68 Section HUMAN DIVERSITY adult part of the whole population were generated. <...> Our analysis of metric and nonmetric facial variation in Bedouin subtribes revealed a number of signifi cant differences (in transverse dimensions and height of the face, eye and hair pigmentation, and hair form). <...> Various Bedouin tribes, then, differ in important features of appearance. <...> Results of visual analysis of composite portraits are comparable with biometric data. <...> For example, CP of the homogeneous Muzeina tribe and the tribal group “others” are consistent with metric information; CP of Gebelia subtribes of patchy origin visualize and complete the numeric information. <...> The profi le portrait adds information on a number of features: height and overall profi le of nasal bridge, morphology of supraorbital and chin areas. <...> Key words: ethnic anthropology, composite portraits, South Sinai Bedouins, facial morphology Contact information: Chumakova Anna, e-mail: achumakova@mail.ru. <...> CEPHALOMETRIC VARIATION AMONG THE TURKMEN OF SOUTHERN RUSSIA Dubova Nadezhda Institute of Ethnology and Anthropology, Russian Academy of Sciences, Moscow, Russia Cephalometric data on Turkmen males (N=240) living in the Stavropol and Astrakhan Provinces of Russia since the XVII century are analyzed (the material was collected by this author and O. Babakov in 1987 in eight settlements of the Stavropol Province and in two in the Volga delta). <...> Thirteen traditional head and face dimensions were included. <...> All statistics were performed using Statistica 8.0 software. <...> The analysis of variance showed no signifi cant differences between the Chovdur, Igdir, Abdal and Suyundzhadzhi tribes (Wilks’ lambda = 0.712135; F=1.53). <...> Differentiation between villages is stronger (Wilks’ lambda = 0.405974; F=1.80). <...> Discrimination of the total population by either criteria (“tribe” and “village”) shows no effect for total head and face measurements. <...> Differences concern mainly details of facial morphology (tribes differ in zygomatic and nasal breadth; villages, in these features plus minimal frontal breadth, zygomatic breadth, and upper lip height). <...> Based on the above traits, correct tribe attribution is possible only in 48.75% of individuals ranging from 39.5% in Abdal to 84.6% in Suyundzhadzhi. <...> Correct village attribution was possible in 37.9% (Funtovo, 60.4%;Sharahalsun, 51.3%). <...> Variation among all the Turkmen tribes (N=1064 individuals) is greater than in southern Russia alone <...>