PUBLISHED VERSION

Ingrid Sierp, Maciej Henneberg Can ancestry be consistently determined from the skeleton? Anthropological Review, 2015; 78(1):21-31

© 2015 Polish Anthropological Society. This work is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 3.0 License. (CC BY-NC-ND 3.0) Published version available at: http://dx.doi.org/10.1515/anre-2015-0002

PERMISSIONS http://degruyteropen.com/you/journal-author/copyright-and-license-statement/ http://degruyteropen.com/you/journal-author/repository-policy/

18 May 2016 http://hdl.handle.net/2440/97180 Ingrid Sierp, Maciej Henneberg Can ancestry be consistently determined from skeleton?

Anthropological Review • Vol. 78 (1), 21–31 (2015)

ANTHROPOLOGICAL REVIEW Available online at: www.degruyter.com Journal homepage: www.ptantropologiczne.pl

Can ancestry be consistently determined from the skeleton?

Ingrid Sierp, Maciej Henneberg

Biological Anthropology and Comparative Anatomy Research Unit, School of Medical Sciences, University of Adelaide, Adelaide 5005, Australia

Abstract: Although the concept of race has been thoroughly criticised in biological anthropology, forensic anthropology still uses a number of methods to determine the ‘race’ of a skeleton. The methods must be evaluated to see how effective they are given large individual variation. This study used 20 cases of skeletons of varied provenance to test whether the nine published methods of ‘race’ determination, using a range of various approaches, were able to consistently identify the ethnic origin. No one individual was identified as belonging to just one ‘major racial class’, e.g. European, meaning that complete consistency across all nine methods was not observed. In 14 cases (70%), various methods identified the same individ- ual as belonging to all three racial classes. This suggests that the existing methods for the determination of ‘race’ are compromised. The very concept of ‘race’ is inapplicable to variation that occurs between pop- ulations only in small ways and the methods are limited by the geographic population from which their discriminant functions or observations of morphological traits were derived. Methods of multivariate linear discriminant analysis, e.g. CRANID, are supposed to allocate an individual skull to a specific population rather than a ‘major race’. In our analysis CRANID did not produce convincing allocations of individual skeletons to specific populations. The findings of this study show that great caution must be taken when attempting to ascertain the ‘race’ of a skeleton, as the outcome is not only dependent on which skeletal sites are available for assessment, but also the degree to which the unknown skeleton’s population of origin has been investigated.

Key words: human skeletal identification; race; discriminant function analysis; non-metric variation

Introduction due to the nature of human variability (Kaszycka et al. 2009). Some elements The concept of assigning skeletal re- of variability can be attributed to genet- mains into a defined ‘race’ is problematic ic drift in small populations (Henneberg

Original Article: Received: September 4, 2014; Accepted for publication: November 19, 2014 DOI: 10.1515/anre-2015-0002 © 2015 Polish Anthropological Society Brought to you by | University of Adelaide Barr Smith Library Authenticated Download Date | 5/18/16 3:57 AM 22 Ingrid Sierp, Maciej Henneberg

2006; Rhine 1993) or to natural selection for determining ‘race’ from skeletal re- in antecedent populations, however, with mains. Other such studies exist, however the increasing levels of migration and the their methods are comparatively similar resultant gene flow, levels of variability to those mentioned previously. In this of specific racial identifiers are increasing study nine methods of ‘race’ determina- within local populations. In general, the tion, chosen to represent their common concept of ‘race’ is ill suited to the study types, have been applied to 20 cases of of human variation because this varia- skeletons requiring racial identification tion is quasi-continuous and most of it as if they were a subject of forensic in- occurs among individuals in the same vestigation. Thus each of the skeletons populations while only about a quarter is to be considered a separate case. The is attributable to geographic distribution ability of the nine methods to consist- of people (Brace 2005; Henneberg 2010; ently determine the ‘race’ of a skeleton Lewontin 1976). Anthropological meth- was evaluated. ods available to identify a race are derived from specific skeletal samples represent- Method ing particular geographic groups. Thus, discriminant functions or morphologi- The nine methods (Table 1) were ap- cal categories recommended by a meth- plied to 20 skeletons held by The Ray od may not be adequate to identify the Last Laboratory at The University of Ad- racial affinity of a skeleton of unknown elaide. The origin of these skeletons is provenance (Iscan 1983; Iscan & Steyn unknown, however, they are most likely 1999; Patriquin et al. 2002). Some mem- to come from two sources; (1) donated bers of a particular ‘race’ may share some skeletons of Australians of European de- features within a population, such as scent with a slight possibility of Austral- the presence of a wide nasal aperture in ian Aboriginal admixture, and (2) teach- African Americans; however, individual ing skeletons bought by the University variability can often cause an overlap of from India early in the 20th century. such features with other ‘races’. The dis- The ability of the methods to consist- crimination ability of available methods ently determine the ‘race’ of an individual should be investigated in skeletal sam- was evaluated in three different ways. (1) ples from outside the geographic popu- Counting in how many cases the major- lations, for which the original methods ity of methods gave the same result, i.e. were derived, to be able to assess their at least five of the nine methods consist- general usability. In the literature there ently identified the skeleton as belonging is a wide variety of methods used to as- to the same ‘race’. (2) Counting in how certain the ancestry of skeletal samples. many cases results were fully ambiguous, These methods used craniometric (Giles i.e. the skeleton was identified as belong- & Elliot 1962; Gill 1984; Wright 2008) ing to one ‘race’ by the same number of and morphologic (Bass 1995; Brues methods as belonging to the other ‘race’. 1990; Gill 1998; Rhine 1993) assess- (3) Counting in how many cases meth- ments of the skull and metric assess- ods identify the skeleton as belonging at ment of the pelvis (Iscan 1983; Patriquin the same time to all three of the ‘racial et al. 2002). These skeletal elements are classes’, with at least one method in each commonly accepted as the gold standard class. The above categories are not mu-

Brought to you by | University of Adelaide Barr Smith Library Authenticated Download Date | 5/18/16 3:57 AM Can ancestry be consistently determined from skeleton? 23

Table 1. Methods used for the determination of ancestry in this study Method Description of method Possible Results Collection Used Author & Date Number 1 3× craniofacial indices White or Black/ Indi- Terry collection, Smithso- Gill 1984 an/Eskimo nian Institution as well as secondary sources 2 2× cranial discrimi- White, Black or Terry collection, St. Louis Giles & Elliot 1962 nant functions American Indian Todd collection, Cleve- land 3 11× morphological Caucasoid, Mongoloid Meta-analysis of previous Bass 1995 traits of skull or Negroid studies

4 20× morphological Caucasoid, Mongoloid Meta-analysis of previous Rhine 1993 traits of skull or Negroid studies 5 12× morphological White, Black or East Meta-analysis of previous Gill 1998 traits of skull Asian/American Indi- studies an/Polynesian 6 Morphological study of Caucasoid, Mongoloid Peabody Museum Col- Brues 1990 the nasal root or Negroid lection 7 4× Pelvic discriminant White or Black Terry collection, Smithso- Iscan 1983 functions nian Institution 8 4× Pelvic discriminant White or Black Dissection specimens, Patriquin et al. functions University of Pretoria 2002 Raymond Dart collection, Johannesburg 9 Multivariate statistics Many populations Howells’ public data set CRANID by Wright on cranial dimensions 2008

tually exclusive except for (1) and (2). the value ‘1’; an outcome of ‘White’ was Authors of each method gave somewhat given the value ‘2’; to increase precision different names for the ‘racial categories’ of analysis, the ‘Other’ class was divid- into which their method is supposed to ed into an outcome of ‘Mongoloid’ that classify and individual. For purposes of was given the value ‘3’ and all other out- comparing the methods we have grouped comes were given the value ‘4’. Consist- their results into three general classes; ency of the nine methods was then an- Black, White and Other. ‘Black’ includes alysed using a non-parametric One-way any determination pertaining to Sub-Sa- ANOVA (Kruskal-Wallis test). Inter-cor- haran African ancestry, ‘White’ includes relation between the nine methods was any determination pertaining to Euro- also analysed using the non-parametric pean ancestry, while ‘Other’ includes Spearman test. any determination that is pertaining to Asian, Amerindian, Indigenous Austral- Results ian and Oceanian ancestry. Outcomes of ‘racial’ determination Table 2 shows that no one individual was were transformed into numerical val- identified as belonging to only one- ‘ra ues for the purposes of statistical anal- cial class’. Twelve individuals fall under ysis. An outcome of ‘Black’ was given the ‘majority’ category (see ‘consistency’

Brought to you by | University of Adelaide Barr Smith Library Authenticated Download Date | 5/18/16 3:57 AM 24 Ingrid Sierp, Maciej Henneberg

12 other black/ stency Consi- White/ 7/1/1 8/1/0 4/2/3 4/3/2 6/1/2 - - 9 11 CRANID mensions Cranial Di Zalavar Hungarian medieval Berg Austria medieval South Austral ian Indigenous Patagonian North Japan Hokkaido 8 10 PDF 1) White 2) White 3) White 4) White 1) White 2) White 3) White 4) White 1) White 2) Black 3) Black 4) Black 1) Black 2) Black 3) Black 4) Black 1) White 2) White 3) White 4) White 2 7 9 PDF 1) White 2) White 3) White 4) White 1) White 2) White 3) White 4) White 1) Black 2) Black 3) Black 4) Black 1) Black 2) Black 3) Black 4) Black 1) White 2) White 3) White 4) White

6 8 root Nasal Mongo- loid Cauca- soid Mongo- loid Mongo- loid Mongo- loid

5 7 Cranial descriptive (morphological) traits –10/12 traits indicate White –2/2 traits indicate East American Indian Asian, or Polynesian –9/12 traits indicate White –1/12 traits indicate Black –2/12 traits indicate East American Indian Asian, or Polynesian –3/12 traits indicate White –4/12 traits indicate Black –5/12 traits indicate East American Indian Asian, or Polynesian –7/12 traits indicate White –4/12 traits indicate Black –1/12 traits indicate East American Indian Asian, or Polynesian –6/12 traits indicate White –1/12 traits indicate Black –5/12 traits indicate East American Indian Asian, or Polynesian ------4 6 traits (morphological) Cranial descriptive –12/19* traits indi cate Caucasoid –2/19* traits indi cate Negroid –5/19* traits indi cate Mongoloid –14/19* traits indi cate Caucasoid –4/19* traits indi cate Negroid –1/19* traits indi cate Mongoloid –8/20 traits indicate Caucasoid –5/20 traits indicate Negroid –7/20 traits indicate Mongoloid –10/20 traits indi cate Caucasoid –6/20 traits indicate Negroid –4/20 traits indicate Mongoloid –14/20 traits indi cate Caucasoid –3/20 traits indicate Negroid –3/20 traits indicate Mongoloid ------3 5 traits (morphological) Cranial descriptive –7/9* traits indi cate Caucasoid –1/9* traits indi cate Negroid –1/9* traits indi cate Mongoloid –6/9* traits indi cate Caucasoid –2/9* traits indi cate Negroid –1/9* traits indi cate Mongoloid –7/11 traits indi cate Caucasoid –2/11 traits indi cate Negroid –2/11 traits indi cate Mongoloid –8/11 traits indi cate Caucasoid –1/11 traits indi cate Negroid –2/11 traits indi cate Mongoloid –11/11 traits indi cate Caucasoid - - - - - 2 4 CDF 1) Black 2) Ameri can Indian 1) Black 2) Ameri can Indian 1) White 2) Ameri can Indian 1) Black 2) Ameri can Indian 1) Black 2) Ameri can Indian 1 1 3 CDF White White White White White 2 1 2 3 B A Label anthropology texts. Numbers 1) 2) etc. in columns indicate results of different discriminant function equations by the same authors anthropology texts. 2 3 4 5 1 1 Methods Specimen No Table 2. Table Results of determination of ‘race’ of a series of male skeletons, of different provenance, by a number of methods recommended by forensic

Brought to you by | University of Adelaide Barr Smith Library Authenticated Download Date | 5/18/16 3:57 AM Can ancestry be consistently determined from skeleton? 25 12 7/1/1 4/4/1 4/4/1 4/5/0 6/2/1 - - 11 London medi eval Egypt 26–30 dynasty Punjab East Teita African Sydney Abo riginal 10 1) White 2) White 3) White 4) White 1) White 2) Black 3) Black 4) Black 1) Black 2) Black 3) Black 4) Black 1) White 2) Black 3) Black 4) Black 1) White 2) Black 3) Black 4) Black 9 1) White 2) White 3) White 4) White 1) Black 2) Black 3) Black 4) Black 1) Black 2) Black 3) Black 4) Black 1) Black 2) Black 3) Black 4) Black 1) Black 2) White 3) White 4) White

8

Mongo- loid Ne- groid Cauca- soid Ne- groid Ne- groid 7 –6/10* traits indicate White –4/10* traits indicate American East Asian, Indian or Polynesian –4/12 traits indicate White –2/12 traits indicate Black –6/12 traits indicate East American Indian Asian, or Polynesian –8/12 traits indicate White –3/12 traits indicate Black –1/12 traits indicate East American Indian Asian, or Polynesian –6/12 traits indicate White –2/12 traits indicate Black –4/12 traits indicate East American Indian Asian, or Polynesian –7/12 traits indicate White –2/12 traits indicate Black –3/12 traits indicate East American Indian Asian, or Polynesian - - - - - 6 –9/15* traits indi cate Caucasoid –2/15* traits indi cate Negroid –4/15* traits indi cate Mongoloid –11/20 traits indi cate Caucasoid –5/20 traits indicate Negroid –4/20 traits indicate Mongoloid –7/20 traits indicate Caucasoid –8/20 traits indicate Negroid –5/20 traits indicate Mongoloid –9/20 traits indicate Caucasoid –7/20 traits indicate Negroid –4/20 traits indicate Mongoloid –13/20 traits indi cate Caucasoid –5/20 traits indicate Negroid –2/20 traits indicate Mongoloid ------5 –8/8* traits indi cate Caucasoid –8/11 traits indi cate Caucasoid –2/11 traits indi cate Negroid –1/11 traits indi cate Mongoloid –7/11 traits indi cate Caucasoid –2/11 traits indi cate Negroid –2/11 traits indi cate Mongoloid –8/11 traits indi cate Caucasoid –1/11 traits indi cate Negroid –2/11 traits indi cate Mongoloid –8/11 traits indi cate Caucasoid –2/11 traits indi cate Negroid –1/11 traits indi cate Mongoloid - - - - - 4 1) Black 2) Ameri can Indian 1) Black 2) Ameri can Indian 1) Black 2) Ameri can Indian 1) Black 2) Ameri can Indian 1) White 2) Ameri can Indian 3 White White White White White 2 4 5 6 7 AM1 6 7 8 9 1 10

Brought to you by | University of Adelaide Barr Smith Library Authenticated Download Date | 5/18/16 3:57 AM 26 Ingrid Sierp, Maciej Henneberg 12 3.5/1.5/4 7.5/1.5/0 3.5/3.5/2 6.5/2.5/0 4/5/0 - 11 Eskimo Green land Italian Post-Medieval Egypt 26–30 dynasty UK Poundbury Zulu South African 10 1) White 2) White 3) Black 4) Black 1) White 2) White 3) Black 4) Black 1) Black 2) Black 3) Black 4) Black 1) White 2) Black 3) White 4) Black 1) Black 2) Black 3) Black 4) Black 9 1) Black 2) Black 3) Black 4) Black 1) White 2) White 3)White 4) White 1) Black 2) Black 3) Black 4) Black 1) Black 2) Black 3) Black 4) Black 1) White 2) Black 3) Black 4) Black

8

Mongo- loid Cauca- soid Mongo- loid Ne- groid Ne- groid 7 –4/12 traits indicate White –3/12 traits indicate Black –5/12 traits indicate East American Indian Asian, or Polynesian –7/12 traits indicate White –4/12 traits indicate Black –1/12 traits indicate East American Indian Asian, or Polynesian –3/12 traits indicate White –7/12 traits indicate Black –2/12 traits indicate East American Indian Asian, or Polynesian –8/12 traits indicate White –1/12 traits indicate Black –3/12 traits indicate East American Indian Asian, or Polynesian –7/12 traits indicate White –1/12 traits indicate Black –4/12 traits indicate East American Indian Asian, or Polynesian - - - 6 –8/20 traits indicate Caucasoid –3/20 traits indicate Negroid –9/20 traits indicate Mongoloid –14/20 traits indi cate Caucasoid –3/20 traits indicate Negroid –3/20 traits indicate Mongoloid –8/20 traits indicate Caucasoid –8/20 traits indicate Negroid –4/20 traits indicate Mongoloid –12/20 traits indi cate Caucasoid –5/20 traits indicate Negroid –3/20 traits indicate Mongoloid –12/20 traits indi cate Caucasoid –5/20 traits indicate Negroid –3/20 traits indicate Mongoloid ------5 –6/11 traits indi cate Caucasoid –2/11 traits indi cate Negroid –3/11 traits indi cate Mongoloid –9/11 traits indi cate Caucasoid –1/11 traits indi cate Negroid –1/11 traits indi cate Mongoloid –6/11 traits indi cate Caucasoid –4/11 traits indi cate Negroid –1/11 traits indi cate Mongoloid –10/11 traits indi cate Caucasoid –1/11 traits indi cate Mongoloid –9/11 traits indi cate Caucasoid –2/11 traits indi cate Mongoloid - - - - - 4 1) White 2) Ameri can Indian 1) Black 2) Ameri can Indian 1) White 2) Ameri can Indian 1) White 2) Ameri can Indian 1) Black 2) Ameri can Indian 3 White White White White White 2 041 HS- AM2 AM3 AM4 AM5 1 11 12 13 14 15

Brought to you by | University of Adelaide Barr Smith Library Authenticated Download Date | 5/18/16 3:57 AM Can ancestry be consistently determined from skeleton? 27 Some * 12 6/3/0 4/3/2 4/3/2 5/3/1 5/3/1 Actual skeletons of skeletons Actual 11 1–5 Italian Post-Medieval New Tolai Britain Berg Austria Medieval Punjab Punjab 10 1) White 2) Black 3) Black 4) Black 1) White 2) Black 3) Black 4) Black 1) White 2) Black 3) Black 4) Black 1) Black 2) Black 3) Black 4) Black 1) White 2) Black 3) Black 4) Black 9 1) White 2) Black 3) Black 4) Black 1) White 2) Black 3) Black 4) Black 1) Black 2) Black 3) Black 4) Black 1) Black 2) Black 3) Black 4) Black 1) Black 2) Black 3) Black 4) Black

8 Cauca- soid Mongo- loid Mongo- loid Cauca- soid Cauca- soid Skeletons of individuals of unknown sex, age and race from the of individuals unknown sex, Skeletons AM# 7 A skeleton of an adult from the Abbie Museum of Anatomy, University of an adult from the Abbie Museum Anatomy, A skeleton A –6/12 traits indicate White –2/12 traits indicate Black –4/12 traits indicate East American Indian Asian, or Polynesian –5/12 traits indicate White –3/12 traits indicate Black –4/12 traits indicate East American Indian Asian, or Polynesian –3/10* traits indicate White –7/10* traits indicate American East Asian, Indian or Polynesian –9/11* traits indicate White –2/11* traits indicate Black –7/12 traits indicate White –2/12 traits indicate Black –3/12 traits indicate East American Indian Asian, or Polynesian ------Boxed half-skeletons of unknown sex, age and race from the University of Adelaide; age and race from the University of Adelaide; of unknown sex, Boxed half-skeletons 6 HS–0##/SC–0## –17/20 traits indi cate Caucasoid –1/20 traits indicate Negroid –2/20 traits indicate Mongoloid –12/20 traits indi cate Caucasoid –2/20 traits indicate Negroid –6/20 traits indicate Mongoloid –12/15* traits indi cate Caucasoid –3/15* traits indi cate Mongoloid –10/19* traits indi cate Caucasoid –5/19* traits indi cate Negroid –4/19* traits indi cate Mongoloid –14/20 traits indi cate Caucasoid –1/20 traits indicate Negroid –5/20 traits indicate Mongoloid ------PDF – Pelvic Discriminant Function; Discriminant Function; PDF – Pelvic 2 5 –8/11 traits indi cate Caucasoid –3/11 traits indi cate Mongoloid –7/11 traits indi cate Caucasoid –4/11 traits indi cate Mongoloid –7/8* traits indi cate Caucasoid –1/8* traits indi cate Mongoloid –8/10* traits indi cate Caucasoid –2/10* traits indi cate Mongoloid –10/11 traits indi cate Caucasoid –1/11 traits indi cate Mongoloid - - - - - 4 1) Black 2) Ameri can Indian 1) Black 2) Ameri can Indian 1) Black 2) Ameri can Indian 1) Black 2) Ameri can Indian 1) Black 2) Ameri can Indian 3 A skeleton of known sex, age and race from a donated cadaver – white male Australian who died at age XX years; B White White White White White 2 SC- SC- 038 039 012 002 003 HS- HS- HS- 1 16 17 18 19 20 CDF – Cranial Discriminant Function; CDF – Cranial Discriminant Function; 1 of Adelaide; age and race used as teaching aids in dissection rooms; individuals of unknown sex, University of Adelaide; Abbie Museum of Anatomy, due to their absence on the skeleton. traits were excluded

Brought to you by | University of Adelaide Barr Smith Library Authenticated Download Date | 5/18/16 3:57 AM 28 Ingrid Sierp, Maciej Henneberg column for specimen numbers 1, 2, 5, 6, importance when no combination of the 9, 10, 12, 14, 15, 16, 19, 20), where in nine methods was able to consistently most cases skeletons were identified as identify a single skeleton as belonging to being White (specimens 9 and 16 were just one and the same ‘racial group’. identified as black). Three individuals fall The ability of methods of forensic an- under the ‘ambiguous’ category (speci- thropology to determine ‘race’ from the men numbers 7, 8, 13), where in all cases skeleton has been questioned by Brues skeletons were identified as being equal- (1992). She states that paradoxical di- ly White and Black. Fourteen individuals agnoses of ‘race’ from the skull may be were identified as belonging to all three due to the use of a subset of a single pop- of the racial classes by at least one meth- ulation to represent a large geographic od in each group (specimen numbers 1, area. She does, however, commend the 3, 4, 5, 6, 7, 8, 10, 11, 13, 17, 18, 19, 20). work of Giles and Elliot (1962) describ- A non-parametric analysis of the ing their method as the ‘standard’ of ra- transformed data found a significant dif- cial determination. Contrarily, Snow et ference between the results of the nine al. (1979) found that in a collection of different methods (p<0.0001). Non- 42 skulls from White, Black and Indian para­metric analysis of the inter-corre- Americans, the racial origin of only 30 lation between the nine methods found skulls was identified correctly using the that only two methods (methods 7 and 8) Giles and Elliot (1962) method. They displayed significant correlation (r=0.56, concluded that the geographically limit- p=0.01). ed crania used to develop the discrimi- nant functions were not representative Discussion of the crania of present-day Americans. This concept can be applied to all the The combination of the nine methods methods used in this study, as the ori- used in this study failed to produce ful- gin of the 20 skeletons may not have ly consistent identification of the ances- matched with the populations used to try of even a single skeleton. Although, derive the methods. Multivariate discri- with a larger sample size, or with yet minant function approaches, such as that ano­ther method, it may have been pos- of Wright (2008), proposed to improve sible to find an individual who belonged precision of identification of ancestry, only to one ‘racial group’; such an event however, they do not seem to perform would still be a small portion of all cases. significantly better as evidenced by our Were any of the skeletons studied here results and those of Kallenberger and Pil- an actual forensic case, the fact that in brow (2012) who found that the CRA- none of the instances a fully consistent NID program was only able to accurately result has been achieved is disturbing. It assign 39% of specimens to geographi- indicates that the reliability of the meth- cally closest matching reference samples. ods is below 5% level while in criminal No single individual was identified as cases the proof required is that “beyond belonging to only one ‘racial group’. It is reasonable doubt”. The ethnic origins of concerning that nine methods of forensic the skeletons used in this study where anthropology, whose sole purpose is to not precisely known, with one exception ascertain the ‘race’ of an individual, did (specimen 2), however, this is of little not yield consistent results. In 14 cases

Brought to you by | University of Adelaide Barr Smith Library Authenticated Download Date | 5/18/16 3:57 AM Can ancestry be consistently determined from skeleton? 29

(70%), individuals were identified as be- praorbital ridge. Despite this, the ability longing to all three racial classes. Speci- of the methods to determine the ‘race’ of men 18 was identified as White by four this individual was ambiguous at best, methods (Bass 1995; Gill 1984; Rhine with results showing that three and a half 1993; Wright 2008), Black by three meth- methods indicated White; three and a half ods (Giles & Elliot 1962; Iscan 1983; indicated Black, and two indicated either Patriquin et al. 2002) and either Asian, Asian, Amerindian, Indigenous Austral- Amerindian, Indigenous Australian or ian or Oceanian. This suggests that de- Oceanian by two methods (Brues 1990; termination of ‘race’ through individual Gill 1998). Specimen 10 was found to be morphological features of the skull (Bass majority White (six of the nine methods) 1995; Gill 1998; Rhine 1993) is limited however the CRANID software (Wright by the specific sites available for exami- 2008) identified this individual as- be nation. If, in the case of specimen 13, re- ing a Sydney Aboriginal. Results such as mains were fragmented and the mid-fa- these show that the range of inconsist- cial region was absent, the individual encies found in assignment of ‘race’ does may have been identified as being White; not improve using more sophisticated a conclusion that would not be accepted software. had the skull been fully intact. The accurate determination of ‘race’ Multivariate metric methods based is virtually impossible with distribution on linear discriminant analysis (LDA) of human variation within and between such like FORDISC (Ousley et al. 2009) populations. For this reason, no-matter or CRANID (Wright 2008) also have how sophisticated the method, there is their fallacies. If even one anthropo- no way to consistently identify an indi- metric point (out of the 29 necessary) vidual as belonging to one specific ‘race’. is missing, say due to a fractured skull, In court cases it is advisable to abstain CRANID will not work. When less than from stating the ‘race’ of a skeleton, even 21 measures, available in FORDISC are if desired by the court, because a mistake used, results are not convincing, too in assignment is likely and it will com- (Williams et al. 2005). Wright (2008) promise the proceedings. If an ‘ethnic’ states that results of the distributable identification is required, it is better to version of CRANID can only be taken as base it on the evidence of lifestyle, such reliable if the most likely sample popu- as tooth wear or limb characteristics than lation has a high probability and if the on any ‘racial’ characters. sample populations that immediately According to the discriminant func- follow are consistently from contiguous tion of Gill (1984) all 20 specimens in geographical areas. In our analysis of 20 this study were White. This is question- skulls, CRANID returned a result ‘poorly able as only 12 specimens were consist- catered for’ for eleven of our specimens ently found to be White by a majority of (Specimens 1, 2, 3, 5, 7, 11, 13, 15, 16, the methods. In addition to this, speci- 17, 18). Wright (2008) states that this men 13 displayed morphological features lack of fit may be due to measurements congruent with an African American being wrongly made or entered, the ge- or Australian Aboriginal person. These ographical area from which these indi- features included a wide nasal aperture, viduals came is poorly catered for in the significant prognathism and a heavy- su database, the individuals being morpho-

Brought to you by | University of Adelaide Barr Smith Library Authenticated Download Date | 5/18/16 3:57 AM 30 Ingrid Sierp, Maciej Henneberg logically atypical, or the individuals are Instances of such identification may be of mixed ancestry. This however, does successful in particular local populations not help to understand what specifi­cally created by recent migrations. caused an ambiguous result for each of our 11 abovementioned specimens. In Acknowledgments the present day forensic applications it is likely that increased number of mixed We would like to thank W Howells for ancestry individuals may appear due to his generosity in making his craniomet- an increase in migration, thus limiting ric data publicly available. Thanks are the applicability of metric methods. also due to Dr J Kumaratilake for improv- Statistical analysis found a significant ing the English of the manuscript. difference between the results of the nine methods. This indicates that, statistically, Authors’ contributions the methods are not consistent. In a prac- tical sense, this means that forensic inves- IS and MH were responsible for project tigators using one, or even a few of these design. IS made measurements and de- methods, will be obtaining results differ- termined ancestry. IS and MH conducted ent from investigators who may be using statistical analysis, made conceptual con- a different combination of methods. tributions and wrote the manuscript. A number of methods of forensic an- thropology were unable to consistently Conflict of interest determine the ‘race’ of any of the 20 skel- etons. Racial determination is an identi- The authors declare that there is no con- fier of an individual’s uniqueness, regu- flict of interest. larly sought after by law enforcement in cases of discovery of skeletal remains. We Corresponding author have shown that even with 20 non-frag- mented sets of skeletal remains none Ingrid Sierp could be consistently placed into a single e-mail address: racial category. Individual variability may [email protected] have played a significant role leading to inconsistency of the results found in this References study, which further confirms the ideas of Brace and Ryan (1980), Henneberg Bass WM. 1995. Human Osteology: A Labo- (2010) and Lewontin (1976); that most ratory and Field Manual. Fourth edn. Co- human variation occurs between individ- lumbia: Missouri Archaeological Society. uals of the same population rather than Brace CL. 2005. “Race” is a four-letter word: the genesis of the concept. Oxford: Ox- being attributable to geographic distribu- ford University Press. tion. Since the majority of the biological Brace CL & Ryan AS. 1980. Sexual Dimor- variation in the human species occurs phism and Human Tooth Size Differences. among individuals with the minority be- J Hum Evol 9(1): 417–435. ing due to geographic differences (Brace Brues AM. 1990. The once and future diagno- 2005; Henneberg 2010; Lewontin 1976), sis of race. In: GW Gill & S Rhine, editors. it seems impossible to construct a pre- Skeletal Attribution of Race. Albuquer- cise method of ‘racial’ identification. que: Maxwell Museum of Anthropology.

Brought to you by | University of Adelaide Barr Smith Library Authenticated Download Date | 5/18/16 3:57 AM Can ancestry be consistently determined from skeleton? 31

Brues AM. 1992. Forensic Diagnosis of Race- Kaszycka KA, Strkalj G & Strzalko J. 2009. General Race vs Specific Populations. Soc Current views of European anthropolo- Sci Med 34(2): 125–128. gists on race: Influence of educational and Giles E & Elliot O. 1962. Race identification ideological background. Am Anthropol from cranial measurements. J Forensic Sci 111(1): 43–56. 7(1): 147–157. Lewontin RC. 1976. The Fallacy of Biological Gill GW. 1984. A forensic test case for a new Determinism. The Sciences 16(2): 6–10. method of geographical race determina- Ousley S, Jantz R & Freid D. 2009. Under- tion. In: TA Rathbun & JE Buikstra, editors. standing race and human variation: Why Human Idnetification: Case Studies in Fo- forensic anthropologists are good at iden- rensic Anthropolgy. Springfield: Thomas. tifying race. Am J Phys Anthropol 139(1): Gill GW. 1998. Craniofacial criteria in the 68–76. skeletal attribution of race. In: KJ Reichs, Patriquin ML, Steyn M & Loth SR. 2002. Met- editors. Advances in the identification of ric assessment of race from the pelvis in human remains. Springfield: Charles C. South Africans. Forensic Sci Int 127(1): Thomas. 293–317. 104–113. Henneberg M. 2006. The rate of human mor- Rhine S. 1993. Skeletal Criteria for Racial At- phological microevolution and taxonomic tribution. NAPA Bulletin 13(1): 54–67. diversity of hominids. Studies in Histori- Snow CC, Hartman S, Giles E & Young FA. cal Anthropology 4(1): 49–59. 1979. Sex and Race Determination of Cra- Henneberg M. 2010. The Illusive Concept of nia by Calipers and Computer: A Test of Human Variation: Thirty Years of Teach- the Giles and Elliot Discriminant func- ing Biological Anthropology on Four tions in 52 Forensic Cases. Washington, Continents. In: G Strkalj, editors. Teach- DC. ing Human Variation: Issues, Trends and Williams F, Belcher R & Armelagos G. 2005. Challenges. New York: Nova Science Pub- Forensic misclassification of Ancient Nu- lishers. bian crania: Implications for assumptions Iscan MY. 1983. Assessment of Race From about human variation. Current Anthro- the Pelvis. Am J Phys Anthropol 62(1): pology 46(1): 340–346. 205–208. Wright R. 2008. Detection of likely ancestry Iscan MY & Steyn M. 1999. Craniometric De- using CRANID. In: M Oxenham, editors. termination of Population Affinity in South Forensic approaches to death, disaster Africans. Int J Leg Med 112(1): 91–97. and abuse. Queensland: Australian Acha- Kallenberger L & Pilbrow V. 2012. Using demic Press. CRANID to test the population affinity of known crania. J Anat 22(5): 459–464.

Brought to you by | University of Adelaide Barr Smith Library Authenticated Download Date | 5/18/16 3:57 AM