Index Introduction Materials and Methods Man Bac samples Comparative samples Statistical procedures Results Cranial metric affinity Dental metric affinity Discussion Acknowledgments References

Introduction

The site of Man Bac is located in Yen Mo district, Ninh Binh province, northern Vietnam (109°59′17″ East and 20°08′00″ North) (Figure 1). The site is almost completely surrounded by karst limestone mountains, and extends in an arc on a terrace of soil and sand lying to the south of the mountain range. The coast is approximately 25 km from the site.

View Details

Figure 1. Location of the Man Bac site in Ninh Binh Province, Vietnam.

This site was discovered in 1998 during reconnaissance work carried out by the Vietnamese Institute of Archaeology and the Ninh Binh Provincial Museum. According to preliminary findings, the site was considered to be a large-scale settlement of the late Neolithic/early Bronze Age. Initial excavation was undertaken in 1999 by the Institute of Archaeology, Ha Noi, Vietnam, and uncovered a substantial amount of stone and ceramic artifacts. The stone artifacts comprised axes, adzes, chisels, grinding stones, hammer stones, saw blades, and various types of semi-precious stone ornaments and jewelry. Ceramic artifacts comprised net sinkers, high-footed bowls, pediform clay support stands, mushroom-shaped ceramic anvils (used for pottery making), and over 10000 potsherds. In addition, five well-preserved burials were uncovered, in which six individuals were buried.

In 2001, the Vietnamese Institute of Archaeology, in collaboration with the Ninh Binh Provincial Museum, conducted the second excavation. This resulted in the finding of a substantial volume of stone and clay objects, associated with ten burials containing skeletons in good condition (Cuong, 2001; Phung, 2001; Hiep and Phung, 2004).

In order to undertake a more extensive investigation of Man Bac, especially regarding the human remains and burials, an international team, consisting of members from Vietnam, Japan, Australia, and the USA, carried out a third excavation season during December 2004 and January 2005, and revealed 31 additional inhumation burials associated with a considerable number of material objects (Figure 2). In the first three excavation seasons, therefore, a total of 48 skeletons were recovered from the combined grids, covering an area of over 90 square meters.

View Details

Figure 2. View of the excavation trench and inhumation burials in the Man Bac site (top: over view of trench looking from North, middle left: burial C30, middle right: burial C31, bottom left and right: burial C29).

The temporal context of the Man Bac site was determined to be a transitional period from the late Neolithic to the early Bronze Age (3800 years–3500 years BP). In terms of local cultural chronologies, the material culture displays many characteristics close to those of the Phung Nguyen period, whose people inhabited more northerly areas of Vietnam (Tri and Tan, 1999), primarily along the upper reaches of the Red River. In a Southeast Asian regional context, the overall material evidence suggests that the ways of life of the Man Bac people had much in common with those seen at other contemporaneous sites in Vietnam and Southeast Asia, such as evidence for agriculture, land clearance, ceramic manufacture, hunting, marine resource gathering and trade (Hiep and Phung, 2004; Huffer, 2005).

The aim of the present morphometric study utilizing the Man Bac skeletal sequence revolves around two questions. The first focuses on the local population history of this region of northern Vietnam. The authors seek to investigate whether or not the Man Bac people were direct descendants of early Neolithic Da But peoples, or earlier Neolithic Bac Son peoples. Or, can their origins be traced even further back to the pre-Neolithic Hoabinhian peoples who first settled the surrounding area? Furthermore, an exploration of possible biological relationships with later early historic Dong Son people, and the present-day Vietnamese, is also relevant to this question.

The second aim, dictated by the results of the above-mentioned micro regional analysis presented below, is to challenge the currently dominant model of the population history of Southeast Asia from a macro regional perspective. Since the early 20th century, it has been argued that Southeast Asia was initially occupied by an ‘Australo-Melanesian’ population that later underwent substantial genetic admixture with East Asian immigrants associated with the spread of agriculture from the Neolithic period onwards. This model is currently referred to as the ‘two layer’ hypothesis (Callenfels, 1936; Mijsberg, 1940; Barth, 1952; Von Koenigswald, 1952; Coon, 1962; Thoma, 1964; Jacob, 1967, 1975; Brace, 1976; Howells, 1976; Brace et al., 1991). A number of recent reviews of archeological studies conclude that a rice farming people with associated Neolithic material culture spread south from the Yangtze Basin into mainland and island Southeast Asia from the fifth millennium BP (Spriggs, 1989; Bellwood, 1987, 1997; Bellwood et al., 1992; Glover and Higham, 1996). In order to further refine, or reject, this scenario of the expansion of East Asian farming populations into Southeast Asia, we need to focus more on the biological relationships between the Neolithic and pre-Neolithic occupants of each region of Southeast Asia. Given these perspectives, the Man Bac skeletons are crucial materials to assess the ‘two layer’ hypothesis. Along with the above-mentioned aims, this paper presents results pertaining to population affinities of the Man Bac skeletal sequence with early and modern population samples from the area covering East/Southeast Asia and the West Pacific regions based on cranial and dental metric data.

Materials and Methods

Man Bac samples

The skeletal human remains unearthed at the Man Bac site through three excavation seasons, 1999, 2001, and 2004–2005, are listed in Table 1 with their age estimates, sex determinations, and conditions of preservations. Age estimates were made based on tooth eruption, extent of tooth attrition, cranial suture closures, postcranial epiphyseal unions, pubic symphyseal face morphology, sequential changes at sternum and rib and severity of osteoarthritis. Sex determination was based on pelvic features, cranial morphology and epiphyseal size of the long-bones. The age and sex were assessed using standard anthropological methods recommended by Buikstra and Ubelaker (1994). This sex estimation is not highly reliable for infant skeletons, so determination of sex relies on Schutkowski’s (1993) techniques based on mandibular morphology as necessary. Of a total of 48 individuals, 17 specimens are adults; the remaining 31 skeletons are children or adolescents. From this sequence, the present study used male adult skulls for cranial metric analysis and male permanent teeth for dental metric analysis. The adult C16 was not used for this study despite the whole body remaining, because the skull was not recovered as it was beyond the excavation trench.

The cranial and dental metric measurements recorded for the Man Bac specimens combining all three season’s datasets, are given in Table 2 and Table 3. The cranial measurements were undertaken following Martin’s definitions (Bräuer, 1988), and Yamaguchi’s (1973) method was utilized for facial flatness measurements. The tooth crown diameters were recorded as maximum diameters according to Fujita’s system (1949).

Figure 3 displays nearly complete adult and near-adult (C13) skeletons recovered from our 2004–2005 excavation at Man Bac. In general, except for one individual described below, the cranial morphology is characterized by relatively narrow faces, flat glabellas, superciliary arches and nasal roots, and round orbits. As shown in Figure 2 and Figure 4, only the male cranium from C29 presents quite unique features such as a dolichocephalic calvarium, large zygomatic bones, a remarkably prominent glabella and superciliary arches, a concave nasal root and a low and wide face with a protruding mandible.

View Details

Figure 3. Views of the representative adult skulls unearthed from the Man Bac site by 2004–2005’s excavation.

View Details

Figure 4. Frontal and oblique views of the No. C29 adult male skull excavated from the Man Bac site.

Comparative samples

Cranial and dental measurements from representative late Pleistocene and Holocene human specimens from the surrounding mainland Southeast Asian region, including China and the Southwest Pacific, were used as a basis for comparison with the Man Bac skeletons. A list of comparative cranial and tooth samples and data sources are given in Table 4. All the comparative data are from males.

Data from pre- and proto-historic Vietnamese samples were from the following four skeletal series. The Mai Da Nuoc specimen is from the early Holocene, and is a nearly complete skull of the late Hoabinhian period (c. 8000 years BP) excavated from the Mai Da Nuoc rock shelter in Thanh Hoa Province, northern Vietnam (Cuong, 1986). Bac Son refers to an early to middle Holocene Neolithic culture, and the skeletal series used here are from the sites of Lang Cuom, Pho Binh Gia, Khac Kiem, and Keo Phay in northern Vietnam (c. 10000 years–6000 years BP; Mansuy and Colani, 1925; Huard and Saurin, 1938; cf. Pietrusewsky and Douglas, 2002, pp. 223–224). The Da But culture belongs to the middle Holocene Neolithic period, and is represented by samples from the Con Co Ngua site from Ha Trung district, Than Hoa province (c. 5000 years BP; Patte, 1965; Thuy, 1990; Bui, 1991). Dong Son refers to the early Iron Age culture of north Vietnam, and is represented here by specimens from the sites of Vinh Quang, Chau Son, Doi Son, Nui Nap, and Minh Duc located near Ha Noi, which date from c. 3000–1700 years BP (Thuy, 1993).

Early to Middle Holocene sites of Malaysia and Thailand provided three samples used in this study. Gua Cha is a rock shelter site located in the State of Kelantan, Malaysia, dug by Sieveking (1954). The skeletal series from this site consists mainly of middle Holocene specimens belonging to the late Hoabinhian, but also includes remains from the subsequent Neolithic period (c. 3000 years BP). Although the Hoabinhian and Neolithic burials clearly derive from different strata, they are combined here since previous research found few morphological differences between the two phases (Bulbeck, 2000). Cranial metrics were provided only for the specimen numbered Gua Cha H12, because no other complete skulls exist. The Guar Kepah series was discovered in a shell midden site in Lenggong district, Malaysia, and belongs to the late Hoabinhian period (Mijsberg, 1940; Callenfels, 1936). Even though some pottery was also recovered, its precise stratigraphy is unclear (Bellwood, 1997; Bulbeck, 2000). The Ban Chiang series comes from an early Metal Age (Bronze Age, c. 3500 BP) site in Thailand, with direct evidence for local agriculture (including domesticated rice) and early metallurgy (including early bronze) (Gorman and Charoenwongsa, 1976; Pietrusewsky and Douglas, 2002).

From China, three Neolithic to early historic samples were used for comparison. The Weidun site is located near Shanghai in the Lake Tai region south of the Yangtze River in Jiangsu Province. The human remains from Weidun belong to the Majiabang culture, which corresponds to the earliest Neolithic period in the Lake Tai region (Chang, 1986). Radiocarbon dating places the earliest phase of the Majiabang culture at c. 7040 years BP and the latest at c. 5245 years BP (Nakahashi and Li, 2002). The frequent recovery of carbonized rice along with the human remains suggests that these people were associated with a rice-farming culture. The Anyang series, dated to c. 3400–3200 years BP, and classified as Bronze Age, was excavated from Shang period sites in Henan Province (IHIA and CASS, 1982). These specimens are from sacrificial burial pits in and around the imperial tombs at Anyang. The Jiangnan series consists of specimens dated to the Zhou and Western Han period (c. 2770–1992 years BP), excavated from various sites in the Chang Jiang basin and Jiangsu Province, south of the Yangtze River. The details of these specimens and sites are described in Nakahashi and Li (2002), in which a close biological relationship with the Yayoi immigrants in Japan is advocated.

The Jomon represent early native hunter-gatherers who lived in the Japanese archipelago from c. 13000–2300 years BP (Akazawa and Aikens, 1986). The samples used here are from various regions in Japan, dating from the middle to final Jomon phases (c. 4000–2300 years BP). The Yayoi samples, dated from c. 2400–1750 BP (Hudson, 1990), came from western Japan, and include materials from the Kanenokuma site in Fukuoka Prefecture, as well as from the sites of Doigahama and Nakanohama in Yamaguchi Prefecture. All of these specimens are considered to be among the first rice cultivators in the Japanese archipelago, and are believed to represent immigrants from the Asian continent or their descendants (Kanaseki et al., 1960; Nakahashi, 1989). In addition to these ancient series, data recorded for twelve modern samples, as listed in Table 4, were used for cranial or dental metric comparisons.

Statistical procedures

Eleven cranial measurements (Martin’s numbers 1, 8, 17, 40, 45, 46, 48, 51, 52, 54 and 55) were selected for statistical comparisons, because all of the comparison series, including Man Bac, had these variables in common and readily accessible.

Similarities in cranial metric proportions were estimated by Q-mode correlation coefficients based on male data from the eleven cranial measurements above. The measurement data were then standardized using grand mean values of all comparative samples and standard deviations of Vietnamese sample recorded by Cuong (1996).

Before grouping Man Bac specimens as a single population sample, in order to confirm impressions of morphological uniqueness of C29, cranial similarities were analyzed between individually separated specimens.

To aid in the interpretation of the matrix of inter-sample phenetic distances, cluster analysis using the un-weighted pair-group method (UPGMA: Sneath and Sokal, 1973) were applied to the Q-mode correlation coefficients.

In the same way, dental metric comparisons were made using the mesiodistal and buccolingual crown diameters of males, excluding the third molars. Before calculating Q-mode correlation coefficients among the populations, the 28 crown diameters were summarized into a small number of new values, taking correlations between the same teeth, and between mesiodistal and buccolingual diameters, into account. In order to reduce the measurements to such a small number of values, the results of the factor analysis conducted on Japanese males in a separate study (Matsumura, 1994) were utilized by assuming that the obtained factors derived from such a single representative population are shared among all human populations. Table 5 gives loading values greater than 0.6 for the six primary factors whose eigenvalues were greater than 1.0. On the basis of these factor loadings and the mean diameters for each population sample, scores of the six factors were calculated. Q-mode correlation coefficients were computed using the factor scores in order to measure similarities in tooth size proportion. The grand mean scores and standard deviations obtained from the Japanese samples given in Table 5 were then used.

Calculations of Q-mode correlation coefficients using individually separated data were not made in the dental metric analysis, because the procedure based on the factor scores requires nearly full battery of crown diameters which was obtained from only a few individual specimens. Nevertheless in order to assess the uniqueness of individual C29, only this specimen was separated from others in calculation of mean values of crown diameters as given in Table 3.

Finally, we applied the cluster analysis to the Q-mode correlation coefficients to provide a summarized pattern of population affinities.

Calculating distances of Q-mode correation coefficients and cluster analysis were processed using the data analysis software ‘STATISTICA version 06J’ (StatSoft Inc.).

Results

Cranial metric affinity

Figure 5 represents the results of cluster analysis applied to the Q-mode correlation coefficients, based on the eleven cranial measurements of six individuals from Man Bac and 21 comparative population samples. This analysis yields a separation of Man Bac C29 and the other five Man Bac individuals. C29 specimen was placed in the cluster containing the Hoabinhian/early Neolithic Southeast Asians and Australo-Melanesians, while the other five Man Bac specimens were grouped with the early Metal Age to modern East/Southeast Asians and Neolithic Weidun sample.

View Details

Figure 5. Dendrogram of a cluster analysis applied to the distances of Q-mode correlation coefficients between the six individuals of Man Bac and comparative samples, based on eleven cranial measurements.

Since individual C29 was thus clearly separated from other comparable individuals of Man Bac, this individual was dealt with as a separate sample, and the Q-mode correlation coefficients were calculated using the mean values of Man Bac males excluding specimen C29, i.e. as given in Table 2. Table 6 gives distance values from the pooled Man Bac males excluding C29 (hereafter referred to as Man Bac) and those from individual C29 of Man Bac, which were transformed (1–r) from Q-mode correlation coefficients (r). The closest sample to the Man Bac skulls is the early Iron Age Dong Son people in Vietnam and the next closest is the Zhou and Western Han Period Jiangnan sample. On the other hand, individual C29 shows close similarities to the Neolithic Jomon, Hoabinhian Mai Da Nuoc, early Neolithic Bac Son and Da But Vietnamese.

Figure 6 shows a dendrogram resulting from a cluster analysis applied to the Q-mode correlation coefficients. The Man Bac and Dong Son people, together forming a sub-cluster, are connected with one of two main clusters consisting of the modern Vietnamese and early Metal Age to modern samples from East/Southeast Asians and Neolithic Weidun sample. The Jomon sample and individual C29 of Man Bac are neighbors, and they branch off from another major cluster, in which the other prehistoric Vietnamese, represented by the Mai Da Nuoc Hoabinhian specimen, and the early Neolithic Bac Son and Da But samples, are grouped together, and adjacent to the Gua Cha Malay and Australo-Melanesian samples.

View Details

Figure 6. Dendrogram of a cluster analysis applied to the Q-mode correlation coefficients between individual C29 of Man Bac, pooled other Man Bac specimens and comparative samples, based on eleven cranial measurements.

Dental metric affinity

Table 6 gives the distances of Q-mode correlation coefficients from individual C29 and those from the pooled other Man Bac specimens, based on of the six factor scores of the crown diameters. The Man Bac sample is quite close to the Weidun sample from Neolithic Southern China and the modern Indochinese samples, such as the Laotians, Vietnamese, and Thailanders. The Dong Son Vietnamese are also relatively close to Man Bac. In contrast, individual C29, which is far removed from the other Man Bac specimens, is closest to the early Neolithic Da But, followed by the Jomon and Australian aborigine series.

Figure 7 displays the results of the cluster analysis applied to the Q-mode correlation coefficients. The Man Bac sample, together with the Laotians, is loosely connected with the cluster consisting of the early and modern samples from China, the Yayoi migrants, the Dong Son series, and the modern Vietnamese and Thailanders. On the other hand, individual C29 of Man Bac, as well as the early Neolithic Bac Son and Da But Vietnamese, are placed in the other major cluster, which consists of the Australo-Melanesians, Andaman, Jomon and Hoabinhian, and Neolithic Malaysian samples.

View Details

Figure 7. Dendrogram of a cluster analysis applied to the Q-mode correlation coefficients between individual C29 of Man Bac, pooled other Man Bac specimens and comparative samples, based on six factor scores of tooth crown diameters.

Discussion

The large morphological gaps between the late Neolithic Man Bac people, except the individual buried in grave C29, and the early Neolithic Vietnamese as represented by the Bac Son and Da But series, were evident in the results of the multivariate analyses of both cranial and dental metrics. Further, these analyses demonstrated the considerable dissimilarity of the Man Bac people, except for individual C29, to the pre-Neolithic Hoabinhian period samples in the Indochina region. The earlier prehistoric Vietnamese, as well as the early Malaysians, resembled the Australo-Melanesian samples the most. Closer affinities to the Man Bac people amongst the series analyzed here were found in the subsequent early Iron Age Dong Son sample, and in the early groups from southern China. Some discrepancies are noticeable when cranial and dental affinities are compared in terms of the closest Chinese samples to Man Bac, i.e. the Jiangnan sample of the Zhou and Western Han periods for the cranial metrics, and the Neolithic Weidun sample for the dental metrics. Regardless of the chronological differences, it may be concluded that the Man Bac people were immigrants affiliated with the populations along the Yangtze River region in southern China, and that they connect, through the subsequent early Iron Age Dong Son people, to the present-day majority of the Vietnamese. Thus, the Man Bac population (excluding individual C29) is not directly descended from the Hoabinhian, Bac Son, or Da But indigenous populations.

The ‘immigration’ hypothesis, also called the ‘two layer’ model, as a means of understanding the population history of Southeast Asia, has been supported by a wide array of archaeological, genetic, and historical linguistic studies. These researchers have shown that the pre-modern expansion of language families, specifically the Austronesian and Austroasiatic families, can often be linked with the dispersal of rice-cultivating populations during the Neolithic period (Renfrew, 1987, 1989, 1992; Bellwood, 1991, 1993, 1997; Hudson, 1994, 1999, 2003; Higham 1998, 2001; Hill, 2001; Bellwood and Renfrew, 2003; Diamond and Bellwood, 2003).

Formerly, the conceptualization of this scenario has been driven by many earlier analyses of preceramic period human remains, which described morphological features akin to those of Australian Aborigines or Melanesians (e.g., Evans, 1918; Duckworth, 1934; Mijsberg, 1940; Trevor and Brothwell, 1962; Jacob, 1967). As a result, it had been argued that Southeast Asia was occupied by an indigenous population, sometimes referred to as ‘Australo-Melanesian,’ before immigrants from East Asia dispersed widely into this region (Callenfels, 1936; Mijsberg, 1940; Barth, 1952; Von Koenigswald, 1952; Coon, 1962; Thoma, 1964; Jacob, 1967, 1975; Brace, 1976; Howells, 1976; Bellwood, 1987, 1997; Brace et al., 1991).

Even today, molecular anthropological studies using classic genetic markers and mtDNA have demonstrated many biological similarities between Chinese and Southeast Asians (Ballinger et al., 1992; Cavalli-Sforza et al., 1994; Omoto and Saitou, 1997; Tan, 2001), which suggest some degree of genetic influence of southern Chinese populations on Southeast Asian populations. Ding et al. (2000) also found that populations in East and Southeast Asia were genetically very similar, and the result may suggest north-south migration, although the authors themselves prefer to interpret the genetic patterns as simply due to isolation by distance.

However, there have long been criticisms of the traditional ‘two layer’ model. Although genetic and cranial data have also been used, teeth have figured prominently in such criticisms. Turner’s (1987) Sundadont/Sinodont hypothesis uses dental evidence to propose an alternative to the ‘two layer’ model, and other researchers have argued that the immigration hypothesis is not necessarily supported by recent analyses of cranial and dental data. Turner (1987, 1989, 1990, 1992) assumed that the array of nonmetric dental traits, so-called ‘Sundadont’ traits, possessed by present-day Southeast Asians, are the product of long-standing continuity uninterrupted by significant admixture with ‘Sinodont’ peoples from the north. Multivariate craniometric analyses by Pietrusewsky (1992, 1994, 1999) and Hanihara (1992a, b, 1993a, b, c, 1994) have demonstrated relatively close affinities between prehistoric and modern Southeast Asians, coupled with a distinct dissimilarity to Australo-Melanesians. The prehistoric Southeast Asians used by these authors are mainly Neolithic to early Metal Age samples from Thailand, Vietnam and Laos, and the finding of similarities with modern Southeast Asians is accepted as evidence for regional continuity in Southeast Asian population history.

These previous analyses supporting regional continuity, however, did not include extensive Mesolithic/Hoabinhian samples from Southeast Asia. Up until now, only a few pre-Neolithic human remains have been discovered from Southeast Asia, as represented by the Gua Gunung Runtuh skeleton from Malaysia (Zuraina 1994; Jacob and Soepriyo, 1994; Matsumura and Zuraina, 1995, 1999), the Moh Khiew Cave specimen from Thailand (Pookajorn 1991, 1994; Matsumura and Pookajorn, 2005), the Mai Da Nuoc, and Mai Da Dieu skulls, and the Hang Cho skeleton from Vietnam (Cuong, 1986; Matsumura et al., 2004). Analyses of their skeletal morphologies demonstrated Australo-Melanesian characteristics, suggesting ancestral biological connections with each other. Furthermore, a broad comparison of dental traits conducted by Matsumura and Hudson (2005) found a close affinity between the Hoabinhian and Australo-Melanesian samples, but identified northern source in the contemporary Southeast Asians, regarded as supporting the immigration hypothesis rather than regional continuity. Although the main source of the ‘East Asian’ populations who expanded into Southeast Asia, it is argued, was China, those people were not Chinese in the historical sense of that term. In the present study, nonmetric dental traits, as well as cranial nonmetric ones, were not analyzed, because comparisons of these traits, based on frequency data, need large sample sizes. Accumulation of human remains by further excavation at the Man Bac site is expected to address the issue from the perspective of Turner’s Sinodont/Sundadont model.

In Vietnam, concerning the early Iron Age Dong Son people, Cuong (1996) found in his cranial study that they have close relations with the present day southern Chinese, Vietnamese, and Japanese, and stated that these so-called ‘Mongoloid’ features could have came from northern and eastern peripheral areas into northern Vietnam. The Man Bac skeletons allow present-day Vietnamese origins to be traced back to the pre-Dong Son period in this region i.e. the late Neolithic period. This may suggest a larger impact of these pre-Dong Son immigrants on the population history of Vietnam.

Concerning the individual buried in grave C29, this male exhibits morphologically unique cranial features that are considerably different from those of the other Man Bac specimens. Most of the Man Bac skulls possess flat and narrow faces, while the C29 cranium, as mentioned above, presents much more robust features. Such a unique, robust, cranial morphology does not rule out the possibility that individual C29 derived from a different population source. The cranial and dental metric analyses demonstrated that C29 has a close relationship, in regards to these atypical features, to the early Neolithic Bac Son and Da But people in Vietnam. This individual was probably a member of an indigenous group descended from pre/early Neolithic settlers, represented by the Bac Son and Da But peoples, and who integrated into the Man Bac community when it was settled by later Neolithic migrants. This individual was buried in association with a greater quantity of ceramics (this being the predominant grave goods category by far) than any other burial to date, including a peculiarly large vessel, suggesting that this man was fully accepted into the Man Bac community, and was buried as an individual of high status according to the mortuary customs already established (for details, see Huffer, 2005). Although the other Man Bac skeletons are regarded as constituting definitive evidence for the existence of initial immigrants from northern areas, the man in grave C29 clearly proves that this site did not consist of immigrants alone.

Acknowledgments

The authors are graterul to Director Dr. Ha Van Phung, and Vice Director Dr. Nguyen Giang Hai, Vietnamese Institute of Archaeology, for their permission to excavate the Man Bac site and for their cooperation.

This study was supported in part by a Grant-in-Aid in 2003–2005 (No. 15405018) from the Japan Society for the Promotion of Science, and by the Toyota Foundation (No. D06-R-0035).