TECHNICAL NOTE Sophie A. Veyre-Goulet, 1 M.Sc.; Catherine Mercier, 2,3,4 Ph.D.; Olivier Robin, 1 Ph.D.; and Claude GuØrin, 3,5 Ph.D. Recent Human Sexual Dimorphism Study Using Cephalometric Plots on Lateral Teleradiography and Discriminant Function Analysis ABSTRACT: The examination of skull sexual dimorphism has been the subject of numerous morphologic and craniometric studies, but the disad- vantage of these studies is that they are dependent on the experience of the operator and involve subjectivity. In 1996, a team from Taiwan refined the methods enabling the sex of an individual to be determined using cephalometric plots made from lateral teleradiography. To validate their work using a European population, 114 dry skulls (59 men and 55 women) were examined. Cephalometric plots were made on lateral teleradiography with an orthodontic software and 18 cephalometric variables were analyzed. Sex was determined with 95.6% accuracy using the 18 variables discriminant function. A subset of eight variables was selected and could predict sex with the same accuracy. In conclusion, it can be said that skull-sexing meth- ods using lateral teleradiography seem always suitable but the most indicative variables could differ relative to the ethnic population concerned. KEYWORDS: forensic science, physical anthropology, sex determination, skull, lateral roentgenographic cephalometrics, discriminant function analysis When human bones are discovered, the first question asked is ‘‘Are they male or female?’’ This is of interest in two fields: anthropology and forensic science. Among the bones of the human skeleton, the pelvis is the most determinant (1), but, because of its complex shape, it is delicate and often found in a very poor condi- tion. The skull, on the other hand, is usually better preserved and more readily exploitable (2). For this reason, many authors have concentrated on the skull for determining the sex of an individual. Several methods have been refined. Initially, morphologic examinations were developed. These were qualitative methods making use of descriptive criteria (3–5). The disadvantage of such methods is that they lack objectivity and are dependent on the experience of the operator (6). Craniometric examination was also developed, and this is a quantitative examination involving taking direct measurements of the skull (7,8). In 1958, CØballos and Rentschler (9) were the first to work on teleradiography for determining sex from the skull. Other authors succeeded them (10–15) using both posteroanterior and lateral teleradiography. In 1996, Hsiao et al. (16) conducted lateral teleradiography on a sample of 50 males and 50 females from Taiwan. Using 18 variables from cephalometric plots obtained from the teleradiogra- phy plates, they claimed to be able to determine the sex of an indi- vidual with 100% accuracy. Furthermore, of the 18 variables, three are more indicative than the others are, and the authors say that they can determine the sex of a subject to 98% accuracy by using these three variables alone. At the end of their study, the Taiwanese authors suggest their method should be tested on a different ethnic population. The pur- pose of our research therefore was to validate the Taiwanese method on a European population. Materials and Methods The sample we studied comprised 114 dry skulls (59 men and 55 women) from the ‘‘Museum d’Histoire Naturelle’’ in Lyon. This collection dates from the end of the 19th century and comes from deceased people in hospital and the dead body where not claimed by the family. For each body, sex and age were listed in an index. All the subjects came from the Rhône-Alpes region in France. The subjects selected were aged between 20 and 55 at the time of their death, in other words, after puberty and before signs of senility appeared (4). Lateral teleradiography was conducted on each skull. The plates were made using radiography equipment that established a focal- plate distance of 4 m. They were digitized with Epson Expression 1640XL Scanner (Epson America, Inc., Long Beach, CA). Then, the cephalometric traces were made by an orthodontic software. Nineteen cephalometric points were identified (Table 1) which enabled the identification of 18 cephalometric variables as described in Hsiao et al. (16) (Table 2). There were eight angles (°), nine linear measurements (mm), and a proportional measure- ment (%). The same operator conducted all the cephalometric plots. 1 UFR d’Odontologie, UniversitØ Claude Bernard Lyon 1, France. 2 Hospices Civils de Lyon, Service de Biostatistiques, Lyon, F-69424, France. 3 UniversitØ de Lyon, UniversitØ Lyon 1, Villeurbanne, F-69622, France. 4 CNRS, UMR 5558, Laboratoire Biostatistique SantØ, Pierre-BØnite, F-69495, France. 5 UMR 5125 «palØoenvironnement et palØobiosphre», CNRS, Lyon, France. Received 28 July 2007; and in revised form 18 Oct. 2007; accepted 18 Nov. 2007. J Forensic Sci, July 2008, Vol. 53, No. 4 doi: 10.1111/j.1556-4029.2008.00759.x Available online at: www.blackwell-synergy.com 786 Ó 2008 American Academy of Forensic Sciences