008)Japanese population structure, based on SNP genotypes from7003 individuals compared to other ethnic groups: effects onpopulation-based association studies. American Journal ofHuman Genetics, 83: 445–456.

Yoshiura K., Kinoshita A., Ishida T., Ninokata A., Ishikawa T.,Kaname T., Bannai M., Tokunaga K., Sonoda S., Komaki R.,and others. (2006) A SNP in the ABCC11 gene is the deter-minant of human earwax type. Nature Genetics, 38: 324–330.