Forensic Science International 216 (2012) 199.e1–199.e7

Contents lists available at SciVerse ScienceDirect

Forensic Science International

journal homepage: www.elsev ier .com/ locate / forsc i in t

Forensic Anthropology Population Data

Estimation of stature from hand and handprint dimensions in a WesternAustralian population

Nur-Intaniah Ishak a,b, Naomi Hemy a, Daniel Franklin a,*a Centre for Forensic Science, The University of Western Australia, M420, 35 Stirling Hwy, Crawley, 6009 Western Australia, Australiab Institute of Biological Sciences, Faculty of Science, University of Malaya, Kuala Lumpur, Malaysia

A R T I C L E I N F O

Article history:

Received 6 June 2011

Received in revised form 11 July 2011

Accepted 19 September 2011

Available online 11 October 2011

Keywords:

Stature

Forensic anthropology

Hand anthropometry

Population standards

Handprints

Forensic anthropology population data

A B S T R A C T

As part of the formulation of a biological profile, the estimation of stature is an important element that

provides useful data towards narrowing the pool of potentially matching identities. Recent literature has

demonstrated that anthropometry of the hand has considerable promise for the accurate estimation of

stature; although the technique has only been tested in a relatively limited range of populations. The aim

of the present study, therefore, is to assess the reliability and accuracy of using anthropometric hand

measurements for the estimation of stature in a contemporary Western Australian population; we also

evaluate whether stature can be accurately estimated from the measurement of handprints.

The study sample comprises 91 male and 110 female adult individuals. Following the measurement of

stature, seven measurements are taken on each hand and its corresponding print. To establish the

reliability of acquiring these measurements, a precision study was performed prior to primary data

collection. Measurements data are analysed using basic univariate statistics and simple and multiple

regression analyses. Our results show that the degree of measurement error and reliability are well

within accepted standards. Stature prediction accuracy using hand and handprint measurements ranges

from �4.74 to 6.53 cm, which is comparable to established skeletal standards for the hand. This study

provides new forensic standards for the estimation of stature in a Western Australian population and also

demonstrates that the measurement and analysis of handprints affords a novel source of profiling data that is

statistically quantified.

� 2011 Elsevier Ireland Ltd. All rights reserved.

1. Introduction

The modern role of a forensic anthropologist has evolved toinclude not only the study of skeletal remains, but in an era ofglobal migration (both legal and illegal), and incidences ofincreased terrorism, crime and natural disasters involving massfatalities, the analysis of living individuals and/or body parts ofdecedents is frequently required. The minimum expertiserequirement of a practicing forensic anthropologist mustinclude the ability to formulate a biological profile (osteobio-graphy – sex, age, ethnicity and stature). A biological profile,alongside other skeletal markers (e.g. ante-mortem pathology)can provide useful information towards establishing theidentity of unknown skeletal remains; for investigating authori-ties, this profile effectively narrows down the pool of potentialmatching (tentative) identities – positive identification canthen be established using traditional markers (e.g. DNA ordental) [1].

* Corresponding author. Tel.: +61 8 6488 1232; fax: +61 8 6488 7285.

E-mail address: daniel.franklin@uwa.edu.au (D. Franklin).

0379-0738/$ – see front matter � 2011 Elsevier Ireland Ltd. All rights reserved.

doi:10.1016/j.forsciint.2011.09.010

It is generally accepted that the most accurate biological profileis formulated using contemporary population specific standards.This issue has been considered many times in the literature andbasically relates to population specific variability in the relativeexpression of morphological characteristics used to estimate, forexample, sex, age and stature (e.g. [2–4]). Populational variabilityis also not ‘temporally static’, thus contemporary standards,devised on modern representative samples, are required. Seculartrends in stature are an example of the importance of contempo-rary forensic standards. For example, under conditions of adequateaccess to nutrition and health services, children and adults mayattain a larger stature compared to individuals from the samepopulation from several generations earlier; this is known as apositive secular trend. Conversely, the opposite effect (negativesecular trend) has also been demonstrated to occur underconditions of poor nutrition and health [5].

In instances where a forensic investigator is presented withcommingled, dismembered and/or incomplete remains, statureestimation standards requiring a complete skeleton (anatomicalmethod – e.g. [6–8]) or complete long bones (mathematicalmethod – e.g. [9,10]) may not be applicable. In such circumstancesstandards devised from different parts of the skeleton or body

Table 1Definition of the hand and handprint measurements used in the present study with

reference to source (where appropriate).

Measurement Definition

Hand breadth (HB)a Distance between the most lateral point

on the head of the 2nd metacarpal to the

most medial point on the head of

the 5th metacarpal [18]

Hand length (HL) Distance between the mid-point of the

distal transverse crease of the wrist to

the most anterior projection of the skin

of the middle finger [23]

Palm length (PL) Distance from the mid-point of the distal

transverse crease of the wrist to the

proximal flexion crease of the middle

finger [37]

Thumb (1D); Index (2D);

Middle (3D);

Ring (4D) Finger length

Distance between the proximal flexion crease

of the finger to the tip of the

respective finger

Handprint breadth (HPB) Distance from the most laterally projected

part of the palm print at the 2nd metacarpal

to the most medially projected part of the

palm print at the distal transverse crease

Handprint length (HPL) Distance from the baseline of the print

(transverse line from the most inferior

point of the medial border of the palm) to

the tip of the middle finger (measurement

modified from a technique based on

footprints [see 38])

a Requires manual palpation of the hand to locate the required bony anatomy.

N.-I. Ishak et al. / Forensic Science International 216 (2012) 199.e1–199.e7199.e2

limbs may provide a practical alternative. There are numerouspublications describing anthropometric approaches to estimatingstature from different body parts. The feet (and to a lesser degreefootprints) have been shown to be a relatively accurate biologicalcharacteristic from which stature can be estimated (e.g. [11–13]).Other studies have considered the relationship between statureand head dimensions in living individuals (e.g. [14–16]) andprovided standards for their respective populations. It is importantthat the forensic investigator select the most appropriatemethod(s) and any further related studies are solidly based on ameaningful biological, and not just statistical, relationshipbetween body parts and living stature (see [17]).

Previous research in various populations has demonstrated thatstature can be accurately estimated from hand dimensions: e.g.North Indian, SEE �3.16–5.60 cm [18]; South Indian, SEE �3.65–5.73 cm [19]; Turkish, SEE �3.91–4.59 cm [20]; Egyptian, SEE �5 cm[21], SEE �4.22–5.48 cm [22]; Mauritanian, SEE �4.16–4.96 cm [23].Most recently, Ahemad and Purkait [24] demonstrated that staturecan also be accurately estimated from hand impressions in an Indianpopulation; SEE �4.64–5.50 cm. The studies outlined above demon-strate the forensic utility of anthropometric measurements of thehand and handprints for estimating stature and, importantly, with anaccuracy closely approaching that of established long bone standards(e.g. [9]). Clearly there is an abundance of research on populationsprimarily from India and the Middle East. It would be useful, however,to assess whether this method of stature estimation has forensicpotential in other populations.

There are currently no population specific standards forestimating stature in a Western Australian population. To thatend, the purpose of the present study is to assess the reliability andaccuracy of using anthropometric hand measurements for theestimation of stature in that population. These standards wouldafford a means of estimating stature in forensic investigationsinvolving disarticulated and/or incomplete fleshed humanremains. We also examine if stature can be accurately estimatedfrom the measurement of handprints; as prints are commonlyfound at crime scenes, such standards may prove useful in criminalprofiling.

2. Materials and methods

2.1. Materials

This study examines measurements taken from 201 adult subjects (91 male; 110

female) currently residing in Western Australia. Subjects were given an information

sheet and were required to sign a consent form before participating in the study.

Each subject was required to fill in a questionnaire containing basic demographic

(e.g. sex; age; ethnicity) and general (e.g. handedness) questions. The mean age of

the male individuals is 38.2 years (range 19–68) and for the females it is 36.5 years

(range 18–63). The study sample comprises individuals of various ethnic

backgrounds (primarily Caucasian) but overall it is representative of a ‘typical’

Western Australian population (see [25] for specific frequency statistics). Only

individuals without any medical history of hand, foot and backbone problems were

recruited for the study. Ethical approval to undertake this research was approved by

the Human Research Ethics Committee of the University of Western Australia (RA/

4/1/2382).

2.2. Methods

2.2.1. Handprint acquisition

A flatbed scanner (CanoScan LiDE 100) was used to acquire images (400 dpi) of

the hands, which were then converted to handprints. The cover of the scanner was

removed and modified to incorporate a mounting box, photographic scale and

polystyrene plate on the plenum; this allowed subjects to place their hand in a

standardised position and provided a uniform scanning background that is devoid

of ambient light and has a reference scale. To obtain the most accurate

approximation of a handprint the scanned images are edited using the Photoshop1

software package (SC3 edition); this involves image conversion to grey-scale and

adjustments to brightness and contrast. The resultant images are then printed at

1:1 and measured (see below) using a 150 mm digital sliding caliper (Mitutoyo 700-

128 – stated accuracy �0.1 mm); a measuring tape (Paleo-Tech Concepts, Inc.) was

used for any measurements beyond the maximum size limit of the digital caliper.

2.2.2. Measurements

Stature (living height) is measured using a stadiometer (Seca 204); each subject

was asked to stand bare-footed on the flat platform, with heels placed together and

touching the base of the vertical board. The head is positioned in the Frankfort

Horizontal (FH) plane against the vertical board; the subject was then requested to

maintain an erect position with their back in contact with the vertical board and arms

placed on the side of the thigh. The horizontal sliding bar is then positioned on the

contact point of the vertex of the head and stature is recorded in centimeters [26].

Seven anthropometric measurements were taken on each individual hand and

corresponding handprint using a digital caliper and (where required) tape (see

above); measurements are defined in Table 1 and illustrated in Figs. 1 and 2. The

definition of the handprint measurements are the same as those provided for the

anthropometric measurement of the hand, with the exception of handprint breadth

(HPB) and handprint length (HPL); it is not possible to locate the skin or bone

landmarks required to define those measurements. Therefore, the measurement of

HPB and HPL are defined accordingly in Table 1.

2.2.3. Statistical analyses

A precision study was performed prior to primary data collection. Stature, hands

and handprints of the same four subjects were measured on four different

evaluation days, with a minimum of 24-h between re-measurement. Intra-observer

error was determined to be within accepted standards for all measurements

(R > 0.9; rTEM < 5%) [27].

Following the calculation of normal descriptive statistics, bilateral asymmetry in

the measurement data was evaluated using a paired samples t-test. The accuracy of

stature estimation using hand and handprint measurements was then assessed

using sex-specific simple linear regression analyses; measured stature is regressed

onto the individual measurements from both the left and right hand. Accuracy is

quantified using standard errors of the estimate (the standard deviation of the

residuals). A series of stepwise regression analyses were also performed to assess

whether multiple measurements improve prediction accuracy. Statistical analyses

were performed using the SPSS 19.0 software program.

3. Results

3.1. Descriptive statistics and bilateral asymmetry

The mean male (n = 91) measured stature was 178.5 cm (SD7.05; range 162.4–200.5) and for females it was 163.7 cm (SD 7.14;range 149.4–191.3). The mean, standard deviation and correlationto stature for each of the seven hand and handprint measurements(left and right for both sexes) are shown in Table 2. Overall it isreadily apparent that there appears to be very little bilateral

[(Fig._1)TD$FIG]

Fig. 1. Hand measurements: (a) hand breadth (HB); (b) hand length (HL); (c) palm length (PL); (d) 3rd digit length (3D); (e) 1st digit length (1D); (f) 2nd digit length (2D); and

(g) 4th digit length (4D).

N.-I. Ishak et al. / Forensic Science International 216 (2012) 199.e1–199.e7 199.e3

variation in the hand and handprint dimensions. The t-testsrevealed that only the measurement of hand breadth wassignificantly different between the left and right sides (male:t = 4.496, P < 0.05; female t = 8.077, P < 0.05); the difference,however, was very small (male: 0.06 cm; female 0.09 cm) with theright side larger in both instances (Table 2). Handprint breadth wasalso bilaterally significantly different in females (t = 5.403;P < 0.05). All of the measurements are significantly correlated tostature (P < 0.01). It is evident that for the hand measurements, thevariable most strongly correlated to stature is hand length – thesame relationship was expressed for the left and right hand in each[(Fig._2)TD$FIG]

Fig. 2. Handprint measurements: (a) handprint breadth (HPB); (b) handprint length. (H

length (1DP); (f) 2nd digit print length (2DP); and (g) 4th digit print length (4DP).

sex (r = 0.69–0.74). Similarly, the most strongly correlated printvariable is handprint length (r = 0.64–0.65). It is evident that thecorrelation to stature is generally weaker for the hand printvariables (Table 2).

3.2. Simple linear regression

Bilateral regression equations for each individual hand andhandprint measurement are presented separately for males andfemales in Tables 3 and 4 respectively. For the analysis of handmeasurements (Table 3), the standard error of the estimate ranged

PL); (c) palm print length (PPL); (d) 3rd digit print length (3DP); (e) 1st digit print

Table 2Descriptive statistics for hand and handprint measurements (in cm) in males and females.

Measurementa Male Female

Left Right Left Right

Mean SD r Mean SD r Mean SD r Mean SD r

Hand

HB 9.04 0.49 0.51 9.10 0.48 0.54 7.84 0.45 0.50 7.93 0.45 0.45

HL 19.56 0.92 0.74 19.54 0.93 0.73 17.60 0.82 0.70 17.59 0.82 0.69

PL 11.22 0.51 0.67 11.21 0.51 0.66 10.01 0.48 0.77 10.00 0.49 0.67

1D 6.74 0.47 0.47 6.75 0.47 0.47 6.07 0.40 0.57 6.07 0.39 0.53

2D 7.56 0.48 0.69 7.55 0.48 0.67 6.91 0.41 0.64 6.91 0.40 0.62

3D 8.35 0.50 0.67 8.34 0.51 0.66 7.61 0.41 0.62 7.61 0.40 0.61

4D 7.79 0.45 0.66 7.76 0.47 0.69 7.11 0.39 0.63 7.10 0.39 0.62

Handprint

HPB 8.21 0.49 0.57 8.21 0.48 0.57 7.18 0.43 0.47 7.25 0.42 0.50

HPL 17.94 0.95 0.64 17.94 0.94 0.64 16.17 0.83 0.65 16.17 0.83 0.65

PPL 10.02 0.53 0.58 10.00 0.54 0.57 9.00 0.51 0.64 8.99 0.52 0.62

1DP 6.28 0.58 0.52 6.31 0.56 0.50 5.68 0.40 0.46 5.70 0.42 0.45

2DP 7.03 0.50 0.56 7.03 0.48 0.57 6.36 0.40 0.51 6.37 0.40 0.45

3DP 7.92 0.53 0.56 7.93 0.51 0.56 7.17 0.42 0.50 7.19 0.41 0.51

4DP 7.39 0.46 0.57 7.36 0.47 0.59 6.68 0.42 0.50 6.68 0.41 0.49

a Definition of measurements in Table 1 and Figs. 1 and 2.

Table 3Linear regression equations for estimation of stature (in cm) from measurements of hands on the left and right side.

Male Female

Equationa SEE r Equation SEE r

Left

S = 113.211 + 7.223 HB �6.11 0.51 S = 102.335 + 7.820 HB �6.22 0.50

S = 66.728 + 5.714 HL �4.74 0.74 S = 55.934 + 6.123 HL �5.10 0.70

S = 75.165 + 9.209 PL �5.27 0.67 S = 63.711 + 9.988 PL �6.53 0.77

S = 131.838 + 6.919 1D �6.27 0.47 S = 102.105 + 10.136 1D �5.91 0.57

S = 102.152 + 10.101 2D �5.14 0.69 S = 86.060 + 11.234 2D �5.51 0.64

S = 99.422 + 9.472 3D �5.26 0.67 S = 82.703 + 10.635 3D �5.65 0.62

S = 96.611 + 10.512 4D �5.30 0.66 S = 81.751 + 11.522 4D �5.55 0.63

Right

S = 106.334 + 7.929 HB �5.96 0.54 S = 106.176 + 7.249 HB �6.40 0.45

S = 69.723 + 5.567 HL �4.83 0.73 S = 57.135 + 6.057 HL �5.17 0.69

S = 75.976 + 9.149 PL �5.32 0.66 S = 66.573 + 9.713 PL �5.34 0.67

S = 130.931 + 7.045 1D �6.26 0.47 S = 104.845 + 9.685 1D �6.08 0.53

S = 103.644 + 8.725 2D �5.24 0.67 S = 87.499 + 11.017 2D �5.61 0.62

S = 102.964 + 9.056 3D �5.34 0.66 S = 80.111 + 10.976 3D �5.68 0.61

S = 99.223 + 10.222 4D �5.16 0.69 S = 82.505 + 11.434 4D �5.63 0.62

a Definition of measurements in Table 1 and Fig. 1; S: stature.

N.-I. Ishak et al. / Forensic Science International 216 (2012) 199.e1–199.e7199.e4

between �4.74 cm (left hand length) and �6.27 cm (left thumblength) in the male sample; in the female sample the range was�5.10 cm (left hand length) to �6.40 cm (right hand breadth). Forthe analysis of handprint measurements (Table 4), the standard error

Table 4Linear regression equations for estimation of stature (in cm) from measurements of ha

Male

Equationa SEE r

Left

S = 111.379 + 8.179 HPB �5.81 0.57

S = 93.836 + 4.718 HPL �5.48 0.64

S = 102.476 + 7.584 PPL �5.80 0.58

S = 138.869 + 6.310 1DP �6.05 0.52

S = 122.120 + 8.018 2DP �5.86 0.56

S = 118.502 + 7.575 3DP �5.85 0.56

S = 113.868 + 8.745 4DP �5.82 0.57

Right

S = 109.775 + 8.375 HPB �5.83 0.57

S = 91.858 + 4.831 HPL �5.42 0.64

S = 104.265 + 7.424 PPL �5.82 0.57

S = 138.481 + 6.345 1DP �6.13 0.50

S = 119.814 + 8.353 2DP �5.82 0.57

S = 116.747 + 7.783 3DP �5.87 0.56

S = 113.129 + 8.877 4DP �5.75 0.59

a Definition of measurements in Table 1 and Fig. 2; S: stature.

of the estimate ranged between �5.42 cm (right handprint length)and �6.13 cm (right thumbprint length) in the male sample; in thefemales the range was �5.46 cm (left handprint length) to �6.42(right thumbprint length).

ndprints on the left and right side.

Female

Equation SEE r

S = 107.029 + 7.890 HPB �6.33 0.47

S = 73.282 + 5.589 HPL �5.46 0.65

S = 83.115 + 8.950 PPL �5.51 0.64

S = 116.846 + 8.245 1DP �6.37 0.46

S = 105.284 + 9.186 2DP �6.17 0.51

S = 103.177 + 8.434 3DP �6.22 0.50

S = 106.514 + 8.556 4DP �6.22 0.50

S = 102.598 + 8.430 HPB �6.22 0.50

S = 73.648 + 5.568 HPL �5.47 0.65

S = 86.405 + 8.599 PPL �5.60 0.62

S = 120.085 + 7.653 1DP �6.42 0.45

S = 112.854 + 7.978 2DP �6.40 0.45

S = 100.262 + 8.821 3DP �6.17 0.51

S = 106.170 + 8.610 4DP �6.25 0.49

Ta

ble

5M

ult

iple

reg

ress

ion

eq

ua

tio

ns

for

est

ima

tio

no

fst

atu

re(i

ncm

)fr

om

me

asu

rem

en

tso

fh

an

ds

on

the

left

an

dri

gh

tsi

de

.

Ma

leFe

ma

le

Eq

ua

tio

na

SE

Er

Eq

ua

tio

nS

EE

r

Left

S=

65

.74

4+

0.4

59

HB

+5

.55

3H

L�

4.7

60

.74

S=

52

.75

6+

0.9

99

HB

+5

.36

0H

L+

0.6

73

PL�

8.4

18

3D

+1

.68

71

D+

4.5

12

2D

+3

.47

44

D�

5.0

40

.73

S=

69

.35

9�

0.1

67

HB

+4

.68

4H

L�

1.8

14

1D

+1

.35

22

D+

2.7

04

4D

�4

.77

0.7

5S

=5

4.1

38

+1

.14

7H

B+

5.7

14

HL

�5

.10

0.7

1

S=

67

.15

7+

0.0

31

HB

+1

6.7

67

HL�

11

.75

4P

L�

14

.93

13

D�

1.6

23

1D

+1

.71

22

D+

4.8

35

4D

�4

.77

0.7

6S

=5

5.2

88

+1

.11

4H

B+

4.3

30

HL

+1

.39

91

D+

2.6

38

2D�

0.4

60

4D

�5

.12

0.7

1

Rig

ht

S=

69

.72

3+

5.5

67

HL

+3

.93

94

D�

4.7

40

.75

S=

56

.88

4+

6.8

65

PL

+5

.01

23

D�

5.1

70

.70

S=

64

.23

7+

0.9

75

HB

+6

.48

3H

L�

1.2

47

PL�

6.4

70

3D�

1.5

49

1D

+0

.88

52

D+

6.5

03

4D

�4

.75

0.7

6S

=5

5.4

89

+6

.66

4P

L+

4.7

62

3D

+0

.66

9H

B�

5.1

80

.70

S=

67

.15

6+

0.9

39

HB

+4

.46

2H

L+

1.5

89

1D�

1.5

56

2D

+4

.91

24

D�

4.7

70

.75

S=

54

.13

8+

1.1

47

HB

+5

.71

4H

L�

5.1

90

.69

aD

efi

nit

ion

of

me

asu

rem

en

tsin

Ta

ble

1a

nd

Fig

.1

;S:

sta

ture

.

N.-I. Ishak et al. / Forensic Science International 216 (2012) 199.e1–199.e7 199.e5

3.3. Multiple regression

To assess if stature prediction accuracy is improved by utilisingmultiple variables, a series of sex-specific bilateral multipleregression equations were formulated; these are presentedseparately for hand and handprint measurements in Tables 5and 6 respectively. In the interest of brevity, only the three mostaccurate models are presented for each sub-sample. With regard tothe hand measurements, it is evident that in males the smalleststandard error of the estimate is �4.74 cm and in females it is�5.04 cm (Table 5); for the handprint measurements the smalleststandard error of the estimate for males is�4.74 cm and in females itis�5.39 cm (Table 6). The variables generally weighted most stronglyin the multiple regression models (both hand and handprint) in bothsexes are hand length, palm length and in some instances handbreadth (see Tables 5 and 6).

4. Discussion

Stature estimation is an important part in any forensicinvestigation involving the anthropological profiling of unknownhuman remains. The primary aim of this study was to assess thereliability and accuracy of using hand and handprint measure-ments to estimate stature in a Western Australian population.With regard to establishing the reliability of acquiring stature andhand and handprint measurements, we have demonstrated thatthe degree of measurement error and reliability are well withinacceptable standards (R > 0.9; rTEM < 5%) [28–30]. The statisticalquantification of error and uncertainty in the forensic sciences isvital; this applies to not only the degree of error associated withforensic standards (see below), but also the accuracy and precisionof the raw data (measurements) from which they are formulated;the latter is often neglected in forensic anthropological research ofthis nature.

In evaluating bilateral asymmetry in hand and handprintmeasurements, hand breadth was the only variable that wassignificantly different; the relationship was apparent in both malesand females. Where significant bilateral differences have beendescribed, it is generally for the measurement of hand breadth (e.g.[18,24]); other published studies found only minor (non-signifi-cant) bilateral variation (e.g. [19,21,23,31]) which accords with theremainder of the data obtained from the Western Australianpopulation. Hand and handprint length was found to have thestrongest correlation to stature; the former is again a relationshippreviously described in related literature (e.g. [18,19,22,24]).

In assessing the accuracy of our simple linear regressionmodels, the standard error of the estimate (SEE) was lowest usingthe measurement of hand length (SEE �4.74–5.17 cm). Expectedly,other research has demonstrated that this is the most accuratemeasurement for estimating stature, albeit their regression modelshave a higher stated accuracy: e.g. Sanli et al. [20] (�3.50–4.59 cm);Rastogi et al. [19] (�3.65–5.04 cm); Krishan and Sharma [18] (�3.78–5.22 cm); Agnihotri et al. [23] (�4.23–4.96 cm); Habib and Kamal [22](�4.54–5.48 cm). This improved accuracy may possibly be attributedto the lack of diversity in their sample populations, specifically interms of genetic variability. The most accurate handprint regressionmodel was, as expected, handprint length (SEE �5.46 cm); theaccuracy is slightly lower compared to taking a direct physicalmeasurement of the hand (SEE �4.74 cm). Ahemad and Purkait [24]similarly demonstrated that handprint length had the lowest SEE(�4.64 cm) of all their hand impression regression models.

With regard to our multiple regression models for handmeasurements, we demonstrated an equivalent degree of predic-tion accuracy (�4.74 cm) compared to the simple linear models (seeTables 5 and 6). A selection of the aforementioned studies, however,demonstrated increases in prediction accuracy when using multiple

Ta

ble

6M

ult

iple

reg

ress

ion

eq

ua

tio

ns

for

est

ima

tio

no

fst

atu

re(i

ncm

)fr

om

me

asu

rem

en

tso

fh

an

dp

rin

tso

nth

ele

fta

nd

rig

ht

sid

e.

Ma

leFe

ma

le

Eq

ua

tio

na

SE

Er

Eq

ua

tio

nS

EE

r

Left

S=

83

.90

1+

3.3

76

HP

L+

4.1

46

HP

B�

5.2

60

.68

S=

73

.46

7+

7.3

78

PP

L+

3.3

18

3D

P�

5.4

20

.66

S=

82

.79

7+

3.8

31

PP

L+

2.8

81

3D

P+

4.2

02

HP

B�

5.2

90

.68

S=

73

.37

2+

7.3

49

PP

L+

3.2

82

3D

P+

0.0

85

HP

B�

5.4

40

.66

S=

85

.00

5+

4.1

28

HP

B+

3.1

05

HP

L+

2.0

94

1D

P�

1.4

38

2D

P+

0.1

16

4D

P�

5.2

90

.69

S=

71

.17

2�

0.1

01

HP

B+

6.8

25

PP

L�

0.9

22

3D

P+

2.0

99

1D

P+

2.2

55

2D

P+

1.8

48

4D

P�

5.4

40

.67

Rig

ht

S=

83

.07

3+

3.5

59

HP

L+

3.8

51

HP

B�

4.7

40

.67

S=

69

.28

8+

1.3

61

HP

B+

6.3

96

HP

L+

2.8

94

1D

P�

6.5

32

2D

P+

0.9

33

4D

P�

5.3

90

.68

S=

81

.42

2+

3.0

96

HP

B+

25

.16

7H

PL�

20

.58

9P

PL�

26

.47

33

DP

+1

.26

21

DP�

0.2

67

2D

P+

4.0

93

4D

P�

5.2

00

.71

S=

69

.37

0+

1.3

53

HP

B+

5.1

28

HP

L+

1.4

42

PP

L+

0.2

49

3D

P+

2.8

61

1D

P�

6.1

32

2D

P+

1.4

39

4D

P�

5.4

40

.68

S=

83

.07

3+

3.8

51

HP

B+

3.5

59

HP

L�

5.2

70

.67

S=

74

.22

5+

6.8

11

PP

L+

3.9

29

3D

P�

5.4

70

.65

aD

efi

nit

ion

of

me

asu

rem

en

tsin

Ta

ble

1a

nd

Fig

.2

;S:

sta

ture

.

N.-I. Ishak et al. / Forensic Science International 216 (2012) 199.e1–199.e7199.e6

variables (e.g. [18,21,22]). Rastogi et al. suggest that when predictingstature ‘‘. . .multiple regression equations give more accurate results’’[19: p. 188] but the statement was supported using correlationcoefficients; their data actually indicates that their simple linearmodels are slightly more accurate (smallest SEE for a single variable:�3.65 cm; for multiple variables: �3.67 cm – see Tables 6 and 7 inRastogi et al. [19]). It is apparent that in the present study, multipleregression equations are only likely to be beneficial in circumstanceswhere the direct linear measurement of hand length is not possible,but other parts of the hand are available in combination (e.g. palmlength, breadth and other finger lengths). In other words, our dataindicates that when the parts of the hand that individually have aweaker correlation to living stature are used in combination,prediction accuracy is increased.

Conversely, the accuracy of the multiple regression modelsusing handprint measurements showed an increase in predictionaccuracy compared to using single variables. The increase inaccuracy was greater in the male sample (6.8 mm) but very minorfor the females (less than 1 mm); the lowest SEE was �4.74 cm and�5.39 cm for males and females respectively (see Tables 4 and 6).Overall, however, it is apparent that the accuracy of the handprintmodels (both single and multiple) is generally lower compared to thedirect measurement of the hand. The measurement and analysis ofhandprints, however, offer a novel source of profiling data that isstatistically quantified. For example, measurements taken from a fullor partial handprint left at a crime scene could be useful towardsoffering an estimated stature of a suspect; this could supplement anyavailable eyewitness information (e.g. ethnicity; sex; age). It is alsoworth considering that eyewitness estimations of a suspect’s heightare very likely to be less accurate compared to the novel handprintstandards outlined in the present study [32,33].

The hand and handprint standards outlined here offer apractical means of estimating stature with an overall accuracycomparable to established skeletal standards for the hand: e.g.[34]: SEE �4.70–8.14 cm; [35]: standard error of the mean, 5.10–5.67 cm. The clear value of anthropometric measurements of thehand for estimating stature has thus been confirmed in a WesternAustralian population. Furthermore, we have also established that themeasurement of handprints afford a practical means of estimatingstature. It is important to note that in the present paper we have notconsidered sexual dimorphism in the data; this is the topic of arecently completed research project [36] and a manuscript ispresently in preparation that highlights the forensic utility of handand handprint data for the estimation of sex.

5. Conclusions

The present study is part of a larger federally funded ongoingresearch project that is developing population specific standardsfor Australia in the form of an interactive Human IdentificationPackage (HIP). In the current study we have outlined a series ofnew forensic standards for the estimation of stature in a WesternAustralian population; standard error rates range between �4.74and 6.53 cm. The accuracy of anthropometric hand measurements forestimating stature is well established in a variety of populations; theuse of handprint data, however, is still relatively novel and theapplicability of that approach for use in other populations and inforensic case-work should be evaluated by the wider anthropologicaland/or forensic community.

Acknowledgements

The authors would like to thank Senior Constable GrahamByard, Forensic Division (Fingerprint Bureau), Western AustraliaPolice, for his invaluable advice. We also thank Sergeant Brad Nind,Forensic Training & Development Unit, Western Australia Police,

N.-I. Ishak et al. / Forensic Science International 216 (2012) 199.e1–199.e7 199.e7

for his assistance. Lastly, we would like to thank all the volunteerswho participated in this project. Funding was provided by an ARCDiscovery Grant (DP1092538) held by DF.

References

[1] D.A. Komar, J.E. Buikstra, Forensic Anthropology: Contemporary Theory andPractice, Oxford University Press, New York, 2008.

[2] E.H. Kimmerle, A. Ross, D. Slice, Sexual dimorphism in America: geometricmorphometric analysis of the craniofacial region, J. Forensic Sci. 53 (2008) 54–57.

[3] D. Franklin, Forensic age estimation in human skeletal remains: current conceptsand future directions, Leg. Med. 12 (2010) 1–7.

[4] A.H. Ross, M.J. Manneschi, New identification criteria for the Chilean population:estimation of sex and stature, Forensic Sci Int. 204 (2011) 206.e1–206.e3.

[5] H.V.F. Cardoso, J.E.A. Gomes, Trends in adult stature of peoples who inhabited themodern Portuguese territory from the Mesolithic to the late 20th century, Int. J.Osteoarch. 19 (2009) 711–725.

[6] G. Fully, Une Nouvelle Methode de determination de la Taille, Ann. Med. Legale 35(1956) 266–273.

[7] J.K. Lundy, The mathematical versus anatomical methods of stature estimationfrom long bones, Am. J. Forensic Med. Pathol. 6 (1985) 73–76.

[8] M.H. Raxter, B.M. Auerbach, C.B. Ruff, Revision of the fully technique for estimat-ing statures, Am. J. Phys. Anthropol. 130 (2006) 374–384.

[9] M. Trotter, Estimation of stature from intact long limb bones, in: T.D. Stewart (Ed.),Personal Identification in Mass Disasters, National Museum of Natural History,Washington, 1970, pp. 71–83.

[10] R.J. Wilson, N.P. Herrmann, L.M. Jantz, Evaluation of stature estimation from thedatabase for forensic anthropology, J. Forensic Sci. 55 (2010) 684–689.

[11] K. Krishan, Estimation of stature from footprint and foot outline dimensions inGujjars of North India, Forensic Sci. Int. 175 (2008) 93–101.

[12] T. Kanchan, R.G. Menezes, R. Moudgil, R. Ramneet, M.S. Kotian, R.K. Garg, Statureestimation from foot dimensions, Forensic Sci. Int. 179 (2008) 241.e1–241.e5.

[13] G. Zeybek, I. Ergur, Z. Demiroglu, Stature and gender estimation using footmeasurements, Forensic Sci. Int. 181 (2008) 54.e1–54.e5.

[14] K. Krishan, Estimation of stature from cephalo-facial anthropometry in northIndian population, Forensic Sci. Int. 181 (2008) 52.e1–52.e6.

[15] A.K. Agnihotri, S. Kachhwaha, K. Googoolye, A. Allock, Estimation of stature fromcephalo-facial dimensions by regression analysis in Indo-Mauritian population, J.Forensic Leg. Med. 18 (2011) 67–172.

[16] D. Sahni, Sanjeev, P. Sharma, Harjeet, G. Kaur, A. Aggarwal, Estimation of staturefrom facial measurements in northwest Indians, Leg. Med. 12 (2010) 23–27.

[17] J.C. Reed, B.F.B. Algee-Hewitt, Comments on Estimation of stature from cranialsutures in a South Indian male population, Int. J. Legal Med. 125 (2011) 469–471.

[18] K. Krishan, A. Sharma, Estimation of stature from dimensions of hands and feet ina North Indian population, J. Forensic Leg. Med. 14 (2007) 327–332.

[19] P. Rastogi, K.R. Nagesh, K. Yoganarasimha, Estimation of stature from handdimensions of North and South Indians, Leg. Med. 10 (2008) 185–189.

[20] G. Sanli, E.D. Kizilkanat, N. Boyan, E.T. Ozsahin, M.G. Bozkir, R. Soames, H. Erol, O.Oguz, Stature estimation based on hand length and foot length, Clin. Anat. 18(2005) 589–596.

[21] A.K. Abdel-Malek, A.M. Ahmed, S.A.E.A.E. Sharkwai, N.A.E.M.A., E. Hamid, Predic-tion of stature from hand measurements, Forensic Sci. Int. 46 (1990) 181–187.

[22] S.R. Habib, N.N. Kamal, Stature estimation from hand and phalanges lengths ofEgyptians, J. Forensic Leg. Med. 17 (2010) 156–160.

[23] A.K. Agnihotri, S. Agnihotri, N. Jeebun, K. Googoolye, Prediction of stature usinghand dimensions, J. Forensic Leg. Med. 15 (2008) 479–482.

[24] N. Ahemad, R. Purkait, Estimation of stature from hand impression: a noncon-ventional approach, J. Forensic Sci. 56 (2011) 706–709.

[25] Australian Bureau of Statistics, Census Data [Internet]. 2006 [cited 2011 May 1];Available from: http://www.abs.gov.au/websitedbs/d3310114.nsf/home/census+data.

[26] C. Gordon, W. Chumlea, A. Roche, Stature, recumbent length, and weight, in: T.Lohman, A. Roche, R. Martorell (Eds.), Anthropometric Standardisation ReferenceManual, Human Kinetics Books, Champaign, IL, 1991, pp. 3–8.

[27] S. Ulijaszek, D.A. Kerr, Anthropometric measurement error and the assessment ofnutritional status, Br. J. Nutr. 82 (1999) 165–177.

[28] T.R. Knapp, Technical error of measurement a methodological critique, Am. J.Phys. Anthropol. 87 (1992) 235–236.

[29] T.A. Perini, G.L. Oliveira, J.D.S. Ornellas, F.P. Olivera, Technical error of measure-ment in anthropometry, Rev. Bras. Med. Esporte 11 (2005) 81–85.

[30] M. Reynolds, D. Franklin, M.A. Raymond, I. Dadour, Bloodstain measurementusing computer fitted theoretical ellipses a study in accuracy and precision, J.Forensic Ident. 58 (2008) 469–484.

[31] L.M. Schell, D.R. Johnston, D.R. Smith, A.M. Paolone, Directional asymmetry ofbody dimensions among white adolescents, Am. J. Phys. Anthropol. 67 (1985)317–322.

[32] B.L. Cutler, S.D. Penrod, T.K. Martens, The reliability of eyewitness identification:the role of system and estimator variables, Law Hum. Behav. 11 (1987) 233–258.

[33] C.A. Meissner, S.L. Sporer, J.W. Schooler, Person descriptions as eyewitnessevidence, in: R.C.L. Lindsay, D.F. Ross, J.D. Read, M.P. Toglia (Eds.), The Handbookof Eyewitness Psychology: vol. II – Memory for People, Lawrence ErlbaumAssociates, Mahwah, NJ, 2007, pp. 3–34.

[34] J.H. Musgrave, N.K. Harneja, The estimation of adult stature from metacarpal bonelength, Am. J. Phys. Anthropol. 48 (1978) 113–120.

[35] L. Meadows, R.L. Jantz, Estimation of stature from metacarpal lengths, J. ForensicSci. 37 (1992) 147–154.

[36] N.I. Ishak, Sex and stature estimation using hand and handprint measurements ina Western Australian population, Unpublished MFORSci, Centre for ForensicScience, University of Western Australia, Crawley, 2010.

[37] T. Kanchan, P. Rastogi, Sex determination from hand dimensions of North andSouth Indians, J. Forensic Sci. 54 (2009) 546–550.

[38] L.M. Robbins, Estimating height and weight from size of footprints, J. Forensic Sci.31 (1986) 143–152.