You are on page 1of 7

AMERICAN JOURNAL OF PHYSICAL ANTHROPOLOGY 685’9-85 (1985)

Accuracy and Direction of Error in the Sexing of the Skeleton:


Implications for Paleodemography
RICHARD S. MEINDL, C. OWEN LOVEJOY, ROBERT P. MENSFORTH,
AND LYDIA DON CARLOS
Departments o f Anthropology and Biology, Kent State Uniuersity, Kent,
Ohio 44242 (R.S. M., C. O.L., R.P.M.); Department of Orthopaedic Surgery,
Case Western Reserve University, Cuyahoga County Coroner’s Office, and
Cleveland Museum ofNatural History, Cleveland Ohio 44106 (C. 0.L.);
Department o f H u m a n Anatomy, Northeast Ohio Universities College of
Medicine, Rootstown, Ohio 44272 (C.O.L., L.D. C.)

KEY WORDS Forensic, Sex-determination, Sexing, Demography,


Paleodemography

ABSTRACT Determinations of sex by subjective assessment of the skulls


from a skeletal series of known sex were compared to fully independent assess-
ments based on pelves of the same specimens. Within-sex correlations of
cranial and pelvic morphologies measured on a n android-gynecoid scale were
smaller than expected. Subjective assessment by means of the skull compared
favorably to that of the linear discriminant functions of Giles and Elliot;
however, the direction of error was similar for both procedures. Of course,
estimations based on the pelves were generally superior to both in terms of
frequency and overall bias of error. The bias of sex estimation for paleodemo-
graphic purposes is contingent upon completeness of skeletal remains.

The correct determination of skeletal sex is metric determination of sex in paleodemo-


a critical requirement in physical anthropol- graphy.
ogy, paleodemography, and forensic medi- MATERIALS AND METHODS
cine. Accuracy is clearly contingent upon
skeletal completeness, and Krogman esti- One hundred adult skeletons from the Ha-
mates the accuracy of descriptive sexing to mann-Todd Collection (from Sample 11, see
be about 90% for the lone cranium, 95% for Lovejoy, et al., 1985) with complete skulls
the lone pelvis, and 98% if both are available and pelves were selected. All but five also
(Krogman, 1962). Others have generally not had a complete femur, which was used as a n
been as optimistic and have advocated the approximate measure of body size. Age, sex,
use of more “objective” metric discrimina- and race of all specimens were unknown to
tors, which are presumed to be more repeat- the primary assessors (R.S.M., C.O.L.)
able and more easily taught (Washburn, throughout the tests as were the proportions
1948; Thieme and Schull, 1957; Giles, 1970; of age, sex, and race in the sample of 100.
Acsadi and Nemeskeri, 1970; Weiss, 1973). Without examining the individual skulls,
This paper presents the results of a blind test the two assessors each arranged the com-
of the accuracy of observational sexing of plete pelves on a continuum from most male
skulls (crania plus mandibles) and, indepen- to most female on the qualititative basis of
dently, pelves (innominates plus sacra) by (1)morphology of the pubic face and inferior
assessors with above average experience in ramus (Phenice, 19691, (2) greater sciatic
skeletal identification. Its primary purposes notch and preauricular region, (3) shape of
are (1)to report the nature of within-sex cor- the lesser pelvis, and (4)overall rugosity and
relation between these two separate discrim- relative size of the acetabulum. Since the
inators of sex, (2) to examine the likely rank-orders determined by the two assessors
direction of error conditional upon age-at-
death and skeletal completeness, and (3) to Received August 11, 1983; revised May 1, 1985; accepted May
examine the implications of descriptive and 3, 1985.

0 1985 ALAN R. LISS. INC


80 R.S. MEINDL, C.O. LOVEJOY, R.P. MENSFORTH, AND L. DON CARLOS

were quite similar, the continua were com- RESULTS


bined without difficulty, and the pelves were
then arranged on a simple six-level ordinal After final sex determinations had been
scale ranging from “very male” (score = 0 ) made and notes on questionable cases pre-
to “very female” (score = 5). pared, the actual sex of each specimen was
Skulls were later seriated without knowl- revealed by the population constructor
edge of the associated pelves, independently (R.P.M.). The sex ratio of the sample was 59
a t first by each assessor. Sex was determined males to 41 females. Eight errors had been
by (1) overall rugosity of the occipital, mas- made based on observations of only the skull.
toid, and zygomatic regions; (2)relative depth Four errors had been made on only the pel-
of the mandibular rami, morphology of the vis. With both considered, a total of three
chin, and degree of gonial eversion; (3) size of errors (3%)had been made. Table 1provides
the teeth and palate; (4)size of the pterygoid verbatim notes for all specimens in the series
plates; (5) prominence of the supraorbital that were suspected to have been incorrectly
ridges; and (6)frontal profile. No postcranial sexed. Of the 100 cases examined, six had
information was used. Interobserver discrep- been considered questionable, and all three
ancies were somewhat more numerous and errors occurred in this subsample of six.
pronounced in the case of the skulls; how- Skull and pelvic sex were judged indepen-
ever, the assessors eventually agreed on a dently so that the joint distribution of scores
single six-level ordinal scale as above. could be examined and their correlations
The independent sex continua based on within each sex estimated (Table 2). The six
skulls and pelves were then compared, and a ordinal levels reflect the degree of “female-
final judgment was made for each case. This ness’’ in both the skull (rows) and pelvis (col-
usually posed no difficulty except in marked umns). As a n example, 11actual males were
failures of agreement between skull and pel- judged to be moderately male in skull mor-
vis scores. In such cases, the pelvis was phology (score = 1)and, independently, very
weighted more heavily. male in pelvic morphology (score = 0).

TABLE 1. Evaluations of questionable cases


Case Pelvis Skull Estimate Actual Comment
HT0854 Female? Female? Female Male “Very borderline case”
HT1312 Female Male? Female Male “Uncertain”
HT1366 Female Male? Female Male “Uncertain”
HT1443 Female? Male Male Male “Pelvis not
informative”
HT1476 Male? Female? Male Male “Uncertain”
HT1584 Female? Female? Female Female “Uncertain”

TABLE 2. Bivariate distribution of estimated sex by skull andpeluis, within actual sex classes
Score on pelvis
Male+ Male Male- Female- Female Female+
Score on skull (0) (1) (2) (3) (4) (5) Total
Males
Female+ (5) -_ -
Female (4) - -
Female- (3) 2 6
Male- (2) 8 14
Male (1) 11 19
Male+ (0) 13 20
Total 34 59
Females
Female+ (5) -.
19
Female (4) -. 4
Female- (3) -.
16
Male- (2) -.
1
Male (1) -.
1
Male+ (0) -~ -
Total -. - - 1 2 38 41
‘Accuracy of finai assessment doubted, but correct.
‘Accuracy of final assessment doubted, and incorrect.
SEX DETERMINATION 81

Females displayed little variation on the female sample was significant (Table 3). It is
pelvic scale with the result that no female negative (7 = - .39) because the females were
was sexed incorrectly on the basis of the pel- assigned the higher values. Thus these data
vis, and such judgments could not have been indicate that greater age produces a n in-
overruled by any cranial morphology. Only creasingly male morphology. The same pat-
three of the females obtained less than ex- tern obtained within the male subsample (7
treme scores on the pelvis. One of these was = - .33). This implies that a n increase in the
over two standard deviations above the fe- accuracy of discrimination is possible if the
male mean for femoral length; the other two assessor considers some estimate of skeletal
were of very average stature yet possessed age (see Giles, 1970; Israel, 1973, 1977).
intermediate morphologies on the ventral arc “Most head and face measurements continue
and subpubic concavity (Phenice, 1969). The to increase after adolescence steadily, though
males, on the other hand, were quite varia- very slowly, to at least age 60 years. The
ble on both scales. Six (10.2%) would have increase from 20 to 60 years amounts to be-
been incorrectly sexed on the basis of skull tween 2% and 4% of the 20-year-old value”
morphology alone, and four (6.8%)on the ba- (Harrison et al., 1977:317). In the present
sis of the pelvis alone. study, the two peaks of the female skull dis-
The human pelvis plays a critical role in tribution were clearly related to age since
both locomotion and in parturition. In evolu- the 16 “slightly female” skulls were of indi-
tionary perspective, the interacetabular di- viduals who averaged 48.1 i 11.3 years of
mension in females is subject to opposing age a t death, while the 19 “very female”
selection pressures. While the relatively skulls were of individuals who average 36.5
large brain of the fetus requires a n expanded f 7.9 years, a difference that is highly signif-
birth canal, locomotor needs of the female icant (t = 3.46, p < .01). Age bore no rela-
limit the divergence of the hip joints. This tionship to the morphology of the pelvis as
forms the very basis of the sex dimorphism ranked by the assessors (Table 3).
of the human pelvis, and results in the very At the start of this project, a strong associ-
distinctive female subpubic concavity de- ation between gynecoid features in skull and
scribed by Phenice (1969) and the informa- pelvis within the female subsample was an-
tional content of the traditional ischiopubic ticipated. The results were equivocal. Since
index as well. Equally important is that it true variation among the female pelves was
results in the extreme limitation of female extremely limited, skull variation had noth-
variability relative to that of the male. Since ing with which to covary (7 = .09; Table 3).
neither female nor male skull anatomy is Males were moderately variable in both skull
held by selection in such a fine balance, and pelvic anatomy: a little over half (57.6%)
within-sex variations on that scale are equiv- exhibited “very male” pelves; about a third
alent, and skull dimorphism itself is less de- (33.9%) had “very male” skulls. However,
finitive. Yet the direction of error was the the correlation between the sex scores on
same as it was for the pelves: 10.2% of the skull and pelvis within the male subsample
males were sexed incorrectly, as compared to only approached significance (7 = .16; Table
4.9%of the females. 3). Body size (approximated by femoral
It was found that the sex-dependent mor- length) exhibited no measurable association
phology of the skull was significantly af- with the cranial or pelvic (subjective) assess-
fected by age. A nonparametric correlation ments within either sex (Table 3). Therefore,
between cranial score and age within the controlling for body size would not have sig-

TABLE 3. Correlations within sexes: Kendalt‘s 7

Sample (n) Variable pair 7 Significance


Females (41) Skull’iage - .39 p < ,001
Males (59) Skull/age - .33 p < ,001
Females (41) Pelvis’/age - .08 ns, p .25
Males (59) Pelvisiage - .03 ns, p > .25
Females (41) Pelvisiskull + .09 ns, p > .25
Males (59) Pelvidskull +.16 .10> p > .05
Females (38) Skull/size2 - .09 ns, p > .25
Males (57) Skullisize + .07 ns, p > .10
Females (38) Pelvis/size -.lo ns, p > .10
Males (57) Pelvis/size + .04 ns. n > -26
‘Scores ranged from 0 (very male) to 5 (very female) for both skull and pelvis
‘“Size” is the bicondylar length of the femur.
82 R.S. MEXNDL, C.O. LOVEJOY, R.P. MENSFORTH. AND L. DON CARLOS

nificantly affected the degree of within-sex black male subsample of this study and the
association between skull and pelvis. one selected by Giles and Elliot. The average
discriminant function scores for females
METRIC DISCRIMINATION OF CRANIA
within both races for both studies were quite
Following Giles and Elliot, nine standard similar, and the average scores for the white
craniometrics of proven sex-discrimination males in both studies were nearly as close
power were taken to the nearest millimeter (Tables 5,6).However, the scores of the black
on each specimen (Giles and Elliot, 196358- males in this study (especially on functions 5
59): The measures were glabello-occipital and 8) were significantly different from those
length, maximum width, basion-bregma in the original Giles and Elliot study. This
height, maximum bizygomatic diameter, reflects a size differential between these sub-
basion-prosthion length, prosthion-nasion samples that is most probably a function of
height, basion-nasion length, external age. No control over race and age was exer-
breadth of the palate, and mastoid height. cised in the selection of the 100 specimens in
All are lengths, widths, and chords and are the present sample, the black males of which
defined in Giles and Elliot (1963) or in Hoo- tended to be quite young (Table 7).
ten (1946). Nine discriminant functions were Had the subjective methods of this study
selected from that study. Those functions for sexing the skull also failed to include
numbered 4-6 utilize eight measures; 7-9 mandibular data, the error rates might have
use all but three; and 19-21 use only five. approached those of the linear functions. In
The last of each group of three (i.e., discrimi- numerous instances the morphology of the
nant functions 6, 9, and 21) were designed chin and the degree of gonial eversion re-
for use with race unknown and were applied solved the status of otherwise difficult crania.
to the entire sample in this study. The re- However, metric discriminators that cannot
maining six are race-specific and were ap- adequately reflect these rather subjective
plied to the appropriate subsamples.' The mandibular traits are not as useful. Giles
sectioning points used for discrimination of (1964) provides nine mandibular functions,
the specimens in this study were those deter- which were not used extensively in this study
mined by Giles and Elliot (1963). for two reasons. First, the adult mandible is
The cranial discriminant functions were perhaps the most environmentally modified
less accurate than the simple observational of all bones. It is subject to extensive remod-
methods employed in this study, which was eling, which makes it too sensitive for paleo-
partly due to our decision to incorporate sub- demographic purposes. Second, remodeling
jective information from the mandible (Table accelerates after the common event of tooth
4). Error rates for the linear functions were loss, and quite a few of the Hamann-Todd
remarkably close to those predicted by the skeletons are functionally (post-canine) ed-
authors, except in the case of the black sub- entulous. When the Giles mandible functions
sample, which suffered misclassification
rates of about one in five (usually males
'As an aid to these procedures, the assessors did not attempt
called females). This was probably a conse- to estimate race; rather, appropriate functions were chosen only
quence of a size discrepancy between the after the revelation of race for each specimen.

TABLE 4. Error rates in sex discrimination


Discriminator Actual sex of errors
(function) Actual errors % Predicted %' (males; females)
6 (race unknown) 141100 14.0 14.5 10; 4
9 (race unknown) 14/100 14.0 15.1 10; 4
21 (race unknown) 16/100 16.0 15.9 10; 6
4 (whites only) 4/29 13.8 13.6 1;3
7 (whites only) 4/29 13.8 13.6 1; 3
19 (whites only) 3/29 10.3 15.1 1; 2
5 (blacks only) 1817 1 25.4 13.4 14; 4
8 (blacks only) 1717 1 23.9 13.5 13; 4
20 (blacks only) 14/71 19.7 15.0 11; 3
Cranium and 81100 8.0 6; 2
mandible, subjective
'General error rate predicted by Giles and Elliot with application of functions to appropriate samples
(1963:631.
SEX DETERMINATION 83

TABLE 5. Mean discriminant function scores, females

Discriminator Giles and Elliot Test sample, this study


(function) sample means Mean SD n
6 5,769.25 5,755.24 194.76 41
9 5,919.72 5,907.84 200.35 41
21 862.41 866.65 29.71 41
4 2,497.10 2,486.36 81.82 18 whites
7 1,248.86 1,245.80 42.59 18 whites
19 4,903.33 4,897.85 144.51 18 whites
5 3,935.03 3,922.74 151.81 23 blacks
8 2,423.22 2,430.06 92.36 23 blacks
20 2,485.22 2,492.79 90.64 23 blacks

TABLE 6. Mean discriminant function scores, males

Discriminator Giles and Elliot Test sample, this study


(function) sample means Mean SD n
6 6,174.82 6,140.96 194.51 59
9 6,319.28 6,288.32 192.35 59
21 920.55 919.22 31.25 59
4 2,687.53 2,663.03 76.70 11 whites
7 1,343.54 1,335.37 38.21 11 whites
19 5,230.05 5,199.00 138.24 11 whites
5 4,223.20 4,162.78 139.69 48 blacks
8 2,608.60 2,570.99 86.20 48 blacks
20 2,652.71 2,636.17 92.32 48 blacks

TABLE Z Ages of racesex classes


Giles and Elliot samples Samples, this study
Mean SD n Mean SD n
White males 44.6 13.7 75 49.9 6.3 11
White females 44.6 14.1 75 44.9 11.0 18
Black males' 44.3 14.1 75 34.6 9.6 48
Black females 43.9 14.3 75 40.3 10.4 23
'Black male means significantly different, p < ,001.

were applied only to those skulls incorrectly vational methods because important mandi-
sexed on one or more crania functions (and bular information is not fully utilized; (2)
provided they had teeth), the bias was pro- cranial dimensions appear to increase with
nounced. Nearly all of the many incorrectly age, thereby affecting the discriminant func-
sexed males remained so on the mandible tion score; and (3)the direction of error in the
functions, while all of the fewer errors on estimation of sex based on the Giles and El-
females were contradicted by the mandible liot discriminant functions is the same as
functions. Thus, systematic application of the that of descriptive sexing. That is, the crania
mandible discriminant functions would only of young adult males, which constitute a siz-
serve to increase the bias (i.e., males called able segment of those mortuaries that are
female) in this study. Giles and Elliot's (1963) demographically useful, will probably be
original decision to limit the metric discrim- misclassified at a higher rate than those of
inators to crania is a sound one. other age-sex classes, regardless of the
The results of the application of linear dis- method employed.'
criminant functions to biological popula-
tions-from which they were derived and for
which they were intended to be used-has 'Movement of the sectioning point, a makeshift procedure
advocated by Henke (1977),would have resulted in only a small
several implications for their use in a paleo- material improvement in the results of our application of the
demographic context: (1) the power of the Giles and Elliot discriminant functions to the present sample.
The relocation of the discrimination boundaries for other popu-
battery of metric discriminators in current lations remains difficult ta implement, largely untested, and
use for skulls is limited compared to obser- theoretically problematic.
84 R.S.MEINDL, C.O. LOVEJOY, R.P. MENSFORTH, AND L. DON CARLOS

DISCUSSION a feeling of trust in their sexing of skeletal


remains” (Stewart, 1954). Our own numer-
The present study has demonstrated that ous attempts to resolve metrically the sex of
actual female skeletons are rarely misclassi- those very few cases in which the pelvic mor-
fied, whereas males can on occasion be mis- phology is indeterminant have never proved
takenly called female. No single cranial more successful than ordinary observational
character emerged as a hallmark of gender. methods.
This is fully expected since masticatory, fa- CONCLUSIONS
cial, and calvarial characters play no sex-
specific role. However, there are definable Any statistical technique that uses multi-
secondary characters that are present in even ple linear dimensions to establish a rule for
the least dimorphic population, and these are discriminating unknown specimens is popu-
useful in descriptive sexing of the skull. lation-specific. The best such functions in
The statistical discriminant functions in current use for the forensic analysis of the
current use for crania derive nearly all of cranium are the result of the extensive and
their power from size and its allometric ef- carefully controlled study of Giles and Elliot.
fects, which are environmentally labile and Certainly we were unable to improve upon
therefore population specific. That a presum- these types of discriminant functions (i.e.,
ably appropriate Giles and Elliot function those dealing with the face and calvarium
was unable to sex correctly more than 65% only), despite considerable effort. However,
of a series of Finnish crania (of known sex) is their functions appear to lose accuracy when
a case in point (Kajanoja, 1966). It must be applied to statistical populations which differ
emphasized that the material of the present in some systematic way from the 300 speci-
study was drawn from virtually the same mens in their core sample.
biological populations as those used by Giles It is concluded that the overall sex-ratio
and Elliot (indeed, from one of the same skel- and specific age-class sex ratios of prehistoric
etal collections). Yet because of an average cemeteries must be estimated from only those
age differential in one of the sex-race classes, adult burials with fully preserved pelves.
the overall misclassification was biased. Even so, anatomists who adhere to this rule
The use of the complete skull overcomes will probably underestimate the proportion
some but not all of these difficulties. Also of adult males, while erring in very few of
many of the secondary sexual traits present their assessments of true females. This may
measurement problems that would violate be a curious conclusion in the light of the
the assumptions of traditional parametric dis- demographic survey by Weiss, who compared
criminant function analysis (i.e., continuity, the high sex ratios of extinct populations to
multivariate normality, and equal variances the consistently balanced ratios of living prim-
and covariances). It is true that many of these itives (1972). We have argued elsewhere (Lo-
traits serve as primary evidence in the sub- vejoy et al., 1977) that extinct pre-urban pop-
jective anatomical estimation of sex with the ulations present distinctly different biological
skull. Other evidence is less liable to linear and demographic conditions than those ob-
measurement. It is not that the linear dis- served in extant, ethnographic ones.3
criminant functions in their present state of While the direction of error in judging pre-
development fail to formalize what the eye historic sex-ratios by means of either metric
discerns; rather, the two approaches utilize
overlapping but somewhat different sets of
cranial information.
Discrimination of sex of the pelvis is an realize that uneven sex-ratios frequently occur in the
recovery of skeletal populations. An important source of error
altogether different type of problem. There is would he the disproportionate inclusion of one sex over the other
a marked functional difference between the as a result of special mortuary practices. Given the results of
this study and Weiss’ survey (1972), we must conclude that of
two pelvic anatomies that may be regarded the archeologicalsamples of reported high sex-ratios, many may
as primary sexual characters. Furthermore, even represent underestimates of the excess of males. As sug-
the basic elements of this dichotomy are com- gested in this study, a primary reliance on cranial sexing may
be responsible for some of the bias in paleodemographic studies.
mon to all human populations. Even statis- Primary use of pelvic sexing can greatly reduce the estimation
tical functions are very successful here (see error, and a careful analysis of archeological evidence can fre-
quently clarify actual imbalances. However, we wish to point
Thieme and Schull, 19571, but a s one experi- out that the accuracy of sex determination must he gauged on
enced forensic specialist has remarked, there tests such as those presented here, and not by comparison to
living ethnographic populations, which may well evince sub-
is not much point in their application other stantially different demographic parameters than the cemeter-
than to allow beginners “to convey to others ies of extinct archaeological populations.
SEX DETERMINATION 85

or descriptive assessment of skulls alone can- LITERATURE CITED


not be determined, one aspect of cranial mor- Acsadi, G, and Nemeskeri J (1970) History of Human
phology could produce a systematic bias in Life Span and Mortality. Budapest: Akademiai Kiado.
some applications. Skulls of the oldest female Giles, E (1964) Sex determination by discriminant func-
decedents are 2% to 5% larger in most linear tion analysis of the mandible. Am. J. Phys. Anthropol.
dimensions (Israel, 1973, 1977; Harrison et 22:129-135.
Giles, E (1970) Sexing crania by discriminant function
al., 1977) and in our study appear to have analysis. Proc. 8th Int. Cong. Anthropol. Ethnol. Sci.
accumulated other qualitative male charac- A-259-61.
teristics as well. A likely result of using only Giles, E, and Elliot, 0 (1963) Sex determination by dis-
the skull for some specimens would be sex criminant function analysis of crania. Am. J. Phys.
ratios underestimated for the bottom of an Anthropol. 21:53-68.
Harrison, GA, Weiner, JS, Tanner, J M and Barnicot, NA
adult age pyramid, in the same fashion as (1977) Human Biology, 2nd ed. Oxford: Oxford Univer-
the overall error, but less so or even overes- sity Press.
timated for the top of the pyramid. The dis- Henke, W (1977) On the method of discriminant function
tortion of the resulting sex differentials in analysis for sex determination of the skull. J. Hum.
adult survivorship is one of the strongest ar- Evol. 6:95-100.
guments to the paleodemographer for pri- Hooten, EA (1946) Up From the Ape. New York:
Macmillan.
mary reliance on the pelvic determination of Israel, H (1973) Age factor and the pattern of change in
sex. craniofacial structures. Am. J. Phys. Anthropol. 39:lll-
The 6%of the adult skeletons in this study 128.
that were not sex-distinctive (Table l),half of Israel, H (1977) The dichotomous pattern of craniofacial
which were sexed incorrectly, probably rep- expansion during aging. Am. J. Phys. Anthropol.
47:47-52.
resents a n extreme figure. The late-Archaic Kajanoja, P (1966) Sex determination of Finnish crania
Carlston-Annis (Mensforth, unpublished) and by discriminant function analysis. Am. J. Phys. An-
late-Woodland Libben skeletal populations thropol. 24:29-34.
(Lovejoy et al., 1977) contained relatively Krogman, WM (1962) The Human Skeleton in Forensic
fewer of what we would have labeled “ques- Medicine. Springfield IL: Charles C. Thomas.
Lovejoy, CO, Meindl, RS, Mensforth, RP, and Barton, TJ
tionable” cases. Despite reduced average (1985)Multifactorial determination of skeletal age a t
body size, archaeological populations tend to death: A method and blind tests of its accuracy. Am. J.
be more sex-dimorphic and genetically ho- Phys. Anthropol. 68:l-14.
mogeneous than the racially and geographi- Lovejoy, CO, Meindl, RS, Pryzbeck, TR, Barton, TS, Hei-
cally mixed samples used in this and other ple, KG, and Kotting, D (1977) Paleodemography of
the Libben site, Ottawa County, Ohio. Science 198:291-
replicative forensic studies. Sex determina- 293.
tions by descriptive methods may be re- Mensforth, RP: Paleodemography of the Carlston Annis
garded as sufficiently accurate so as not to BT-5 Skeletal Population. Unpublished manuscript,
differ significantly from the actual sex ratios Kent State University.
of cemetery populations, so long as care and Phenice, TW (1969) A newly developed visual method of
sexing the 0s pubis. Am. J. Phys. Anthropol. 30297-
skill are employed in the application of tra- 301.
ditional anatomical methods. Stewart, TD (1954) Sex determination of the skeleton by
guess and by measurement. Am. J. Phys. Anthropol.
ACKNOWLEDGMENTS I2:385-392.
Thieme, FP, and Schull, W J (1957) Sex determination
The research reported in this paper was from the skeleton. Hum. Biol. 29242-273.
funded by the National Science Foundation, Washburn, SL (1948) Sex differences in the pubic bone.
Division of Behavioral and Neural Sciences, Am. J. Phys. Anthropol. 6:199-207.
Award No. BNS-77-07958. The Cleveland Weiss, KM (1972) On the systematic bias in skeletal
sexing. Am. J. Phys. Anthropol. 37:239-250.
Museum of Natural History generously Weiss, KM (1973) Demographic models for anthropology.
loaned skeletal material from the Hamann- Mem. Soc. Am. Archaeol. Am. Antiq. 38: (No. 2, Part
Todd Collection. 2).

You might also like