RACE DETERMINATION FROM THE POSTCRANIAL SKELETON.

(1)

Race

Determİnation from the Postcranİal

Skeleton

MEHMET YAŞAR İş CAN

Department of Antlıropology, Florida Atlantic University, Boca Raton, FL, USA

KAFATASı DlŞINDAKİ KEMİKLERDEN IRK SAPTANMASI

Özet

Bu çalışmanın amacı, insan iskelet kalıntılarındaki ırksal farklılıkları tesbit etmek amacıyla

kullanılabilecek çeşitli diskriminant fonksiyon formüllerini geliştirmektir. Cleveland Doğa Tarihi Müzesi (Museum of Natural History)'ndeki Hamann-Todd kolleksiyonuna ait 224 ömek kullanıldı. III beyaz (56 erkek, 55 kadın) ile 113 zenci (52 erkek, 61 kadın) pelvis, femur ve tibialannda 21 osteometrik boyutun ölçümleri yapıldı. Tek kemik ölçümü kadar kombine ölçümlerin kullanılmasıyla 5 diskriminant fonksiyon formülü geliştirildi. Pelvisten alınan tüm boyutların fonksiyonlannın femur veya tibiadan elde edilenlerle kombinasyonu halinde, geçerliliği en üst düzeyde (%95 oranında doğru sınıflandırma) olan veriler elde edildi. Cinsiyet ayınmında, erkeklerin uzun kemiklerinin kadınlarinkilere oranla daha verimli olduğu saptandı. Bu çalışmanın sonuçlan ile kafatası ve pelvis boyutlarının ölçümüne dayanan incelemelerin önemli derecede uyumlu olduğu görüldü. Ancak, Hamann-Todd örneklerindeki sağlık, fiziksel özellikler ve ırk kompozisyonu-nun, Kuzey Amerika'nın farklı yörelerindeki çağdaş toplumlarla aynı olmadığının unutulmaması gerektiği

vurgulandı.

Summary

The purpose of this paper is to develop various diseriminant function formulae to assess the racial affinity of human skeletal remains. The sample consisted of 224 speeimens from the Hamann-Todd Colleetion of the

Cleveland Museum of Natural History. Twenty-one osteometrie dimensions were taken from the pclvis, femur and tibia of III White s (56 males, 55 females) and 113 Blaeks (52 males,61 females). Five discriminant function formulae were derived using measurements from single as well as eombinations of bones. The results indicated that funetions incorporating dimensions from the pelvis in combination withthose from the femur or tibia provided the highest accuracy (as high as 95% correct Cıassification). It was observed that the long bone s of males showed more raeial differenees than females. The results of this research eompared favorably with other sıudies us ing skull or pelvie dimensions. However, it must be pointed out that the hcahh, physical characteristics, and racial composition of the Hamann-Todd speeimens may not be the same as those of eontemporary populations from different rcgions of NOM America.

Kcy words: Race determination - Pelvis· Femur - Tibia -Discriminantfunctionformulae

Adli Tıp Derg., 6, 129 - 140 (1990)

İ TIP DERGİSİ

Journal of Forensic Medicine

(2)

130 M.Y.1ŞCAN

INTRODUCTION

Determination of race from the postcranial skeleton has not beena major focus in

skeletal anthrapology un til recently (1 -3). Differences between races have been noted by

Derry

(4), Todd and Lindala (5), Howells and Hotelling (6) and Torpin (7) using the

pe

l

vis and

its

components and by Flower (8), Munter (9), Farrally and Moore (10) and

Flander

(11) using other bones and indexical relationships

between long

bones.

Assessment of the anterior curvature of the femur has

probably

been the most

eommonly used method

in

postcranial racial identification

for

forensic cases (12,13).

Only recentlY has

the

"objective" discriminant function approach been utilized to

determine race

from

the postcranial skeleton. Studies published include those on the

pelvis

(14,15) and on both the

innominate

bone and femur (16,17). While the

contribution of these studies

to physical

anthropology is important, all bu

t

one

require

a

minimum of a complete pelvic girdie (two innominate bones and sacrum) (14,15) or

complete innominate bone and femur (16). Moreover, in the case of the latter study 15

dimensions were essential to caleulate a discriminant function score. The study by

Schulter-Ellis

and Hayek (17) is the lone exception because only the acetabular rcgion

and the

pubis

are necessary for the analysis.

Since many forensic cases are incomplete and damaged, it is preferable that a set of

discriminant function models

be

developed to determine race using a minimal number of

measurements without seriously sacrificing accuracy of prediction. The purpose of this

paper is to present a number of discriminant function statistics that may be

u

sed to

de

t

erm

i

ne

r

ace fram the pelvis, femur, and tibia of American Blacks and Whites and to

introduce a graphic technique that can be used to assess the posterior probab

i

lity of

racial

identity. Furthermore,

this

paper details the development of the disc

r

iminant

function formulae which were first published in Krogman and

lşcan

(3).

MATERIALS AND METHODS

The sample consisted of 224 specimens from four race/sex groups (56 White males, 55 White females, 52 Black males, and 61 Black females) drawn from the Hamann-Todd Collcction houscd at the Cleveland .\1useum of Natural History, Ohio. Twenty-one measurements (in mm) were taken from each specimen; six from the pelvis, eight from the femur, and eight from the tibia. With a few exceptions, standard measurements were

used. The list and descripüon of the se dimensions are provided in Appendix ı.

Seven stepwise discriminant analyses were performed separately on each of the two sex groups using various combinations of bones and measurements, to produce a total of 14 different functions. Discriminant function analysis involves weighting a set of original measurements to produce one or more "discriminant function" scores. These scores represent weighted line ar composites of the original measurements that are created in a fashion designed to maximize the variance or dispersions between two or more criterion groups relative to the variance or dispersion of the scores within the groups. This procedure is similar to multiple regression analysis in that it involves a weighted linear sum of a number of different predictor variables. in the case of multiple regression analysis, the predictor scores are weighted so as to produce a "predicted" or "estimated" score that has a maximal correlation with a mpre or less continuous criterion variable.

(3)

Race Detcnnination from the Postcranial Skeleton 131

Discriminant funetion analysis differs in the respect thatthe eriterion variable is usually a categorieal one denoting group membcrship. In the presenı study the criıerion groups consisted of specimens that had becn labcled White or Black, since race was the eriterion variable. The initia! set of predicıor variables consisted of the 21 measurements obtained from each specimen.

SPSS was used to perfonn all discriminant function analyses(l8,19). All the analyses used stepwise method s that employcd the Wilks lambda minimization procedure. 'This procedure begins by sclecting the single origina! variable that provides the maximum discriminaıing power, i.e., the variable that has the ability to maximize the variance between the erilerion groups relative to the variance wiıhİn the groups. At each succcssive "stcp" a new variable is added to those aıready in the analysis. This new variable is selceted so as to maximize the multivariate F ratio between the eriterion groups, which has the same effect as mİnimizing

Wilks lambda. Additional variables are added in this stepwise fashion until further inclusİons fail to statistically improve the funelion's ability to diseriminate betwcen the groups.

RESULTS

Table

i

presents

the means, standard deviations, and univariate

F

ratios

for

21 dimensions. The data

in

Table

i

are based solely on those

speeimens

that had complete

sets

o

f

measuremenls for all variables. The F ratios in Table II apply to the d

i

fferenees

between

the raees

wilhin caeh

sex.

Raw (unstandardizcd) discriminant

funetion

eocfficicnts and constants for each

of

the

seven

analyses carried ou

t

separately

on maJes

and

females are

presented in Tables LI and III, rcspcetivcly. The eocffieicnts and eonstant

for the

first analysis

involving all 21 measures are, for example, presented in the fifst

eolumn

of

the table. Only nine variables were

seleeted

by the

stepwise function and

are

thus required

to

eompute the funetion

score.

None of the remaining 12 mcasurements

signifieantly improved the ability of the funetion to

correetly elassify

males on the

basis of raee.

To eompute, fOf example, a male's diseriminant funetion seore on

the

basis of

dimensions obtained from the pelvis plus the femur and tibia (F(p+f+t)) one would use

the weights and eonstant presented in

eolumn

ı

of Table II. Specimens having negative

scores on this

funetion

would be classified as White,

while

speeimens having posilive

seores

would classify as

Black.

The diseriminant funetion seores produeed by the raw

eoeffieien

t

and eonstant have amean of

zero

and a standard deviation

of

one for the

sample

of speeimens from whieh the funerion was derived. Speeimens having seores

close to zero are more likely to be miselassified than speeimens whose scores

are

higher.

Table IV prescnts the overall classifiealion aceuracy for eaeh sex/raee group.

Examinalion

of the male side of this table

indicaıes

that 94.0%

of

the 50 White males

were eorrectly classified

by

the funelion while this figure was 95.8% for the 48 Blaeks

in

the

sample. The average eonect classifieation among males, when using a

diseriminant funetion based on the partieular set

of

9 variables

chosen

from the pelvis,

(4)

Tablc ı. Mcans. standard dcvİaıions and unİvariale F·raıios. Whites IJlack~ Variabks (mm) Mearı

SO

Mean SD Fa N 49 Males

48 - - -

- -- ---",,--

-Biıliac br. 274.4 16.77 256.0 1195 32.73 b Transverse br. 124.6 7.41 114.8 8.23 38.29 b A·p. ht. 106.6 8.25 103.5 X.54 3.74 Innommaıe ht. 219.1 10.53 214.1 9.61 5.90 d mac br. 163.1 8.31 156.7

lI.n

n69c lschiopııbıc br. 123.5 5.81 117.5 6.45 22.96 b Femorallength 452.9 22.63 479.2 24.42 30.26 b Bicoııdylcr Iength 450.0 22.58 475.8 24.91 28.48 b

Femoral a-p. dİa. 29.1 3.72 30.0 3.03 2.52

Fcm. trans. dİa. 29.2 2.70 28.2

2.99

335

Fcm. di,t. ep. br. 83.0 4.11 83.2 3.83 0.1 1

Fem. mid. cİrc. 91.1 4.77 9U 5.88 003

Fem. he..1d dİa. 41.2 2.31 47.8 2.36 0.76

fıbi.al length 363.6 17.88 400.4 24.86 70,45 b

Tibial a.p. dia. 35.6 3.15 ]6.3 3.70 1.19

Tib. trans. dia. 26.3 2.55 28.4 3.39 12.02 c

Tib. prox. ep. br. 76.8 3.68 77.9 3.32 1.65

Tib. dis\. cp. br. 46.7 2.58 46.6 2.74 0.01

Circ. al Nutr. foram. 98.5 4.90 101.8 8.ü7 6.10 d

Mm. cıre. 76.6 3.70 79.8 6.83 8.59 c Midshaft cire. 84.8 4.88 89.\ 6.60 13.!-n b

-

- ---"._"".-

- -

-~,-""-""-""-""-

- - -

-""._""._""._"".-

- -

- " " . _ " " - " " - " , , - , , , , - , , , , -

-

--""._""._""._""._""._" -

- -

_"". N 48 Fcmaı.cs -~

- -

-~----""-

-

- - - -

-,_.~---Biiliac br. 270.3 20.18

251.3

17.33 Transverse br. 132,4 9.21 122.1 7.28 A-p. ht. 113,4 17.76 117.6 9.12 rnnommaıe hL 201.3 8.73 196.9 950 lliac hr. 1590 8.57 151.5 9.31 Ischiopubıc br 120.3 5.20 116,4 6.6 ı Fem()ra! length 4139 2.3.';9 440.2 24.21 8icondyler length 419.8 23.13 436.2 24.12

Fenıoral a-p. dia. 26.0 2.46 273 1.93

Fcrn. trans. dia. 26.0 211 25A 2.07

Fem. di,t. cp. br. 74.0 3.32 74.1 3.51

Fem. mid. circ. 82.0 5.31 831 4.8S

F~m. head dia. 42.1 2.36 rı2J 2.39

Tibial length 3413 21,46 163,4 20.96

Tıbial a.p. rliQ. 305 319 3Ll 3. LO

Tıb. trans. dia. 24.0 3.09 24.7 2.50

Tib. prox. ep. br. 68.2 3.14 69.6 3.35

Tib. disı. ep. br. 41.3 298 42.6 2.23

Circ. at l'\lltr. foram. 86.1 6.36 89.7 6.17

Miıı. eiıe. 67.7 5.37 70.3 4.17

M idshaft ci re. 73.6 5.54 76.8 4.6[

d) de. ı ,95 for nıales and! ,Q} f"r females. Ns hased arı subjects heving cornplcte d31. on all variable,

b) signiiıcant aı p<ılOOl ; c) signiJicarıl at p<O.Ol ; d) signiIicaııt aı p<O.o5.

47 24.89 b 35.79 b 0.01 5.48 d ltı.H3 c 9.96 c 10.98 c 11.44 c 7.38 c 1.80 0.01 1.21 0.12 25.R6 b 0.76 1.47 4.48 d 639

ct

8.21 c 6.98 c 8.84 c:

(5)

Race Detemıinaıion from the Posteranıal Skclcton 133 Tablc IL Unstandardi7-ed discriminant function coefficients for determining race on the basıs of varİom

combinations of posıcranial skelctal fcmaim for males".

Variables Pelvis+ Pelvis Pelvis Femur Pelvis or

Femur+ + + + Fernur Oı

Tihia Femur Tibıa Tibia Tibia

Bıuıac br. -0.02978638 -0.02401147 -0.02483145 -0.03851971 Transverse br. ~0.02512998 ·0.02999086 ·0.09033262 A~p. ht. -(L.04706068 -0.04271974 -0.03503142 ~0.02()4943 7 Innominaıe ht. -0.05620039 -0.03525313 -0.03927115 0.07786652 Ischiopubic br. 0.05877165 -0.08663778 -0.04752680 -0.06306601 Coııstaııt 13.92409000 1'eınofal lengtn 0.04345227 0.04990413 Bieondyler Iength -0.02165001

Fcmoıal a-p. dıa. 0.16775940 0.12899130 0.08249423 0.10412800

Fcın. lr3ns. dia. -0.06266954

Ff'm. dı sı' ep. br. 0.1358881 0.12266100

Fcm. mid. cire. -0.0643007 -0.12500940 -0.09894327

Fem. head dİa. -0.15981070 -0.22412480

Constaııt 5.3798l'l00 -6.53728600

- " " " - " " . _ " , , , , - " ' - - -- -

-

-~

-Tibialle.ng1.h 0.05386101 005056249 0.06294315 0.05343540

Tıoia! a-p. dİa,

Tib. trans. dia. 0.13559940 0.12086070

Tib. prox ep. br. 0.08405210

Tıb. disL ep. br. -0.06583214 -0.09308883

Cire . .ıl J'{utr. [oranı. -0.08137693 ~0.08117219

Min. cİre.

Midshaft ei re. 0.09099263

Conslanl '1.23663500 2.28482700 -6.71339800 -11.23612000

*

Disenminan! $c"re. kss ıhan O classuies as White

As part of its

optioııa! ouıput

lhe SPSS DISCRIMINANT program produccs

classi-fication probabilitics and c!iscriminant function scorcs for individua! cases. Figurcs 1 and

2

illııstrate

the probabilily of correct racial classification as a function of the

discrimi-nant

scorc obtained

for

males and

females.

Thesc figures were produced

by

pIotling the

actual discriminant function score and highest probability

of

group memhership

for

each

ease and then simply

"conneetiııg

the

dOLS" WiLh

slraiglıı

lines. They

ınay

be usd, for

examplc, as

a guiclc in

esLiınuLing

the

probabiliıy

of

correctly classifying

spcciıncııs

(6)

134

M.Y.

tŞCAN

Table III. Uıısıandanjized discriminant function coefficicııts for deıenTıining race on the basİ. of various combinations of posterarıial skeletal remain,; for feınales *.

Variables Pelvis+ Pelvis Pelvis Femur Pelvis or

Femur+ + + + Femur or

Tibia Femur Tibia Tibia Tibia

Transverse br. -0.0571·1166 ·0.06984539 ·0.05781593 0.12701030 A-p. ht. 0.01978165 0.01667620 -0.01769865 -0.02079786 Irınom uıaıe ht. -0.06156737 -0.04903466 -0.06580186 lliac br. -0.04924215 -0.06120747 -0.05325227 Constant -13.7J472000 Femoralleıı gtb 0.03314700 0.02352210 Bicondyler length 0.03795900 Femoral a p. dia. 0.1220/\'\90 0.14509370 0.24875630 0.22115100

Feın. trans. dİa. -0.11076730 -0.18791270 -0.24670730

Fcm. disl. ep. br. 0.09076933

Feın. ınid. dıe. -0.10187190

Fem. head dİa. -0.13333740 -0.17151780

Constant 1.28662900 -8.557B100

- ----

---Tıbia! length 0.03981776 0.04017946 0.06nl?13 0.04753136

Tibial a-p. dia. -0.0591745 ı -0.15·402670 -0.11774900

Tibi:ıl prox. cp. 0.12227440 0.13076620

Tio. dist. cp. br. 0.10824140 0.08261725 0.16448920 0.09595671

Circ. al Nutr. foram. 0.07255529

Conslarıt -0.72537240 0.01681537 -7.86855600 -17.14047000 • Discriminant score !ess than O classifies as White.

corrcclly lütle better than chance, whilc cxtrcmely

lıigh

positivc scores al most certainly

represem Blacks and extremely low negative scores almost certainly represcm Whites.

To demonstration a practka! application of

ıhcsc

findings. consider a hypothetical

ınale speciınen

in which measures can only be obtained

from

the femur and" tibia.

Examination of Table

II

indicates thal. four femoral and four tibial

mea.;;urcmenl.S (mm)

should be taken. These

ıneasurcments

and the corresponding values

from

the

hypoıhetieal

specimen are preseme,d below:

Bicondylar length

477 Femaral anteroposterior height

31

Feınoral

transverse diametcr

29 Femora! midshaft circumference

91 Tibia1 length

403 Tibial transverse diameter

27 Tihial circumference at nutrient

fomınen

100

(7)

Race Detcrminaıion from the Posteranial SkelelOlI

'rhe

caktılation

or

ıhe

approptiate

discriminanı

[unetion score would he as follows:

F (fH)

=

- 0.02165001 X (477)

+

(),Q8249423 X CH) 0.062611954 X (29)

0.12500940 X (91)

+

0.06294315 X (401)

i-

0.13559940 X (27)

0.08137693 X (100)

+

0.09099263 X (90) 6.71339800

+

1.40252

135

Sincc the

coınpuıed

value of 1.40252 is pasilive, the hypothctical spccimcn would be

classiricu as Black.

ExaminaLİon

of Tablc IV,

iııdicatcs

that. on the average, one can

cxpecı

to classify R6.0 percent of Blaek males correctly using Libial and femoral

mcasurements togetiler. The

posteıior

probability of an individual's being a member of a

panicular raee

group

can be obtained from the

appropriate

curvc

ın

Figurcs 1 or 2.

In

this case the posterior prohabi1ity of corrcctly classifying

ıhis

spccimcn,

wilh

a scorc of

1.40252, as Black, is

approximately

96%.

Tablc IV. Cl:ısslficaıion results and the percentage üf cürrccıly prcdicted individuals.

FuncLİons /'tfale Female

an:! White I3!ack Ave. While Black Ave.

Variables

---

-~---% N %

N

0/0 % N 0/0 N % Pelvis + 94.0 50 95.8 48 94.9 _87.5 ₄₈ ₉₃₉ ₄₉ _90.7 Fcmur + T:bia Pelvis + 92.2 Si 93.9 49 93.0 _81.6

₄₉

_86.5 ₅₂ _84.2 Femur l'dvis t 96.0 50 91.8

49

93.9 87,5 48 94.0 50 908 Tibia femur t- S8.9 54 86.0 50 87.5 77.4 53 81.4 59 79.5 Tibia Pclvi, 78.4 51 82.0 50 80.2 _72.0 ₅₀ _SO.X _:;2 _76.5 Femur 78.6 56 76.0 50 77.4 _64.8 ₅₄ _68.9 ₆₁ _67.0 Tibia 87.0 54 78.4 51 82.9 7S.)

53

66.1 :"9 70.S

(8)

136 MY lŞCAN

DISCUSSION AND CONCLUSION

Skeletal physical anthropology has profited by the increasing use of multivariate

statistieal

proecdures, parlicularly the

use

of

diseriminant function analysis in

the areas

of sex and race determinalion, Since

Hanihara's

1958 (20)

paper on sex determination

by

diseriminant funetion analysis,

the use

o[

this

meıhodology

in skeJcta! and forensie

studies

rcmained

infıequem uınilthc

last

decade

(21).

lşcan

(14)

emphasized

that

the pelvis as

a

whole

as

well

as the

skull,

varies

metrieruly

by racc. He

also

raised

the possibility that racial varjation

elsewhere

in

the

postcranial

skeletol1

may

be osteomeııically

quantifiable.

Of

the 21

mcasurements

seleeted

from

the three postcranial stmetmes

examincd here,

seven dimensions

yield a

success

mtc

of 95%. The [emur shows the kast

meırieal

varimion

wiıh

race (67% accuraey)

İn spiıe

of

rnorphological variation

in

the

anteıim

curvatlife and intereondylar notch (12,13,22).

A

comparison

with

other imestigations

suggests

that

the

present study is quitc

slIcccssful.

For example, the maximum aecuracy in classifying raee reached

by

Ci/es and

El/iot (23, 24)

was

95%

using

cıght

eranial dimensions.

This approach procluces much

less

accuracy as populations

are

gcographically more removed from the

original

samplc

area

(25),

Howells

(26),

using

16 cranial

climensıons

on

a

geographiu111y diverse

sampk

was

able to

ohtain an

aeeuraey

of

9OCJo.

Using a single variable at a

time,

Gill (27)

prcdicts

race with

simila! accuracy in his study dealing wiLh American Jndians

and

Whites.

Most

studies emphasize

the

danger inherenl. in au.cmpting

to

extencl

even

a

statistically elegant procedure to

differerıl

populations.

When fmther

comparisoils

were made with

studies on

the postcranial

skelcton,

the

pre.sent study

sıi!!

provides improved

discriminaıory

aceuraey. For examp!e,

Jşcan

(14)

obtained a

maximum

aecuracy of 85% using

the

pelvis. When lhe age of the spccimen

was

added

to Lhe

funclion

the accllfacy was

soınewlıat higher

(lS). Because of the

dirrerenees in age betwcen

Whiıcs

(older mean age) and Blacks (younger mean age)

İn

the Terry CoIlectiol1, a controlled study dcsigned to remove the complicating factor of

age

wiıh

face

was

performed

(28).

The 8ccuraey of

ıhe racin1 elassificaıiol1

wa:'i not

drastieaııy

affected

when

the effects of age \Vereeliminated and

the

rcsulting accuracy

was somewhat lower than that reporled in

tlıe

present study. Arecent

study

(17)

indicaıed

that

by

ıısing

[our variables

(measurernents and

indiccs

laken from

the

acetabulum and pubis) aecUfacy of prcdicting rilce was as high as 85

%.

Perhaps the only study

comparabIe

to the preseni one in tcrms of accuracy was

carried

out

by

DiBennarda

and

Taylor (16).

The

aCClIraey

of this

study

was also 97%,

However, the number of dimensions taken from both the innominate and femur was 15

compared to 9 for males

and 10

for females

in the present

study.

(9)

c o "6 ü ~ if} </) o U u

'"

ts

v

....

_o >-.Q

"

.0 2 o..

Race Determinatian fr0ın th" Posıcranial Skeleıon lTı

Graphic tcchniqucs (Figures 1 anel 2) rnay be

ııscd

to

qııickly

assess the probable

Tacial identit)' of a given spccimen. Asimilar approach for calculating individual

probabilities was published

by Tflylor

ei

al

(29).

The prescnt study

sinıply

extcnds

1

heir

approach by plotting the calculations for all bone combinations. This probability is

USdııl

to

ıhe

anthropologist,

partieıılarly

for

coıırlroom prcsemaı

ions. It provides a

relatively simplc and casily grasped picture of the rclationship bctwecn the discriminant

funcıion

score

and classification

accııraey

th(lt c(ln

be

ıındersıaod

hy a jury membcr or

attorney.

I. ~~.~.~--,---~---r---r---.--~~~

....

. 9

,

'. p+ f·ı p+t p ..

!

ht ~ -2. -1.5 -!. -.5 0 . 5 1.5:2

Discrim:nanl Function Score

Figure 1. Posterıor prohabiliıy of CorIfOct racıal c1assificaıion as a funelion of discriminant scores using dimensions from the pelvıs (P), femur (r) and tıbıa (t) for males.

Ackrıuwlcdgemeııts

§

"6 .<e

....

,n ci> O U "o ~ O O

'o

..

~ 15 O .o

ct

~ .8 .7 .6 , p+f f + t P f 5 ! ~--~--~--~--~~---7.. -1.5 -I. -.5 O. .5 L 1.5 2.

Diseriminant Funclion Score

Figurc 2. Postenor l'robahitity of correCi racial

classifieaıion as a fl1l1ction of discriminant scores ming dimensions from the pelvis (pJ. femur (t) and !ibta (I) for females

The amhcr i. very graıeful to Tim Cotton for his uıvaluable assisıance with the complex statıstical analysis necessary for this sıudy. He also wisheı lo thank Patricia Mil1er-Sha,viız for her heir in data collecıion, and Roben Gray and Sıeve Weldcn for ,heir assisımce in the progra,."ming rcquired to produce the lwo seıs of ploıs. The author alsa appreciaıes the eoopemtlon pro"idcd by Lyman M. JcUcma of the Cleveland lvluseum of Natural IIistol)' during ıJıe coJlECtlon of dala. He is graıcful to Walda E. !şcan for her ediıoda[

assisıance. This study was supportcd by an FAU Faculıy Development Granı, dIld !.he Hoh family Chariıable Trust of Borida.

(10)

138

REFERENCES

Kerley, E.R. (1978) Yrbk. Phys'-Anthropol., 21, 160-173. 2 Snow,

c.c.

(1982) Ann. Rev. Anthropol., 11, 97-13l.

M.Y.IŞCAN

3 Krogman, W.M., Işean, M.Y. (1986) The Human Skeleton in Forensic Medicine, Charles C. Thomas, Spri.ngfield, IL.

4 Derry, D.E. (1923) 1. Anat, 58, 71-83.

5 Todd, T.W., Lindala, A. (1928) Am. J. Phys. Anthropol., 12, 35-119. 6 Howells, W.W., Hotelling, H. (1936) Am. 1. Phys. Anthropol., 21,91-106. 7 Torpin, R (1951) Am. 1. Obstet. Gynecol., 62, 279-293.

8 Flower, W.H. (1879) J.Anat, 14, 13-17. 9 Munter, A.H. (1936) Biometry, 28, 258-294.

10 Farrally, M.R, Moore, W.J. (1975) Am. J. Phys. Anthropol., 43, 63-69. II Flander, L.B. (1978) Am. J. Phys. Anthropol., 49, 103-110.

12 Stewart, T.D. (1962) llum. Biol., 34, 49-62. 13 Walensky, N.A. (1965) Anat. Record, 151, 559-570. 14 İşean, M.Y. (1981) Ossa, 8, 95-100.

15 Işean, M.Y. (1983) Am. J. Phys. Anthropol., 62, 205-208.

16 DiBennardo, R., Taylor, J.V. (1983) Am. J. Phys. Anthropol., 61, 305-314. 17 Sehulter-Ellis, F.P., Hayek, L. C. (1984) Col!. Antropol., 8, 155-162.

18 Nie, N.H., Hull, C.H., Jcnkins, J.G., Steinbrenner, K., Bent, D.H. (1975) SPSS Update 7-9, McGraw-llill, New York.

19 Hull, C.A., Nie, N.H. (1981) SPSS Update 7-9, MeGraw-Hill, New York. 20 Hanihara, K. (1958) Zinruigaku Zassi (J. Anthropol. Soc.), Nippon, 66, 39-48. 21 ışean, M.Y. (1988) Yrbk. Phys. Anthropol., 31,203-230.

22 Gilbert, B.M. (1976) Am. 1. Phys. Anthropol., 45, 601-604.

23 Giles, E. and E!liot, O. (1962) Vle Cong. Intl. Sci. Anthropol. Ethnol., Paris, 1, 179-184. 24 Giles, E., Elliot, O. (1962) 1. Forensic Sci., 7, 147-157.

25 Birkby, W. H. (1966) Am. 1. Phys. Anthropol., 24, 21-28.

26 Howells, W.W. (1970) in Personal Identijication in Mass Disasters (Stewart, T.D., ed.), National Museum of Natural History, Washington, D.C.

27 Gill, G.W. (1984) in Human Identijication: Case Studies in Forensic Anthropology (Rathbun, T.A., Buikstra, J.E., eds.) Charles C. Thomas, Spri.ngficld, IL.

28 İşean, M.Y., Cotton, T.S. (1985) 1. }Jum. Evol., 14, 275-282.

29 Taylor, J.V., DiBennardo, R, Linare, G,H., Goldman, A.D., DeForest, P.R. (1984) 1. Forensic Sci.,

29,798-805.

Reprints request to: Prof.Dr. Mehmet Yaşar Işean

Department of Anthropology Horida Atlantic University Hoca Raton, FL 33431-0991, USA

(11)

Race Deıennination from the Postcranial Skeleıon 139

APPENDIX 1 D

E

FINITION OF OSTEOMETRIC DIMENSIONS

This Appendix is included in order to provide a quick reference for those individuals

who do not have access to the major sources on osteometry. As mentioned earlier,

twenty-one osteometric dimensions were taken for this paper. All of these

measurements are also describe in Krogman and

lşcan

(3).

PeIvic girdIe and innominate bone

Biiliac (bicristal) breadth: maximum distance between the iliac crests measured

o

n

an osteometric board

af

ter innominate bones were

arLİculated