Beruflich Dokumente
Kultur Dokumente
4
doi: 10.1111/1556-4029.12133
TECHNICAL NOTE Available online at: onlinelibrary.wiley.com
ANTHROPOLOGY
Richard M. Thomas,1 Ph.D.; Douglas H. Ubelaker,2 Ph.D.; and John E. Byrd,3 Ph.D.
ABSTRACT: A common task in forensic anthropology involves pair-matching of left and right skeletal elements. This can be achieved
through visual pair-matching by evaluating similarities in morphology, and through osteometric sorting, a quantitative technique. To simplify
the process of osteometric sorting, this article explains the use of a statistic (M), which captures the amount of size variation found between
homologous bones from single individuals. A database of skeletal measurements for all major paired postcranial bones is used to calculate val-
ues of M from a variety of sources. The maximum value and the 90th and 95th percentiles of M are provided in tabular format, and values of
M from forensic cases can be compared to these tables as an objective means for determining whether homologous bones could have originated
from the same individual. This simple technique can be combined with visual pair-matching to be particularly effective in cases involving com-
mingling of skeletons.
KEYWORDS: forensic science, forensic anthropology, pair-matching, osteometric sorting, commingling, bilateral asymmetry
A common task in forensic anthropology involves pair-match- remains. Another study developed a simple formula for identify-
ing of left and right skeletal elements, especially in cases involv- ing pairs, but focused on nonhuman bones (4).
ing commingling of skeletons from multiple graves, mass Research in the areas of asymmetry, including bilateral, direc-
graves, or disaster scenes. However, pair-matching on a smaller tional, or fluctuating asymmetry, has long focused on metric dif-
scale is used to determine whether two homologous bones found ferences between the right and left side of an organism (59).
at different sites or at different times could have originated from Simple formulae to measure asymmetry have been developed as
a single individual. There are two methods of pair-matching: a part of these studies, but they have explored asymmetry with
visual pair-matching, a subjective technique that uses gross relationship to handedness and growth and development and
visual evaluation of similarities in bone morphology and taphon- did not discuss the use of these formulae for the purpose of
omy to match pairs; and osteometric sorting, a quantitative tech- pair-matching.
nique that allows statistical evaluation of size similarities The reference data tables presented here show the maximum
between homologs to evaluate possible matches (1). value, as well as the 90th and 95th percentiles of a statistic (M),
This article uses standard skeletal measurement data to statisti- which is similar to previous measures of asymmetry and is
cally evaluate possible pair matches. The data are from a data- designed to capture the range of variability between the left and
base developed to aid the metric assessment of skeletal elements right elements within human individuals. These values of M are
by producing statistical tests to resolve commingling (2,3). These shown for a number of standard skeletal measurements for paired
previously published tests can be used for pair-matching pur- elements and can be easily used to metrically test the null hypothesis
poses, but also apply to osteometric sorting of other skeletal ele- that homologous bones originated from the same individual.
ments and therefore are more laborious for use in matching
single pairs. These tests also require multiple measurements for
Methods
each bone, which may be impossible in the case of fragmentary
The reference set of measurement data utilized in this study
were compiled for osteometric sorting methods and come from
1
Laboratory Division, Trace Evidence Unit, Federal Bureau of Investiga- several skeletal collections and databases, including the Forensic
tion, 2501 Investigation Parkway, Quantico, VA. Data Bank and William M. Bass Donated Skeletal Collection at
2
Department of Anthropology, Smithsonian Institution, National Museum the University of Tennessee, Knoxville; the Robert J. Terry Ana-
of Natural History, Washington, DC.
3
Central Identification Laboratory, Joint POW-MIA Accounting Com- tomical Skeletal Collection at the Smithsonian Institutions
mand, 310 Worchester Avenue, Hickam AFB, HI. National Museum of Natural History Department of Anthropol-
*Disclaimer: Names of commercial manufacturers are provided for identi- ogy; the Hamann-Todd Osteological Collection at the Cleveland
fication purposes only, and inclusion does not imply endorsement of the Museum of Natural History; the International Commission on
manufacturer or its products or services by the FBI. The views expressed are
those of the authors and do not necessarily reflect the official policy or posi-
Missing Persons, Bosnia and Herzegovina; the Peabody Museum
tion of the FBI or the U.S. Government. of Archaeology and Ethnology at Harvard University; and the
Received 3 June 2011; and in revised form 23 April 2012; accepted 6 Central Identification Laboratory, Joint POW/MIA Accounting
May 2012. Command. The sample used here includes 108 women and 283
2013 American Academy of Forensic Sciences
952 Published 2013. This article is a U.S. Government work and is in the public domain in the U.S.A.
THOMAS ET AL. . PAIR-MATCHING TABLES 953
TABLE 1Continued.
men of Black (48 men, 32 women), White (215 men, 68 of L and R. Thus, the statistic M expresses the difference in val-
women), Asian (13 men, 4 women), Hispanic/Mexican (5 men, ues between left and right homologs as a proportion of the aver-
3 women), and other (2 men, 1 female) descent. A sample of age value of the two bones. Left and right homologs with the
Vietnamese men was excluded due to the measurements having same measurement will have an M-value of zero. M is similar to
been collected for one side only. A more detailed description of the measure of standardized asymmetry developed by Van Valen
the reference set can be found in previous publications (2,3). An (5), which has been used in several papers discussing asymmetry
attempt was made in the development of this database to incor- and handedness (69). However, M is unrelated to previously
porate diversity in terms of ancestry and sex in hopes that the described osteometric sorting statistics and was not utilized in
statistical assessments would be conservative in the following previous osteometric sorting papers (24).
sense: greater diversity creates higher variance that reduces the Values of M were calculated for all samples in the database
likelihood of erroneous segregations. Of course, this conserva- where the appropriate measurements were available. Individual
tism comes at a cost of power in cases where the ancestry and measurements were evaluated for inclusion in the current refer-
sex are known in advance. It is important to note that the bilat- ence tables and suspected data errors were excluded. Measure-
eral measurements were not available for every individual in the ments applying to paired elements were included unless there
database. In some instances, only a subset of the measurements was inadequate sample size (total n < 45). The 90th and 95th
could be obtained. Thus, the reported sample sizes vary for each percentiles of M for each measurement were calculated using
measurement. The test described below makes no assumptions Microsoft Excel (Redmond, WA), which uses an algorithm that
regarding population affinity or sex, and the user must take heed conducts linear interpolation between data points and is therefore
of the appropriateness of the reference sample when interpreting capable of returning percentile values that are not present in the
results. original data set (12). These percentiles are reported in Table 2,
The database consists of standard postcranial skeletal measure- along with the maximum value of M from the data set for each
ments (10) as well as additional measurements added by Byrd (3) measurement.
and Byrd and Adams (2) designed for use with fragmented bones.
All 51 measurements used in this study are described in Table 1.
Results and Discussion
Interobserver variation was tested on a difficult subset of these
measurements and was found to show only modest error rates Table 2 includes the 90th and 95th percentiles, and the maxi-
(generally <3%) and, thus, high reliability (11, p. 1199). mum M-value for men and women separately, as well as for the
The statistic (M) used in this study can be expressed as sexes combined, a situation that may be more reflective of an
follows: actual forensic case where sex may be undeterminable. The val-
ues of M for men and women were subjected to a t-test to deter-
M jL Rj=L R=2 mine whether there were differences between the two sexes with
regard to this statistic. When evaluated (a = 0.05), three mea-
where L and R are the measurements of the left and the right surements showed a statistically significant difference in mean
bone, respectively, and the expression (L+R)/2 gives the average values between men and women: the physiological length of the
THOMAS ET AL. . PAIR-MATCHING TABLES 955
TABLE 2Maximum values and the 90th and 95th percentiles for the statistic M. Measurements in bold show a statistically significant difference in mean
values between men and women (a = 0.05).
Measurement # n 90th 95th Max M n 90th 95th Max M n 90th 95th Max M
Clavicle Max Length 35 26 0.040 0.045 0.047 78 0.053 0.058 0.081 104 0.049 0.056 0.081
Clavicle A-P Diameter Midshaft 36 26 0.191 0.200 0.222 67 0.113 0.151 0.194 93 0.182 0.154 0.222
Clavicle M-L Diameter Midshaft 37 26 0.236 0.250 0.353 66 0.148 0.184 0.200 92 0.180 0.200 0.353
Scapula Height 38 31 0.028 0.034 0.050 71 0.032 0.040 0.077 102 0.031 0.039 0.077
Scapula Breadth 39 35 0.028 0.036 0.053 80 0.033 0.038 0.064 115 0.032 0.040 0.064
Scapula Maximum Height Glenoid Fossa 39A 12 0.033 0.040 0.048 55 0.057 0.061 0.092 67 0.050 0.061 0.092
Scapula Maximum Breadth Glenoid Fossa 39B 12 0.053 0.055 0.058 56 0.052 0.058 0.082 68 0.054 0.058 0.082
Humerus Maximum Length 40 38 0.020 0.021 0.023 114 0.023 0.030 0.071 152 0.021 0.026 0.071
Humerus Epicondylar Breadth 41 35 0.037 0.050 0.057 100 0.034 0.044 0.079 135 0.034 0.047 0.079
Humerus Capitulum-Trochlea Breadth 41A 9 0.022 0.028 0.034 48 0.025 0.055 0.067 57 0.039 0.052 0.067
Humerus Head Diameter 42 30 0.028 0.037 0.047 98 0.040 0.041 0.089 128 0.040 0.043 0.089
Humerus A-P Head Breadth 42A 10 0.022 0.030 0.038 36 0.038 0.039 0.043 46 0.034 0.038 0.043
Humerus Maximum Diameter Midshaft 43 33 0.059 0.102 0.133 85 0.091 0.097 0.160 118 0.091 0.099 0.160
Humerus Min Diameter Midshaft 44 38 0.069 0.074 0.162 100 0.082 0.105 0.162 138 0.074 0.101 0.162
Humerus Min Diameter Diaphysis 44 14 0.039 0.042 0.047 61 0.069 0.081 0.101 75 0.066 0.079 0.101
Radius Length 45 29 0.024 0.031 0.032 105 0.016 0.026 0.064 134 0.019 0.029 0.064
Radius A-P Diameter Midshaft 46 25 0.101 0.105 0.105 79 0.080 0.090 0.154 104 0.089 0.098 0.154
Radius M-L Diameter Midshaft 47 25 0.078 0.092 0.143 79 0.106 0.118 0.139 104 0.097 0.117 0.143
Radius Maximum Diameter at Radial Tuberosity 47A 9 0.035 0.042 0.049 48 0.064 0.078 0.096 57 0.057 0.079 0.096
Radius Maximum Diameter of Diaphysis 47B 9 0.032 0.040 0.047 47 0.081 0.107 0.167 56 0.075 0.106 0.167
Distal to Radial Tuberosity
Radius Minimum Diameter of Diaphysis Distal 47C 13 0.056 0.062 0.070 59 0.068 0.076 0.091 72 0.068 0.073 0.091
to Radial Tuberosity
Ulna Length 48 33 0.024 0.027 0.039 96 0.020 0.025 0.041 129 0.022 0.026 0.041
Ulna Dorso-Volar Diameter 49 34 0.122 0.188 0.400 82 0.133 0.158 0.533 116 0.129 0.186 0.533
Ulna Transverse Diameter 50 38 0.089 0.128 0.182 94 0.117 0.202 0.375 132 0.116 0.160 0.375
Ulna Physiological Length 51 26 0.023 0.027 0.039 59 0.018 0.026 0.032 85 0.022 0.027 0.039
Ulna Min Diameter Osseous Crest 51A 8 0.078 0.090 0.102 40 0.069 0.086 0.091 48 0.071 0.087 0.102
Ulna Min Diameter 51B 8 0.044 0.049 0.053 42 0.065 0.082 0.095 50 0.062 0.081 0.095
Os Coxa Height 56 32 0.013 0.022 0.035 101 0.019 0.023 0.051 133 0.019 0.023 0.051
Os Coxa Iliac Breadth 57 35 0.021 0.023 0.048 97 0.025 0.031 0.077 132 0.025 0.030 0.077
Os Coxa Maximum Thickness at Sciatic Notch 59A 9 0.076 0.098 0.120 59 0.091 0.116 0.158 68 0.092 0.118 0.158
Os Coxa Maximum Diameter of Acetabulum 59E 6 0.034 0.040 0.046 40 0.031 0.033 0.037 46 0.031 0.035 0.046
Femur Maximum Length 60 27 0.014 0.015 0.019 82 0.013 0.015 0.020 109 0.014 0.015 0.020
Femur Epicondylar Length 61 25 0.012 0.014 0.015 74 0.014 0.016 0.021 99 0.014 0.015 0.021
Femur Epicondylar Breadth 62 23 0.028 0.040 0.063 85 0.013 0.024 0.026 108 0.022 0.025 0.063
Femur Head Diameter 63 25 0.025 0.025 0.048 97 0.028 0.037 0.045 122 0.026 0.037 0.048
Femur A-P Subtrochlear Diameter 64 37 0.076 0.092 0.118 103 0.069 0.094 0.129 140 0.071 0.095 0.129
Femur Transverse Subtrochlear Diameter 65 34 0.063 0.072 0.102 92 0.069 0.097 0.163 126 0.066 0.094 0.163
Femur A-P Diameter Midshaft 66 27 0.064 0.071 0.083 52 0.041 0.055 0.071 79 0.053 0.067 0.083
Femur S-I Neck Diameter 68D 5 0.058 0.059 0.061 48 0.056 0.065 0.082 53 0.056 0.063 0.082
Tibia Length 69 28 0.012 0.014 0.033 108 0.014 0.016 0.028 136 0.014 0.015 0.033
Tibia Maximum Breadth of the Prox Epiphysis 70 22 0.031 0.031 0.041 82 0.025 0.027 0.041 104 0.026 0.031 0.041
Tibia Maximum Breadth of the Dist Epiphysis 71 22 0.044 0.057 0.078 79 0.040 0.046 0.059 101 0.042 0.051 0.078
Tibia Maximum Diameter at Nutrient Foramen 72 34 0.085 0.097 0.127 104 0.069 0.090 0.126 138 0.073 0.095 0.127
Tibia Transverse Diameter at Nutrient Foramen 73 31 0.095 0.154 0.424 91 0.080 0.092 0.251 122 0.083 0.097 0.424
Tibia Maximum A-P Diameter Distal to 74A 7 0.044 0.045 0.046 40 0.062 0.077 0.090 47 0.060 0.073 0.090
Popliteal Line
Tibia Min A-P Diameter Distal to Popliteal Line 74B 6 0.028 0.028 0.029 43 0.074 0.090 0.094 49 0.074 0.086 0.094
Fibula Length 75 22 0.014 0.020 0.031 85 0.013 0.016 0.041 107 0.013 0.016 0.041
Fibula Maximum Diameter Midshaft 76 19 0.100 0.119 0.133 56 0.087 0.109 0.133 75 0.092 0.118 0.133
Fibula Minimum Diameter of Diaphysis 76B 8 0.052 0.057 0.127 50 0.108 0.124 0.149 58 0.108 0.128 0.149
Calcaneus Length 77 16 0.024 0.026 0.029 57 0.029 0.038 0.056 73 0.029 0.033 0.056
Calcaneus Middle Breadth 78 15 0.046 0.051 0.054 51 0.045 0.048 0.085 66 0.045 0.050 0.085
ulna; the anteriorposterior diameter of the clavicle at midshaft; The reference tables presented here provide a simple tool for
and the anteriorposterior diameter of the femur at midshaft. use in the metric assessment of possible skeletal pair matches.
These measurements are shown in bold in Table 2. It should be Analysts can test the null hypothesis that two homologs are from
noted that when performing 51 tests with a = 0.05, there is a the same individual by calculating the value of M for each mea-
greater than 80% probability of obtaining three significant results surement for the bones in question and comparing it to the 90th
by chance alone. Thus, the reader should consider the differ- and 95th percentiles or the maximum value of M from the data-
ences between the sexes as generally of minor consequence in base to make a statistical evaluation of the difference in size
osteometric sorting, and given the small sample size for women between the left and right elements. If the value of M is greater
for certain measurements, it is suggested that the percentile val- than that for the percentile chosen, the null hypothesis can be
ues with the sexes combined be used. rejected, providing evidence that it is unlikely that the two bones
956 JOURNAL OF FORENSIC SCIENCES
originated from the same individual. However, the reverse is not 3. Byrd JE. Models and methods for osteometric sorting. In: Adams BJ,
necessarily true; failure to reject the null hypothesis is not suffi- Byrd JE, editors. Recovery, analysis, and identification of commingled
human remains. Totowa, NJ: Humana Press, 2008;199220.
cient to conclude that the elements are from the same individual. 4. Lyman RL. Identifying bilateral pairs of deer (Odocoileus sp.) bones:
Additional analyses, such as visual pair-matching, would com- how symmetrical is symmetrical enough? J Archaeol Sci 2006;33:
plement this metric assessment. Investigators utilizing this 125665.
method are encouraged to consider the results in conjunction 5. Van Valen L. A study of fluctuating asymmetry. Evolution 1962;16:
12542.
with other lines of evidence when sorting commingled remains.
6. Steele J, Mays S. Handedness and directional asymmetry in the long
bones of the human upper limb. Int J Osteoarchaeol 1995;5:3949.
T, Leben-Seljak P, Stefan
7. Cuk cic M. Lateral asymmetry of human long
Acknowledgments
bones. Variability and Evolution 2001;9:1932.
The following institutions provided skeletal collections for the 8. Auerbach BM, Ruff CB. Limb bone bilateral asymmetry: variability and
commonality among modern humans. J Hum Evol 2006;50:20318.
measurement database: Central Identification Laboratory, Joint 9. Blackburn A. Bilateral asymmetry of the humerus during growth and
POW/MIA Accounting Command; Forensic Anthropology Cen- development. Am J Phys Anthropol 2011;145:63946.
ter, University of Tennessee, Knoxville; Department of Anthro- 10. Moore-Jansen PH, Ousley SD, Jantz RL. Data collection procedures for
pology, Smithsonian Institution; Cleveland Museum of Natural forensic skeletal material. Knoxville, TN: University of Tennessee,
History; International Commission on Missing Persons, Bosnia- Department of Anthropology, 1994;Report of Investigations No. 48.
11. Adams BJ, Byrd JE. Interobserver variation of selected postcranial skele-
Herzegovina; and Peabody Museum, Harvard University. We tal measurements. J Forensic Sci 2002;47:1193202.
also thank FBI Laboratory personnel and anonymous reviewers 12. Hyndman RJ, Fan Y. Sample quantiles in statistical packages. Am Stat
for reviewing earlier drafts of the manuscript. 1996;50:3615.