Home About Us Laboratory Services Forensic Science Communications Back Issues July 2002 "Sexual Dimorphism in Femora: An Indian Study,...

"Sexual Dimorphism in Femora: An Indian Study, Forensic Science Communications, July 2002

"Sexual Dimorphism in Femora: An Indian Study, Forensic Science Communications, July 2002


July 2002 - Volume 4 - Number 3

Research and Technology

Sexual Dimorphism in Femora: An Indian Study

Ruma Purkait
Department of Anthropology
Saugor University
Saugor, India

Heeresh Chandra
Former Director and Founder
Medico-Legal Institute
Bhopal, India

Abstract | Introduction | Materials and Methods | Results | Discussion | References


In the field of forensic osteology, determining sex from skeletal remains, especially from isolated bones, has been an age-old problem. This study documents efforts to determine sex by using five measurements from the femur. The study is based on 200 male and 80 female femora from central India. The data are analyzed using discriminant function procedures, and results of different measurements are reported independently and in various combinations. Three variables combined into a function could correctly assign sex to 92 percent of males and 96.3 percent of females.


The role of the skeleton in estimating attributes such as age, sex, race, stature, and the presence of disease is discussed by Krogman and Iscan (1986). They stated that the record of organic evolution is largely written by the hard parts of the body recognizable even after many years of death. Sex determination from skeletal remains, which forms an important component in the identification procedure, sometimes becomes a difficult task for the forensic anthropologist, especially in the absence of the pelvis. Therefore, most of the long bones, either individually or in combination, have been subjected to statistical and morphological analysis for the purpose of determining sex. So far, several studies conducted for assessing sex from various skeletal parts have reiterated that there is a size difference between populations and that metric standards must be developed for each group. Citing the example of the femur, studies have been reported on various populations including the Finns (Lofgren 1956); French (Godycki 1957); Japanese (Hanihara 1958); Australian aborigines (Davivongs 1963); English (Steel 1972); American blacks, whites, and Indians (Black 1978; DiBennardo and Taylor 1979 and 1982; Iscan and Miller-Shaivitz 1984 and 1986); Italians (Pettener 1979); Czechs (Cerny and Komenda 1980); prehistoric Scottish (MacLaughlin and Bruce 1985); archeological remains of Sudanese Nubians, Pecos Pueblo Indians, and Arikaras (France 1988); Chinese (Liu 1989; Iscan and Shihai 1995); Spanish (Trancho et al. 1997); Nigerians (Asala et al. 1998); Thais (King et al. 1998); South African whites and blacks (Asala 2001; Steyn and Iscan 1997); and Germans (Mall et al. 2000). Little work on the subject has been reported from India except for the study by Singh and Singh (1972A and 1972B) on the head of femur. To date, nothing has been published on other measurements of the femur, which may be useful if the bone is fragmented. This study is an attempt to examine the sexual dimorphism in femora of Indian origin.

This study on sexual dimorphism is based on the principle that the axial skeleton weight of the male is relatively and absolutely heavier than that of the female (William et al. 1989), and the initial impact of this weight is borne by the femur in transmission of the body weight. Another factor that makes its indentation on the femur is the modification of the female pelvis with respect to its specialized function of reproduction. Therefore, the stress and strain experienced by the femur is different in a male than it is in a female.

Materials and Methods

Data for this study are comprised of 280 dry adult femora from 200 male and 80 female residents of central India. The collection was housed at the Medico-Legal Institute of Bhopal, India. Abnormal or pathologically deformed bones were excluded from the study. Most of the bones in the collection, stored since 1973, are forensic specimens, and only a few are unclaimed specimens. Every care had been taken by the authors to include bones from a homogenous population. Information on probable age at death, race, sex, date of arrival, and probable cause of death were well documented in a register after examination. The bones were preserved in iron boxes coded with a serial number.

In order to test for bilateral variation in the measurements, 20 sets of femora were subjected to a paired t-test. The difference was found to be insignificant at the 0.05 level, thus allowing the bones of both sides to be grouped together. However, only one bone, either left or right, has been included in the analysis. A set of five anthropometric measurements was taken on each femur: maximum length, maximum diameter of the head, midshaft circumference, maximum anteroposterior diameter of the femoral shaft, and epicondylar width. Maximum length, midshaft circumference, and epicondylar width were measured following the standard techniques recommended by Martin and Saller (1957). Maximum head diameter and maximum anteroposterior diameter were measured following the technique given by Brauer (1988) and MacLaughlin and Bruce (1985). The latter had measured the maximum anteroposterior diameter of the shaft between superiorly, the inferior margin of gluteal ridge, and inferiorly, the level at which the two lips of the linea aspera diverge to form the supracondylar lines. A Mitutoyo-dial caliper was used to measure the maximum head diameter, the maximum anteroposterior diameter, and the epicondylar width nearest to 1/100 mm.

Data were analyzed using the SPSSX Subroutine software (SPSS Incorporated, Chicago, Illinois). Stepwise discriminate function analysis employing measurements was used to determine the optimal combination of variables for assessing sex. Variables, alone and in combination, were also subjected to direct analysis to develop functions to allow sex determination from fragmentary remains. To further test the efficiency of the discriminant functions derived from the previous analysis, they were applied to a randomly chosen test group of 43 femora (29 males and 14 females). The test group was not a part of the original sample but did consist of bones of the same class and population as the original one.


This study used measurements taken for five variables. Table 1 shows the routine statistical analysis of these variables accounting for various measurements. The standard deviations are given with an F-ratio for each measurement. With the average male skeleton being longer, more muscular, and heavier than the average female, all the measurements in Table 1 exhibit a highly significant sex difference. The combined coefficient of variation for male and female in the table exhibit highest relative variation in the sample for maximum anteroposterior diameter of the femoral shaft. Table 2 gives the summary of the stepwise discriminant function analysis. The data are reported under three headings: the Wilk's Lambda, equivalent F-ratio, and calculated degree of freedom. The table shows that out of five variables entered, four were selected for the analysis. Once the epicondylar width was analyzed, the remaining variables were reassessed and selected according to the Lambda level. The maximum anteroposterior diameter having the least Lambda value was entered from the remaining variables as Step 2. Included in the analysis were the maximum head diameter as Step 3 and the maximum length as Step 4. The analysis was terminated after Step 4, probably because of the extremely low value of F-ratio of the remaining variable, which was below the threshold of criteria for entrance.

After the stepwise discriminant function analysis, the variables were entered directly to provide various combinations, some of which may be used for fragmentary remains. The functions and their coefficients are presented in Table 3. The raw coefficients are used to calculate the discriminant scores for the functions. The sectioning point has been set to zero. As a result, when the product of the predictor variable and its coefficient added to the constant is above zero, the individual can be classified as male. If the product added to the constant is below zero, the individual can be classified as female. The standard coefficient column indicates the contribution of a variable to the discriminant score relative to other variables. In this study, epicondylar width has the maximum discriminating power. The structure coefficient, the next column, gives an idea of what a variable contributes to a function on its own. Again, epicondylar width has the highest contribution (0.88844).

Table 4 presents the percentage of correct group membership. This gives the accuracy of prediction for each function. The first column shows the accuracy for males, the second for females, and the last shows the average for both sexes. These functions can be grouped into four categories. In the first category using a single variable (Functions 1-5), the accuracy ranges from 84.3 to 91.1 percent, and the best discriminator is maximum head diameter. It is interesting to note that, although the F-ratio value for epicondylar width is greater than maximum head diameter, it is probably the latter's smaller relative variation in the sample (coefficient of variation = 5.59) that is responsible for its edge over the former (coefficient of variation = 5.77).

The second category uses two variables in combination (Functions 6-8), which can be used when one or more anatomical areas are damaged. The combination of epicondylar width and maximum head diameter gives 92.1 percent accuracy. The third category, consisting of Function 9 using three variables in combination, shows a marked increase to 93.2 percent in prediction accuracy. The last category (Function 10) is the result of stepwise analysis; although four variables are used in combination, the prediction accuracy remains unchanged. The results of discriminant functions that were applied on the original sample were applied on the test cases and presented in Table 5. The success rate of identification was lower when compared to the original sample on Table 4. This was expected, because the sample from which the functions had been drawn gives maximum accuracy for the same sample compared to any other sample, although drawn from the same population. Table 6 exhibits the results of cross testing the Indian data using formulae of Thai, Chinese, American whites, and South African whites and contrasts them with accuracies obtained from their original studies. The American white formula identified only 19 percent of the males and classified most of the males as females. The result of the South African formula is similar, exhibiting a slight improvement (27 percent for males). The results of Chinese (58.5 percent) and Thai (63.5 percent) formulae are comparatively better, although nowhere near the accuracy attained on the Indian sample using its own population-specific formula (93.2 percent).


It is known that the average male skeleton is longer and more robust than the average female, although the magnitude of difference varies from population to population. This sex difference can be the result of genetic factors, environmental factors affecting growth and development (nutrition, physical activity, and pathologies), or the interaction of these factors (Trancho et al. 1997). The results of this study show that the femoral extremities display higher classification accuracy (91.1 percent for maximum head diameter and 89.6 percent for epicondylar width) than shaft dimensions. The extremities of the bone are the areas where a number of muscles make their insertions and are subjected to more pull than at the point of origin. Also, as suggested by France (1988), the articular surfaces of the bone receive a portion of the force being applied across them, and as such, the extremities of the femur will react to such forces.

Pons (1955), while working on Portuguese femora, had opined that the head diameter and width of the lower end discriminated sex better than any other part of the bone. Of the 17 variables analyzed by Van Gerven (1972) for sex difference by discriminant function, epicondylar width in isolation produced the greatest male-female discrimination. Dittrick and Suchey (1986) also concluded that the ends of the femur produced 10 percent greater accuracy than either femoral length or midshaft circumference. In most of the recent studies on Spanish (Trancho et al. 1997), Chinese of Qingdao and Changchun cities (Iscan and Shihai 1995), and South African whites (Steyn and Iscan 1997), the extremity measurements proved better, with epicondylar width outdoing head diameter. It is the reverse for the present sample and also among northeastern Chinese, as reported by Liu (1989).

Although sex determination with maximum anteroposterior diameter was discussed by MacLaughlin and Bruce in 1985, its efficacy has not been tested on many populations. The results of this study confirm that the diameter is a good indicator of sex, with classification accuracy reaching 85.7 percent. Maximum anteroposterior diameter of the femoral shaft is directly related to the muscle attachment to the bone. Several muscles make their insertion at the linea aspera, which is the area of measurement. Males generally use their muscles more powerfully due to both heavier body weight and additional action, resulting in greater pull at the insertions. Moreover, the diameter of the shaft is related to the weight borne by the femur. In males, the axial skeleton weight is relatively and absolutely more than that of female (William et al. 1989).

It is a common experience for the forensic expert to be confronted with poorly preserved or fragmentary bones. Because of the tubular structure of long bones, they are often better preserved than other shorter bones. This long bone measurement has an additional advantage. Unlike some of the previous studies, this study shows that the midshaft circumference measurement displays less classification accuracy than length measurement.

Table 4 demonstrates that there is a gap in the accuracy between sexes for all the functions, with the females being consistently on the higher side. This is probably due to the combined effect of the unequal sample size and the intrasex variation.

Figure 1 is a line graph showing the comparative data of male measurments for Indian, Thai, Chinese, Sourth African white, American white, and American black populations.
Figure 1 Comparative Data of Male Measurements for Six Populations Click to enlarge image.
Figure 2 is a line graph showing the comparative data of female measurments for Indian, Thai, Chinese, Sourth African white, American white, and American black populations.
Figure 2 Comparative Data of Female Measurements for Six Populations Click to enlarge image.

The population variation is graphically represented in Figures 1 and 2. Comparison has been made between the results of those recent studies where at least four variables are common. All the measurements of the present study, both male and female, are comparable to Thai data except for male maximum femoral lengths that are closer to American whites. The nearness of data of the present study to the Thai sample does not indicate any racial proximity. It does indicate that, on the average, the girth (midshaft circumference) of the Indian male is nearly equal to the girth of the American female, white and black. As can be seen in Table 6 for American whites and Thai populations, stepwise discriminant function formulae have been used for cross testing. This is in contrast to Chinese and South African whites where other suitable functions were selected to be applied on the present sample, because stepwise formula used variables not included in the present study. Ignoring this limitation, when the results of the cross testing are compared, it is evident that among the four populations, the Thai formula gave the best results. This was expected, because all the dimension values of the Indian population (except for maximum length in males) were closest to the Thai sample.

Thus, this study reconfirms the fact that osteometric assessment is highly population- specific. It may be added that more studies are required in south Asia to give a better picture of the racial variation that exists there and to offer more osteometric standards for assessing sex.


Asala, S. A. Sex determination from the head of the femur of South African whites and blacks, Forensic Science International (2001) 117:15-22.

Asala, S. A., Mbajiorgu, F. E., and Papandro, B. A. A comparative study of femoral head: Diameters and sex differentiation in Nigerians, Acta Anatomica (1998) 162:232-237.

Black, T. K. A new method for assessing sex of fragmentary remains: Femoral shaft circumference, American Journal of Physical Anthropology (1978) 48:227-231.

Brauer, G. Osteometrie in anthropologie: Handbuch der vergleichenden biologie des Menschen. Ban 1, Wesen und methoden der anthropologie, Teil 1. Wissenschaftstheorie, Geschichte, Morphologische Methoden. R. Knussman, ed., Gustav Verlag, Stuttgart, 1988.

Cerny, M. and Komenda, S. Sexual diagnosis by the measurement of humerus and femur, Sbornik Praci Pedogogicke Fakulty, Univetzity Palackeho v Olomouci, Biologie (1980) 2:147-167.

Davivongs, V. The femur of Australian aborigine, American Journal of Physical Anthropology (1963) 21:457-467.

DiBennardo, R. and Taylor, J. V. Sex assessment of the femur: A test of new method, American Journal of Physical Anthropology (1979) 50:635-638.

DiBennardo, R. and Taylor, J. V. Classification and misclassification in sexing the black femur by discriminant function analysis, American Journal of Physical Anthropology (1982) 58:145-151.

Dittrick, J. and Suchey, J. M. Sex determination of prehistoric central California skeleton remains using discriminant analysis of the femur and humerus, American Journal of Physical Anthropology (1986) 70:3-9.

France, D. L. Osteometry at muscle origin and insertion in sex determination, American Journal of Physical Anthropology (1988) 76:515-526.

Godycki, M. Sur la certitude de determination de sexe d'apres le femur le cubitus, et l'humerus, Bulletin et Mem de la Soc d'Anthropologie de Paris (1957) 110:405-410.

Hanihara, T. Sexual diagnosis of Japanese long bones by means of discriminant function, Journal of Anthropological Society Nippon (1958) 67:21-27.

Iscan, M. Y. and Miller-Shaivitz, P. Determination of sex from femur in blacks and whites, Collegium Anthropology (1984) 8:169-177.

Iscan, M. Y. and Miller-Shaivitz, P. Sexual dimorphism in the femur and tibia. In: Forensic Osteology: Advances In Identification of Human Remains, K. J. Reichs, ed. Charles C. Thomas, Springfield, 1986.

Iscan, M. Y. and Shihai, D. Sexual dimorphism in the Chinese femur, Forensic Science International (1995) 74:79-87.

King, C. A., Iscan, M. Y., and Loth, S. R. Metric and comparative analysis of sexual dimorphism in the Thai femur, Journal of Forensic Sciences (1998) 43:954-958.

Krogman, W. M. and Iscan, M. Y. Human Skeleton in Forensic Medicine. 2nd ed., Charles C. Thomas, Springfield, 1986.

Lofgren, L. Some anthropometric anatomical measurements of the femur of Finns from the view point of surgery, Acta Chir Scandinavia (1956) 110:477-484.

Liu, W. Sex determination of Chinese femur by discriminant function, Journal of Forensic Sciences (1989) 34:1222-1227.

MacLaughlin, S. M. and Bruce, M. F. A simple univariate technique for determining sex from fragmentary femora: Its application to a Scottish short cist population, American Journal of Physical Anthropology (1985) 67:281-288.

Mall, G., Graw, M., Gehring, K. D., and Hubig, M. Determination of sex from femora, Forensic Science International (2000) 113:315-321.

Martin, R. and Saller, K. Lehrbuch der Anthropologie. Gustav Fisher Verlag, Stuttgart, 1957.

Pettener, D. La determinazione del sesso mediante analisi multivariatadi carrater metrici del femore, Rivista di Antropologia (1979) 60:281-288.

Pons, J. The sexual diagnosis of isolated bones of the skeleton, Human Biology (1955) 27:12-21.

Singh, S. P. and Singh, S. The sexing of adult femora: Demarking points for Varanasi zone, Journal of the Indian Academy of Forensic Sciences (1972A) 11:1- 6.

Singh, S. P. and Singh, S. Identification of sex from the head of the femur: The demarking points for Varanasi zone, Indian Medical Gazette (1972B) 11:45-49.

SPSS/PC: A Statistical Guide for IBM PC/XT/AT and PS/2. SPSS Incorporated, Chicago. 1988.

Steel, F. L. D. The sexing of the long bones with reference to the St. Bride's series of identified skeletons, Journal of the Royal Anthropological Institute of Great Britain and Ireland (1972) 92:212-222.

Steyn, M. and Iscan, M. Y. Sex determination from the femur and tibia in South African whites, Forensic Science International (1997) 90:111-119.

Trancho, G. J., Robledo, B., Lopez-Bueis, I., and Sanchez, A. Sexual determination of femur using discriminant function analysis of a Spanish population of known sex and age, Journal of Forensic Sciences (1997) 42:181-185.

Van Gerven, D. P. The contribution of size and shape variation to patterns of sexual dimorphism of human femur, American Journal of Physical Anthropology (1972) 37:49-60.

William, P. L., Warwick, R., Dyson, M., and Bannister, L. H. Gray's Anatomy. Churchill Livingstone, Edinburgh, 1989.

Top of the page