A Revised Method of Sexing the Human Innominate Using Phenice’s Nonmetric Traits and Statistical Methods Alexandra R. Klales, 1 * Stephen D. Ousley, 2 and Jennifer M. Vollner 3 1 Department of Anthropology, University of Manitoba, Manitoba, MB R3T 5V5, Canada 2 Department of Applied Forensic Sciences, Mercyhurst University, Erie, PA 16546 3 Department of Anthropology, Michigan State University, East Lansing, MI 48824 KEY WORDS phenice traits; pelvis; sex estimation; nonmetrics ABSTRACT The traits of the pubis described by Phe- nice (Am J Phys Anthropol 30 (1969) 297–302) have been used extensively by physical anthropologist for sex esti- mation. This study investigates all three of Phenice’s characteristics in an approach similar to Walker’s (Am J Phys Anthropol 136 (2008) 39–50) study using observa- tions from the cranium and mandible. The ventral arc, the subpubic contour, and the medial aspect of the ischio- pubic ramus were scored on a five-point ordinal scale from a sample of 310 adult, left innominates of known ancestry and sex from the Hamann-Todd Human Osteo- logical Collection and the W.M. Bass Donated Skeletal Collection. Four observers with varying levels of experi- ence blindly scored each trait using new descriptions and illustrations adapted from those originally created by Phenice. The scores were then analyzed with ordinal logistic regression. Using all three traits for sex classifica- tion, the mean correct classification rate was 94.5% cross- validated for experienced observers. Intra- and interob- server error in trait scoring was low for all three traits and agreement levels ranged from moderate to substan- tial. Tests of the method on an independent validation sample provided a classification accuracy of 86.2%. This revision of the Phenice (Am J Phys Anthropol 30 (1969) 297–302) technique is a reliable and valid method of sex estimation from the human innominate that meets the Daubert criteria for court admissibility. Am J Phys Anthropol 149:104–114, 2012. V V C 2012 Wiley Periodicals, Inc. Accurate sex estimation is critical for forensic anthro- pologists establishing the biological profile of unidentified individuals and also for bioarcheologists recreating the demographics of past populations. Historically, the bones most frequently used for sex estimation have been the cra- nium and the bones of the pelvis. The innominate and the pubic bone in particular, has generally been accepted as the single best indicator of sex (Letterman, 1941; Phenice, 1969; Stewart, 1979; Krogman and Isc ¸an, 1986; MacLaughlin and Bruce, 1986; Walker, 2005) due to differ- ences between males and females related to childbirth and locomotion. Because of these aforementioned differences, the innominate has been examined previously through multiple metric and nonmetric studies to develop methods of sex estimation for use by physical anthropologists. Numerous metric approaches have been developed for sex estimation in the innominate (Day and Pitcher- Wilmott, 1975; DiBennardo and Taylor, 1983; Patriquin et al., 2005; Bytheway and Ross, 2010), but nonmetric methods continue to be more frequently used for sex esti- mation due to ease of use and the inherent problems with defining measurement landmarks on the innominate (Walker, 2008; Klales, 2011). Interobserver errors using measurements of the innominate have been notoriously high because of poorly defined or difficult-to-locate mea- surement landmarks, though recent research by the authors has shown the utility of using discriminant func- tion analyses of interlandmark distances for sex estima- tion (Klales et al., 2009; Vollner, 2009). Furthermore, many metric methods of sexing have proven to be highly specific to the population from which the sample origi- nated and may not distinguish between the sexes in other populations (MacLaughlin and Bruce, 1986). Finally, apprehension for the use of statistical methods may be further limiting the widespread use of metric methods. On the contrary, nonmetric methods continue to be taught and utilized because of their practicality: they can be per- formed quickly and do not require specialized equipment. Nonmetric methods typically can be used even with frag- mented remains and are the suggested method of sex esti- mation for the pelvis in many popular introductory osteol- ogy textbooks (for example Stewart, 1979; Krogman and Isc ¸an, 1986; Bass, 2005; Byers, 2005). However, nonmet- ric methods are not without their limitations. Shortcom- ings of nonmetric methods in general, as summarized by Bruzek (2002: 158) include a ‘‘(1) high degree of observer subjectivity, (2) a lack of consistency in evaluation of traits, and (3) a strong dependence on the results of previ- ous experiences of the observer.’’ To these shortcomings, one could add that most nonmetric methods generally avoid the use of statistical methods for classification, which often optimize correct classification rates. This research was presented in part at the 78 th Annual Meeting of the American Association of Physical Anthropologists in Chicago (April 2009). *Correspondence to: Alexandra R. Klales, 1515 E. Woodbank Way, West Chester, PA 19380, United States. E-mail: alexandra.klales@g- mail.com Received 9 October 2011; accepted 4 May 2012 DOI 10.1002/ajpa.22102 Published online 19 June 2012 in Wiley Online Library (wileyonlinelibrary.com). V V C 2012 WILEY PERIODICALS, INC. AMERICAN JOURNAL OF PHYSICAL ANTHROPOLOGY 149:104–114 (2012)