.S .l. 9...:
3-. {31:};

99.4
$5.. an.» ..
unmet. .

 

‘1!

viva:

P r! 1...!
A" RE
I .0 I

SPVVM

vl“ I};

b.‘ .37
A... |11
33.0....

 

 

V Iv
1A.. .. t...
;2.) :va

.i-r, ens .
...J.a...v..:?

Ipvt

‘1'

.3?!

if
if

t

1.1.54.1‘:
.l .5! .an‘.i...

I

:7

 

\iiiiiigiiiiiiiiiﬁiigii

This is to certify that the

dissertation entitled

BILATERAL ASYMMETRY PATTERNS IN LINEAR DIMENSIONS
OF THE HUMERUS AND FEMUR IN THREE HUMAN
SKELETAL POPULATIONS

presented by

Brian David Brown

has been accepted towards fulﬁllment
of the requirements for

. PhD degree in Anthropology

 

 

/ 7
[ﬂag/mar? Q toJ/t/té/L/

r professor

Date 11-17-95

 

MS U i: an Afﬁrmative Anion/Equal Opportunity Institution 0-1277 1

 

LIBRARY :
Michigan State,
University i

 

 

PLACE It RETURN BOX to romovo this chockout from your rooord.
TO AVOID FINES rotum on or baton duo duo.

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

usu to An Afﬂrmdivo ActiorVEquol Opportunity institution
WM‘

 

BILATERAL ASYMMETRY PATTERNS IN LINEAR DIMENSIONS
OF THE HUMERUS AND FEMUR IN THREE HUMAN SKELETAL POPULATIONS

By

Brian David Brown

A DISSERTATION
Submitted to
Michigan State University
in partial fulﬁllment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY

Department of Anthropology

1995

ABSTRACT
BILATERAL ASYMMETRY PATTERNS IN LINEAR DIMENSIONS
or THE HUMERUS AND FEMUR IN THREE HUMAN SKELETAL POPULATIONS
By

Brian David Brown

Analysis of bilateral asymmetry in human limb bones is an increasingly prominent aspect
of the biocultural approach to human skeletal biology, and several recent studies have
assessed asymmetry in the cross-sectional geometry of long bone diaphyses. While the
presence of directional asymmetry in linear dimensions of long bones is well
documented, the nature of these asymmetry patterns among human populations has not
yet been well characterized. This study lays a foundation for future biocultural work
with linear asymmetry by assessing two skeletal samples ﬁom Georgian-era London
church crypts (St. Bride’s, Fleet Street, and Christ Church, Spitalﬁelds), and comparing
their asymmetry with ﬁgures derived ﬁ'om the Forensic Data Bank (at the University of
Tennessee). The analysis focuses on Six paired measurements: maximum length,
vertical head diameter, and biepicondylar width of the humerus; and maximum length,
head diameter, and biepicondylar width of the femur. Study hypotheses test whether the
samples and the sexes display statistical independence in the distribution of (a) direction
of asymmetry, (b) signed and unsigned magnitude of asymmetry; and (c) size of the
linear dimensions themselves. The two Georgian samples are ﬁrst assessed separately,
then pooled and compared with the Forensic sample, and in each case the sexes are

treated separately. All male-female comparisons display signiﬁcant diﬂ‘erences in each

of the long bone dimensions, with both Georgian samples signiﬁcantly smaller than their
Forensic counterparts in each of the dimensions. The Georgian females display the
greatest asynunetry in humerus length, and the Georgian males display the greatest '
asymmetry in femur length. Asymmetry patterns are more often signiﬁcantly diﬁ‘erent in
Georgian-Forensic comparisons than in comparisons between the sexes of each sample,
or between same-sex subgroups of the Georgian samples; this indicates that setting in
time and Space, but not sex is a determinant of asymmetry patterns in long bone
dimensions. Several methodological issues involving the assessment of asymmetry and

historic skeletal populations are also discussed.

Copyright by
BRIAN DAVID BROWN

1995

To my parents

ACKNOWLEDGMENTS

No work of this magnitude can be accomplished by a single person working in
isolation, and this certainly is no exception. I wish to acknowledge the assistance of a
multitude of professional colleagues, ﬁiends, and family; without their support this
project would never have come to any level of completion.

First I thank my colleagues who have primary responsibility for the St. Bride’s
and Spitalﬁelds skeletal material and the Forensic Data Bank. At St. Bride’s I thank Dr.
J. Louise Scheuer, Dr. Sue M. Black, and Dr. Jacqui Bowman for their selﬂess assistance
and camaraderie while I was in the crypt; Canon John Oates for giving me formal access
to the collection; and The Leverhulme Foundation for funding the rehabilitation of the St.
Bride’s skeletal material; it made my task much easier. At the British Museum of
Natural History in particular I thank Miss Theya Molleson for giving me access to the
Spitalﬁelds collection and the resources of the museum, and the English Heritage
Foundation for funding the initial excavation and research on the Spitalﬁelds material. I
thank Steve Ousley at the University of Tennessee for his assistance in providing me with
raw ﬁgures ﬁ'om the Forensic Data Bank. In addition, I thank the numerous researchers
across the country who initially collected the forensic data, and Dr. Christopher Ruﬁ‘ for

providing extremely useful reprints of his work on skeletal asymmetry

Secondly I thank my committee for their thoughtful consideration and reviews of
my work. Bill Lovis and J en'y Voss each spent many grueling hours reviewing several
drafts of the manuscript, and the product is much improved as a result of their careful and
diligent efforts. I also appreciate Larry Robbins’ willingness to assist in the latter stages
of the project in spite of his myriad other commitments. Most importantly I thank Norm
Sauer for his unwavering support throughout my entire graduate career, a career he
inﬂuenced in many more ways than can be articulated here. A student could not ask for a
better collection of faculty to serve on a guidance committee, and I am glad they were
here.

Thirdly I thank my family and ﬁiendsuincluding students, faculty, staﬁ‘, and
people far ﬁ'om the University-for sticking with me throughout the whole process. Their
individual contributions would ﬁll volumes; I can only say that their support and

steadfast acceptance were and are invaluable to me. I have been fortunate.

TABLE OF CONTENTS

CHAPTER l—AN INTRODUCTION .................................... 1
THE BIOCULTURAL PERSPECTIVE ............................. 5
OSTEOMETRY AND TIE BIOCULTURAL PERSPECTIVE ........... 5
METRIC ASYMMETRY ........................................ 6
PLAN OF THE FOLLOWING CHAPTERS ......................... 10

CHAPTER 2—TIE PROBLEM ........................................ 12
ASYMMETRY ASSESSMENT AS AN ANALYTICAL TOOL .......... 13
ISSUES IN THE POSTCRANTAL ASYMMETRY LITERATURE ....... 28

CHAPTER 3—TIE STUDY POPULATIONS ............................. 32
THE NATURE OF SKELETAL STUDY POPULATIONS .............. 32
GEORGIAN LONDON AND THE CRYPT POPULATIONS ........... 48
THE SKELETAL COLLECTIONS ................................ 59

CHAPTER 4-—METHODS ............................................ 61
THE MEASUREMENTS ....................................... 61
STUDY SAMPLE SELECTION .................................. 66
TI-E SHRINKAGE PROBLEM .................................. 71
ASSESSING DISTRIBUTION OF THE LINEAR DIMENSIONS ........ 75
ASSESSING ASYNINETRY .................................... 78
CALCULATING ASYMMETRY MAGNTTUDE ..................... 79
ASSESSING CROSS SYMMETRY ............................... 83
TESTING THE SIGNIFICANCE OF ASYMMETRY ................. 83
HYPOTIESIS TESTING ....................................... 90
SUMMARY ................................................. 93

CHAPTER 5—RESULTS ............................................. 95
LINEAR DIMENSIONS DISTRIBUTION PATTERNS ................ 96
ASYMMETRY DIRECTION ................................... 127
ASYMMETRY MAGNITUDE .................................. 153
PATTERNS IN THE HUMERUS ................................ 205
PATTERNS IN THE FEMUR ................................... 207
PATTERNS IN THE HUMERUS AND FEMUR .................... 209
SUMMARY: RESULTS OF HYPOTI-ESIS TESTING ............... 212

CHAPTER 6—DISCUS SION ......................................... 218
SUMMARY OF THE STUDY STRATEGY ........................ 218
SIGNIFICANT ISSUES IN TI-E STUDY FINDINGS ................ 221
ISSUES INVOLVED WITH DOCUMENTED HISTORIC POPULATION8226
CONCLUSION .............................................. 23 1

viii

LIST OF TABLES

Table l Asymmetry Distributions: Normal vs. Skewed (Signed Values) .......... 16
Table 2 Asymmetry Distributions: Normal vs. Skewed (Unsigned Values) ........ 17
Table 3 Examples of Three Fundamental Asymmetry Patterns. ................. 22

Table 4 Humerus and Femur Asymmetry Direction (Lowrance and Latirner 1957) . . 25
Table 5 Crossed Symmetry in Humeri and Femora (Latimer and Lowrance 1965) . . 26
Table 6 Changes in Humerus Asymmetry Over Time (ﬁom F resia eta], 1990) ..... 31

Table 7 Attributing Sex from Femur Head Diameter (adapted ﬁ'om Bass 1987) . . . . 34

Table 8 Sample Size—Georgian Subpopulations ........................... 68
Table 9 Sample Size—Identiﬁed Subpopulations ........................... 69
Table 10 Eﬂ’ect of Character Size on Asymmetry Magnitude .................. 81
Table 11 Sample Data for a Mann-Whitney U Calculation .................... 86
Table 12 Sample Data Placed into Ordinal Ranks ........................... 87
Table 13 Paired Georgian Humerus Length Summary Statistics ................ 96
Table 14 Intact Georgian Humerus Length Summary Statistics ................ 98
Table 15 Pooled Georgian Humerus Length Summary Statistics ................ 99
Table 16 Forensic Data Bank Humerus Length Summary Statistics ............ 100
Table 17 Georgian Humerus Head Diameter Summary Statistics .............. 102
Table 18 Humerus Head Diameter Summary Statistics ...................... 104
Table 19 Georgian Humerus Biepicondylar Width Summary Statistics .......... 106

Table 20 Humerus Biepicondylar Width Summary Statistics .................. 108

ix

Table 21 Paired Georgian Femur Length Summary Statistics ................. 111

Table 22 Intact Georgian Femur Length Summary Statistics ................. 113
Table 23 Georgian Femur Length Summary Statistics ...................... 114
Table 24 Forensic Data Bank Femur Length Summary Statistics .............. 115
Table 25 Georgian Femur Head Summary Statistics ........................ 117
Table 26 Femur Head Diameter Summary Statistics ........................ 118
Table 27 Georgian Femur Biepicondylar Width Summary Statistics ............ 121
Table 28 Femur Biepicondylar Width Summary Statistics .................... 123
Table 29 Summary: Diﬁ’erences in Humerus Size Distribution Patterns ......... 125
Table 30 Summary: Differences in Femur Size Distribution Patterns ........... 126
Table 31 Georgian Humerus Length Asymmetry Direction (2 1 mm) .......... 131
Table 32 Combined Humerus Length Asymmetry Direction (2 1 mm) .......... 131
Table 33 Georgian Humerus Length Asymmetry Direction (2 2 mm) .......... 132
Table 34 Combined Humerus Length Asymmetry Direction (2 2 mm) .......... 133
Table 35 Georgian Humerus Head Asymmetry Direction .................... 136
Table 36 Combined Humerus Head Asymmetry Direction ................... 137
Table 37 Georgian Humerus Biepicondylar Width Asymmetry Direction ........ 138
Table 38 Humerus Biepicondylar Width Asymmetry Direction ................ 139
Table 39 Georgian Femur Length Asymmetry Direction (2 1 mm) ............ 140
Table 40 Combined Femur Length Asymmetry Direction (2 1 mm) ............ 141
Table 41 Georgian Femur Length Asymmetry Direction (2 2 mm) ............ 141
Table 42 Combined Femur Length Asymmetry Direction (2 2 mm) ............ 147
Table 43 Georgian Femur Head Asymmetry Direction ...................... 148

X

Table 44 Combined Femur Head Asymmetry Direction ..................... 148

Table 45 Georgian Femur Biepicondylar Width Asymmetry Direction .......... 149
Table 46 Combined Femur Biepicondylar Width Asymmetry Direction .......... 150
Table 47 Summary: Differences in Direction of Humerus Asymmetry .......... 151
Table 48 Summary: Diﬂ‘erences in Direction of Femur Asymmetry ............ 152
Table 49 Mean Humerus Length AsymmetriesuSigned vs. Unsigned ........... 154
Table 50 Paired Georgian Humerus Length Asymmetry Magnitude ............ 157
Table 51 Combined Humerus Length Asymmetry Magnitude ................. 158
Table 52 Humerus Length Asymmetry—Ordinal Ranking ................... 161
Table 53 Summary: Diﬂ'erences in Humerus Length Asymmetry Magnitude ..... 162
Table 54 Georgians Humerus Head Asymmetry Magnitude .................. 163
Table 55 Combined Humerus Head Asymmetry Magnitude .................. 168
Table 56 Humerus Head Asymmetry—Ordinal Ranking ..................... 169
Table 57 Summary: Diﬁ‘erences in Humerus Head Asymmetry Magnitude ...... 170
Table 58 Georgian Humerus Biepicondylar Width Asymmetry Magnitude ....... 171
Table 59 Combined Humerus Biepicondylar Width Asymmetry Magnitude ...... 172
Table 60 Humerus Biepicondylar Width—Ordinal Ranking .................. 177
Table 61 Humerus Biepicondylar Width Asymmetry Magnitude Diﬁ‘erences ...... 178
Table 62 Mean Femur Length Asymmetries—Signed vs. Unsigned ............ 179
Table 63 Paired Georgian Femur Length Asymmetry Magnitude .............. 181
Table 64 Combined Femur Length Asymmetry Magnitude ................... 182
Table 65 Femur Length Asymmetry—Ordinal Ranking ..................... 183

Table 66 Summary: Diﬂ‘erences in Femur Length Asymmetry Magnitude Patterns . 184

xi

Table 67 Intact Georgian Femur Head Asymmetry Magnitude ................ 189

Table 68 Combined Femur Head Asymmetry Magnitude .................... 194
Table 69 Femur Head Diameter Asymmetry—Ordinal Ranking ............... 195
Table 70 Summary: Diﬁ'erences in Femur Head Asynunetry Magnitude ......... 196
Table 71 Georgian Femur Biepicondylar Width Asymmetry Magnitude ......... 202
Table 72 Combined Femur Biepicondylar Width Asymmetry Magnitude ........ 203
Table 73 F emur Biepicondylar Width Asymmetry—Ordinal Ranking ........... 203
Table 74 Femur Biepicondylar Width Asymmetry Magnitude Differences ........ 204
Table 75 Georgian Diﬂ‘erences in Humerus Asymmetry Direction Patterns ....... 206
Table 76 Summary: Differences in Humerus Asymmetry Direction Patterns ..... 207
Table 77 Georgian Differences in Femur Patterns .......................... 208
Table 78 Summary: Diﬂ‘erences in Femur Asymmetry Direction Patterns ....... 209
Table 79 Georgian Cross Symmetry in Humerus and Femur Lengths ........... 210
Table 80 Combined Cross Symmetry in Humerus and Femur Lengths .......... 211
Table 81 Signiﬁcant Differences in Humerus and Femur Dimensions ........... 212

Table 82 Signiﬁcant Differences in Humerus Asymmetry Among the Samples . . . . 213
Table 83 Signiﬁcant Diﬂ’erences in Femur Asymmetry Among the Samples ...... 214
Table 84 Signiﬁcant Diﬂ’erences in Asymmetry Direction Between Measurements . 215
Table 85 Summary of Statistical Signiﬁcance Between the Sexes .............. 234

Table 86 Summary of Statistical Signiﬁcance Among Populations ............. 235

xii

Table 67 Intact Georgian Femur Head Asymmetry Magnitude ................ 189

Table 68 Combined Femur Head Asymmetry Magnitude .................... 194
Table 69 F emur Head Diameter Asymmetry—Ordinal Ranking ............... 195
Table 70 Summary: Differences in Femur Head Asymmetry Magnitude ......... 196
Table 71 Georgian Femur Biepicondylar Width Asymmetry Magnitude ......... 202
Table 72 Combined Femur Biepicondylar Width Asymmetry Magnitude ........ 203
Table 73 F emur Biepicondylar Width Asymmetry—Ordinal Ranking ........... 203
Table 74 Femur Biepicondylar Width Asymmetry Magnitude Differences ........ 204
Table 75 Georgian Diﬂ‘erences in Humerus Asymmetry Direction Patterns ....... 206
Table 76 Summary: Differences in Humerus Asymmetry Direction Patterns ..... 207
Table 77 Georgian Differences in Femur Pattems .......................... 208
Table 78 Summary: Differences in Femur Asymmetry Direction Patterns ....... 209
Table 79 Georgian Cross Symmetry in Humerus and Femur Lengths ........... 210
Table 80 Combined Cross Symmetry in Humerus and Femur Lengths .......... 211
Table 81 Signiﬁcant Diﬂ’erences in Humerus and Femur Dimensions ........... 212

Table 82 Signiﬁcant Differences in Humerus Asymmetry Among the Samples . . . . 213
Table 83 Signiﬁcant Differences in Femur Asymmetry Among the Samples ...... 214
Table 84 Signiﬁcant Diﬂ‘erences in Asymmetry Direction Between Measurements . 215
Table 85 Summary of Statistical Signiﬁcance Between the Sexes .............. 234

Table 86 Summary of Statistical Signiﬁcance Among Populations ............. 235

LIST OF FIGURES

Figure 1 Relative Location of St. Bride’s and Christ Church, Spitalﬁelds ......... 44
Figure 2 Humerus Length Distribution — Georgian Males .................... 97
Figure 3 Humerus Length Distribution — Georgian Females .................. 97
Figure 4 Humerus Length Distribution — Identiﬁed Males .................. 101
Figure 5 Humenis Length Distribution — Identiﬁed Females ................. 101
Figure 6 Humerus Head Distribution — Georgian Males .................... 103
Figure 7 Humerus Head Distribution — Georgian Females .................. 103
Figure 8 Humerus Head Distribution — Identiﬁed Males .................... 105
Figure 9 Humerus Head Distribution — Identiﬁed Females .................. 105
Figure 10 Humerus Biepi Distribution — Georgian Males ................... 107
Figure 11 Humerus Biepi Distribution — Georgian Females ................. 107
Figure 12 Humerus Biepi Distribution — Identiﬁed Males ................... 110
Figure 13 Humerus Biepi Distribution — Identiﬁed Females ................. 110
Figure 14 Femur Length Distribution — Georgian Males .................... 112
Figure 15 Femur Length Distribution — Georgian Females .................. 112
Figure 16 Femur Length Distribution — Identiﬁed Maloes ................... 116
Figure 17 Femur Length Distribution — Identiﬁed Females .................. 116
Figure 18 Femur Head Distribution — Georgian Males ..................... 119
Figure 19 Femur Head Distribution — Georgian Females ................... 119
Figure 20 Femur Head Distribution — Identiﬁed Males ..................... 120

Figure 21 Femur Head Distribution — Identiﬁed Females ................... 120

Figure 22 Femur Biepi Width Distribution — Georgian Males ................ 122
Figure 23 Femur Biepi Width Distribution — Georgian Females .............. 122
Figure 24 Femur Biepi Width Distribution — Identiﬁed Males ................ 124
Figure 25 Femur Biepi Width Distribution — Identiﬁed Females .............. 124
Figure 26 Humerus Asymmetry Direction —- Spitalﬁelds Males ............... 128
Figure 27 Humerus Asymmetry Direction — St. Bride’s Males ............... 128
Figure 28 Humerus Asymmetry Direction — Spitaiﬁeids Females ............. 129
Figure 29 Humerus Asymmetry Direction —- St. Bride’s Females .............. 129
Figure 30 Humerus Asymmetry Direction — Georgian Males ................ 134
Figure 31 Humerus Asymmetry Direction — Georgian Females ............... 134
Figure 32 Humerus Asymmetry Direction -— Forensic Males ................. 135
Figure 33 Humerus Asymmetry Direction — Forensic Females ............... 135
Figure 34 Femur Asyrm'netry Direction — Spitalﬁelds Males ................. 143
Figure 35 Femur Asymmetry Direction — St. Bride’s Males ................. 143
Figure 36 Femur Asymmetry Direction — Spitalﬁelds Females ............... 144
Figure 37 Femur Asymmetry Direction — St. Bride’s Females ................ 144
Figure 38 Femur Asymmetry Direction — Georgian Males .................. 145
Figure 39 Femur Asymmetry Direction — Georgian Females ................. 145
Figure 40 Femur Asymmetry Direction — Forensic Males ................... 146
Figure 41 Femur Asymmetry Direction — Forensic Females ................. 146
Figure 42 Georgian Male Signed Humerus Length Asymmetry ............... 155
Figure 43 Georgian Female Signed Humerus Length Asymmetry ............. 155

xiv

Figure 44 Identiﬁed Male Signed Humerus Length Asymmetry ............... 156

Figure 45 Identiﬁed Female Signed Humerus Length Asymmetry .............. 156
Figure 46 Georgian Male Humerus Length Asymmetry ..................... 159
Figure 47 Georgian Female Humerus Length Asymmetry ................... 159
Figure 48 Identiﬁed Male Humerus Length Asymmetry .................... 160
Figure 49 Identiﬁed Female Humerus Length Asymmetry ................... 160
Figure 50 Georgian Male Signed Humerus Head Asymmetry ................ 164
Figure 51 Georgian Female Signed Humerus Head Asymmetry ............... 164
Figure 52 Identiﬁed Male Signed Humerus Head Asynunetry ................ 165
Figure 53 Identiﬁed Female Signed Humerus Head Asymmetry .............. 165
Figure 54 Georgian Male Humerus Head Asymmetry ....................... 166
Figure 55 Georgian Female Humerus Head Asymmetry ..................... 166
Figure 56 Identiﬁed Male Humerus Head Asymmetry ...................... 167
Figure 57 Identiﬁed Female Humerus Head Asymmetry ..................... 167
Figure 58 Georgian Male Signed Humerus Biepi Asymmetry ................. 173
Figure 59 Georgian Female Signed Humerus Biepi Asymmetry ............... 173
Figure 60 Identiﬁed Male Signed Humerus Biepi Asymmetry ................. 174
Figure 61 Identiﬁed Female Signed Humerus Biepi Asymmetry ............... 174
Figure 62 Georgian Male Humerus Biepi Asymmetry ....................... 17 5
Figure 63 Georgian F ernale Humerus Biepi Asymmetry ..................... 175
Figure 64 Identiﬁed Male Humerus Biepi Asymmetry ...................... 176
Figure 65 Identiﬁed Female Humerus Biepi Asymmetry ..................... 176
Figure 66 Georgian Male Signed Femur Length Asymmetry ................. 185

XV

Figure 67 Georgian Female Signed F emur Length Asymmetry ................ 185

Figure 68 Identiﬁed Male Signed Femur Length Asymmetry ................. 186
Figure 69 Identiﬁed Female Signed Femur Length Asymmetry ................ 186
Figure 70 Georgian Male Femur Length Asymmetry ....................... 187
Figure 71 Georgian Female Femur Length Asymmetry ..................... 187
Figure 72 Identiﬁed Male Femur Length Asymmetry ....................... 188
Figure 73 Identiﬁed Female Femur Length Asymmetry ..................... 188
Figure 74 Georgian Male Signed F emur Head Asymmetry .................. 190
Figure 75 Georgian Female Signed Femur Head Asymmetry ................. 190
Figure 76 Identiﬁed Male Signed Femur Head Asymmetry .................. 191
Figure 77 Identiﬁed Female Signed Femur Head Asymmetry ................. 191
Figure 78 Georgian Male Femur Head Asymmetry ......................... 192
Figure 79 Georgian Female Femur Head Asymmetry ....................... 192
Figure 80 Identiﬁed Male Femur Head Asymmetry ........................ 193
Figure 81 Identiﬁed Female Femur Head Asymmetry ....................... 193
Figure 82 Georgian Male Signed Femur Biepi Asymmetry ................... 198
Figure 83 Georgian Female Signed Femur Biepi Asymmetry ................. 198
Figure 84 Identiﬁed Male Signed Femur Biepi Asymmetry .................. 199
Figure 85 Identiﬁed Female Signed Femur Biepi Asymmetry ................. 199
Figure 86 Georgian Male Femur Biepi Asymmetry ......................... 200
Figure 87 Georgian Female Femur Biepi Asymmetry ...................... 200
Figure 88 Identiﬁed Male Femur Biepi Asymmetry ........................ 201
Figure 89 Identiﬁed Female Femur Biepi Asymmetry ....................... 201

xvi

CHAPTER l—AN INTRODUCTION

There is popular appeal to the belief that a physical anthropologist can construct a
meaningful narrative about the life of an individual or a group of people by carefully .
examining skeletal remains. This appeal is exempliﬁed by a recent series of mystery
novels which feature the exploits of the affable forensic anthropologist Gideon Oliver.l
In each book Oliver is presented with troublesome cases of death under mysterious
circumstances. He proceeds to decipher seemingly obscure aspects of skeletal
morphology to reveal facts about a deceased individual—facts which could only be
determined by the trained eye of an anthropologist, and facts which lead to resolution of
an engrossing mystery.

In recent years the Gideon Oliver mysteries have been joined by nonﬁction
accounts written by or about real-life physical anthropologists. They include W
W (Ubelaker and Scamell 1992). witnessesﬁamths
MW (Joyce and Stover 1991), and MW
(Schwartz 1993). One need not even read past the titles of these three volumes to grasp
their fundamental message that skeletal material is a text that can be read and accurately
interpreted by an expert. It is a message that comes as little surprise to skeletal
biologists, for whom W is not just the title of a
specialist volume (Iscan and Kennedy 1989) but an explicit statement of a dominant

theme in contemporary human skeletal biology.

 

‘Although Gideon Oliver is a ﬁctitious character, he is modeled on a composite of
contemporary forensic anthropologists. The creator of the series, Aaron Elkins, authored

(magmas, which was recipient of an Edgar Award for Best Mystery in 1988. Other
Gideon Oliver Mysteries include W195 (1990) and W (1983).

1

2

For physical anthropologists the task of reconstructing life from skeletal remains
is fraught with diﬁculty. Recent exchanges in W regarding “the
osteological paradox” oﬁ‘er the sobering suggestion that enthusiastic researchers have
lost sight of the limits of osteological analysis as a tool for reconstructing lifeways
(Wood M. 1992, Byers 1994). The paradox arises when the same Skeletal evidence can
be employed to support contradictory views about human lifeways. For example, one
may consider a scenario in which one of two contemporaneous and spatially local
skeletal populations consists of many individuals who display lesions associated with
tuberculosis, while the other population exhibits no such lesions. One researcher may
conclude that the population manifesting the disease markers displays a deterioration in
health status relative to the p0pulation without the markers. Another researcher might
argue that the population with the disease markers is the one which is relatively healthier,
since its members were able to survive the disease adequately to be able to develop the
skeletal lesions. In the latter interpretation, the absence of disease markers in the second
population indicates not that the disease is absent, but rather that individuals succumbed
to the disease process before it could progress to the extent that they could develop
skeletal lesions. The paradox is that skeletal markers of disease may be viewed as both a
sign of good health and a sign of poor health when comparing skeletal populations.

Even if there are potential pitfalls to the interpretation of skeletal biology, human
remains are often the only primary data available for deriving information about an
individual or a population ﬁ'orn the past. However, a physical anthropologist can only
successfully reconstruct facets of an unknown person’s life by drawing on the research of

others who have studied known individuals and have identiﬁed meaningful aspects of

3
their skeletal morphology. While such research has traditionally employed both

qualitative and quantitative strategies, the popular accounts cited above focus primarily
on non-metric techniques of osteological analysis. The contemporary emphasis on
qualitative assessments contrasts markedly with the basic osteometric analyses which
were prevalent earlier this century. In the early 1900s two major journals, the American
W981 and the British mm reported numerous
craniometric studies performed on skulls ﬁ'om around the world. The popularity of
craniometry at that time reﬂected what were thought to be the interesting anthropological
questions of the day. Speciﬁcally, craniometry was employed as a tool for discriminating
genetic relationships between human populations, and it was often the basis for drawing
conclusions about the relative intelligence of diﬂ’erent human racial groups (Gould 1981,
Annelagos an], 1982).

While early researchers subjected human crania to intense study, they showed
relatively little interest in the postcranial skeleton. The few published studies which did
address the major limb bones, however, still emphasized the comparison of long bone
morphology among racial groups (Hrdliéka 1932, Miinter 1936, Schultz 1937). Like
most so-called “racial” characteristics, variation in limb morphology between populations
did not serve as a particularly useﬁrl analytical tool for categorizing humans into discrete
groups. As a result, the perceived value of postcranial morphology as an indicator of
broader patterns in human biological variation waned as the century progressed. At the
same time so did the number of descriptive reports based on the osteometry of major limb

bones.

4

In their place, today’s osteometric analyses of limb bones serve primarily to aid
researchers in determining two pieces of information about an unknown individual: sex
and living stature.2 Sex determination on the basis of long bone measurements reﬂects
the general size dimorphism between human males and females. For example, the
diameter of the head of the humerus or the femur is commonly employed as a univariate
technique for attributing sex to skeletal material when the cranium and pelvis are not
present. Even more accurate multivariate techniques have been developed for assessing
sex ﬁom the post-cranial skeleton, and these are widely reported in the standard
osteological texts.3 Because the long bones of the lower limb contribute signiﬁcantly to
an individual’s stature, skeletal biologists have developed a series of formulae for
estimating living stature based on the lengths of limb bones among several diﬂ‘erent
human groups. Stature formulae have not been limited to the bones of the lower limb,
although upper limb measurements tend to give less accurate estimations of stature.
These formulae have also been well documented in the standard osteological texts and

are commonly in use by physical anthropologists.

 

2Estimation of sex, stature, age, and ethnic aﬂiliation are the four most
fundamental aspects of human identiﬁcation based on skeletal criteria. There are
published discriminant ﬁrnctions for the latter which employ a combination of pelvic and
femoral characteristics (DiBennardo and Taylor 1983), but none which rely solely on
measurements of limb bones. Likewise, limb bone osteometry is not a useful tool for
estimating the age of an unknown individual.

’Bennett (1993) provides currently the most up-to-date reference guide for
skeletal identiﬁcation, and the univariate and multivariate techniques described here are
outlined in his volume. Bass (1987), Steele and Bramblett (1988), Krogman and Iscan
(1986), and Stewart (1979) are other commonly cited references, and these constitute the
“standard osteological texts.”

S
W

Recent decades have seen an expansion in osteological research that reﬂects the
popularity of a biocultural approach toward skeletal biology. The biocultural perspective
is predicated on the empirical ﬁnding that the skeleton is a dynamic organ system which
is constantly being modiﬁed in interaction with the environment (Bush and Zvelebil
1991). This ﬁ'esh perspective has led skeletal biologists to focus their energy on
assessing lifeways based on characteristics of the skeleton that have been subjected to
modiﬁcation by factors in the environment (Annelagos eta], 1982).

Notwithstanding the popularity of qualitative techniques for assessing skeletal
elements in the biocultural approach, quantitative assessments of long bone morphology
have not been totally supplanted. For example, patterns which indicate an increase in
mean femur length over time within a population have been put forward as evidence for
secular increase in the population’s mean stature. The argument follows that an increase
in a population’s stature over time reﬂects a general improvement in that population’s

general health status.‘

W

One prevailing theme in the biocultural literature is that the skeleton retains

features which are indicative of an individual’s health status during life (Cohen 1989;

 

‘The relationships among stature, nutrition, and health status have been discussed
widely among clinicians (Acheson and Fowler 1964), skeletal biologists (Steegman 1985,
1986, 1991), and historians (Floud et al. 1990, Floud et al. 1993; Fogel et a1. 1983; Fogel
1986; Komlos 1993). Researchers generally agree that the issues are linked, but there is
debate about the historical signiﬁcance of the differences. Henneberg and VandenBerg
(1990) report that secular trends are not obviously associated with socioeconomic status.

6
Goodman eLal. 1988). These features include markers of generalized stress, such as

enamel hypoplasias in the dentition (Goodman and Rose 1990); they also include
markers of a speciﬁc type of stressor such as porotic hyperostosis, which is associated
with anemia (Stuart-Macadam 1989a; 1989b; 1992).

Increasingly sophisticated osteometric techniques have also been developed in
recent years to reconstruct patterns of lifeways ﬁ'om postcranial skeletal material.
Beginning in the 19803 studies which assessed diaphyseal morphology of limb bones
began to appear in the literature (Ruff and Jones 1981). In these studies new
technologies, such as three-dimensional scanning of gross skeletal morphology, were
applied to examine the cross-sectional geometric characteristics of long bones.
Diaehronic studies of spatially local native North American skeletal populations revealed
that the diaphyseal geometry of long bones—most notably the humerus—showed
signiﬁcant modiﬁcations over time. Researchers hypothesized that such changes in the
shape of the bone shaft were caused by changing patterns of mechanical loading on the
long bones. In short, they argued that evidence for changes in physical activity patterns
were preserved in the cross-sectional geometry of the bone shaﬂs. They suggested
further that changes in the size and shape of the shaft of the humerus provide evidence
for modiﬁcations in subsistence patterns within the population over time (Bridges 1989,

Fresia eLaL 1990).

W
One component of diaphyseal morphology which has been subject to considerable

recmt research is bilateral asymmetry (Ruﬂ‘eLaL 1993, Trinkaus e111, 1994, Roy 9131,

7
1994). As Ruﬂ‘ (1992250) suggests, “diﬂ’erences in the average degree of asynunetry
present within populations or subsets of populations may be indicative of signiﬁcant
diﬁ‘erences in behavioral characteristics.”

Fresia ﬁll. (1990) oﬂ’er a typical example of how asymmetry patterns in long
bones have been addressed from a biocultural perspective. Speciﬁcally, they draw
attention to changes in patterns of bilateral asymmetry in humerus morphology which,
they argue, accompanied the shift from a pre-agricultural to an agricultural way of life
among Native Americans in Georgia. While Fresia e131, are primarily interested in the
nature of bilateral asymmetry in diaphyseal cross-sectional geometry, they also brieﬂy
report ﬁndings related to asymmetry in the length of the humerus.

The growing prominence of cross-sectional analysis appears to have eclipsed
contemporary critical examination of bilateral asynunetry in the linear dimensions of
long bones. For example, the discussion by Trinkaus 21.11 (1994) of humerus
morphology in modern and premodem Home populations brieﬂy notes the presence of
linear asymmetry, indicating that the magnitude of the asymmetry among their studied
individuals is relatively small. The magnitude of asymmetry in the length of the humerus
they cite (less than two percent, in most cases) is consistent with that reported in most
studies of long bone asymmetry published to date. Missing ﬁom their discussion,
however, is a systematic appraisal of how patterns of linear asymmetry vary within and
between the skeletal populations.

The fact that the magnitude of asymmetry is relatively small appears to be
commonly used as a prim facie argument for neglecting a methodical appraisal of

bilateral asymmetry patterns in linear dimensions of human long bones across

populations. This may be an unfortunate circumstance, since patterns of asymmetry may
reveal valuable and heretofore unrecognized information about the lifeways of past and
present populations. Such an assessment would have been diﬁcult for Trinkaus e111
(1994), since they were employing small samples—sometimes only single individuals
ﬁ'om a particular site. It is striking, however, that no one to date has reported a study of
linear asymmetry patterns within or between human skeletal populations of a reasonable

size.

The Assumption of Symmetry

Stirland (1986) describes the median sagittal plane as “the central plane of the
body which passes along the central sagittal suture in the top of the skull, and about
which the body is bilaterally symmetrical and divided into right and left halves” (p. 15)
All unpaired bones (mandible, sternum, vertebrae, etc.) are regarded as virtually
symmetrical across the median sagittal plane. Likewise, all paired bones in the skeleton
are paired left and right, never anterior and posterior or superior and inferior, and “side
identiﬁcation” is one of the fundamental techniques described in introductory osteology
texts.

In spite of the general assumption that paired bones are symmetrical, the
empirical ﬁnding that asymmetry occurs regularly in human skeletal elements has been
well documented since the mid-nineteenth century; Ruﬂ‘ and Jones (1981) cites several.
of these early reports. In general, these studies have brieﬂy acknowledged the
phenomenon of bilateral asymmetry in the skeleton, but almost always as an aside to

discussing other aspects of skeletal morphology in general. As a result, the nature of

9

asymmetry in skeletal elements among members of skeletal populations has eluded

systematic investigation to date.

Explanations for Asymmetry

If researchers assume that bilateral asymmetry is the normal condition for paired
skeletal elements, then they require some way to explain the presence of asymmetry in
bones. A number of studies (which are reviewed in detail in the following chapter)
indicate that a certain level of asymmetry is, in fact, the norm for both the length of the
humerus and femur. The current understanding of the factors associated with the
presence of asymmetry in the skeleton is summarized by this passage from Helrnkamp
and Falk’s study of rhesus macaque forelimb asymmetry:

In sum, we must consider that there could be numerous and pervasive genetic,

epigenetic, hormonal factors, among others, that vary with age and sex and when

combined with environmental interaction present a complex causal hierarchy that
is played out through ontogenetic stages of development. (Helmkarnp and Falk

1990:212)

It is diﬁcult to assess the relative importance of the various factors that interact to
aﬂ‘ect the direction and magnitude of long bone asymmetry,. To come to a better
understanding of how they interact, it would be necessary to compare asymmetry patterns
among fairly closely controlled and well-documented series of skeletal material.
Individuals in the series would also need to be unambiguously identiﬁed with respect to
age and sex. Unarnbiguously identiﬁed means that individuals are identiﬁed on the basis

of documentary evidence; this is set in contrast to the identiﬁcation of skeletal material

on the basis of anatomical criteria alone. In addition, environmental factors such as

10

health and socioeconomic status would also need to be documented and accounted for in
the analysis.

Two skeletal populations ﬁom London appear to meet the criteria for a uniquely
informative study of skeletal asymmetry. The research program which forms the basis of
this dissertation is designed to study pattems of postcranial metric variation patterns
among Georgian-era skeletons which had been interred in the crypts of two London
churches: St. Bride’s (Fleet Street) and Christ Church (Spitalﬁelds). These two crypt
populations are unusually well-documented, as well as being essentially contemporary
and spatially local. A twentieth century American population, culled from the Forensic

Data Bank at the University of Tennessee, provides a further basis for comparing

asymmetry patterns.

W

Chapter 2 reviews current and historic literature relating to quantitative
assessments of long bone morphology—in particular, issues surrounding studies of
skeletal asymmetries. Chapter 3 addresses the nature of skeletal reference populations in
general, and then focuses on the speciﬁc skeletal series that form the basis for the study.
The description of the two core populations, the crypt interments from St. Bride’s (Fleet
Street) and Christ Church (Spitalﬁelds) are placed in context by a discussion of
eighteenth century London life. Chapter 4 describes the particular measurements which
were taken on the skeletal samples, and also describes the strategy by which the study
hypotheses are tested. The speciﬁc hypotheses of this study are presented at the end of

the chapter. Chapter 5 presents the results of the quantitative analyses and hypothesis

ll

testing. Chapter 6 discusses some of the implications of the ﬁndings. In addition, it
outlines a number of interesting ﬁndings related to the study which were unexpected, but
which may provide important direction for future research. Finally, it summarizes the
major conclusions of the study and suggests avenues for ﬁrrther research with these

p0pulations.

CHAPTER 2—THE PROBLEM

The past ﬁfteen years have seen an increase in studies focusing on asymmetry in
the diaphyseal geometry of human long bones (for example, F resia M. 1990; Ruﬂ’
1992; Ruﬂ‘and Hayes 1983a, 1983b; Ruﬂ’ and Jones 1981; Trinkaus e131. 1994). These
studies interpret asymmetry patterns, particularly of the humerus, to address research
questions generated by the biocultural approach to human osteology. A handﬁrl of
studies from earlier hr the century acknowledged the presence of asymmetry in the linear
dimensions of long bones. This earlier research was not driven by a biocultural
perspective, resulting in a gap in the literature regarding the biocultural signiﬁcance of
asymmetry patterns in the linear dimensions of long bones.

Researchers of diaphyseal asymmetry have asserted that patterns of variation may
be inﬂuenced by population variation, sex, and physical activity, and the same may well
be true of linear dimension asymmetry. However, because there has been no standard
protocol for reporting patterns of linear dimension asymmetry there is no basis for studies
that compare asynunetry patterns across populations. Comparative studies may reveal
variation based on sex or physical activity, among other factors; once these factors are
documented then it would be possible to construct research programs which explicitly
address biocultural questions.

Before asking the biocultural questions researchers must ﬁrst address these more
basic questions about asymmetry in the linear dimensions: Are there documentable sex-
related patterns to asymmetry in the dimensions of long bones? Do patterns of

dimensional asymmetry in long bones vary widely among disparate populations, and are

12

13

they consistent among related populations? Is there evidence that long bone length
asymmetry is associated with activity patterns?

This dissertation seeks to answer these questions through the analysis of long
bone asymmetry patterns among related populations and between unrelated populations.
The answers to these questions will set the stage for future studies of linear asymmetry
with an explicitly biocultural focus. They will also provide a methodological basis for
future comparative studies of linear dimensional asymmetry in a variety of human
skeletal populations. This chapter reviews issues surrounding the phenomenon of
bilateral asymmetry in long bones. The following chapter describes the populations

selected and the rationale for their inclusion in this study.

WW

Use of bilateral asymmetry to address structure/function questions has several
inherent advantages, including control over total body size, systemic
physiological environmernt (e.g., diet, hormone status), and various life history
variables (e.g., age, general activity level, past disease stress, etc.). (Roy 9131,
1994:203-4).

In this statemernt Roy £1.31. succinctly summarize the value of postcranial
asymmetry analysis as a tool for osteometry, a tool that is particularly appropriate for the
biocultural approach to osteology. However, assessing patterns of bilateral asynunetry
among skeletal dimensions is a complicated proposition. Linear osteometric dimensions
vary only in terms of magnitude, but there is an additional discrete aspect of direction in
the asymmetry of any paired skeletal eiernents. There is no single statistical strategy that
ﬁrlly characterizes both aspects; this problem has led to wide variation in how postcrarniai

asymmetry is assessed and reported in the literature, hindering attempts at comparing

14
work involving different studies. In addition, there are practical diﬁculties with using
assessrnernts of postcranial asymmetry as analytical tools: the deﬁnition of asymmetry,
the measurement of skeletal material to determine the extent of asymmetry, and the

interpretation of asymmetry as a biological phenomenon.

Defining Asymmetry

Because it is highly unlikely that any pair of long bones is truly symmetrical in
any linear dirnernsion, attributing labels of symmetrical or asymmetrical to paired
biological structures is dependent on the precision with which they are measured. In
practical terms this means that any apparent symmetry in paired structures results
essentially ﬁom a lack of precision in measurement technique. This is an important
methodological issue since any reference to paired skeletal elements as being
symmetrical is essentially a designation based on the limitations of the measurement
instrument; if one were to remeasure the bones with increasingly higher levels of
precision, then the prevalence of symmetry would be reduced. That is, there is less
likelihood that both elements of a bone pair would produce the same measurement value
when humerus length is measured to the nearest tenth of a millimeter, rather than to the
nearest rrnillirneter; in the former instance, a smaller proportion of paired bones would be
labeled as symmetrical.

A factor that complicates the assessment of osteometric asynnrnetry is the
diﬁculty in distinguishing actual asymmetry ﬁ'om apparent asymmetry which may be
attributable to such factors as variations in measurement techrnique or measurement error.

It is particularly important to distinguish between inaccuracy due to measurement error

15

and the seemingly random variation associated with the phenomenon of ﬂuctuating
asymmetry, which is associated with organisms subjected to high levels of physiological
stress.s .
One strategy for reducing the possibility of inaccuracy is to increase the tlnreshold
for making the distinction between symmetry and asymmetry. The standard osteometric
board, the tool with which long bone lengths are measured, is calibrated in increments of
one millimeter, as a result, a one-millimeter threshold might be inferred for the
distinction between symmetrical and asymmetrical paired bones. One might consider the
scenario wherein the length of a left humerus is recorded as 310 mm and a right humerus
is recorded as measuring 311 mm in length. It is possible that both bones are really
closer to 310.5 mm in length, and the recorded diﬂ‘erence can be attributed to
intraobserver variation. Ifthe threshold for asymmetry were one millimeter, then the
bone pair would be nnistakenly labeled as asymmetrical. However, if the asymmetry
threshold were increased to two millimeters, that bone pair would be considered
symmetrical. This would diminish the intraobserver variation in decisions concerning
asymmetry, provided that measurements are performed with consistent technique.
Unfortunately, this strategy also risks masking actual asymmetry that is small in
magnitude. Table 1 presents a series of observations for two hypothetical samples of
paired humeri. In both samples, the length of the left humerus is subtracted from the
length of the right, and the direction and magnitude of the asymmetry are recorded as

signed values. That is, if the left humerus is 310 mm in length and the right is 311 mm

 

’Fluctuating asymmetry is perhaps the form of asymmetry most widely reported
in the literature, and is discussed below in the section on “Interpreting Asymmetry”.

16

long, the signed magnitude is 1 m; if the leﬁ humerus were the longer of the pair, the

signed magnitude would be -1 m. Table 1 lists the number of individuals ﬁom each

Table l Asymmetry Distributions: Normal vs. Skewed (Signed Values)

 

 

(Right - Left) .4 -3 -2 -1 o 1 2 3 4 s 6
(mm)

Normal 0 2 6 12 15 12 6 2 o o o

Skewed o 2 3 6 9 12 14 13 no 5 2

 

sample who exhibit a given signed magnitude of asynunetry. The ﬁrst sample has a
normal distribution and the second sample is skewed.

The Table demonstrates that if asymmetry has a normal distribution with a mean
value of zero, then changing the threshold from one millimeter to two would draw
equally from both the right-dominant individuals and the left-dominant individuals, both
in terms of the number of individuals and the proportion of the study sample being
shifted. Using the ﬁgures in the sample data Table, the shift would result in an increase
ﬁ'om ﬁfteen to thirty-nine symmetrical individuals. However, if the distribution pattern
of the asymmetry is skewed, then a shift in the threshold will draw a geater mm; of
individuals from the more dominant side into the symmetrical category. In the sample
data, this means a shift of twelve individuals ﬁom the more dominant side and six ﬁ'om
the less dominant side. At the same time, the shift draws a greater 12132122111911 of
individuals from the less dominant side into the symmetrical category. In the case of the

sample, the six individuals ﬁ'om the less dominant side represent 55% of the lett-

l7

dominant cases; the twelve from the more dominant side represent 21% of the right-
domirnant cases. Thus, the researcher’s choice of a threshold value may have a
signiﬁcant eﬂ’ect on the appearance of asymmetry in a population if the actual
distribution is skewed. There are practical implications in this distinction because past
studies have indicated that the distributions of length asymmetry in the human humeri

and femora are consistently skewed.

Table 2 Asymmetry Distributions: Normal vs. Skewed (Unsigned Values)

 

 

 

0 1 2 3 4 5 6
Normal 15 24 12 4 0 O O
Skewed 9 18 17 15 10 5 2

Table 2 displays the same distribution scenario, but in this case the ﬁgures
represernt the magnitude of asymmetry alone, without direction being taken into account.
In this case the mean value for the normal population is 1.09, in contrast with 2.29 for the
skewed population. If the signed values are used to calculate means, the mean for the
normal sample is zero, and the mean for the skewed population is 1.82. Shiﬁing ﬁom the
use of signed to unsigned values results in a geater increase in the calculated mean for
the normal distribution than for the skewed distribution.

The points raised in the preceding paragaphs are neither profound nor new, but
they indicate that the strategies that researchers choose to employ in deﬁning asymmetry

in long bones can have a signiﬁcant impact on the results they report. They also

l8

underscore the diﬁculty of making comparisons between studies in which researchers

apply diﬁ’erent techniques for evaluating asymmetry in paired skeletal elements.

Measuring Asymmetry

In 1936 A Heinrich Milnter published a comprehensive review of the existing
literature regarding comparative studies of long bone lengths among humans. Mi'mter
lamented the diﬁculties he conﬁ'onted in assessing long bones from archaeological
contexts, as well as the problems that arise when he compared his ﬁndings with those
reported by other researchers. For example, Manter recognized that the summary
statistics he reported might not be those which would prove most useful to ﬁrture
researchers, so he published his raw measurement data as well. This strategy allows
contemporary researchers to apply their own statistical techniques to the data. The
methodological issues that Mi'rnter identiﬁed oﬂ’er insight into the diﬂiculties irnlnerent in
designing a research progam involving an osteometric analysis of postcranial skeletal
material. His observations also provide a basis for a critical review of more recent
studies of postcrarnial osteometry.

Miinter’s research was designed to assess the lengths of the six major long bones
of Anglo-Saxon irndividuals from a number of a skeletal collections in the Urnited
Kingdom In the majority of cases he determined bone lengths by “obtaining the
maximum separation between the ﬁxed and movable vertical surfaces of the
[osteometric] board making contact with opposite extremities of the bone” (1936:260).
In fact, Munter’s method parallels that which is recommended irn the University of

Tennessee’s “Data Collection Procedures for Forensic Skeletal Material,” which is the

19

reference document for the Forensic Data Bank at the University of Tennessee (Moore-
Jansen arnd Jantz 1989). It is also the method reconnrnended by standard osteological
reference texts (Hrdliéka 1939; Bass 1987).

One nnight easily assume that all researchers would assess the maximum length of
long bones in the same way. However, Munter found that methods could vary
considerably among researchers. The interobserver variations he noted fall into two
general categories: (1) the positioning of the bone on the osteometric board for taking
the length measuremernt; and (2) the use of speciﬁc landmarks on bones to measure
length—landmarks which do not necessarily correspond to the “extremities of the bone”.

The manner in which a bone is postured on the osteometric board can inﬂuence
tine resulting measurement if a researcher does not shift the bone adequately to determine
the true maximum length. Indeed, Milnter himself chose to deviate ﬁ'om his stated
method when he determined the maximum length of the femur. Speciﬁcally, he chose to
measure femur length with the bone resting on the horizontal surface of the osteometric
board. That is, he moved the bone ﬁom side to side on the board, but not up and dowrn,
to determine the maximum length. As a result, the value that he recorded was slightly
smaller than the actual maximum length of the bone.

In describing his methods, Munter did explicitly describe his modiﬁed technique
for measuring femur length. However, his choice to iderntify the measurernernt as
“maximum length” was misleading, since other investigators might easily assume that
Munter was referring to the actual maximum. This may lead to spurious comparisons
with other data sets if investigators were to examine the Munter’s raw data tables without

ﬁrst consulting his narrative description of measurement technique. Mi'rnter’s discussion

20
of his modiﬁed method suggests that he chose the modiﬁed femur measurement to

maintain consisterncy with what he saw to be the predominant technique listed irn the
literature at the time. While Milnter’s choice of techrnique might seem rational in that
context, that fact that it is at variance with what is considered the standard technique
today underscores the problem of ambiguous standards for measurements in the
literature.

A recernt example ﬁ'om the literature reaﬂirms the persistence of such
methodological issues in long bone measurements. Jantz e111, (1994) report that the
measurement techniques employed by Trotter in detemnining maximum length of the
tibia to construct stature formulae were remarkably inconsistent. Even though her
narrative description of her technique unambiguously indicates that the medial malleolus
is to be included in the measurement, in fact her reported values appear to have been
derived from measurements that exclude the malleolus. The inconsistency uncovered by
Jarntz ELIL is particularly unsettling, because Trotter’s description of measurement
techrniques was so clearly spelled out in her reports—and yet they are clearly in conﬂict
with her published data

The work of Adolph Schultz (193 7) provides another example of how postcranial
measurements may be assessed differently by diﬂ‘erent researchers. Schultz determined
his length measurements using speciﬁc anatomical landmarks which might appear
counterintuitive to a researcher interested in the maximum lengths of long bones, and
which are not consistent with the standards listed in the standard reference texts of human
osteology. In humarns, the maximum length of the humerus extends ﬁ'om the most

proximal aspect of the humeral head through the tip of the trochlear projection at the

21

distal end of the bone. When Schultz measured humerus length, he consistently
irnterpreted the furthest extension of the capitulum as the most distal point on the bone.
In much the same way, Schultz measured the length of the femur from the most distal
aspect of the lateral epicondyle to the most superior projection of the geater trochanter.
In humans, the true maximum length of the femur is measured from the head of the femur
to the lateral epicondyle (Bass 1987). As a result, Schultz’s measurements were less than
the actual maxima.

Like Munter, Schultz unambiguously described his measurement techniques.
Unlike Milnter, Schultz took care not to use the term maximum to refer to his length
measurements. It is important to note that Schultz’s choice of landmarks was not
arbitrary, but rather driven by his research problem. The goal of his study was to
compare the length of human bones with those of chimpanzees, gorillas, orangutans,
siarnangs, gibbons, and nnacaques. In these non-human primates, the landmarks he
employed are coterminous with the extremities of the long bones. As a result, he was
compelled to use consistent measurement landmarks among the diﬂ’erent species.
Researchers reviewing Schultz’s ﬁgures, however, would risk making inaccurate

comparisons if they did not note the modiﬁcation in his measurement technique.

Interpreting Asymmetry
Even if researchers could agee on a standard technique for measuring linear
dimensions of long bones, such as the guidelines set out in Moore-Jansen and Jantz

(1988), they still have to contend with the problem of interpreting the results of their

observations in a consistent manner. For example, researchers typically have chosen to

22

focus on only the magnitude of asymmetry, and not its direction. One possible reason
why direction of asymmetry has not been well studied to date is the tradition of using
paramaric statistics to assess the signiﬁcance of osteometric variation. Because the '
direction of asymmetry is a nominal variable, testing its statistical signiﬁcance requires
the use of nonparametric techniques. In contrast, asymmetry magnitude has generally
been assessed using parametric tests, such as t-test comparison of sample means.
However, the small magnitude of asymmetry coupled with the small sample sizes which
are typical of most skeletal populations reduces the likelihood of generating a statistically
signiﬁcant refutation of a null hypothesis on the basis of parametric techniques.

Theoretically, asymmetry in bilateral structures can occur in one of three basic

Table 3 Examples of Three Fundamental Asymmetry Patterns.

 

(Right - Left) -4 -3 -2 -1 0 1 2 3 4
(mm)
Non-directional 1 2 5 8 10 8 5 2 1
Directional 0 1 2 5 8 10 8 3 1
Anti-symmetric 2 4 6 4 2 4 6 4 2

 

_
patterns: non-directional, directional, and anti-symmetric. The distribution pattern of
non-directional asymmetry resembles a typical bell-Shaped curve; it is unimodal and
normal, with a mean value of zero. Directional asymmetry is typically also unimodal, but
the curve is shifted such that the mean value is not at zero; in most cases a directional

distribution is not normal but skewed. An anti-symmetric distribution is bimodal, with

23

one peak located on each side of the zero value. Table 4 contrasts these three asynunetry

patterns, with the top row indicating the difference in size between the right and leﬂ sides

of a paired skeletal element (like humerus length, for example).

Fluctuating Asymmetry

Most published studies of morphological asymmetry focus on the phenomenon of
ﬂuctuating asymmetry (Van Valen 1962). Fluctuating asymmetry is understood to
reﬂect the eﬂ‘ects of physiological stress (nutritional stress, environmental stress, etc.) on
bilaterally symmetrical body elements (Parsons 1990). In theory, orgarnisms that are
subjected to increased levels of physiological stress will display an increased range of
morphological variation. This means that there is an increased likelihood that within
paired elements a feature on one side (buccal-lingual diameter of a tooth, for example)
would manifest a diﬂ‘erence in size ﬁ'om its partner. Within individuals subjected to
higher levels of stress there is a concomitant increase in the likelihood that the size
diﬂ‘erence between the two sides will be geater as well. As a result, asymmetry
magnitude would be relatively high in a population of organisms that has been subject to
higher levels of stress.‘ It is important to note that a deﬁning characteristic of ﬂuctuating
asymmetry is that it is non-directional. That is, there is equal likelihood that either the

right-sided or the left-sided element would be the larger of the pair.

 

6Mailer and Pomiankowski (1993) suggest that the presence of increased levels of
ﬂuctuating asymmetry in an evolutionary population may be an indicator of rapid
evolutionary change. Drawing on the theoretical phenomenon of punctuated equilibriunn,
they argue that rapid evolutionary processes render organisms more susceptible to stress,
and hence more likely to display patterns of asymmetry.

24

The presence of ﬂuctuating asymmetry in humans has received much attention
over the past two decades, particularly in studies of human dernnatoglyph patterns
(Markow and Martin 1993). Fluctuating asymmetry has been demonstrated in non-
human postcranial skeletal morphology (reviewed in Livshits and Kobylianski 1990), but
research involving human skeletal material has traditionally focused on the dentition
(Bailit £111. 1970; Perzigian 1977, Hershkovitz e111. 1993). Recent studies of dental
asymmetries have involved prenatal exposure to tobacco smoke (Kieser and Groenevel
1993) and maternal consumption of alcohol (Keiser 1992). Manning and Chamberlain
(1994) have also studied the relationship between asymmetry in the canines of lowland
gorillas and environmental stressors associated with the degadation of their habitat.’

Trinkaus e131, (1994) suggest that linear asymmetry in humerus length may be
ultimately attributed to the eﬂ'ects of ﬂuctuating asymmetry. However, the fundamental
nature of ﬂuctuating asymmetry is inconsistent with it being presented as a
comprehensive explanation for humerus length asymmetry in humans. By deﬁnitiorn,
ﬂuctuating asymmetry is a non-directional characteristic, and yet virtually all studies
(Including those cited in Trinkaus e111.) report asymmetry in humerus length as highly
directional in nature. That is, most studies of skeletal samples report a much geater
prevalence of right-dominance in the length of the humerus compared with left-
dominance. Therefore, while there may be a ﬂuctuating asymmetry component to

humerus asymmetry patterns, there must be other factors operating as well.

 

1Some researchers have criticized the methodology associated with quantifying
ﬂuctuating asymmetry (Smith £111. 1982 ), and with the diﬁculty in distinguislning
between measurement error and actual ﬂuctuating asymmetry in odontometric and
anthropometric observations (Fields M. 1995).

25

Table 4 Humerus and Femur Asymmetry Direction (Lowrance and Latimer 1957)

 

 

Humerus Femur

Left-dominant 17.2% 50.5%
No difference 9.5% 19.0%
Right-dominant 73.3% 30.5%

 

Very few researchers have addressed the issue of asymmetry in linear dimensions
of human limb bones. In 1957 Lowrance and Latimer assessed the length of humeri and
femora in a series of 105 skeletons, identiﬁed as Asian in origirn, which had been
purchased from a commercial supplier for student use. Table 4 sununarizes their
ﬁndings, with the ﬁgures representing the percentage of the sample in each of three
categories. The authors did not explicitly state the criteria by which they made the
distinction between symmetrical (i.e., No diﬁ‘erence) and asymmetrical bone pairs.
Because length of mq'or long bones is typically measured on an osteometric board
gaduated at one-millimeter intervals, it is likely that bones with the same measurement
were labeled as synunetrical in their study-that is, with precision to the nearest
millimeter.

In 1965, Latimer and Lowrance reassessed the same Asian adult skeletons ﬁ'om
their 1957 publication, this time to compare the relationship between symmetry in the
humeri and femora. They found that ﬁfteen individuals showed no remarkable level of

asynunetry, but that 50% of those which showed asymmetry manifested a pattern called

crossed symmetry (Table 5).

26

Table 5 Crossed Symmetry in Humeri and Femora (Latimer and Lowrance 1965)

 

 

Left Humerus Right Humerus
Left Femur 8.9% 47.8%
Right Femur 2.2% 41.4%

 

 

The concept characterizes a pattern of asymmetry wherein upper limb bones are larger
on one side (typically the right) at the same time that lower limb bones are larger on the
opposite side (typically the left). The term itself was ﬁrst introduced into the literature
by Schaeﬂ’er (1928), and the phenomenon has received brief comment by researchers
over the subsequent decades (Singh 1970; Chhibber and Singh 1972). However, the
extent of crossed symmetry has not been well documented among human populations,
nor has there been comment on whether there are any sex-related patterns of crossed
W097-

There is a signiﬁcant gap in the literature surrounding an important series of
questions: Is the directionality of asymmetry that is reported in population means the
result of broad and regular patterns of asymmetry across a population? Alternatively,
does it reﬂect a high level of asymmetry among a small subgoup of the population, with
the geater majority being essentially symmetrical? At ﬁrst glance, these questions nniglnt
appear to be somewhat trivial. Their importance, however, lies in the testing of
assumptions related to the nature of skeletal asymmetry in general.

An obvious suggestion is that asymmetry in long bones is linked to differential

levels of physical activity on the two sides of the body; a recent review is found in

27
Stirland (1993a; see also 1993b). In his comparison of limb asymmetry patterns in
humarns and martens, Pierre Jolicoeur (1963) ackrnowledges the theory that “the geater
stability of bilateral synunetry in external organs is interpreted as a locomotor adaptation:
the symmetrical development of limbs and sense organs would make it easier for an
animal to reach its goal directly” (1963 :410). In another often-cited study in the limb
asymmetry literature, Buskirk e111. (1956) demonstrated that unilateral physical activity
in the upper limb of tennis players is associated with asymmetrical development of bone
as well as muscle. In humans, then, this type of activity would result in relatively
symmetrical lower limbs; the upper limbs, however, would not require the same levels of
symmetry, and unilateral activity would result in geater levels of asymmetry in upper
limb bones. It has been found that asynunetry in the length of the humerus is generally
geater than asymmetry in the femur (Schultz 193 7), but the relationslnip between activity
and asymmetry is not straightforward, and has been subject to debate irn recent years.
Ruﬂ’and Jones extended this cautionary note followed their assessment of Native
American remains ﬁ'om Califorrnia wherein they observed apparent sex-related
diﬂ‘erences in cortical asynunetry of the tibia and humerus:
the evidence for a direct activity-bone response explanation for bilateral
asymmetry is somewhat conﬂicting. One of the problems in evaluating diﬂ‘erent
hypotheses in this area has been the lack of data on relevant bone dimensions in a
normal unselected population sample. . . . Another area which has not been
systematically investigated is the effect of physiological factors other than activity
levels, such as sex and age, on bilateral asymmetry. (Ruﬂ‘ and Jones 1981 :71)
They were not the ﬁrst to suggest that there was a component to long bone

asymmetry that was not attributable to physical activity. Hrdliéka, too, observed sex-

related variation in humerus length:

28

[Another] point, both striking and quite new, I believe, is the behavior of the
bones on the two sides. The relation of the female to the male humerus on the
two sides of the body shows that in all the goups of the whites, and equally in the
Indians and the negoes, the left female bone is on the average and relatively to
themale, shortertlnantheright. aharmonyindetails snrchasthesegoesfar
toward convincing one of the fundamental unity of the human species. (Hrdliéka
1932:43 7).

It is obviously very difﬁcult to isolate the eﬁ‘ects of physical activity ﬁ'om other
features that aﬂ'ect the deveIOpment of skeletal limb asynunetries. For example,
Baskerville (1992) has suggested that asymmetries in the forelimb may be associated
with asymmetries in the main trunk of the body—and Helrnkamp and Falk agee “that
sex hormones probably play a signiﬁcant role in causing asymmetrical development”

(1992:498).

WW

Published studies of asymmetry in the humerus and femur of humans (Trotter and
Gleser 1952, Laubach and McConville 1967, etc.) are in general ageement on tlnree
points. First, the length of the major upper limb bones show a geater magnitude of
asymmetry than lower limbs. Secondly, on the upper limb the right humerus is generally
larger than the left and, thirdly, on the lower limb, the left femur is generally larger than
the right. One intriguing exception to this pattern is reported in Graham and Yarbrough’s
(1968) study of the Shell Mound population, where they found a consistent pattern of
longer lS‘ﬂ humeri and longer ﬁght femora in both males and females.

These conclusions are based on comparing mean values of asymmetry
observations witlnin skeletal series, and are presented with little if any discussion of their

nnearning, except perhaps for a reference to Schaeﬂ‘er (1928). Ifthese patterns have

29
signiﬁcance in a biocultural context they have not been addressed. Instead, recent years

have seen an apparent reduction in the number of straightforward osteometric studies of
human postcranial skeletal material. They appear to have been overtaken by more '
technologically sophisticated metric studies designed to explore the lifeways of
individuals represented by the skeletal material. In the past two decades, reference to
asymmetries in studies of long bone morphology have addressed biocultural hypotheses
by focusing on geometric properties of the diaphyses. While contemporary authors (for
example, Bridges 1989; Fresia et al. 1990) may make a passing reference to asymmetry
irn bone lengths, they generally dismiss the phenomenon as inconsequential to their
biocultural analyses.

Ruﬂ‘ and Jones (1981) provide one of the ﬁrst attempts to investigate age- and
sex-related patterns of bilateral asymmetry in cortical bone. They assessed paired humeri
and tibiae from seventy-nine archaeological specimens ﬁom Californnia. Based on
skeletal criteria, the authors divided their adult study population into four goups on the
basis of age and sex; that is, older males, younger males, older females, and younger
females. They found males showed a good deal more asymmetry than females, and that
cortical bone area decreased with age.

The phenomenon of bilateral asynunetry in the morphology of long bones has
recently become popular as a strategy for addressing biocultural questions, as reviewed
by Ruﬂ'(l992). One particular study by Fresia M. (1990) exempliﬁes the biocultural .
approach to studies of humerus morphology. Speciﬁcally, Fresia 911], compared
bilateral asynunetry of the humeri in three temporally discrete goups ﬁ'om the Georgia

coast (see Table 6). They found that bilateral asymmetry in humerus length increased

30

over time, while asymmetry in humerus strength (based on cross-sectional biomecharnical
analysis) decreased. The authors suggested that the diﬂ‘erence in strength was related to

changes in side-dominant activity patterns (but did not speculate on why the length of the
humeri varied).

Their study population consisted of ﬁfty-one individuals from ﬁve sites in
Georgia The authors assessed the humeri of ﬁfty-one individuals which they partitioned
into three categories: Precontact preagicultural, Precontact agicultural, and Contact.
Their study samples are small, and it is somewhat speculative to draw conclusions ﬁ'om
such small numbers, but the limits of sample size is a problem which commonly arises in
analyses of skeletal remains.

Fresia eLaL characterized length asymmetry of the humerus by use of Equation 1,

one of several which have been applied to the studies of asymmetry. This particular

 

Amway = (100 )[R'gzighf‘ﬁJ (1)

equation characterizes asymmetry in percentage terms, using the right humerus as a
baseline. Ifboth humeri are the same length, then the asynunetry value is zero. Ifthe
right humerus is geater in length than the left, the value is positively signed, and if the
left humerus is longer the value is negatively signed.

Using this formula, Fresia :13], reported that asymmetry in the preagicultural
goup was less than 0.5% for both males and females, and that the same held true for the
females in the precontact agicultural goup. In contrast, they found that the males of the

precontact agicultural goup, as well as both the males and females of the Contact goup,

31

Table 6 Changes in Humerus Asymmetry Over Time (ﬁom Fresia e131, 1990)

 

Male Female

Precontact Preagricultural < .5 % < .5 %

Precontact Agricultural l % < .5 %
Contact l % 1 %

 

 

presernted approximately 1% asymmetry in humerus length (Table 6).

The ﬁndings for humerus length asymmetry reported in this Table exemplify the
ambiguity of summarizing asymmetry patterns in terms of population means. It is
unclear whether a stated mean asymmetry value of .5% means that the riglnt humerus was
longer than the left in all cases, or whether there were some pairs where the left humerus
is longer than the right. In the latter instance a few left-dominant pairs, being negatively
signed, would substantially reduce the mean asymmetry value for the population ﬁ'om
what it would be if the absolute (unsigned) values of asymmetry magnitude were
assessed. To date there have been no studies which have addressed the issue of bilateral

asymmetry in linear dimensions of long bones at the level of the population.

CHAPTER 3—THE STUDY POPULATION S

W

The term population has several meanings in the context of skeletal biology. At
the most elementary level, the individuals who comprise a skeletal collection or a
subgroup of a skeletal collection are collectively referred to as a population (Steegnan
1991; Waldron 1991; Whittaker and Hargeaves 1991). The larger goup of living
persons whom the skeletal material represents is a population in a more restricted and
more analytically valuable sense. That is, inferential statistics can be applied to
observations taken ﬁom the sample of the larger population to draw inferences about the
nature of the population as a whole (Weiss and Hassett 1982). These conceptions are
distinct ﬁom the population genetics deﬁnition of a population as “a local or breeding
goup; a goup in which any two individuals have an equal probability of mating with
each other” (Campbell 19922539).

One cannot assume that a skeletal sample necessarily corresponds to a population
in the latter sense of the term. As Wood e131. (19922344) warnn, “There is one, and
perhaps only one, irreﬁrtable fact about the cases making up a skeletal series: they are
dead.” Nonetheless, osteologists commonly assess skeletal series as a collective whole
when they engage in osteometric analyses, and draw conclusions about the p0pulations
that they supposedly represent.

In general ternns, analytical osteometry research of populations is directed toward
one of two goals. The ﬁrst of these is to examine a known reference series in order to
construct a set of standards or guidelines which are meant to be applied to future

investigations of other skeletal material. The second goal is to assess an unknown

32

33
individual (or population) in terms of the standards which have previously been derived

ﬁom a reference source, in order to come to a better understanding of the unknown(s).

AS noted in the previous chapter, estimating stature on the basis of long bone
lengths illustrates these complementary goals of contemporary osteometry. In a number
of studies, researchers have collected measurements from suﬁciently large skeletal series
for which living stature is known (for example, Trotter and Gleser 1951; 1952; 1958).
Based on their observations, the researchers have derived formulae which reﬂect the
relationship between bone lengths and stature. These formulae are then applied to
measurements taken from bones belonging to an unknown individual in order to
determine living stature—within a reasonable margin of error.

The assessment is not quite as straightforward as this simpliﬁed description
suggests. For example, males and females ﬁ'om a single homogeneous population tend to
show diﬂ’erent patterns of relationship between bone lengths and living stature.
Researchers typically contend with this problem by constructing two different series of
stature formulae—one for males and another for females. Likewise, persons of diﬂ’erent
ancesz ternd to show dissimilar patterns of relationship between bone length and stature.
As a result, stature formulae for diﬁ‘erent “racial” goups have been derived and they
feature pronninernﬂy in skeletal biology reference volumes.

In addition to stature estimation, another common application of analytical
osteometry is the determination of an unkrnown individual’s sex from diﬂ'erences in linear
dimensions of long bone elements. In humarns the long bones of males are, on average,
larger than females; this fact has been used by several researchers to construct univariate

and multivariate discriminant ﬁrnctions for sex estimation. Bone lengths, head diameters,

34

and biepicondylar widths of the humerus and femur have all been used to assign sex to

skeletal material (Dittrick and Suchey 1986; France 1988).

Table 7 Attributing Sex from Femur Head Diameter (adapted ﬁom Bass 1987)
_
Diameter (mm) Female Female? Indeterminant Male? Male
Pearson <41.5 41.5-43.5 43.5-44.5 44.5-45.5 >45.5
Stewart <42.5 42.5-43.5 43.5-46.5 46.5-47.5 >47.5

 

 

Diﬂ'erences between populations affect the attribution of sex to skeletal remains,
just as they aﬁ’ect stature estimation. Bass’s (19872219-20) ﬁeld manual acknowledges
the population diﬂ’erence phenomenon by reporting Pearson’s (1919) “Rules for Sexing
the Femur” on the basis of femoral head diameter, as well as Stewart’s (1979)
modiﬁcation of Pearson’s ﬁgures for sexing “American Whites” (Table 7).' The
diﬂ’erence between Pearson’s and Stewart’s ﬁgures can be attributed to the diﬂ‘erent
“White” populations they use as their sources. Pearson used seventeenth century
Londoners as his reference series while Stewart used a series of specimens from the
Terry Collection in the Urnited States as his reference. The geatest certainty that a
stature or sex estimation is accurate occurs when the unkrnown individual is drawn ﬁom
the same population as the reference source. In other words, a researcher asked to
attribute sex to a femur ﬁom a 17th century London plague pit would more wisely

employ Pearson’s standard than Stewart’s.

 

'The ﬁgures Bass attibutes to Pearson are taken ﬁom Pearson and Bell (1919).

35
Unfortunately it is rarely the case that there is such a close match between skeletal

unknowns and available reference series. Therefore, researchers choose the next best
strategy by assessing an unknown individual with reference to skeletal standards derived
ﬂour a series that is most representative of the living population associated with the
individual being investigated. Thus, if a femur to be sexed were from a 20th century
forensic case, then it would be more appropriate to employ Stewart’s standard than
Pearson’s. Likewise, a femur ﬁom 18th century France would be best assessed using
Pearson’s ﬁgures. In either case, since any reference skeletal series is limited in its
ability to be broadly enough representative of the larger population from which it is
drawn, there will always be a measure of uncertainty in the assessment.

The limited number of available reference series also complicates comparative
approaches to osteometry. Over the past century published research of human long bone
morphology has involved a considerable range of skeletal series. The osteological
collections which fornn the basis for the research vary widely in terms of numbers of
individuals represented, level of supporting documentation, preservation quality, and a
number of other factors. These diﬁ‘erences inﬂuence the choice of the questions that can
be addressed by studying a given skeletal collection, as well as the explanatory power of
the results of an analysis. AS a result, it can be diﬂicult to make direct comparisons

between the ﬁndings of different research progarns.

Reference Series Populations
Traditionally, skeletal biologists have drawn on ﬁve types of reference series for

skeletal analysis: (1) anatonnical collections; (2) military dead; (3) undocumented

36
prehistoric and historic archaeological series; (4) documented historic archaeological
series; (5) living clinical populations; and (S) forensic cases. Each of these groups
possesses a series of characteristics that make them appropriate for comparative
osteometric studies; at the same time, each has other characteristics which make them

less suitable for making legitimate comparisons.

Anatomical Collections

Many osteological and osteometric standards in the United States have been
derived ﬁom research involving two anatomical series: the Terry Collection which is
housed at the National Museum of Natural History, and the Hamann-Todd Collection
located at the Cleveland Museum of Natural History (Stewart 1979:84). These particular
skeletal series appeal to researchers because they consist of a relatively large number of
individuals and the skeletal material itself is typically well preserved. Perhaps most
importantly, the individuals in these collections are well documented with respect to age
and sex.

On the other hand, the individuals which comprise the collections are not
representative of a well—deﬁned living population. Their lifeways are not well-
documented, insofar as their diet and physical activity patterns are understood only in
general terms. In many cases the skeletons represent indigent members of society who
were relegated to anatomical collections; as such, they may not be adequately

representative of the larger society from which they are drawn (Ericksen 1982).

37
Military Dead

Several of the most commonly cited studies which estimate human stature
resulted ﬁ'om research performed on the skeletal remains of American war casualties
(Trotter and Gleser 1951, 1952, 1958; Trotter 1970). Like the anatomical collections, the
number of individuals in these collections is relatively large. Also like the anatomical
collections, the individuals themselves are well documented with respect to sex and age
at death. In contrast with the anatomical collections, military dead oﬁ‘er the additional
advantage of being particularly well documented in other ways. The living stature of
military personnel, for example, is recorded as part of a documentary record which is
atypical of most skeletal samples.

Perhaps the most obvious disadvantage of military dead as a reference skeletal
series is that it represents a relatively restricted population of living persons. All of the
individuals from these collections are men; in addition, these men represent a narrow
range of ages that is not representative of the larger population. Like the anatomical
collections the reference populations of military dead do not represent a genetically or
culturally homogeneous group of people. As a result, their value in comparative studies

of populations is somewhat limited.

Undocumented Archaeological Series
Archaeological series of undocumented individuals provide some of the largest
skeletal series available to researchers today (Graham and Yarbrough 1968). The
advantage gained by the large number of individuals represented in these series is

countered by a number of shortcomings related to their undocumented nature. For

38
example, the age and sex of individuals can only be determined by anatomical criteria,
which lack the certainty of the documentary record.

A more substantial concern for comparative studies is that, in many cases, an
archaeological skeletal series may represent a series of occupations of a given site which
took place over a period of several hundred years. During that time a number of changes
may have occurred in lifeways and migration patterns which may call into question
whether indeed the skeletal material can be said to represent a single population of
interbreeding individuals who lived at a place over an extended period of time .

Arguably the most prevalent and challenging dimculty with assessing long bones
from archaeological contexts is that the majority of the bones are alien too fragmented to
provide accurate length measurements. In cases where there might be only one or two
specimens from a population, this can be a very serious problem. Trinkaus M. (1994)
were required to contend with this issue in their assessment of humerus morphology in
early Home.

In studies of asymmetry in paired skeletal elements a study series can be reduced
signiﬁcantly in size if only a relatively small proportion of individuals retain paired long
bones which are suﬁiciently preserved to provide accurate bilateral observations. For
example, Munter’s (193 6) study of postcranial measurements on Anglo-Saxon remains
from British museum sources was based on measurements taken ﬁom 233 male and 93
female skeletons. Only 113 males and 43 females retained paired femora and 67 males
and 30 females retained paired humeri for which maximum length measurements could
be obtained. Of these, only 53 males and 18 females retained both paired humeri and

paired femora.

39

Another problem with assessing archaeological skeletal material is the ambiguity
of anatomical sexing. The problem of attributing sex to skeletal remains is a challenging
one. In those cases where entire discrete skeletons are available, the morphology of the
pelvis is typically used to assess sex, and this technique is regarded as being quite
accurate (Dittrick and Suchey 1986, Ruﬁ’and Jones 1981). In cases where the skeletal
material is commingled, such as the plague pit skeletons reported in W (Morant
1926; Hooke 1926) in the early part of this century or where an intact pelvis is not
available, long bones can only be sexed on the basis of intrinsic anatomical criteria,
which are even more subject to various degrees of error (Machughlin 1987). One way
to avoid the sexing problem is to assess only those series for which each individual’s sex
is identiﬁable by non-anatomical (i.e., documentary) criteria.

A signiﬁcant problem for analytical osteometry is the relatively small number of
females in many reference series. As noted above, the larger skeletal series which have
been used to develop standard osteometric formulae for assessing stature, for example,
are military dead and anatomical collections which consist predominantly of males. Even
in archaeological reference series there is a relative paucity of female representation.
One cause of this may involve bias in anatomical sexing techniques which over-classiﬁes
males and under-classiﬁes females (Weiss 1972).

The less than optimal preservation of skeletal material exacerbates the problem of
female under-representation. Because human males tend to have larger and more robust
skeletons than females, male skeletons are more likely to remain well preserved in harsh

environmental conditions. Mi’mter’s study series, for example, only had a relatively

40

small number of females in his sample—much less than the 50:50 ratio that would be

expected in a normal population.

Documented Historic Series

In contrast with non-documented archaeological series, historic cemetery or
plague pit samples have the advantage that the lifeways of the populations represented
are oﬁen well described by supporting documentation. Unfortunately, in the case of most
cemetery series the individuals themselves are not identiﬁed (Angel M. 1987; Lanphear
1988, 1990; Saunders e111. 1993, Sirianni 1993). In the case of plague pit or cemetery
clearance series, the skeletons of hundreds of individuals may be represented (MacDonell
1904, 1906; Hooke 1926). However, their appropriateness as reference populations is
severely restricted because bones ﬁ'om many individuals are commingled. In spite of
their commingled nature, several seventeenth century London plague pit and clearance
series form the basis for Pearson and Bell’s (1919) extraordinarily comprehensive three-

volume study of the English femur.

Living Clinical Populations
There have been a number of studies on clinical populations of contemporary
humans for which radiographs or computed tomography are used to represent skeletal
morphology. Such populations oﬁ‘er a powerful tool for contemporary research because
they are typically well documented by data ﬁom clinical records. One diﬁculty with
employing clinical populations in osteological research is that they typically represent a

pathological subset of the larger population. That is, generally healthy individuals are

41

not selected for radiographic study; for example, dialysis patients were studied by Garn :1
al. (1976) to investigate patterns of skeletal asymmetries. Another methodological
consideration is that radiographic examination is relatively expensive, which means that
relatively few groups have been studied in this way. However, in a few cases
longitudinal analyses of individuals have included the radiographic assessments of
healthy individuals, and there have been cases where specialized series have been studied

radiographically (for example, athletes and early humans in Trinkaus 91.81. (1994)).

Forensic Cases

One of the most common applications of analytic osteometry in skeletal biology
today is the assessment of skeletal material in forensic cases to identify the individual(s)
represented by the remains. It follows that the most appropriate reference source for
standards of forensic analysis should be other contemporary forensic cases. Recognizing
this, forensic anthropologists at the University of Tennessee have collected data from
forensic cases provided by their colleagues and have formed a computerized database of
these cases (Jantz and Moore-Jansen 1988). The information in the Forensic Data Bank
has already been used by Bennett (1993) to derive updated discriminant ﬁrnctions for
determining sex and ethnic aﬁiliation (see also Giles 1970).

The individuals listed in the Data Bank themselves are typically well-documented
with respect to age at death, year of birth, sex, and ethnic afﬁliation. In many cases, they
have also been subjected to a comprehensive battery of metric and noncmetric analyses.
Most importantly, the individuals in the Forensic Data Bank represent members of

contemporary American society. On the other hand, they cannot truly be said to

42
represent a single population, since they are, by deﬁnition, drawn singly from unique
locations across the United States. In addition, the lifeways of the individuals themselves
are not well documented.

One other potential diﬁculty with the Forensic data is that it consists of
measurements made by a number of researchers, so there is a possibility that differences
in measurement technique might inﬂuence some observations. The curators of the Data
Bank have attempted to mitigate this problem by explicitly stating the techniques that
should be applied in taking measurements (Moore-Jansen and Jantz 1989), but there is no

assurance that all contributors to the Data Bank follow those techniques to the letter.

London Crypt Populations as a Basis for Study

None of the types of skeletal series described here is ideal for deriving standards
for a phenomenon such as long bone asymmetry. Indeed, ﬁnding an ideal skeletal series
is problematic since those populations which have the highest level of supporting
documentation either reﬂect a small sample size or are representative of a only a limited
subset of the larger living population. At the same time, those series with the largest
sample sizes are typically those which are the least well-documented, or those which
consist of commingled skeletal material.

While there may be no ideal reference series available to osteologists, the crypt
populations ﬁ'om two churches on the periphery of the City of London—St. Bride’s
(Fleet Street) and Christ Church (Spitalﬁelds)——oﬁ‘er a unique basis for a comparative

study of long bone morphology. Not only are the churches themselves well-documented

43
by historical records, but each person represented in the sample is individually identiﬁed
with respect to name, sex, age, and year of birth.

The two churches’ crypts contain individuals who are contemporaries of the
Georgian era, with dates of death ranging from the early 17003 to the mid-18503.9 The
fact that these two skeletal populations are localized in terms of time and space suggests
that they were subject to the same environmental conditions. In addition, because crypt
interment in central London churches usually required a reasonably high level of income
or social status, there is evidence for a commonality in socioeconomic status between the
two groups (Cox and Molleson 1993; Scheuer personal communication). In short, taken
together these skeletal series oﬁ‘er a unique and valuable opportunity to engage in a
comparative study with a particularly high level of documented control over
environmental factors.

By typical standards of time and space, the two churches (and their crypt
populations) are very much alike. The most obvious similarity is their close proximity to
each other, as they are separated by a distance of only approximately 2.5 kilometers.
Each is found just outside the boundary of the City of London proper, St. Bride’s to the
west, and Christ Church to the east (Figure 1). Their crypt populations are also roughly
contemporaneous; the earliest identiﬁed crypt interment at Christ Church was in 1729,
and the latest in 1859. The earliest identiﬁed crypt interment at St. Bride’s was in 1740,
and the latest in 1852. The earliest born individual in the St. Bride’s crypt, Mr. Samuel .»

Holden, was born in 1676 and died in 1740. In the Spitalﬁelds population, Miss

 

’The two crypt populations are referred to here collectively as the “Georgians,”
even though technically the Georgian era ends in 1837 when the reign of Queen Victoria
begins. Virtually all of the individuals in both series were born prior to 183 7.

 

LONDON l

Fournler Street

 

 

 

 

 

 

 

. o
:5”: ti. W I Christ Church
" °" I Spitamelds
g London Wu! K
Fleet snug ' " a
* , St P:ul 5 Bank of
St I ' England
Bride's
River Thames
__,._ London
Budge
Scale
3 firm

 

 

 

Figure 1 Relative Location of St. Bride’s and Christ Church, Spitalﬁelds.

Susannah Hull was born in 1647 and died in 1732. Crypt interments were halted in both
churches within a few years of each other; the mandate which ended interments at St.
Bride’s was dated 1854, and for Christ Church was dated 1858.

The demographic patterns of the two populations are also similar. The adult
population from both crypts show a roughly even ratio of males to females. The
Spitalﬁelds crypt showed a substantially larger number of juveniles than St. Bride’s; this
was likely due to the number of family vaults excavated at Christ Church. Age at death
for the adults of both populations also showed a comparable pattern. In both groups,
approximately half of the population died between the age of 55 and 75 years.

Finally, there is a basis for arguing that the two crypt populations had a roughly

equivalent socioeconomic status. In Georgian London the ﬁnancial cost of interment in a

45

crypt was substantial, and the poorer members of a parish were typically relegated to the
churchyard for burial (Litten 1991). This is not always the case, though, since a few
individuals of rather meager income were interred in crypts, either because of speciﬁc

ties to the Church, or because more well-to—do members of the family were associated
with the Church. Nonetheless, the parishoners of St. Bride’s and Christ Church reﬂected
a fairly broad range of middle-class Londoners, which included shopkeepers, artisans,
weavers, and businessmen.

Since individuals are identiﬁed by name in both the St. Bride’s and the
Spitalﬁelds series, previous researchers have referred to trade and vestry records to
clarify the social and economic context in which speciﬁc individuals lived and worked
(Cox 1989; Waldron and Cox 1989; Bowman and Scheuer, personal communication).
Cox (1989) divided members of the male Spitalﬁelds population into occupational
subgroups of artisans, master craftsmen, and professionals, and others. Cox and Scott
characterize the women of Spitalﬁelds as “a middle-class group, they were largely of
high nutritional status and, by the standards of the day, lived in sanitary and comfortable
conditions” (Cox and Scott 19922431). In addition, twelve skulls from Spitalﬁelds
showed evidence of dental restorations or artiﬁcial teeth, another indication of middle-
class socioeconomic status (Whittaker and Hargreaves 1991).

Bowman suggests that the St. Bride’s males be partitioned into two main
occupational groups—one consisting of lawyers, doctors, and gentlemen; and the other
consisting of businessmen and shopkeepers. It appears that few if any artisans are

present in the St. Bride’s series, while many are available in the Spitalﬁeld’s sample; this

46
is not surprising, given that weaving was a well-known occupation in Spitalﬁelds, but not
in the area of Fleet Street (George 1965).

One way in which the Spitalﬁelds and St. Bride’s collections have already
effectively been compared is with respect to speciﬁc indicators of stress on the skeleton,
such as cribra orbitalia, and enamel hypoplasias. The Spitalﬁelds material shows some
cribra orbitalia, but none that could be regarded as severe; the St. Bride’s series shows
very little cribra orbitalia. Neither series shows vault lesions of porotic hyperostosis, and
enamel hypoplasias are uncommon. These ﬁndings are not totally unexpected, given that
London middle-class crypt populations not be likely to suffer the signiﬁcant nutritional
deﬁciencies which are commonly associated with these markers (Molleson and Bowman,
personal communication).

On the basis of these criteria it would appear that it would be diﬁcult to ﬁnd a
better matched pair than the St. Bride’s and Spitalﬁeld populations. However, there are
three reasons for considering the populations as distinct: geographic separation, the
possible occupational diﬂ‘erences discussed above, and population history. The majority
of persons in Spitalﬁelds in the early decades of the crypt sample appear to represent an
immigrant population—primarily Huguenots ﬁom France and their descendants. As time
went on, the proportion of individuals with French sumames became smaller, but it is not
certain if this was due primarily to intermarriage or to a replacement of the Huguenots
with persons of strictly British heritage (Cox and Molleson 1993). Secondly, there are no

surnames held in common between the documented persons interred in Christ Church

1m

2m

COT?

47

and those ﬁ'om St. Bride’s. '° Even the existence of common surnames would not
necessarily indicate interbreeding between the two groups, but the fact that there appears
to be no commonality would suggest that these are indeed two distinct populations.

Even if there are meaningﬁll diﬂ‘erences between these skeletal series, there is a
marked disadvantage to relying on two skeletal populations which are as closely linked in
time, space, and social status as the St. Bride’s and Spitalﬁelds crypt collections for a
comparative study of long bone morphology. Even if there were apparent sex-related
patterns in asymmetry in both skeletal series, it would be impossible to discount the
possibility that the patterns were also associated with environmental conditions in
London at the time. For this reason, a reference population drawn from the computerized
Forensic Data Bank at the University of Tennessee is employed in this study (Jantz and
Moore-Jansen 1988). This is a skeletal series of documented twentieth-century
Americans, and therefore drawn ﬁom a very different setting in time and space ﬁ'om the
Georgians. Although the individuals comprising the Forensic series are identiﬁed with
respect to age, sex, and year of death, they are not documented with respect to their ways
of life. They cannot be understood to represent a population in any other sense than
having lived in the United States this century; nonetheless, these characteristics are
suﬁcient to set them apart for a comparative study with the Georgians.

The individuals comprising the Forensic Data Bank sample are drawn ﬁom a
much broader range of socio-economic background and geographic area than the

Georgians. It is likely, therefore, that the former group would manifest a much greater

 

“’The records associated with married women in the Georgian populations
typically list the maiden name as well as the married name of the individuals. There is no
evidence that there any common surnames in either married or maiden names.

48
range of variation in asymmetry patterns—particularly asymmetry magnitude—than the
latter. If setting in time and space does affect asymmetry patterns, the diﬁ'erence should

be recognizable in the comparison of these skeletal samples.

W

While there is no documentation that characterizes the lifeways of the individuals
comprising the Forensic series, there is a rich documentary record of Georgian-era
London. Samuel Johnson provided the world with perhaps the most famous—and
certainly the most succinct—characterization of eighteenth century London ever written
in his diary entry of 20 September 1777: “When a man is tired of London, he is tired of
life; for there is in London all that life can afford.” In the preceding century London had
undergone a tremendous transformation. The 16605 had been disastrous for the city,
which had suﬂ‘ered in quick succession ﬁ'om both the Great Plague (1665) and Great Fire
(1666). By the 1770s Britain had emerged victorious ﬁom the Seven Years War, making
London the capital of the greatest colonial power of its time. The country had also begun
to feel the economically vitalizing effects of the Industrial Revolution. Although not
itselfan industrial center, London’s population grew substantially as immigrants from the
provinces, as well as ﬁom other countries, ﬂocked there in search of prosperity. By
1800, London had surpassed Paris in size to become the largest city in all of Europe
(Rude 1971).

Even at the beginning of the eighteenth century the area popularly referred to as
“London” radiated well beyond the boundaries of the City, which is itself only one

square mile in area (Beeton and Chandler 1969). The City itself lies on the north bank of

49

the River Thames, with the Tower of London marking its easternmost point. Its western
boundary is formed by the Fleet River, now covered over by Farringdon Road and Bridge
Street, which is the approach to Blackﬁiars Bridge.

For hundreds of years the City was enclosed by a protective wall which ran in a
roughly semicircular arc ﬁom the mouth of the Fleet northward, eastward on the southern
boundary of the Moor Fields, and back southward to the Tower. The wall was
punctuated by seven access gates: Ludgate was easternmost, and nearest the Fleet;
continuing clockwise it was followed in turn by Newgate, Aldersgate, Cripplegate,
Moorgate, Bishopsgate, and ﬁnally Aldgate, located just north of the Tower. St. Bride’s
Church is located just a few hundred meters west of Ludgate, on the south side of Fleet
Street, which was (and is) the major thoroughfare running westward ﬁ'om the City
parallel to the Thames. Christ Church is located at the southeast corner of Fournier Street
(formerly Church Street) and Commercial Street in the area called Spitalﬁelds that
extends northeastward ﬁ'om the area around Bishopsgate on the east side of the City.

If one were to take notice of the multitude of people passing by on Fleet Street or
Commercial Street in London today, it would be difﬁcult to ﬁnd a representative
Londoner. Likewise, it would have been almost as dimcult in the eighteenth century,
because of the wide diversity of the population. The increase in London’s population
following the Plague and Fire of the mid-16603 had many causes. In part, London’s
growth in the eighteenth century reﬂected the general increase in the population of
England as a whole. However, patterns of migration within the country were dominated

by a net shift in population ﬁom rural to urban areas (Wrigley 1987; Wrigley and

50
Schoﬁeld 1989). As a result, London’s population grew at a greater rate than most of the
country.

The majority of social historians agree that living conditions in urban areas across
England improved with the social and sanitary reforms of the mid-nineteenth century
(Walvin 1984). There is less agreement about the preceding 150 years. A prominent
debate among social historians of eighteenth- and early nineteenth-century England is
referred to, in fact, as the standard-of-living controversy (Floud, Wachter, and Gregory
1990; Royle 1987). Resolving the controversy is challenging because researchers can
cite valid evidence supporting contradictory positions. For example, economists can
point to the fact that per capita income increased consistently in England across the
century as evidence for an ever increasing standard of living (Burnett 1969). On the
other hand, the plight of the growing industrial working class is also well-documented
(Thompson 1966).

From the mid-eighteenth through the mid-nineteenth century the Industrial
Revolution brought about changes in the standard of living across the nation, particularly
in larger cities. When the rural poor ﬂocked to the cities to ﬁnd work, they simply added
to the growing number of poor working-class urban dwellers. At the same time that the
disparity between the very rich and very poor was increasing, there was also a growing
number of persons who could best be described as “middle class” (cf. Cox and Molleson
1993). Dorothy George (1965) argues that London differed signiﬁcantly ﬁ'om other
urban areas in England during the years of the Industrial Revolution—that is, in the years
1750 through 1850. While the large cities of the industrialized North were burdened by

the Industrial Revolution, she posits that London was able to escape large-scale

5 1
industrialization yet still beneﬁt from the economic grth that resulted from the output

of the industrial centers further north. George’s conclusion ﬁ'om a comprehensive review
of available documentary evidence was that, with the exception of a setback between
1720 and 1750, living conditions for most inhabitants of London progressively and
consistently improved throughout the century. Nonetheless, she agrees with other
historians (Burnett 1969; Porter 1990; Royle 1987; Thompson 1966; Walvin 1984) that
any general improvement in living conditions during the eighteenth century beneﬁted the
middle and upper classes much more than the working class.

In addition to migration from rural to urban areas within the country, another
factor leading to the tremendous increase in London’s population was the large number
of immigrants who descended upon London from abroad to escape religious persecution
and economic deprivation. Because of their alien status, they were not eligible to become
citizens, and they could not live within the City itself. However, they wanted to live as
near as possible to the population center to be able to engage in a reasonable livelihood.
Liberties on the fringe of the City, which were not under control of the City, became their
reﬁrge.

One of the most famous immigrant groups was the Huguenots, the French
Protestant followers of John Calvin. For decades they had lived in relative peace in their
predominantly Catholic homeland, owing to the religious tolerance they enjoyed
following the Edict of Nantes in 1598. When the Edict was revoked by Louis XIV in
1685, the Huguenots were forced to ﬂee religious persecution by emigrating to other
countries. They established a number of expatriate communities in England, as well as

Germany, Switzerland, and the Dutch Republic (Gwynn 1985; Cottret 1991).

rte
Spit

500

g-

the

iii

90."

hr.- .
((54

ff»

“l:

52

Unfortunately for France, the loss of the Huguenots meant the loss of a large
number of highly skilled weavers and artisans. A substantial community of these
weavers found their reﬁrge in an area of London just northeast of the City known as
Spitalﬁelds. Gwynn ( 1985) notes that “There they congregated in the outskirts, where
food and housing were cheaper and guild control less eﬂ‘ective...”; Rose (1951 :43) adds
that the Huguenots were drawn to Spitalﬁelds “partly by the opportunity of practising
their craft, and partly by the cosmopolitan and non-conformist atmosphere which was
already typical of the area.”

Although the Huguenots were clearly an ethnic minority in Spitalﬁelds, their
community was well-respected and prospered for generations (Smith 193 9). Until the
time of the Industrial Revolution, the market for Huguenot hand-woven fabrics was
stable. But while the Industrial Revolution was a boon for most of the country, the effect
on the Spitalﬁelds silkweavers was devastating. Large mechanical looms located in the
northern part of the country could produce fabric now at a price signiﬁcantly lower than
that which had been charged by the handloom weavers. For a short time the Spitalﬁelds
Act of 1773 saved the handloom weavers from ﬁnancial ruin. Unfortunately, the stopgap
measures were not enough and the prosperity of the weavers community was reduced
signiﬁcantly in the early nineteenth century. In short, Spitalﬁelds suﬁ'ered ﬁom
economic decline in the midst of the Industrial Revolution (Smith 1939). This fact was
reﬂected in the population of Christ Church: looking at the burial register at Christ
Church, Cox noted that “the ratios of master craftsmen to artisan are almost reversed on

either side of 1800" (1989130), with a marked reduction in master craftsmen post-1800.

53

It is a truism that the East End of London and the West End are worlds
apart—-and have been since before the Great Fire of 1666. More prosperous individuals
have moved westward from the City proper, while those persons living to the east and
north of the Tower lived in relative squalor. This assessment is clearly an
oversimpliﬁcation, since many of the merchants and master weavers living in the area of
Spitalﬁelds, for example, were quite prosperous themselves in the years preceding the
Industrial Revolution. However, the arrival of mechanical looms introduced for the
manufacture of cloth did tremendous damage to the livelihood of the Spitalﬁelds weavers
(Smith 193 9). The relative poverty of the Spitalﬁelds neighborhood has persisted
throughout the nineteenth and twentieth centuries, as has its reputation as an abode for
recent immigrants.

Margaret Cox and Theya Molleson have outlined these circumstances in Ihe
Middh'ngjgn, their monograph of the anthropology of the individuals ﬁ'om the crypt of
Christ Church, Spitalﬁelds. The term “middling sort” has traditionally been used as a
reference to the rising middle class of the 18th century. A true picture of the
communities represented by the two crypt populations is certainly more complex that
what could be recorded in that simple phrase, but it does characterize the lifeways of the

people of St. Bride’s, Fleet Street, and Christ Church, Spitalﬁelds.

St. Bride’s (Fleet Street)
The Great Fire of 1666 decimated roughly eighty percent of the area within the

Wall; only the northeastern reaches of the City were spared. However, the Fire had also

54
been allowed to spread several hundred meters westward ﬁom the City. At the end of

three days, 13,000 houses and eighty-seven churches were destroyed (Morgan 1973:134).

One church which fell victim to the Fire was St. Bride’s on Fleet Street. St.
Bride’s is located just outside the west boundary of the City proper, a few hundred meters
beyond Ludgate (today’s Ludgate Circus). As one of his eﬁ‘orts in the reconstruction of
London, the architect Christopher Wren designed a new St. Bride’s, and the church was
rebuilt on its original site in 1708. During World War H, catastrophe again struck St.
Bride’s. On the evening of 29 December 1940, the church was gutted by ﬁre following
an incendiary bombing raid by the German air forces (Morgan 1973). Several years after
the War a decision was made to rebqu the Church yet again on the same site, and the
newest incarnation of St. Bride’s was completed in 1957.

Excavations on the building site in preparation for the construction revealed an
array of archaeologically intriguing ﬁndings. Human remains which have been identiﬁed
as Celtic in origin, dating to the ﬁfth century AD, were unearthed below Wren’s crypt.

In addition, a medieval chamel house was located below the ﬂoor of the church.

As noteworthy as these pre—Norman and medieval remains were, the ﬁnd of
greatest osteological signiﬁcance was excavated ﬁ'om Wren’s crypt itself. Here
researchers discovered that well over two hundred individuals had been interred in
coﬁns which bore name plates listing the occupant’s name, age, and year of death.

Dr. J. C. Trevor, then Director of the Duckworth Laboratory of Physical Anthropology at
Cambridge University, asserted that this was “outstandingly the most important

collection in the world” (Harvey 1968263).

55

Trevor’s enthusiasm was fueled by the realization that it was possible now, for the
ﬁrst time in history, to study a skeletal population wherein each individual was
unambiguously identiﬁed with respect to age, sex and year of birth. Unlike dissecting
room populations, St. Bride’s could oﬂ‘er a reference source of individuals who
represented a middle class lifestyle." By itself, this information could form the basis for
a range of studies on the eﬂ‘ects of age and sex on skeletal morphology. Moreover, it was
thought that this data could be corroborated with church records to determine family
relationships, and reference to guild records could reveal further evidence about
occupation and lifeways.

In the years following the excavation, the St. Bride’s material was used primarily
by researchers to study the relationships among age, sex, and the morphology of the
skeleton. F. L. D. Steel, in particular, published metric analyses of the skeletal material
(Steel 1962), and as recently as 1987 Sue Maclaughlin used the St. Bride’s sample to test
the eﬁ‘ectiveness of techniques for sexing the human skeleton based on morphological

Although some of the St. Bride’s material was removed to Cambridge for study
for a short period of time in the 19503, all of the identiﬁed individuals were returned to
the crypt and placed in storage containers. Acknowledging the continued scholarly value
of the collection, the Church has allowed the skeletal material to remain accessible to

M researchers who wish to study it—with a' stipulation that the skeletal material is

 

"At the time there were virtually no published studies of the skeletal biology of
middle class individuals. Since then, Angel (1975) has broached the subject, but a lack
of available skeletal material makes it difﬁcult undertake such analyses. In this respect,
Trevor’s enthusiasm was and is quite ﬁtting.

56

not to physically leave the crypt. The stipulation that material not leave the crypt makes
it impossible to undertake a radiographic study of the St. Bride’s long bones. Even if
permission could be granted for transporting a portable radiographic instrument into the
crypt, there is only a very limited supply of electricity to the crypt laboratory. It would
be extremely valuable to have radiogrpahic data from St. Bride’s available as a
comparative sample for studies of asymmetry in diaphyseal morphology.

Rosemary Powers undertook comprehensive cataloguing of the identiﬁed St.
Bride’s material in 1960. In addition to listing the names and coﬁn plate information of
the individuals, she included a series of pen and ink drawings showing the extent of
preservation of each skeleton and a typewritten description of the skeletal elements and
dentition. In some cases, she was also able to oﬂ‘er comments about documented family
relationships.

Recognizing that ﬁlrther deterioration of the collection had taken place since
Powers’ catalogue was completed, a team of researchers (Louise Scheuer, Susan
Maclaughlin-Black, and Jacqui Bowman) secured funding from the Leverhulme
Foundation in the mid-1980s to recatalogue the material and establish a modest
laboratory facility within the crypt itself. In addition, they attempted to further explore
the historical background related to the St. Bride’s individuals. By making reference to
guild and municipal records, they were able to determine the cause of death for 64% and
address at death for 84% of the adults, as well as the occupation of 20% of the males.
This project is just being completed and the skeletal material is now again being made
available for osteological analysis. The study on which this dissertation is based is the

ﬁrst since the re-establishment of the St. Bride’s crypt.

57

All told, there are 237 individuals currently in the identiﬁed St. Bride’s series; of
the 212 adults, 110 are males and 102 are females. As might be expected, there is wide
variation in the degree of preservation of the St. Bride’s material. Many of the I
individuals are nearly completely intact, but the majority have shown signiﬁcant
deterioration over the years. Indeed, for a small number of individuals skeletonized

remains are virtually nonexistent.

Christ Church (Spitalﬁelds)

One consequence of London’s population increase in the decades following the
Great Fire was the establishment of an ambitious mandate by Queen Anne in 1711 to
construct ﬁfty new churches in and around the city to accommodate the increasing
population (Smith 1939:99- 100). One of the seventeen which was eventually completed
was Christ Church, located in Spitalﬁelds, a neighborhood situated just to the north of
Whitechapel village, and northeast of the segment of the Wall bounded by Aldgate and
Bishopsgate. In terms of modern geography, Christ Church is located approximately
one-half kilometer north of Aldgate Underground station on Commercial Street, between
Fournier Street and Fashion Street.

Thomas Hawksmoor, a student of Wren, was commissioned to design Christ
Church. Construction was begun in 1714, and the church was completed and consecrated
in 1729 (Adams and Reeve 1987). The ﬁrst recorded interment in the crypt was listed in.
the Burial Register that same year. The crypt of the church was used for interments until
1859, when it was decreed that the further crypt interments be prohibited, in the interests

of public health. In 1867 a further mandate was executed that the crypt be sealed.

58

While Christ Church has never experienced ﬁery tragedies like those which befell
St. Bride’s, deterioration over the years brought about the need for a broad program of
restoration to the physical inﬁastructure of the Church. These repairs necessitated that
the crypt be cleared. The eastern halfof the crypt vaults had already been cleared in the
1960s to provide space for a shelter for transients within the church (Adams and Reeve
1987). However, the western halfof the vaults still held the remains of nearly 1000
individuals. In the early 1980s an agreement was reached with staﬂ‘ at the British
Museum (Natural History) to conduct a scientiﬁc excavation of the crypt, with the
stipulation that the skeletal remains would be made available for a period of time for
osteological study (Cox 1989; Adams and Reeve 1987).

As was the case at St. Bride’s, several hundred persons ﬁom Spitalﬁelds were
interred in cofﬁns with lead plates that provided unambiguous documentation of name,
age, sex, and year of death. However, the number of identiﬁed individuals at Christ
Church far exceeded that of St. Bride’s. Of the 968 total individuals whose remains were
excavated ﬁ'om the crypt at Christ Church, 378 could be identiﬁed on the basis of coﬁn
plate information. There were a relatively large number of juveniles in the Spitalﬁelds
collection: 50 males and 37 females under 20 years of age. Of the remaining 290 adults,
144 were identiﬁed as male, and 146 as female.

Unfortunately, the general state of preservation of the Spitalﬁelds sample is
poorer than that of St. Bride’s. Nonetheless, they still have provided a reference
papulation which parallels St. Bride’s. To date, studies of the Spitalﬁelds identiﬁed
skeletons have provided new criteria for determining the sex of juvenile skeletons

(SChutkowski 1993), as well as provocative assessments of bone density in the women of

59

Spitalﬁelds (Lees £1.11. 1993) and the presence of osteoarthritis among the silkweavers

(Waldron 1991).

W

St. Bride’s

The entire identiﬁed St. Bride’s collection is currently housed in a laboratory
located in the crypt of the Church itself. A large segment of the crypt is currently utilized
to house an public exhibition on the history of St. Bride’s-—and London in general—from
Roman times to the present. However, the area of the crypt which houses the skeletal
material and lab is not accessible to the public. The skeletal remains are individually
boxed, and are located within the conﬁnes of the laboratory. Typically there are two
boxes per individual, one containing the cranium and the other containing postcranial
remains. Each box is labeled with an identiﬁcation code, as well as an indication of the
age and sex of the individual. Names and other useﬁrl information are available in
Powers’ (1960) catalogue, which has now been superseded by a card ﬁle developed by

the Leverhulme project.

Spitall'relds

The Spitalﬁelds collection is currently housed in the basement storage area of the
British Museum (Natural History). Like with the St. Bride’s collection, cranial and post-
cranial remains are boxed separately. Unfortunately, the Spitalﬁelds collection diﬂ‘ers
ﬁ'om St. Bride’s in that there is no listing of the skeletal elements associated with each

individual. The situation is further complicated by the fact that the identiﬁed as well as

60

the unidentiﬁed Spitalﬁelds individuals are stored together. Storage boxes indicate the
catalogue number, but not whether a given individual is from the identiﬁed subsample or
not.

Each of the Spitalﬁelds individuals is identiﬁed by a four-digit catalogue number,
beginning with the digit 2. That is, the Spitalﬁelds catalogue numbers run ﬁom 2001
through 2987. Because the St. Bride’s collection has no four-digit catalogue numbers,

any potential problem of confusion between the two collections is easily avoided.

CHAPTER 4-—METHODS

W
This study tests hypotheses that characterize the relationship between apparent

sex-related factors and environmentally-related factors that aﬂ‘ect the phenomenon of
long bone asymmetry. It employs speciﬁc osteometric observations which (a) extract the
greatest possible amount of information ﬁ'om a relatively small number of measurements
and (b) meaningfully supplement the existing literature regarding long bone asymmetry.
To meet this goal, the measurements included in this study were selected to satisfy four
criteria:
1. Wm. To facilitate comparison of the results with others in the literature,
each measurement is associated with an unambiguous measurement technique and has
also been employed by previous researchers.
2. 2mm. To reduce as much as possible the potential for measurement error, each
measurement employs distinct and unambiguous anatomical landmarks.
3. Man. To permit the largest available study samples, each measurement
involves skeletal elements which are best preserved in skeletal series.
4. WW. To allow an assessment of the relationship between
asymmetry and activity, selected measurements include both those that are subject to
activity-related morphological changes as well as those that are not likely to be directly
modiﬁed by physical activity.

These criteria are best met in three paired measurements of the humerus and their
counterparts in the femur. Because these two long bones are large and robust, they are

likely to survive intact in skeletal samples (Dittrick and Suchey 1986). Moreover,

61

‘0
I“

‘l.

elf

62
detailed analyses of the morphology of the humerus and femur have a long tradition in

the skeletal biology literature (Aiello and Dean 1990). The humerus and femur are
analogous components of the upper and lower limbs; however, the femur is a weight-
bearing bone, while the humerus is not. This suggests that the humeri would be more
likely subject to activity which favors one side over the other. In contrast, the femora
experience a greater amount of loading, but loading which would be shared more equally
between the paired bones.

The speciﬁc measurements are maximum length, head diameter, and
biepicondylar width. ‘2 Each is a standard measurement that has been well studied by
other researchers in the past. Because the epicondyles are sites of muscle and tendon
attachments, biepicondylar widths are subject to activity-related changes in morphology.
Head diameters are located on the articular surfaces of the humerus and femur; as such,
their morphology is not directly subject to activity-related variation in size. ‘3

The speciﬁc techniques for taking the six paired measurements are based on the

guidelines outlined in l 2 .:

 

 

”The terms “biepicondylar width” and “epicondylar wid ” are used
interchangeably in the skeletal biology literature. Some authors use the term
“biepicondylar breadth” or “epicondylar breadth” to refer to the same measurement.

l3While humerus and femur head diameters are not directly subject to activity-
related morphological variation, France suggests that their sexually dimorphic character
is associated with their being positioned in close proximity to areas of large muscle
insertion (France 1988:523).

63

(Moore-Jansen and Jantz 1989), as described below. The measurements were taken on

both the St. Bride’s and Spitalﬁelds material in accordance with these guidelines.“

Maximum Length
40. MW: The direct distance from the most

superior point on the head of the humerus to the most inferior point on the
trochlea....

Instrument: osteometric board

Comment: Place the humerus on the osteometric board so that its long axis
parallels the instrument. Place the head of the humerus against the vertical
endboard and press the movable upright against the trochlea. Move the bone up,

down and sideways to determine the maximum distance... (Moore-Jansen and
Jantz 1989272; references to ﬁgures and literature citations omitted).

60. WW1: The distance ﬁom the most superior point
on the head of the femur to the most inferior point on the distal condyles....

Instrument: osteometric board
Comment: Place the femur parallel to the long axis of the osteometric board and
resting on its posterior surface. Press the medial condyle against the vertical
endboard while applying the movable upright to the femur head. Raise the bone
up and down and shift sideways until the maximum length is obtained... (Moore-
Jansen and Jantz 1989279; references to ﬁgures and literature citations omitted).
The proximal and distal ends of both the humerus and femur are located on the
articular surfaces of the bones. In contrast, the distal ends of the other major long bones
are sites of muscle and ligament attachments (the malleoli of the tibia and ﬁbula, and
styloid processes of the radius and ulna). These latter attachment sites are prone to
osteophytic deposits which can modify the length measurements of the bones, and render

those bones unsuitable for analysis of length asymmetry.

 

1"The numbers associated with each measurement reﬂect the measurement

number in ﬂIeDataQQlLeztinnlimss-rdurss manual.

Ha

Head Diameter

 

distancebetween the most superior and inferior points on theborder of the
articular surface...

Instrument: sliding or spreading caliper
Comment: Measure the vertical distance perpendicular to the maximum
transverse diameter of the head of the humerus. This diameter is not necessarily

equal to the maximum diameter.... (Moore-Jansen and Jantz 1989:72-3;
references to ﬁgures and literature citations omitted).

63. Maximurnﬂametmftheﬁemunﬂsarlt The maximum diameter of the
femur head measured on the border of the articular surface...

Instrument: sliding caliper

Comment: Place the femur head between the branches of the instrument to ﬁnd

the maximum diameter. This measurement is in contrast to the separate vertical

and transverse diameters recommended by Martin... (Moore-Jansen and Jantz

1989279; references to ﬁgures and literature citations omitted)"

In both the humerus and femur, the proximal articular surface—the head—is
located within a joint capsule. As a result, the morphology of the head is not subject to
osseous modiﬁcation at sites of tendon and ligament attachments. Although the area
surrounding the head of the humerus is a site of attachment for a number of rotator cuﬂ’
muscles, they are all located beyond the margin of the head. At the proximal end of the
femur, the major sites of muscle attachment are the greater and lesser trochanters. The
ligarnentum teres is attached to the foveal depression in the center of the femoral head,

but in no case does that site of attachment complicate the measurement of maximum head

diameter.

 

1"The reference to Martin in the description of measurements is Martin and
Saller (1957)

65

Biepicondylar Width
41. Epicondylar Breadth of the Humerus: The distance of the most laterally
protruding point on the lateral epicondyle from the corresponding projection of
the medial epicondyle...
Instrument: osteometric board
Comment: Place the bone with its posterior surface resting on the osteometric
board. Place the medial epicondyle against the vertical endboard and apply the
movable upright to the lateral epicondyle... (Moore-Jansen and Jantz 1989:72;
references to ﬁgures and literature citations omitted).

62. Epicondylar Breadth of the Femur: The distance between the two most
laterally projecting points on the epicondyles. . ..

Instrument: osteometric board

Comment: Place the femur on the osteometric board so that it is resting on its

posterior surface. Press one of the epicondyles against the vertical endboard

while applying the movable upright to the other condyle. The measurement is
parallel to the distal surfaces of the condyles.... (Moore-Jansen and Jantz

1989279; references to ﬁgures and literature citations omitted).

By deﬁnition, biepicondylar width of the humerus and femur is measured as the
maximum distance between the medial and lateral epicondyles, which are found at the
distal end of the bone. The epicondyles are located outside the joint capsule of the elbow
and knee, and are the sites of muscle and ligament attachments.

In the humerus, the medial epicondyle is a common origin site for several of the
ﬂexor muscles of the forearm, as well as the pronator teres muscle. The lateral
epicondyle is a common origin site for several of the extensor muscles of the forearm and
the anconeus. As such, variation in the size and morphology of the humeral epicondyles
have been linked to: patterns of physical activity (France 1988, Dittrick and Suchey
1986). The epicondyles of the femur are primarily sites of ligamentous attachments.

Speciﬁcally, the medial (tibial) collateral ligament attaches to the medial epicondyle; the

66

lateral (ﬁbular) collateral ligament and the lateral head of the gastrocnemius muscle are

associated with the lateral epicondyle.

W

The guiding principle for assessing the two Georgian series was to measure all
available adult skeletal material ﬁ'om each that met documentation and preservation
criteria. No juvenile skeletons were measured because the epiphyses of the humerus and
femur would not be suﬁciently fused to ensure that maximum length measurements
would be accurate. Both crypts contained a large number of undocumented individuals;
that is, those for whom name, age, and sex are not identiﬁed on the basis of information
listed on coﬁn plates. At Spitalﬁelds, for example, over 950 individuals were removed
ﬁ'om the crypt, but less than 400 were “identiﬁed” (Adams and Reeve 1987).

Individuals are included in the study sample only if they retain a suﬁcient level
of preservation for accurate maximum length measurements to be taken on either paired
humeri or paired femora In many cases, however, these bones also displayed variable
levels of disintegration of the head or epicondyles, and a substantial number of humeri
manifest osteophytic lesions, particularly on the medial epicondyle. Because the
presence of osteophytes has a strong potential to distort asymmetry assessments for the
biepicondylar humerus widths, those bones which presented with osteophytic lesions are
discarded from the analysis sample. In the case of the femora, there is virtually no
evidence of osteophytic lesions on the epicondyles; however, the distal ends of femora
often suﬁ‘er disintegration of the trabecular bone to the extent that it is impossible to

accurately measure the biepicondylar width.

67
Likewise, arthritic changes can be observed on the margins of the head of the

humerus and femur in some mature adults. Osteophytic lipping, in particular, can distort
the measurement of vertical head diameter of the humerus, which is measured from the
most superior to the most inferior margins of the head. In the femur, osteophytic lipping
rarely deforms the aspect of the head where diameter is measured. In a few cases, it is
possible that arthritic ebumation can distort the shape of the femoral head to the extent
that its diameter cannot be accurately measured. In any case where the diameter of the
humerus or femur head appeared to be grossly modiﬁed by pathological conditions that
bone pair was removed ﬁ'om the study sample.

There is a necessary trade-off in assessing osteometric patterns in skeletal
samples which are not well-preserved. Ifone were to only include those paired bones for
which all three measurements (length, head diameter, and biepicondylar width) are
available, the resulting sample size is relatively small. If one were to investigate each
measurement individually, then the sample sizes increase accordingly. However, because
these larger samples consist of diﬁ‘erent individuals it is not possible to make accurate
comparisons between them. To resolve this problem the assessment of humerus length
patterns are based on two samples. The mired sample is larger, and consists of all
humeri for which paired length measurements are available. The smaller intact sample
consists of humeri for which all three paired measurements are available. To be included
in the intact sample then, a bone pair would need to show no osteophytic lipping on
either head, and no osteophytic deposits on either of the epicondyles. In addition, the
head and epicondyles would need to be suﬁciently preserved for accurate measurements

to be taken. Likewise, each series consists of two samples of femora, partitioned in the

68

Table 8 Sample Size—Georgian Subpopulations

 

Spitalﬁelds Spitalﬁelds St. Bride’s St. Bride’s
Males Females Males Females
Paired Humeri 42 53 30 32
Paired Femora 18 33 36 35
Paired Humeri 16 3O 23 25
and Femora
Intact Humeri 36 47 27 22
Intact Femora 17 27 24 19

 

 

same manner as the humeri. Therefore, each of the series is partitioned into six
subgroups, each with a different sample size, as summarized in Table 8.

The St. Bride’s and Spitalﬁelds skeletal series are considered both as individual
populations and also as combined “Georgians” when they are compared with the
Forensic Data Bank sample. This two-level assessment of the Georgians is designed to
help assess the problem of whether two contemporaneous and spatially local skeletal
groups can be legitimately understood to be a single population.

Data ﬁom a total of 225 individuals (130 males and 95 females) were provided
from the Forensic Data Bank at the University of Tennessee. The individuals culled from
the Data Bank for the comparative sample were adults unambiguously identiﬁed with
respect to age and sex, coded as “Caucasian” in the Data Bank, and expressing the six
paired measurements as outlined above. The criteria by which individuals were selected
for inclusion in the ﬁnal study series was the same as for the Georgian populations. The

resulting sample sizes are summarized in Table 9 as well.

69

Table 9 Sample Size—Identiﬁed Subpopulations

 

Georgian Georgian Forensic Forensic

Males Females Males Females
Paired Humeri 72 85 85 64
Paired Femora 54 68 93 65
Paired Humeri and Femora 39 55 75 58
Intact Humeri 63 69 80 58
Intact Femora 41 46 82 58

 

Reduction in Size of the Skeletal Series

Among the methodological problems which plague studies of asymmetry in
human bones, probably the most serious is that researchers draw conclusions ﬁom
skeletal populations with relatively small sample sizes. The problem of small sample
size is common to studies of human skeletal biology—particularly in comparative studies
of archaeological populations. Given the size of the St. Bride’s and Spitalﬁelds skeletal
collections, it was ﬁrst thought that small sample size would not be an issue for this
study. The two collections seemed to provide an abundant study population, with 212
adults in the total St. Bride’s identiﬁed population and 290 in the Spitalﬁelds identiﬁed
population.

Diﬁculties arose, however, with individuals lacking an adequate level of

Preservation to enable accurate measurements to be taken on the appropriate skeletal

elements. It was disheartening to remove a well-preserved individual removed from

70
consideration on the basis of a minor deterioration at the distal end of femur, or if a chip
of bone had been removed ﬁom the trochlear projection of the humerus. It was even
more disheartening as the number of individuals who fell into this category increased
steadily as the measuring process progressed. One way to mitigate the problem is to
divide the skeletal series into multiple subsarnples in an attempt to retain the largest
possible sample sizes for each measurement without having a separate study sample for
each paired measurement. This application of this strategy is described in the following
chapter-

Degree of preservation was not the only factor which determined which
individuals were selected for inclusion in the skeletal samples. The greatest diﬁculty
with selecting a sample ﬁom the St. Bride’s collection for analysis (once degree of
preservation was accounted for) was that of non-secure identiﬁcation. When the crypt
population was initially excavated following WWII, the environment for methodological
data collection was clearly less than optimal. One consequence was that the skeletal
remains ﬁom a substantial number of individuals was mixed during initial storage.

The extent of the mixing was perhaps underestimated prior to the systematic
recataloguing of the collection in the Leverhulme project. In some cases, it was
relatively easy to separate the major skeletal elements of mixed individuals. For
example, in a few instances the remains of a husband-wife pair were placed in the same
storage box; if the sex of the individuals was obvious from observing the length and
robusticity of the skeletal elements, then the separation of the two persons was a fairly

straightforward process. In cases where two persons of the same sex and similar age

71

were placed in the same storage box, it was much more difﬁcult to accurately associate
the elements with discrete individuals (Scheuer, personal communication).

The Leverhulme researchers divided the St. Bride’s population into three
categories, based on certainty of identiﬁcation. Those individuals for whom
identiﬁcation was certain were placed in the “secure” category. Ifthe researchers felt
conﬁdent about the identiﬁcation of an individual, but could not be absolutely certain that
there was no mixing of some skeletal elements, it was placed in a second “likely secure”
category. A third “unsecure” category consisted of those individuals about whose
identiﬁcation researchers could not feel conﬁdent. While it may well be that the majority
of the “unsecure” individuals are in fact correctly identiﬁed,- they were excluded ﬁom
consideration for the study. On the other hand, personal observation of the skeletal
material, in conjunction with conversation with Dr. Scheuer, conﬁrmed that the
researchers were conservative in her designation of “secure” skeletons, to the extent that
individuals with “likely secure” designations were included in the study. The
opportunity to discuss the nature of skeletal material directly with the curators of the
collection has a value that cannot be overstated, and it made much easier the task of

sorting through possible inconsistencies in the St. Bride’s collection.

W
In the early stages of the study, a particularly vexing problem arose. The team of
researchers who originally excavated the crypt of Christ Church (Spitalﬁelds) recorded a
substantial number of qualitative and quantitative observations shortly after the skeletons

were removed ﬁom their coﬁins. Among these were a series of post-cranial

72

measurements, including paired measurements of the long bones of the upper limb, and
measurements taken ﬁ'om the left lower limb bones. The data ﬁom these measurements
were then compiled into a computerized database (Cox and Molleson 1993).

When approached with the proposal for this project, the staﬁ’ of the British
Museum of Natural History generously oﬁ‘ered to make available those data ﬁles in order
to reduce the task of measuring the skeletal elements again. Given that the measurements
selected for this study were chosen in part because they were not subject to idiosyncrasies
in measurement technique, the possibility that an accurate data set could be constructed
without remeasuring the skeletal elements was viewed with optimism.

A small series of measurements had been collected on a number of Spitalﬁelds
humeri and femora in a pilot study during the previous summer. To be certain that there
indeed was no difference in measurement technique, ﬁgures from the computerized
database were matched with those ﬁom the pilot study to ensure that the results were
consistent. The comparison showed a small but marked diﬂ’erence in the two series of
measurements. A few of the bones which showed the widest discrepancy were measured
yet again, and the results were consistent with those taken in the pilot study, and not
consistent with those taken shortly after the skeletons were excavated. As a result, no
ﬁgures from the initial data collection in the 19803 were used in the current study;
instead, each of the bones was remeasured.

The discrepancy between the measurements taken in the 19803 and the 19903 was
troubling. The most obvious explanation for such an inconsistency is simple random
interobserver variation. Interobserver error has been well-documented in the physical

anthropology literature, both in relation to metric and non-metric analyses (Uterrnohle

73

and Zegura 1982). However, interobserver error did not seem to be the case in this
instance. In fact, the discrepancy appeared to be markedly unidirectional, since literally
none of the recorded measurements ﬁ'om the 19803 were larger than the 19903
measurements. The great majority of them were smaller, but a few showed no difference;
the magnitude of the diﬁ‘erence averaged approximately 2 mm. Ifall measurements were
within 2 mm of each other, it might have been possible to regard some manner of
interobserver error as the cause of the diﬁ‘erence, but here the difference was too great in
magnitude in too many cases. Likewise, the 19903 measurements were taken using the
British Museum’s own osteometric board, which ruled out the likelihood that the
diﬂ'erence could be attributed to a miscalibration of the measurement device.

The next suggested possibility was a systematic diﬁ'erence in measurement
technique. As noted earlier, in this study the true maximum length of the bones were
measured, positioning the bone against the ends of the osteometric board until a
maximum separation of the uprights is attained. Any alternative technique would have
resulted in a shorter value for a length measurement, since by deﬁnition there is no other
measurement technique that could have resulted in a larger value than the true maximum.

The only remaining possibility appears to be that the bones literally were shorter
in the 19903 than in the 19803. Studies have shown that exposure of cranial material to
humidity is associated with larger values for craniometric measurements (Albrecht 1983;
Uterrnohle and Zegura 1982; Utermohle en], 1983), but there has not been a recent
comparable study for long bones. However, Krogman and Iscan report that Rollet’s 1888
thesis indicated a 2 mm diﬂ‘erence in the length of cadaver bones when measured in a

“fresh state” and a “dry state” ten months later (Krogman and been 19862302).

74

According to Cox and Molleson’s monograph (1993) the state of the skeletal
material when removed ﬁ'om the crypt at Christ Church was highly variable. Most of the
individuals were interred in sealed lead coﬂins which contained various amounts of
coﬁn ﬂuid, which is consistent with the suggestion that some of the skeletal material
was maintained in a humid environment ﬁom the time of interment until the excavation
irn the 19803. The most likely explanation is that the bones were indeed longer when they
were removed ﬁ'om the comns than they were when remeasured after several years in
storage in the relatively dry environment of the Natural History Museum.

This explanation is supported by the ﬁnding that, on average, the femora in the
Spitalﬁelds populations presented a greater magnitude of discrepancy in the length
measurements than did the humeri. Such a ﬁnding is consistent with the phenomenon of
shrinkage, since the amount of shrinkage would be relative to the length of the bone.
Because femora are consistently longer than humeri, it would be expected that the
absolute magrnitude of shrinkage associated with the femora would be greater than that
associated with the humeri. It is also important to note that the ﬁnding that the humeri
and femora marnifest diﬂ‘erent magrnitude of discrepancy also eﬁ‘ectively rules out the
likelihood that a problem with the calibration of the osteometric board led to the
irnconsistent measurements. Ifthe osteometric board were poorly calibrated, there would
have been a more consistent magrnitude of length differences for the humeri and femora.

Indeed, the magnitude of discrepancy in length measurements among the femora
was not consistent, nor was the magrnitude of shrinkage among the humeri. Given an
explanation that humidity-related shrinkage caused the different measurements, this is

not a surprising ﬁnding. The environment in the coﬁns ﬁ'om which the skeletal material

75

was removed showed a range of variation. In some cases, the original researchers
reported that bones were in a much drier state than others. Even in 1919, Pearson and
Bell were critical of studies which attempted to compare the ﬁndings of “wet” bone and
“dry” bone studies. It is important to consider that interment in humid coﬂin
environments may lead to diﬂ‘erent osteometric observations than would be evident in

dry bones.

W

Before addressing the issue of asymmetry, frequency distributions and summary
statistics are generated and reported for each of the six paired measurements in each of
the ten subpopulations listed in the previous section. Characterizing the distribution
patterns of the linear dimensions themselves is an important prelude to assessing
asymmetry in that it clariﬁes the extent and nature of variation in general long bone
morphology among the study populations. For example, if the patterns of femoral head
diameter ﬁgures reported by Dwight (1905) and by Stewart (1979) persist in the
populations involved in this study one should observe that the modern forensic
population would consist of long bones which are consistently larger than those of the
Georgians. Ifthe Forensic sample consists of bones which are larger than the Georgians,
then it would not be surprising if they displayed an equally larger magrnitude of
asymmetry in the linear measuremernts. However, if the smaller Georgian bones were to
display an absolutely greater magrnitude of asymmetry the ﬁndings would be more

notable.

76

A consisternt strategy was used in determining the frequency distribution patterns
for the linear dirnernsions. Patterns are reported only for those individuals who have
paired measurements available. In each case, the mean of the observations from the
right and left bones is used to calculate the value of that dimension for each individual.
For example, a person with a right femur length of 420 mm and a left femur length of 423
mm would be listed as having a mean femur length of 421 .5 mm. The ﬁndings for each
subpopulatiorn, partitioned by sex, are assessed for mean, standard error of the mean,
standard deviation, and .95 conﬁdence level."

The subgroups were tested for independence on the basis of standard 2 or t tests.
Both tests assess the likelihood that two samples are drawn ﬁom the same population.
The 2 test is appropriate if both series under consideration consist of thirty or more
individuals; the t test is employed for smaller samples. As the number of individuals in

the study samples increase, the t score approaches the value of the z score.
I _

x 2

2 2
Sr 52
— + —
"r "2

The 2 score is calculated on the basis of Equation 2, where >‘< is the sample mean,

2:

 

 

(2)

 

 

s2 is the sample variance, and n is the sample size; the subscripts refer to the two samples.
The 2 score is deﬁned as the number of standard deviations by which a sample mean

diﬂ’ers ﬁom another p0pulation mearn, and can be used to test the signiﬁcance ofthe

 

“These ﬁgures were calculated using the built-in routines in Quattro Pro for
Windows, Version 5.0 (Borland).

77

diﬂ‘erence betweern two sample means. That is, a comparison of two samples that results
inazscoreof1.5 meansthattheirmeansdiﬂ‘erby 1.5 standard deviations. Inatwo-
tailed test of statistical sigrniﬁcance at an alpha of .05, the critical value of z is 1.96; if a z
score exceeds this value, the hypothesis that the samples are drawn ﬁ'om the same

population must be rejected. For an alpha of .01 the critical value of z is 2.58.

J"

_ ”72
s: s: (3)
,7, + 72?

Thesigrniﬁcanceofdiﬂ‘erencesinsmallsamplemeansareassessedbythettest.

 

 

The typical 1 test requires that the two samples under study be normally distributed with

equal variances. Because equal variances cannot be assumed for this study the 1 statistic

 

 

 

 

 

is calculated using Equation 3.
ﬂ 2 2 ‘2
2". l ’_2.
" "2
df = 2 2 l 2 2 (4)
[ii-L 1* ’- [—‘ l
"1 (5'1) "2 ('2'1)

 

 

The calculated value fort is essentially the same as the z score, but its
sigrniﬁcanceisassessednotbyusingthenormalcurvebutratherthetdistributionwith
degrees of freedom calculated by the formula in Equation 4 (rounded to the nearest

integer).

78

The presentation of the results follows a pattern which is applied consisterntly
throughout the following chapter. Two general Tables are presented for each
measurement; one compares the Spitalﬁelds and St. Bride’s males and females, the other
pools the Georgian samples and contrasts them with the Forensic Data Bank sample.
This underscores a dominant theme in this researcln, since in one case the two Georgian
groups are assessed as diﬂ‘erent sample populations, while in the other they are presented
as representative of a single population. Along with the Tables of summary statistics, the
distribution patterns are graphically represented in a series of Figures which follow the
same format as the Tables. That is, in one set of Figures the Georgians are compared as
if they were separate populations, while in the other they are pooled and contrasted with
the Forensic Data Bank. In all cases the males and the females of each skeletal group are

treated separately.

AW

The methodological caveat from Chapter 2 bears repeating: To designate paired
skeletal elements as symmetrical is to apply a label which is directly related to the
precision with which the skeletal elements are measured. One of the thorny problems
that one conﬁonts in assessing asymmetry is that intraobserver measurement error may
be misconstrued as an indication of asymmetry. Chapter 2 suggested that since the
standard osteometric board is calibrated in increments of one millimeter, intraobserver ,
variation in recording an observation could understandably be 1 mm. However, it would
seem unlikely that intraobserver variation would be 2 mm or larger, provided that

measurements are performed with consistent technique. It is important to recognize that

79

if one were to restrict the label “asymmetric” to bone pairs whose lengths differ by a
nnagrnitude of two millimeters or larger, an indeterminant number of cases of true
asymmetry are being lost in the process.

With respect to the maximum lengths of humeri and femora, the standard
procedures for taking the measurements minimize the likelihood for variation. That is, if
the bone is moved back and forth between the ﬁxed and moveable ends of the
osteometric board until the true maximum length is determined, there should be no
ambiguity as to the accuracy of the measurement. The same statement can also be nnade
about the biepicondylar widths of the humerus and femur. For the purposes of this study,
each paired measurement is coded as symmetrical if the magrnitude of diﬁ’erence between
the dirnernsion on the left bone and the right bone was less than one millimeter.
Diﬁ‘erences greater than or equal to one millimeter are then coded left-dominant or right-
donninant, depending on which side is larger.

Nonetheless, it is worthwhile to assess how the distribution patterns of asymmetry
direction change if a two millimeter threshold is employed in place of a one millimeter
tlnreshold. For this reason, patterns of asymmetry direction are compared for the lerngths
of the humerus and femur as reported using a two-millimeter threshold as well as a one-
millimeter threshold. The two-millimeter threshold is arguably a more valid indicator of

true asymmetry by mitigating the effects of measurement error.

WW
Once the distinction between “symmetry” and “asynunetry” is made, calculating

the direction of asymmetry is a straightforward process. In contrast to the ease of

80
determining the direction of asymmetry in paired linear dimensions, there are several
diﬂ‘erent ways to report the magnitude of that asynunetry. One way is to simply calculate
the diﬂ‘erence in the corresponding measurements on the right and left sides. That is, if a
riglnt femur were 480 mm in length and the left were 482 mm long, the magnitude of
asymmetry is reported as 2 mm. Ifthe right side were shorter than the left, the result
would still be reported as 2 mm.

Some researchers attempt to include both magnitude and direction in their
calculations, typically by applying subtracting the leﬁ-side measurement ﬁ'om the right-
side measurement and reporting a signed magnitude value. Using this strategy, the two
femora pairs described above would be reported as having asymmetry magnitude of 2
mm and -2 man, respectively. Whether or not the diﬂ‘erence is reported as sigred or
unsigned will gently affect the summary statistics derived for a population. For this
reason, summary statistics and frequency distributions based on both signed and unsigned
asymmetry values are reported for each measurement for each subpopulation.

Another consideration further complicates the calculation of asymmetry. It is
plausible that the magnitude of asymmetry is affected by the size of the character being
measured. For example, asymmetry of 2 mm between paired femoral heads is a geater
proportion of difference that between the lengths of paired femora. Table 10 describes
thisdiﬂ‘erenceusingmockdata, showingthatinthecaseoftlne head diarnetersthe
variation represents a difference of ﬁve percent. In assessing lengths, however, the same
two millimeters represents a diﬁ‘erence of only one-half percent. It is possible that the

magnitude of asymmetry is correlated with the size of the character being measured. If

81

this were the case then it is valuable to report the asymmetry as a percentage ofthe total
character size.

Table 10 Eﬁ’ect of Character Size on Asymmetry Magnitude

 

Right Left Difference Difference
(mill) (mm) (mm) (’/-)
Femur Head 42 40 2 5%
Femur Length 402 400 2 .5%

 

There is yet another potential source of inconsisterncy between reported studies,
evern when magnitude of asymmetry is scaled on character size. There are several

different ways to calculate the size of a characteristic in paired bones.

 

z Bight-Lei?)
Asymmetry ( Right )(100) (5)

One way is to choose one or the other side arbitrarily as the character size
(Equation 5). Fresia et al. (1990) chose this technique, which results in a negative value if
the left-sided element is the larger of the paired characteristics.

Another strategy is to scale the asymmetry against the smaller of the two side
measurements, as indicated in Equation 6. Trinkaus et al (1994) employed this formula,

which maximizes the perceived level of asynunetry in a population.

82

It also results in no negative values, which makes it more amenable to statistical

 

(Maximum-Minimum)) (100) (6)

Asymmetry = ( Minimum
analysis. However, the direction of the asymmetry is not available from the calculation.

Palmer and Strobeck (1986) list a number of ways that researchers have
calculated asymmetry patterns in assessing biological asymmetry (in their case,
ﬂuctuating asymmetry). Both signed and unsigned calculations ﬁgure prominently in
their discussion. In this study asymmetry is reported both in terms of millimeters
diﬂ'erence (both signed and unsigned) and as a percentage of mean character size (again
both signed and unsigned).

In short, asymmetry magnitude is calculated in four diﬂ‘erent ways for each
individual, and summary statistics are reported for each. In addition, each subpopulation
is ordirnally ranked with respect to mean asymmetry values of each paired measurement,

based on each of these four calculation strategies:

Signed Value = (Right - Left) (7)

Unsigned Value = [Right - Left1 (8)

83

 

S' P ___ (Right - Leﬂ)
rgned ercentage ( Right + Leﬂ) (9)
2

Unsigned Percentage = Right ’ Left I
(Right + Leﬂ) (no)

 

2

W

Cross symmetry refers to the phenomenon of reversed length dominance between
a component of the upper and lower limb—for example, a pattern wherein the right
humerus is longer than the left, while the left femur is longer than the right. For a
substantial subset of the core and comparative populations, preservation is suﬁcient to
allow length measurements to be taken on both humeri and both femora. In many cases,
it is also possible to determine all six paired measurements from the skeletal material
available. Having all these observations together permits an assessment of cross

symmetry in the postcrarnial skeleton.

W
The most challenging problem in this analysis is to determine how to indicate the
level of signiﬁcance for the asymmetry that is observed. When sample sizes are small, as
they commonly are in skeletal biology research, a relatively large amount of variation
needs to be presernt for the difference to be regarded as statistically signiﬁcant. In the

past most authors have avoided these methodological problems by primarily describing

84
the patterns that they have observed, and providing a series of summary statistics
consisting solely of means, standard deviatiorn, and/or standard error (cf. Milnter 193 6;
Schultz 193 7).

In inferential statistics a ﬁrndamental distinction is made among nominal, ordinal,
interval, and ratio data. The nature of bilateral asymmetry is such that the different
aspects of it are best assessed in terms of each of these scales of data. For example,
nominal data is that which falls into discrete categories, as illustrated by the Mm: of
asymmetry. In terms of side dominance, paired bones are either right-dominant, leﬁ-

dorninant, or symmetrical. These categories are arbitrarily constructed, but a given
observation in a given individual can only ﬁt into one of the categories. As sucln, the
appropriate test statistic for determining the statistical signiﬁcance of diﬂ‘erences between
goups is based on the chi-square (36') distribution.

A deﬁning characteristic of nominal data is that the categories have no
hierarchical ranking, but the magnitude of asymmetry can be ranked. The magnitude of
asymmetry can be viewed as an interval variable as well, to which standard parametric
analyses can be applied. A signiﬁcant difﬁculty arises when one attempts to assess both
the direction and magnitude of asymmetry in a single statistical framework. Those
researchers who did employ inferential statistics (F alk e111. 1988) used paired t-tests to

compare the size of paired skeletal elements.

Any attempts to assess simply the direction of asymmetry as a nominal variable
suﬂ‘er from problems as well. Researchers who have exarrnined direction of asymmetry in
limb bones (Latimer and Lowrance 1965, for example) have avoided statistical problems

by relying solely on describing the asymmetry patterns they observed. Their strategy of

85
characterizing asymmetry as left-dominant or right-dominant can lend itself to chi-square
analysis.

The most ﬁ'ustrating issue is that there is no well-reported statistic that can
represent both the direction and the magnitude of skeletal asymmetries. As will be seen
in the following chapter, reporting asymmetry in terms of unsigned magnitude values
alone provides a very diﬂ‘erent picture than when signed values are used. For example, if
the study populations are ordinally ranked ﬁ'Om the most symmetrical to the least
symmetrical on the basis of signed measurements there is no guarantee that the same
ordinal ranking will persist with unsigned measurements. The implications of this
problem are discussed more fully in Chapter 6.

For each of the measurements in each of the populations, asymmetry direction is
partitioned into one of three categories: left-dominant, symmetrical, or right dominant.
For the head diameters and biepicondylar widths a one-millimeter threshold distinguishes
between the dominant and symmetrical categories. For the bone lengths both a one-
nnillimeter and a two-millimeter tlnreshold is employed to categorize asymmetry
direction. Each of the study subpopulations is compared with the others to determine the
probability that diﬁ'erences in asymmetry direction ﬁndings could be due to chance.

To maintain consistency with reports in the literature that treat asymmetry as if it
met the criteiia for parametric assessment, the unsigned magnitude values are subjected
to parametric irnferential statistical analysis. Summary statistics are reported for all the ,.
study subsamples following the four conventions listed irn the previous section, and the

populations are ordinally ranked.

86

The Mann-Whitney U Test for Statistical Signiﬁcance
Trinkaus et a1 (1994) employed the Mann-Whitney U statistic to test the

signiﬁcance of diﬁ‘erences in asymmetry patterns within a population. The Mann-
Whitrney U was designed to test ordinally ranked data ﬁom two independent samples to
determine the probability that they are drawn from the same population (Mann and
Whitney 1947). It is a particularly useful tool for assessing bilateral asymmetry since it
does not rely on assumptions that underlie parametric statistical analysis. Speciﬁcally,
U-based statistics do not require that the variable under study follow a statistically
normal distribution pattern.

Table 11 Sample Data for a Mann-Whitney U Calculation

 

Humerus Length
(Right - Left)
Sample 1 0 2 2 -l 3 -2 1
Sample 2 0 0 2 4 -1 3

 

The Mann-Whitney U tests the probability that two unmatched and ordinally
ranked samples are drawn from a single population. The U is computed by pooling the
scores of the two samples, assigning ranks to each score, and then summing the rank
values for each goup individually. The procedure is somewhat time-consuming,
particularly with large samples, and it is worthwhile to lay out the exact procedure by
which the U is calculated. Table 11 lists asymmetry values for several individuals ﬁ'om
two samples. Observations with a negative value reﬂect left-dominance, while those

with a positive value are right dominant.

87

Table 12 Sample Data Placed into Ordinal Ranks

 

 

Sample 1 Ranks Sample 2 Ranks

-2 1 -1 2.5
-1 2.5 0 5

0 5 0 5

l 7 2 9

2 9 3 11.5
2 9 4 13

3 11.5

 

 

 

 

Table 12 shows the sample data ranked ordinally for the two samples. Because
the smallest observed value is “-2", it is given the lowest rank. There are two
observations at the next level (“-1 ”), so their ranks are averaged; rank 2 and rank 3
become a shared rank 2.5. Three values of “0" mean that they share the average ranking
of 4, 5, and 6-that is, 5. The same strategy is employed to determine the ranks of all
observations in both samples.

The Mann-Whitney U is the smaller of the two values derived ﬁom the following

two equations:

(11)

2 (12) '

88

In these equations, n1 is the number observations in the smaller of the two samples, and
n1 is the number of observations in the larger sample. Rl is the sum of the ranks for the

smaller sample, and R2 is the sum of the ranks in the larger sample.

 

U=(5)(7)+(6)(26’1)-46=17 (13)

Using the sample data above, nl = 6 and n2 = 7; R1 = 46 and R2 = 45. Placing the
values irnto the formula, a U of 17 is calculated. For values of n2 s 20, consultation of a
Mann-Whitney table is required to determine the probability associated with the test. ’7 In
this case, the table gives a probability of .3 14 when nl = 6, n2 = 7, and U=17.
Therefore, the null hypothesis that both samples were drawn ﬁom the same population

cannot be rejected. To reach an 3.213911 rejection value of p < .05 with samples of this

 

 

 

sizewould requireacalculated U of 8.
z = 2 14
J"r"2("l+"2 +1) ( )
12

A powerful characteristic of the Mann-Whitney U test is that when n2 > 20 it is
possrble to derive a z score ﬁom U, using the formula presented in Equation 14. From 2
it is easily possible to determine the statistical probability p that the two samples were

drawn ﬁ'om a single population.

 

"The original tables are found in Mann and Whitney (1947), and have been
reproduced in textbooks of nonparametric statistics such as Siegel (1956) and Sprent
(1989).

89

 

 

 

 

 

 

 

(1-32.
__ 2
2‘ (15)
n 3..
1"2 N N—ET
N(N-1) 12
3_ 3_ 3_ 3_
zr=zz+22+33+33=l+l+2+2=5 (16)
12 12 12 12 2 2

The presence of ties (i.e., multiple observations of the same value) does have a
slight eﬁ‘ect on the value of 2 derived ﬁ'om this equation. The length of the ties is the
characteristic that modiﬁes the z score, by changing the variability in the set of ranks, and
hence the standard deviation of the sampling distribution of U. Siegel (1956: 124-5)

oﬁ‘ers a correction for ties based on the formula in Equation 15. In this equation N equals

3
x

x

the sum of nl and n2. The length of each tie is t, and £T= Z
1

sample data from the Table above displays two ties of length 2 and two ties of length 3."

.1;

 

. For example, the

Equation 16 shows how this translates into a 27' of 5.

Two additional points about the correction of ties in the Mann-Whitney statistic
merit note. First, the correction has the eﬂ‘ect of increasing the z score slightly, which
will in turn cause a small reduction in the value of p. In other words, not correcting for
ties leads to a slightly more conservative test. Secondly, a long run of ties contributes

signiﬁcantly more to the value of {T than would a shorter run of ties. For example, a t of

 

"The correction for ties applies ornly in the case where a z score can be derived
from U— that is, where n, > 20. The sample data from the Table does not contain an
adequate number of observations for the correction formula to be applied. In this case,
the sample Table data is applied for convenience simply to demonstrate the calculation of

ET.

9O

6 results in a t of 17.5, and a t of 12 results in a t of 143.833. Emphasizing these two
points, Siegel (19562126) recommends that “one should correct for ties only if the
proportion of ties is quite large, if some of the t’s are large, or if the p which is obtained
without the correction is very close to one’s previously set value of a.” The report of
results in Chapter 5 presents a correction for ties for all comparisons which display an
uncorrected p < .10; the corrected 2 and p values are listed in parentheses below the
uncorrected values.

The Mann-Whitney U is a median-based statistic. In addition to not assuming an
underlying normal distribution patterrn, it also minimizes the inﬂuence of extreme outliers
in the sample distribution. If, for example, the largest asymmetry value in Sample 2 ﬁom
the Table were 6, rather than 4, it would still retain the same rank value. In a parametric
test, such an outlier would have a geater impact on the calculated relationship between
the two samples. This is a noteworthy consideration in discussions of asymmetry, since
in a given data set there is a possibility for a outlier demonstrating an unusually large
magnitude of asymmetry. When there is no evidence of measurement error, irnaccurate
pairing of the bones, or pathology, it would be unreasonable to remove the outlier from
the study sample. At the same time, if that outlier were assessed in terms of its
magnitude, rather than its ordinal ranking, it might lead to a spurious representation of

the population mean.

W
The hypothetical questions raised at the conclusion of Chapter 2 are assessed ﬁrst

by tests that compare the two Georgian populations with each other, then pool the

91

Georgians to make a comparison with the Forensic sample. In each case the pertinent
testing statistic determines the probability that the two samples being compared are
drawn ﬁ'om a single statistical population. Each of the asymmetry hypotheses is tested at
three levels. For the direction of asynunetry, x3 tests are the basis for testing statistical
signiﬁcance of the diﬂ’erences. For both signed and unsigned magnitude values, the
Mann-Whitney U statistic is employed. The 91.213913 signiﬁcance level (a) is set at .05 for
all the assessments, and other more conservative levels of p are identiﬁed below as
appropriate. The .05 alpha value is chosen to maintain consistency with the literature;
none of the asymmetry studies in the existing literature which employ signiﬁcance levels

use values which are less conservative.

The speciﬁc hypothetical testing strategies are as follows:

 

,2”, ,_ ,3 ,1 ..H H H h. , ,1. - WW - ﬂ”. , '. ,gn _“ f, ._. , J. , ..
lune. Before assessing diﬁ’erences in asymmetry patterns between the skeletal series it is
ﬁrst necessary to assess how the series differ in terms of the magnitude of the dimensions '
of the bones themselves. This hypothesis would be supported if the two series of
Georgians display statistically similar distribution patterns in long bone dimensions

which differ signiﬁcantly ﬁ'om those of the Forensic series. This means that comparisons
of bone dirnernsions between the two series of Georgian males would not reject the null
hypothesis that they are drawn ﬁ'om the same statistical populatiorn, and that comparisons
of bone dimensions between the pooled Georgian males and the Forensic males would
reject the null hypothesis that they are drawn from the same statistical population.

Likewise, comparisons of bone dimensions between the two series of Georgian females

92

would not reject the null hypothesis that they are drawn from the same population, and
that comparisons of bone dimensions between the pooled Georgian females and the

Forernsic females would reject the null hypothesis that they are drawn ﬁ'om the same

po 'on.

 

mm. This hypothesis would be supported if the two Georgian male
populations display a similar pattern of asymmetry which is statistically consistent with
the Forernsic males patterrn, and if the Georgian females display asymmetry patterns
which are consisternt with the Forensic females. This mearns that comparisons of
asymmetry direction and magnitudebetween the sexes in each of the three populations
(St. Bride’s, Spitalﬁelds, and Forensic Data Bank) would reject the null hypothesis that
they are drawn from the same statistical population; and comparisons of asymmetry
direction arnd magnitude between the two series of Georgian males and between the two
series of Georgian females would not reject the null hypothesis that they are drawn ﬁom
the same statistical population.

(3) Ir: '
Wm. This hypothesis would be supported if the two series of

Georgians display a statistically similar pattern of asymmetry which differ signiﬁcantly

 

ﬁom the Forernsic series. To avoid conﬁrsion with possible sex-related factors the two
sexes are tested separately. This means that comparisons of asymmetry direction and
magnitude between the two series of Georgian males would not reject the null hypothesis
tlnat they are drawn from the same statistical population, and that comparisons of

asymmetry direction and magnitude between the pooled Georgian males and the Forensic

93

nnales would reject the null hypothesis that they are drawn ﬁ'om the same statistical
population. Likewise, comparisons of asymmetry direction and magnitude between the
two series of Georgian females would not reject the null hypothesis that they are drawn
ﬁom the same statistical population, and that comparisons of asymmetry direction and
magnitude between the pooled Georgian females and the Forensic females would reject
the null hypothesis that they are drawn ﬁom the same statistical population.

(4) Ir:
this is the case then the direction of asymmetry in biepicondylar widths (which are sites

 

of tendon and ligament attachments) within individuals should correspond to direction of
asymmetry in the corresponding bone lengths. In statistical terms this means that in any
given subsample the results of x3 analysis comparing proportions of right- and left-
dorninance in biepicondylar width with direction of dominance in bone lerngth would not
reject the null hypothesis that they are drawn ﬁ'om a common statistical population. This
hypothesis is based on France’s (1988) assertion that the morphology of muscle

insertions on bones are associated with physical activity levels.

W

In sum, six paired long bone measurements are to be assessed for the two skeletal
samples representing Georgian-era crypt populations from London—St. Bride’s and
Spitalﬁelds. The measurements are analogous components of the humerus and femur:
maximum length, head diameter, and biepicondylar width. The statistical signiﬁcance of
the population diﬁ‘erences are assessed in four ways: (a) 2 test for comparing population

means of the long bone dimensions, or attest if the two samples being compared do not

94
each meet the nninirnum for sample size (n230); (b) 1’ test for comparing patterns of
asymnnetry direction; (c) Mann-Whitney U test for comparing the signed and unsigned
magnitude of asymmetry; and (d) binomial test for comparing asymmetry direction
patterns between measurements within bones, and for crossed symmetry between
lnumerus and femur lerngth.

The two Georgian series are dealt with both as individual populations and as a
common population. When pooled, the Georgians are contrasted with individuals from
the 20th cerntury United States, based on measurements derived ﬁ'om the Forensic Data
Bank at the University of Tennessee. For each individual measurement, comparisons are
made at three levels. First, the males and females of each series are compared.
Secondly, the males of the two Georgian series are compared, and females of the two
Georgian series are compared. Thirdly, the pooled Georgian males are compared with
the Forensic males, and the pooled Georgian females are compared with the Forensic
females.

CHAPTER S—RESULTS

The overall strategy of this research project is unique in two respects. First, it
addresses both the magnitude and direction of asymmetry of each of the six
measurements in detail. Secondly, it investigates relationships among asymmetry
patterns at the population level both within individual bone pairs and between the
humerus and femur. Accordingly, this chapter is divided into several sections. The ﬁrst
section describes how the individual linear dimensions are distributed among the study
subgoups. The second section focuses speciﬁcally on the direction of asymmetry for
each measurement among the goups. The third section presents the patterns of signed
and unsigned asymmetry magnitude for each measurement among the subgoups. The
fourth and ﬁfth sections evaluate the patterns of asymmetry witlnin the paired humeri and
the paired femora of each goup, respectively, and the sixth section addresses the issue of
cross symmetry in the maximum length of the humerus and femur. The ﬁnal section
summarizes the results in terms of the hypotheses laid out in the previous chapter.

Throughout this chapter a series of Tables present summary statistics for the study
subgoups, partitioned by sex. In addition, a series of Figures display in a gaphic format
the distribution patterns for each of the measurements. In most cases, the vertical axes of
the gaphs are scaled to represent the percentage of each subgoup represented at each
interval, rather than the raw numbers of individuals represented. Because the subgoups
vary substantially in size, tlnis strategy for illustrating ﬁndings facilitates the reader’s
ability to make comparisons of patterns among them. Likewise, all Figures associated

with a givern measurement employ a consistent scale along the horizontal axis.

95

96
WW

Preliminary to reporting patterns of asymmetry patterns in the bones it is

necessary to describe the distribution patterns of the measurements themselves. This
comparison gives some indication of how the Georgian goups compare with each other,

and how they contrast with the Forensic Data Bank individuals.

Humerus Length—HI.

Summary statistics for each of the four Georgian subgoups are presented in
Table 13, and the distribution patterns themselves are shown in Figures 2 and 3 The two
Georgian male samples show very little diﬁ‘erence in mean humerus length but the
females display a statistically signiﬁcant mean diﬂ‘erence of 4.9 mnn, with the Spitalﬁelds

females larger (2 = 1.710, p < .05).

Table 13 Paired Georgian Humerus Length Summary Statistics

 

Spitalﬁelds Spitalfields St. Bride’s St. Bride’s
Males Females Males Females
Mean 321.3 293.9 320.9 289.0
SE 2.92 2.24 2.75 1.76
SD 18.89 16.34 15.06 9.94
.95 Confidence 5.71 4.40 5.39 3.45
N 42 53 3O 32

 

As would be expected, the differences between the sexes irn the two Georgian samples are

highly statistically signiﬁcant (St. Bride’s z = 9.750, Spitalﬁelds z = 7.442).

97

 

 

  

 

 

 

 

   

HUMERUS LENGTH DISTRIBUTION
GEORGIAN MALES
sons
30".. .............................................................................................................................................
4
i 1
m -1 ..................................................................................................................................
3
x i
10%.. ............................................ , ............................................. . ..................................................
\
k
“'2 310'33‘0 '330 so 'sdo‘sio
LEVGTHﬂnm)
a—ST. amass mm; + SPITALFIELDS (Ir-42)

 

 

 

Figure 2 Humerus Length Distribution - Georgian Males

 

HUMERUS LENGTH DISTRIBUTION
GEORGIAN FEMALES

 

 

 

 

 

 

 

 

x“ n 1 L n r r
270 '230'31’0 33b '330'310'330'410
LBUGTHnnm)

~I-ST. BRIDE‘S (W2) —ar- SPITALFELDS (II-ﬂ)

 

 

 

Figure 3 Humerus Length Distribution - Georgian Females

98

Table 14 Intact Georgian Humerus Length Summary Statistics

 

Spitall'relds Spitalfields St. Bride’s St. Bride’s
Males Females Males Females
Mean 322.8 293.1 320.6 289.9
SE 2.96 2.46 2.94 1.88
SD 17.77 16.86 15.28 8.83
.95 Conﬁdence 5.81 4.82 5.76 3.69
N 36 47 27 22

 

 

Table 14 partitions intactl9 humerus length summary statistics by sex between the
Georgian goups. The reduction in the study sample size in the irntact goups is
associated with a slight change in summary statistics when compared with the paired
bones. Nonetheless, the Spitalﬁelds and St. Bride’s males are very similar to each other,
likewise, the females of the intact goup present the same general pattern as they do in
paired humeri.

Table 15 pools the Georgian males and females for comparison of the paired and
intact samples. The pooled goups display very similar summary statistics, which
suggests that there is little diﬁ‘erence in the character of the more well preserved intact
long bones when compared with the less well preserved paired bones. The mean

diﬂ‘erence in humerus lengths between the paired humeri samples and the intact humeri

 

1"The intact sample is the subset of the paired sample for which head diameter and
biepicondylar width were also observed.

99

Table 15 Pooled Georgian Humerus Length Summary Statistics

 

Paired d' Paired 9 Intact o‘ Intact 9
Mean 321.1 292.1 321.8 292.1
SE 2.04 1.56 2.10 1.78
SD 17.29 14.40 16.66 14.79
.95 Confidence 3.99 3.06 4.11 3.49
N 72 85 63 69

 

samples is negligible—well within the standard error of each sample.

Among the Georgian males, a reduction in sample size ﬁ'om 72 to 63 is associated
with only a 0.7 mm difference in mean humerus length. For the females, the sample is
reduced from 85 to 69, with virtually no change in mean humerus length. This is a
noteworthy ﬁndirng, for it suggests that bones are not likely to be diﬁ‘erentially preserved
on the basis of their size. It also argues against the assumption that larger and more
robust bones would be less likely to suﬁer ﬁom the eﬂ’ects of poor preservation than
smaller bones. Ifthat were truly the case, then the intact sample would show a larger
mean length value than the paired sample.

Table 16 presents a comparable picture of the sununary statistics of humerus
length for the paired and the intact humeri of the Forensic Data Bank. The results are
consistent with those of the Georgians, in that the difference in mean humerus length

between the paired and intact samples is negligible.

100

Table 16 Forensic Data Barnk Humerus Length Summary Statistics

 

Paired d' Paired 5? Intact d' Intact 5?
Mean 334.5 306.7 334.2 308.2
SE 1.97 1.80 2.05 1.77
SD 18.20 14.37 18.35 13.44
.95 Confidence 3.87 3.52 4.02 3.46
N 85 64 80 58

 

In Figure 4 humerus length distributions of the pooled Georgian males are
displayed against the Forensic Data Bank males; Figure 5 does the same for the females.
The diﬂ'erences between the 18th and 20th century samples are striking for both sexes.
The mean male humerus length for the Forensic Data Bank is nearly 13 mm geater that
for the Georgians (z = 4.73). Likewise, the Forensic Data Bank females show a mean
humerus length approximately 15 mm geater than the Georgians (z = 6.13).

The diﬂ‘erence between the males and females of the Forensic sample parallels
that of the two Georgian goups, in that the males display a signiﬁcantly larger mean
humerus length (2 = 10.44). It is not surprising that the males of each of the three study
samples show a substantially larger mean humerus length than their female counterparts.
It is noteworthy, however, that the male-female diﬂ‘erence in means for the Forensic Data
Bank sample (27.8 mm) is slightly smaller than for the pooled Georgians (29.0 mm),
even though the Georgian humeri are smaller in length overall than the Forensic Data

Bank humeri.

101

 

HUMERUS LENGTH DISTRIBUTION

 

 

 

 

 

 

 

 

 

 

IDENTIFIED MALES
sent,
26% f,
It A
\n
gm f \ x
315% ,[ .1 \s
8 : / ,’ \r l‘
was, , ,
1 / 1/ \ \\
a,“ ..
on W J! W\
'2 2 2 ’sﬁjssa'ssa's s so
LENGTI-Hmm)

i-LONDONMIH) «kWh-85)

 

 

 

Figure 4 Humerus Length Distribution — Identiﬁed Males

 

HUMERUS LENGTH DISTRIBUTION

 

 

 

 

 

 

 

 

 

 

IDENTIFIED FEMALES

36%
: h
1 I \

30% l t
: r \
A. r

23%, 1
a: / A r

20%, i \
1 / f 1

315*: I i
* 1: J I \

10% ,1 l

i / I \ L\
a“ 7A I R
1 / \ \
my ’1 l n a a a a a
2 2 2 so 3 ‘3103330'410
LENGTI-Hmm)

fLONDON (n-UG) *W(M)

 

 

 

Figure 5 Humerus Length Distribution — Identiﬁed Females

102

Humerus Head Diameter—HH

The diameter of the humerus head typically is employed by osteologists as a tool
for sexing the humerus. As expected, in both Georgian goups the diﬂ'erences between
the male and female means are statistically signiﬁcant (St. Bride’s: t = 10.47, df= 43,
p < .0001; Spitalﬁelds: t= 12.06, df= 70, p < .0001). The summary statistics for vertical

head diameters in the Georgian goups are listed in Table 17.

Table 17 Georgian Humerus Head Diameter Summary Statistics

 

 

Spitall'relds Spitalfields St. Bride’s St. Bride’s
Males Females Males Females
Mean 46.8 40.9 46.9 41.0
SE 0.39 0.30 0.47 0.31
SD 2.35 2.07 2.47 1.43
.95 Confidence 0.77 0.59 0.93 0.60
N 36 47 27 22

 

 

In the case of both the males and females, the mean head diameter is only slightly
larger (0.1 mm) in the St. Bride’s humeri than in their Spitalﬁelds counterparts, and this
is not statistically signiﬁcant for either the males (t = 0.061, df= 55, p = 0.95) or females
(t = .197, df= 57, p = .84). For the Georgians the male and female curves intersect
between 43 and 44 mm (Figures 7 and 8). In contrast, the humerus head distribution
curves for the Forensic Data Bank sample intersect between 46 and 47 mm. Just as there
is a signiﬁcant diﬂ‘erence between the Georgian males and females, the mean head

diameter of the Forensic males and females is also signiﬁcantly different (2 = 16.18).

103

 

HUMERUS HEAD DISTRIBUTION
GEORGIAN mes

 

 

 

 

 

 

45 60
(RIGHT - LET) mm
+812 BRIDE'S (mm + SPITALFIELDS (mad)

 

 

 

Figure 6 Humerus Head Distribution — Georgian Males

 

 

 

 

 

 

 

HUMERUS HEAD DISTRIBUTION
GEORGIAN FEMALE

35% 0
316% 43
* I,
”9‘ 1.
5% 2’

ass-i ‘

36 40 45 so 55 so

(RIGHT- LEFT) nun
... ST. BRIDE'S (n-22)-.- SPITALFIELDS (nu-41)

 

 

 

Figure 7 Humerus Head Distribution — Georgian Females

104

Summary statistics for the pooled Georgian males and females are compared with
those of the Forensic Data Barnk in Table 18. The average head diameter for Forensic
Data Bank males is a statistically signiﬁcant 2.2 nnrn larger than that of the pooled .
Georgian males (2 = 5.51). For the females the mean diﬁ‘erence in humerus head
diameter is 1.7 man, which is also signiﬁcant (2 = 4.57). The distribution patterns of male
head diameter between the Forensic Data Bank and the pooled Georgians are oﬂ’set by
approximately two millimeters (Figure 8). For the Georgian females (Figure 9) the range
of variation is quite narrow, with over ﬁfty percent maintaining a diameter between 39

and 41 m.

Table 18 Humerus Head Diameter Summary Statistics

 

Georgian Georgian Forensic Forensic
Males Females Males Females
Mean 46.9 40.9 49.1 42.6
SE 0.30 0.23 0.29 0.29
SD 2.38 1.88 2.57 2.18
.95 Conﬁdence 0.59 0.44 0.56 0.56
N 63 69 8O 58

 

105

 

HUMERUS HEAD DISTRIBUTION
IDEN‘ITFIED MALES

 

 

 

 

 

 

 

1 \
1. / j x
ens“ ‘3
11 A \ A
oat: “
as 40 45

50
(RIGHT- LH-‘D nun

+LONDON (II-83) -a- FORENSIC (Is-80)

 

 

 

 

Figure 8 Humerus Head Distribution — Identiﬁed Males

 

HUMERUS HEAD DISTRIBUTION
IDENTIFIED FEMALES

 

1
259“
1

 

 

 

 

 

 

 

‘ \
l I 1 a
59‘ i l ..... 2
1 1 \
I A A-
036 r

45 50 66
(RIGHT-LEI) nun

...LONDON (nu-s9) .... FORENSIC (rt-es)

 

 

 

 

Figure 9 Humerus Head Distribution — identiﬁed Females

l 06
Humerus Biepicondylar Width—HR
The biepicondylar width of the humerus has also been identiﬁed as a useful
univariate characteristic for distinguishing between the sexes. Unlike the diameter of the
humerus head, however, biepicondylar width is subject to modiﬁcation in size and shape
associated with the role of the epicondyles as points of attachment for muscles that ﬂex
the wrist and pronate the forearm. For this reason, it is expected that this measurement

would manifest a geater range of variation within a sample.

Table 19 Georgian Humerus Biepicondylar Width Summary Statistics
—

 

Spitalﬁelds Spitalfields St. Bride’s St. Bride’s
Males Females Males Females
Mean 61.9 54.3 61.8 53.7
SE 0.50 0.48 0.63 0.52
SD 3.00 3.30 3.28 2.44
.95 Confidence 0.98 0.94 1.24 1.02
N 36 47 27 22

 

 

Statistics for humerus biepicondylar width in the Georgian goups are
summarized irn Table 19, and the distributions are gaphically represented in Figures 10
and 11. In contrast with the humerus head diameter, the mean values of biepicondylar
width are slightly higher in the Spitalﬁelds sample than in their St. Bride’s counterparts.
In the case ofthe males, the difference of0.1 mm is negligible (1 = .167, df= 53,p = .87),

as is the difference of 0.6 mm between the Georgian females (t = .878, df= 54, p = .38).

107

 

HUMERUS BIEPI WIDTH DISTRIBUTION
GEORGIAN MALES

 

 

 

 

 

 

48 60 66 60 66
(RIGHT- LEFT) mm
-— ST. BRIDE‘S (n-z7)-.- SPITALFIELDS (In-08)

 

 

 

Figure 10 Humerus Biepi Distribution — Georgian Males

 

HUMERUS BIEPI WIDTH DISTRIBUTION
GEORGIAN FEMALES

 

 

 

 

45 50 56

60 66
(RIGHT- LE-‘D mm
+ ST. BRIDE'S (m22)+ SPITALFELDS (nu-47)

 

 

 

Figure 11 Humerus Biepi Distribution — Georgian Females

The diﬁ’erernces between the sexes within the two Georgian goups are highly

108
statistically signiﬁcant, however (St. Bride’s t = 9.89, df= 47, p < .0001; Spitalﬁelds
t = 10.96, df= 79, p < .0001). The diﬁ‘erence between the sexes is also signiﬁcant within
the Forensic sample (2 = 18.66).
Biepicondylar width statistics for the pooled Georgian males and females are
shown in Table 20, as well as for the Forensic Data Bank nnales and females. The
Forensic sample is relatively larger for both sexes, and the diﬂ‘erence in mearns between

the Georgians and the twentieth-century sample is statistically signiﬁcant. Among the

Table 20 Humerus Biepicondylar Width Summary Statistics

 

Georgian Georgian Forensic Forensic
Males Females Males Females
Mean 61.9 54.1 64.8 55.7
SE 0.39 0.37 0.36 0.33
SD 3.10 3.04 3.20 2.52
.95 Conﬁdence 0.77 0.72 0.70 0.65
N 63 69 80 58

 

males, the diﬂ’erence in the mean biepicondylar width is 2.9 mm (2 = 5.51); among the
females the diﬂ‘erence between the means is 1.6 (2 = 3.15, p < .001). These diﬁ‘erences in
distribution pattern are evident in Figures 12 and 13, as well. An intriguing similarity
between the Georgian females is apparent in both the distribution polygons for humerus
head diameter and biepicondylar width. That is the high unimodal peak of the
distribution curve, which appears much less prominent in the Georgian males, the

Forernsic Data Bank nnales, and the Forensic Data Bank females.

109

In the St. Bride’s sample, no female shows a width larger than 57 mar, although a
few of the males have smaller biepicondylar width than that. There is slightly more
overlap within the Spitalﬁelds sample, with several of the females displaying a
biepicondylar width geater than 57 mm. The distribution patterns of the Forensic Data
Bank sample show a comparable pattern of overlap, but in this case the male and female

distributions intersect at a value of approximately 60 mm.

110

 

 

 

 

 

 

 

 

 

 

HUMERUS BIEPI WIDTH DISTRIBUTION
IDENTIFIED MALES
26% ‘
gm 3’
3am
at 3
0,23
46 50 66 60 65 70 75
(RIGHT - LET) nun
+LONDON (n-63) + FORENSIC (rt-M)

 

 

 

Figure 12 Humerus Biepi Distribution — Identiﬁed Males

 

HUMERUS BIEPI WIDTH DISTRIBUTION
IDENTIFIED FEMALES

 

 

 

 

 

 

 

 

 

+LONDON (M) + FORENSIC (II-58)

 

 

 

Figure 13 Humerus Biepi Distribution — Identiﬁed Females

110

 

HUMERUS BIEPI WIDTH DISTRIBUTION

 

 

 

 

 

 

 

 

IDENTIFIEDMALES
25!“
11
2036*
it
Eran" H
1 I‘
s 1‘ /.
10% i
I8 1. II
1 K 1
“1: /\ gr! 5 \
d 1 \
05-"
as as

+LONDON (II-63) -a- FOREISIC (II-80)

 

 

 

Figure 12 Humerus Biepi Distribution — Identiﬁed Males

 

HUMERUS BIEPI WIDTH DISTRIBUTION
IDENTIFIED FEMALES

 

 

 

 

 

 

 

 

 

+LONDON (M) + FORWIC (M)

 

 

 

Figure 13 Humerus Biepi Distribution — Identiﬁed Females

1 1 1
Femur Length—FL

As was the case for the humerus, femur length assessments in this study reﬂect
two series. The larger paired sample includes those femora for which maximum length
measurernernts are available. The subset of that series for which biepicondylar width and

head diameter measurements were also available is referred to as the intact sample.

Table 21 Paired Georgian Femur Length Summary Statistics

 

Spitalﬁelds Spitalfields St. Bride’s St. Bride’s
Males Females Males Females
Mean 447.1 408.7 455.3 412.1
SE 5.77 3.92 5.05 3.41
SD 24.49 22.54 30.30 20.16
.95 Confidence l 1.31 7.69 9.90 6.68
N 18 33 36 35

 

Summary statistics for the maximum lengths of paired femora in the Georgian
samples are listed in Table 21. On the basis of the total range of paired femora for which
paired maximum lengths are available, the St. Bride’s males (Figure 14) and females
(Figure 15) appear to be somewhat larger than their Spitalﬁelds counterparts, but the
diﬂ‘erences are not statistically signiﬁcant (males t = 1.07, females 2 = .646). The
diﬂ‘erences between the sexes are substantially diﬂ‘erent, as expected (Spitalﬁelds

t= 5.50, df= 33, p < .0001; St. Bride’s t= 7.10, df= 61, p < .0001).

 

112

 

 

FEMUR LENGTH DISTRIBUTION
GEORGIAN MALES
26% .
it.
20" ............................................................................... g..\\. .............................................................
i’ '\
9"4 ‘ 7—1
E15“ ........................................ , ‘t\ l' ‘1‘ ..............................................
I n l n
3 ll \\ i \n
I 4 I A
*fa,‘ ............................................. l , .............. 1 ......... i ................... i ..............................................
I r ‘
a ........................................... 7'. hit ........................ a. ....... ﬂ ..........
1' ‘1 ’1 '1
m‘ A ‘\I ‘I II
360 380 H 4 4' I 460 5"
LENGTH (mm)

 

 

 

 

{-ST. BRIDE'S (II-36) + SPITALFIELDS (M8)

 

Figure 14 Femur Length Distribution — Georgian Males

 

 

FEMUR LENGTH DISTRIBUTION
GEORGIAN FEMALES

 

 

 

 

  

 

Elsa Tsdoisdo '420 'do '460 rule '660 '520 '540
LEIGH-Harm)
+STZBRIDE'S(n-36)-a-SPITALF£LDS(M)

 

Figure 15 Femur Length Distribution — Georgian Females

 

113

Table 22 Intact Georgian Femur Length Summary Statistics

 

Spitalfields Spitalfields St. Bride’s St. Bride’s
Males Females Males Females
Mean 446.7 411.9 447.8 410.9
SE 6.11 3.97 6.07 4.20
SD 25.18 20.63 29.75 18.29
.95 Confidence 11.97 7.76 l 1.90 8.22
N 17 27 24 19

 

As is the case with the humeri, a subset of intact femora was extracted from the
paired femora data set. Table 22 partitions the intact femur length summary statistics
between the Georgian samples. A comparison with the Figures for the paired femora
shows a geater diﬂ‘erence in two of the subgoups than was the case with the humeri.
The Spitalﬁelds females show a diﬁ‘erence in mean femur length between the paired and
intact samples of 3.2 mm. Although this is within the standard error of the mean
calculated for the samples, it is twice the diﬂ‘erence for either the St. Bride’s females or
the Spitalﬁelds males. More remarkable is the difference of 7.5 mm between the mean
length of paired and intact femora of the St. Bride’s males. Tlnis ﬁgure is well in excess
of the standard error of the mean that was calculated for either of the samples.

When the Georgian males and females are pooled for comparison of the paired
arnd irntact samples, the pattern is still evident (Table 23). It is notable that, for the males,

the intact femora show a smaller mean length than the paired femora. This is the

l 14
opposite of what one rrnight expect if the expectation is that larger bones are more likely
to survive intact than smaller bones. The pattern is reversed for the females, but the
magnitude of diﬂ‘erence in mean femur lengths is quite small in contrast with that of the

males.

Table 23 Georgian Femur Length Summary Statistics

 

Paired d' Paired 9 Intact d' Intact 9
Mean 452.6 410.5 447.4 411.5
SE 3.88 2.58 4.31 2.87
SD 28.53 21.25 27.62 19.5
.95 Conﬁdence 7.61 5.05 8.45 5.63
N 54 68 41 46

 

Summary statistics of femur length for the paired and intact femora of the
Forensic Data Bank sample are listed in Table 24. Unlike the Georgians, the Forensic
Data Bank samples show consistency between mean femur lengths in the paired and
irntact samples. When the Georgian males are pooled, they display a signiﬁcantly smaller
femur length than the Forensic Data Bank males (2 = 4.61)(Figure 16).

There is also a distinct pattern of birnodality in the femur length distributions of
the Georgian males which is not evident in the Forensic sample. That is, there are two
distinct and roughly equivalent peaks separated by at least two measurement intervals;
this suggests that the apparent pattern reﬂects an underlying bimodal distribution and is

not simply an artifact of how observations are placed into interval categories. The

115

Table 24 Forensic Data Bank Femur Length Summary Statistics

 

Paired d' Paired 9 Intact d' Intact 9
Mean 473.9 437.5 473.9 437.6
SE 2.50 2.59 2.76 2.67
SD 24.09 20.85 25.00 20.35
.95 Conﬁdence 4.90 5.07 5.41 5.24
N 93 65 82 58

 

diﬂ‘erence in lengths between the 18th and 20th century samples are also evident for the

females (Figure 17), but the distribution patterns for the two goups are clearly unimodal.

116

 

FEMUR LENGTH DISTRIBUTION
IDENTIFIED MALES

 

 

 

 

 

 

 

 

 

+LONDON (W) -t- FMENSIC (n43)

 

 

 

Figure 16 Femur Length Distribution — Identiﬁed Maloes

 

FEMUR LENGTH DISTRIBUTION
IDENnFIED FEMALES

 

 

 

 

 

 

 

 

 

LENGTH (mm)

-I- LONDON (II-68) 1r- FOREWIC (II-65)

 

 

 

Figure 17 Femur Length Distribution — Identiﬁed Females

117

Femur Head Diameter—FH

Table 25 Georgian Femur Head Summary Statistics

 

Spitalﬁelds Spitalﬁelds St. Bride’s St. Bride’s
Males Females Males Females
Mean 48.0 41.8 46.8 41.5
SE 0.74 0.51 0.58 0.52
SD 3.06 2.67 2.84 2.29
.95 Conﬁdence 1.46 1.01 1.14 1.03
N 17 27 24 19

 

It is interesting to note that while the St. Bride’s males and females presented a
larger mean femur length than their counterparts from Spitalﬁelds, the patterrn is reversed
for the diameter of the femoral head (Table 25). The diﬂ'erences, however, are not
statistically signiﬁcant (males t = 1.31, af= 33,p = .198; females t = .377, df= 42,

p = .708). When the diﬂ‘erences are presented gaphically, the males ﬁ'om Spitalﬁelds
display a modal value (50 mm) which is three millimeters geater than that of the St.
Bride’s males (47 mm) (Figure 18). In contrast, the St. Bride’s females display a mode
(44 mm) which is only one millimeter geater than their Spitalﬁelds counterparts
(Figure 19).

There are, however, statistically signiﬁcant diﬁ‘erences between the sexes in all
tlnree of the study samples. This is not surprising, given the popularity of femoral head
diameter as a univariate technique for determining sex from skeletal material (Spitalﬁelds

t== 6.90, df= 31, p < .0001; St. B1ide’st= 6.74, df= 41, p < .0001; Forensic z = 17.30).

118

Table 26 Femur Head Diameter Summary Statistics

 

Georgian Georgian Forensic Forensic
Males Females Males Females
Mean 47.3 41.7 48.9 42.5
SE 0.46 0.37 0.28 0.25
SD 2.96 2.50 2.52 1.91
.95 Conﬁdence 0.91 0.72 0.55 0.49
N 41 46 82 58

 

When the pooled Georgian males and females are compared with the modern
sample of the Forensic Data Bank, there are marked differences in mean femur head
diameter (Table 26). The average head diameter for the Georgian males is a statistically
signiﬁcant 1.5 mm less than that of the Forensic Data Bank (2 = 3.015, p < .005); the
diﬂ‘erence between the Georgian and Forensic females is less than one millimeter, but
still signiﬁcant (2 = 1.740, p < .05). The apparent bimodal distribution that was present
in the Georgian males femur length distributions also appears in their femur head
distribution, again in contrast with the Forensic sample (Figure 20). For the females, the

distribution patterns of the Georgians and the modern goup are similar (Figure 21).

119

 

FEMUR HEAD DISTRIBU'HON
GEORGIAN muss

 

 

 

 

 

 

 

 

 

36 40 w on

45 so
(RIGHT-Hm

 

+ ST. BRIDE'S (II-24) + SHTALFIELDS (M17)

 

 

Figure 18 Femur Head Distribution -— Georgian Males

 

FEMUR HEAD DISTRIBUHON
GEORGIAN FEMALES

 

..................................................................................................

 

 

 

 

    

45 50 66
(RIGHT-LE1) mm

+ ST. ”mm-19b.- SPITALFIELDSMIID

 

 

 

Figure 19 Femur Head Distribution — Georgian Females

120

 

 

 

 

 

 

 

 

FEMUR HEAD DISTRIBUTION
IDENTIFIED MALES
m1
m3
2 x
15,“ f/ \
a , \
3mg ,1 ﬂ
21 ,1 V \
are / :1
3* Z :11 \j’jkx
0964 ~

35 40 55 00

45 50
(RIGHT - LEFT) mm
+LONDON (Ir-41) + FORENSIC (II-82)

 

 

 

Figure 20 Femur Head Distribution — Identiﬁed Males

 

 

 

 

 

 

 

 

 

FEMUR HEAD DISTRIBUTION
IDENTIFIED FEMALES
we
‘r
20% ‘[
2 8.
grew \\
17 \
3w“ 1 \|
at 1 \ k
5,,1 ll )1
“a: V V
35 4o 45 so 55 so
(RIGHT -LE=1) mm
...Louoou (nus) ..- FORENSIc (mas)

 

 

 

Figure 21 F emur Head Distribution — Identiﬁed Females

121
Femur Biepicondylar Width—FR
Because the epicondyles of the femur are the sites of attachment for ligaments
rather than tendons, they are less likely to display osteophytic projections that would
corrupt measurements. However, in the Georgian skeletal collections a large number of
individuals suﬂ‘ered from disintegration of the distal ends of the femur that rendered them

unsuitable for measurement. This contributed to a profound reduction in the number of

intact paired femora available for observation.

Table 27 Georgian Femur Biepicondylar Width Summary Statistics
—

 

Spitalfields Spitalﬁelds St. Bride’s St. Bride’s
Males Females Males Females
Mean 80.9 72.5 80.8 73.2
SE 0.84 0.75 0.98 0.77
SD 3.47 3.90 4.78 3.35
.95 Conﬁdence 1.65 1.47 1.91 1.51
N 17 27 24 19

 

 

Unlike the case with femoral head diameter, the biepicondylar width of the femur
diﬁ‘ers very little between the males of St. Bride’s and Spitalﬁelds (Table 27). When the
distributions are presented graphically, the shapes of the patterns diﬁ‘er signiﬁcantly ﬁ'om
the relatively unimodal (or bimodal) distributions of the bone length and head diameter.
Particularly in the case of the males, there is a broad pattern of variation in biepicondylar

width (Figure 22). In contrast, the females display patterns of birnodality (Figure 23).

122

 

 

 

 

 

 

 

 

FEMUR BIEPI WIDTH DISTRIBUTION
GEORGIAN MALES

26% ‘

I

20%:
is: i
I I‘
3,0,1 Ink 3.1 {'1

* , In

“I I\ M‘ ‘ .
1’ If s s s I\
1 I I

as.

00 65 70 75 U 85 so as 100
(RIGHT-LET) mm
-- ST. BRIDE'S (rt-24) + SPITALHELDS (II-17)

 

 

 

Figure 22 Femur Biepi Width Distribution - Georgian Males

 

FEMUR BIEPI WIDTH DISTRIBUTION
GEORGIAN FEMALES

 

 

 

 

 

 

 

 

 

 

 

00 66 70 76 00 85 90 95 100
(RIGHT- LEFT) mm

... sr. amass (Ir-19) ... SPITALFIELDS (nu-21)

 

 

 

Figure 23 Femur Biepi Width Distribution - Georgian Females

123

As was the case for the diameter of the femoral head, biepicondylar widths
manifest a statistically signiﬁcant difference between the sexes in each of the study
samples (Spitalﬁelds t = 7.45, df= 37, p < .0001; St. Bride’s t = 6.08, af= 40, p < .0001;

Forensic z = 15.79).

Table 28 F emur Biepicondylar Width Summary Statistics

 

Georgian Georgian Forensic Forensic
Males Females Males Females
Mean 80.8 72.8 85.7 75.0
SE 0.66 0.54 0.51 0.45
SD 4.24 3.66 4.60 3.39
.95 Confidence 1.30 1.06 1.00 0.87
N 41 46 82 58

 

Although the males of the two Georgian groups are very similar with respect to
their mean biepicondylar widths, Table 28 and Figure 24 show that when they are pooled
they contrast markedly with the males in the Forensic Data Bank sample. The diﬁ‘erence
is statistically signiﬁcant (2 = 5.83). The same is true for the females (Figure25)

(z = 3.198,p < .001).

124

 

 

 

 

 

 

 

 

 

 

FEMUR BIEPI WIDTH DISTRIBUTION
IDENTIFIED MALES
209‘ ‘
I
155“
10% ‘ I If
3 I I I ‘1’ I
g I I I
‘ ’r' I a
’ 1' \ / \
0% II I \ l \ m.
on ‘5 70 75 80 85 M as too
(RIGHT-L531) mm
+LONDON (II-41) + FMENSIC (II-82)

 

 

Figure 24 Femur Biepi Width Distribution - Identiﬁed Males

 

FEMUR BIEPI WIDTH DISTRIBUTION
IDENTIFIED FEMALES

 

 

...:p
a»

0""-

 

-a
i
a»
a.

~1b-

)d_——-’"'

l A - A A
~
-‘
o"
f
--

‘

i
- ~F-~
~

3 1’ N
* 1
an

60 85 70

I.

75 M 85
(RIGHT - LED mm
-0- LONDON (M) -o- FOREmn (II-ﬂ)

p
b

 

 

 

M 95 100

 

 

Figure 25 Femur Biepi Width Distribution -- Identiﬁed Females

 

125

Summary

The distribution patterns for each of the six individual measurements indicate that
there is a very clear diﬁ‘erence in means between the sexes of all three groups. However,
the St. Bride’s males and females display a notably greater mean femur length than their
Spitalﬁelds counterparts, although the male/female distinction was still clearly evident
between the samples. For all six of the measurements, the pooled Georgians diﬁ‘er
considerably ﬁ'om the Forensic Data Bank ﬁgures, with the latter individuals
considerably larger. Again the magnitude of the diﬁ‘erences appear to be fairiy
consistent among the males and among the females.

For both the humeri and femora, the more well-preserved intact Georgian samples
did not show a greater mean length than the less well preserved paired samples. In the

humeri the diﬁ‘erence between the two samples was negligible, and for the femora the

Table 29 Summary: Diﬂ‘erences in Humerus Size Distribution Patterns

 

 

Maximum Vertical Head Biepicondylar
Length Diameter Width

z p z/I“ p z/t“ p

Georgian 6' .109 —'- .061 — .167 —

Georgian 9 1.71 <.05 .19‘7"I — .878" —
Spitalfields d'/ 9 7.44 <.0001 12.1‘ <.0001 11.0" <.0001

St. Bride’s d'l 9 9.75 <.0001 10.5“ <.0001 9.89" <.0001 ,

Forensic d'/ 9 10.4 <.0001 16.2 <.0001 18.7 <.0001
Georgian lForensic d' 4.73 <.0001 5.51 <.0001 5.51 <.0001
Georgian [Forensic 9 6.13 <.0001 4.57 <.0001 3.15 <.001

126

paired samples were slightly longer. This is not consistent with the idea that larger (and

supposedly more robust) bones are better preserved than smaller bones.

Table 30 Surrunary: Diﬂ‘erences in Femur Size Distribution Patterns

 

 

 

Maximum Head Biepicondylar
Length Diameter Width

M" p z/t* p z/t" p

Georgian 6' 1.07" -— 1.31" — .080“ —

Georgian 9 .646 — .377“ — .695“ —
Spitalﬁelds d'/ 9 5.50“ <.0001 6.90“ <.0001 7.45“ <.0001
St. Bride’s o‘/ 9 7.10 <.0001 6.74"I <.0001 6.08“ <.0001
Forensic d'/ 9 10.1 <.0001 17.3 <.0001 15.8 <.0001
Georgian / Forensic d' 4.61 <.0001 3.02 <.005 5.83 <.0001
Georgian [Forensic 9 7.41 <.0001 1.74 <.05 3.20 <.001

127

ASYMNIETRY DIRECTION

Results of assessments of asymmetry direction are summarized in a series of
Tables in this section. Each Table partitions asymmetry into one of three nominal
categories: left-dominant (represented by “L” in the Table), symmetrical (represented by
“O” in the Table), and right-dominant (represented by “R” in the Table). Following the
same presentation strategy as the previous section, the four Georgian subgroups are listed
in one Table, followed by a listing of the pooled Georgians in contrast with the Forensic
data. The chi-square ()8) statistic is used to test the statistical probability of

independence of the subgroups.

Humerus Length—EL

Direction of humerus length was assessed on the basis of two thresholds of
diﬁ‘erence. In the ﬁrst case, paired bones were considered symmetrical if the magnitude
of their diﬁ‘erences was coded as less than one millimeter. In the second case, they were
considered symmetrical if the diﬂ'erence in magnitude was coded as less than two
millimeters.

Figures 26 through 33 display the eﬁ‘ects of shifting the threshold for symmetry in
the study samples. In each Figure, the X-axis represents the assessment of asymmetry:
left-dominant (L), symmetrical (O), or right-dominant (R). Two series are plotted; the
solid series represents the number of individuals who display each dominance pattern
with a one-millimeter threshold, and the bashed series represents the number who

manifest the pattern with a two-millimeter threshold.

128

 

HUMERUS LENGTH ASYMMETRY
SPITALFIELDS MALES

 

 

r
I
I.
I
m 1.. .................................................................................................
I _—____E
I» =
o =
II =
” ... ............................................................................................ 4 = .........
o =
o =
" =
‘} —
m .. .................................................................................................. = ........
‘ #
o =
0 =
1’ =
’5 1p ------------------------------------------------------------------------------------------------- —, ........
0 =
v =
v =
4 =
.... = ..........
—
—
—
—
—
—
—
_
_
* .........
_
..—
—
_

 

 

 

 

 

 

 

 

 

 

 

 

 

O R
DIRECTION OF ASYMMETRY

[jutnmgxzmm

 

 

 

Figure 26 Humerus Asymmetry Direction - Spitalﬁelds Males

 

HUMERUS LENGTH ASYMMETRY
sr BRIDE'S MALES

 

 

..........

{I

OOFNDIVDUALS

 

 

 

 

 

 

 

O
DIRECTION OF ASYMMETRY
[:]>- 1 «ME» 2 nun

 

 

 

Figure 27 Humerus Asymmetry Direction - St. Bride’s Males

129

 

HUMERUS LENGTH ASYMMETRY
SPITALFIELDS FEMALES

 

 

...................................................................................................

.................................................................................................

 

 

 
 

 

 

 

 

IIIIIIIIIIIIIIIIIIHIHIHIIIllllll|llIlllllIIIIIIIIIIIIUIIIIU.

 

 

 

DIRECTION OF ASYMMETRY

[:I» 1 mung» 2mm

 

 

 

Figure 28 Humerus Asymmetry Direction -- Spitalﬁelds Females

 

HUMERUS LENGTH ASYMMETRY
sr BRIDE'S FEMALES

 

 

 

 

 

0 E

v E

15 w ............................................................................................ ...: ____.

. E

10 .L ................................................................................................ . E

3 I E

3 h :—

. 0 '——=

5 .1 .............................................................................................. 1 =

I E

I E

. r—r mgr. ————_
O

 

DIRECTION OF AS YMMETRY

 

[:P- 1 mung» 2 mm

 

 

Figure 29 Humerus Asymmetry Direction - St. Bride’s Females

130
In all cases, a number of individuals shift from the left-dominant and right-

dorrninant categories to the symmetrical category in the second series. The extent of the
shift varies, however, depending on the magnitude of asymmetry in the series.

Figures 26 and 27 contrast the relationship between the two thresholds for
Georgian males. One of the Spitalﬁelds males shifts from left-dominant to symmetrical,
and three shift from right-dominant to symmetrical. In the St. Bride’s males, two left-
dominants shiﬁ to the center, and two right-dominants shift to the center. The pattern is
even more pronounced among the Georgian females. 0f the 47 Spitalﬁelds females, only
one right-dominant and one left-dominant pair are shifted to the symmetrical category
(Figure 28). Of the 22 St. Bride’s females, only the single leﬁ-dominant pair is absorbed
with the increased threshold (Figure 29).

Because the Georgian males are similar in their asymmetry patterns, when they
are pooled their patterns do not appear to vary greatly (Figure 30); the same is true for the
females (Figure 31). However, the Georgians contrast markedly with their counterparts
from the Forensic sample. Among the Forensic males, nine make the shiﬁ from left-
dominant and nine make the shift from right-dominant to symmetrical (Figure 32).
Among the females, three shift ﬁ'om left-dominant to symmetrical, and ﬁve shift from
right-dominant to symmetrical (Figure 33).

Table 31 contrasts the four Georgian subgroups under the one-millimeter
threshold. The overall right—dominance in humerus length is striking, particularly in the
females. The St. Bride’s males are the only one of the subgroups with less than 75% of
the individuals right-dominant in humerus length. There are no statistically signiﬁcant

diﬂ'erences in asymmetry direction among any of the Georgian subsamples.

131

Table 31 Georgian Humerus Length Asymmetry Direction (2 1 mm)
—

”OF

I-l

 

St. Bride’s
Females

Spitalﬁelds Spitalﬁelds St. Bride’s
Males Females Males

3 8.3% 5 10.6% 4 14.8% 1 4.5%
1 2.8% 1 2.1% 3 11.1% 1 4.5%

32 88.9% 41 87.2% 20 74.1% 20 91.0%
36 47 27 22

Table 32 Combined Humerus Length Asymmetry Direction (2 1 mm)

”OF

 

Forensic
Females

Georgian Georgian Forensic
Males Females Males

7 11.1% 6 8.7% 29 36.3% 11 19.0%
4 6.3% 2 2.9% 12 15.0% 9 15.5%
52 82.5% 61 88.4% 39 48.8% 38 65.5%
63 69 8O 58

Whern the Georgian samples are pooled and compared with the ﬁgures from the

Forensic Data Bank, a signiﬁcant contrast becomes apparent: the latter individuals are

less side-dominant (Table 32). Fewer than 50% of the males from the Forensic Data

Bank sample are right-dominant, and only 65.5% of the females are right-dominant.

Likewise, the percentage of left-dominant individuals in the Forensic Data Bank samples

are substantially higher than is the case with the Georgians. The diﬁ’erences between the

Georgian and Forensic males are statistically signiﬁcant (x2 = 17.53, p <0.001); the same

132

is true for the relationslnip between the Georgian and Forensic females (x2 = 10.39,

p < .01).

Table 33 Georgian Humerus Length Asymmetry Direction (2 2 mm)

Spitalfields Spitalﬁelds St. Bride’s St. Bride’s
Males Females Males Females

2 5.6% 4 8.5% 2 7.4% 0 0.0%
5 13.9% 3 6.4% 7 25.9% 2 9.0%
29 80.6% 40 85.1% 18 66.7% 20 91.0%

36 47 27 22

 

 

”OI"

 

The revised ﬁgures for the Georgian samples under the two-millimeter threshold
are found in Table 33. As expected, with the larger threshold there is a net shiﬁ in the
number of individuals who are asymmetrical into the symmetrical category. Reduction
in the apparent right-dominance is slight; in all but the St. Bride’s males more than 80%
of the individuals still show right-dominance with the higher threshold. Under the
revised threshold there is still no statistically signiﬁcant diﬁ‘erence in humerus length
asymmetry direction between any of the Georgian subgoups.

When the pooled Georgians are compared with the Forensic Data Bank
individuals under the more revised threshold, the strong contrasts between the two goups
persist (Table 34). Although there is still a net right-dominance pattern in all of the
subgoups, the Forensic Data Bank males in particular display relatively little side-

dominance in asymmetry direction.

133

Table 34 Combined Humerus Length Asymmetry Direction (2 2 mm)

Georgian Georgian Forensic Forensic
Males Females Males Females

4 6.3% 4 5.8% 20 25.0% 8 13.8%
12 19.0% 5 7.2% 30 37.5% 17 29.3%
47 74.6% 60 87.0% 30 37.5% 33 56.9%
63 69 80 58

”OF

 

Because there is relatively less directionality in the Forensic sample, increasing
the symmetry threshold to two nnillimeters has an eﬁ‘ect of increasing the x3 values when
they are compared with the Georgians; there is a conconnitant increase in the calculated
probability that the two samples are independent. Using the ﬁgures in Table 35 the x3 for
Georgian and Forensic males is increased to 20.40 (p < .0001); for the females the x3 is
increased to 14.88 (p < .001). The implications of a shift in direction-of-asymmetry

patterns are discussed more fully in Chapter 6.

134

 

HUMERUS LENGTH ASYMMETRY
GEORGIAN MALES

 

A_A

 

...........................

I

IHllIlllllllllllIHill!lIlllllllllllllllllllllllllllllllllilll

.........

 

 

 

 

 

 

 

 

O
DIRECTION OF ASYMMETRY

[:j» 1 mung” 2 mm

 

 

 

Figure 30 Humerus Asymmetry Direction - Georgian Males

 

HUMERUS LENGTH ASYMMETRY
GEORGIAN FEMALES

 

 

 

 

 

 

70 .1
1I
.
{I
w 0 ...........................................................
‘ —
0 E
‘ =
«I =
w ... ........................................................................................ .I _____,
(I _—
.1 :—
1? ——
1I ——__
‘0 ......U = .........
.1 2
1p _—
{I =
1 =
x . ...................................................................................... —.
1 z
‘ —
II =
4b =
.m .1... ....................................................................................... = .........
{I :—
1» —-——____—-
In =
{I —
1a . .................................................................................................... = .........
‘ =
I 2.
‘ —-
a " [his—- ——

MECTION OF ASYMMETRY

D» 1 mg» 2mm

 

 

 

Figure 31 Humerus Asymmetry Direction - Georgian Females

135

 

HUMERUS LENGTH ASYMMETRY
FORENSIC OA rAGASE MALES

 

 

{I
x . .......................................... ....................

  

  

lllllllllllllllllll

lllllllllIllllllllllllIIIlllllIll!IIIIIHIIHIIIIHIHIIIHII

......................................

 
   

 

...............................

 

 

 

 

 

 

llllllllllIlllllllllillllllIlIIIllllllll
IIIIIIHIllll|lllIIIIIIIIIIIIIIIIIIIIII

 

 

L O
DIRECTION OF ASYMMETRY

[3x 1 mung» 2mm

 

 

 

Figure 32 Humerus Asymmetry Direction - Forensic Males

 

HUMERUS LENGTH ASYMMETRY
FORENSIC OA rAGASE FEMALES

 

 

.........

 

 

lIllllllllIIIIIIIIHIIIIIHIIHHlIIIIUIIIIIHIIIIIIIIIIIIIIHIII

 

 

 

 

 

 

O
DIRECTION OF ASYMMETRY

[3» 1 ﬂung» 2 mm

 

 

 

Figure 33 Humerus Asynunetry Direction - Forensic Females

136

Humerus Head Diameter—BB

Because the head of the humerus is relatively small in size, a one-millimeter
diﬂ‘erence in size between paired heads represents a fairly large amount Of asymmetry.
Therefore, in contrast with humerus length it would be expected that more individuals in
each of the study samples would fall into the symmetrical category. The striking pattern
for the Georgians is that the diﬂ‘erences appear to be geater between the groups, rather
than between the sexes. That is, head diameter is more symmetrical among the St.
Bride’s sample than the Spitalﬁelds sample. For the females the diﬂ‘erence in pattern is

statistically signiﬁcant (x2 = 7.026, p < .05).

Table 35 Georgian Humerus Head Asymmetry Direction

 

 

 

 

 

Spitalﬁelds Spitalﬁelds St. Bride’s St. Bride’s
Males Females Males Females
L 5 13.9% 5 10.6% 2 7.4% 3 13.6%
0 16 44.4% 23 48.9% 18 66.7% 17 77.3%
R 15 41.7% 19 40.4% 7 25.9% 2 9.1%
T 36 41.7-r 27 22

 

When the Georgian males and females are pooled, they appear to display similar
patterns in direction Of head diameter asymmetry. Just as there is little gender-related
diﬁ'erence in humerus head diameter among the pooled Georgians, there is also little
diﬁ‘erence between the males and the females Of the FOrensic Data Bank sample

(Table 36). Again, the trend suggests that population rather than sex trend appears when

137

the Georgians are compared with the Forensic sample, but none Of the diﬁ‘erences are

statistically signiﬁcant. Speciﬁcally, the latter goup displays double the prevalence of

Table 36 Combined Humerus Head Asymmetry Direction

Georgian Georgian Forensic Forensic
Males Females Males Females

 

”OI"

 

7 11.1% 8 11.6% 18 22.5% 14 24.1%
34 54.0% 40 58.0% 35 43.8% 28 48.3%
22 34.9% 21 30.4% 27 33.8% 16 27.6%
63 69 8O 58

leﬁ-dominance than the fornner. This ﬁnding is consistent with the asymmetry patterns

which appeared in the length Of the humeri. The contrast in these patterns point to the

one of the conundrums of asymmetry assessment. In the sense that a larger percentage of

the Georgians fall into the symmetrical category, it can be asserted that they are more

symmetrical than the modern sample. However, because the extent of directionality in

asymmetry patterrns is so much geater for the Georgians, it can also be said that the

Forensic Data Bank sample appears more symmetrical than the Georgiarns.

138

Humerus Biepicondylar Width—BB

As is the case with head diameter, biepicondylar width Of the humerus displays a
compelling population-related pattern among the Georgians (Table 37). The males and
females of the St. Bride’s display a similar pattern of variation in asymmetry direction,
with nearly 50% Of the goup showing a symmetrical pattern. In contrast, the Spitalﬁelds
males and females display geater levels of asymmetry. (The diﬂ'erences are not quite
statistically signiﬁcant at the p < .05 level. For the males, 1’ = 5.89, p w .0526; for the

females x3 = 4.70, p s .0952.)

Table 37 Georgian Humerus Biepicondylar Width Asymmetry Direction

 

 

Spitalfields Spitalfields St. Bride’s St. Bride’s
Males Females Males Females
L 9 25.0% 11 23.4% 4 14.8% 3 13.6%
0 7 19.4% 13 27.7% 13 48.1% 12 54.5%
R 20 55.6% 23 48.9% 10 37.0% 7 31.8%
_ T 36 47 27 22

 

When the Georgian males and Georgian females are pooled, they appear to
display comparable patterns of asymmetry (Table 38). However, irn this case the
Georgian and modern Forensic samples are very similar in asymmetry direction patterns,
with no diﬂ’erences approaching statistical signiﬁcance. This is a notable contrast with

the pattern for either humerus length or head diameter.

139

Table 38 Humerus Biepicondylar Width Asymmetry Direction
—

Georgian Georgian Forensic Forensic
Males Females Males Females

13 20.6% 14 20.3% 21 26.3% 8 13.8%
20 31.7% 25 36.2% 23 28.8% 17 29.3%
30 47.6% 30 43.5% 36 45.0% 33 56.9%
63 69 80 58

 

”OI.“

140

Femur Length—FL

As was the case with humerus length, direction of asymmetry patterns in femur
lengths are assessed using both a 1 mm and a 2 mm threshold. Because the femora, in
general, display a left-dominant pattern in length that is smaller in magnitude than the
patterns for humeri, shifting the threshold for labeling asymmetry results in a
considerable reduction in the number of individuals who are asymmetric. Table 39

describes how asymmetry direction patterns differ for the Georgian subgoups.

Table 39 Georgian Femur Length Asymmetry Direction (2 1 mm)
—

Spitalfields Spitalﬁelds St. Bride’s St. Bride’s
Males Females ' Males Females

10 58.8% 15 55.6% 15 62.5% 6 31.6%
1 5.9% 4 14.8% 1 4.2% 6 31.6%
6 35.3% 8 29.6% 8 33.3% 7 36.8%
17 27 24 19

 

”OP

As has been reported by other researchers, there is a general pattern Of leﬁ-
dominance in femur length asymmetry, but the pattern is much less one-sided than in the
humerus. The St. Bride’s females are the only subgoup for which there is a slight
pattern of right dominance which results in a statistically signiﬁcant contrast with the St.
Bride’s males (x2 = 7.01, p < .05).

The pooled Georgians do not display a signiﬁcantly diﬂ‘erent pattern of

asymmetry direction than the Forensic sample (Table 40). The contrast between the

141

Table 40 Combined F emur Length Asymmetry Direction (2 1 mm)

Georgian Georgian Forensic Forensic
Males Females Males Females

25 61.0% 21 45.7% 43 52.4% 33 56.9%
2 4.9% 10 21.7% 8 9.8% 4 6.9%

 

”O!"

14 34.1% 15 32.6% 31 37.8% 21 36.2%
41 46 82 58

eighteenth- and twentieth- century groups that appeared so striking for the humeri does
not seem to be the case when the patterns are compared for the femora

Because the femur is substantially larger than the humerus, a one millimeter
variation in length represents a smaller percentage difference, so it would not be
surprising if there were a smaller proportion of symmetrical femora than humeri in any of
the samples. However, this turns out not to be the case. In fact, among all the study

samples more femora are symmetrical than humeri, and when the symmetry threshold is

Table 41 Georgian Femur Length Asymmetry Direction (2 2 nnm)

Spitalfields Spitalﬁelds St. Bride’s St. Bride’s
Males Females Males Females

9 52.9% 14 51.9% 14 58.3% 3 15.8%
3 17.6% 10 37.0% 4 16.7% 10 52.6%

 

”O1“

5 29.4% 3 11.1% 6 25.0% 6 31.6%

17 27 24 19

142
raised to two millimeters, there is a relatively large shiﬁ into the symmetrical category
ﬁom both the right-dominant and left dominant sides. Table 42 shows the numbers and
percentages of the Georgians who fall into each direction-of-asymmetry category with
the higher asymmetry threshold.

The Spitalﬁelds males (Figure 34) and St. Bride’s males (Figure 35) display
remarkably similar femur length asymmetry patterns under the two tlnresholds, with a
shift Of one or two individuals from each side to the center under the higher tlnreshold. In
the case of the Spitalﬁelds females, ﬁve individuals shift ﬁ'om right-dominant to
symmetrical, and one individual shifts from left-dominant to symmetrical (Figure 36). In
the St. Bride’s females, tlnree leﬁ-dominants shift to symmetrical, as well as one right-
dominarnt (Figure 37). While the Georgian females appear to display a geater magnitude
Of asymmetry in the humeri, the pattern is reversed in the femora. In both Georgian
samples the males display geater asymmetry than the females, and in all cases the
pattern is left-dominant (Figures 38 and 39).

In assessing the humeri it was apparent that shitting ﬁom a one-millimeter to a
two-millimeter threshold increased the )8, associated with an increased statistical
probability that the samples are independent. There is a comparable phenomenon in
assessing relationships for femur length. The new threshold further accentuates the
right-dominance pattern in St. Bride’s female femora, which renders that sample
signiﬁcantly diﬁ‘erent ﬁ'om not only the St. Bride’s males (x2 = 9.23, p < .01), but also

the Spitalﬁelds females ()0 = 6.94, p < .05).

143

 

FEMUR LENGTH ASYMMETRY
SFIrALFIELOS MALES

 

 

 

....................................................

.........................................................

..aoeceseuuau--..----~oo--- —— ------------

 

 

 

 

O ‘ R
DIRECTION OFASVMME'IRY

 

 

 

 

[:j» 1 mung»- 2mm

 

 

 

Figure 34 Femur Asymmetry Direction - Spitalﬁelds Males

 

FEMUR LENGTH ASYMMETRY
ST GRIOE'S MALES

 

 

..................................................................................................

......................................................................................................

 

 

 
   

OOFNDIWU
‘

as a

‘1 .

...........................

...........................

 

 

 

 

llllllllllllllllIllllllllllllllllllllllllllllllllllllllillllllllllllll

 

 

R
DIRECTION OF ASYMMETRY

[3» 1 mung» 2mm

 

 

 

Figure 35 Femur Asynnrnetry Direction - St. Bride’s Males

144

 

FEMUR LENGTH ASYMMETRY
SFIrALFIELOS FEMALES
16

14 1, ........ I

 

...................................................................................................

 

 

 

 

 

 

 

 

 

 

 

 

 

 

E1» 1 mung» 2mm

 

 

 

Figure 36 Femur Asymmetry Direction - Spitalﬁelds Females

 

FEMUR LENGTH ASYMMETRY
sr BRIDE'S FEMALES

 

 

 

    

.........
—

.........

.........

 

 

 

 

  
   

 

 

 

L R

O
DRECTION OF ASYMMETRY

D» 1 mung)- 2mm

 

 

 

Figure 37 Femur Asynunetry Direction - St. Bride’s Females

145

 

FEMUR LENGTH ASYMMETRY
GEORGIAN MALES

 

.....................................................................................................

......................................................................................................

 

   
 

O R
DIRECTION OF ASYMMETRY

 

 

 

 

 

 

 

[3»- 1 mung» 2 IIIIII

 

 

Figure 38 Femur Asymmetry Direction - Georgian Males

 

FEMUR LENGTH ASYMMETRY
GEORGIAN FEMALES

 

 

 

 

 

 

 

 

 

 

 

O R
DIRECTION OF ASYMMETRY

D» 1 mung» 2mm

 

 

 

Figure 39 Femur Asymmetry Direction - Georgian Females

146

 

FEMUR LENGTH ASYMMETRY

 

 

 

 

FORENSIC DATABASE MALES
50
I:
40 .‘L ........ t ...............................................................................................................
3 1;
g” I —_
II g
’0 .3? ........ I g
I E
a { =

 

 

 

 

 

 

 

O R
DIRECTION OF ASYMMETRY

[:j» 1 mung” 2mm

 

 

 

Figure 40 Femur Asymmetry Direction - Forensic Males

 

FEMUR LENGTH ASYMMETRY
FORENSIC OA TABASE FEMALES

 

 

 

 

 

 

 

 

 

 

 

 

 

35 0
4+
{I
{I
m . ........ I ...................................................................
d
«I
{I I. I
0 =
u .. ......... = .......
o =
{I =
o =
t» =
m .... ........ m .........................................................
I
_
_
—
: n =
15 ‘p ........ —‘
{I —
.1 =
3 .. =
‘ {I =
’0 .1. ........ =
—
_
{I —
0 m
6 I ——
_
—
1 =
(I =
0 =
a I. =

 

 

 

 

O R
DIRECTION OF ASYMMETRY

[:j» 1 mug» 2mm

 

 

 

Figure 41 Femur Asymmetry Direction - Forensic Females

147

Table 42 Combined Femur Length Asymmetry Direction (2 2 mm)

Georgian Georgian Forensic Forensic
Males Females Males Females

23 56.1% 17 37.0% 35 42.7% 27 46.6%
7 17.1% 20 43.5% 32 39.0% 15 25.9%
11 26.8% 9 19.6% 15 18.3% 16 27.6%
41 46 82 58

”OF

 

In contrast with the Georgians, the Forernsic Data Bank individuals arerernarkably
similar in their side-dominance distribution patterns (Table 42; Figures 40 and 41). In
both sexes, a substantial number of individuals shiﬁ ﬁom left- or right-dominant to
symmetrical when the threshold is raised from 1 mm to 2 mm. The higher threshold
renders the Georgian and Forensic males statistically diﬁ‘erent (x2 = 6.14, p < .05), but
not the ferrnales. However, it also results in a signiﬁcant diﬁ‘erence between the pooled

Georgian males and the pooled Georgian females (x2 = 7.09, p < .05).

Femur Head Diameter—FE

It is not surprising that the head of the femur would appear relatively
symmetrical, much like the head of the humerus. What is striking, however, is that even
though the length Of the femur shows an obvious pattern of leﬁ-dominance, the right side

is dominant in femur head diameter (Table 43). The Georgians are remarkably similar in

148

their distribution patterns, and none of the diﬁ‘erences among them are statistically

signiﬁcant.

Table 43 Georgian Femur Head Asymmetry Direction

 

 

 

 

 

 

 

 

 

Spitalfields Spitalfields St. Bride’s St. Bride’s
Males Females Males Females
L 2 11.8% 1 3.7% 2 8.3% 2 10.5%
0 10 58.8% 12 44.4% 16 66.7% 13 68.4%
R 5 29.4% 14 51.9% 6 25.0% 4 21.1%
T I 17 27 24 19
Table 44 Connbined Femur Head Asymmetry Direction
Georgian Georgian Forensic Forensic
Males Females Males Females
L 4 9.8% 3 6.5% 11 13.4% 10 17.2%
0 26 63.4% 25 54.3% 44 53.7% 32 55.2%
R 11 26.8% 18 39.1% 27 32.9% 16 27.6%

 

'-1
t
3
on
N
III
on

The same right-dominant pattern appears in the Forensic Data Bank males and

females (Table 44), although it is less profound than in the pooled Georgians. There

does not appear to be any Obvious sex-related pattern Of variation in the direction of

149

femoral head asymmetry, and there are no statistically signiﬁcant diﬁ‘erences between

the Georgian and Forensic samples.

Femur Biepicondylar Width—FB

Similar to the case of femoral head diameter, biepicondylar width asymmetry in
the femur shows a pattern Of right-dominance in directionality among the Georgians,
which contrasts with the direction of dominance in femur length. The sole exception is
the St. Bride’s males, but the slight apparent right dominance could easily be an artifact
of sampling error (Table 45). Unlike the case for the head of the femur, there is a
substantially higher level of symmetry in the St. Bride’s males than in the other Georgian
subsamples, and the greatest proportion Of asymmetric individuals is seen in the

Spitalﬁelds males.

Table 45 Georgian Femur Biepicondylar Mdth Asymmetry Direction

Spitalfields Spitalfields St. Bride’s St. Bride’s
Males Females Males Females

3 17.6% 3 11.1% 4 16.7% 0 0.0%

4 23.5% 10 37.0% 17 70.8% 9 47.4%

10 58.8% 14 51.9% 3 12.5% 10 52.6%
7 27 24 19

NO!"

 

Because so few individuals in any Of the Georgian subgoups display leﬁ-

donninance, minimal cell-size requirements for a x3 comparison are not met and the usual

150

x3 assessment cannot legitimately be performed. However, if the left-dominant and
symmetrical categories are pooled, minimum cell-size requirements are met and it is
possible to perform a two-by-two 1’ comparison of the subgoups (with a Yates
correction for continuity, since there is only one degree of ﬁeedom). In this scenario, the
St. Bride’s male sample diﬁ'ers signiﬁcantly from each of the Spitalﬁelds males

(12 = 7.84, p < .01), Spitalﬁelds females (x2 = 7.17, p < .01), and St. Bride’s females

(x2 = 6.31,p < .05).

Table 46 Combined Femur Biepicondylar Width Asymmetry Direction

 

Georgian Georgian Forensic Forensic
Males Females Males Females

7 17.1% 3 6.5% 16 19.5% 12 20.7%

21 51.2% 19 41.3% 31 37.8% 26 44.8%
13 31.7% 24 52.2% 35 42.7% 20 34.5%
41 46 82 58

 

”01"

When the Georgians are pooled and compared with the Forensic Data Bank
individuals, the right-dominance pattern persists (Table 46). As was the case for the
femur head, there are no obvious diﬁ‘erences between the Georgiarns’ patterns and those

of the more recent sample, and none which are statistically signiﬁcant.

151

Summary

Table 47 Summary: Differences in Direction ofHumerus Asymmetry
_

Length Length Head Biepicondylar
(1 mm) (2 mm) Diameter Width

Georgian 6' — — - —
Georgian 9 — — <.05 —
Spitalﬁelds d'/ 9 — — — —
St. Bride’s d'/ 9 — — — —
Forensic d'/ 9 — — — —

Georgian / <.001 <.0001 — —
Forensic 01'

Georgian I <01 <00] — —
Forensic 9

 

The strong directional bias in the length of the humerus in the two Georgian
samples is apparent in comparison with their Forensic counterparts. Because the latter
goup is so much less directional, the differences between the goups are highly
signiﬁcant (Table 47).

When a two-millimeter symmetry threshold replaces the one-millimeter tlnreshold
the population diﬁ‘erences appear to be even more statistically signiﬁcant. There is no
signiﬁcant difference in length asymmetry directionality among any of the Georgian
subgroups.The only signiﬁcant diﬁ‘erence in direction of head asymmetry inexplicably
occurs between the Georgian females. At ﬁrst glance it seems surprising that there are no
statistically signiﬁcant differences among any of the subgoups with respect to

biepicondylar width of the humerus.

152

Table 48 Summary: Differences in Direction Of Femur Asymmetry

 

Length Length Head Biepicondylar

(1 mm) (2 mm) Diameter Width

Georgian 6' — — — <.01"I
Georgian 9- — <.05 — —
Spitalfields d‘/ 9 — — — —

St. Bride’s 01'! 9 <05 <01 — <.05*
Forensic d' I 9 -— — — —
Georgian IForensic o‘ — <.05 — —
Georgian / Forensic 9 — — — —

 

 

Table 48 describes the patterns of asymmetry direction among the measurements
of the femur. As was the case with the humerus length, increasing the symmetry
tlnreshold for the length of the femur results in more highly signiﬁcant diﬁ'erences among
some Of the study subgoups. The St. Bride’s males and females display the geatest

diﬁ‘erence in directionality patterns, but the Spitalﬁelds and St. Bride’s males diﬁ‘er

signiﬁcantly

153

ASYNINIETRY MAGNITUDE

In this section, the magnitude of asymmetry for each of the measurements is .
described in terms Of (1) signed values; (2) unsigned values; (3) signed percentage Of
total character size; and (4) unsigned percentage of total character size. For each
measurement, the study subgroups are ordinally ranked from the most symmetrical to the
least symmetrical on the basis of each Of these assessments of asymmetry.

In addition to the ordinal ranking, summary statistics are provided in tabular form
for the unsigned magnitude values of each measurement for the study subgroups.
Summary statistics include the mean, standard error Of the mean, standard deviation, and
95% conﬁdence level. The Mann-Whitney U statistic is employed to assess the
independence of the paired samples as a means for testing the core hypotheses Of the

study.

Humerus Length—BL

The magnitude of asymmetry in paired elements can be reported either in terms of
signed or unsigned values. When a number of positive and negative values are averaged,
the resulting mean is consistently smaller than if only unsigned asymmetry values are
averaged. Because of the strong directional nature Of humerus length asymmetry (as
identiﬁed in the previous section) there is a relatively small diﬁ‘erence between the means
reported with the signed and unsigned values. However, the extent Of the directionality"
will inﬂuence the amount of difference between the means.

Table 49 shows how the mean value of an asymmetry distribution changes when

signed and unsigned values are compared among the study samples. Those subgoups

154

Table 49 Mean Humerus Length Asymmetries-Signed vs. Unsigned

 

 

SIGNED UNSIGNED

Spitalﬁelds Males 3.57 3.90
Spitalﬁelds Females 5.09 5.58
St. Bride’s Males 3.57 4.47
St. Bride’s Females 4.66 4.72
Georgian Males 3.57 4.14
Georgian Females 4.93 5.26
Forensic Males 0.76 2.69
Forensic Females 1.61 2.61

—
which display the geatest level Of directionality (St. Bride’s females, for example) have
the smallest change in means; those which are less directional (Forensic males, for
example) have a much geater change in means.

The extent of the directionality among the Georgians is displayed irn the Figures
representing the frequency distributions of the asymmetry (Figures 42 and 43). The
distributions of humerus length asymmetry in each of the male subgoups display a
pattern of birnodality that appears to be absent from their female counterparts. The
implications of this apparent pattern are discussed in Chapter 6. This gaphic display Of
the shape of the asymmetry patterns is instructive, since it reveals information that cannot

be found in a straightforward reporting of summary statistics (Table 50).

155

 

HUMERUS LENGTH ASYMMETRY
GEORGIAN MALES

 

 

 

 

 

 

 

 

 

 

+ST. BRIM’S (II-30) + SHTALFHDS (II-42)

 

 

 

Figure 42 Georgian Male Signed Humerus Length Asymmetry

 

HUMERUS LENGTH ASYMMETRY
GEORGIAN FEMALES

 

 

 

 

 

 

 

 

6
(R-U mm
+31: amass (fl-32) ... SPITALFIELDS (In-as)

 

 

 

Figure 43 Georgian Female Signed Humerus Length Asymmetry

156

 

HUMERUS LENGTH ASYMMETRY
IDENTIFIED MALES

 

 

 

 

 

 

 

 

 

-10 -5 0 10 15 20

5
IR-l-I mm
+ LONDON min) + FMEIBIC (n-86)

 

 

 

Figure 44 Identiﬁed Male Signed Humerus Length Asymmetry

 

HUMERUS LENGTH ASYMMETRY
IDENnFI- FEMALES

 

 

 

 

 

 

 

 

 

 

 

 

... LONDON (II-85) ..- FORENSIC (In-u)

 

 

 

Figure 45 Identiﬁed Female Signed Humerus Length Asymmetry

157

Table 50 Paired GeorgianI-Iumerus Lerngth Asymmetry Magnitude

a

 

Spitalﬁelds Spitalﬁelds St. Bride’s St. Bride’s
Males Females Males Females
Mean 3.90 5.58 4.47 4.72
SE 0.40 0.49 0.60 0.48
SD 2.59 3.55 3.30 2.70
.95 Conﬁdence 0.78 0.95 1.18 ' 0.93
N 42 53 3O 32

 

 

The pooled Georgians diﬁ‘er markedly in asymmetry magnitude ﬁom the Forernsic
goup (z = 3.049, p < .005)” Figure 44 contrasts the males, showing the relatively large
number of left-dominant individuals in the Forensic sample. There are fewer leﬁ-
donninant females in the Forensic sample, a pattern consistent with the geater level of
right-dominance in humerus lengths among all the female study samples (Figure 45).
However, the Georgian'females are decidedly more asymmetric than their Forensic
counterparts (z = 5.606, p < .0001).

The shape of the asymmetry magnitude distribution patterns changes when
unsigned values are employed. Figures 46 and 47 show that the Georgian males appear
much less similar when only the absolute magnitude of their asyrnrnnetry in considered.

When the unsigned magnitude of asymmetry patterns is compared between the pooled

 

”Assessments of statistical signiﬁcance of sample diﬁ’erences in asymmetry
magnitude for the remainder of this chapter are based on 2 scores derived ﬁorn the Mann-
Whitrney U statistic.

158

Table 51 Combined Humerus Length Asymmetry Magnitude

 

Georgian Georgian Forensic Forensic
Males Females Males Females
Mean 4.14 5.26 2.69 2.61
SE 0.34 0.35 0.26 0.23
SD 2.90 3.26 2.41 1.88
.95 Conﬁdence 0.67 0.69 ' 0.51 0.46
N 72 85 85 64

 

Georgians and the Forensic samples, the diﬁ‘erences are even more profound (males:

2 = 4.905, p < .0001; females: 2 = 5.346, p < .0001.) (Table 51). Among the Forensic
nnales, asymmetry geater than ﬁve rrnillimeters is unusual, but that is not the case for the
Georgians (Figure 48); the same is true for the females (Figure 49).

The ordinal ranking of the four Georgian subgoups does not shiﬁ when signed
values are replaced by ﬁgures representing unsigned magnitude alone (Table 52)”. The
Georgian females present the most asymmetry, followed by the Georgian males, with the
Forensic Data Bank sample presenting with substantially less asymmetry. Even when the
asymmetry patterns are scaled on character size, there is no change in the ordinal ranking.
The only diﬁ‘erence that appears among the four ranking methods is the slightly geater

absolute asymmetry of the Forensic males in comparison with the Forensic females.

 

21In this Table, and in several that follow, SP = Spitalﬁelds; SB = St. Bride’s;
FD = Forensic Database; and CG = Combined Georgians.

159

 

HUMERUS LENGTH ASYMMETRY
GEORGIAN MALES

 

 

 

 

 

 

 

 

 

 

«I—ST. BRIM’S (ll-30) ne- SPITALFIELDS (II-42)

 

 

 

Figure 46 Georgian Male Humerus Length Asymmetry

 

HUMERUS LENGTH ASYMMETRY
GEORGIAN FEMALES

 

 

 

 

 

 

 

10
MAGNITIDE (mm)
a— ST. BRIM'S (NZ) * WALHELDS (W)

 

 

 

Figure 47 Georgian Female Humerus Length Asymmetry

160

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

HUMERUS LENGTH ASYMMETRY
IDENTIFIED MALES
25,“
1 In
2096‘ ,’\
I], ‘ R
£st I ‘V \
a / .
81096‘ \e R
n
n
..I \V/ \A
n
0% ‘1 /A\ A- A ‘
o 10 20
MAGNITIDE (mm)
*LONDON (Ir-72) a.- FORENSIC (ud5)
Figure 48 Identiﬁed Male Humerus Length Asymmetry
HUMERUS LENGTH ASYMMETRY
IDENTIFIED FEMALES
25% f‘
I \
. I t
: l 1
20% .1 I \
.. I n
‘ 1
W
k
\
\
‘. R
1‘ M—
\ A
A’ \ A
ﬂ ' 5 v v 10 15 20
MAGNITIDEﬂnm)
{LONDON (II-85) qr FOREIIBIC (n-M)

 

 

 

Figure 49 Identiﬁed Female Humerus Length Asymmetry

161

Table 52 Humerus Length Asyrnmetry—Ordirnal Ranking

 

SIGNED (mm) UNSIGNED(mm) SIGNED (%) UNSIGNED (%)
SP9 5.09 SP9 5.58 SP9 1.72% SP9 1.88%
CG9 4.93 CG9 5.26 CG9 1.68% CG9 1.79%
SB9 4.66 8B9 4.72 SB9 1.61% SB9 1.63%
SBd' 3.57 SBd' 4.47 SBd' 1.14% SBo‘ 1.40%
C66 3.57 CGo‘ 4.14 CGd' 1.11% CGd' 1.28%
SPO‘ 3.57 SPd' 3.90 SPd‘ 1.09% SPo‘ 1.20%
m9. 1.61 FDd' 2.69 FD9 0.53% FD9 0.85%
FDd' 0.76 m9 2.61 FDd' 0.22% FDd' 0.81%

 

 

 

 

The statistical signiﬁcance of diﬁ‘erences between the Georgians and Forensic
samples has already been noted, but there is also a signiﬁcant diﬁ‘erence between the
males and females of Spitalﬁelds in signed asymmetry magnitude (2 = 2.027; p < .05).
The Forensic males and females are also signiﬁcantly diﬁ‘erent ﬁ'om each other in terms
of signed asymmetry (z = 2.121;p < .05). There is not a corresponding signiﬁcant
diﬁ‘erence for the St. Bride’s sample. When patterns of unsigned asymmetry are
compared between the sexes of the three samples, the signiﬁcance of the Spitalﬁelds
diﬁ‘erence increases slightly (2 = 2.338; p < .05). At the same time the Forensic males
and females are no longer statistically different, and the St. Bride’s sample continue to
not Show statistically signiﬁcant differences.

Table 53 summarizes the ﬁndings for the paired comparisons; note that there is no
hint of a signiﬁcant diﬁ‘erence between the males of the Georgian samples, or between

the females of the Georgian samples. Values in parentheses are based on Mann-Whitney

162

Table 53 Summary: Differences in Humerus Length Asymmetry Magnitude

 

 

SIGNED UNSIGNED
z p z p
Georgian 6‘ .011 — .348 —
Georgian 9 .698 — 1.066 —
Spitalﬁelds d‘/ 9 2.027 < .05 2.338 < .05
(2.038) (< .05) (2.357) (< .05)
St. Bride’s d' I 9 1.000 — .549 —
Forensic d'/ 9 2.121 < .05 .218 —
(2.134) (< .05)
Georgian / Forensic d' 4.906 < .0001 3.049 < .005
(4.924) (< .0001) (3.077) (< .001)
Georgian I Forensic 9 5.606 < .0001 5.346 < .0001
(5.637) (< .0001) (5.391) (< .0001)

 

U statistics that were calculated with a correction for ties (as described in Chapter 4).

The p values that are listed in this Table are for a two-tailed probability.

163
Humerus Head Diameter—HH
The distribution of humerus head asymmetry for the Georgian males is
represented in Figure 50 and for the females in Figure 51. When the pooled Georgians
are compared with the individuals from the Forensic Data Barnk, the slight right-
domirnance in the London groups contrasts markedly with the almost precise symmetry of

the Forensic males and females (Figures 52 and 53).

Table 54 Georgians Humerus Head Asynnmetry Magnitude

 

 

Spitalﬁelds Spitalﬁelds St. Bride’s St. Bride’s
Males Females Males Females
Mean 0.97 1.04 1.04 0.82
SE 0.20 0.13 0.15 0.14
SD 1.18 0.86 0.76 0.66
.95 Conﬁdence 0.39 0.25 0.29 0.28
N 36 47 27 22

 

Summary statistics for the magnitude of head diameter asymmetry (Table S4) in
the Georgians suggests a comparable range of asymmetry among all four of the study
subgoups. Figures 54 and 55 indicate that most of the asymmetry is on the order of a

one-millimeter diﬁ’erence.

164

 

HUMERUS HEAD ASYMMETRY
GEORGIAN MALES

 

 

 

 

 

 

 

 

 

 

0 2 4
(RIGI'IT- LE7) mm
1» ST. BRIw'S (In-soy... SPITALFELN (II-42)

 

 

 

Figure 50 Georgian Male Signed Humerus Head Asymmetry

 

HUMERUS HEAD ASYMMETRY
GEORGIAN FEMALES

 

 

 

 

............................................................................................

 

 

 

5.2
A ‘ A l J
V I I I

 

 

0 2 4
(RIGHT-LEFT) IIIIII
.- S'I'. BRIDE’S anny—.- SPITALFIELDS (Ir-53)

 

 

 

Figure 51 Georgian Female Signed Humerus Head Asymmetry

165
HUMERUS HEAD ASYMMETRY

 

 

 

 

 

 

 

 

 

 

 

IDENTIFIED MALES
50%
I 1
40% I “
I \
I I \
gm ,"
I \
3 ‘ II/ x
20%
* / “\
tr \
10% .1/ \
// kw
/ L ‘ . \
I 0 2 4 6 8
(RIGHT - LET) mm

a” .
-4 .2
{LONDON (nu-72) + FORENSIC (II-85)

 

 

Figure 52 Identiﬁed Male Signed Humerus Head Asymmetry

 

HUMERUS HEAD ASYMMETRY

 

 

 

 

 

 

 

 

 

 

 

 

IDENTIFI. FEMALES
60%
It
1 [I \
40% I X
‘ " A
gears ,’ \
I II/ *
320% 1 ‘-,
at / \ \
1 / n
10% r/ ‘\
. ,1 n
0% 9 e T. ‘2‘. : : :
-4 -2 0 2 4 8 8
(RIGHT . LEI) mm
-.-LONDON (In-05) -.a- FORENSIC (In-u)
Figure 53 Identiﬁed Female Signed Humerus Head Asymmetry

166

 

HUMERUS HEAD ASYMMETRY
GEORGIAN MALES

 

 

 

 

 

 

 

 

 

 

 

 

’.,a-““-.‘-
t" ﬂ ‘1
5

O 7

 

3 4
NAGNIIIDE (mm)
a-ST. BRIDE‘S (mac) -n- SPITALHELDS (n-42)

 

 

 

Figure 54 Georgian Male Humerus Head Asymmetry

 

HUMERUS HEAD ASYMMETRY
GEORGIAN FEMALES

 

 

 

 

 

....................................................... t“
/ 9 \e r at
4 5
NAGNITIDE (mm)
+ST. BRILE’S (I'D-32) -e- WALFIELDS (II-ﬂ)

 

 

 

 

 

“ﬂ

 

 

 

Figure 55 Georgian Female Humerus Head Asymmetry

167

 

HUMERUS HEAD ASYMMETRY
IDENTIFIED MALES

 

 

 

 

 

 

 

X
89-83““
v c \II -
,9-

 

0 1 2 3 4 6 8 7
NA GNITLDE (mm)

-C- LONDON (II-72) 1r- FOREMIC (ud5)

 

 

 

Figure 56 Identiﬁed Male Humems Head Asymmetry

 

HUMERUS HEAD ASYMMETRY
IDENTIFIED FEMALES

 

 

 

 

 

 

 

16
9 9.9.- $199:
ﬂ T
/
\k
, */

 

 

 

II 1
II \>\.
I : ‘4 a as 4-
s 7

3 4
MAGNITIDE (mm)
-I- LONDON (II-85) ... FOREAEIC (nIG4)

 

 

 

Figure 57 Identiﬁed Female Humerus Head Asymmetry

168

Table 55 Combined Humerus Head Asymmetry Magnitude

 

Georgian Georgian Forensic Forensic
Males Females Males Females
Mean 1.00 0.97 0.76 0.60
SE 0.13 0.10 0.09 0.08
SD ' 1.02 0.80 0.83 0.65
.95 Conﬁdence 0.25 0.19 0.18 0.17
N 63 69 80 58

 

When the Georgians are compared with the Forensic Data Barnk individuals, the
apparent lack of asymmetry in the latter sample persists (Table 55). When the magnitude
distributions are displayed gaphically, however, the diﬁ‘erences between the older and
the more recent goups appear to be slight, for both the males and females (Figures 56
and 57).

Table 56 indicates that when the individuals are ordinally ranked, the Georgiarns
are clearly more asymmetrical than the Forensic sample. It is also noteworthy that the
Forensic females are the only subgroup which display a left-dominance pattern in head

diameter asymmetry—albeit a very slight dominance.

169

Table 56 Humerus Head Asymmetry—Ordinal Ranking
_

 

SIGNED (mm) UNSIGNED SIGNED (%) UNSIGNED (%)
(mm)
SBd' 0.67 SP9 1.04 SBd' 1.41% SP9 2.56%
C60“ 0.65 SBd' 1.04 C60“ 1.36% CG9 2.38%
SPd‘ 0.64 CGo' 1.00 SPd' 1.32% SBd' 2.20%
SP9 0.49 CG9 0.97 SP9 1.18% C66“ 2.11%
CG9 0.42 SPd' 0.97 CG9 1.01% SPo‘ 2.05%
SB9 0.27 8B9 0.82 SB9 0.65% SB9 2.00%
FDd' 0.16 FDd' 0.76 FDd' 0.32% FDd' 1.57%
m9 002 FD9 0.60 m9 004% FD9 1.42%

 

 

 

The only strong statistical differences between the study samples occur between
the same-sex samples of the Georgians and the Forensic Data Barnk, as described in

Table 57.

Table 57 Summary: Diﬂ‘erences in Humerus Head Asymmetry Magnitude

170

 

SIGNED UNSIGNED
2 J 2 P
Georgian d' .660 — .840 —
Georgian 9 1.094 — .933 —
Spitalﬁelds o‘/ 9 .262 — .822 —
St. Bride’s d'/ 9 1.397 — 1.005 —
Forensic d‘/ 9 .748 — .8303 —
Georgian lForensic d' 2.249 < .05 1.397 < .10
(2.348) (< .05) (1.450) (< .10)
Georgian lForensic 9 2.396 < .05 2.396 < .05
(2.510) (< .05) (2.510) (< .05)

 

171

Humerus Biepicondylar Width—BB

Table 58 Georgian Humerus Biepicondylar Width Asymmetry Magnitude
—

 

Spitalfields Spitalﬁelds St. Bride’s St. Bride’s
Males Females Males Females
Mean 1.56 1.15 1.63 1.23
SE 0.25 0.15 0.27 0.28
SD 1.48 1.00 1.42 1.31
.95 Conﬁdence 0.48 0.29 0.53 0.55
N 36 47 27 22

 

 

Biepicondylar width of the humerus is the measurement in this study which is
most likely to be directly modiﬁed by the effects of physical activity; therefore,
asymmetry in these measurements would be most likely associated with side-activity
diﬁ'erences among the populations. The Georgian males (Figure 58) display a greater
range of variation in the distribution of asymmetry than the females (Figure 59;

Table 58). All groups display a pattern of right-dominance; among the Georgians the St.
Bride’s females are the only subsample to have a modal asymmetry values of zero. A
mode of zero is also present in the Forensic Data Bank males; in contrast with the
Georgian males they appear more symmetrical (Figure 60). The pooled Georgian
females manifest a normal distribution with a Slight right-shill, not unlike their Forensic
counterparts (Figure 61). When the magnitude of the asymmetry alone is considered,

there is a gender diﬁ'erence between the male and female Georgians, but in this case the

172

males, rather than the females, display more asymmetry

Table 59‘ Combined Humerus Biepicondylar Width Asymmetry Magnitude

 

Georgian Georgian Forensic Forensic
Males Females Males Females
Mean 1.58 1.17 1.20 0.84
SE 0.18 0.13 0.13 0.09
SD 1.44 1.10 1.18 0.67
.95 Confidence 0.36 0.26 0.26 0.17
N 63 69 80 58

 

A graphic display of the asymmetry magnitude shows very little diﬁ‘erence among
the Georgian males (Figure 62) and the Georgian females (Figure 63). When the pooled
Georgians are compared with the Forensic samples, a new pattern emerges. The Forensic
males display a greater magnitude of asymmetry than the Georgian females; this is the
one instance in the study where the Georgians are not consistently the most asymmetric
(Table 59). The Forensic males still display less asymmetry than the Georgian males
(Figure 64), and the Georgian females are still somewhat more asymmetric than the

Forensic females (Figure 65).

173

 

 

 

 

 

 

 

 

 

 

 

 

HUMERUS BIEPI WIDTH ASYMMETRY
GEORGIAN MALES
3596 ‘
30% i I
§25K I. *
20% i r." \
15% “ , ﬁ ;“
8 3, I 1 \ ‘\
10% ' ’ x 1.
: ‘l/ \ \
5,, 1 A
: Y ‘
m .1 /‘~ / \ /A\
J -6 4 -2 0 2 O I
(RIGHT-LEW) MM
-I- ST. ”DES (M) 11- 8mm (II-42)

 

 

 

Figure 58 Georgian Male Signed Humerus Biepi Asymmetry

 

HUMERUS BIEPI WIDTH ASYMMETRY
GEORGIAN FEMALES

 

 

 

 

 

 

or
‘
g
\L
/
"r

 

 

 

 

 

 

 

 

-I- 81'. ”JOE'S (Iv-32) + SHTALFELDS (rt-ﬂ)

 

 

Figure 59 Georgian Female Signed Humerus Biepi Asymmetry

174

 

HUMERUS BIEPI WIDTH ASYMMETRY

 

 

 

 

 

 

 

 

 

IDENTIFIED MALES
50%
(L
40%
J1
£3076
A
.1 I \
326x 4’
* \N-
0 I \
10% K“
K
1) \
0% '1 ~¢¢¥+— W

J 4 0

-2 2
(RIGHT- LE1) mm

 

C-LONDON (It-72) + FOREINIC (nllﬁ)

 

 

Figure 60 Identiﬁed Male Signed Humerus Biepi Asymmetry

 

HUMERUS BIEPI WIDTH ASYMMETRY

 

 

 

IDENTIFIED FEMALES
50%
‘ R
40% ’
I
l I
I
gm
1 1

 

\

l

l

1

1

\

\
*‘ / \x
10% [I A,

1) I, \\

-8 -6 4 -2 0 2 4 6 l
(RIGHT - LET) mm

 

 

 

 

 

i-LONDON (ll-85) -ar- FORENSIC (n-M)

 

 

Figure 61 Identiﬁed Female Signed Humerus Biepi Asymmetry

175

 

HUMERUS BIEPI WIDTH ASYMMETRY
GEORGIAN MALES

 

 

 

 

 

 

 

 

+872 BRIDE'S (M) 1.- SPITALFELDS (ud2)

 

 

 

Figure 62 Georgian Male Humerus Biepi Asymmetry

 

 

 

 

 

 

 

 

 

 

 

HUMERUS BIEPI WIDTH ASYMMETRY
GEORGIAN FEMALES
00% ‘i
: ,‘\
I \
g i x, \
30% 1.
3 (I, \
:20“ ‘ a
j \
. \
10%: \VA
31 \k \ a
0% 1 J. ‘
0 1 2 3 4 5 6 7 I
MAGNITIDE (mm)
-I-ST. BRDE'S (mm) aa- SPITALFELDS (M)

 

 

 

Figure 63 Georgian Female Humerus Biepi Asymmetry

176

 

HUMERUS BIEPI WIDTH ASYMMETRY
IDENTIFIED MALES

 

 

 

 

 

 

 

3 6
NAGNITIDE (mm)
a— LONDON (nan) + FORENSIC (ads)

 

 

 

Figure 64 Identiﬁed Male Humerus Biepi Asymmwy

 

HUMERUS BIEPI WIDTH ASYMMETRY
IDENTIFIED FEMALES

 

 

 

 

 

 

 

 

 

4 5 O 7 I
MAGNITIDE (nun)

-I- LONDON (rt-85) q.- FOREIBIC (II-64)

 

 

 

Figure 65 Identiﬁed Female Humerus Biepi Asymmetry

177

Table 60 Humerus Biepicondylar Width—Ordinal Ranking

 

SIGNED (mm) UNSIGNED SIGNED (%) UNSIGNED (%)
SPo" 0.78 SBd' 1.63 spa- 1.30% 830' 2.62%
CGd' 0.76 CGd' 1.59 CGd' 1.25% CGd' 2.58%
886' 0.74 SPd' 1.56 330“ 1.19% SPd' 2.55%
839 0.59 S39 1.23 $39 1.09% 839 2.28%
339 0.57 FDo‘ 1.20 FD9 1.02% CG9 2.18%
CG9 0.45 CG9 1.17 CG9 0.83% SP9 2.13%
SP9 0.38 SP9 1.15 SP9 0.71% FDd' 1.84%
FDd' 0.38 339 0.84 FDd' 0.60% FD9 1.51%

 

 

 

 

In the ordinal ranking of the groups, the place of the Forensic males presents a
telling example of the importance of distinguishing between signed and unsigned values
when making comparisons between study samples (Table 60). As noted above, in terms
of magnitude alone the Forensic males display a level of asymmetry which is greater than
the pooled Georgian females. However, when signed values are compared the picture
looks very diﬁ‘erent. In fact, in the latter case the Forensic males are the least asymmetric
of all the study subgroups. This is due to the relatively larger number of leﬁ—dominant
individuals among the Forensic males, which counters the right-dominant individuals to
make the group as a whole appear more symmetrical than it actually is. The same
phenomenon appears to occur for the Forensic females as well.

In terms of the statistical signiﬁcance of the diﬁ‘erences among the studied

groups, the only notable diﬁ'erences occur between the pooled Georgian males when they

178

Table 61 Humerus Biepicondylar Width Asymmetry Magnitude Diﬁ‘erences

 

SIGNED UNSIGNED

z p z p
Georgian d' .055 — .299 —
Georgian 9 .051 — .180 —
Spitalﬁelds d'/ 8 .786 — 1.029 —
St. Bride’s d' / 9 .502 — 1.065 —
Forensic d'/ 9 .895 — 1.268 —
Georgian I Forensic d' 1.214 -— 1.590 —
Georgian I Forensic 9 .452 — 1.384 —

 

are compared with the Forensic males. However, these are not signiﬁcant at an alpha of
.05 for a two-tailed test of signiﬁcance. The results of the Mann-Whitney tests are

summarized in Table 61.

179

Femur Length—FL

Table 62 Mean Femur Length Asynunetries—Signed vs. Unsigned

 

 

SIGNED UNSIGNED

Spitalﬁelds Males -1.22 3.33
Spitalﬁelds Females -1.82 2.73
St. Bride’s Males -0.97 3.64
St. Bride’s Females -0.23 2.43
Georgian Males -1.06 3.54
Georgian Females -l.00 2.57
Forensic Males -0.73 2.47
Forensic Females -0.61 2.92

 

A number of studies suggest that the right-dominance in humerus length in human
populations is oﬁen accompanied by a left-dominance in femur length. They also
suggest that the magnitude of asymmetry is relatively smaller in femora when compared
with humeri. These earlier ﬁndings are aﬂirmed in the assessments of the study samples
(Table 62). In all cases there is a net leﬁ-dominance in the asymmetry patterns. More
importantly, there is no straightforward relationship between the Signed and the unsigned
values. The Signed values suggest that femora display a very small magnitude of
asymmetry; however, when unsigned values are reported it is clear that there is a clear .-
pattern of asymmetry in femur length.The distribution of asymmetry for the Georgian
males (Figure 66) and Georgian females (Figure 67) Show both the left-shift and the

relative symmetry of the femora in contrast with the humeri. When the Georgians are

180

pooled and compared with the Forensic sample the same patterns persist. An intriguing
phenomenon appears in the male samples wherein a bimodal distribution is apparent in
both the pooled Georgians and the Forensic sample. This is consistent with the ﬁnding
that femora maintain relatively equal loading on both the right and leﬁ sides to the extent
that there should be some level of symmetry in the asymmetry distributions (Figure 68).
There is a more indistinct picture of bimodality in the Forensic females, but there seems
to be no such pattern in the pooled Georgian females (Figure 69). However, the
Spitalﬁelds females appear to display a bimodality in their femur length asymmetry, as
well.

The combination of relatively small magnitude in asymmetry coupled with a lack
of strong directionality in the asymmetry patterns suggest that the reporting of signed
values has very limited explanatory value in the assessment of asymmetry in femur
length. The Georgian males display a signiﬁcantly greater magnitude of asymmetry than
the females (Table 63); this is the reverse of the case for the length of the humeri. The St.
Bride’s males have a modal asymmetry value of four millimeters, compared to a mode of
two millimeters for the Spitalﬁelds males (Figure 70). The Georgian females both have a
modal femur length asynunetry value of one millimeter, but the Spitalﬁelds females also
have another distinct peak at the four-millimeter level which substantially increases their

mean (Figure 71).

181

Table 63 Paired Georgian Femur Length Asymmetry Magnitude

 

Spitalﬁelds Spitalfields St. Bride’s St. Bride’s
Males Females Males Females
Mean 3.33 2.73 3.64 2.43
SE 0.62 0.46 0.36 0.35
SD 2.61 2.64 2.17 2.08
.95 Confidence 1.21 0.90 0.71 0.69
N 18 33 36 35

 

Table 64 compares the pooled Georgian sample with the Forensic Data Bank
individuals. The Georgian males follow the typical pattern in which they display a
substantially greater asymmetry than the Forensic males. In fact the latter sample has a
modal asymmetry of one millimeter, and few individuals displaying greater than three
millimeters of length difference; the Georgians show much more variation (Figure 72).
The pattern is reversed for the females. The Forensic females display a greater
magnitude of asymmetry than their Georgian counterparts, with a substantial proportion
of the Forensic women manifesting asymmetry of up to ﬁve millimeters, and a modal
value of three millimeters (which contrasts with the Georgian mode of one millimeter).
The relatively high level of asymmetry in the Forensic females femur length reverses the

pattern of the humeri wherein all three measurements wherein the Georgians displayed a

182

Table 64 Combined F emur Length Asymmetry Magnitude
—

 

Georgian Georgian Forensic Forensic
Males Females Males Females
Mean 3.54 2.57 2.47 2.92
SE 0.31 0.28 0.19 0.23
SD 2.31 2.35 1.87 1.88
.95 Confidence 0.62 0.56 0.38 0.46
N 54 68 93 65

 

 

greater magnitude of asymmetry (Figure 73).

Ifmagnitude alone is a basis for comparison, the ordinal ranking of the four
Georgian subgroups in terms of symmetry is modiﬁed as well. For example, the
Spitalﬁelds females appear to have the greatest mean levels of asymmetry when signed
values are used in the calculation, but in terms of magnitude, both the Spitalﬁelds and St.
Bride’s males show greater asymmetry. Likewise, the St. Bride’s males present the
largest mean values for asymmetry when the magnitude alone is considered; when signed
values are compared, however, bOth the Spitalﬁelds males and females appear to show
greater asymmetry. There is a somewhat different pattern for the Forensic Data Bank
males and females. The males Show Slightly more mean asymmetry than the females

when the signed values are compared, but the females Show a greater magnitude of

183

Table 65 Femur Length Asymmetry—Ordinal Ranking

 

 

SIGNED (mm) UNSIGNED (mm) SIGNED (%) UNSIGNED (%)
SP9 -l.82 SBd' 3.64 839 0.45% SBO‘ 0.80%
SPd' -122 CGd' 3.54 SPd' 0.27% CGo‘ 0.78%
CGd' -1.06 SPo‘ 3.33 CGd' 0.24% SPd' 0.75%
CG9 -1.00 339 2.92 CG9 0.24% FD9 0.67%
536' .097 SP9 2.73 830' 0.22% SP9 0.67%
FDd' -073 CG9 2.57 FDd' 0.15% CG9 0.63%
FD9 -0.61 FDo‘ 2.47 339 0.14% S39 0.59%
839 .023 S39 2.43 839 0.04% FDd' 0.52%

 

 

 

asymmetry than the males (2.9 mm vs 2.5 m) (Table 65).
The statistical signiﬁcance of the sample diﬁ‘erences in the length of the femur are

much less profound than for the humerus (Table 66).

184

Table 66 Summary: Differences in Femur Length Asymmetry Magnitude Patterns

 

 

SIGNED UNSIGNED
z p z p
Spitalﬁelds vs St. Bride’s Males .119 — .477 —
Spitalfields vs St. Bride’s 1.736 < .10 .626 —
Females (1.748) (< . 10)
Spitalfields Males vs Females .384 — .985 —
St. Bride’s Males vs Females .914 — 2.421 < .05
(2.449) (< .01)
Forensic Males vs Females .011 — 1.715 < .10
(1.744) (< .10)
Georgian vs Forensic Males .627 -— 2.583 < .01
(2.622) (< .01)
Georgian vs Forensic Females .232 — 1.710 < .10
(1.734) (< .10)

185

 

FEMUR LENGTH ASYMMETRY
GEORGIAN MALES

 

 

 

 

 

 

 

 

5 10 15 20

+81. BRDE'S (ls-06) ... SPITALFELDS 07-18)

 

 

 

Figure 66 Georgian Male Signed Femur Length Asymmetry

 

 

 

 

 

 

 

 

FEMUR LENGTH ASYMMETRY
GEORGIAN FEMALES
2096
ms 0 T.
I» g It
1» [I {I
" t 1‘1."
096 0 11 I ‘1
8 J '1 I 1 *
’ II I
S 1. ‘1‘; 1‘ A ’11‘
.. 1%.. a: 1 ,4
J’ R a I 1 VLA! I ‘
”l\ l\
I \ I \
.51 .
45 ~10 .5 0 5 10 15 20
034)"!!!
..ST. BRDE'S M6) ...- SprrAu-‘rELDS our)

 

 

 

Figure 67 Georgian Female Signed Femur Length Asymmetry

186

 

FEMUR LENGTH ASYMMETRY
IDENTIFIED MALES

 

53'.

 

 

 

C v 7 C v v v
5-. ?
ﬂ *
"
7 ‘~
~
-
'P-""

a
8
T k
if;
:"<
)

 

 

 

4r \ I
096-1 '
-15 -10 .5 O 5 10 15 2O
III-Um

-- LONDON (II-50 + FORMS 0H3)

 

 

 

Figure 68 Identiﬁed Male Signed Femur Length Asymmetry

 

 

 

 

 

 

 

 

 

 

 

FEMUR LENGTH ASYMMETRY
IDENTFED FEMALES
2076
159‘ 4 ........
East 0
a I:
5x
096"
.15 .10 .5 o 5 10 15 20
("rum
...Louoou (II-68) + FORENSIC (II-65)

 

 

 

Figure 69 Identiﬁed Female Signed Femur Length Asymmetry

187

 

FEMUR LENGTH ASYMMETRY

GEORGIAN MALES
5516

 

2516

 

 

I
I
16 i
‘ M; “/\
I: I
516" : ‘1 3.

 

8 3;]; 1‘ X".
*mr"“ * ’ \A

 

1./ ‘x I, Y\ \
o \ I ‘
5%" X kw!

 

 

... sr. BRIDE'S muss)... muses on“)

 

 

 

Figure 70 Georgian Male Femur Length Asymmetry

 

 

 

 

 

 

 

 

 

 

 

 

FEMUR LENGTH ASYMMETRY
GEORGIAN FEMALES
f
H,
’ '1
l
l
1
l
‘1
A
Y‘,\
‘1
A\ I \ /A\
\ I \ / \
4 : x +
4 s s 10 12 :4 rs
MAGrsersI-n)
..SE BRDE'S me) + SFrrALFssLDS muss;

 

Figure 71 Georgian Female Femur Length Asymmetry

 

188

 

FEMUR LENGTH ASYMMETRY
IDENTIFIED MALES

 

 

 

 

 

 

 

s s 10 12 u rs
MAGNIwDE m)

-I- LONDON (II-54) + FORBISIC (1H3)

 

 

 

Figure 72 Identiﬁed Male F emur Length Asymmetry

 

 

 

 

 

 

 

 

 

 

FEMUR LENGTH ASYMMETRY
IDENTIFIED FEMALES
3016
11
25’s 1’. ..........................................................................................
11
13 ......... A
0 I ‘\
it
:96 ” f ‘38 ,A,
3 i I ‘\ 1
I k \
*1096 j 1
1 1
ll \1‘
5%“ \
.. 1
1 \
0% J. : t . 4 M
0 2 4 5 5 10 12 14 15
MAGMTUDE pun)
...Lomou (II-68) + FORENSIC 01-05)

 

 

 

Figure 73 Identiﬁed Female Femur Length Asymmetry

189

Femur Head Diameter—F11

Table 67 Intact Georgian Femur Head Asymmetry Magnitude
_

 

Spitalﬁelds Spitalﬁelds St. Bride’s St. Bride’s
Males Females Males Females
Mean 0.65 0.85 0.79 0.68
SE 0.21 0.17 0.15 0.13
SD 0.86 0.86 0.72 0.58
.95 Confidence 0.41 0.33 0.29 0.26
N 17 27 24 19

 

Just as the diameter of the humeral head is not expected to vary substantially with
physical activity, neither is the femoral head expected to do so. In fact the asymmetry
distribution patterns for femur head diﬁ‘erences are virtually the same as for the humerus
heads. There is a very slight right-shift in the pattern, and the males and females are very
similar. Among the Georgian samples, the St. Bride’s males and females are Slightly
more asymmetrical than their Spitalﬁelds counterparts (Figures 74 and 75). The pooled
Georgian males are very similar in distribution pattern to the Forensic males (Figure 76),
but the Georgian females display a right-Shift which contrasts with the very slight left-
shiﬁ in the Forensic females (Figure 77). Table 67 summarizes the ﬁndings for the

Georgian samples in terms of unsigned magnitude values.

190

 

FEMUR HEAD ASYMMETRY
GEORGIAN MALES

 

 

 

 

 

 

 

 

 

 

 

 

a» 81'. BRIW‘S (II-35hr SHTALHHDS (II-18)

 

 

 

Figure 74 Georgian Male Signed Femur Head Asynunetry

 

FEMUR HEAD ASYMMETRY
GEORGIAN FEMALES

 

 

 

 

 

 

 

 

.» ST. BRIOE'S (II-19)-* SPITALFIELDS (II-27)

 

 

Figure 75 Georgian Female Signed Femur Head Asymmetry

191

 

 

 

 

 

 

 

 

 

 

 

 

FEMUR HEAD ASYMMETRY
IDENTIFIED MALES
60%
50% ‘ #4 ‘1
I \
I 11A \\
' \
I ‘ ﬂ \1
83096
. \
1.». xi S
J .- \._ ‘ ‘\.
0% 1M ‘- - ' g
-4 -3 -2 -1 0 ‘I 2 3 4
(RIGHT - LEI-‘0 mm
.- LONDON (II-54) + FORENSIC (n-93)

 

Figure 76 Identiﬁed Male Signed Femur Head Asymmetry

 

FEMUR HEAD ASYMMETRY
IDENTIFIED FEMALES

 

 

 

 

 

 

I
,/
I

 

.1/

 

 

 

4

 

 

/
. /
J
r , g

-2 -1
(RIGHT

 

0
4,51)

 

1
mm

*LONDON (M) + FOREMIC (II-55)

 

Figure 77 Identiﬁed Female Signed Femur Head Asymmetry

 

 

192

 

FEMUR HEAD ASYMMETRY
GEORGIAN MALES

 

 

 

 

2
MAGNITIDE (mm)
-I~ ST. BRIDE’S (II-35%.- WALHELDS (II-15)

 

 

 

Figure 78 Georgian Male Femur Head Asymmetry

 

FEMUR HEAD ASYMMETRY
GEORGIAN FEMALES

 

 

 

 

 

2
MAGNITIDE (mm)
+ST. BRIDE'S (II-35) —a~ SPITALFIELDS (II-33)

 

 

 

Figure 79 Georgian Female F emur Head Asymmetry

193

 

 

 

 

 

 

 

 

 

 

FEMUR HEAD ASYMMETRY
IDENTIFIED MALES
50% ‘
J».
50% ‘ x
i \\
9%
gm 1, \\
: \
309‘ x
3 I \\
*2096 " ‘5.
1 \ \\
. x
4 \\
10% ‘ \
l .- ~\\.
0% i 4 r
0 2 3 4
MAGNITIDE (um)
-I- LONDON (n054) + FORENSIC (II-93)

 

 

 

Figure 80 Identiﬁed Male F ernur Head Asymmetry

 

FEMUR HEAD ASYMMETRY
IDENTIFIED FEMALES

 

 

 

 

 

 

 

 

 

MAGNITIDE (mm)
‘- LONDON (RI-68) + FORENSIC (RICE)

 

 

 

Figure 81 Identiﬁed Female Femur Head Asymmetry

194

In this instance, there is no obvious relationship between either the sexes or the
populations with respect to the asymmetry patterns. That is, the St. Bride’s males and the
Spitalﬁelds females display greater mean asymmetry magnitude than the Spitalﬁelds
males and the St. Bride’s females. Both the St. Bride’s males and females do have a
modal head asymmetry value of one millimeter, which contrasts with the mode of zero

that applies to the Spitalﬁelds samples (Figures 78 and 79 ).

Table 68 Combined Femur Head Asymmetry Magnitude

 

Georgian Georgian Forensic Forensic
Males Females Males Females
Mean 0.73 0.78 . 0.57 0.50
SE 0.12 0.11 0.08 0.08
SD 0.78 0.76 0.72 0.60
.95 Confidence 0.24 0.22 0.16 0.15
N 41 46 82 58

 

The Forensic males and females also display a mode of zero for femoral head
asymmetry, and in the end the distribution patterns for the pooled Georgians is quite
similar to the Forensic sample, although the Georgians display a slightly greater
magnitude of asymmetry (Table 68; Figures 80 and 81). Because head diameters are
more symmetrical than are bone lengths or biepicondylar widths, the magnitude of

diﬁ'erences among the study subgroups are small.

195

Table 69 F emur Head Diameter Asyrnmetry—Ordinal Ranking

 

SIGNED UNSIGNED (mm) SIGNED (%) UNSIGNED (%)
SP9 0.63 SP9 0.85 SP9 1.45% SP9 2.02%
CG9 0.57 333 0.79 CG9 1.31% CG9 1.85%
839 0.47 CG9 0.78 839 1.11% 830' 1.71%
830' 0.46 CGd' 0.73 830' 0.96% S39 1.62%
CGd‘ 0.44 839 0.68 CGd' 0.89% CGd' 1.55%
SPo‘ 0.41 SPd' 0.65 SPd' 0.80% spa 1.32%
FDd' 0.21 FDd‘ 0.57 FDd' 0.43% FDd' 1.17%
1739 0.09 FD9 0.50 339 0.22% 339 1.17%

 

 

 

When asymmetry magnitude patterns among the study subgroups are ordinally
ranked, there is again a general pattern in which the Georgian females display greater
asymmetry than the males (Table 69). Likewise, the Forensic Data Bank samples are
consistently more symmetrical than the Georgians, irrespective of which of the four
techniques for calculating asymmetry iS employed.

In terms of statistical signiﬁcance of differences, the only remarkable diﬂ'erences
occur between the forensic females and the Georgian females. The results of the

calculations for all groups are presented in Table 70

196

Table 70 Summary: Diﬂ'erences in Femur Head Asymmetry Magnitude
_

 

 

SIGNED UNSIGNED

2 P 2 P
Spitalﬁelds vs St. Bride’s Males .304 — .807 —
Spitalfields vs St. Bride’s .435 — .413 —

Females
Spitalfields Males vs Females .928 — .819 —
St. Bride’s Males vs Females .147 — .367 —
Forensic Males vs Females .783 — .309 —
Georgian vs Forensic Males .888 — 1.046 —
Georgian vs Forensic Females 2.549 < .01 1.731 < .10
(2.756) (< .01) (1.920) (< .10)

197

Femur Biepicondylar Width—EB

The epicondyles of the femur are primarily Sites of ligamentous attachments,
which makes them much less prone to direct activity-related changes than the
epicondyles of the humerus. Among the study samples, they also do not display the wide
range of variation that was present in the humeral biepicondylar widths. Among the
Georgian males the St. Bride’s sample is virtually symmetrical in its asymmetry
distribution pattern, and the Spitalﬁelds sample displays a distinctive right-shift
(Figure 82). Both of the Georgian female sample have a right-Shift, as well as a modal
asymmetry value of one millimeter (Figure 83)

When the Georgians are pooled and compared with the Forensic sample, the
males display virtually the same distribution pattern (Figure 84). The Forensic females
display a virtually normal distribution as well, with only the slightest hint of a right-shift
in distribution pattern (Figure 85). As was the case for the femoral head diameter, the
Georgians do not display a distinctive population-related or gender-related pattern of
asymmetry magnitude (Table 71). However, there is an unusual trend among the
Georgians. The Spitalﬁelds males and the St. Bride’s females display the greatest
magnitude of asymmetry, which is the reverse of the case of the femoral head diameters.
The numbers are too small to suggest that there is any real inverse relationship between
head diameter asymmetry and biepicondylar width asymmetry, but they do suggest that
there is no obvious relationship between the two measurements. (See “Patterns in the
Femur” for the results of the statistical testing of the relationships among measurements

within the study samples.)

198

 

FEMUR BIEPI WID11-I ASYMMETRY
GEORGIAN MALES

 

 

 

/ \A
II \
30% 4’ ‘~
g ‘ / I, \‘\
320% j K ‘1
at I’

I

1

/' ‘\
II A /
// \\ I
015- 9 e

-5 -4 .2 0 2 4 5
(RIGHT-LET) arm

-I- ST. BRME'S (0824) + SPITALMLDS (M17)

 

 

 

 

 

 

 

 

 

Figure 82 Georgian Male Signed Femur Biepi Asymmetry

 

FEMUR BIEPI WIDTH ASYMMETRY
GEORGIAN FEMALES

 

 

 

 

 

 

 

 

 

 

-I- ST. BRIDES (II-19) + SPITALFIELDS (II-27)

 

 

 

Figure 83 Georgian Female Signed Femur Biepi Asymmetry

199

 

FEMUR BIEPI WIDTH ASYMMETRY
IDENnFIED MALES

 

 

 

 

 

as .1 ’1 \
1096
WA \
U \
09H 4 9 i 5
.5

-4 -2 0 2 4 5
(RIGHT- LET) mm

iLOWON (II-41) -t- FOREABIC (II-52)

 

 

 

 

 

 

 

 

Figure 84 Identiﬁed Male Signed Femur Biepi Asymmetry

 

FEMUR BIEPI WIDTH ASYMMETRY
IDENnFIED FEMALES

 

 

 

 

 

 

 

 

 

 

0 2 4 5
(RIGHT- LEI) mm

-4 .2

i-LONDON (II-45) 1.- FOREmIC (II-55)

 

 

 

Figure 85 Identiﬁed Female Signed Femur Biepi Asymmetry

200

 

FEMUR BIEPI WIDTH ASYMMETRY
GEORGIAN MALES

 

 

 

 

 

AA.
W
\
/,
/y//
/
/’

 

.. X
.. \ \
0 \g
.. ‘\
1?
1r ‘\
I? l l
I I

0 1 2 3 4 6 8
MAGNITIDE (nun)

-.- ST. BRIDE’S (II-24) -.a- SPITALHELDS (II-17)

 

 

 

 

 

 

 

Figure 86 Georgian Male Femur Biepi Asymmetry

 

FEMUR BIEPI WIDTH ASYMMETRY
GEORGIAN FEMALES

 

 

 

 

 

 

 

 

 

 

3 4
MAGNITIDE (mm)
-.-ST. BRIDE'S (nﬂS) + SPITALFELDS (II-33)

 

 

 

Figure 87 Georgian Female Femur Biepi Asymmetry

201

 

FEMUR BIEPI WIDTH ASYMMETRY
IDENTIFIED MALES

 

 

50%
50,"
”/‘1
§ ( \
*2096‘

10%

0%

 

 

 

 

 

A 1 ~

 

2 3 4 6 5
MAGNITIDE (um)

-I- LONDON (n854) 1- FORENSIC (M)

 

 

 

Figure 88 Identiﬁed Male Femur Biepi Asymmetry

 

FEMUR BIEPI WIDTH ASYMMETRY
IDENnFIED FEMALES

 

 

 

 

 

 

 

 

 

 

 

2 3 4
MAGNITIDE (mm)
-I- LONDON (nl58) 11- FOREIIBIC (II-55)

 

 

 

Figure 89 Identiﬁed Female Femur Biepi Asymmetry

202

Table 71 Georgian Femur Biepicondylar Width Asymmetry Magnitude

 

Spitalfields Spitalfields St. Bride’s St. Bride’s
Males Females Males Females
Mean 1.06 0.93 0.63 1.21
SE 0.20 0.23 0.15 0.21
, SD 0.83 1.17 0.71 0.92
.95 Confidence 0.39 0.44 0.28 0.41
N 17 27 24 19

 

All four of the Georgian subsamples display a modal value of one millimeter diﬁ‘erence
in biepicondylar widths (Figures 86 and 87).

When the pooled Georgians are compared with the Forensic Data Bank sample,
there is again no clear pattern of either a gender or a population similarity in asymmetry
magnitude (Table 72). The Forensic males are virtually the same as the Georgian males
(Figure 88) and there is only a slight diﬁ‘erence for the females. Because of the

extremely high level of symmetry in the Forensic females, however, the modal value for

203

Table 72 Combined F emur Biepicondylar Width Asymmetry Magnitude

 

Georgian Georgian Forensic Forensic
Males Females Males Females
Mean 0.80 1.04 0.85 0.84
SE 0.12 0.16 0.10 0.14
SD 0.78 1.07 0.86 1.04
.95 Confidence 0.24 0.31 0.19 0.27
N 41 46 82 58

 

 

that group is zero millimeters (in contrast with the other subgroups) (Figure 89).
When the biepicondylar width asymmetries are ordinally ranked, there is again no
clear relationships among the study subgroups (Table 73).

Table 73 Femur Biepicondylar Width Asymmetry—Ordinal Ranking

 

SIGNED (mm) UNSIGNED (mm) SIGNED (%) UNSIGNED (%)
839 1.11 S39 1.21 839 1.53% S39 1.67%
CG9 0.87 SPd' 1.06 CG9 1.19% CG9 1.44%
SP9 0.70 CG9 1.04 SP9 0.96% SPd' 1.29%
SPd' 0.47 SP9 0.93 SPd' 0.56% 1739 1.12%
FDd' 0.39 1730‘ 0.85 FDo“ 0.47% FDO' 1.01%
CGd' 0.27 1739 0.84 1739 0.36% CGO‘ 0.98%
1739 0.26 CGd‘ 0.80 CGd' 0.32% SBd' 0.77%
SBd‘ 0.13 SBd' 0.63 SBd' 0.14% SP9 0.66%

 

 

 

204

Table 74 Femur Biepicondylar Width Asymmetry Magnitude Diﬂ‘erences

 

SIGNED UNSIGNED
z p z p
Georgian 6' 1.508 — 1.680 < .10
(1.859) (< .10)
Georgian 9 1.629 — 1.439 —
Spitalﬁelds o‘/ 9 .205 -— 1.012 —
St. Bride’s d'/ 9 2.935 < .005 2.103 < .05

Forensic d'/ 9
Georgian / Forensic d‘
Georgian / Forensic 9

 

(3.063) (< .005)

.651 —

.335 —

2.541 < .05
(2.564) (<05)

(2.294) (< .05)
.459 —
.166 ——
1.096 —

Table 74 summarizes the results of the Mann-Whitney signiﬁcance tests. In terms

of statistical signiﬁcance, there are notable differences between the sexes in the Georgian

samples, and between the same-sex groups of St. Bride’s and Spitalﬁelds. Interestingly,

however, the pooled Georgian males and females do not diﬁ‘er signiﬁcantly from their

Forensic counterparts, and there is no signiﬁcant diﬁ'erence between the Forensic males

and females in terms‘of femoral biepicondylar width asymmetry magnitude.

205
RAW

One might assume that in any paired set of humeri a ﬁnding of right-dominance
in bone length would be necessarily coupled with right-dominance in head diameter
and/or biepicondylar width, as well. This assumption would make particular sense if
physical activity patterns were the primary determinants of side-dominance patterns. To
determine whether this was indeed the case, the Study samples were used to test the
hypothesis that there is a relationship between the direction of asymmetry in
biepicondylar width of the humerus and the maximum length of the bone. Each
individual who displays asymmetry in both of those measurements at a magnitude of one
millimeter or larger was assessed to determine whether the direction of asymmetry in
biepicondylar width was the same as, or diﬂ‘erent ﬁ'om, the direction of asymmetry in
length. Persons who were coded as symmetrical in either of the measurements were not
included in the assessment. Likewise, paired comparisons were made with the
relationship between length and head diameter, as well as biepicondylar width and head
diameter.

Table 75 shows the results of the comparisons for the four Georgian subgroups.
The calculated p values represent the probability that the paired observed values would
arise from a population where there is equal likelihood that individuals would fall into the
two categories. For both the males and females of Spitalﬁelds, the relationships between
humerus length and biepicondylar width are highly signiﬁcant, and the relationships
between humerus length and head diameter are Signiﬁcant. The relationship between
length and head diameter is signiﬁcant for the St. Bride’s males as well. None of the

Georgian samples display a statistically signiﬁcant relationship between the direction of

206

Table 75 Georgian Diﬂ‘erences in Humerus Asymmetry Direction Patterns

 

 

 

Spitalfields Spitalfields St. Bride’s St. Bride’s
Males Females Males Females
Length vs Same 21 24 9 7
Bk'xgﬁﬁym 3111‘ 7 9 3 2
p <01 <01 <.10 <.10

Length vs Same 15 17 8 3
Danger Diﬂ‘ 5 7 l 2

p <05 <05 <05 —

Biepicondylar Same 1 1 7 5 0
WEE,“ 313‘ 4 10 2 1

Diameter p <10 _ _ _

 

 

biepicondylar width and head diameter.

Sample size has a notable effect on calculated p values in a binomial test, and
pooling the Georgians by sex increases the level of Signiﬁcance in this assessment, as
evident in Table 76. The relationship between the biepicondylar width and length of the
humerus is highly signiﬁcant for both the Georgian males and females, in contrast with
the Forensic males in particular. The Forensic females display a Signiﬁcant diﬁ‘erence as
well, the only statistically signiﬁcant ﬁnding for the Forensic sample in the humerus.

The difference between the Georgian and Forensic samples is noteworthy, for it indicates
a signiﬁcant relationship for the more highly directional group, but a lack of relationship

for the less directional contemporary sample.

207

Table 76 Summary: Diﬁ‘erences in Humerus Asymmetry Direction Patterns

 

 

 

Georgian Georgian Forensic ForenSic
Males Females Males Females
Length vs Same 30 31 29 25
Biepicondylar .
Width Drﬂ‘ 10 11 20 10
p <.005 <.005 — <.01
Length vs Same 23 20 24 15
Head .
Diameter Drﬂ‘ 6 9 16 1 1
p <.005 <.05 — _
Biepicondylar Same 15 7 20 10
Width vs ,
Head Drﬁ‘ 6 11 12 9
Diameter p (.05 _ _ _

 

 

RAW
In order to assess the patterns of relationship between asymmetry direction among

the three paired observations of paired femora, the same analytical scheme was applied to
the femur as described for the humerus above. Those individuals who displayed a
measurable asymmetry in both of the paired measurements under consideration were
included in the assessment, and those which were symmetrical in either measurement
were not included.

The ﬁndings for the Georgians are listed in Table 77. There are no statistically .-
signiﬁcant deviations from the null hypothesis that observations would fall equally into
the same and different categories. For the Georgians, at least, there is no evidence for a

relationship between direction of asymmetry in any of the paired measurements.

208

Table 77 Georgian Diﬁ'erences in Femur Patterns

 

 

 

Spitalfields Spitalfields St. Bride’s St. Bride’s
Males Females Males Females

Length vs Same 8 8 2 3
““355“,?” Diff 5 8 5 3

p _ _ _ _

Length vs Same 3 6 2 2
Dilalzrger M 3 7 5 2

p _ _ ... _
Biepicondylar Same 3 7 2 1
WEE?" Diﬁ‘ 2 2 2 1

Diameter p _ _ _ _

 

 

When the Georgians are pooled and compared with the Forensic sample, the
pattern persists, as evident in Table 78. Even with the larger sample sizes, there are
virtually no statistically signiﬁcant deviations ﬁom the null hypothesis that same-side
and opposite-side dominance would occur with equal probability. The one exception is
the Forensic males, who display a highly signiﬁcant pattern of reversed relationship
between length of the femur and femoral head diameter. Earlier in this chapter the
general pattern of right-dominance in head diameter and biepicondylar width of the
femur in contrast with the left-dominant length of the femur were noted, but this is the

only instance where the pattern is revealed to be statistically Signiﬁcant.

209

Table 78 Summary: Diﬁ'erences in F ernur Asymmetry Direction Patterns

 

 

 

Georgian Georgian Forensic Forensic

Males Females Males Females
Length vs Same 10 11 23 14
“egg?" Diff 10 11 21 16
p _ ._ _ __
Length vs Same 5 8 13 10
35:3" Diﬁ‘ 8 9 29 13
p - — <.005 —
Biepicondylar Same 5 8 14 10

Wgﬂ'd" Diﬁ‘ 4 3 9 5

Diameter p _ _ __ _

 

 

The suggestion that there may be a tendency toward a diﬁ‘erence in asymmetry
direction of observations within a bone leads to the question of whether there is a notable
pattern of cross symmetry in the study samples. To test this, individuals manifesting
measurable asymmetry in length of the humerus and length of the femur were assessed in
a manner very much like that described above for the individual bones. Asymmetry
thresholds of both one millimeter and two millimeters were both used, to determine if
they would have an effect on the results.

Table 79 lists the results for the Georgian subgroups. In each group there is a
greater proportion of individuals displaying a crossed Symmetry pattern rather than a

same-Side asymmetry pattern, typically by a two to one margin. In most cases, however,

210

Table 79 Georgian Cross Symmetry in Humerus and Femur Lengths

 

 

 

 

Same-Side Crossed p

Spitalfields 1 mm 4 9 —
Md" 2 mm 3 8 —
Spitalfields 1 mm 9 l8 —
Fem” 2 mm 4 15 <.01
St. Bride’s 1 mm 8 10 —
Md” 2 mm 3 7 —-
St. Bride’s 1 mm 6 12 —
Females 2 mm 4 6 _

 

the diﬁ‘erences are not large enough to reject the null hypothesis that the study samples
were drawn from a population in which there is an equally likelihood of same—side and
crossed asymmetry. The only notable ﬁnding is that the use of a two-millimeter
asymmetry threshold results in a scenario where the Spitalﬁelds females display a highly
statistically signiﬁcant pattern of crossed symmetry.

Table 80 pools the two Georgian samples and contrasts them with the Forensic
sample. There is a statistically signiﬁcant pattern of crossed symmetry in both the
Georgian males and the Georgian females at the .05 level, if a two-millimeter threshold is
employed to distinguish between symmetry and asymmetry. In contrast, there is
substantially less crossed symmetry in the Forensic sample, and no evidence of a

statistically signiﬁcant pattern of cross symmetry in that sample.

211

Table 80 Combined Cross Symmetry in Humerus and Femur Lengths

 

 

 

 

 

Same-Side Crossed p

Georgian 1 mm 12 19 —

Md” 2 mm 6 15 <.05
Georgian 1 mm 15 30 —
Fund” 2 mm 8 21 <.05
Forensic 1 mm 31 29 —

mu” 2 mm 15 17 —
Forensic 1 mm 19 27 —
Fund“ 2 mm 12 21 —

 

Summary

Well over half of the Georgian males and females display a pattern of crossed
symmetry in humerus and femur lengths, with the females presenting a greater extent of
the pattern than the males. In contrast, the Forensic males and females are relatively
more likely to display same-side symmetry, and the Forensic males demonstrate slightly
more same-Side asymmetry than crossed symmetry at the one-millimeter symmetry
threshold. In all cases, raising the symmetry threshold from one millimeter to two
millimeters accents the crossed symmetry phenomenon. By far, the predominant pattern

is one of a right-dominant humerus coupled with a left-dominant femur.

212
W
Throughout this chapter, the hypotheses introduced in Chapter 2 have been tested
by using statistical techniques that determine the probability that the two samples being
compared are drawn from a single population. An alpha of .05 maintains consistency
with the literature, and is used here to indicate p values that are adequately signiﬁcant to

indicate independence.

Table 81 Signiﬁcant Diﬁ'erenoes in Humerus and Femur Dimensions
—

Humerus Femur

L H B L H B

 

Georgian 01'
Georgian 9
Spitalfields d'/ 9
St. Bride’s o“ I 9
Forensic d'/ 9

Georgian / Forensic d'

\\\\\\
51““
\‘R“
R‘Tx“
‘\\\\
\‘\\\

 

 

Georgian / Forensic 9

(1) '13. I‘ no 324 1'911.-‘.€.110_' It: I 1011:. 10111101 1101001 1.2.
‘O'IIIJJ (131311 I'll ’. . .' ' 1%.11'. Filll . ... “.1. .‘W. .
This hypothesis iS supported by z and t test comparisons of means for the study samples,
as indicated in Table 81. I

For ﬁve of the six paired measurements (length (L), head diameter (H), and

213

biepicondylar width (B)) there is no statistical diﬁ‘erence between the two Georgian male
samples nor between the two Georgian female samples. The lone exception is maximum
length of the humerus, which was signiﬁcantly different (p < .05) for the Georgian
females. These results do not allow rejection of the null hypothesis that both Georgian
samples were drawn ﬁ'om a single population. In contrast, the pooled Georgians diﬂ‘er
very signiﬁcantly from the Forensic sample in each of the six paired measurements, with
the twentieth-century group larger in each instance. For both males and females,
comparisons of bone dimensions between the pooled Georgians and the Forensic sample
clearly indicate a rejection of the null hypothesis that they are drawn ﬁ'om the same
population.

Table 82 Signiﬁcant Diﬁ‘erences in Humerus Asymmetry Among the Study Samples

 

Length Head Biepicondylar
Diameter Width

DUSDUSDUS

 

 

 

 

Georgian 13'
Georgian 9 V
Spitalfields «fl 9 V V
St. Bride’s d' I 9
Forensic d' I 9 V
Georgian I Forensic d‘ V V V V
Georgian IForensic 9 V V V V V

 

m. Table 82 reviews the results of the study regarding asymmetry patterns in the

214

Table 83 Signiﬁcant Differences in F emur Asymmetry Among the Study Samples

 

 

 

Length Head Biepicondylar
Diameter Width
U S D U S D U S
Georgian 0‘ V
Georgian 9
Spitalfields (H 9
St. Bride’sd'l9 V V V V
Forensic d'I 9
Georgian I Forensic d' V
Georgian I Forensic 9 V V

 

 

 

humerus for the three paired measurements under consideration. Table 83 reviews the

study’s results regarding the comparisons of asymmetry patterns in the three paired

measurements in the femur. Taken together, the information contained in these two

Tables summarizes the study ﬁndings for hypothesis two and hypothesis three. In the

Tables, statistical signiﬁcance at the .05 level is indicated for direction of asymmetry (D),

unsigned magnitude (U), and signed magnitude (S). Given these results, there is not

strong support for this hypothesis. The Spitalﬁelds males and females diﬁ‘er signiﬁcantly

in both unsigned and signed humerus length asymmetry, but this pattern is not shared by

their counterparts from St. Bride’s. Forensic males and females diﬂ‘er signiﬁcantly in

terms of signed humerus length magnitude, but this is the only measurement among the

six in which they display signiﬁcant sex-related variation. In the femur, the only sex-

related diﬂ‘erence in asymmetry patterns is seen among the St. Bride’s males and females.

 

unequivocal support for this hypothesis. In humerus length, both the males and the

females of the Georgians and Forensic samples differ Signiﬁcantly in direction of
asymmetry, as well as its signed and unsigned magnitude. There are also signiﬁcant
diﬁ‘erences between these samples in the magnitude of humerus head asymmetry. In the
femur, the Georgian and Forensic males diﬁ‘er signiﬁcantly in bone length, and the
females diﬁ‘er in terms of signed magnitude of head diameter and unsigned magnitude of
biepicondylar width.

Table 84 Signiﬁcant Diﬁ‘erences in Asymmetry Direction Between Measurements

 

 

 

 

 

 

Spitalfields St. Georgian Forensic
Bride’s
o‘ 9 d' 9 d' 9 d' 9
Length vs Humerus V V V V V
Biepicondylar
Width Femur
Length vs Humerus V V V V V
Head
Diameter Femur 8
Biepicondylar Humerus V
Width vs
Head Diameter Femur
Humerus vs Femur Length 8 8 8

 

m. Table 84 summarizes the results of comparing the direction of asymmetry

216

among various pairs of measurements. Statistically signiﬁcant patterns of same-Side
dominance are indicated by a check (V) and opposite-side dominance by a cross (8). If
physical activity were associated with bone length asymmetry, then there should a
signiﬁcant same-side relationship between direction of asymmetry in biepicondylar width
and length of the bone.

This is clearly the case for the humerus, except for the Forensic males. The
Spitalﬁelds sample displays a signiﬁcance at p < .01, and the St. Bride’s males and
females both exhibit signiﬁcance at the level of p < .10. The apparent lack of
signiﬁcance for the St. Bride’s sample may be an artifact of small sample size, on which
the binomial test is highly dependent. Ifthe St. Bride’s sample size were doubled,
assuming the same ratio of same-side to opposite-side dominance, both the males and
females would exhibit Signiﬁcance at p < .05. When the Georgians are pooled, the extent
of same-Side dominance in humerus length and biepicondylar width is very highly
signiﬁcant for both males and females (p < .005).

There is virtually no such relationship in the femur for any of study samples, and
there appears to be no relationship between directional asymmetry, femoral length, and
biepicondylar width. It is noteworthy that femur length is longer on the left in all the
groups, yet both head diameter and biepicondylar width are generally right-dominant.
These trends are not statistically signiﬁcant except for the Forensic males, which
inexplicably display a very highly signiﬁcant opposite-side dominance pattern between
femoral length and head diameter. It is noteworthy that the comparison of side-
dominance patterns in humerus and femur length asymmetry demonstrates a signiﬁcant

proportion of cross symmetry in the Georgian groups which is not seen as strongly in the

217
Forensic sample.

In contrast, the distinct patterns of same-side dominance in the humerus support
this hypothesis. Because paired humeri are more likely to be subject to bilaterally
dissimilar activity patterns than are the femora, and because the epicondyles of the
humerus are more associated with muscle attachments than are the epicondyles of the
femora, the evidence from the humerus for an activity-related eﬂ‘ect on humerus
asymmetry is more compelling than the seemingly contradictory evidence from the

femur.

CHAPTER 6—DISCUSSION

The analysis of bilateral asymmetry in human limb bones is an increasingly
prominent aspect of the biocultural approach to human skeletal biology. The literature
review in Chapter 2 indicates that several recent studies draw attention to asymmetry in
the cross-sectional geometry of long bone diaphyses. Researchers have acknowledged
the presence of directional asymmetry in linear dimensions of long bones as well;
however, the existence of these asymmetry patterns are either reported with little
comment, or attributed to the effects of ﬂuctuating asymmetry. As a result, the nature of
metric asymmetry patterns in the linear dimensions of long bones has not been well
characterized in human populations. The purpose of this study is to set a foundation for
ﬁrture biocultural analyses by assessing the nature of asymmetry variation in related and

unrelated skeletal populations.

W

This study assesses limb bone asymmetry patterns in two skeletal samples ﬁ'om
Georgian—era London church crypts (St. Bride’s, Fleet Street, and Christ Church,
Spitalﬁelds), which are compared with ﬁgures derived ﬁ'om the Forensic Data Bank at
the University of Tennessee. The Georgian crypt samples are particularly valuable
because they are drawn from a group of individuals associated with a common setting in
Space and time. These characteristics set them apart from the Forensic sample, which
represents individuals ﬁ'om the United States in the twentieth century. The lifeways of
the Forensic individuals are not documented, so it is diﬁicult to conﬁdently oﬂ‘er a basis

of comparison between them and the Georgians in terms other than their setting in space

218

219

and time. It is important to note-and this cannot be emphasized too strongly—that the
hypotheses in this study are constructed to draw inferences about the nature of bilateral
asymmetry patterns in major long bones, and not to draw inferences about the skeletal
samples being studied. The only assumptions that can conﬁdently be made about the
samples is that the Georgians are associated with a common setting in time and space
which is distinctly diﬁ‘erent ﬁ'om that of the Forensic sample.

Six paired measurements are compared among the three skeletal samples:
maximum length, vertical head diameter, and biepicondylar width of the humerus; and
maximum length, head diameter, and biepicondylar width of the femur. Study
hypotheses test whether the samples and the sexes display statistical independence in the
distribution of (a) direction of asynunetry, (b) signed and unsigned magnitude of
asymmetry; and (c) size of the linear dimensions themselves. The two Georgian samples
are ﬁrst assessed separately, then pooled and compared with the Forensic sample; in each
case the sexes are treated separately. The linear dimensions are assessed using I and 2
test statistics, the direction of asymmetry is assessed using x3 analysis, and the signed and
unsigned magnitude of asymmetry are assessed using the Mann-Whitney U test statistic.
The signiﬁcance of same-side versus crossed symmetry patterns both within and across
bone pairs are assessed using the binomial test.

AS one would expect, in each measurement there is a clear pattern of sermal
dimorphism in character size, with the males Signiﬁcantly larger than the females in all
cases. There is also a highly signiﬁcant diﬂ‘erence in the bone dimensions when the
Georgian males and compared with the Forensic males, and when the Georgian females

are compared with the Forensic females; in this instance the Georgians are consistently

220

smaller in size. It is not surprising that the Georgians are relatively smaller in size than
the twentieth-century sample; however, the Georgians display a substantially greater
magnitude of asymmetry than the Forensic sample. Speciﬁcally, the Georgian females
display the greatest asynunetry in humerus length, and the Georgian males display the
greatest asymmetry in femur length. Asymmetry patterns are more oﬁen signiﬁcantly
diﬂ‘erent in Georgian-Forensic comparisons than in comparisons between the sexes of
each sample, or between same-sex subgroups of the Georgian samples; this indicates that
setting in time and space, but not sex, is a determinant of asynunetry patterns in long
bone dimensions. One may feel compelled to speculate about speciﬁc environmental or
activity-related factors that account for the diﬁ‘erences between the eighteenth- and
twentieth-century populations, but the nature of those factors is elusive because little is
known about the speciﬁc lifeways of the individuals that comprise the Forensic sample.
It is possible to draw some inferences (see Hypothesis 4) about physical activity
in these samples as they are revealed in skeletal morphology, if one is willing to accept as
given France’s (1988) assertion that those aspects of bone which are muscle attachments
are particularly subject to modiﬁcation by activity. There is a statistically signiﬁcant
same-side relationship between biepicondylar width and maximum length of the humerus
which suggests a correlation between physical activity and bone length asymmetry. The
fact that the same-side relationship pattern does not persist in femora is still consistent
with the suggestion, since the epicondyles of the femur are more associated with
ligamentous attachments, rather than muscle attachments, and would be less likely

subject to activity-related modiﬁcation.

221

Another noteworthy ﬁnding of this study relates to direction-of-asymmetry
patterns as contrasted between the humerus and femur. Notwithstanding the existence of
variation between the Georgian and Forensic samples, in each there is a net same-side
dominance in each of the dimensions measured on the humerus, yet this is not the case
for the femur. Among those individuals who display measurable asymmetry in both
humerus length and epicondylar width, in particular, there is a statistically signiﬁcant
same-side relationship between these measurements. In contrast, there is a general
tendency for left-dominance in femoral length to be coupled with right dominance in
femoral head diameter and biepicondylar width. These tendencies in the femur are not
statistically signiﬁcant, except for the case of the Forensic male sample, which shows the
crossed pattern to be highly statistically signiﬁcant. Nonetheless, these observations
underscore the fact that a given bone pair may be right-dominant in some dimensions at

the same time that they are left-dominant in others.

W

In addition to testing the hypotheses that were presented at the end of Chapter 4,
the primary value of this research is as a methodological critique of asymmetry
assessments. Although the issues discussed in this section—and which were stated in
Chapter 2—are concerned speciﬁcally with the assessment of linear asymmetry patterns,
some of the implications apply to the assessment of cross-sectional asymmetry patterns .-
as well. This section outlines the most signiﬁcant issues, and their implications for future

research.

222

Given the lack of attention directed to linear asymmetry patterns in the
osteological literature, the ﬁrst signiﬁcant ﬁnding to come ﬁ'om this project is the
unequivocal afﬁrmation that the directional nature of asymmetry in both humerus and
femur lengths is a real phenomenon, and that it is inappropriate to simply attribute bone
length asymmetry to the phenomenon of ﬂuctuating asymmetry. Because the asymmetry
truly is distributed in a directional manner, and because the extent of the directionality
varies between the humerus and the femur, there arises two troublesome methodological
issues relating to how the asymmetry should be reported, particularly when making

comparisons between skeletal samples.

Reporting Asymmetry Magnitude—Signed or Unsigned?

The ﬁrst issue is whether to employ signed or unsigned values in reporting the
magnitude of the asymmetry that is used as the basis for comparison between
populations. The choice of strategy will aﬁ‘ect the comparison, and the eﬂ‘ect will be
diﬁ‘erent in diﬁ‘erent bones insofar as their asymmetry distribution patterns vary. In the
case of the humerus, where the prevalence of directionality to one side within a
population is relatively more pronounced than in the femur, there is relatively greater
concordance between signed and unsigned magnitude values. In contrast with the
prominent right-dominance in humerus length, within a population the directionality of
femur length dominance is less one-sided. That is, even though a majority of individuals
are left-dominant in femur length, there is also a substantial minority of individuals that

are right-dominant. Because of this, the average signed value for magnitude of

223
asymmetry within a population will appear to be quite small (as seen in the sample data
example from Chapter 2).

There is no clear preference for using either the signed or unsigned values for
determining population averages, Since each has distinct disadvantages. When signed
values are used the true magnitude of the asymmetry is obscured by the canceling eﬂ‘ect
of positive and negative values. When unsigned values are used the direction of the
asymmetry is totally removed from the assessment. Even if unsigned values are used in
conjunction with a separate assessment of directionality the relationship between
direction and magnitude is still not clearly portrayed. The only effective way to
characterize asymmetry patterns, then, is to report them in terms of each of the three
attributes described here: unsigned asymmetry magnitude, signed asymmetry magnitude,
and directionality.

While the use of these three distinct attributes may be adequate for characterizing
the ﬁndings for a single skeletal sample, it also renders comparisons between skeletal
samples substantially more complex. However, the use of nonparametric statistical
techniques which involve rank ordering of the individuals in the samples being compared
can be a powerful tool for undertaking the assessment of signed asymmetry magnitude.
In the situation where two samples are being compared the Mann-Whitney U statistic
oﬁ'ers an attractive tool for assessment. In the case of nominal variable assessment of

asymmetry direction, x3 analysis is appropriate if sample sizes are suﬁciently large.

224
The Asymmetry Threshold Issue

The second primary issue surrounds the arbitrary nature of asymmetry labels, and
the manner in which increasing the measurement threshold of symmetry diﬁ‘erentially
affects populations depending on their asymmetry distribution patterns. These
considerations are relatively trivial in cases of non-directional asymmetry, but they can
have serious implications in situations where asymmetry patterns are clearly directional.

As outlined in Chapter 2, metric observations of paired skeletal elements can be
placed into three mutually exclusive categories of leﬁ-dominant, symmetrical, and right-
dominant. Ifthe distribution of asymmetry is normal with a mean of zero, increasing the
asymmetry threshold will draw an equal number and proportion of individuals from both
the right-dominant and left-dominant categories into the symmetrical category.
However, if the distribution is directional then increasing the threshold will draw a
greater number of individuals from one side and a greater proportion of individuals ﬁ'om
the other. In the case of humerus length, which has a strong tendency toward right
dominance, this means that increasing the threshold ﬁom one millimeter to two
millimeters will shiﬁ more individuals ﬁ'om the right-dominant category to the
symmetrical category than are shifted from the left-dominant category to the symmetrical
category. However, a greater proportion of the left-dominant individuals will shiﬁ ﬁ'om
the left-dominant category to the symmetrical category than are shifted from the left-
dominant category to the symmetrical category.

When patterns of direction in asymmetry of humerus length and femur length are
assessed using two diﬂ‘erent thresholds of asymmetry (one-millimeter and two-

millimeters) the statistical signiﬁcance of the differences between the sexes was shown to

225

vary. Speciﬁcally, diﬁ‘erences in the direction of humerus length between the Georgian
and Forensic samples appear statistically signiﬁcant with a greater level of conﬁdence
when assessed at the higher symmetry threshold; this is true for both the males and
females. Likewise, the higher asymmetry threshold is associated with greater conﬁdence
in the statistical signiﬁcance of direction of asymmetry patterns in femur length between
the sexes in both Georgian samples. Researchers need to bear in mind that the threshold
of asymmetry chosen for a given study will affect the results of the analysis; to facilitate

eﬂ‘ective comparisons between sample groups, it is important that the threshold values be

stated explicitly in any study of metric asymmetry.

Are Asymmetry Distribution Patterns Unimodal?

Another intriguing ﬁnding derived ﬁom this research is that the distribution of
asymmetry in humerus and femur lengths in a skeletal population does not appear to be
unimodal. The relatively small sample sizes associated with each group here leads this
assertion to be made with caution, but the consistency in the bimodal pattern among the
males of the populations is remarkable. While it is possible that the apparent bimodality
reﬂects two distinct subpopulations in each sample, there is no obvious documentary
evidence to support such a suggestion. Detailed discussion of the bimodality issue goes
beyond the hypotheses in this study, but it is clearly an issue that merits follow-up in a
future investigation.

Nonetheless, one implication of this ﬁnding is that statistical techniques which
rely on assumptions of unimodal normal distributions may be essentially inappropriate

for characterizing the true nature of postcranial bone length asynunetry. Whether the

226

asymmetry of other components of the long bones (head diameter, biepicondylar width,
etc.) is also non-normally distributed is unclear. The data presented in the current study
do not suggest a unimodal normal in those measurements. Because nonparametric
statistical techniques do not assume an underlying unimodal normal distribution in the
character being studied, there is additional reason for employing these techniques in the

assessment of directional asynunetry patterns.

Summary

Perhaps the most important contribution of the present study is to provide a
framework for ﬁrture studies of the nature of skeletal asymmetry in populations that
accounts for both the direction and magnitude of asymmetry. It is critical to underscore
that parametric tests of asymmetry magnitude offer an incomplete picture of the true
nature of directional asymmetry, and that the signiﬁcance of asymmetry patterns between
populations is most appropriately assessed by median-based statistics (such as the Mann-
Whitney U) or nominal scale statistics (such as 12 analysis of asymmetry direction
between populations). Future studies of asymmetry, both of linear dimensions and of

diaphyseal morphology, should consider the use of these statistical techniques.

I In. IIIIIIIOQIUILIIIII 0'1 '9'Ié'.\
Typically skeletal biologists are frustrated by the lack of supporting
documentation associated with the populations that they study. In the case of prehistoric
American populations, for example, the lifeways of grOups is spoken of in very broad

terms. Such is not the case with the Skeletal populations of 18th century London. Not

227

only were parish and municipal records well-maintained, there have also been many
books written on the topic of Georgian London. Some have dealt with London in general
(Bayne-Powell 1938; George 1965; Marshall 1968; Porter 1994), others with London as
the focal point of the nation as a whole (Porter 1990), and yet others with speciﬁc
geographic areas within the metropolitan area of London. For example, volumes have
been written on the East End of London by Smith (1939) and Rose (1951), among others.
In fact, several books have been written about the history of Fleet Street itself (for
example, Bell 1912, Boston 1990, and Morgan 1973).

It would seem that this abundance of information could greatly simplify the task
of characterizing the lifeways of the individuals interred in the crypts of Christ Church
and St. Bride’s. To some extent this is true; however, the historical data make clear that
there is a wide level of variability in lifeways even among the so-called “middling sort”
of 18th century London. One might argue that it would be possible to determine a
meaningful subset of the population, and try to derive a representative sample of those
individuals. In the case of the church crypt sample, for example, the interred individuals
shared some characteristics. With respect to socioeconomic status, they were typically of
the “middling sort”, as Cox and Molleson (1993) indicates. Indeed, the ﬁnancial cost of
a lead coﬁn was not cheap, and most of less well-to-do individuals were interred in the
Churchyard (Litten 1991). This is not always the case, though, Since a few individuals of
rather meager income were interred in crypts, either because of Speciﬁc ties to the
Church, or because more well-to-do members of the family were associated with the

Church.

228

The issue of representativeness has taken a more prominent role in discussions of
osteological analysis of the past decade, and reﬂected by the fact that a symposium on
representativeness was held at the 1993 meetings of the American Association of .
Physical Anthropologists. The problem is generally phrased as such: How accurately
does a Skeletal population represent the community of individuals ﬁ'om which it was
derived? With the recent surge in studies involving historic Skeletal populations,
particularly those populations for which contemporary documents exist, this question has
become much more important.

Another consideration sets these Skeletal populations apart ﬁ'om the generally
accepted understanding of most archaeological skeletal populations. It is important to
recognize that persons interred in a given crypt did not necessarily reside in close
proximity to the church. Cox located the addresses at baptism and marriage for twenty-
eight individuals from the Spitalﬁelds identiﬁed population, and found that “only four
were baptized, married, and died resident in the parish of Christ Church” (Cox 1989226).
A review of address listings at the time of death for the St. Bride’s population reveals a
similar phenomenon. Nonetheless, the majority of individuals from both Georgian
samples were residents of greater London (Cox and Molleson 1993; Scheuer, personal

communication).

Reliability of Written Documentation
Other assumptions about the persons located in the crypts are dashed in the
process of research. One of the most important ones has to do with the reliability of

written documents. Bowman 3131,0993) found this to be a troubling aspect of their

229

research of the St. Bride’s collection. They identiﬁed three particular problematic areas:
(1) Contradictory information; for example, when church records and death registration
records gave different information about a single individual. (2) Ambiguous information;
with respect to interpreting cause of death ﬁ'om these documentary evidence, for
example, it is difﬁcult to associate the term “Decline” with a particular pathological
process. (3) Deliberate corruption of records; the classic example ﬁom the St. Bride’s
collection is the falsiﬁcation of cause of death so as to not publicly reveal a case of
suicide (Bowman £1.11. 1993).

Cox (1993) is stronger in her critical stance toward the relationship between
skeletal biology and history. She argues that skeletal biologists risk drawing spurious
conclusions if they are not critical in their assessment of the historical record of the
population they investigate. Waldron’s (1991) study of the correlation between
occupation and osteoarthritis among the Spitalﬁelds weavers exempliﬁes Cox’s concern.
Since a signiﬁcant number of the persons interred in Christ Church were listed in historic
documents as silkweavers, on the surface Waldron’s task appeared to be quite
straightforward. However, on closer examination Waldron found that the job title of
silkweaver could have a range of meaning. Joumeymen weavers, for example, were
known for working extremely long hours, while master weavers had a much more relaxed
way of life. Because master weavers were typically journeymen themselves at a younger
age, it is unclear what meaningful distinction Should best be drawn between the two

groups of weavers.

230
A Sex-Gender Distinction?

Another intriguing issue relates to the apparent differences in asymmetry patterns
between the males of St. Bride’s and Spitalﬁelds, as well as between the females of the
two sites. Ifthe women of St. Bride’s and Spitalﬁelds are comparable in terms of space,
time, and socioeconomic status, as well as sex, they should therefore Show similar
asymmetry patterns; likewise, the men of the two sites should be comparable in terms of
asymmetry as well. However, they do not seem to be similar in this regard, and the most
obvious explanation for the discrepancy is that there are issues related to gender and the
environment which are different for the two populations.

One way to characterize a distinction between genetic sex differences and
environment- or activity-related differences between the sexes is to refer to the
diﬂ‘erences as gender diﬁ‘erences, rather than sex diﬂ'erences. The concept of gender
does not appear regularly in the skeletal biology literature. In studies of past human
populations, skeletal biologists have accepted that there is a high level of consistency in
gender-stratiﬁed physical activity patterns. In other words, it appears to be assumed that
all women within a skeletal sample (that represents a population) engage in comparable
physical activities, and that the same is true for the men within the sample. The
assumption is embraced to the extent that no efforts are taken to separate the genetic
effects of sex from the environmental effects of gender-related diﬁ‘erences in skeletal
morphology. To be fair, those researchers studying undocumented skeletal populations
in particular do not have any way of knowing with any certainty the extent of consistency
in gender-stratiﬁed physical activity patterns, and hence have no basis for distinguishing

between the eﬂ‘ects of sex and gender.

231
Perhaps with the availability of historical documentation and well marked

populations for study there iS room in the literature of skeletal biology to clarify a
distinction between “sex” differences and “gender” differences in the human skeleton.
The latter term can be a useful tool applied to variation in skeletal morphology which is
not readily attributable to genetic inﬂuences. In cases where males and females within a
skeletal population manifest diﬁ‘erential patterns of physiological stress markers which
would be associated with differential access to food resources, the term “gender” is more

appropriately descriptive than “sex” to label the distinction.

CONCLUSION

The introductory chapter opened with reference to a ﬁctional detective who
applies principles of osteological analysis to reconstruct the lives of individuals on the
basis of their skeletal remains. In reality, such reconstruction is not a simple and
straightforward task, as those who perceive the pitfalls of the osteological paradox are
quick to warn. Even in the seemingly simple task of assessing bilateral asymmetry in
linear dimensions of long bones, there are several methodological issues, and problems in
interpretation, that complicate the assessment. Because of the growing importance of
bilateral asymmetry in biocultural osteology studies, these issues and problems merit a
close examination.

Drawing on recent scholarly interest in biocultural interpretations of bilateral
asymmetry in limb bone morphology, this project was initiated to determine the extent to
which three factorso-gender, physical activity, and setting in time and space—are

observed to aﬂ‘ect patterns of bilateral asymmetry in the linear dimensions of long bones.

232
There has been little research on this topic, since earlier studies of asymmetry predated

the biocultural approach and were primarily descriptive in nature. The more recent
research has focused on cross-sectional analysis of limb bone diaphyses, and not
addressed the phenomenon of directional asymmetry in the linear dimensions of the

bones.

The Hypotheses

The four hypotheses presented in Chapter 4 were constructed to assess the
likelihood that the skeletal samples under study would be drawn ﬁ'om a single
population. Results of the hypothesis testing are presented at the end of Chapter 5, and

summarized here, as well (Table 85):

(1) II:

 

‘°‘I.'_ 13-! ‘I‘I'l I'll I' 0 AI . rill. 'I 91111" 'l' 'x'l’ I...Il")l,"l‘
This hypothesis is unequivocally supported by z and t test comparisons of means for the

study samples.

 

W5. This hypothesis is not supported either by 36' tests of asymmetry

direction or Mann-Whitney U assessments of the signed and unsigned magnitude of

asymmetry patterns.

 

- _ . The study results oﬁ‘er strong

 

support for this hypothesis, but the diﬂ‘erences are not statistically signiﬁcant for all

measurements for both sexes.

 

This hypothesis is strongly supported by the ﬁndings of the study. There is a statistically
signiﬁcant same-side relationship between biepicondylar width of the humerus and
humerus length for each of the study subgroups except the Forensic males. There is not
a Similar relationship in the femur, however. This is not inconsistent with the hypothesis,
however, since the biepicondylar width of the femur is not associated with muscle

attachments in the same way that the biepicondylar width of the humerus is.

234

Table 85 Summary of Statistical Signiﬁcance Between the Sexes
_

ST. BRIDE’S SPITALFIELDS FORENSIC

Direction (1 mm) <.05
(2 mm)

HL Signed <05 <05
Unsigned <.05

 

Direction
EH Signed
Unsigned

 

Direction
HB Signed
Unsigned

 

Direction (1 mm) <.05 —
(2 mm) <.01 <.05

FL Signed
Unsigned <.05

 

Direction
FH Signed
Unsigned
Direction
FB Signed <.005
Unsigned <.05

 

235

Table 86 Summary of Statistical Signiﬁcance Among Populations

 

 

 

 

 

 

GEORGIANS GEORGIAN I FORENSIC
d' 9 d' 9 -
Direction (1 mm) <.001 <01
(2 mm) <.0001 <.001
HL Signed <.0001 <.0001
Unsigned <.005 <.0001
Direction
HH Signed <.05 <.05
Unsigned <.05
Direction
FIB Signed
Unsigned
Direction (1 mm) —
(2 mm) <.05
FL Signed <.01
Unsigned <01 <05
Direction
FH Signed <.01
Unsigned I
Direction <.01 <.OI
FB Signed <.05
Unsigned

 

..

236

Implications of the Study for Biocultural Questions

While the explicit goal of this dissertation is to set a foundation for future
biocultural research, there are biocultural implications in the study results themselves.
The most important implication arises from the evidence that setting in time and space,
but not sex, appears to inﬂuence patterns of asymmetry in the linear dimensions of long
bones. This means that there is a legitimate basis for addressing biocultural questions by
assessing long bone linear asymmetry since some aspect (or aspects) of culture—physical
activity, environmental stressors, nutritional status, and/or socioeconomic
factors—aﬁ‘ects the manisfestation of these asymmetry patterns. Further research with
better controlled populations will be necessary to establish the relative importance of
these factors in determining asymmetry patterns, but given the results of this study it
appears that such research would be fruitﬁrl.

In addition to testing the hypotheses listed above, this study revealed several other
interesting patterns that speak to bioculturally focused research questions. They include
the following:

(1).. n 'I I'-1'°3_ JIIIIOIOEI'IO, ~juttiz._1,-’t an 91941.19-
..- H ”H, . ..1 H". :- “ru . -i ”11712.1 .- - - y t,- ”H“ . ._
W This characteristic was
consistent in all three of the study populations, and was evident in both sexes. There are
two implications of this ﬁnding. First, it indicates that researchers cannot assume that a
side-dominance pattern in one linear dimension of a bone pair should be associated with
a similar side-dominance pattern in other aspects of the bone. In other words, a

researcher who studies asymmetry in femoral head diameter in one Skeletal sample would

237

not have a legitimate basis for comparing those patterns with results of another
researcher’s study of asymmetry in femur length. A second implication is that, within the
femur, asymmetry in bone length, head diameter, and biepicondylar width are not

associated with physical activity and therefore would have limited application in

biocultural studies that address activity patterns.

 

the sexes which do not appear to reﬂect genetic differences between males and females.
Ifthat were the case, then the Georgian and Forensic females would show greater
consistency in their asymmetry patterns. Because these diﬁ‘erences result ﬁ'om the
interaction of culture and biology, characterizing the diﬁ‘erences as “sex-related” may be
inaccurate. When the impact of culture is juxtaposed with sex factors to aﬂ‘ect skeletal
morphology, it may be more appropriate to apply a concept that acknowledges the
inﬂuence of culture, such as gender, to describe these differences. At the same time, it is
important to not Simply employ the term “gender-related” in lieu of “sex-related” to
characterize all differences in male and female skeletal material, since these are two
distinct concepts. That is, there are clearly some diﬁ‘erences in skeletal morphology,
such as sexual dimorphism in the size of the femoral head, that are truly sex-related and
that do not result directly from an interaction of culture and genetic diﬁ‘erences.

(3) It- . -.' I: ._ = 7-1-. gm -' y”; 1 112-12- It or ‘MIIII an
Wm. Although the Forensic humeri and femora are
signiﬁcantly larger than the Georgians in all six of the linear dimensions under study in

no instance do they present the greatest magnitude of asymmetry. This is in spite of the

238
fact that asymmetry magnitude is calculated four different ways (signed, unsigned,
percentage signed, and percentage unsigned). In fact, in the case of humerus length,
humerus head diameter, and femoral head diameter the Forensic males and females
display the smallest magnitude of asymmetry in all four variations of its calculation.

This ﬁnding is particularly surprising given the obvious diﬁ‘erences between the
Forensic sample and the Georgian sample. Although both samples are derived ﬁ'om a
relatively restricted temporal range the Forensic sample reﬂects a much broader range of
socioeconomic status and geographic origin than the Georgians. This would suggest that
the Forensic sample Should Show a greater range of variation than the Georgians—not the
reverse.

It remains to be seen exactly what factors underlie the diﬂ’erence. One possibility
is that the Georgians represent a culturally restricted pOpulation which shows an
abnormally high level of directionality to asymmetry. Secondly, there may be something
signiﬁcantly diﬁ‘erent about the lifeways of the more modern Forensic Data Bank sample
which is associated with the difference. It may be that twentieth-century youths engage
in more sedentary behaviors than their eighteenth-century counterparts; this would be
consistent with less directionality in asymmetry patterns. This is an interesting
speculation, since it suggests that the female Georgians, in showing greater asymmetry
than the other groups, were the least sedentary of all the sample skeletal populations in
the study.

(4) . .,10_°I t- r o 1‘er 99.10“on .5 - 2011.1.13.0'.ll0‘ll‘ tic-1: .I'

:1" Ili.i‘~‘?.l‘ 9111?.93‘ '3 '2‘ 3. 1.1191215 '1" II’JJ HII FII‘II IL! I‘d.

WW.

239

Because femur length is the bone measurement most commonly used in univariate
estimations of living stature, this ﬁnding suggests that the St. Bride’s population were
taller than the Spitalﬁelds population. The differences in mean femur length between the
two skeletal samples are not statistically signiﬁcant, however. This ﬁnding is consistent
with the assessment of East London vs. West London diﬂ‘erences in socioeconomic status
which were outlined in Chapter 3. That is, the celebrated poverty of London’s East End
may be manifest in the shorter stature of the Spitalﬁelds skeletal population. In spite of
the two crypt populations having a “middle class” socioeconomic status at the time of
death, it is possible that the effects of relative poverty in childhood might persist as a
permanent reduction in adult stature. Addressing this issue is an excellent example of a
future comparative endeavor involving these two samples and the documentary record

that surrounds them.

Suggestions for Further Study

As long as researchers are realistic in their expectations of studies involving
documented historic skeletal samples, the St. Bride’s and Spitalﬁelds identiﬁed skeletal
collections offer skeletal biologists a unique opportunity for assessing skeletal
morphology within a well-documented and culturally homogeneous pair of skeletal
samples. The trend noted in the previous paragraph indicates that it is still unclear the
extent to which the two Georgian skeletal samples should be interpreted as a single
population or as two distinct populations. Further studies on these two groups which

combine skeletal analysis with ethnohistorical research will reﬁne our understanding of

240

meaningful similarities and differences in the Skeletal morphology of the people of St.
Bride’s, Fleet Street and Christ Church, Spitalﬁelds.

In addition to setting the stage for future comparative studies involving these two
Georgian Skeletal samples, this research also sets a baseline for comparative studies of
asymmetry in the linear dimensions of long bones in other reference skeletal samples to
further characterize if asymmetry patterns can be used to draw inferences about
biocultural issues in unknown skeletal populations. As records of osteometric
observations become computerized it will become easier to undertake the statistical

analysis that would adequately compare varied populations.

LIST OF REFERENCES

Acheson RM, and Fowler GB (1964) Sex, Socio-economic status, and secular Increases

in stature: A family study. BdﬁshloumalﬂmenmandjmialMEdicm
18. 25-34.

Adams M, and Reeve J (1987) Excavations at Christ Church, Spitalﬁelds. Ann'gujm
61:247-56.

Aiello L. and Dean C (1990) Aalntmdrrctiamoﬁumanﬂoluﬁonammwx
LondonzAcademic Press.

Albrecht GH (1983) Humidity as a source of measurement error in osteometrics.

WW 60(4)=517-20

Angel JL (1975) Middle class Skeletal diﬂ‘erences (abstract). WW
EhxsisaLAnthmmlm 42 233

Annelagos GJ, Carlson DS, and Van Gerven DP (1982) The Theoretical foundations

and development of skeletal biology. In W531
W289 (Frank Spencer, ed.) New York: Academic Press.

Angel JL, Kelly JO, Parrington M, and Pinter S (1987) Life stresses of the free black
community as represented by the First Aﬁican Baptist Church, Philadelphia,

1823- 1841 Amsdmloumﬂﬂhxsisalﬂthmolm“ 213-29

Bailit HL, Workrnen PL, Niswander JD, and MacLean CJ (1970) Dental asymmetry as
an indicator of genetic and environmental conditions in human populations.

Hummﬂiszlm 421526-33-

Baskerville RF (1992) On the directional asymmetry of rhesus macaque forelimb bones.
WW 87: 497-8

BassWM (1987) 6112.1 01.100“; ‘ ' . = :_.(3rded..) Special
Publication No.2 of the Missouri ArcheOIogicaI Society: Columbia MO.

Bayne-Powell R (1938) W. New Yorsz. P. Dutton.

 

241

242

Beeton MM. and Chancellor EB (1969) AlmthmLLandonAhothslcaLllZiL

191-11 2.0011 o artr“; 0. -or‘ [10' we I
irar 0.1211191" “inf-or to 101001. 2. «1.91.1 I' 50
163711111 :0 0.310 0 1.1...117'21‘0 unuxexv 32.1‘ .2901.
LondonzBenjamin Blom, Inc.

Bell WG (1912) W. London:Sir Isaac Pitrnan and Sons,
Ltd.

Bennett KA (1993) W (2nd ed).
SpringﬁeldzCC Thomas.

Boston R (1990) The Essential Fleet Street: it’s History and Inﬂuence.
London:Blanford.

Bowman JE, MacLaughlin SM, and Scheuer JL (1993) A study of documentary and
skeletal evidence relating to 18th and 19th century crypt burials in the city of
London: The St. Bride’s Project. Paper delivered at the 1993 Meeting of the

American Association of Physical Anthropologists. Wm
Aammpalm 91 (Special Meeting Supplement): 60.

Bridges PS (1989) Changes in activities with the Shift to agriculture in the southeastern
United States. CunenLAnthmaclm 30(3):385-94.

 

Buskirk EFL Andersen KL, and Brozek J (1956) Unilateral Activity and bone and
muscle development in the forearm. W 27: 127-3 1.

Byers S (1994) On stress and stature in the “osteological paradox”. Cam
Anthtsnolcsx 35(3):282-4-

Campbell B (1992) Wag. New Yorszarper-Collins.

Chhibber SIL and SinghI (1972) Asymmetry in muscle weight in the human upper
limbs. W93 81:462-5.

Cohen MN (1989) W. New Haven: Yale University

Press.

CottretB (1991) II‘ ° ' :2.
Cambridge: Cambridge U Press.

 

Cox M (19898) The Huguenots of Spitalﬁelds: the evidence from the Christ Church
PrOI'eet WW 25(1) 21-38

Cox MJ (1993) A dangerous assumption: Anyone can be an historian! The lessons
from Christ Church, Spitalﬁelds, London. Paper delivered at the 1993 Meeting of
the American Association of Physical Anthropologists. Ameﬁmloumalaf

W 91 (Special Meeting Surrplement): 75

Cox MJ and Molleson T (1994) mm. OxfordzBritish Archaeological
Report...

Cox M and Scott A (1992) Evaluation of the obstetric signiﬁcance of some pelvic
characters in an 18th century British sample of known parity status. American

Won 39(4) 431-40

Dittrick J, and Suchey JM ( 1986) Sex Determination of Prehistoric Central California
Skeletal Remains Using Discriminant Analysis of the Femur and Humerus.

WWW 70: 3-9

Dwight T (1905) The size of the articular surfaces of the long bones as characteristics of
sex: an anthrepelogieal study. Amsdmloummm 4: 19-32-

Elkins A (1988) gum. New YorkzMysterious Press.

Ericksen MF (1982) How‘ ‘representative” is the Terry collection? Evidence from the
Proximal femur WW 59(3) 345-50

Falk D, Pyne L, Helrnkamp RC, and DeRousseau CJ (1988) Directional Asymmetry 1n

the Forelimb of W.
77: 1-6.

Fields SJ, Spiers M, Herskovrtz I and Livshits G (1995) Reliability of reliability

coeﬁcients 1n the estimation of asynunetry. WW
133111313219er 96 83- 7

Floud K Wachter K. and Gregory A (1990) WWW
Will-.1289 Cambridge Cambridge University

Press.

Floud R, Wachter K, and Gregory A (1993) Measuring historical heights: short cuts or
the long way round: a reply to Konrlos. W 46(1). 145- 54.

244

F ogel RW (1986) Physical grth as a measure of the economic well-being of
populations: The eighteenth and nineteenth centuries. In W
W (2nd ed., Vol. 3; F Falkner and JM Tanner, eds). New
York: Plenum Press; p. 263-81.

Fogel RW 9131, (1983) Secular Changes in American and British Stature and Nutrition.
9 I" 5,.199‘1'5 'I' ll‘i'l' 13.1911 991'.H.'1'I3.I' 01111.99!

W. Carnbridge2Cambridge University Press; p. 247-83.

France DL (1988) Osteometry at muscle origin and insertion in sex determination.

Amensanloumalaﬁﬁhxasalﬂthmom 75 15-26

Fresia AE, Ruﬁ‘ CB, and Larsen CS (1990) Temporal decline 1n bilateral asymmetry of

the upper limb on the Georgia coast. In Larsen CS, W2.
New York: Anthropological Papers of the American Museum of Natural History
#68.

Garn SM, Mayor GH, and Shaw HA (1976) Paradoxical bilateral asymetry in bone size

and bone mass in the hand WWW“: 209-
10.

Garn SM, Nagy JM, and Sandusky ST (1972) Diﬁ‘erential sexual dimorphism in bone

diameters of subjects of European and Aﬁican ancestry. WW
besicalAnthronolosy 37 127-9

GOOTSCMD (1965) WWI! Londonil’ot‘esﬁne BOOKS-

Giles E (1970) Discriminant Function Sexing of the Human Skeleton. In: 2mm]

111W (1‘ Dale Stewart. ed) Washinstoni Nattionttl
Museum of Natural History, pp. 99-109.

Goodman AH, and Rose JC (1990) Assessment of systemic physiological perturbations
from dental enamel hypoplasias and associated histological structures. W
ofﬁhysisalAnthronoley 33 59-110

Goodman AH, Thomas RB, Swedlund AC, and Annelagos GJ (1988) Biocultural
perspectives on stress in prehistoric, historical, and contemporary population

research WNW 311169-202
Gould SJ (1981) W. New York:WW Norton.

Graham TM, and Yarbrough JD (1968) Anthropometric studies of the long bones of the
“Shell Mound” Indians Amsdcmloummmmxsnmmaolm 28: 85-92

 

GwynnRG (1985) t .- - - 2.
'mﬁatnia. Routledge and Kegan Paul, London.

Harvey Warren (1968) Some dental and social conditions of 1696—1852 connected with
St. Bride’s Church, Fleet Street, London. W 12:62-75.

Helrnkarnp RC, and Falk D (1990) Age- and sex-associated variations in the directional

asynunetry of rhesus macaque forelimb bones. Ameﬁcanloumalﬂhysical
Anlhmmlou 33(2) 211-3

Hehnkarnp RC, and Falk D (1992) Reply to Baskerville. WW
Anmtmlﬂﬂ! 87:498.

Henneberg M, and VandenBerg ER (1990) Test of socioeconomic causation of secular
trend: Stature changes among favored and oppressed South Aﬁicans are parallel.

AmencaaloumalaﬂﬁbxucaLAatthlon 83 459-55

Hershkovitzl, Livshits G, Moskona D, Arensburg B, and Kobyliansky E (1993)
Variables aﬁ'ecting dental ﬂuctuating asymmetry in human isolates. Ammgan

MW 91(3) 349-65

Hooke, BGE (1926) A Third Study of the English Skull, with Special Reference to the
Farringdon Street Crania. Bjamejﬁka 18: 1-55.

Hrdlicka, A (1932) The principal dimensions, absolute and relative, of the humerus 1n
the white race WWW 16(4)I431-50

Hrdlicka, A (1939) Wm. Philadelphia: Wistar Institure of Anatomy
and Biology.

lsean MY. and Kennedy KAR (1989) Wu New
Yorszlan R Liss.

J antz RL, Hunt DR and Meadows L ( 1994) Maximum length of the tibia: How did
Trotter measure it? WWW ”(4)1525-8

Jantz RI- end Moore-Jensen PH (1988) W

W Report of Investigations #47
Knoxville: University of Tennessee Department of Anthropology.

Jolicoeur P (1963) Bilateral symmetry and asymmetry in limb bones omeas
uncricura and man RmaQanadrennedeErolm 22(34): 409-32

193’“ C, ““4 Stover E (1991) WWW

BostonzLittle-Brown.

246

Kieser JA (1992) Fluctuating odontometric asymmetry and maternal alcohol
consumrition. Aaaeluﬂﬂumeaﬁielm 19(5)3513-20-

Kieser JA and Groeneveld (1994) Effects of prenatal exposure to tobacco smoke on
develOPmental stability In children loumeleﬂcreaiefeeielﬁeaetieuad
Dexeleamearelﬂelesx l4(1)143-7

Komlos J (1993) The secular trend in the biological standard of living in the UK 1730-
1860. MW 46(1): 1 15-44-

Krogman WM, and lscan MY (1986) IheHummSkelereajaioLeasieMedieiae. 2nd
ed., Springﬁeld ILzCC Thomas.

Lanphear KM (1988) Health and Mortality in a Nineteenth Century Poorhouse Skeletal
Sample. Ph.D. diss., State University of New York-Albany.

Lanphear KM (1990) Frequency and distribution of enamel hypoplasias 1a a historic
skeletal samvle Ameamlorrmeleflhxsieemathmelm 81 ”-43

Latimer HB, and Lowrance EW (1965) Bilateral asymmetry in weight and in length of
human bones. W 152217.24.

Laubach LL, and McConville JT (1967) Notes on anthropometric technique:
Anthropometric measurements—right and left sides. WM

thsieeLAmhmmlm 26 357-70

Lees B, Molleson T, Amett TR and Stevenson JC (1993) Diﬁ‘erences in proximal femur
bone density over two centuries. Lang; 341(8846):673-5.

Littenl (1991) MW
Londoanobert Hale.

Livshits G, and Kobyliansky E (1991) Fluctuating asymmetry as a possible measure of
developmental homeostasis i humans: a review. Wm 63(4):441-66.

Lowrance EW, and Latimer HB (1957) Weights and linear measurements of 105 human
skeletons from Asia WW 101:445-59

MacDonell WR (1904) A Study of the Variation and Correlation of the Human Skull,
with Special Reference to Human Crania. him 3: 191-244.

MacDonell WR (1906) A Second Study of the Human Skull, with Special Reference to
the Moorﬁelds Crania. Bjomgtﬁka 5:86-104.

247

MacLaughlin SM (1987) An Evaluation of Current Techniques for Age and Sex
Determination from Adult Human Skeleton Remains. PhD. thesiszUniversity of
Aberdeen.

Mann HB, and Whitney DR (1947) On a test of whether one of two random variables rs

stochastieally larger than the other AaaelufMuhemetieuadjteﬁaies 18: 52-
4

Manning JT, and Chamberlain AT (1994) Fluctuating asymmetry in gorilla canines: a
sensitive indicator of environmental stress.

WW 2550343) 189-93.

Markow TA and Martin JF (1993) Inbreeding and developmental stability in a small
human population. WWW 20(4):389-94.

Marshall D (1968) Wm. New Yorkzlohn Wiley and Sons.

MartinPs and SellerK (1957) uhrbrrehdemmhronelmraazaemeaseher

 

(Band 1). Stuttgart: Gustav Fischer Verlag.

Maller AP, and Ponriankowski A (1993) Punctuated equilibria or gradual evolution:
ﬂuctuating asymmetry and variation in the rate of evolution. 1911M

W192! 161(3):?159-67.

Moore-Jansen PH, and Jantz RL (1989) WWW ‘
Weds]. Report of Investigations #47. Knoxvillerniversity of
Tennessee Department of Anthropology.

Morant GM (1926) A ﬁrst study of the craniology of England and Scotland from
neolithic to early historic times, with special reference to the Anglo-Saxon skulls
in London museums. Bigmetﬂka 18:1-55.

M0188“ D (1973) BMW; London:C Knight.

Manter AH (1936) A study of the lengths of the long bones of the arms and legs in man,
with special reference to Anglo-Saxon skeletons. Bigmetdka 281258-94.

Palmer AIL and Strobeck C (1986) Fluctuating Asymmetry: Measurement, Analysis,
Patterns We: 17: 391-421

Parsons PG (1914) The characters of the English thigh bone. WM
Physiology 48:238o67.

248

Pearson K, and Bell 1 (1919) WWW
Drapers’ Company Research Memoirs; Biometric Series X. Cambridge
University Press.

Perzigian AJ (1977) Fluctuating dental asymmetry: variation among skeletal
correlations WW 47I 81-8

Plato CC, Wood JL, and Norris AH (1980) Bilateral asymmetry in bone measurements

of the hand and lateral hand dominance. MW
Anthem 52127-31

Porter R (1990) WWW (211d ed-)- LondoniPenBUin-

Porter R (1994) W. LondonzHamish Hamilton.

Powers R (1960) Catalogue of the documented skeletal collection at St. Bride’s.
Unpublished, but available for review in the crypt of St. Bride’s Church, Fleet St.,
London.

RoseM (1951) W. London2Cresset Press.

Royle E (1987) WWW LondonIEdward Arnold.

Roy TA, Ruﬁ‘ CB, and Plato CC (1994) Hand dominance and bilateral asymmery in the

structure of the second metacarpal MW
94: 203- 11.

Rude G (1971) W. Berkeley: University of California

Press.

Ruﬂ‘CB and Jones HH ( 1981) Bilateral asymmetry in cortical bone of the humerus and
tibia—sex and age factors. W 53(1):69-86.

Ruff CB (1992) Biomecharrical analyses of archeological human skeletal samples.
Chapter 3 1n . - - ' : ... ‘ .SR Saunders
and A Katzenburg, eds, NY. erey-Liss,1nc.

 

Ruff CB, and Hayes WC (1983a) Cross-sectional geometry of Pecos Pueblo femora and
tibiae—A biomechanical investigation: 1. Method and general patterns of

variation WWW 601359-81

Ruﬂ‘, CB and Hayes WC (1983b) Cross-sectional geometry of Pecos Pueblo femora and
tibiae—A biomechanical investigation: 11. Sex, age, and side differences.

WW 601383-400

249

Ruﬂ‘CB, Trinkaus E, Walker A, and Larsen CS (1993) Postcranial robusticity 1n Hm.

1: Temporal trends and mechanical interpretation. Ameﬁmlmmﬂﬂhysjm
Anthremlesx 911-21-53

Saunders SR, Hering DA, and Sawchuk LA (1993) The nineteenth century cemetery at
St. Thomas’ Anglican church, Belleville. Skeletons, parish records, and censuses.
Paper delivered at the 1993 Meeting of the American Association of Physical

Anthr0polosists. MW 91 (Special
Meeting Supplement): 173.

Schaeﬂ‘er AA (1928) Spiral movement in man. Ioumalanemholomadﬁhysiolou
45:293-398.

Schell LM, Johnston FE, Smith DR, and Paolone AM (1985) Directional asymmetry of

body dimensions among white adolescents. W
Anthropology 67: 317-22

Schultz AH (193 7) Proportions, variability and asymmetries of the long bones of the
limbs and the clavicles in man and apes. W 9(3):281-328.

Schutkowski H (1993) Sex determination of infant and juvenile skeletons: I.

Morphosnostic features AmeaeealerrmeletﬁhxsieuAmhmolm 90(2) 199-
205.

Siegel S (1956) NeaneruaetﬁeSreﬂuieafouheBeheximLScienees New
YorkchGraw-Hill.

SinghI ( 1970) Functional asymmetry in the lower limbs. We; 77 :13 1-8.

Sirianni JE (1993) A comparison of death records from the Monroe County Poor House
with skeletal evidence from the associated Highland Park cemetery. Paper
delivered at the 1993 Meeting of the American Association of Physical

Anthropologists mmmammmm 91 (Special
Meeting Supplement): 181.

Smith BH, Garn SM, and Cole PE (1982) Problems of sampling and inference 1n the

study of ﬂuctuating dental asymetry. Ameaeealorrmelefﬁhxsirel
Aatbtomlesx 58I 281-9

SmithHL (1939) [1‘ 1‘ .
W. London: MacMillan and Co.

Sprent P (1989) We“ LondonIChapman and
Hall.

 

250

Steegmarr AT Jr (1985) 18th Century British Military Stature: Grth Cessation,
Selective Recruiting, Secular Trends, Nutrition at Birth, Cold and Occupation.

Walnut 57(1):77-95.

Steegrnan AT Jr (1986) Skeletal Stature Compared to Archival Stature in Mid-
Eighteenth Century America: Ft. William Henry. W
Amhmmlggy 71 :431-5.

Steegrnan AT Jr (1991) Stature m an early mid-19th century poorhouse population:

Highland Park, Rochester New York Amerieealoumeuﬂlhxsieal
Anthtepelos! 85(3)1261-8

Steel FLD (1962) The sexing of long bones, with reference to the St Bride’s series of

identiﬁed skeletons WW
Batuaeadlrelurd 92: 212-22.

Steele D6. and Bramblett CA (1988) IheAauammiﬂielcmﬂheﬂma
Skeleton. College Stationsz A&M University Press.

Stewart TD (1979) WWW Springﬁeld ILICC Thomas-

Stirland A (1986) W. Shire Archaeology: Aylesbury.

Stirland A (1993a) Asymmetry and activity-related change in selected bones of the
human male skeleton. Ph.D. ThesiszUniversity of London.

Stirland A (1993b) Asymmetry and activity-related change in the male humerus. Paper
delivered at the 1993 Meeting of the American Association of Physical

Anthropologists Ameamloumelﬂhreieemmhmmlm 91 (Special
Meeting Supplement): 190.

Stuart-Macadam P (1989a) Nutritional deﬁciency diseases: A survey of scurvy, rickets,
and iron-deﬁciency anemia. In MY Iscan and KAR Kennedy (eds).

Reecnstaretreagfhfeﬁerruhﬁrkelmn New York: Alan R Liss

Stuart -Macadam P (1989b) Porotic hyperostosis: Relationship between orbital and vault
lesions W 80:187-93-

Stuart- Macadam P (1992) Porotic hyperostosis: A new perspective. Ameﬁmlgumd
efﬁhysieuAnthreaelm 87 3947.

Thompson 131’ (1966) W New YorkIRandom

House.

251

Trinkaus E, Churchill SE, and Ruﬂ’ CB (1994) Postcranial robusticity 1n Hgmn.11:

Humeral bilateral asymmetry and bone plasticity. WWW
Anthropology 93:1-34.

Trotter M, and Gleser GC (1952) Estimation of stature from long bones of American
Whites and Negroes MW 101 463- 514

Trotter M, and Gleser GC (1958) A re-evaluation of stature based on measurements

taken during life and of long bones aﬁer death. WW
Anthropology 16: 79-123-

Ubelaker D. and Scammell H (1992) WWW

Utermohle CJ and Zegura SL (1982) Intra- and interobserver error in craniometry: a

cautionary tale. WWW 57(3) 303 10

Uterrrrohle CJ, Zegura SL, and Heathcote GM (1983) Multiple observers, humidity, and
choice of precision statistics: factors inﬂuencing craniometric data quality.

AmeacealoumeleﬂihxsleeLAathtoaolm 61(1)I 85-95I
Van Valen L (1962) A study of ﬂuctuating asymmetry. Exnlntinn 16:125-42.

Waldron HA (1991) Prevalence and distribution of osteoarthritis in a population from
Georgian London. AanelmﬂRherrmetieDiseese 50(5):301-7.

Waldron HA, and Cox M (1989) Occupational arthropathy: Evidence ﬁ'om the past.
BatishlorrmelafladustaeLMedieine 46:420-422.

Walvin J (1984) W. LondonzHutchinson.

Weiss NA, and Hassett MJ (1982) W. Reading MAzAddison-
Wesley.

Whittaker DK and Hargreaves AS (1991) Dental restorations and artiﬁcial teeth in a
Georgian population. WM 171 (11-12):371-6.

Wood JW, Milner GK Harpending HC, and Weiss KM (1992) The osteological
paradox: Problems of inferring prehistoric health from skeletal samples. Current

Anthropology 33(4)::143-70.

Wrigley EA (1987) ' -. ~ "
Ensign. Oxford: Basil Blackwul.

Wrigley EA, and Schoﬁeld RS (1989) MW
Wan. Cambridge: Cambridge University Press.

 

111011an STATE UNIV. LIBRnRIEs
"WWW WW "W” lllllll ml NIINIIW ll
31293014203339