PDA

View Full Version : Craniometric Investigation of the Bronze Age Settlement of Xinjiang



Polak
Sunday, October 5th, 2003, 05:02 PM
Craniometric investigation of the Bronze Age settlement of Xinjiang


American Journal of Physical Anthropology
Early View (Articles online in advance of print)

©2003 Wiley and Sons, Inc. All Rights Reserved.


Research Article
Horse-mounted invaders from the Russo-Kazakh steppe or agricultural colonists from western Central Asia? A craniometric investigation of the Bronze Age settlement of Xinjiang
Brian E. Hemphill 1 *, J.P. Mallory 2
1Department of Sociology and Anthropology, California State University at Bakersfield, Bakersfield, California 93311-1099
2School of Archaeology and Palaeoecology, Queen's University, Belfast, Northern Ireland BT7 1NN, UK

email: Brian E. Hemphill (bhemphill@csub.edu)

*Correspondence to Brian E. Hemphill, Department of Sociology and Anthropology, California State University at Bakersfield, Bakersfield, CA 93311-1099

Keywords
craniometry • phenetic distance • Tarim Basin • Bactria • China


Abstract

Numerous Bronze Age cemeteries in the oases surrounding the Täklamakan Desert of the Tarim Basin in the Xinjiang Uyghur Autonomous Region, western China, have yielded both mummified and skeletal human remains. A dearth of local antecedents, coupled with woolen textiles and the apparent Western physical appearance of the population, raised questions as to where these people came from. Two hypotheses have been offered by archaeologists to account for the origins of Bronze Age populations of the Tarim Basin. These are the steppe hypothesis and the Bactrian oasis hypothesis. Eight craniometric variables from 25 Aeneolithic and Bronze Age samples, comprising 1,353 adults from the Tarim Basin, the Russo-Kazakh steppe, southern China, Central Asia, Iran, and the Indus Valley, are compared to test which, if either, of these hypotheses are supported by the pattern of phenetic affinities possessed by Bronze Age inhabitants of the Tarim Basin. Craniometric differences between samples are compared with Mahalanobis generalized distance (d2), and patterns of phenetic affinity are assessed with two types of cluster analysis (the weighted pair average linkage method and the neighbor-joining method), multidimensional scaling, and principal coordinates analysis. Results obtained by this analysis provide little support for either the steppe hypothesis or the Bactrian oasis hypothesis. Rather, the pattern of phenetic affinities manifested by Bronze Age inhabitants of the Tarim Basin suggests the presence of a population of unknown origin within the Tarim Basin during the early Bronze Age. After 1200 B.C., this population experienced significant gene flow from highland populations of the Pamirs and Ferghana Valley. These highland populations may include those who later became known as the Saka and who may have served as middlemen facilitating contacts between East (Tarim Basin, China) and West (Bactria, Uzbekistan) along what later became known as the Great Silk Road. Am J Phys Anthropol, 2003. © 2003 Wiley-Liss, Inc.



--------------------------------------------------------------------------------
Received: 25 October 2002; Accepted: 2 June 2003
Digital Object Identifier (DOI)

10.1002/ajpa.10354 About DOI


Article Text



Schoolchildren in the United States are taught that the peoples of western Asia and Europe remained ignorant of the populations of East Asia for many centuries, but that this changed with the incredible journey of Marco Polo. According to popular account, Marco Polo, a Venetian merchant, left Venice along with his father and uncle in A.D. 1269 and traveled east across Central Asia along the route of what later came to be known as the Great Silk Road, arriving at the court of Kublai Khan at Beijing in A.D. 1275. There they were received by the great Khan and remained for the next 16 years, only to return to Venice with news of the wonders of the East in A.D. 1295 (Komroff, [1930], p. vii-xx). Yet, despite popular perception, the numerous inaccuracies, apparent plagiarisms, and profound gaps of observation evident in The Travels of Marco Polo raise serious doubts as to whether Polo ever went to China at all (Mallory and Mair, [2000], p. 71).

Those with more than a passing familiarity often identify the date 132 B.C. as the beginning of contacts between East and West along the Great Silk Road (Barber, [1999], p. 19; Di Cosmo, [1996], p. 87; Mair, [1995], p. 302; Mallory and Mair, [2000], p. 56). It was at this time that the Chinese explorer Zhang Qian embarked on a 13-year mission westward from Gansu, through Xinjiang, and across the Ferghana Valley into Central Asia. Soon the Emperor Wudi gathered an army of some 60,000 soldiers to secure the Great Silk Road, so that between 114 and 108 B.C. no less than 10 caravans a year were making the journey from China in the east to the Ferghana Valley in the west.

Other scholars maintain that the beginning of contacts between East and West occurred even earlier, for silk appears in Europe by the sixth century B.C., throughout the northern Mediterranean basin by the fifth century B.C, and perhaps as early as 1000 B.C. in North Africa (Mair, [1995], p. 285). In the fifth century B.C., Herodotus mentioned transit trade occurring across great distances in Central Asia along a route that stretched from the River Don in the Urals in the west to the Altai and the Minusinsk Basin in the east (Chlenova, [1983]).

Archaeological excavations at the site of Sapalli tepe, located in the North Bactrian oasis of southern Uzbekistan, during the 1960s and 1970s yielded the earliest evidence of silk outside of China and raised the possibility that contacts between East and West along the Great Silk Road may be far older than previously thought (Askarov, [1973], p. 133-134, [1974], [1977], [1981], [1988]). No longer were contacts dated to 13th century A.D., nor to the second century B.C., nor even to the 10th century B.C. Rather, the presence of silk at Sapalli tepe raised the possibility that contacts along the Great Silk Road may have occurred near the end of the third and the beginning of the second millennia B.C. (Hiebert, [1994]; Kohl, [1984]). Yet, as provocative as this discovery was, few scholars outside the former Soviet Union knew of these discoveries (Kohl, [1981], [1992]).

One of the major archaeological events of the past decade has been the proliferation of popular articles, books, and television documentaries devoted to the discovery of prehistoric Bronze Age Caucasoid or Europoid populations in the western Chinese Xinjiang Uyghur Autonomous Region (Mair, [1995], p. 281). Here along the oases of the Tarim Basin have been recovered some 300 mummies, many of which have been found along with their clothes in an extraordinary state of preservation, dating from ca. 1800 B.C. until the Chinese conquest of the region in the first centuries B.C. A far greater assemblage of skeletal remains has been recovered from Bronze and Iron Age cemeteries of the region (an assemblage that should be numbered in the many hundreds, if not thousands), though only a few more than 300 have been examined. Where it has been possible to identify a phenotypic pattern, the majority of these individuals are identified as possessing a stronger resemblance to a Western pattern (e.g., fair hair, high-bridged noses, heavy beard) rather than the phenotypic pattern common to the Han of China (Barber, [1999], p. 19; Mallory and Mair, [2000], p. 16; Wang, [2001]). Not surprisingly, such identifications led many to ask (Mair, [1995], p. 289), Who were the[se] corpses from the Tarim Basin? Where did they come from? And how did they get there?

Three lines of evidence have been offered to demonstrate that the Bronze Age inhabitants of Xinjiang were neither long-term indigenous inhabitants of this region, nor ethnic Han Chinese immigrants from the East. These lines of evidence include the textiles worn by these individuals, the evidence of Indo-European languages in Xinjiang, and previous biological analyses of the human remains themselves. The purpose of this paper is to compare craniometric variation among Bronze Age inhabitants of Xinjiang, western China, with Aeneolithic and Bronze Age samples from the Russo-Kazakh steppe, south Central Asia, southern China, Iran, and the Indus Valley, in order to test which of the hypotheses best explain the origins and subsequent interactions of the Bronze Age inhabitants of the Tarim Basin.

Mair ([1995], p. 295) considered the textiles worn and associated with the Xinjiang remains to be highly diagnostic, perhaps still more so than DNA analysis, for identifying the origins and affiliations of the Tarim Basin Bronze Age people. These textiles encompass an array of items including string skirts, fur-lined and fur-trimmed coats, long stockings, and pants that Mair ([1995], [1998]), Mallory and Mair ([2000]), and Barber ([1999]) claim are indicative of a close relationship with Indo-European-speaking pastoralist nomads from the Russian steppe. The most ancient Bronze Age remains found in Xinjiang derive from the site of Qäwrighul (ca. 1800 B.C.), located along the Könchi River at the southeastern edge of the Tarim Basin (Fig. 1). These remains were found clad in simple cloaks, mantles, and wraps in shades of natural brown or beige that lack any evidence of piping, sleeves, or trouser legs. Barber ([1999], p. 71) maintained that these individuals not only appear to have introduced weaving into the Tarim Basin, but the mere presence of woolen textiles reveals that domestic sheep from the West had been introduced into the Tarim Basin by the beginning of the second millennium B.C. (Mallory and Mair, [2000], p. 219).


Figure 1. Geographic location of craniometric samples. Sample abbreviations from Table 1. Xinjiang samples (QAW, ALW, and KRO) and Chinese sample from Hainan (HAI) are represented by asterisks; North Bactrian samples, by stars; Iranian samples, by pentagons; Turkmenian, Caucasus, and Tajik samples, by triangles; Indus Valley samples, by circles; and Russo-Kazakh samples, by squares.
[Normal View 104K | Magnified View 274K]


Table 1. Samples considered in study

--------------------------------------------------------------------------------

Code Maximum sample size
--------------------------------------------------------------------------------
Site/region Period Dates Reference
Males Females

--------------------------------------------------------------------------------

ADM 22 11 Andronovo/Minusinsk Late Bronze Age 2100-900 B.C. Alexeev ([1961])
AFA 17 7 Afanasievo/Altai Early Bronze Age 3500-2500 B.C. Alexeev ([1961])
AFM 18 11 Afanasievo/Minusinsk Early Bronze Age 3500-2500 B.C. Alexeev ([1961])
AND 25 31 Andronovo/Kazakhstan Late Bronze Age 2100-900 B.C. Alekseev ([1967])
ALT 40 42 Altyn depe/Turkmenistan Namazga V 2500-2200 B.C. Kiiatkina (1967); Hemphill ([1999])
ALW 33 25 Alwighul/Xinjiang, China Late Bronze Age 650-200 B.C. Han ([1998])
CEMH 10 18 Harappa/Indus Valley Late Harappan 1900-1600 B.C. Gupta et al. ([1962])
DJR 17 33 Djarkutan/North Bactria Djarkutan phase 2000-1800 B.C. Hemphill ([1998])
GKS 38 32 Geoksyur/Turkmenistan Namazga III 3500-3000 B.C. Kiiatkina ([1987]); Hemphill ([1999])
HAI 45 38 Hainan/South China Living Living Howells ([1989])
HAR 23 41 Harappa/Indus Valley Mature Harappan 2500-2000 B.C. Gupta et al. ([1962]); Hemphill et al. ([1991])
KAM 133 118 Karasuk/Minusinsk Late Bronze Age Rykusina ([1976])
KAR 14 13 Kara depe/Turkmenistan Namazga III 3500-3000 B.C. Ginzberg and Trofimova ([1972])
KOK 14 10 Kokcha III/Turkmenistan Late Bronze Age 1800-1500 B.C. Trofimova ([1961])
KRO 4 2 Kroran/Xinjiang, China Bronze/Iron Age 202 B.C.-A.D. 150 Han ([1998])
KUZ 13 14 Djarkutan/North Bactria Kuzali Phase 1800-1650 B.C. Hemphill ([1998])
KUZ 13 14 Djarkutan/North Bactria Kuzali phase 1800-1650 B.C. Hemphill ([1998])
MOL 18 28 Djarkutan/North Bactria Molali phase 1650-1500 B.C. Hemphill ([1998])
QAW 11 7 Qawrighul/Xinjiang, China Early Bronze Age 1800 B.C. Han ([1998])
SAMB 14 13 Samtavro/Caucasus Late Bronze Age 1400-800 B.C. Abduelivili ([1954])
SAP 13 28 Sapalli tepe/North Bactria Sapalli phase 2200-2000 B.C. Hemphill ([1998])
SHS 45 43 Shahr-i Sokhta/Eastern Iran SHS I, II, III 3000-2200 B.C. Pardini and Sarvari-Negahban ([1976]) Pardini ([1977], [1979-1980])
TH2 9 7 Tepe Hissar/Northern Iran Tepe Hissar II 3300-2500 B.C. Krogman ([1940])
TH3 102 36 Tepe Hissar/Northern Iran Tepe Hissar III 2500-1700 B.C. Krogman ([1940])
TMG 9 11 Timargarha/Indus Valley Late Bronze/Early Iron Age 1400-800 B.C. Bernhard ([1967])
TMM 26 21 Tigrovaja-Makoni Mor/ Tajikistan Molali phase 1650-1500 B.C. Kiiatkina ([1976])


--------------------------------------------------------------------------------



Some 500 years later, during the closing centuries of the second millennium B.C., mummies recovered from Zaghunluq (near Chärchän), located along the southern margin of the Täklamakan Desert in the southern Tarim Basin, document the introduction of an entire array of new techniques in clothing manufacture (Barber, [1999]). Textiles from the site of Qizilchoqa (near Qumul/Hami), in the easternmost part of the region, are marked by weaving of improved quality that includes twill as well as plain weave. A detailed examination of a textile fragment by Good ([1995]) yielded evidence of the same decorative technique as that found in Scottish tartans which, in turn, exhibited similarities to tartans found at Hallstatt in Austria dated to the late second millennium B.C. (Mallory and Mair, [2000]).

The second line of evidence is linguistic, and involves the discovery of written documents in the Tarim Basin that attest to the presence of a series of Indo-European languages collectively known as Tocharian (Barber, [1999], p. 115; Jettmar, [1998], p. 216; Mallory, [1998], p. 189; Renfrew, [1988], p. 63-66). These languages exhibit no close similarities to Indo-Iranian (which is well-represented by a number of languages in the Tarim Basin) or any other eastern (satem) Indo-European languages (other than loan words). Within the phylogeny of the Indo-European languages, Tocharian languages are either placed with the western (centum) languages of Europe (Adams, [1984]; Hamp, [1998]), or are regarded as languages that split from the rest of the Indo-European languages at such an early date that they lack many of the isoglosses found between other Indo-European languages (Ringe et al., [1998]). Mallory and Mair ([2000], p. 240-246) suggested that the separation of Tocharian speakers from other Indo-European-speaking communities may have occurred as early as the fourth millennium B.C., and they identified the Afanasievo culture (ca. 3500-2500 B.C.), a primarily pastoralist culture found in the Altai and Minusinsk regions of the Eurasian steppe, as a possible source for a Tocharian presence in the Tarim Basin (Mallory, [1989], p. 62, 263, [1995], p. 380-381, [1998], p. 189; see also Parpola, [1998]). Wall paintings of Tocharian speakers depict these individuals as possessing red or blonde hair, long noses, blue or green eyes, and wearing broadswords inserted in scabbards hanging from their waists (Mair, [1995], p. 299).

The third line of evidence comes from analyses of the biological features of the human remains themselves. This evidence derives from the obviously Western appearance of many of the mummified remains, analyses of ancient DNA, and craniometry. Francalacci ([1995]) obtained tissue samples from 11 mummies, but only two of these samples were permitted out of China, and the DNA from one was too damaged for analysis. Hence, at present, the genetic evidence for the history of the Bronze Age inhabitants of Xinjiang rests on results obtained from a single individual (Mallory and Mair, [2000], p. 246-247). The mtDNA of this mummy was identified by Francalacci ([1995]) as belonging to haplogroup H, one of nine subtypes of mitochondrial lineages largely associated with Europeans. Yet, while haplogroup H is the most common marker of European populations (40%), it is also found in 15% of individuals from the Near East. Hence, despite claims to the contrary (see Cavalli-Sforza, [2000]), the ancient DNA evidence currently available offers little resolution concerning the precise origins of the Bronze Age inhabitants of Xinjiang.

Over the course of the last two decades, Han ([1994a], [b]; see also Mair, [1998]) conducted craniometric analyses of some 302 adult Bronze Age inhabitants of Xinjiang. Han ([1998], p. 568) concluded that metric variation among these individuals revealed that the majority (89%) may be attributed to at least three branches of the Caucasoid type, while a minority (11%) may be ascribed to two branches of the Mongoloid type. Han ([1994a], [1998], p. 566-568) contended that the temporal and geographic patterns observed among these cranial types document a three-stage settlement of Xinjiang during the Bronze Age. According to Han ([1998]), the first wave of immigration involved colonization by a proto-Europoid type with Nordic characteristics that he attributed to Afanasievo and Andronovo populations who migrated to Xinjiang and the Tarim Basin from the steppelands located to the north and northwest. Han ([1998]) maintained that a second wave of immigration, this time involving an Eastern Mediterranean type and possibly a Pamir-Fergana type, entered Xinjiang and the Tarim Basin from the west around 500 B.C. The third wave of immigration involved a westward migration of an Eastern Mongoloid type into the eastern regions of Xinjiang and the Tarim Basin, most likely from the adjacent province of Gansu and points further east (see also An, [1992b]; Shui, [1993]).

MODELS FOR XINJIANG POPULATION ORIGINS


Contemporary researchers are divided over the most likely explanation for the origins of the Bronze Age inhabitants of Xinjiang, and this division is reflected by the wide array of explicative accounts advanced by archaeologists, biological anthropologists, historical linguists, and others. Nevertheless, these explanations can be grouped into two general models. These models may be designated the steppe hypothesis and the Bactrian oasis hypothesis.

Proponents of the steppe hypothesis maintain that the Tarim region experienced at least two population influxes from the Russo-Kazakh steppe region (Anthony, [1998]; Han, [1998]; Kuzmina, [1994], [1998]; Mallory, [1995]; Mallory and Mair, [2000]; Parpola, [1998]). They suggest that the initial wave of immigration into Xinjiang may have originated among members of the Afanasievo culture found in the Altai and Minusinsk regions of the steppe north of the Tarim Basin. This attribution is based on numerous similarities in material culture, including bronze metallurgy, burial practices, and textiles, found between Afanasievo culture sites and the earliest Bronze Age sites in Xinjiang, such as Qäwrighul (Kuzmina, [1994], p. 241, [1998], p. 69; Mallory, [1995]; Mallory and Mair, [2000]).

The Afanasievo culture itself is believed intrusive to the eastern steppe and is held to share the closest similarities to the Yamnaya culture (3500-3000 B.C.; Mallory and Mair, [2000]) found in the Pontic-Caspian region of the Russian steppe (Alexeev, [1961]; Anthony, [1998], p. 104; Chernykh, [1992]; Gryaznov and Vadezkaya, [1968]; Kuzmina, [1998]; Mallory, [1989], p. 62, [1995], [1998], p. 189; Mallory and Mair, [2000]; Parpola, [1998]; Posrednikov, [1992]; but see Shishlina and Hiebert, [1998], p. 222-223). As noted by Mallory ([1995]) and Mallory and Mair ([2000], p. 381-382), an eastward emigration and subsequent isolation of Yamnaya-derived Afanasievo populations in the eastern steppe provides a possible explanation for the appearance of Tocharian in the Tarim Basin of Xinjiang. Hence, the apparent similarities between Tocharian and such western Indo-European languages as Celtic and Italian are due to a common retention of proto-Indo-European archaicisms (Mallory, [1989], p. 61), while the differences between Tocharian and neighboring Indo-Iranian might be explained by the peripheral position of proto-Tocharian populations with respect to the emergence of Indo-Iranian languages. This assertion appeared to be supported by Han's ([1998]) contrast of cranial and facial indices which indicated that the earliest individuals from Qäwrighul (type I tombs; see below) were most similar to individuals from Afanasievo cultural contexts (see Mallory and Mair, [2000], p. 240-243).

Many proponents of the steppe hypothesis contend that the immigration of Afanasievo populations to the Tarim Basin was followed by a later influx of populations derived from the Late Bronze Age Andronovo culture complex (ca. 2100-900 B.C.; Mallory and Mair, [2000]) found to the west, northwest, and north in the Pamirs, the Ferghana Valley, Kazakhstan, and the Minusinsk/Altai region (Chen and Hiebert, [1995]; Kuzmina, [1998]; Mei and Shell, [1998]; Parpola, [1998]). As with earlier Afanasievo populations, those accompanied by artifacts assigned to the Andronovo culture are believed to have originated in the western Russian steppe. In this latter case, ultimate stylistic origins of the artifacts are traced to the Sintashta culture (ca. 2300-1900 B.C.; Mallory and Mair, [2000]) of the southeast Urals (Anthony, [1998]; Anthony and Vinogradov, [1995], p. 36; Kuzmina, [1994]; Mallory, [1998]). The eastward expansion of these likely Indo-Iranian-speaking peoples was facilitated by the development of the war chariot (Anthony, [1998], p. 94; Di Cosmo, [1996], p. 90-91; Mallory, [1989]; Parpola, [1998]) and is reflected by the rapid appearance of such regional variants of the Andronovo culture designated as Alakul', Federovo, Tazabag'yab, Beshkent, and Vakhsh (Gupta, [1979]; Hiebert, [1994]; Kohl, [1984], p. 183-184; Kuzmina, [1994]; Masson, [1992b], p. 350-351; Sulimirski, [1970], p. 261, 263; Zdanovich, [1988]). The initial appearance of Andronovo-derived populations in the Tarim Basin is held to be signaled by the introduction of new clothing styles, ceramic wares, and burial customs as well as objects of tin bronze, and objects associated with horses around 1200 B.C. (Barber, [1999]; Kuzmina, [1998], p. 73; Mei, [2000], p. 72-75; Mei and Shell, [1998]).

The second model for the origin of the Bronze Age inhabitants of Xinjiang is the Bactrian oasis hypothesis. Proponents of this model assert that settlement of this region of western China came not from nomadic pastoralists of the steppe, but from sedentary, agriculturally based populations of the Oxus civilization (the Bactrian-Margiana archaeological complex (BMAC); Hiebert, [1994]) found west of Xinjiang in Uzbekistan (north Bactria), Afghanistan (south Bactria), and Turkmenistan (Margiana) (Barber, [1999]; Chen and Hiebert, [1995], p. 287). Proponents of this model emphasize the environmental similarities between the desert basins of eastern (Xinjiang) and western Central Asia (Bactria, Margiana) and maintain that the sophisticated irrigation techniques developed in the oases of Margiana and Bactria permitted the colonization of the river deltas and oases surrounding the north, east, and southern margins of the Täklamakan Desert (Barber, [1999]; Chen and Hiebert, [1995]).

Barber ([1999]) suggested that the archaeological record provides better evidence that the initial colonizers of the Tarim Basin were agriculturalists from Bactria than nomadic pastoralists from the Russian steppe. This evidence not only includes irrigation systems, evidence of western cultigens such as wheat, and bones of sheep and goats, but also evidence of carefully bundled bags containing Ephedra sp. found accompanying many Bronze Age Xinjiang burials. Use of ephedra is well-known in Oxus civilization urban centers, where Hiebert ([1994]) and Sarianidi ([1987], [1990], [1993a], [b], [1994]; see also Kussov, [1993]; Meyer-Melikyan, [1998]; Meyer-Melikyan and Avetov, [1998]) found evidence of specialized areas known as white rooms where it is believed a ritual drink, known as haoma in Iranian and soma in Indic, was consumed (but see Nyberg, [1995], p. 400; Parpola, [1995], p. 371). Ephedra does not grow on the Russo-Kazakh steppe, nor is it associated with either Afanasievo or Andronovo cultures (Barber, [1999], p. 165; but see Parpola, [1998], p. 126-127).

Barber ([1999]) suggested that the Bronze Age settlement of the Tarim Basin was a two-step process in which initial immigration came from the oases of Bactria and Margiana, and is represented by remains found at such southeastern sites as Qäwrighul. This was followed by a second wave of immigration soon after 1200 B.C. This time, immigrants came from Andronovo populations located to the northwest, and participants in this wave of immigration may be associated with the remains found at northern Tarim Basin sites such as Alwighul, Hami, Turfan, and Kucha. Barber ([1999]) claimed that evidence of these two separate waves of immigration is provided not only by dramatic differences in textile manufacture, but also by textual evidence from the first centuries A.D. which documents that inhabitants of the southern Tarim oases spoke Iranian languages (such as Saka and Sogdian), while those inhabiting the northern oases spoke Tocharian.

Measurements of the neurocranium and facial skeleton have been used for many years to provide an assessment of the degree of biological relatedness among samples of past and living populations. Although it is clear that these measurements actually provide assessment of an unknown combination of environmental and hereditary factors (Cavalli-Sforza and Bodmer, [1971]), and may be affected by masticatory mechanics (Carlson and Van Gerven, [1977]; Van Gerven, [1982]) and environmental variation (Beals, [1972]; Guglielmino-Matessi et al., [1979]), twin studies (Clark, [1956]; Lundstrom, [1954]; Nakata et al., [1974a]; Orczykowska-Swiatkowska and Lebioda, [1975]; Saunders et al., [1980]), familial studies (Devor, [1987]; Howells, [1966]; Nakata et al., [1974b]; Susanne, [1975], [1977]), and worldwide comparisons of craniometric variation revealed a moderate degree of genetic control (Susanne, [1975], [1977]), and demonstrated the utility of such variables for reconstructing patterns of biological interactions among populations (Howells, [1973], [1989]). Since all of the samples included in this study derive from either sedentary, agricultural communities or pastoralist populations who received regular supplies of agricultural produce, and from sites that differ little in latitude, a comparison of craniometric variation should suffer no systemic biases due to differences in masticatory stresses or natural selection for dramatically different environments (Hemphill, [1998], [1999]).

MATERIALS AND METHODS


Materials

Analyzed Bronze Age skeletal samples from Xinjiang are few and are underrepresented in the literature. Although many individuals included in the current study were the subject of craniometric comparisons by Han ([1990], [1994b], [1998], [2001]) and Mallory and Mair ([2000], p. 236-244), these comparisons are limited to contrasts of cranial and facial indices that provide no assessment of covariation among cranial indicators or of the significance of phenetic separation between samples. Further, results obtained from these contrasts are interpreted within a typologically grounded ethnogenetic paradigm that identifies human variation, even in individual cemeteries, as attributable to the presence of multiple physical types and subsequent interbreeding among them (Han, [1994b], [1998]; Mair, [1995], p. 291-292; Mallory and Mair, [2000], p. 235). Hence, reification of such fixed physical types encourages a static perspective of human populations that fails to accommodate the known evolutionary forces of natural selection, mutation, gene flow, and genetic drift. It is these processes that result in the inherent dynamism in the genetic foundation of all populations, as emphasized in modern population genetics (Falconer, [1981]; Hartl and Clark, [1997]; Hedrick, [2000]).

Through the efforts of Mallory ([1995]), measurements made in accordance with standards established by Martin ([1928]) by Han ([1998]) on crania recovered from three sites from the Tarim Basin of Xinjiang were made available to B.E.H. for further statistical analyses and interpretation. These sites include Qäwrighul, Alwighul, and Krorän (Fig. 1).

Discovered in 1979 and excavated by Wang ([1987], [1996]), Qäwrighul represents the most ancient Bronze Age cemetery in Xinjiang (Debaine-Francfort, [1988], p. 15-16; Han, [1994a]; Jettmar, [1992]; Kuchera, [1988]; Wang, [1987], [1996]). A series of five radiocarbon dates places this cemetery between 2300-1430 B.C. (An, [1998]; Chen and Hiebert, [1995]). The cemetery is located on the bank of the Könchi River, 70 km west of ancient Lake Lopnur along the eastern edge of the Täklamakan Desert (Kuzmina, [1998]; Mallory and Mair, [2000], p. 137). Excavation of the cemetery resulted in the recovery of 42 burials, each containing a single individual, from two different types of tombs (Wang, [1982], [1983]).

The first, designated as Qäwrighul type I tombs, account for 36 of the burials. All are shaft pit graves, and a minority of these graves feature wooden poles placed at east and west ends. The actual burial chamber was lined with small wooden boards, and the top of the chamber was sealed with animal skins, carpets, or a basket-shaped cover. The bodies were placed in a supine, extended position, with their heads to the east and their feet to the west. Qäwrighul type II tombs account for six burials, and they appear stratigraphically later than type I tombs. Type II tombs are identical to type I tombs with respect to the use of small wooden boards to line the burial pit and placement of the body, but differ by featuring an elaborate arrangement of wooden poles embedded in the ground surface. The most elaborate of these type II tombs featured seven concentric rings of wooden stakes that radiate outward in what some (Wang, [1983], [1984]) interpreted as a solar pattern from the center of the burial chamber, to encompass an area 50-60 m in diameter.

Individuals interred in the Qäwrighul cemetery wore no clothing apart from leather shoes and cloaks made of plain woven textiles fastened at the front with bone pins. Felt hats were placed on the head, and in several instances, small bags containing twigs of Ephedra sp. were found on their chests (Barber, [1999]; Chen and Hiebert, [1995], p. 253). No ceramics accompany any of the burials, but one bone and five wooden anthropomorphic figurines, some still wearing fragments of textiles, were recovered. A few jade beads and fragments of either copper or bronze represent the extent of additional burial accoutrements (Wang, [1982], [1983], [1984]). All 18 adult crania recovered from this site (11 male, 7 female) were identified by Han ([1986a], [1994a], [1998]) as proto-European and possessing closest affinities to crania recovered from southern Siberia, Kazakhstan, and the Volga River region of southern Russia (Han, [1998], p. 559-560; Mair, [1995], p. 290; Mallory and Mair, [2000], p. 241). Han ([2001], p. 234) suggested that the earliest (type I) burials were most closely related to the Afanasievo type, while later burials bore greater similarity to Andronovo populations.

The cemetery of Alwighul is located near the Ordos grasslands along the southern slopes of the Tian Shan mountains (Han, [1998]; Ma and Wang, [1994], p. 213). Featuring a series of radiocarbon dates that cluster between 800-200 B.C., the Alwighul cemetery dates to the Late Bronze Age, a period that is marked by an increase in the frequency of bronze objects and, in some regions of Xinjiang, the initial appearance of large bronze objects (An, [1998], p. 58; Ma and Wang, [1994], p. 213).

Three major inhumation types were observed at Alwighul. The first two feature distinctive burial chambers in which the perimeter walls are lined with pebbles. The early pebble graves (type I) feature multiple interments of at least 10-20 individuals piled atop one another in a supine position, with the head to the west and the feet to the east. Late pebble graves (type II) are similar in construction to those of type I, but feature an additional wooden bench supported by four pillars and contain the remains of only 1-2 individuals. Type III graves appear as heaps of stones at ground level beneath which there is a large vertical pit that leads to a wooden coffin-chamber.

In total, 58 adult crania (33 males, 25 females) were recovered from Alwighul. All of these remains were recovered from a single type I mass interment grave and were associated with a wide array of burial goods, including hand-made gray-red ceramic vessels decorated with triangular, net, whorl and pine-needle motifs painted in light black; a considerable number of bronze plates and knives; many strings of beads made from bone, shell, agate, and jade; and earrings of both bronze and gold (Ma and Wang, [1994], p. 213-215). The craniometric analysis by Han ([1994b], [1998], p. 560-561, [2001], p. 231-232) of these remains revealed the presence of at least three different racial types, including two different forms of Caucasoids (Pamir-Ferghana type and Eastern Mediterranean type) and a single form of Mongoloid. Han ([1990], [1998]) suggested that the apparent lack of conformity of some of the crania to these racial types indicated some degree of mixture between the two Caucasoid types as well as between Caucasoids and Mongoloids.

The cemetery of Krorän is located in one of the southeastern group of oases on the southern bank of the Könchi River near the outskirts of Loulan, the capital of the Shan-Shan state during the first century B.C. (Ma and Wang, [1994], p. 211). There is considerable controversy over the correct dates for the human remains from this cemetery. According to some researchers, the six adult crania recovered from this cemetery derive from the Western Han period (202 B.C.-A.D. 220; Han, [1998]; Mair, [1995]), but recent Chinese excavations suggest that these remains are associated with artifacts that span the period between the seventh to first centuries B.C. (Ma and Wang, [1994], p. 211). Nevertheless, a pair of radiocarbon dates obtained from the cemetery suggest that the Western Han period is most likely correct (Mallory and Mair, [2000], p. 335).

Individuals recovered from the cemetery at Krorän were interred in graves featuring a chamber of wooden planks within a shallow pit. The body was placed in an extended position, and wrapped in a woolen cloth. One extremely well-preserved individual was buried wearing hide boots and a peaked brown felt hat with bird feathers. In this case, the woolen cloth was gathered into a pouch on the upper chest and filled with fragments of Ephedra sp. (Ma and Wang, [1994], p. 211-212). The graves also contained wooden and stone figurines with long, round faces, but most funerary objects represent articles of everyday use and decorative ornaments. In earlier tombs, there is no pottery, and utensils are made of woven grass, wood, bone, or horn. Ornaments include beads made of bone, amber, agate, or jade, and were usually found encircling the neck or ankles. Groups of bone tubes about 10 cm long were sometimes linked together and worn around the waist. In addition to these locally produced items, many artifacts such as brocades, rough silk, silk floss, bronze mirrors, lacquerware, and wuzhu coins typical of the Han Dynasty of the middle-lower Yellow River were also recovered.

Of the six adult crania recovered, only one, a female, was identified by Han ([1986b], [1994a], [1998], p. 562-563) as Mongoloid. The remaining five individuals were identified by Han as possessing Eastern Mediterranean characteristics most similar to those found among sixth century B.C. Saka of the southern Pamirs (Ma and Wang, [1994], p. 212; Mair, [1995], p. 292).

Cranial series used to provide a comparative foundation for the Xinjiang remains encompass 22 samples, numbering 1271 individuals (665 males, 606 females) from the Russo-Kazakh steppe, southern China, Central Asia, Iran, and Indus Valley. Together, all 25 skeletal samples span a timeframe from 3500 B.C. to the present. Abbreviations, sample sizes, sources, and sample locations for all cranial samples are provided in Table 1 and Figure 1.

Eight cranial variables (two for the neurocranium, and six for the facial skeleton) of those defined by Martin ([1928]) provide the metrical basis for the current study (Table 2). While this small battery of measurements is far from representing the ideal array of variables for capturing the morphological complexity of the human cranium, increases in the number of variables do not automatically result in greater insight into the patterning of phenetic distances (Kowalski, [1972], p. 121; Oxnard, [1973], p. 39). In addition, when employing remains recovered from archaeological contexts, the often fragmentary nature of these remains leads to a concomitant decrease in sample size for every increase in variables considered. Quite simply, these eight variables represent the best combination of those measurements for which data were available for all of the samples compared and those adequately represented, due to differential preservation relative to other measurements, within each of these samples.


Table 2. Craniometric variables used to generate Mahalanobis generalized distances between samples

--------------------------------------------------------------------------------

Variable1

--------------------------------------------------------------------------------

Neurocranium
Maximum cranial length (GOL) 1
Maximum cranial breadth (BEB) 8
Facial skeleton
Upper facial height (NPH) 48
Nasal height (NH) 55
Nasal breadth (NB) 54
Orbital height (OH) 52
Orbital breadth (OB) 51
Bizygomatric breadth (BZB) 45


--------------------------------------------------------------------------------


1 Numbers of variables as defined by Martin ([1928])



When utilizing data collected by other workers, the degree of interobserver differences in assessment of these variables represents an important source of potential error that can compromise meaningful results. In this study, the degree of interobserver error between the authors and describers of comparative cranial series could be assessed for Tepe Hissar (Krogman, [1940]), Harappa (Cemetery R37), Cemetery H at Harappa (Gupta et al., [1962]), Altyn depe, and Geoksyur (Ginzberg and Trofimova, [1972]; Trofimova, [1961]; Kiiatkina, [1976], [1977], [1987]). Repeated-measures analysis of variance (Hemphill, [1998], [1999]; Hemphill et al., [1991]) indicated no significant measurement differences between different observers. Although the degree of interobserver error could not be directly assessed between the authors and samples obtained from Alekseev and Gochman ([1983]), these researchers incorporated measurements taken by Trofimova ([1961]) and Kiiatkina ([1976], [1977], [1987]) with those of Alexseev ([1961], [1967]) and Abduelivili ([1954], [1960], [1966]) and found no significant differences. Logically, then, there should be no significant differences between measurements taken by the author and those obtained by Alexseev ([1961], [1967]) and Abduelivili ([1954], [1960], [1966]) as well. Interobserver error could not be assessed for published measurements for individuals recovered from Shahr-i Sokhta, Timargarha, or Hainan (southern China).

Methods

The covariance matrix for each sample was obtained for males and females pooled together with listwise deletion. Although pairwise deletion permits greater effective sample sizes within each sample, listwise deletion was used to avoid systematic biases caused by overrepresentation and underrepresentation of individual variables (Wilkinson, [1990]). A pooled covariance matrix was obtained for all samples for which individual data were available and bias-adjusted to accommodate differences in sample size. Hence, those samples represented by group-level data only (Andronovo-Minusinsk (ADM), Afanasievo-Altai (AFA), Afanasievo-Minusinsk (AFM), Andonovo-Kazakhstan (AND), Karasuk-Minusinsk (KAM), Kara depe (KAR), Kokcha III (KOK), Samtavro (SAMB), and Tigrovaja Balka-Makoni Mor (TMM)) were not used to construct the pooled covariance matrix. Variable averages were calculated for both males and females. Sex-standardized group values for each variable were obtained by taking the average of male and female mean values for each sample (Table 3). The bias-adjusted pooled covariance matrix and sex-standardized group values were used to obtain Mahalanobis generalized distances (d2) between each pair of samples. The diagonal matrix of Mahalanobis d2 values is provided in Table 4. The significance of pairwise d2 distance contrasts for those samples in which individual data were available were assessed by means of F-tests, conducted according to the method of Konigsberg et al. ([1993]).


Table 3. Mean values of craniometric variables1

--------------------------------------------------------------------------------

GOL BEB NPH NH NB OH OB BZB

--------------------------------------------------------------------------------

Male sample
ADM 186.0 145.0 67.8 50.2 25.8 31.7 44.4 140.7
AFA 191.7 142.4 71.7 53.1 27.1 32.3 43.7 141.6
AFM 192.1 144.1 71.8 52.1 26.1 32.9 44.9 138.4
ALT 189.5 135.9 70.7 51.6 25.2 32.4 40.7 129.1
ALW 184.2 141.9 68.7 52.2 25.0 33.1 39.0 130.8
AND 186.4 140.4 69.2 51.5 25.4 32.9 42.1 131.8
CEMH 188.2 141.3 67.9 50.8 26.3 32.9 41.3 134.8
DJR 186.9 134.7 69.9 50.7 24.8 30.9 37.5 131.3
GKS 190.1 134.5 71.1 52.0 25.5 33.0 40.1 127.6
HAI 176.4 138.4 69.7 52.4 27.3 33.6 38.7 134.0
HAR 187.3 134.5 69.2 51.4 26.5 33.2 41.4 131.5
KAM 183.0 147.4 73.4 51.6 25.8 33.7 44.1 139.7
KAR 194.8 134.9 72.6 51.2 26.6 31.8 42.5 129.9
KOK 186.1 138.1 68.4 51.5 23.5 30.9 43.2 133.4
KRO 187.9 139.1 74.4 53.6 24.6 34.1 38.6 131.0
KUZ 190.9 138.9 68.7 49.6 26.4 30.9 39.9 134.0
MOL 185.6 138.1 69.4 51.5 25.1 31.8 38.3 126.6
QAW 183.0 137.9 66.5 50.9 26.2 31.6 40.5 136.2
SAMB 189.3 137.1 76.7 53.8 23.8 35.0 42.0 128.3
SAP 183.5 134.9 70.2 51.3 24.2 32.7 37.7 129.1
SHS 185.8 136.4 70.2 50.6 25.7 31.8 42.1 129.4
TH2 188.8 132.0 70.3 50.4 25.1 31.6 41.0 125.3
TH3 188.4 134.1 69.8 50.6 25.4 32.1 41.2 127.3
TMG 190.2 132.0 70.3 50.0 22.9 33.3 41.5 133.0
TMM 188.4 136.9 71.8 51.6 24.7 31.2 42.4 131.8
Female sample
ADM 175.9 140.6 67.5 48.9 23.9 33.0 42.5 127.4
AFA 182.6 138.2 64.8 47.8 25.4 31.1 46.5 129.8
AFM 180.4 135.8 67.4 49.5 25.6 33.1 44.3 131.8
ALT 181.4 135.1 67.3 49.6 24.2 32.7 38.9 121.4
ALW 173.9 135.3 64.9 49.4 24.2 32.0 37.5 124.4
AND 177.6 136.0 67.3 48.4 24.6 32.2 41.8 128.2
CEMH 179.2 132.4 62.7 46.0 24.4 33.3 39.9 119.5
DJR 184.7 134.0 69.5 50.2 25.6 33.0 38.5 123.9
GKS 185.8 132.9 69.8 50.9 25.2 32.9 39.0 123.4
HAI 170.6 135.0 65.4 49.3 26.0 32.8 37.6 125.6
HAR 180.9 132.1 66.2 48.3 24.2 34.1 40.6 123.9
KAM 173.2 143.4 68.0 48.4 24.6 32.9 42.1 131.8
KAR 183.0 132.1 67.5 47.9 24.7 32.0 41.6 123.8
KOK 177.6 136.4 66.2 49.4 23.8 31.8 41.2 128.5
KRO 181.0 133.5 69.0 50.6 24.1 33.2 36.8 129.5
KUZ 179.3 132.6 65.1 46.8 23.6 30.7 36.3 122.4
MOL 183.5 134.2 70.6 49.7 25.0 32.6 38.8 126.5
QAW 178.1 128.8 62.6 47.4 24.5 32.3 38.9 125.0
SAMB 180.1 136.5 69.5 48.4 23.4 32.1 39.3 122.5
SAP 181.5 134.1 67.5 49.2 24.8 33.0 37.2 124.4
SHS 179.1 133.3 67.6 50.0 24.5 31.9 40.7 122.7
TH2 178.3 132.1 67.6 48.3 23.7 33.6 38.7 118.7
TH3 179.4 131.8 66.1 48.3 23.9 31.7 39.6 120.2
TMG 180.2 130.9 66.6 48.1 22.9 33.1 40.0 122.3
TMM 179.8 133.2 69.2 49.2 23.7 31.9 40.9 124.5
Sex-standardized sample
ADM 181.0 142.8 67.7 49.6 24.9 32.4 43.5 134.1
AFA 187.3 140.3 68.3 50.5 26.3 31.7 45.1 135.7
AFM 186.3 140.0 69.6 50.8 25.9 33.0 44.6 135.1
ALT 185.5 135.5 69.0 50.6 24.7 32.6 39.8 125.3
AND 182.0 138.2 68.3 50.0 25.0 32.2 42.6 132.4
ALW 179.0 138.6 66.8 50.8 24.6 32.6 38.3 127.6
CEMH 183.7 136.8 65.3 48.4 25.3 33.1 40.6 127.1
DJR 185.8 134.3 69.7 50.5 25.2 32.0 38.0 127.6
GKS 190.1 134.5 71.1 52.0 25.5 33.0 40.1 127.6
HAI 173.5 136.7 67.5 50.8 26.7 33.2 38.1 129.8
HAR 184.1 133.3 67.7 49.9 25.4 33.6 41.0 127.7
KAM 178.1 145.4 70.7 50.0 25.2 33.3 43.1 135.8
KAR 188.9 133.5 70.1 49.6 25.7 31.9 42.1 126.9
KOK 181.9 137.3 67.3 50.5 23.7 31.4 42.2 131.0
KRO 184.4 136.3 71.7 52.1 24.4 33.7 37.7 130.2
KUZ 185.1 135.7 66.9 48.2 25.0 30.8 38.1 128.2
MOL 184.5 136.1 70.0 50.6 25.1 32.2 38.6 126.5
QAW 180.5 133.3 64.5 49.1 25.3 32.0 39.7 130.6
SAMB 184.7 136.8 73.1 51.1 23.6 33.6 40.7 125.4
SAP 182.5 134.5 68.8 50.2 24.5 32.9 37.5 126.7
SHS 182.5 134.8 68.9 50.3 25.1 31.8 41.4 126.0
TH2 183.5 132.1 69.0 49.4 24.4 32.6 39.9 122.0
TH3 183.9 133.0 69.9 49.4 24.7 31.9 40.4 123.8
TMG 185.2 131.5 68.4 49.1 22.9 33.2 40.8 127.7
TMM 184.1 135.1 70.5 50.4 24.2 31.6 41.7 128.2


--------------------------------------------------------------------------------


1 Abbreviations for craniometric variables are from Table 2. Abbreviations for samples are from Table 1.



Table 4. Matrix of Mahalanobis d2 generalized distances1

--------------------------------------------------------------------------------

ADM AFA AFM ALT ALW AND CEMH DJP GKS HAI HAR KAM KAR KOK KRO KUZ MOL QAW SAMB SAP SHS TH2 TH3 TMG TMM

--------------------------------------------------------------------------------

ADM 0.0
AFA 2.146 0.0
AFM 1.423 0.651 0.0
ALT 7.482 8.216 6.208 0.0
ALW 7.237 11.187 8.787 2.879 0.0
AND 1.094 1.900 0.953 3.870 4.961 0.0
CEMH 5.033 5.942 4.134 2.006 3.396 2.864 0.0
DJR 11.681 12.879 10.467 1.671 3.169 6.790 4.367 0.0
GKS 10.844 9.969 7.823 0.969 4.836 6.037 3.465 1.429 0.0
HAI 10.554 14.863 11.760 7.800 3.043 7.931 6.800 6.824 9.582 0.0
HAR 8.735 8.921 6.318 3.114 5.479 4.355 2.029 3.727 3.435 6.989 0.0
KAM 1.752 6.396 4.352 10.608 9.040 3.695 8.526 14.107 14.601 9.924 12.042 0.0
KAR 8.138 5.592 4.677 2.526 9.317 4.242 4.002 5.058 2.576 14.058 4.187 12.367 0.0
KOK 2.481 3.160 2.804 4.381 5.416 1.076 4.722 7.647 6.834 10.962 6.375 6.724 6.035 0.0
KRO 13.039 16.850 12.914 3.878 2.903 8.714 6.607 1.593 3.804 11.857 5.530 13.567 9.922 9.700 0.0
KUZ 9.531 10.922 9.408 2.519 3.537 5.703 3.748 0.952 3.154 7.896 4.176 12.471 5.182 6.472 3.445 0.0
MOL 9.209 10.987 8.567 0.759 2.327 5.304 3.399 0.428 1.296 6.218 3.979 11.105 4.189 6.280 1.954 1.342 0.0
QAW 6.914 7.599 6.263 4.270 3.055 3.435 2.581 4.154 5.180 4.693 2.069 10.941 7.091 4.269 5.783 3.558 4.544 0.0
SAMB 8.160 10.378 7.277 2.009 6.324 5.297 5.477 4.431 3.313 5.297 6.149 9.208 3.535 2.158 9.440 5.838 2.676 9.440 0.0
SAP 11.583 14.700 11.396 2.298 1.963 7.188 4.240 0.669 2.731 4.981 3.544 13.146 7.571 8.194 0.597 1.812 0.962 3.814 4.895 0.0
SHS 4.751 4.366 3.326 1.296 4.932 1.865 2.815 4.432 2.866 8.819 3.893 8.204 1.521 2.439 7.851 4.700 2.952 4.579 2.854 5.733 0.0
TH2 9.455 9.497 7.309 0.726 5.768 5.137 3.163 2.963 1.617 10.480 3.077 12.896 1.573 5.886 6.162 4.035 2.019 6.049 1.873 4.016 1.427 0.0
TH3 8.245 6.399 7.828 0.608 6.390 4.106 2.410 3.045 1.656 12.592 2.941 14.296 1.152 2.811 10.382 3.426 2.122 4.669 5.995 4.387 0.586 0.384 0.0
TMG 8.486 9.195 6.706 3.044 6.898 4.411 4.033 4.462 3.696 12.528 2.211 12.344 4.028 4.318 5.786 4.666 4.459 4.688 3.790 4.593 3.940 2.828 5.276 0.0
TMM 4.901 4.800 3.713 2.106 6.273 2.000 4.564 4.608 3.535 11.403 4.880 8.076 1.675 6.790 7.567 4.565 3.367 5.782 2.403 6.126 0.754 2.121 0.689 2.863 0.0


--------------------------------------------------------------------------------


1 Abbreviations for samples are from Table 1.



The diagonal matrix of Mahalanobis d2 values was used as input for cluster analyses. Different associating algorithms were used to provide two perspectives on the patterning of intersample phenetic affinities. These associating algorithms include the weighted pair average linkage method (WPGMA) (Sneath and Sokol, [1973]) and the neighbor-joining method (Felsenstein, [1989]; Saitou and Nei, [1987]). The cophenetic correlation coefficient, rcs (Sneath and Sokol, [1973]), was computed with the NTSYS-pc statistical package to measure the degree of correspondence between the obtained phenogram from WPGMA cluster analysis and the original resemblance matrix.

The diagonal matrix of Mahalanobis d2 values was used as input for nonmetric multidimensional scaling, to provide a third perspective on the patterning of intersample affinities. The coefficient of alienation of Guttman ([1968]) was used to calculate distances between individual points. The goodness of fit obtained by multidimensional scaling was assessed through calculation of the degree of stress through 100 iterations. Multidimensional scaling was accomplished with the SYSTAT statistical package (Wilkinson, [1990]). Results obtained were ordinated in three-dimensional space, and a minimum spanning tree (Hartigan, [1975]) was imposed on the array of data points to ease interpretation of the patterning of intersample associations.

Principal coordinates analysis was used to provide a fourth perspective on intersample craniometric variation (Hair et al., [1971]). The symmetric matrix of Mahalanobis d2 values was double-centered prior to entry into NTSYS-pc statistical software (Rohlf, [2000]). The first three principal coordinate axes were retained, group scores were calculated along these axes, and ordinated into three-dimensional space. As with results from multidimensional scaling, a minimum spanning tree was imposed on the array of principal coordinate scores to ease interpretation of intersample associations. The cophenetic correlation coefficient was computed to assess the goodness of fit of the obtained eigenvectors with the matrix of Mahalanobis d2 values. This latter step is especially important, because the cophenetic correlation coefficient provides more information on the patterning of relative phenetic distances among samples than the absolute distance (as indicated by the percentage of total variation explained by the first three eigenvectors) (Rohlf, [1972], [2000]), and it is the patterning of these relative distances that is most useful for understanding processes of past population interactions.

As a final step in assessment of the nature of intersample craniometric variation, spatial distance and temporal distance matrices were computed among all sample pairs. Congruence between the Mahalanobis d2 matrix and these latter two matrices was assessed by means of the Mantel test (Mantel, [1967]) and Mantel correlation coefficient (Smouse et al., [1986]). These procedures provide a test to determine if differences between samples may simply be a product of geographical propinquity or differences in antiquity. The significance of these associations was obtained through 1,000 permutations at random by rows and columns.

RESULTS


The bias-adjusted matrix of Mahalanobis d2 values was calculated according to the procedures outlined above (Table 4). F-tests for those 16 samples in which individual data are available (Table 5) reveal that the majority of d2 values between samples are significant (98/120; 81.7%). Of the 98 pairwise contrasts exhibiting a significant difference, 7 (7.1%) are significant at the 0.05 level, while 91 (92.9%) are significant at the 0.01 level.


Table 5. F-tests and probability values of pairwise Mahalanobis d2 generalized distances1

--------------------------------------------------------------------------------

ALT ALW CEMH DJR GKS HAI HAR KRO KUZ MOL QAW SAP SHS TH2 TH3 TMG

--------------------------------------------------------------------------------

ALT 0.000 0.003 0.000 0.001 0.000 0.000 0.067 0.000 0.081 0.000 0.000 0.000 0.425 0.004 0.000
ALW 9.423 0.000 0.000 0.000 0.000 0.000 0.197 0.000 0.000 0.000 0.001 0.000 0.000 0.000 0.000
CEMH 3.254 5.378 0.000 0.000 0.000 0.007 0.086 0.004 0.001 0.061 0.001 0.000 0.026 0.000 0.008
DJR 4.512 8.280 6.102 0.000 0.000 0.000 0.660 0.228 0.554 0.000 0.358 0.000 0.002 0.000 0.000
GKS 3.443 16.507 5.769 4.002 0.000 0.000 0.066 0.000 0.003 0.000 0.000 0.000 0.027 0.000 0.000
HAI 31.612 11.799 12.222 21.312 40.812 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000
HAR 9.300 15.797 3.031 9.003 10.669 24.431 0.018 0.000 0.000 0.014 0.000 0.000 0.014 0.000 0.009
KRO 1.975 1.457 2.229 0.735 1.969 6.421 2.672 0.359 0.564 0.187 0.980 0.000 0.162 0.000 0.187
KUZ 4.251 5.825 3.701 1.380 5.468 14.805 6.481 1.198 0.102 0.012 0.061 0.000 0.006 0.000 0.003
MOL 1.839 5.467 4.406 0.861 3.247 17.226 8.291 0.854 1.802 0.000 0.183 0.000 0.038 0.000 0.000
QAW 6.043 4.228 2.208 5.113 7.511 7.301 2.712 1.739 3.139 5.209 0.000 0.000 0.002 0.000 0.008
SAP 4.585 3.812 4.768 1.130 5.613 11.172 6.447 0.233 2.107 1.497 3.816 0.000 0.002 0.000 0.001
SHS 4.919 17.950 4.874 13.115 11.404 40.557 12.834 4.160 8.480 7.782 6.883 12.318 0.041 0.001 0.000
TH2 1.027 7.983 2.706 3.647 2.345 16.304 4.033 1.853 3.560 2.314 4.638 4.019 2.145 0.733 0.070
TH3 2.948 29.446 4.768 10.984 8.508 77.572 12.042 5.928 7.091 6.702 7.929 11.006 3.294 0.652 0.000
TMG 4.308 9.547 3.450 5.492 5.359 19.491 2.898 1.740 4.117 5.111 3.594 4.596 5.922 2.168 8.960


--------------------------------------------------------------------------------


1 F-values are below diagonal. Probability values (P-values) are above diagonal. Abbreviations for samples are from Table 1. Only those samples with individual data are included. This resulted in elimination of Andronovo-Minusinsk (ADM), Afanasievo-Altai (AFA), Afanasievo-Minusinsk (AFM), Andronovo-Kazakhstan (AND), Karasuk-Minusinsk (KAM), Kara depe (KAR), Kokcha III (KOK), Samtavro (SAMB), and Tigrovija Balka-Makoni Mor (TMM).



WPGMA cluster analysis

The dendrogram obtained by means of the WPGMA associating algorithm (Fig. 2) indicates that the Hainan sample (HAI) from south China represents the most divergent of all samples considered. The major division among remaining samples occurs between steppe samples (except SAMB and TMM) and all other samples. Bactrian samples are segregated from samples obtained from Iran, Turkmenistan, and the Indus Valley. The two later samples from Xinjiang (ALW and KRO) are associated with the Bactrian samples. Krorän (KRO) features a very close affinity with the earliest of the Bactrian samples (SAP), while Alwighul (ALW) joins the later Bactrian samples (DJR, KUZ, and MOL) at a more distant remove. Indus Valley samples are identified as sharing slightly closer affinity to samples from Iran and Turkmenistan than to Bactrian samples. Affinities among Indus Valley samples are rather diffuse. In fact, the early sample from western China, Qäwrighul (QAW), is identified as possessing closer affinities to the two samples from Harappa (HAR and CEMH) than exhibited by the third Indus Valley sample, Timargarha (TMG). The remaining samples form a loose cluster composed of sedentary agricultural groups from Iran (TH2, TH3, and SHS) and Turkmenistan (GKS, ALT, and KAR), as well as steppe samples from the Caucasus (SAMB) and Tajikistan (TMM). Affinities are closest between the two northern Iranian samples from Tepe Hissar (TH2 and TH3), followed by the latest sample from Turkmenistan (ALT). The eastern Iranian sample (SHS) and the steppe Bronze Age sample from Tajikistan (TMM) exhibit close affinities to one another, and moderate affinities to the two samples from Tepe Hissar (TH2 and TH3) and Altyn depe (ALT). The two earlier samples from Turkmenistan (KAR and GKS) join these samples at a more distant remove. The steppe Bronze Age sample from the Caucasus (SAMB) exhibits a peripheral association with these Iranian, Turkmenian, and Tajik samples.


Figure 2. WPGMA cluster analysis of Mahalanobis d2 values. Branch points are Euclidean distances. Sample abbreviations from Table 1.
[Normal View 13K | Magnified View 31K]


Neighbor-joining cluster analysis

Neighbor-joining cluster analysis (Fig. 3) provides a different representation of the distance matrix than that provided by WPGMA cluster analysis, because it is an unrooted tree whose branches have different lengths. Long branch lengths may be interpreted as an indicator of a large degree of morphological separation, while short branch lengths are indicative of a small degree of morphological separation between samples.


Figure 3. Neighbor-joining tree based on Mahalanobis d2 values. Sample abbreviations from Table 1.
[Normal View 10K | Magnified View 24K]


The neighbor-joining tree provides an array of intersample associations that are largely in agreement with those depicted by WPGMA (Fig. 2). Once again, the south China sample from Hainan (HAI) is identified as the most divergent of all samples considered. All three western Chinese samples exhibit closest affinities to samples from Bactria. The two later western Chinese samples, Krorän (KRO) and Alwighul (ALW), feature closest affinities with the earliest of the Bactrian samples, Sapalli (SAP), while the earliest western Chinese sample, Qäwrighul (QAW), is identified as possessing closer affinities to later Bactrian samples (DJR, KUZ, and MOL).

Turkmenian samples from Geoksyur (GKS) and Altyn depe (ALT) serve as a phenetic link between Indus Valley samples (HAR, TMG, and CEMH) that feature the closest affinities to one another. In a departure from the results obtained by WPGMA analysis, the sample from Kara depe (KAR) occupies a unique position among Turkmenian samples by exhibiting much closer affinities to Iranian samples (especially TH3 and SHS) than to samples from Bactria.

Andronovo and Afanasievo steppe samples occupy the left side of the array. Steppe samples from Tigrovaja Balka/Makoni Mor (TMM), Samtavro (SAMB), and Kokcha III (KOK) occupy an intermediate phenetic position; these samples, especially TMM, manifest some affinities to Iranian samples. Afanasievo samples (AFA and AFM) are identified as possessing the closest affinities to one another, and exhibit affinities to the Andronovo samples (AND and ADM) as well. The Karasuk sample from the Minusinsk region (KAM) stands apart as the most divergent of the steppe samples considered.

Cophenetic correlation coefficients

The cophenetic correlation coefficient for the degree of correspondence between the phenogram obtained by WPGMA cluster analysis and the bias-adjusted matrix of Mahalanobis d2 values is low (rcs = 0.496). This suggests that a fair amount of distortion is encountered when attempting to arrange intersample differences in craniometric variation in a hierarchical fashion through cluster analysis (Rohlf, [2000]). Sneath and Sokol ([1973]) recommended that alternative methods of data reduction be used in cases where cophenetic correlations indicate that a fair amount of distortion of the original data matrix is incurred by hierarchical cluster analyses. Specifically, Sneath and Sokol ([1973]) recommended the use of multidimensional scaling and principal coordinates analysis.

Multidimensional scaling

Multidimensional scaling of the bias-adjusted diagonal matrix of d2 values into three dimensions with the coefficient of alienation of Guttman ([1968]) is accomplished with a stress value of 0.097 after 100 iterations. This value falls within acceptable limits, and indicates that multidimensional scaling of these data into three dimensions provides an array of intersample associations only mildly affected by distortion. A plot of multidimensionally scaled values, with a minimum spanning tree imposed between individual data points, is provided in Figure 4.


Figure 4. Minimally spanned plot of sample values for first three multidimensionally scaled dimensions. Sample abbreviations from Table 1. Xinjiang samples (QAW, ALW, and KRO) and Chinese sample from Hainan (HAI) are represented by asterisks; North Bactrian samples, by stars; Iranian samples, by pentagons; Turkmenian, Caucasus, and Tajik samples, by triangles; Indus Valley samples, by circles; and Russo-Kazakh samples, by squares.
[Normal View 18K | Magnified View 43K]


An examination of this array confirms the patterns of interregional affinities identified by neighbor-joining cluster analysis (Fig. 3). Hainan (HAI) reflects the most divergent sample. The two later western Chinese samples, Krorän (KRO) and Alwighul (ALW), feature the closest affinities to Sapalli (SAP), the earliest of the Bactrian samples. Two of the samples from Turkmenistan (Altyn depe (ALT) and Geoksyur (GKS)) span the phenetic space between Iranian samples and Bactrian samples, with Geoksyur exhibiting closer phenetic affinities to Bactrians (especially the latest sample, Molali (MOL)), while Altyn depe shares closer phenetic affinities to Iranians. The steppe Bronze Age sample from the Caucasus (SAMB) represents a phenetic outlier to all other samples, exhibiting only a very distant affinity to the sample from Altyn depe. Indus Valley samples share rather close affinities to one another but are strongly segregated from all other samples, except the early western Chinese sample from Qäwrighul (QAW).

All steppe Bronze Age samples, except Samtavro (SAMB), are found on the left side of the array. Intersample affinities among the Karasuk (KAM), Afanasievo (AFA and AFM), and Andronovo (AND and ADM) samples are relatively close. However, steppe Bronze Age samples from Turkmenistan (KOK) and Tajikistan (TMM), while exhibiting distant affinities to other steppe Bronze Age samples, appear more closely aligned to sedentary agricultural samples from Turkmenistan (Kara depe (KAR)) and, to a lesser degree, eastern Iran (Shahr-i Sokhta (SHS)).

Principal coordinates analysis

A principal coordinates analysis of the double-centered Mahalanobis d2 matrix yields three coordinate axes that combine to explain 89.9% of the total variance. Comparison of the eigenvector matrix with the d2 matrix yields a cophenetic correlation coefficient whose value (rcs = 0.948) indicates that the first three eigenvectors provide an excellent fit of the data (Rohlf, [2000]). An ordination of group scores for the first three coordinate axes is provided in Figure 5, and a minimum spanning tree was imposed on this array to clarify associations between samples.


Figure 5. Minimally spanned ordination of sample scores for first three principal coordinate axes. Sample abbreviations from Table 1. Xinjiang samples (QAW, ALW, and KRO) and Chinese sample from Hainan (HAI) are represented by asterisks; North Bactrian samples, by stars; Iranian samples, by pentagons; Turkmenian, Caucasus, and Tajik samples, by triangles; Indus Valley samples, by circles; and Russo-Kazakh samples, by squares.
[Normal View 18K | Magnified View 43K]


The pattern of intersample variation provided by this analysis confirms many of the major features previously identified by neighbor-joining cluster analysis (Fig. 3) and multidimensional scaling (Fig. 4). Once again, the two later western Chinese samples, Krorän (KRO) and Alwighul (ALW), exhibit the closest affinities to the earliest Bactrian sample, Sapalli (SAP). Bactrian samples (SAP, DJR, KUZ, and MOL) exhibit the closest affinities to one another. The two Turkmenian samples from Geoksyur and Altyn depe occupy an intermediate phenetic position between Bactrians and northern Iranians, in which the former (GKS) shares the closest affinities with the latest Bactrian sample (MOL), while the latter (ALT) shares the closest affinities with the earlier northern Iranian sample (TH2). Indus Valley samples (HAR, CEMH, and TMG) are located in the lower left of this array and, once again, the earliest western Chinese sample, Qäwrighul (QAW), is identified as possessing closer affinities to Indus Valley samples than to samples from any other region. Standing somewhat in contrast to results obtained by other analyses, principal coordinates analysis identifies an especially close affinity between the Late Bronze-Early Iron Age sample from the Swat Valley of Pakistan (TMG) and the early northern Iranian sample (TH2). As with other analyses, this array also indicates that the Turkmenian sample from Kara depe (KAR) is strongly separated from other sedentary Turkmenistan samples, but unlike other analyses, principal coordinates analysis indicates that this sample possesses no close affinities with any of the other samples considered.

All steppe Bronze Age samples, regardless of geographic location, occupy the right side of this array. In agreement with other analyses, the Afanasievo samples (AFA and AFM) exhibit the closest affinities to one another. However, unlike other analyses, the patterning of affinities yielded by principal coordinates analysis suggests a moderate degree of distinctiveness between Afanasievo samples and Andronovo samples (AND and ADM). The sample from Tajikistan (TMM) is identified as the steppe sample with closest affinities to nonsteppe samples in general, and with the eastern Iranian sample from Shahr-i Sokhta (SHS) in particular.

Mantel tests

The normalized Mantel statistic, which is equivalent to a correlation coefficient (r), obtained between the Mahalanobis d2 matrix and the matrix of chronological differences between samples, is 0.294. The permutational probability to observe a higher or equal correlation based on 1,000 permutations is P = 0.827. This value suggests that differences in antiquity, ranging from 3500 B.C. to the present, do not contribute significantly to the patterning of craniometric differentiation among these samples. Given ample opportunity for responses to changes in selection pressures due to exposure to agricultural diets and more sophisticated food preparation techniques during passage of the more than five millennia encompassed by these comparative samples, the absence of any chronological effect on the patterning of phenetic distances suggests that either these selective pressures led to an alteration of cranio-gnathic dimensions prior to 3500 B.C. or that this battery of measurements is not affected in any appreciable way by changes in masticatory pressures.

A comparison between the Mahalanobis d2 matrix and the matrix of geographical distances between samples yields a correlation coefficient of r = 0.560. The permutational probability to observe a higher or equal correlation is also not significant, with a value of P = 0.330. Contrary to standard expectations of isolation by distance (Barbujani, [1987]; Cavalli-Sforza et al., [1994]; Fix, [1999]; Kimura and Weiss, [1964]; Malécot, [1967]; Morton et al., [1982]; Piazza and Menozzi, [1983]; Sokol et al., [1986]; Sokal and Wartenburg, [1983]; Wright, [1943], [1946], [1951]), these results indicate that the amount of geographic distance between individual samples does not provide an important contributing factor behind the patterning of craniometric differentiation among these samples.

DISCUSSION


Numerous specific hypotheses have been advanced to account for the initial appearance of Bronze Age populations found at a series of oases skirting the margins of the Täklamakan Desert within the Tarim Basin of Xinjiang, western China, during the final two millennia B.C. These individual hypotheses can be grouped into two general models, and the model currently favored by a small majority of archaeologists working in Central Asia and western China is the steppe hypothesis (Han, [1998]; Kuzmina, [1998]; Mair, [1995]; Mallory and Mair, [2000]; Parpola, [1998]). As a general model, this hypothesis holds that for reasons as yet unknown, Afanasievo-related steppe populations from the north and northwest began to emigrate southward, either directly into the Tarim Basin (Kuzmina, [1998]), or subsequent to contact with more settled agricultural populations in Central Asia (Mallory and Mair, [2000]). These immigrants are thought to be represented by the human remains recovered from such early Bronze Age sites as Qäwrighul (Kuzmina, [1998]; Mallory, [1995]; Mallory and Mair, [2000]). Later, beginning around 1200 B.C., the archaeological record of Xinjiang reveals a series of changes in textile manufacture and clothing design. Although there are always problems in equating changes in material culture with population movements, proponents of the steppe hypothesis suggest that these changes signal the appearance of a second wave of immigration to the Tarim Basin from the Russo-Kazakh steppe. In this latter case, these immigrants are held to be members of the widespread Andronovo culture complex that appears throughout the south Russian steppe, Kazakhstan, and western Central Asia during the middle of the second millennium B.C.

Given both the archaeological (Kuzmina, [1998]) and craniometric (Han, [1998]) arguments, remains recovered from the earliest sample, Qäwrighul, should exhibit broad phenetic similarities to Afanasievo samples from the Altai and Minusinsk, while later Tarim Basin samples from Xinjiang (Alwighul and Krorän) should exhibit closer phenetic affinities to the later Andronovo samples from Kazakhstan and Minusinsk. Since most proponents of the steppe hypothesis envision the immigration of Andronovo populations as limited to several centuries spanning the end of the second and the beginning of the first millennia B.C., late Bronze Age populations of the Tarim Basin are expected to be sequentially more divergent from their Andronovo source populations over time due to genetic drift. Hence, the Alwighul sample (ca. 800-200 B.C.), if it truly predates the sample from Krorän (ca. 202 B.C.-A.D. 220), should exhibit closer affinities to Andronovo samples, while the Krorän sample should be more divergent. Bactrian, Iranian, and Indus Valley populations are thought to have played little to no role in the origins of Bronze Age inhabitants of the Tarim Basin of Xinjiang; therefore, samples from these latter regions should be markedly divergent phenetically.

Most advocates of the steppe hypothesis recognize an East Asian contribution to the Xinjiang gene pool subsequent to that provided by Afanasievo-related steppe populations, but contemporaneous with that provided by the later influx of Andronovo steppe populations (Han, [1998]; Mallory and Mair, [2000]). Han ([1998], [2001], p. 237-239) maintained that this influence is largely restricted to such eastern Tarim Basin samples as Yanbulaq (Han, [1990]), but identified seven of the crania from the earlier graves at Alwighul (Han, [1998]) and a single female from Krorän (Han, [1986b]) as Mongoloid. If Han ([1998]) is correct that East Asian populations contributed to eastern Tarim Basin populations in general and account for a minority of individuals encompassed by Alwighul (12%) and Krorän (17%) samples, these later Tarim Basin samples should be marked by a reduction in phenetic distance from the Han Chinese sample (HAI) relative to that found for the earlier sample from Qäwrighul.

The results of all analyses provide abundant evidence in support of a migration of pastoralist populations across the Russo-Kazakh steppe. This is reflected by the degree of phenetic cohesion found among steppe samples, regardless of the geographic distances that separate them. Further, once steppe samples are removed from consideration, a Mantel test of the correlation between the Mahalanobis d2 matrix and the matrix of geographical distances between samples yields a highly significant (P = 0.001) correlation coefficient (r = 0.871). Thus, the apparent departure of the patterning of phenetic distance from expectations of an isolation-by-distance model appears to be due to a spread of steppe pastoralist populations across an enormous distance from the trans-Ural region in the west to the Minusinsk Basin in the east. In this case, the close similarities in archaeological assemblages attributed to Andronovo and Afanasievo archaeological horizons do appear to document an eastward and southward population expansion.

Nevertheless, there is no support for the hypothesis that steppe populations contributed significantly to Bronze Age populations of the Tarim Basin. Despite numerous similarities between Afanasievo and Andronovo artifacts and Bronze Age artifacts from Xinjiang (Bunker, [1998]; Chen and Hiebert, [1995]; Kuzmina, [1998]; Mei and Shell, [1998]; Peng, [1998]), all analyses of phenetic relationships consistently reveal a profound phenetic separation between steppe samples and the samples from the Tarim Basin (Qäwrighul, Alwighul, and Krorän). Further, neither of the later Tarim Basin samples from Alwighul or Krorän appears phenetically closer to the Han Chinese sample from Hainan, thereby indicating an absence of East Asian influence in these samples.

The second model offered to account for the origins of the Bronze Age inhabitants of the Tarim Basin is the Bactrian oasis hypothesis (Askarov, [1973], [1981], [1988]; Barber, [1999]). Proponents of this model emphasize the similarity in environmental conditions between the oases skirting the Täklamakan Desert and those found in Bactria and Margiana to the west. Proponents of the Bactrian oasis model argue that the very skills developed by the founders and occupants of the urban centers of the Oxus civilization (irrigation agriculture, development of extensive trade networks between locally resource-impoverished oases, and domestication of sheep and goats) are exactly those that accompany the initial appearance of Bronze Age populations in the Tarim Basin (Barber, [1999]; Chen and Hiebert, [1995]). To explain the changes in textile manufacture, clothing styles, and metallurgical technology found in the Tarim Basin beginning around 1200 B.C., some proponents of the oasis model concur with the steppe hypothesis and envisage a second influx of colonists from the steppelands (Barber, [1999])

If this model is true, the earliest Tarim Basin populations, such as Qäwrighul, should possess close similarities to samples from Bactria. Given that the nature of this interaction is thought to have been unidirectional, from Bactria to the Tarim Basin, and limited in duration, phenetic affinities between populations of the two regions should initially be close and then progressively decrease over time as the two gene pools became increasingly distinct due to genetic drift. Tarim Basin inhabitants that postdate 1200 B.C. should represent the impact of Andronovo immigrants from the Russo-Kazakh steppe. Later Bronze Age inhabitants of the Tarim Basin should be marked by a reduction in phenetic distance to steppe populations in general, and to Andronovo samples in particular. Phenetic affinities possessed by the samples from Alwighul and Krorän should be markedly closer to those of steppe samples than those possessed by the earlier sample from Qäwrighul.

The results obtained offer little support for the Bactrian oasis hypothesis. While Tarim Basin samples do exhibit closer affinities to samples from the urban centers of the Oxus civilization than to steppe samples, three aspects of the patterning of interregional phenetic affinities run counter to the expectations of this model. First, rather than identifying that closest affinities occur between early Tarim Basin (Qäwrighul) and earlier or contemporaneous Oxus civilization samples that antedate or are contemporaneous (Sapalli and Djarkutan), the closest affinities actually occur between the earliest of the Oxus civilization samples, Sapalli, and the latest of the Tarim Basin samples, Krorän, followed by the late sample from Alwighul. Second, none of the Tarim Basin samples, not even those that postdate 1200 B.C., exhibit any phenetic affinities to any of the steppe samples included in this analysis. Third, while neighbor-joining cluster analysis (Fig. 3) suggests a distant affinity between Qärwighul, the earliest Tarim Basin sample, and the Oxus civilization samples, this is not confirmed by any other analysis. While a case could be made for greater involvement of Bactrian oasis peoples in the population history of the Tarim Basin during the Bronze Age than by steppe populations, the nature of this involvement is not predicted by the Bactrian oasis hypothesis.

The results fail to demonstrate close phenetic affinities between the early inhabitants of Qäwrighul and any of the proposed sources for immigrants to the Tarim Basin. The absence of close affinities to outside populations renders it unlikely that the human remains recovered from Qäwrighul represent the unadmixed remains of colonists from the Afanasievo or Andronovo cultures of the steppelands, or inhabitants of the urban centers of the Oxus civilization of Bactria.

Three alternative possibilities remain once simple large-scale emigration from a known source population is ruled out. First, the human remains from Qäwrighul may be those of a local, indigenous population from the Tarim Basin itself or the surrounding highlands. Second, the human remains from Qäwrighul may be the product of emigration from a source area other than the Russo-Kazakh steppelands or Oxus civilization urban centers. Third, the human remains from Qäwrighul may derive from one of the suggested source areas, but the separation of this emigrant population from the host population involved fewer founding individuals or occurred earlier than currently thought by proponents of either the steppe or Bactrian oasis hypotheses. Under such conditions, a founder effect, coupled with subsequent genetic drift, may have resulted in amelioration of phenetic similarities detectable through craniometric analyses.

The first alternative is certainly possible, for advocates of both the steppe and Bactrian oasis hypotheses admit the presence, albeit scarce, of a series of Neolithic sites in the basin that antedate the initial Bronze Age (An, [1992a]; Bergman, [1939]; Chang, [1977]; Chen and Hiebert, [1995]; Chen, [1990]; IAX, [1989]; Mallory, [1995]; Olsen et al., [1988]; Teilhard de Chardin and Young, [1932]; Wang, [1985]; Xu, [1995]). The recovery of microliths on the surface in association with painted ceramics at Yi'erkabake and with plain red wares at Xinge'er led Wang ([1985]) to suggest that the transition from the Neolithic to the Bronze Age, in at least the eastern portion of the Tarim Basin, was one of cultural continuity rather than an unprecedented introduction of a foreign Bronze Age cultural complex.

If a resident population, regardless of ultimate derivation, was present in the Tarim Basin at the emergence of the Bronze Age, initiation of the Bronze Age in this region may have been the product of a more subtle interaction between this local population and groups from adjacent regions. From a biological perspective, such interactions may have involved limited levels of unidirectional or bidirectional gene flow. Under such conditions, the biological impact of emigrants from outside would likely be muted relative to the impact of outright wholesale colonization of an essentially unpopulated, or minimally populated, region (Ayala, [1982]; Bodmer and Cavalli-Sforza, [1975]; Cavalli-Sforza et al., [1994]; Cummings, [1997]; Fix, [1999]; Weiss, [1988]; Wijsman and Cavalli-Sforza, [1984]). Unless by some fortuitous circumstance the remains recovered from Qäwrighul are those of some foreign entrepôt, these remains should reflect the impact of such gene flow by moderate to low affinities to those adjacent, non-Tarim Basin populations with whom they were in contact.

The results, however, fail to demonstrate even a low-level phenetic affinity between Qäwrighul and either steppe samples or samples from Oxus civilization urban centers. Not only is there no evidence for substantial immigration into the Tarim Basin by populations of these two adjacent regions; it also appears unlikely that either steppe populations or Oxus civilization populations served as a source of any significant gene flow commensurate with the appearance of the Bronze Age occupation of Qäwrighul. Given the paucity of contemporaneous skeletal remains from other parts of the Tarim Basin, the presence of immigrants from either the steppelands or the urban centers of the Oxus civilization cannot be definitively ruled out.

The second alternative explanation to account for the human remains from Qäwrighul is that they are the product of emigration from a source area other than the Russo-Kazakh steppelands or Oxus civilization urban centers. While the results obtained indicate that there is no evidence that gene flow from either steppe or Oxus civilization populations led to the establishment of the Qäwrighul population, all analyses, except neighbor-joining cluster analysis (Fig. 3), disclose a low-level affinity between the Qäwrighul and Indus Valley samples. Such affinities could be indicative of some early interaction between the populations of these two regions. The implications of such early interaction are potentially profound.

In a reversal of mainstream thought on a western Asian homeland (Urheimat) and eastward dispersal of Indo-European languages into Central Asia and India (Burrow, [1973]; Gamkrelidze and Ivanov, [1990]; Mallory, [1989]; Renfrew, [1988]), there is a body of scholars who have vigorously argued for an Indo-European homeland in the Indus Valley of India and Pakistan (surveyed at length in Bryant, [2001]), or that Indo-European languages disseminated from a locus somewhere in the vicinity of ancient Bactria-Sogdiana (Nichols, [1997], p. 137; see also Sargent, [1997]). If true, the dispersal of these Indo-European languages may have been accompanied by immigration and some gene flow from the Indus Valley homeland to the various historical seats of the Indo-European languages. In this way, Tocharian languages found in the Tarim Basin would be attributed to the influx of populations from Bactria whose ultimate derivation may be traced to the Indus Valley of India and Pakistan.

The results of this study offer little support for such a scenario. The problems are threefold. First, WPGMA cluster analysis (Fig. 2) identifies Qäwrighul as possessing closer affinities to the two samples from Harappa than are possessed by the Late Bronze-Early Iron Age sample from Timargarha. It is difficult to see any archaeological support for such a connection, as there are no material artifacts of mature Harappan or even late Harappan attribution found at Qäwrighul (Chen and Hiebert, [1995]; Wang, [1982], [1983]). Likewise, there are no artifacts reflective of Tarim Basin derivation at either mature Harappan (HAR) or late Harappan (CEMH) levels at Harappa (Allchin and Allchin, [1982]; Kenoyer, [1998]). Second, if Indo-European-speaking populations entered the Tarim Basin from Bactria, Bactrian populations should also show evidence of gene flow from the Indus Valley. None of the analyses presented here or in previous assessments (Hemphill, [1998], [1999]; Hemphill et al., [1998]) provide any evidence of significant interaction between Bactrian and Indus Valley populations prior to the latter half of the first millennium B.C. Finally, greater insight into the relationship between Indus Valley and Tarim Basin populations is provided by multidimensional scaling (Fig. 4) and principal coordinate analysis (Fig. 5). Both merely identify Qäwrighul as occupying a peripheral and opposite phenetic position to whatever Indus Valley sample is least separated from other regional samples. Such positioning is best interpreted as evidence of outlier status to all samples considered in this multidimensional array, rather than of any peripheral association to Indus Valley samples per se.

With neither biological nor archaeological support, there is no compelling evidence to uphold the idea that Indo-European languages were introduced into the Tarim Basin from populations emigrating from the Indus Valley commensurate with the initiation of the Xinjiang Bronze Age. Given the information currently available, it is most likely that the early Tarim Basin sample from Qäwrighul occupies an isolated phenetic position, because this sample represents a population of western China to which none of the potential regional contributors represented in this analysis (Russo-Kazakh steppe, Bactria, Indus Valley, and south China) contributed substantially.

Yet while the cranial series from Qäwrighul exhibits no distinct affinities to any of the other samples included in this analysis, this is not the case for the later Tarim Basin samples from Alwighul and Krorän. Although results differ as to whether Krorän has closer affinities to the inhabitants of the earliest of the north Bactrian urban centers (Sapalli) than does Alwighul (Figs. 2, 3), or whether affinities to the inhabitants of Sapalli are equally close (Figs. 4, 5), all results indicate that these later inhabitants of the Tarim Basin manifest a unique affinity to Bactrians. None of the results revealed affinities between Tarim Basin samples and samples from the Russo-Kazakh steppelands, the Indus Valley, or the Han Chinese sample from Hainan.

It appears that neither Han Chinese nor steppe populations played any detectable role in the initial establishment or subsequent interregional biological interactions of Bronze Age Tarim Basin populations. This conclusion, however, must be tempered with a pair of caveats. First, East Asian populations may have played a late role in the eastern oases of the Tarim Basin (Han, [1994b], [1998]). Until adequate samples from such eastern Tarim Basin sites as Yanbulaq (Han, [1990], [1998]) and from temporally and geographically more appropriate Han Chinese contexts become available, this possibility must remain open. Second, while Russo-Kazakh steppe populations appear to have played no role in the establishment of southern (KRO and QAW) and northern Tarim Basin oasis populations (ALW) for which data are available, this does not rule out a role for steppe populations in Xinjiang during the Bronze Age all together. It is possible that Afanasievo populations, Andronovo populations, or both may have played an important role in the development of Bronze Age cultures of the Zhunge'er Basin and the Yili Valley, located north of the Tian Shan Mountains in northern Xinjiang.

This research confirms that populations from the urban centers of the Oxus civilization of Bactria played a role in the population history of the Bronze Age inhabitants of the Tarim Basin. Yet these Bactrian populations were not the direct, early colonizers envisioned by advocates of the Bactrian oasis hypothesis (Barber, [1999]). None of the analyses document the immediate and profoundly close affinities between colonizers and the colonized expected if the Tarim Basin experienced substantial direct settlement by Bactrian agriculturalists.

Likewise, the pattern of affinities between the Tarim Basin and Bactrian samples also fails to support a model in which a sizable extant resident population within the Tarim Basin prarticipated in long-standing, bidirectional gene flow with urban populations of the BMAC. If such were the case, phenetic distances between Bactrian populations and Tarim Basin populations should initially show no affinities to one another, while later samples should demonstrate a progressive reduction in the phenetic distance that separates them over time. None of the analyses document this relationship. Rather, the closest affinities occur between Sapalli, the earliest of the Bactrian samples, and the two later Tarim Basin samples from Alwighul and Krorän. Diametrically opposed to these expectations, all analyses, apart from WPGMA cluster analysis, reveal that remaining the Bactrian samples are marked by decreasing, rather than increasing, phenetic affinities to Tarim Basin samples with the passage of time.

If the relationship between Tarim Basin populations and Oxus civilization populations was neither one of colonization, nor of long-standing bidirectional interaction, what was the nature of the relationship between these two regional populations? Many authorities assert that the dramatic changes in textiles, clothing styles, and metallurgical technology found in the Tarim Basin soon after 1200 B.C. may be associated with a wave of immigration of Iranian-speaking peoples (Barber, [1999]; Kuzmina, [1998]; Mair and Mallory, [2000]). Many identify these immigrants as Andronovo steppe nomads from the north and northwest (Barber, [1999]; Kuzmina, [1998]; Vinogradova and Kuzmina, [1996]). Others, however, suggest that they may represent the Saka (Debaine-Francfort, [1990]; Parpola, [1998], p. 135), a historically known Iranian-speaking population found in the Pamirs dividing western and eastern Central Asia by the sixth century B.C.

Most researchers view the Saka as either direct lineal descendants of a steppe Andronovo population (Kuzmina, [1998]) or a steppe Andronovo population that experienced admixture with local highland populations in western Tajikistan, the Ferghana Valley, and perhaps the upper reaches of the Tian Shan mountains of Xinjiang (e.g., Chen and Hiebert, [1995], p. 285; Hiebert and Shishlina, [1996]). Previous analyses by Han ([1994], [1998]), however, provide no support for the assertion that Saka populations are ultimately of Russo-Kazakh steppe derivation. Rather, Han ([1998], p. 556-558) concluded that a Saka presence may be detected across the southern Tarim oases in a temporal sequence indicative of an initial entrance from the west prior to 1000 B.C., with a subsequent spread eastward to Krorän (see also Mair, [1995], p. 292; Mallory and Mair, [2000], p. 243). Further, the analyses by Han ([1994], [1998]) suggest that the Saka found in and along the southern margin of the Täklamakan Desert are of the Eastern Mediterranean type, while those found at the northern Tarim Basis oasis of Alwighul are placed in the Pamir-Ferghana type (Mallory and Mair, [2000], p. 238). Han ([1998]) explained that the Eastern Mediterranean type may be traced to western Central Asia, while the Pamir-Ferghana type may represent admixture between Eastern Mediterraneans and the local, resident population of the Tarim Basin. He concluded that all of the anthropological materials mentioned above seem to indicate that the opening of the ancient Silk Road from Xinjiang to Central Asia supported an eastward migration of the early Mediterranean population of Central Asia across the Pamir region (Han, [1998], p. 568).

This study confirms the assertion of Han ([1998]) that the occupants of Alwighul and Krorän are not derived from proto-European steppe populations, but share closest affinities with Eastern Mediterranean populations. Further, the results demonstrate that such Eastern Mediterraneans may also be found at the urban centers of the Oxus civilization located in the north Bactrian oasis to the west. Affinities are especially close between Krorän, the latest of the Xinjiang samples, and Sapalli, the earliest of the Bactrian samples, while Alwighul and later samples from Bactria exhibit more distant phenetic affinities. This pattern may reflect a possible major shift in interregional contacts in Central Asia in the early centuries of the second millennium B.C.

Hemphill ([1999]) argued that the populations that lived in the Oxus civilization urban centers of the north Bactrian oasis were the product of largely unidirectional gene flow between the descendents of refugees fleeing the desiccated Tedjen River delta of Turkmenistan and a local Neolithic population whose artifactual remains are designated as the Hissar Culture (Mandelshtam, [1968]; Masson and Sarianidi, [1972]; P'yankova, [1986], [1994]; Ranov, [1982]; Tosi, [1988]). Patterns of phenetic affinities possessed by north Bactrian Oxus civilization samples suggest a profound change in interregional contacts near the beginning of the second millennium B.C. After 2000 B.C., the population history of these Bronze Age north Bactrian urban centers is one of ever-increasing rapprochement in phenetic distances with populations to the west (Turkmenistan and Iran) over time. However, the earliest sample in this sequence, Sapalli, stood apart from all others, and the suggestion was made that, in light of discoveries of silk and millet seeds (cf. Askarov, [1973], [1977], [1981]) at this site, as well as compartmented bronze seals similar to those found in the Ordos region of western China (Hambis, [1956]; Kohl, [1981]; Pelliot, [1931-1932]), earlier contacts may have been oriented to the east: to the Ferghana Valley, and perhaps western China.

This study supports a connection between the early Oxus civilization inhabitants of the north Bactrian oasis and populations of the Tarim Basin, but the relationship was neither temporally immediate nor direct. If such contacts were of such a nature, the sample from Sapalli should be phenetically most similar to the early Tarim Basin sample from Qäwrighul. Since these interactions were believed to end with the shift in interregional contacts heralded by myriad cultural changes at the beginning of the Djarkutan phase (Abdullaev, [1979]; Askarov and Abdullaev, [1983]; Askarov and Shirinov, [1991]), affinities between Sapalli and later Tarim Basin samples should become increasingly diffuse with the passage of time, due to genetic drift. Neither of these expectations is borne out. Rather, the relationship between north Bactrian populations and populations of the Tarim Basin appears more subtle and complex.

Given the patterns of phenetic affinities found in this and previous studies (Hemphill, [1998], [1999]; Hemphill et al., [1998]), it is possible that the sample from Sapalli derives from an indigenous north Bactrian population, largely unaffected by extraregional gene flow, which represents the direct lineal descendants of the manufacturers of the Neolithic Hissar culture. Yet the Hissar culture did not disappear from areas adjoining the north Bactrian oasis with the appearance of the Oxus civilization. Rather, the Hissar culture continued in the mountains of southern Tajikistan (P'yankova, [1994], [1996]), and the many microliths found in Hissar culture assemblages share similarities with Neolithic assemblages found in the Ferghana Valley (Kairak-Kum culture: P'yankova, [1994]). Later, near the middle of the second millennium B.C., the archaeological assemblages of these regions attest to contacts with Molali-phase inhabitants of the Oxus civilization urban centers and steppe Andronovo populations in southern Tajikistan (Vakhsh/Beshkent cultures: Litvinski, [1964], [1973], [1981]; Litvinski and P'yankova, [1992]; P'yankova, [1981], [1994], [1996]; Vinogradova, [1994]) and steppe Andronovo populations in the Ferghana Valley (Chust culture: Askarov, [1992]; Barber, [1999], p. 166; Shui, [1998], p. 166; Zadneprovsky, [1978]). It may be these local, resident populations of the north Bactrian oases (prior to 2000 B.C.), southern Tajikistan, and the Ferghana Valley (prior to ca. 1500 B.C.) who later came to be known as the Saka (Shui, [1998], p. 168). If so, the affinities found between Sapalli, Alwighul, and Krorän are reflective of genetic exchange between East and West through an intermediary - an intermediary that may have had control over one of the most important commodities of the Bronze Age in this region of the world, the tin deposits of the Ferghana Valley (Gupta, [1979]; Kohl, [1984]; Masson, [1992a]; P'yankova, [1994], p. 368; Tosi, [1973-194]) and the western Tian Shan Mountains (Hiebert and Shishlina, [1996], p. 11).

CONCLUSIONS


The Great Silk Road served for many years as a vital artery linking the worlds of East and West. Recent genetic studies confirmed that this avenue served not only as a conduit for commerce and cultural diffusion, but also for the exchange of genes (Comas et al., [1998]; Yao et al., [2000]). Yet while such studies are valuable for documenting the passage of genes, their reliance upon living populations provides little insight as to when and under what circumstances this gene flow occurred. Such knowledge is of primary importance, given recent archaeological evidence for eastern artifacts (silk) in the West and western artifacts (wool, bronze) in the East some 2,000 years earlier than the traditional date (132 B.C.) for the opening of the Great Silk Road.

Many scholars have compared the material cultural remains recovered from Bronze Age sites in Xinjiang to those associated with Afanasievo and Andronovo pastoralist populations of the Eurasian steppe. Others, such as Mair ([1995], p. 281), noted that the mummified human remains appear Caucasoid or Europoid. From these artifacts and observations, proponents of the steppe hypothesis have argued that some of the earliest known Bronze Age populations of Xinjiang owe their origins to migrations from the Russo-Kazakh steppe.

The results of this study provide little support for the steppe hypothesis, for none of the statistical procedures yields the close phenetic distances expected between colonizers (steppe samples) and the colonized (Tarim Basin samples). Similarly, none of the analyses revealed any affinity between the Bronze Age inhabitants of the Tarim Basin and Han Chinese. Thus, it appears that neither steppe populations nor Han Chinese populations played any significant role in the establishment of those Bronze Age populations of the Tarim Basin for which samples were available. Nevertheless, given the paucity of available evidence and appropriate comparative samples, contributions by these two regional populations in the establishment and subsequent interactions of Bronze Age Tarim Basin populations cannot be ruled out entirely.

Other researchers emphasize the similarities in environment and economy between western (Bactria and Margiana) and eastern Central Asia (Xinjiang), and suggest that the very skills in irrigation technology, networks of long-distance exchange, and ovicaprid domestication characteristic of the Bronze Age Oxus civilization of west Central Asia provided the ideal preparation for initial settlement of the Tarim Basin. Proponents of the Bactrian oasis hypothesis maintain that the Bronze Age populations of Xinjiang owe their origins to colonization from the Oxus civilization from the north Bactrian oasis. This study provides no support for this contention. While the earliest Oxus civilization sample, Sapalli, represents the non-Xinjiang sample with closest affinities to Tarim Basin samples, these affinities are with later inhabitants of the basin, not the earliest and temporally most proximate sample, Qäwrighul. Thus, Oxus civilization populations of the north Bactrian oasis may have had some form of contact with Bronze Age populations of the Tarim Basin, but this contact does not appear to have been one of colonization.

The earliest Bronze Age sample from the Tarim Basin (Qäwrighul) exhibits no close affinities to any non-Xinjiang sample, and hence the origins of this population remain, as yet, unknown. The Qäwrighul remains may be those of a local, indigenous population from the Tarim Basin itself or the surrounding highlands, they may represent emigrants from a source area not included in this analysis, or they may actually derive from one of the suggested source areas, but whose separation from the host population may have occurred earlier in time or involved fewer founding individuals than currently envisioned by proponents of either the steppe or Bactrian oasis hypotheses. Later Tarim Basin samples from Alwighul and Krorän are more closely affiliated with the early Oxus civilization sample from Sapalli than with the earlier Tarim Basin sample from Qäwrighul. A possible explanation for this association, despite temporal differences of 1,800 and 1,200 years, respectively, is that these affinities are a consequence of a long-standing local population, interposed between the Tarim Basin and the north Bactrian oasis, that facilitated interactions between the populations of East and West through their control of a vital economic resource (tin) and territory (the Pamirs and the Ferghana Valley).

Acknowledgements


The authors thank Timor Shirinov, Director of the Institute of Archaeology, Uzbek Academy of Sciences, for granting access to the Djarkutan and Sapalli tepe skeletal series. Thanks also go to Viktor Sarianidi and Denis Pezhemsky of the Russian Academy of Sciences in Moscow for granting access to Geoksyur and Altyn depe skeletal series. Thanks go to Victor Mair, Chris Thornton, and two anonymous reviewers for many helpful comments on an earlier draft of this paper. Special thanks are due to Jaymie L. Brauer for numerous helpful editorial comments and insights throughout all stages of this research, and to M. Cassandra Hill and John Rodriguez for preparation of Figure 1.

References

Abdullaev B. 1979. Mogil'nik Dzharkutan. Istorija Materialnoj Kultury Uzbekistana 15: 1-22. Links
Abduelivili MG. 1954. K Paleoantropology Samtavrskogo Mogilnika. Tbilisi: FAN.
Abduelivili MG. 1960. Paleoantropologichesky material iz pozdnich pogrebenij Samtavrskogo moil'nika. Trudy Inso. Eks Morf AN Grusinskoy SSr 8: 281-303. Links
Abduelivili MG. 1966. K Kraniology Drevnego i Sovremennogo Naselenija Kavkasa. Tbilisi: FAN.
Adams DQ. 1984. The position of Tocharian among the other Indo-European languages. J Am Oriental Soc 104: 395-402. Links
Alexeev VP. 1961. Paleoanthropology of the Altai-Saian mountains in the Neolithic and Bronze Age. Antropol Sbornik 3: 107-206. Links
Alekseev VP. 1967. Antropologia Andronovskoj kultury. Sov Archaeol 1: 22-26. Links
Alekseev VP, Gochman II. 1983. Rassengeschichte der Menscheit. Asien II: Sowjet-Asien. Munich: R. Oldenbourg Verlag.
Allchin B, Allchin R. 1982. The rise of civilization in India and Pakistan. Cambridge: Cambridge University Press.
An Z. 1992a. Neolithic communities in eastern parts of Central Asia. In: Dani AH , Masson VM , editors. History of civilizations of Central Asia, volume I. Paris: UNESCO. p 153-189.
An Z. 1992b. Bronze Age communities in eastern parts of Central Asia. In: Dani AH , Masson VM , editors. History of civilizations of Central Asia, volume I. Paris: UNESCO. p 319-336.
An Z. 1998. Cultural complexes of the Bronze Age in the Tarim Basin and surrounding areas. In: Mair VH , editor. The Bronze Age and early Iron Age peoples of eastern Central Asia. Volume 1. Philadelphia: University of Pennsylvania Museum Publications. p 45-62.
Anthony DW. 1998. The opening of the Eurasian steppe at 2000 BCE. In: Mair VH , editor. The Bronze Age and early Iron Age peoples of eastern Central Asia. Volume 1. Philadelphia: University of Pennsylvania Museum Publications. p 94-113.
Anthony DW, Vinogradova NB. 1995. Birth of the chariot. Archaeology 48: 36-41. Links
Askarov AA. 1973. Sapallitepa. Tashkent: FEN.
Askarov AA. 1974. K voprosu o vydelenii kul'tury Sapalli. In: Masson VM , editor. Drevnyaya Baktriya. Leningrad: Nauka. p 26-32.
Askarov AA. 1977. Drevnezemledel'cheskaya kul'tura Epokhi Bronzi Iuga Uzbekistana. Tashkent: FEN.
Askarov AA. 1981. Southern Uzbekistan in the second millennium B.C. In: Kohl PL , editor. The Bronze Age civilization of Central Asia: recent Soviet discoveries. Armonk: M.E. Sharpe. p 256-272.
Askarov AA. 1988. L'Uzbekistan à l'age du Bronze: exploitation des resources naturelles en rapports aves les régions Voisines. In: Gardin JC , editor. L'Asie Centrale et ses rapports avec les civilizasions orientales des origines à l'Age du Fer. Paris: de Boccard. p 103-107.
Askarov AA. 1992. The beginning of the Iron Age in Transoxiana. In: Dani AH , Masson VM , editors. History of civilizations of Central Asia. Volume I. The dawn of civilization: earliest times to 700 B.C. Paris: UNESCO. p 441-458.
Askarov AA, Abdullaev BN. 1983. Dzharkutan. Tashkent: FAN.
Askarov AA, Shirinov T. 1991. Le temple du feu de Djarkutan, le plus ancien centre cultuel de la Bactriane septentrionale. In: Francfort HP , editor. Histoire et cultes de l'Asie Centrale préislamique. Paris: Éditions du Centre National de la Recherche Scientifique. p 129-136.
Ayala FJ. 1982. Population and evolutionary genetics: a primer. Menlo Park, CA: Benjamin/Cummings.
Barber EW. 1999. The mummies of Ürümchi. New York: Norton.
Barbujani G. 1987. Autocorrelation of gene frequencies under isolation by distance. Genetics 117: 777-782. Links
Beals KL. 1972. Head form and climatic stress. Am J Phys Anthropol 37: 85-92. Links
Bergman F. 1939. Archaeological researchers in Sinkiang. Stockholm: Bokförlags Aktiebolaget Thule.
Bernhard W. 1967. Human skeletal remains from the cemetery of Timargarha. Pakistan Archaeol 3: 291-407. Links
Bodmer WF, Cavalli-Sforza LL. 1975. Genetics, evolution, and man. San Francisco: W.H. Freeman and Co.
Bryant E. 2001. The quest for the origins of Vedic culture: the Indo-Aryan migration debate. New York: Oxford University Press.
Bunker E. 1998. Cultural diversity in the Tarim Basin vicinity and its impact on Chinese culture. In: Mair VH , editor. The Bronze Age and early Iron Age peoples of eastern Central Asia. Volume 2. Philadelphia: University of Pennsylvania Museum Publications. p 604-618.
Burrow T. 1973. The Sanskrit language. London: Farber & Farber.
Carlson DS, Van Gerven DP. 1977. Masticatory function and post-Pleistocene evolution in Nubia. Am J Phys Anthropol 46: 495-506. Links
Cavalli-Sforza LL. 2000. Genes, peoples, and languages. New York: North Point Press/Farrar, Straus & Giroux.
Cavalli-Sforza LL, Bodmer WF. 1971. The genetics of human populations. San Francisco: W.H. Freeman.
Cavalli-Sforza LL, Menozzi P, Piazza A. 1994. The history and geography of human genes. Princeton: Princeton University Press.
Chang KC. 1977. The archaeology of ancient China. New Haven: Yale Univesity Press.
Chen G. 1990. On the cultures of the Bronze and early Iron Ages in the Xinjiang area. Kaogu 4: 366-374. Links
Chen K, Hiebert FT. 1995. The late prehistory of Xinjiang in relation to its neighbors. J World Prehist 9: 243-300. Links
Chernykh EN. 1992. Ancient metallurgy in the USSR. Cambridge: Cambridge University Press.
Chlenova NL. 1983. Prehistory of Herodotus' trade road. Sov Archaeol 1: 47-66. Links
Clark PJ. 1956. The heritability of certain anthropometric characters as ascertained from measurements of twins. Am J Hum Genet 8: 49-59. Links
Comas D, Clafell F, Mateu E, Pérez-Lezaun A, Bosch E, Martínez-Arias R, Clarimon J, Facchini F, Fiori G, Luiselli D, Pettener D, Bertranpetit J. 1998. Trading genes along the Silk Road: mtDNA sequences and the origin of Central Asian populations. Am J Hum Genet 63: 1824-1838. Links
Cummings MR. 1997. Human heredity: principles and issues. Belmont, CA: Wadsworth.
Debaine-Francfort C. 1988. Archaéologie du Xinjiang des origines aux Han. I. Paleorient 14: 5-29. Links
Debaine-Francfort C. 1990. Le Saka du Xinjiang avant les Han (206 BCE - 220 AD): critères d'identification. In: Francfort HP , editor. Nomades et sédentaires en Asie Centrale: rapports de l'archéologie et de l'ethnologie. Paris: Éditions du Centre National de la Recherche Scientifique. p 81-95.
Devor EJ. 1987. Transmission of human craniofacial dimensions. J Craniofac Genet Dev Biol 7: 95-106. Links
Di Cosmo N. 1996. Ancient Xinjiang between Central Asia and China: the nomadic factor. Anthropol Archaeol Eurasia 34: 87-101. Links
Falconer DS. 1981. Introduction to quantitative genetics. 2nd ed. New York: Wiley.
Felsenstein J. 1989. PHYLIP - phylogeny inference package (version 3.2). Cladistics 5: 164-166. Links
Fix A. 1999. Migration and colonization in human microevolution. Cambridge: Cambridge University Press.
Francalacci P. 1995. DNA analysis of ancient desiccated corpses from Xinjiang. J Indo Eur Stud 23: 385-398. Links
Gamkrelidze TV, Ivanov VV. 1990. The early history of Indo-European languages. Sci Am 262: 110-116. Links
Ginzberg VV, Trofimova TA. 1972. Paleoantropologia Srednej Azii. Moscow: FAN.
Good I. 1995. Notes on a Bronze Age textile from Hami, Xinjiang with comments on the significance of twill. J Indo Eur Stud 23: 319-345. Links
Gryaznov MP, Vadezkaya EB. 1968. Afanasievskaya kul'tura. Istoriya Sibiri 1: 159-165. Links
Guglielmino-Matessi CR, Gluckman P, Cavalli-Sforza LL. 1979. Climate and the evolution of skull metrics in man. Am J Phys Anthropol 50: 549-564. Links
Gupta P, Dutta PC, Basu A. 1962. Human skeletal remains from Harappa. Mem Anthropol Surv India 9: 3-186. Links
Gupta SP. 1979. Archaeology of Soviet Central Asia and the Iranian borderlands. Delhi: B.R. Publishing Corp.
Guttman L. 1968. A general nonmetric technique for finding the smallest coordinate space for a configuration of points. Psychometrika 33: 469-506. Links
Hair JF, Anderson RE, Tatham RL, Grablowsky BJ. 1971. Multivariate data analysis. Tulsa: PPC Books.
Hambis L. 1956. A propos des ceaux-amulettes Nestoriens. Arts Asiatiq 3: 279-286. Links
Hamp ES. 1998. Who were the Tocharians? - Linguistic subgrouping and diagnostic idiosyncrasy. In: Mair VH , editor. The Bronze Age and early Iron Age peoples of eastern Central Asia. Volume 1. Philadelphia: University of Pennsylvania Museum Publications. p 307-346.
Han K. 1986a. Anthropological characteristics of the human skulls from the ancient cemetery at Qäwrighul (Gumugou), Xinjiang. Acta Archaeol Sin 3: 361-384. Links
Han K. 1986b. Anthropological characteristics of the human crania from Kroran (Loulan) site, Xinjiang. Acta Anthropol Sin 3: 227-242. Links
Han K. 1990. The study of racial elements of bones from the Yanbulaq site of Qumul, Xinjiang. Acta Anthropol Sin 5: 371-390. Links
Han K. 1994a. The study of ancient human skeletons from Xinjiang, China. Sino-Platonic papers no. 51. Philadelphia: Department of Asian and Middle Eastern Studies, University of Pennsylvania.
Han K. 1994b. The racial anthropological study of the ancient Inhabitants of Xinjiang. Urumchi: People's Press of Xinjiang.
Han K. 1998. The physical anthropology of the ancient populations of the Tarim Basin and surrounding areas. In: Mair VH , editor. The Bronze Age and early Iron Age peoples of eastern Central Asia. Volume 2. Philadelphia: University of Pennsylvania Museum Publications. p 558-570.
Han K. 2001. Physical anthropological studies on the racial affinities of the inhabitants of ancient Xinjiang. In: Wang B , editor. The ancient corpses of Xinjiang: the peoples of ancient Xinjiang and their culture. Urumchi: Xinjiang Renmin Chubanshe. p 224-241.
Haritgan JH. 1975. Clustering alogrithms. New York: Wiley.
Hartl DL, Clark AG. 1997. Principles of population genetics. 3rd ed. Sunderland, MA: Sinauer Associates.
Hedrick PW. 2000. Genetics of populations. 2nd ed. Sudbury, MA: Jones and Bartlett.
Hemphill BE. 1998. Biological adaptations and affinities of Bronze Age Bactrians: III. An initial craniometric assessment. Am J Phys Anthropol 106: 329-348. Links
Hemphill BE. 1999. Biological adaptations and affinities of Bronze Age Bactrians: IV. a craniometric examination of the origins of Oxus civilization populations. Am J Phys Anthropol 108: 173-192. Links
Hemphill BE, Christensen AF, Mustafakulov SI. 1998. Biological adaptations and affinities of Bronze Age Bactrians: II. Dental morphology. In: Lukacs JR , editor. Human dental development, morphology, and pathology. University of Oregon anthropological papers no. 54. Eugene: University of Oregon. p 51-77.
Hemphill BE, Lukacs JR, Kennedy KAR. 1991. Biological adaptations and affinities of Bronze Age Harappans. In: Meadow R , editor. Harappa excavations 1986-1990: a multidisciplinary approach to third millennium urbanism. Madison: Prehistory Press. p 137-182.
Hiebert FT. 1994. Origins of the Bronze Age oasis civilization in Central Asia. American School of Prehistoric Research, bulletin no. 42. Cambridge, MA: Harvard University Press.
Hiebert FT, Shishlina NI. 1996. Introduction to between lapis and jade: ancient cultures of Central Asia. Anthropol Archaeol Eurasia 34: 5-12. Links
Howells WW. 1966. Variability in family lines versus population variability. Trans NY Acad Sci 134: 624-631. Links
Howells WW. 1973. Cranial variation in man: a multivariate analysis of patterns of difference among recent human populations. Papers of the Peabody Museum of Archaeology and Ethnology, volume 67. Cambridge, MA: Harvard University Press.
Howells WW. 1989. Skull shapes and the map: craniometric analyses in the dispersion of modern Homo. Papers of the Peabody Museum of Archaeology and Ethnography, volume 79. Cambridge, MA: Harvard University.
IAX (Institute of Archaeology of the Xinjiang Academy of Social Sciences). 1989. Report on the survey of the microlithic sites at the Caiwobao Lake, Xinjiang. Kaogu Yu Wenwu 1989: 10-16. Links
Jettmar K. 1992. Archäologie in Xinjiang und ihre Bedeutung für Südsibirien. Beitr Allg Vergleich Archaol 12: 139-153. Links
Jettmar K. 1998. Early migrations in Central Asia. In: Mair VH , editor. The Bronze Age and early Iron Age peoples of eastern Central Asia. Volume 1. Philadelphia: University of Pennsylvania Museum Publications. p 215-221.
Kenoyer JM. 1998. Ancient cities of the Indus Civilization. Karachi: Oxford University Press.
Kiiatkina TP. 1976. Materialy po paleoantropologii Tajikistana. Dushanbe: FAN.
Kiiatkina TP. 1977. Kraniologieski material iz Altyn-Depe (1971 g.). Karakum Drevnosti Aschabar 5: 170-175. Links
Kiiatkina TP. 1987. Paleoantropologiya zanadnikh rayonov Tsentral'noy Azii Epoki Bronzi. Dushanbe: Akademiya Hauk Tajikistan.
Kimura M, Weiss GH. 1964. The stepping-stone model of population structure and the decrease of genetic correlation with distance. Genetics 49: 561-576. Links
Kohl PL. 1981. The Namazga civilization: an overview. In: Kohl PL , editor. The Bronze Age civilizations of Central Asia: recent Soviet discoveries. Armonk: M.E. Sharpe. p vii-xxxviii.
Kohl PL. 1984. Central Asia: Palaeolithic beginnings to the Iron Age. Editions recherche sur les civilisations, synthèse no. 14. Paris: CNRS.
Kohl PL. 1992. Central Asia (western Turkestan): Neolithic to the Early Iron Age. In: Ehrich RW , editor. Chronologies in Old World archaeology, volume I. 3rd ed. Chicago: University of Chicago Press. p 179-195.
Komroff M. 1930. Introduction. In: Komroff M , editor. The travels of Marco Polo. New York: Boni and Liveright. p v-xxxii.
Konigsberg LW, Kohn LAP, Cheverud JM. 1993. Cranial deformation and nonmetric trait variation. Am J Phys Anthropol 90: 35-48. Links
Kowalski CJ. 1972. A commentary on the use of multivariate statistical methods in anthropometric research. Am J Phys Anthropol 36: 119-132. Links
Krogman WM. 1940. Racial types from Tepe Hissar, Iran from the late fifth to the early second millennium B.C. Verh K Ned Akad Wet Afeel Nat (Tweedie Sectie) 39: 1-87. Links
Kuchera S. 1988. Novoe dostizhenie archeologov I antropologov KNR. Obschestvo I Gosudarstvo Kitae 19: 3-14. Links
Kussov S. 1993. White rooms in the temples of Margiana. Inf Bull Int Assoc Stud Cult Cent Asia 19: 128-135. Links
Kuzmina EE. 1994. Otkuda prishli Indoarii? Moscow: M.G.P. Kalina.
Kuzmina EE. 1998. Cultural connections of the Tarim Basin people and pastoralists of the Asian steppes in the Bronze Age. In: Mair VH , editor. The Bronze Age and early Iron Age peoples of eastern Central Asia. Volume 1. Philadelphia: University of Pennsylvania Museum Publications. p 63-93.
Litvinski BA. 1964. Tajikistan i India (premeri drevnikh svyazey i kontaktov). Moscow: Nauka.
Litvinski BA. 1973. Arkheologicheskiye raboti v Tajikistane v 1962-1977. Arkheol Raboti Tajikistane 10: 5-41. Links
Litvinski BA. 1981. Problemi etnicheskoy istorii Sredney Azii vo II tisyacheletii do n.e. Etnicheskiye problemi istorii Tsentral'noy Azii v drevnosti II tiyacheletiye do n.e. Moscow: Nauka.
Litvinski BA, P'yankova LT. 1992. Pastoral tribes of the Bronze Age in the Oxus Valley (Bactria). In: Dani AH , Masson VM , editors. History of civilizations of Central Asia. Volume I. The dawn of civilization: earliest times to 700 B.C. Paris: UNESCO. p 379-394.
Lundstrom A. 1954. The importance of genetic and non-genetic factors in the profile of the facial skeleton studies in 100 pairs of twins. Rep 30th Congr Eur Orthodont Soc 30: 92-107. Links
Ma Y, Wang B. 1994. The culture of the Xinjiang region. In: Harmatta J , Puri BN , Etemadi GF , editors. History of civilizations of Central Asia, volume II. The development of sedentary and nomadic civilizations: 700 B.C. to A.D. 250. Paris: UNESCO. p 209-225.
Mair VH. 1995. Prehistoric Caucasoid corpses of the Tarim Basin. J Indo Eur Stud 23: 281-307. Links
Mair VH. 1998. Priorities. In: Mair VH , editor. The Bronze Age and early Iron Age peoples of eastern Central Asia. Volume 1. Philadelphia: University of Pennsylvania Museum Publications. p 4-41.
Malécot G. 1967. Identical loci relationships. Proc Fifth Berkeley Symp Math Stat Probl 4: 317-332. Links
Mallory JP. 1989. In search of the Indo-Europeans: language, archaeology and myth. London: Thames and Hudson.
Mallory JP. 1995. Speculations on the Xinjiang mummies. J Indo Eur Stud 23: 371-384. Links
Mallory JP. 1998. A European perspective on Indo-Europeans in Asia. In: Mair VH , editor. The Bronze Age and early Iron Age peoples of eastern Central Asia. Volume 1. Philadelphia: University of Pennsylvania Museum Publications. p 175-201.
Mallory JP, Mair VH. 2000. The Tarim mummies. London: Thames and Hudson.
Mandelshtam AM. 1968. Pamyatniki Epokhi Bronzi v Iuzhnom Tadzhikistane. Moscow: Nauka.
Mantel N. 1967. The detection of disease clustering and a generalized regression approach. Cancer Res 27: 209-220. Links
Martin R. 1928. Lehrbuch der Anthropologie. Zweiter Band: Kraniologie, Osteologie. Jena: Gustav Fischer Verlag.
Masson VM. 1992a. The Bronze Age in Khorasan and Transoxiana. In: Dani AH , Masson VM , editors. History of civilizations of Central Asia. Volume I. The dawn of civilization: earliest times to 700 B.C. Paris: UNESCO. p 225-245.
Masson VM. 1992b. The decline of the Bronze Age civilization and movements of the tribes. In: Dani AH , Masson VM , editors. History of civilization of Central Asia. Volume I. The dawn of civilization: earliest times to 700 B.C. Paris: UNESCO. p 337-356.
Masson VM, Sarianidi VI. 1972. Central Asia: Turkmenia before the Achaemenids. London: Thames and Hudson.
Mei J. 2000. Copper and bronze metallurgy in late prehistoric Xinjiang. British archaeological report no. 865. Oxford: Archaeopress.
Mei J, Shell C. 1998. Copper and bronze metallurgy in late prehistoric Xinjiang. In: Mair VH , editor. The Bronze Age and early Iron Age peoples of eastern Central Asia. Volume 2. Philadelphia: University of Pennsylvania Museum Publications. p 581-603.
Meyer-Melikyan NR. 1998. Analysis of floral remains from Togolok 21. In: Sarianidi V , editor. Margiana and Protozoroastrism. Athens: Kapon Editions. p 178-179.
Meyer-Melikyan NR, Avetov NA. 1998. Analysis of floral remains in the ceramic vessel from the Gonur temenos. In: Sarianidi V , editor. Margiana and Protozoroastrism. Athens: Kapon Editions. p 176-177.
Morton NE, Kenett R, Yee S, Lew R. 1982. Bioassay of kinship in populations of Middle Eastern and controls. Curr Anthropol 23: 157-167. Links
Nakata M, Yu P-L, Nance WE. 1974a. Multivariate analysis of craniofacial measurements in twin and family data. Am J Phys Anthropol 41: 423-430. Links
Nakata M, Yu P-L, Davis B, Nance WE. 1974b. Genetic determinants of cranio-facial morphology: a twin study. Am J Hum Genet 37: 431-443. Links
Nichols J. 1997. The epicenter of the Indo-European linguistic spread and the Eurasian spread zone. In: Blench R , Spriggs M , editors. Archaeology and language. Volume 1: combining archaeological and linguistic aspects of the past. London: Routledge. p 122-148.
Nyberg H. 1995. The problem of Aryans and the soma: the botanical evidence. In: Erdosy G , editor. Language, material culture and ethnicity: the Indo-Aryans of ancient South Asia. Berlin: de Gruyter. p 382-406.
Olsen JW, Reeves RW, Miller-Antonio S, Jiaqiang L. 1988. New discoveries of stone artifacts on the southern edge of the Tarim Basin. Acta Anthropol Sin 7: 294-301. Links
Orczykowska-Swiatkowska Z, Lebioda H. 1975. Variability of cranial shape in twins. Stud Phys Anthropol 1: 21-30. Links
Oxnard C. 1973. Form and pattern in human evolution. Chicago: University of Chicago Press.
Pardini E. 1977. Gli inumati di Shahr-i Sokhta (Sistan, Iran): studio osteologico preliminare- - II parte. Ist Anthropol Univ Firenze Estratto Arch Anthrpol Etnol 107: 159-235. Links
Pardini E. 1979-1980. Gli inumati di Shahr-i Sokhta (Sistan, Iran). Ist Anthropol Univ Firenze Estratto Arch Anthrpol Etnol 109-110: 521-608. Links
Pardini E, Sarvari-Negahban AA. 1976. Craniologia degli Inumati di Shahr-i Sokhta (Sistan, Iran) - studio preliminare. Ist Anthropol Univ Firenze Estratto Arch Anthrpol Etnol 106: 1-49. Links
Parpola A. 1995. The problem of the Aryans and the soma: textual-linguistic and archaeological evidence. In: Erdosy G , editor. Language, material culture and ethnicity: the Indo-Aryans of ancient South Asia. Berlin: Walter de Gruyter. p 353-381.
Parpola A. 1998. Aryan languages, archaeological cultures, and Sinkiang: where did proto-Iranian come into being, and how did it spread? In: Mair VH , editor. The Bronze Age and early Iron Age peoples of eastern Central Asia. Volume 1. Philadelphia: University of Pennsylvania Museum Publications. p 114-147.
Pelliot P. 1931-1932. Sceaux-amulettes de bronze avec croix et colombes provenant de la boucle du fleuve jaune. Rev Arts Asiatiq 7: 1-3. Links
Peng K. 1998. The Andronovo bronze artifacts discovered in Toquztara county Ili, Xinjiang. In: Mair, VH , editor. The Bronze Age and early Iron Age peoples of eastern Central Asia. Volume 2. Philadelphia: University of Pennsylvania Museum Publications. p 573-580.
Piazza A, Menozzi P. 1983. Geographic variation in human gene frequencies. In: Felsenstein J , editor. Numerical taxonomy: proceedings of a NATO advanced study institute. Berlin: Springer. p 444-450.
Posrednikov VA. 1992. Pit-grave culture migration to the east and the Afanasievo-proto-Tocharian problem. Donezky Archaeol Sbornik 17: 9-20. Links
P'yankova LT. 1981. Bronze Age settlements of southern Tajikistan. In: Kohl PL , editor. The Bronze Age civilization of Central Asia: recent Soviet discoveries. Armonk: M.E. Sharpe. p 287-310.
P'yankova LT. 1986. Jungbronzezeitliche Grabefelder in Vachs-Tal, Süd-Tadzhikistan. Munich: C.H. Beck.
P'yankova LT. 1994. Central Asia in the Bronze Age: sedentary and nomadic cultures. Antiquity 68: 355-372. Links
P'yankova LT. 1996. Bronze Age cultures of the steppe and highlands in Central Asia. Anthropol Archaeol Eurasia 34: 13-28. Links
Ranov VA. 1982. Istoki, vneshnye vliania i vnutrennee razvitie Gissarskoi kul'tury. In: Ranov VA , editor. Drevneishie kul'tury Baktrii. Dushanbe: Donish. p 19-21.
Renfrew C. 1988. Archaeology and language: the puzzle of Indo-European origins. Cambridge: Cambridge University Press.
Ringe D, Warnow T, Taylor A, Michailov A, Levison L. 1998. Computational cladistics and the position of Tocharian. In: Mair VH , editor. The Bronze Age and early Iron Age peoples of eastern Central Asia. Volume 1. Philadelphia: University of Pennsylvania Museum Publications. p 391-414.
Rohlf FJ. 1972. An empirical comparison of three ordination techniques in numerical taxonomy. Syst Zool 21: 271-280. Links
Rohlf FJ. 2000. NTSYS - pc numerical taxonomy and multivariate analysis system. Version 2.1. Setauket: Exeter Software.
Rykusina GV. 1976. Antropologia epochi eneolitia-bronzy Krasnojarskogo kraja. In: Rykusina GV , editor. Nektorye problemy etnogeneza i etniceskoy istorii Narodov Mira. Moskow: FAN. p 187-201.
Saitou N, Nei M. 1987. The neighbor-joining method: a new method for reconstructing phylogenetic trees. Mol Biol Evol 4: 406-425. Links
Sargent B. 1997. Genèse de I'Inde. Paris: Payot.
Sarianidi VI. 1987. South-west Asia: migrations, the Aryans and Zoroastrians. Info Bull Int Assoc Cult Cent Asia 13: 44-56. Links
Sarianidi VI. 1990. Drevnosti strani Margush. Ashkhabad: Ylym.
Sarianidi VI. 1993a. Margiana in the ancient Orient. Inf Bull Int Assoc Stud Cult Cent Asia 19: 5-28. Links
Sarianidi VI. 1993b. Recent archaeological discoveries and the Aryan problem. In: Gail AJ , Mevissen GJR , editors. South Asian archaeology 1991. Stuttgart: Franz Steiner Verlag. p 251-264.
Sarianidi VI. 1994. Temples of Bronze Age Margiana: traditions of ritual architecture. Antiquity 68: 388-397. Links
Saunders SR, Popovich F, Thompson GW. 1980. A family study of cranio-facial dimensions in the Burlington Growth Centre sample. Am J Orthod 78: 394-403. Links
Shishlina NI, Hiebert FT. 1998. The steppe and the sown: interaction between Bronze Age Eurasian nomads and agriculturalists. In: Mair VH , editor. The Bronze Age and early Iron Age peoples of eastern Central Asia. Monograph no. 26. Philadelphia: Journal of Indo-European Studies. p 222-237.
Shui T. 1993. A comparative study of Bronze Age cultures in Xinjiang - with a discussion of the process of early cultural exchange between the East and the West. Stud Sinol 1: 449-490. Links
Shui T. 1998. On the relationship between the Tarim and Fergana Basins in the Bronze Age. In: Mair VH , editor. The Bronze Age and early Iron Age peoples of eastern Central Asia. Volume 1. Philadelphia: University of Pennsylvania Museum Publications. p 162-168.
Smouse PE, Long JC, Sokol RR. 1986. Multiple regression and correlation extensions of the Mantel test of matrix correspondence. Syst Zool 35: 627-632. Links
Sneath PHA, Sokol RR. 1973. Numerical taxonomy. San Francisco: W.H. Freeman.
Sokal RR, Wartenburg DE. 1983. A test of spatial autocorrelation analysis using an isolation-by-distance model. Genetics 105: 219-237. Links
Sokal RR, Smouse PE, Neel JV. 1986. The genetic structure of a tribal population, the Yanomama Indians. XV. Patterns inferred by autocorrelation analysis. Genetics 114: 259-287. Links
Sulimirski T. 1970. Prehistoric Russia. London: John Baker, Ltd.
Susanne C. 1975. Genetic and environmental influences on morphological characteristics. Ann Hum Biol 2: 279-287. Links
Susanne C. 1977. Heritability of anthropological characters. Hum Biol 49: 573-580. Links
Teilhard de Chardin P, Young CC. 1932. On some Neolithic (and possibly Palaeolithic) finds in Mongolia, Sinjian and west China. Bull Geol Soc China 12: 83-104. Links
Tosi M. 1973-1974. The northeastern frontier of the ancient Near East. Mesopotamia 8-9: 21-76. Links
Tosi M. 1988. The origins of early Bactrian civilization. In: Ligabue G , Salvatori S , editors. Bactria: an ancient oasis civilization from the sands of Afghanistan. Venice: Erizzo. p 41-72.
Trofimova TA. 1961. Cerepa iz Gulkinskogo mogilnika tazabagyabskoi kultury Kokca 3. In: Trofimova TA , editor. Molgilnik Bronzovogo veka Kokca 3. Moscow: FAN. p 97-146.
Van Gerven DP. 1982. The contribution of time and local geography to craniofacial variation in Nubia's Batn el Hajar. Am J Phys Anthropol 59: 307-316. Links
Vinogradova NM. 1994. The farming settlement of Tangurttut (south Tadjikistan) in the late Bronze Age. Archaeol Mitteilungen Iran 27: 29-47. Links
Vinogradova NM, Kuzmina EE. 1996. Contacts between the steppe and agricultural tribes of Central Asia in the Bronze Age. Anthropol Archaeol Eurasia 34: 29-54. Links
Wang B. 1982. The excavation and preliminary study of the Gumugou cemetery. In: Xinjiang Academy of Social Sciences, editors. Collections of articles of the first meeting on social science research. Urumchi: Xinjiang Academy of Social Sciences. p 327-349.
Wang B. 1983. On several problems of the social-cultural aspects of the Gumugou people. Xinjiang Daxue Xuebao 1983: 86-90. Links
Wang B. 1984. New knowledge on ancient civilization in Xinjiang. Baike Zhishi 1: 19-22. Links
Wang B. 1985. Study of microlithic artifacts of Xinjiang. In: Wang B , editor. Archaeological studies of the Silk Road. Urumqui: Xinjiang Renmin. p 129-145.
Wang B. 1987. Recherches historiques preliminaries sur les Saka du Xinjiang ancien. Arts Asiatiq 42: 31-44. Links
Wang B. 1996. A preliminary analysis of the archaeological cultures of the Bronze Age in the region of Xinjiang. Anthropol Archaeol Eurasia 34: 67-86. Links
Weiss KM. 1988. In search of times past: gene flow and invasion in the generation of human diversity. In: Mascie-Taylor CGN , Lasker GW , editors. Biological aspects of human migration. Cambridge: Cambridge University Press. p 130-166.
Wijsman EM, Cavalli-Sforza LL. 1984. Migration and genetic population structure with special reference to humans. Annu Rev Ecol Syst 15: 279-301. Links
Wilkinson L. 1990. SYSTAT: the system for statistics. Evanston: SYSTAT, Inc.
Wright S. 1943. Isolation by distance. Genetics 28: 114-138. Links
Wright S. 1946. Isolation by distance under diverse systems of mating. Genetics 31: 39-59. Links
Wright S. 1951. The genetical structure of populations. Ann Eugen 15: 323-354. Links
Xu W. 1995. The discovery of the Xinjiang mummies and studies of the origins of the Tocharians. J Indo Eur Stud 23: 357-369. Links
Yao Y-G, Lü X-M, Luo H-R, Li W-H, Zhang Y-P. 2000. Gene admixture in the Silk Road region of China: evidence from mtDNA and melanocortin 1 receptor polymorphism. Genes Genet Syst 75: 173-178. Links
Zadneprovsky I. 1978. Chustskaya kul'tura Fergani I Pamyatniki rannezheleznogo veka Srednei Azii. Unpublished doctoral thesis. Moscow State University, Moscow.
Zdanovich G. 1988. Bronzovy vek Uralo-Kazakhstanskikh stepey. Sverdlovsk: FAN.