Forensic Science International 187 (2009) 109.e1109.

e3

Contents lists available at ScienceDirect

Forensic Science International

journal homepage: www.elsevier.com/locate/forsciint

Forensic Anthropology Population Data

Estimating stature from percutaneous length of tibia and ulna in

Indo-Mauritian population
Arun Kumar Agnihotri a,*, Smita Kachhwaha b, Vandna Jowaheer c, Ashok Pratap Singh b
a
Department of Forensic Medicine, SSR Medical College, Postal Code: 230, Mauritius
b
Department of Anatomy, SSR Medical College, Postal Code: 230, Mauritius
c
Department of Mathematics, University of Mauritius, Postal Code: 230, Mauritius

A R T I C L E I N F O A B S T R A C T

Article history: Stature estimation from percutaneous body measurements forms part of forensic anthropological
Received 23 October 2008 analysis for the purpose of identication. This study is aimed at modeling the stature on the basis of
Received in revised form 28 January 2009 percutaneous tibial and/or ulnar length in human subjects comprising of Indo-Mauritian population. The
Accepted 9 February 2009
study was conducted in the Department of Forensic Medicine and Toxicology, SSR Medical College,
Available online 14 March 2009
Mauritius on 180 young and healthy students comprising of 90 males and 90 females in the age group
ranging from 18 to 28 years. The measurements were taken by using standard anthropometric
Keywords:
instruments. It is remarked that tibial as well as ulnar length show a linear relationship with the stature,
Forensic Science
Forensic Anthropology where stature is normally distributed. However, it is required to transform the measurements on stature
Tibia length to obtain appropriate regression equations. Moreover, since ulna and tibia are signicantly correlated, it
Ulna length is recommended to use the sum of the ulnar and tibial length to estimate stature, in case both are
Stature estimation available for an individual. Our regression models are sufciently validated and highly efcient.
2009 Elsevier Ireland Ltd. All rights reserved.

1. Introduction 2. Materials and methods

2.1. Sample collection

Estimating individuality on the mutilated part of a dead body is
a difcult task in Forensic Medicine. The problem of identication
mainly arises when unknown human remains are brought to
forensic pathologists for examination. The stature of an individual
is an inherent characteristic. It is considered as one of the
important parameters for personal identication. There is an
established relationship between stature and dimensions of
various body parts, which allows forensic experts to estimate
stature from different parts of the body. Many studies have been
carried out to estimate stature from percutaneous body measure-
ments [121]. Linear regression models are widely used to predict
height of individuals on the basis of their body parts.
In circumstances where only mutilated leg and/or forearm
portions are available for a deceased person, it becomes quite
challenging to formulate the proper and adequate regression
models. Not much research work is done in this direction. In this
paper we construct linear regression models to predict heights on
the basis of ulna and tibia measurements for Indo-Mauritian
population.
* Corresponding author at: 4, Malherbes Street, Curepipe, Mauritius.
Tel.: +230 6751318.
E-mail address: agnihotri_arun@hotmail.com (A.K. Agnihotri).
The study was conducted in the Department of Forensic Medicine and
Toxicology, SSR Medical College, Mauritius in the year 2007. Mauritius is an
island, geographically located in the Indian Ocean, approximately 2400 km off the
south east coast of Africa. The material consisted of 180 young and healthy students
(90 males and 90 females), in the age group ranging from 18 to 28 years. Each
student has been studied for the measurements of stature, tibial length and ulnar
length. The measurements were taken by using standard anthropometric
instruments.
2.2. Techniques for taking measurements
The measurements were taken by using standard anthropometric instruments,
in centimeters, to the nearest millimeter according to the technique described by
Vallois [22]. All measurements were taken by the same observer and with the same
instrument, to avoid any technical and/or inter-observer error and to maintain
reproducibility. During data collection, the instruments like sliding caliper,
spreading caliper, etc. were regularly checked for their accuracy. Percutaneous
tibial and ulnar lengths were chosen because their lengths could be easily taken in
living individuals.
Height of the subject was measured in standing position. The subject was
instructed to stand barefooted on the board of a standard standiometer with both
feet in close contact with each other, trunk braced along the vertical board, and the
head adjusted in Frankfurt plane. The measurement was taken in centimeters, by
bringing the horizontal sliding bar to the vertex. The length of tibia was measured as
a distance from the inner border of the medial condyle to the farthest point of the
medial malleolus. The measurement was taken on the sitting subjects with knees in
semiexed position. The length of ulna was measured as a straight distance from
the most proximal point of the olecranon process to the most distal point of the
styloid process, with the forearm exed at a 908 angle.

0379-0738/$ see front matter 2009 Elsevier Ireland Ltd. All rights reserved.
doi:10.1016/j.forsciint.2009.02.010
109.e2 A.K. Agnihotri et al. / Forensic Science International 187 (2009) 109.e1109.e3

2.3. Regression models analyzing data on tibial length, ulnar length and height using linear
Regression models are constructed to estimate stature on the basis of only tibial regression models. Our ndings are quite different from the earlier
length, only ulnar length and ulnar as well as tibial length. A sample of 80 females studies [4,7]. Mohanty [4] suggested gender-based two different
and 80 males was selected randomly out of a total of 90 females and 90 males, who regression equations to predict height among Oriya population
formed part of the study. The data on these 160 subjects were used to construct the when only the leg portion is available. However, this method suffers
models and the data on the remaining 20 subjects were used to test the predictive
ability of the models.
from a serious drawback in its model formulation stage itself. If the
leg portion as such cannot be identied as being male or female; then
3. Results how would one make a choice between two equations in order to
predict height of the dead person by simply using percutaneous
There was no signicant difference in the lengths of right and tibial length? Recently, Duyar et al. [7] have proposed to use a
left tibia in both males (mean difference = 0.009, p-value = 0.893) stature-specic linear regression approach, taking into account the
and females (mean difference = 0.008, p-value = 0.910). Similarly tail observations. The approach is statistically sound. However, the
no signicant differences were observed in the lengths of right and application and validity of the stature-specic regression models as
left sides of ulna in both males (mean difference = 0.024, p- discussed by these authors are not convincing, and rather
value = 0.857) and females (mean difference = 0.020, p- misleading to a great extent. For example, the height is predicted,
value = 0.996). Hence we have used right tibial length and right based on single short ulna using general regression model, as well as
ulnar length for the construction of models. stature-specic regression model. It is well understood that the
The relationship between tibial length (TL) and stature is linear single short ulna belongs only to one person whose height is
in the Indo-Mauritian population. The same pattern exists for ulnar predicted using two different models. Thus two values of estimated
length (UL) and stature. Stature, TL and UL are all normally heights are reasonable, but it is certainly not possible to have two
distributed. It is observed that TL and UL are signicantly different values of the actual height of that person. Also, the
correlated. When stature is regressed on TL the regression equation corresponding t-values and p-values are possible only if predictors
does not satisfy the underlying assumption of constant error are derived for the entire group 2 and not just for one actual
variance. A transformation on the response variable stature is observation. Moreover, several models based on the predictors
therefore identied from the family of power transformations mentioned in this study cannot be compared blindly with each
using Box-Cox procedure, such that other. This is because the model based on ulna is comparable to
model based on tibia, as well as the model based on ulna and tibia,
yl  1 only under circumstances when both ulna and tibia of the same
yl (1.1)
ly l1 person are available for investigation, failing which the comparison
Q will lead to invalid conclusions.
1=n
where l = 2, y represents stature, y ni1 yi and n = 160. The models (1.2)(1.4) are adequate and satisfy all underlying
Hence, the tted normal error regression model with TL as the assumptions. For a known TL value, estimate of stature can be
only covariate is given by obtained by using Eqs. (1.1) and (1.2). In the same way for a known
TL and UL, estimate of stature can be obtained by using Eqs. (1.1)
and (1.4). For a given UL, the estimate of stature can be directly
2
y 21:0 2:83 TL (1.2) obtained by using Eqs. (1.1) and (1.3). Tibial length has stronger
p-val0:000 p-val0:000
linear relation with stature as compared to ulnar length. The
R2 = 81.4%. validity of all the models has been veried by conducting the tests
In case, when UL is the only covariate, the tted regression for no difference between the estimated and the corresponding
model is given by exact values of the stature for the sample of 20 individuals
comprising unused set. For all the models p-values for these tests
turn out to be greater than 0.10 suggesting that all the models bear
y 60:4 4:13 UL (1.3) signicant predictive ability. It is concluded that the stature of a
p-val0:000 p-val0:000
deceased person for whom only mutilated legs or arms are
R2 = 74.0%. available, can be determined with fair degree of accuracy by using
When both UL and TL are available for an individual, it is the model equations derived from the present study.
recommended to construct a linear regression model using the
variable TUL = TU + UL as a single covariate than constructing a 4.1. Limitation of study
multiple linear regression model with two covariates UL and TL.
This is because UL and TL are signicantly correlated (p-val = 0.97). This study has been conducted mainly in Indo-Mauritian
However, when stature is regressed on the TUL, the assumption of population. Therefore other studies in different parts of the world
constant error variance is violated. Hence Box-Cox transformation are required to conrm whether it would be equally applicable
is applied to the response variable. In this case also, l = 2 and elsewhere.
following Eq. (1.1), the tted regression model is obtained as
2
References
y 31:5 1:82 TUL (1.4)
p-val0:000 p-val0:000
[1] D.P. Bhatnagar, S.P. Thapar, M.K. Batish, Identication of personal height from the
somatometry of the hand in Punjabi males, Forensic Sci. Int. 24 (1984) 137141.
R2 = 85.8%. [2] S.K. Saxena, A study of correlations and estimation of stature from hand length,
hand breadth and sole length, Anthropol. Anz. 42 (1984) 271276.
4. Discussion and conclusion [3] A.K. Abdel-Malek, A.M. Ahmed, S.A. el-Sharkawi, N.A. el-Hamid, Prediction of
statue from hand measurements, Forensic Sci. Int. 46 (1990) 181187.
[4] N.K. Mohanty, Prediction of height from percutaneous tibial length amongst Oriya
The results of the present study indicate that the percutaneous population, Forensic Sci. Int. 98 (1998) 137141.
length of tibia and ulna can be efciently used for the estimation of [5] A. Ozaslan, M.Y. Iscan, I. Ozaslan, H. Tugcu, S. Koc, Estimation of stature from body
stature. The hypothesis pertaining to the relationship of stature with parts, Forensic Sci. Int. 132 (2003) 4045.
[6] S.G. Sanli, E.D. Kizilkanat, N. Boyan, E.T. Ozsahin, M.G. Bozkir, R. Soames, H. Erol, O.
the dimensions of different body parts is veried and supported by Oguz, Stature estimation based on hand length and foot length, Clin. Anat. 18
results. This study extends the ndings of previous researches by (2005) 589596.
 A.K. Agnihotri et al. / Forensic Science International 187 (2009) 109.e1109.e3
[7] I. Duyar, C. Pelin, R. Zagyapan, A new method of stature estimation for forensic
anthropological application, Anthropol. Sci. 114 (2006) 2327.
[8] A. Ozaslan, S. Koc, I. Ozaslan, H. Tugcu, Estimation of stature from upper extremity,
Mil. Med. 171 (2006) 288291.
[9] O. Celbis, H. Agritmis, Estimation of stature and determination of sex from radial
and ulnar bone lengths in a Turkish corpse sample, Forensic Sci. Int. 158 (2006)
135139.
[10] M.M. El-Meligy, R.H. Abdel-Hady, R.M. Abdel-Maaboud, Z.T. Mohamed, Estima-
tion of human body built in Egyptians, Forensic Sci. Int. 159 (2006) 2731.
[11] K.R. Nagesh, G. Pradeep Kumar, Estimation of stature from vertebral column
length in South Indians, Leg. Med. (Tokyo) 8 (2006) 269272.
[12] K. Krishan, R. Kumar, Determination of stature from cephalo-facial dimensions in
a North Indian population, Leg. Med. (Tokyo) 9 (2007) 128133.
[13] A.K. Agnihotri, B. Purwar, K. Googoolye, S. Agnihotri, N. Jeebun, Estimation of
stature by foot length, J. Forensic Leg. Med. 14 (2007) 279283.
[14] K. Krishan, A. Sharma, Estimation of stature from dimensions of hands and feet in
a North Indian population, J. Forensic Leg. Med. 14 (2007) 327332.
109.e3
[15] P. Rastogi, K.R. Nagesh, K. Yoganarasimha, Estimation of stature from hand
dimensions of north and south Indians, Leg. Med. (Tokyo) 10 (2008) 185
189.
[16] J. Sen, S. Ghosh, Estimation of stature from foot length and foot breadth among the
Rajbanshi: an indigenous population of North Bengal, Forensic Sci. Int. 181 (2008)
55.e155.e6.
[17] G. Zeybek, I. Ergur, Z. Demiroglu, Stature and gender estimation using foot
measurements, Forensic Sci. Int. 181 (2008) 54.e154.e5.
[18] T. Kanchan, R.G. Menezes, R. Moudgil, R. Kaur, M.S. Kotain, R.K. Garg, Stature
estimation from foot dimensions, Forensic Sci. Int. 179 (2008) 241.e1241.e5.
[19] A.K. Agnihotri, N. Jeebun, S. Agnihotri, K. Googoolye, Prediction of stature using
hand dimensions, J. Forensic Leg. Med. 15 (2008) 479482.
[20] K. Krishan, Estimation of stature from footprint and foot outline dimensions in
Gujjars of North India, Forensic Sci. Int. 175 (2008) 93101.
[21] K. Krishan, Estimation of stature from cephalo-facial anthropometry in north
Indian population, Forensic Sci. Int. 181 (2008) 52.e152.e6.
[22] H.V. Vallois, Anthropometric techniques, Curr. Anthropol. 6 (1965) 127144.