Professional Documents
Culture Documents
e7
A R T I C L E I N F O A B S T R A C T
Article history: Estimating the sex of skeletal remains is critical in creating the biological profile of an adult individual.
Received 10 January 2011 Previous research has shown that analyzing the sternum may lead to an accurate estimation of sex based
Received in revised form 28 August 2011 on studies performed on various populations around the globe, such as Indian, European, African,
Accepted 30 November 2011
Canadian and North American. The motivation of the current study is to develop classification functions
Available online 29 December 2011
and sectioning points for use in forensic investigations in the United States. The majority of previously
published methods are population specific, meaning the data would not prove useful in the United States
Keywords:
[8–12,14,16–23].
Forensic anthropology
Sex estimation
For this study, sternal measurements were collected from the William M. Bass Donated Skeletal
Sternum Collection located at The University of Tennessee, Knoxville following the metric definitions provided by
Forensic anthropology population data Schwartz [41] and Bass [35]. This collection consists of recent skeletal remains with known age at death,
ancestry and sex. Material for the present study consisted of 410 human sterna: 285 male (256 American
White/29 American Black) and 125 female (120 American White/5 American Black).
Previous studies [8,9,12,17,18,23] analyze the applicability of Hyrtl’s law in the estimation of sex. This
law follows that the mesosternum is greater than twice the length of the manubrium in males and in
females the length of the manubrium is greater than half the length of the mesosternum. In this study,
comparisons of the proportion of the length of the manubrium to the length of the mesosternum were
performed to determine if Hrytl’s law is applicable in an American population. Comparisons of these
measurements between individuals identified as American Black and American White were analyzed to
determine whether this method could be used on both population groups. Further, discriminant function
analysis was used to estimate sex and provide a population specific classification function for use in the
United States.
The discriminate function analysis produced an overall cross-validation classification rate of 84.12%
for sex estimation. The cross-validation classification rate for males and females was 80.00% and 88.24%,
respectively.
ß 2011 Elsevier Ireland Ltd. All rights reserved.
0379-0738/$ – see front matter ß 2011 Elsevier Ireland Ltd. All rights reserved.
doi:10.1016/j.forsciint.2011.11.034
290.e2 R. Bongiovanni, M.K. Spradley / Forensic Science International 219 (2012) 290.e1–290.e7
Fig. 2. Examples of different types of fusion: (A) unfused manubrium and mesosternum, (B) fused mesosternum and xiphoid process with arrow detailing partial fusion.
The sternebra widths follow this design as well. The measurements males and American Black males. The results of the ANOVA will
were taken at midpoint of the first and third sternebra on each side of specify whether or not there is a need for separate discriminant
the specimen. In most specimens, the right and left sides of the functions for American White males and American Black males.
sternebra were perpendicular to the length of the mesosternum. In Insufficient data were gathered from American Black females to
some cases, however, the right and left sides were not perpendicular include in the analysis. The assumption was made that if there is a
to the sagittal plane of the mesosternum. In these cases, excluding significant difference between American White males and Ameri-
extreme variations, measurements were still obtained. In addition, can Black males, the same would hold true for American White and
various specimens contained a fracture across the anterior side of Black females.
the mesosternum. The fracture did not exhibit signs of healing and In order to test if the sternum can provide accurate sex
may possibly be attributed to CPR efforts at or around the time of estimation, univariate and multivariate tests were conducted.
death. As long as the posterior surface of the mesosternum was With the understanding that forensic cases do not always contain
complete, the mesosternum length was recorded if the author felt complete and normal sterna, univariate analyses were used in
that the integrity of the bone was not compromised. Lastly, the total order to evaluate whether there is a difference between males and
length of the sternum was calculated by summing the manubrium females with each variable separately using sectioning points.
length and mesosternum length. Again, no estimated values were Averaging the means of the male and female measurements for
used to calculate this value. Pictures of the sterna were taken with a each variable created sectioning points. The multivariate test
Canon Powershotß SD400. included using a discriminant function analysis to find out which
variables provide the best measures for discrimination of sex.
2.1. Intra-observer error
2.3. Sectioning points
Before testing the main hypotheses of this study an intra-
observer test was performed. Descriptive statistics were calculated in order to create
Test-retest reliability of the sample was conducted after the sectioning points for each variable by utilizing the functions in
measurements were taken in order to test intra-observer variation. A Microsoft Excel 2008. Each variable was separated by sex, and the
retest sample includes a subsample of the data to be re-measured at sample size, mean, standard deviation, sectioning point, frequency
a different point in time in order to discover the reliability of the test. rate, and overall classification rate were calculated. The results
Thirty complete sterna were selected at random of which all four from the univariate analysis that are greater than the sectioning
measurements could be collected. These thirty sterna were split into point are considered male individuals, and results that are less
two groups of fifteen to be re-measured. The two sets of data for the than the sectioning point are considered female individuals, with
test-retest were entered into SPSS 16.0 to calculate the correlation results that are equal to the sectioning point being classified as
co-efficient between variables. A Spearman’s rho correlation indeterminate.
analysis was run to discover whether the correlation between the
two data sets is significant at the 0.01 and 0.05 two-tailed level. 2.4. Discriminant function analysis (DFA)
2.2. Analysis of variance (ANOVA) for population group differences A stepwise discriminant (PROC STEPDISC) function was
performed in order to find out which variables provide the best
Next, an analysis of variance (ANOVA) using PROC GLM (the measures for discrimination of sex. This was followed by a linear
general linear model) in SAS 9.1.3 was calculated to ascertain if discriminant function (PROC DISCRIM) analysis (DFA). These two
there are ancestry specific differences between American White analyses were run in SAS 9.1.3. The DFA maximizes between-group
290.e4 R. Bongiovanni, M.K. Spradley / Forensic Science International 219 (2012) 290.e1–290.e7
differences, and reduces error rates by excluding the sample being Table 1
Cross-validated classification rates for females and males by variable.
classified from the function. This produces cross-validation
classification rates [36,37]. The linear discriminant function scores Variable Female (n) Male (n) Female (%) Male (%) Overall (%)
were then utilized to generate a classification function. The ML, BL, S1W 35 89 85.71 79.78 82.75
classification function is a formula that allows a user to input their All variables 34 85 88.24 80.00 84.12
sternal measurements and estimate whether the individual is male
or female. The cross-validation classification rates indicate how
well the classification function performs. respectively, with a sectioning point of 97.09 mm. The classification
rate for males is 80% and 77% for females, with an overall classification
2.5. Sternal index rate of 78.50%.
For sternebra 1 width, the sample size includes 255 males and
Lastly, the sternal index was calculated in Excel by creating a 114 females. The mean is 27.35 mm in males and 24.29 mm in
formula that divided the length of the manubrium by the length of females with a standard deviation of 4.22 mm and 3.20 mm,
the mesosternum and multiplied by 100, following the definition respectively, with a sectioning point of 25.82 mm. The classification
provided by Hunnargi et al. [23]. The average of the sternal index for rate for males is 61% and 71% for females, with an overall
males and females were compared to see if there was a significant classification rate of 66%.
difference between the two sexes. According to Hyrtl’s Law, the For sternebra 3 width, the sample size includes 249 males and
sectioning point should be 50%, since the ratio of manubrium to 109 females. The mean is 34.47 mm in males and 30.15 mm in
mesosternum is thought to be 2:1 in males. In addition to this females with a standard deviation of 6.33 mm and 4.91 mm,
sectioning point, the male and female mean were divided by two in respectively, with a sectioning point of 32.31 mm. The classification
order to generate a true sectioning point, separate from that rate for males is 58% and 68% for females, with an overall
conforming to Hyrtl’s Law. In this case, however, values below the classification rate of 63%.
sectioning point indicate male, and values above the sectioning For total length, the sample size includes 91 males and 36
point indicate females, with values equaling the sectioning point females. The mean is 154.97 mm in males and 136.75 mm in
considered indeterminate. The frequency rate, or the percentage of females with a standard deviation of 10.85 mm and 8.23 mm,
sterna that were correctly classified utilizing the sectioning point, respectively, with a sectioning point of 145.86 mm. The classification
was taken into account to decide whether or not the sternal index rate for males is 77% and 86% for females, with an overall
would prove to be a useful method in estimating sex of the sterna. classification rate of 81.50%.
Adopting methods from previous research [8,12,14,17], a
3. Results ‘‘Rule’’ was calculated for the total length. The midpoint of the
overlap section between the two sexes is found, and treated as a
3.1. Intra-observer error sectioning point. For this sample, no male has a total length that is
less than 132.43 mm and no female has a total length that is
The Spearman’s rho correlation analysis provided a near perfect greater than 158.9 mm. Therefore, the overlap section between the
positive linear relationship. The two data sets for ML, BL and S1W two sexes is 132.44–158.8 mm. The midpoint of the overlap
had a correlation coefficient of 0.998, which was significant at the section is 145.62 mm. The corresponding rule then is the ‘‘145
0.01 confidence level. The two data sets for S3W were also significant Rule’’. This rule is very similar to the original sectioning point of
at the 0.01 confidence level, with a correlation coefficient of 0.997. 145.86 mm. In using the rule, however, any value between
This indicates the reliability of this method is very high. 145.00 mm and 145.99 mm is considered indeterminate. As a
result, the frequency rate is 82% for males and 83% for females, with
3.2. Analysis of variance (ANOVA) for population group differences an overall classification rate of 82.50%.
The ANOVA using the PROC GLM produced a p-value of 0.13, 3.4. Discriminant function analysis (DFA)
indicating no significant difference between American White
males and American Black males. Therefore, American White Significant differences were found between sexes according to
males and American Black males were grouped together in the PROC DISCRIM for both male and female individuals, at p < .001
subsequent analyses. In addition, because there were not enough (Table 1). According to the stepwise results, all variables provide the
females to run this analysis between population groups, the highest classification rate of 84.12% (Table 1). Therefore, the
assumption was made that no difference existed between discriminant function analysis was run using all variables. The
American White females and American Black females and they classification rates can be found in Table 2. A linear discriminant
were also pooled for all further analyses. function for sex was calculated and resulted in the following
formula:
3.3. Sectioning points y = (ML 0.1706) + (BL 0.17032) + (S1W 0.24504) + (S3W -
0.06407) + 28.97601, where y equals the discriminant function
Sectioning points are listed in Table 2 for each measurement. score. To use this formula, multiply each measurement by the factor
Values falling above the sectioning point are male, below are provided above, take the sum of those values, and add the constant.
female, and values equal to the sectioning point are indeterminate. The sectioning point for this formula is zero (0), negative values are
For manubrium length, the sample size includes 234 males and considered females and positive values are considered males, with
104 females. The mean is 51.84 mm in males and 48.24 mm in values equaling zero (0) being indeterminate. The overall classifica-
females with a standard deviation of 4.93 mm and 3.98 mm, tion rate is 84.12%.
respectively, with a sectioning point of 50.04 mm. The classification
rate for males is 61% and 68% for females, with an overall 3.5. Sternal index
classification rate of 64.50%.
For mesosternum length, the sample size includes 103 males The sternal index for males ranged from 36.30% to 66.91%, with a
and 40 females. The mean is 104.80 mm in males and 89.38 mm in mean of 48.81%. The sternal index for females ranged from 40.34% to
females with a standard deviation of 9.83 mm and 8.33 mm, 69.41%, with a mean of 53.68%. While differences are obvious in the
R. Bongiovanni, M.K. Spradley / Forensic Science International 219 (2012) 290.e1–290.e7 290.e5
Table 2
Descriptive statistics, sectioning points, and overall classification rates for all variables.
Female Male
Variable n Mean StdDev %Correct n Mean StdDev %Correct Sect. point Overall classification rate (%)
Table 4
Data from previous and present research including average measurements (mm) and average sternal index values.
Female Male Female Male Female Male Female Male Female Male Female Male
Europe
Ashley [8] 47.90 52.20 90.80 104.70 24.50 26.40 30.50 33.70 138.70 156.90
Mall et al. [16] 150.00 166.00
Teige [14] 43.90 47.70 88.50 103.40 23.20 27.10 30.90 34.70 133.70 153.40 54.30 46.20
East Africa
Ashley [8] 44.20 45.90 82.90 96.50 21.50 24.00 26.80 30.90 127.10 142.60
India
Dahiphale et al. [17] 43.78 48.46 70.19 94.43 24.44 27.17 28.24 31.95 113.87 142.20 63.01 52.0
Gautam et al. [18] 48.00 53.00 76.00 95.00 124.00 149.00
Hunnargi et al. [23] 44.88 51.99 72.38 89.17 63.31 59.21
Hunnargi et al. [21] 44.88 51.99 72.38 89.17 117.25 141.16
Jit et al. [12] 48.42 51.73 78.60 95.35 24.32 27.45 29.19 32.58 127.02 147.08 61.80 55.53
Narayan and Varma [9] 58.98 54.76
Canada
Torwalt and Hoppa [19] 48.79 54.43 93.55 109.41 142.34 163.84
United States
Dwight [6] 46.70 51.80 89.50 105.80 136.20 157.70
Dwight [7] 49.40 53.70 91.90 110.4 141.30 164.10
McCormick et al. [15] 50.30 55.60 138.30 161.60
Stewart and McCormick [13] 137.60 159.60
Present study 48.24 51.84 89.38 104.80 24.29 27.35 30.15 34.47 136.75 154.97 53.68 48.41
populations are, on average, smaller than sterna from European, accurate indicator of sex. DFA achieved the highest classification
United States, and Canadian populations. rate, 84.12%. No difference between population groups was
Other interesting observations due to inter-study comparisons observed in this sample, therefore, American Black and American
were discovered between the North Indian and western Indian White individuals were analyzed together for differences between
populations [9,12,17,18,21,23]. On average, female sterna in North sexes. While population differences did not exist between
India are greater than those observed in western India. The male American Blacks and American Whites, considerable differences
sterna in North India, however, fall within range of the western were observed between the populations of previous studies. A
India populations. In addition, the current study consistently general trend was that sterna from East African and Indian
reported the lowest values for sterna measurements when populations are, on average, smaller than sterna from European,
compared to the other populations in the United States. United States, and Canadian populations. Furthermore, as would
Previous research on the sternum in the United States does not be expected in sexually dimorphic populations, female sterna are
provide the best methods for sex estimation due to shortcomings generally shorter than male sterna.
in materials and methods. As stated above, not only is Dwight’s [6] The present study was in agreement with previous studies that,
sample very small, but it is also out-dated. Stewart and McCormick while the mean values of sternal index conform to Hyrtl’s Law, the
[13] and McCormick et al. [15] both focused on costal cartilage range of values largely overlap, and prove unreliable for sex
ossification patterns in conjunction with sternal measurements estimation. Other measurements that did not show significant
observed in chest plate roentgenograms. This may not prove useful differences between male and female values were manubrium
to forensic anthropologists when presented with a skeletonized length, sternebra 1 width, and sternebra 3 width. Through the use
sternum that is no longer articulated with the rest of the thorax. of metrics of the sternum, the sex of the individual is able to be
The present study makes up for these shortcomings and provides estimated using corresponding accuracy rates. Utilizing metrics
the most reliable methods for use by forensic anthropologists on a yield an objective estimation of sex, unlike the subjective method
recent population from the United States by utilizing a sample of visual assessment. In addition, metrics allow less experienced
representing the U.S. population today, as well as generating individuals to utilize this research and provide accurate estima-
sectioning points with error rates and a classification function tions of sex, without relying on years of experience for confidence
achieving high accuracy rates. or familiarity with methods involving visual assessment. The most
reliable indices of the sternum are total sternal length and
4.1. Future areas of research discriminant function analysis, with accuracy rates of 81.5% and
84.12% respectively. This study was successful in generating a
Lastly, future research in the United States should more reliable classification function for use by forensic anthropologists.
accurately exemplify the modern day heterogeneity of the
population. This research has focused on American Blacks and
American Whites due to a limited sample of other population References
groups in the skeletal collection. Obtaining data on additional
[1] T.W. Phenice, A newly developed visual method of sexing the Os pubis, American
population groups from other collections would increase the Journal of Physical Anthropology 30 (2) (1969) 297–301.
usefulness of the sternum as an indicator of sex in the forensic [2] L. Sutherland, J. Suchey, Use of the ventral arc in pubic sex determination, Journal
of Forensic Sciences 36 (2) (1991) 501–511.
context (Table 4).
[3] W.D. Haglund, M.H. Sorg, Advances in Forensic Taphonomy: Method, Theory, and
Archaeological Perspective, CRC Press, New York, 2002.
5. Conclusion [4] P.L. Walker, Greater sciatic notch morphology: sex, age, and population differ-
ences, American Journal of Physical Anthropology 127 (2005) 385–391.
[5] M.K. Spradley, R.L. Jantz, Skull vs. postcranial elements in sex determination, in:
According to research conducted on recent donated skeletal 55th Annual Meeting of the American Academy of Forencis Sciences, Chicago,
remains from the United States, the sternum proves to be an Illinois, 2003.
R. Bongiovanni, M.K. Spradley / Forensic Science International 219 (2012) 290.e1–290.e7 290.e7
[6] T. Dwight, The sternum as an index of sex and age, Journal of Anatomy and [25] M.Y. Iscan, Forensic anthropology of sex and body size, Forensic Science Interna-
Physiology 15 (1881) 327–330. tional 147 (2) (2005) 107–112.
[7] T. Dwight, The sternum as an index of sex, height, and age, Journal of Anatomy 24 [26] P.L. Walker, Sexing skulls using discriminant function analysis of visually
(1890) 527–535. assessed traits, American Journal of Physical Anthropology 136 (2008) 39–50.
[8] G.T. Ashley, The human sternum: the influence of sex and age on its measure- [27] L.W. Konigsberg, B.F.B. Algee-Hewitt, et al., Estimation and evidence in forensic
ments, Journal of Forensic Medicine 3 (1) (1956) 27–43. anthropology: sex and race, American Journal of Physical Anthropology 139 (1)
[9] D. Narayan, H.C. Varma, Sternal index in U.P. Indian males and females, Journal of (2009) 77–90.
the Anatomical Society of India 7 (1958) 71–72. [28] SWGANTH. (2010). Sex Assessment. Retrieved November 17th, 2010, from http://
[10] P. Rother, H. Hunger, et al., Sex differences in the human sternum, Gegenbaurs swganth.startlogic.com/Sex%20Rev0.pdf.
morphologisches Jahrbuch 121 (1) (1975) 29–37. [29] W. Krause, Ueber das weibliche sternum, Monthly International Journal of
[11] M. Dokladal, Sex differences in the human sternum and thin practical importance Anatomy and Histology 14 (S) (1897) 21–26.
for the estimation of the sex of the human skeleton, Scripta Medica 51 (8) (1978) [30] J. Hyrtl, Handbuch der Topographischen Anatomie, K.K. Hofbuchhandler, Wien,
451–468. Wilhelm Braumuller, 1857.
[12] I. Jit, V. Jhingan, et al., Sexing the human sternum, American Journal of Physical [31] J. Hyrtl, Handbuch der topographischen anatomic percentage, Vienna: Wilhelm
Anthropology 53 (2) (1980) 217–224. Braumuller 1 (S) (1893) 348.
[13] J.H. Stewart, W.F. McCormick, The gender predictive value of sterna length, [32] M. Strauch, Anatom. Untersuchungen uber das Brustbein des Menschen, etc.,
American Journal of Forensic Medicine and Pathology 4 (3) (1983) 217–220. Inaugural Dissertation (1881).
[14] K. Teige, Morphometric studies of X-rays of the sternum, Zeitschrift fur Rechtsme- [33] F. Petermoller, Uber den sogen, geschleschstypus des menschlichen brustbeins,
dizin. Journal of Legal Medicine 90 (3) (1983) 199–204. Inaugural Dissertation in Medicine (1890).
[15] W.F. McCormick, J.H. Stewart, et al., Sex determination from chest plate roent- [34] H. Bogusat, Anomalien und varietaten des brustbeins, Inaugural Dissertation in
genograms, American Journal of Physical Anthropology 68 (1985) 173–195. Medicine (1902).
[16] G. Mall, G. Sprinzl, et al., Clinical morphology of the sternum, Biomedical [35] W.M. Bass, Human Osteology: A Laboratory and Field Manual of the Human
Technology (Berlin) 36 (1991) 288–289. Skeleton Columbia, Missouri Archaeological Society Special Publications, Mis-
[17] V.P. Dahiphale, B.H. Baheete, et al., Sexing the human sternum in marathwada souri, 1987.
region, Journal of the Anatomical Society of India 51 (2) (2002) 162–167. [36] R. DiBennardo, The use and interpretation of common computer implementations
[18] R.S Gautam, G.V. Shah, et al., The human sternum – as an index of age & sex, of discriminant function analysis, in: K.J. Reichs (Ed.), Forensic Osteology:
Journal of the Anatomical Society of India 52 (1) (2003) 20–23. Advances in the Identification of Human Remains, Charles C. Thomas, Springfield,
[19] C.R.M.M. Torwalt, R.D. Hoppa, A test of sex determination from measurements of IL, 1986, pp. 171–195.
chest radiographs, Journal of Forensic Sciences 50 (4) (2005) 785–790. [37] S. Institute, SAS/STAT User’s Guide, SAS Institute, Cary, NC, 1990.
[20] R. Selthofer, N. Vasilije, et al., Morphometric analysis of the sternum, Collegium [38] A.M. Christensen, C.M. Crowder, Evidentiary standards for forensic anthropology,
Antropologicum 30 (1) (2006) 43–47. Journal of Forensic Sciences 54 (6) (2009) 1211–1216.
[21] S.A. Hunnargi, R.G. Menezes, et al., Sexual dimorphism of the human sternum in a [39] T. Bohan, Strengthening forensic science: a way station on the journey to justice,
Maharashtrian population of India: a morphometric analysis, Legal Medicine Journal of Forensic Sciences 55 (1) (2010) 5–7.
(Tokyo) 10 (1) (2008) 6–10. [40] L.M. Jantz, R. Jantz, Secular changes in long bone length and proportion in the United
[22] P. Mahakkanukrauh, Thai Sternum and Sexing, 2008. States, 1800–1970, American Journal of Physical Anthropology 110 (1999) 57–67.
[23] S.A. Hunnargi, R.G. Menezes, et al., Sternal index: is it a reliable indicator of sex in [41] J.H. Schwartz., Skeleton Keys: An Introduction to Human Skeletal Morphology,
the Maharashtrian population of India? Journal of Forensic and Legal Medicine 16 Development, and Analysis. Oxford University Press. 2007.
(2) (2009) 56–58. [42] J. Wenzel. 1788, cited by Ashley, G.T. (1956): A comparison of human and
[24] W.M. Krogman, M.Y. Iscan, The Human Skeleton in Forensic Medicine, Charles C. anthropoid mesosterna. American Journal of Physical Anthropology 3: 449-461.
Thomas, Springfield, IL, 1986.