Professional Documents
Culture Documents
2
doi: 10.1111/1556-4029.12959
TECHNICAL NOTE Available online at: onlinelibrary.wiley.com
ANTHROPOLOGY
ABSTRACT: In cases of partial or poor print recovery and lack of database/suspect print, fingerprint evidence is generally neglected. In light
of such constraints, this study was designed to examine whether ridge density can aid in narrowing down the investigation for sex identifica-
tion. The study was conducted on the right-hand index digit of 245 males and 246 females belonging to the Punjabis of Delhi region. Five
ridge density count areas, namely upper radial, radial, ulnar, upper ulnar, and proximal, were selected and designated. Probability of sex origin
was calculated, and stepwise discriminant function analysis was performed to determine the discriminating ability of the selected areas. Females
were observed with a significantly higher ridge density than males in all the five areas. Discriminant function analysis and logistic regression
exhibited 96.8% and 97.4% accuracy, respectively, in sex identification. Hence, fingerprint ridge density is a potential tool for sex identifica-
tion, even from partial prints.
KEYWORDS: forensic science, forensic anthropology, fingerprints, dactyloscopy, ridge density, sex identification
Ever since the first usage in the third century BC, fingerprints even in cases of partial print recovery and lack of database/sus-
have been a promising branch of forensic science and anthropol- pect prints. This study investigates fingerprint ridge density in
ogy. Fingerprint dermatoglyphic traits have been extensively the Punjabi (Khatri & Arora) population of the Delhi region.
used in anthropology, genetics, and evolutionary studies (1). The study explains the relation between sex and ridge density of
However, some features such as fingerprint ridge density still five selected count areas of the dactylogram. Furthermore, the
carry untapped potential. Recovery of partial prints and the lack present research communication provides a general equation for
of suspect print or database are the two major limitations of fin- sex identification from ridge density.
gerprint investigations. In such cases, sex identification can be
the most crucial link to narrow down the investigation. Further-
Methods
more, sexual dimorphism in fingerprints has been of key interest
to the researchers. In a study among the Chuvashian population Prior ethical clearance from the Department Ethical Commit-
of Russia, males were reported with higher total ridge count tee was obtained to conduct the research. Informed written con-
(TRC) and absolute finger ridge counts (AFRC) than females, sent from the subjects participating in the study was obtained
for both the hands (2). Similarly, the quantitative value of der- prior to the actual commencement of the study. The study was
matoglyphic patterns in the Spanish population is lower in males conducted among unrelated and randomly selected 491 subjects,
than in females (3). Variations have also been reported in the comprising of 245 males and 246 females of the Punjabi Khatri
minutiae frequencies of males and females (4). and Arora population groups in the northern Indian territory of
Only few studies have focused on fingerprint ridge density so Delhi region. The age-group of the selected subjects varied
far. These studies presented probability findings of the ridge between 18 and 65 years. Exclusion criteria were defined as
density values and significant difference in the ridge density of subjects belonging to the non-Punjabi (Khatri and Arora) ethnic-
males and females among various population groups, namely ity, subjects with disease or injury in the dermis, as evident
Spanish (5), Argentinian (6), North Indian (7), South Indian (8), from the fingertips. Subjects were encouraged to clean and dry
Chinese (9), Caucasian and African Americans (10). their hands and digits properly before depositing the prints. The
However, no study has provided the practical utilization of right index digit prints were collected, using the standard
ridge density for sex identification. Sex identification from ridge method of inking and rolling the digits (11). Complete pattern
density would give crucial information in forensic investigations area of the distal phalanx of the index digit was recorded.
Dactyloscopic examination was conducted using the 69 magni-
fication hand lens. Five ridge density count areas, namely upper
1 radial, radial, ulnar, upper ulnar, and proximal, were selected
Department of Anthropology, University of Delhi, Delhi 110007, India.
*Funded by the University Grants Commission by providing the research and designated. Each area was a square of 25 mm2. Radial/core
fellow grant to the main author. radial, ulnar/core ulnar, and proximal areas were selected fol-
Received 1 Sept. 2014; and in revised form 9 Mar. 2015; accepted 28 lowing the standardized procedure (5). Upper radial and upper
Mar. 2015. ulnar areas (10) were selected such that they touch the extreme
follows: 70
60
Frequency
PðAÞ ¼ PðBÞ; then PðAÞ 50
X ð1Þ 40
¼ ðA \ Bn Þ; Bn : n ¼ 1; 2; 3. . .A Finite Number Males
30
Females
20
Stepwise discriminant function analysis was performed to 10
determine the discriminating ability of the ridge density areas.
0
Wilks’ lambda method was used for the stepwise analysis. 10 11 12 13 14 15 16 17 18 19 20
Wilks’ lambda method provides the F statistic as discriminant
Ridge Density (Number of ridges/ 25mm2)
criteria, with higher values giving better discrimination (Wilks’
lambda= unexplained variance/total variance and minimum F
value was set at F = 3.84 for variables to enter the equation FIG. 2––Sex-wise frequency distribution of upper radial ridge density
and a maximum at F = 2.71 for variables to be removed, repre- among the study group.
senting significance levels of 0.05 and 0.1, respectively). Logis-
tic regression analysis was performed to attain a predicting
equation for sex estimation from the ridge density. The model 120
was evaluated at 0.5 limit value which explained that the values 100
<0.5 were evaluated as males, and those more than 0.5, as
females. Both discriminant function analysis and the regression 80
Frequency
20
0
11 12 13 14 15 16 17 18 19 20 21 22
Ridge density (Number of ridges/25mm2)
120
100
80
Frequency
60
Males
40 Females
20
0
10 11 12 13 14 15 16 17 18 19 20 21
Ridge density (Number of ridges/ 25 mm2)
FIG. 1––Illustration of ridge density count areas in the right index finger- FIG. 4––Sex-wise frequency distribution of ulnar ridge density among the
print of subject. study group.
426 JOURNAL OF FORENSIC SCIENCES
50
Probability
40 Males
Females
Upper Radial Ridge Density Males Females
30
20 ≤11 1 0
≤12 0.961 0.039
10 ≤13 0.921 0.077
0 ≤14 0.883 0.117
9 10 11 12 13 14 15 16 17 18 19 ≤15 0.748 0.252
Ridge density (Number of ridges/ 25 mm2)
>15 0.072 0.928
>16 0.022 0.979
>17 0 1
FIG. 5––Sex-wise frequency distribution of upper ulnar ridge density
>18 0 1
among the study group.
>19 0 1
60
50 ≤17 0.821 0.180
Males >17 0.058 0.942
40 >18 0.009 0.991
Females
30 >19 0 1
20 >20 0 1
>21 0 1
10
0 Probability
8 9 10 11 12 13 14 15 16
Ridge density (Number of ridges/25mm2) Ulnar Ridge Density Males Females
≤13 0.950 0.050
FIG. 6––Sex-wise frequency distribution of proximal ridge density among
≤14 0.941 0.059
the study group.
≤15 0.877 0.123
≤16 0.769 0.231
>16 0.078 0.922
>17 0 1
TABLE 1––Descriptive statistics of the ridge density values of the various >18 0 1
count areas among the males and females of the selected population group. >19 0 1
>20 0 1
Mean $ SD
Ridge Density Independent t-Test Probability Obtained in Sample
Count Area Males Females t-Value
Upper radial 13.58 $ 1.39 16.14 $ 1.43 % 20.19* Upper Ulnar Ridge Density Males Females
Radial 15.96 $ 1.09 18.52 $ 1.32 % 23.45* ≤11 1 0
Ulnar 14.90 $ 1.81 17.13 $ 1.26 % 15.87* ≤12 1 0
Upper Ulnar 12.69 $ 1.59 16.22 $ 1.20 % 27.76* ≤13 0.982 0.018
Proximal 10.70 $ 1.10 12.88 $ 1.17 % 21.37* ≤14 0.935 0.065
*p is significant at <0.001 level. >14 0.118 0.882
>15 0.027 0.973
>16 0.020 0.980
characteristics of ridge density among males and females. The >17 0 1
>18 0 1
test results showed a significant difference (p < 0.001) with
respect to ridge density between the males and the females in all Probability
the five areas. The mean difference between the males and the
females was maximum in the upper ulnar area (3.53) and mini- Proximal Ridge Density Males Females
mum in the proximal area (2.18). The minimum ridge density
≤9 1 0
was encountered in the proximal area and maximum in the radial ≤10 0.990 0.009
area, in both males and females. ≤11 0.864 0.136
Table 2 shows the probability of sex origin from the ridge ≤12 0.714 0.286
density of various count areas. It was evident that the probability >12 0.062 0.938
>13 0.014 0.986
values were highly indicative of the sex origin. For the upper >14 0 1
radial area, a ridge density of ≤14 was most likely of male origin >15 0 1
and a ridge density of >15 was most likely of female origin.
DHALL AND KAPOOR . FINGERPRINT RIDGE DENSITY 427
Likewise, in radial, ulnar, upper ulnar, and proximal areas, a ation. The discriminating ability of the four ridge density areas
ridge density of ≤16, ≤15, ≤14, and ≤11, respectively, was most was in the order: upper ulnar> radial> upper radial> proximal.
likely of the male origin, whereas >17, >16,>14, and >12 in the Further the discriminant function score was determined using the
radial, ulnar, upper ulnar, and proximal areas, respectively, was constant and the unstandardized canonical discriminant function
most likely of the female origin (Table 2). coefficients (Eq. 2). The sectioning point (% 0.0035) was calcu-
Stepwise discriminant function analysis (Table 3) demon- lated using the group centroids. The subject was classified as a
strated that the ulnar area did not contribute in the sex differenti- female if the score was more than the sectioning point, while a
TABLE 3––Stepwise discriminant function analysis among the five ridge density areas.
Males
60 Mean = -1.8
Std.Dev. = 0.87
N = 245
50
40
30
20
10
0
-5.0 -2.5 0.0 2.5 5.0
Females
Frequency
Mean = 1.8
50 Std.Dev. = 1.114
N = 246
40
30
20
10
0
-5.0 -2.5 0.0 2.5 5.0
score lower than the sectioning point classified the individual lyzed, while the present findings demonstrate consideration of a
as a male. The analysis presented 96.8% accuracy in sex single digit at a time. The approach appears to be more accurate
identification. Separate group plots of the discriminant scores of for forensic investigations, where all the ten digits may not be
males and females exhibited only a marginal overlapping area available. Selection, standardization, and designation of the areas
(Fig. 7). were important dimensions of the present study. Most of the pre-
vious studies have focused on the probability of observing a
Di ¼ % 19:495 þ 0:234 ' upperradial þ 0:378 ' radial ridge density value, when the sex of the donor was known. In
ð2Þ the present study, the probability of sex origin has been deter-
þ 0:364 ' upperulnar þ 0:343 ' proximal
mined when the ridge density value is known. Furthermore, the
present communication attempts to provide a general equation of
Table 4 shows the association of sex with the ridge density prediction and the discriminating ability of selected standardized
count areas using logistic regression. The Neagelkerke’s R2 areas. It is evident that the approach in the present research is
showed that the model fits well for all the areas. The logit equa- more applicable as an investigation tool.
tions were computed for each area as well as for all the areas The frequency distribution highlights lower ridge density in
together. When the areas were analyzed together, it was found the proximal area and, thus, thicker ridges in the area. It is note-
that upper radial, radial, upper ulnar, and proximal areas were worthy that the findings on the Indian population in the present
significant factors in the prediction of sex. However, the ulnar study and findings from a previous study on the Thai popula-
area showed a nonsignificant odds ratio. The upper ulnar area tions (12) show higher range of ridge density values than the
exhibited maximum percentage accuracy in the sex estimation. African Americans and the Caucasians of America (10). The
When sex was identified using all the five areas together, the present findings show that the core values were highest for both
regression model showed a 97.4% accuracy. the sexes and the proximal area exhibited the lowest ridge den-
sity. This is in concordance with the disto-proximal gradient
Discussion extending from the fingertips to the proximal region of the palm
(13). Similar findings have been reported among the Spanish
The present study draws various novel inferences on the population (5), where core values were higher than the proximal
basis of its findings. In previous studies on the ridge density count area values. Findings of previous studies on the frequency
examination, mean ridge density of all the digits has been ana- distribution and the mean of ridge density values among various
Percentage Accuracy
Ridge Density
2
Count Area B Exp (B) Neagelkerke’s R Logit Regression Equation Males Females Total
Upper radial 1.437 4.209* 0.615 % 21.364 + 1.437*Upper radial 73.9 90.2 82.1
Radial 2.211 9.124* 0.736 % 37.871 + 2.211*Radial 95.1 79.3 87.2
Ulnar 1.678 5.356* 0.613 % 26.873 + 1.678*Ulnar 93.9 72.0 82.9
Upper ulnar 1.998 7.374* 0.799 % 29.262 + 1.998*Upper ulnar 87.3 93.9 90.6
Proximal 1.769 5.863* 0.638 % 20.790 + 1.769*Proximal 77.6 87.8 82.7
All areas together B Exp (B) 0.941 % 83.501 + 0.767*Upper radial+1.698*Radial+0.707 97.6 97.2 97.4
Upper radial 0.767 2.153* *Ulnar+1.669*Upper ulnar+0.667*Proximal
Radial 1.698 5.461^
Ulnar## 0.707 2.028
Upper ulnar 1.669 5.309*
Proximal 0.667 1.949^
*p is significant at <0.001 level.
^
p is significant at <0.05–0.01 level.
##
Ulnar region shows nonsignificant odds.
TABLE 5––Frequency distribution and mean of ridge density values among various population groups from previous studies.
Population Males Females Males Females Males Females Males Females References
Caucasian 7.9–14.70 10.60–16.80 10.90 12.61 10
African American 8.2–14.30 9.7–16 11.14 13.32 10
South Indian 11–15 12–16 12.80 14.60 8
South Indian 10–17 10–17 12.57 14.14 11
Puna–Quebrada 13–22 (radial), 13–21 (ulnar), 14–22 (radial), 13–21 (ulnar), 16.67 18.47 16.39 17.62 14.33 16.13 6
11–18 (proximal) 12–20 (proximal)
Ramal 13–21 (radial), 12–20 (ulnar), 15–23 (radial), 14–23 (ulnar), 17.04 19.08 16.10 17.75 14.08 15.12 6
10–17 (proximal) 11–20 (proximal)
Uttarakhand 11–16 11–16 11.90 14.10 14
population
Thais 10–26 (radial), 9–26 (ulnar) 10–26 (radial), 9–26 (ulnar) 15.89 16.19 15.84 16 12
DHALL AND KAPOOR . FINGERPRINT RIDGE DENSITY 429