71 .

.4“?
_ ... .51.. .

i

1:

fix

in .1.

.V 59.9“}.
, l. ..
.... .
5.11....

.3.

 

 

. .‘|.v nuvh.vluu .‘
.n,‘§$fv§mfl£§?w I: ; #

This is to certify that the
dissertation entitled

RECONSTRUCTING BEHAVIOR FROM ARCHAEOLOGICAL
SKELETAL REMAINS: A CRITICAL ANALYSIS OF THE
BIOMECHANICAL MODEL

presented by

GILLIAN BICE

has been accepted towards fulfillment
of the requirements for the

PhD. degree in Anthropologl

r I ’M'ajor Professor’s Signature

M03

Date

MSU IS an Affirmative Action/Equal Opportunity Institun‘or»

 

LIBRARY
MIehI/gan State
University

 

 

 

PLACE IN RETURN Box to remove this checkout from your record.
To AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

 

AUG 1.2.2005

U‘Zzu

V
04071:

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/01 chlRC/DatoDuepGS-pJ 5

 

RECONSTRU ‘l

REILlINS: Al :1'

 

 

RECONSTRUCTIN G BEI—LAVIOR FROM ARCHAEOLOGICAL SKELETAL
REMAINS: A CRITICAL ANALYSIS OF THE BIOMECHANICAL MODEL

By

Gillian Bioc-

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

Department of Anthropology

2003

RECONSTRUCT
REMAINS: A C“

A Illl mm Inn n.

 

irl‘lialns \\ Lb‘ IIl'\ I‘Il I} )t
lal'ldlllln in (Tl m-xt': l
phls‘li‘il dt mitt. 2 (
wilt-155 tlillrrrni l‘s it
$~H‘1fiit.~in' nl‘ m M.

II. , ..
imb. Ihrsr prrilii l

Till}

fIlN‘I’lLllIt in mi

ffllliitllt t‘\ dIlltilt‘ II

9+» :t ‘
haunt Irom It ill ‘w'

A (‘I If? ("I lI1\

.Jnruiimlli' adapt l

i.
13'"

MD. as \\'i illl\ I

Stir-r ‘
wdlt'r ‘ * '
lb} phiq.

t.
hf?” “’ - ' ‘
m“ dI found 1*

$5,

‘ Wm.

m r ii... m

Win;- a
‘1 I\

ABSTRACT

RECONSTRUCTING BEHAVIOR FROM ARCHAEOLOGICAL SKELETAL
REMAINS: A CRITICAL ANALYSIS OF THE BIOMECHANICAL MODEL

By

Gillian Bice

A biomechanical model for reconstructing behavior from archaeological skeletal
remains was developed in the early l9803. The model makes three predictions: 1) Group
variation in cross-sectional geometry reflects group differences in levels and types of
physical activity. ‘2) Group variation in sexual dimorphism of cross-sectional geometry
reflects differences in sexual division of labor. 3) Group variation in upper limb bilateral
asymmetry of cross-sectional geometric properties reflects differential usage of the upper
limb. These predictions have become unquestioned tenets of bioarchaeological research.
This dissertation utilized principles borrowed from informal logic to reconstruct and
critically evaluate the argument presented by anthropologists for their ability to infer
behavior from long bone cross-sectional geometry.

A core construct of the biomechanical model is the premise that bones
functionally adapt to their mechanical environment. This late 19‘h century construct,
known as Wolff’s Law, proposes a causal relationship between the mechanical forces
generated by physical activity and bone architecture. It is proffered as the principal
theoretical foundation for inferring behavior from long bone diaphyseal cross-sectional
geometry.

Since the mid-l9803, the study oflong bone structural variation has become a

virtual subspecialty in the field of biological anthropology, yet the biomechanical model

 

limit (hangi‘tl slut
[llll'IlllIilfI 1P1 ill IL’lt 4t.
time In Illt'l'lltllllt til I

l .lltiinuzli trialled

 

sawmill} aireptt-tl.
milling the l't'ldlll r

ll liftiltlt't ill IL'lt til rt '\

 

mm Mpltumtit!
Iii‘éttl on statistit All\
preemies ttlltill n 'l l.
fivSS—xt‘t‘llnlldl 2m in l‘
ftlnlri‘turc.

This Ill>$t‘l'l.l

human lichai'ii tr an

it i. ' ~ "
Raine ( onlrlllt

13H“ .1 ,, . .
,1 Tlrlnt \tirlallnn

ulllll be USctl to H

fun
,lltil'murt“. \\ llllt'

ET'lijm". .3 . .
* '[nf \flrlttilm‘

has not changed since its conception twenty years ago, and does not reflect on-going,
non-anthropological theoretical, experimental, and clinical research on the adaptation of
bone to mechanical loading. Based on a review of this literature, it is shown that:

1) Although treated as axiomatic, Wolfl’s Law is fundamentally flawed, and not
universally accepted, 2) The argument fails to make use of all available relevant evidence
regarding the relationship of physical activity to bone structure, and 3) The conclusions of
bioarchaeological research are confounded by variables (e.g., genetics, diet) that provide
alternative explanations for the data. Furthermore, behavioral inferences are frequently
based on statistically non-significant group differences in the cross-sectional geometric
properties analyzed. Therefore, behavior-based conclusions resulting from the study of
cross-sectional geometry are not borne out by scientific evidence, and must be rejected as
conjecture.

This dissertation concludes by suggesting that bioarchaeological studies of past
human behavior are based on simplistic and generally unsupported assumptions about
the relative contribution of physical activity, age, genetics, and nutrition to cross-sectional
geometric variation. Because of its complex etiology, long bone cross-sectional geometry
cannot be used to make inferences, or test hypotheses regarding past human behavior.
Furthermore, while specific “causes” cannot be identified, long bone cross-sectional
geometric variation among archaeological populations can potentially be explained

without resorting to behavior-based interpretations.

 

Ila dissertatit III is lit

This dissertation is dedicated to my best friends me husbandJohn and my sister Sigrid

 

l'hrmniplrtit
unwitting supp lfl .t'
ititXttmi lot lit‘Vt‘f L'
littzlfllulmi mini:
1nd Bill lmis. and IL:
and patirncr is tlccph

ltilso gratefull-

 

 

 

lfihrllrtl at the l.t~‘t H
In dlltllllllll. l\

anti: t'tllit‘tlg'tlt‘s \\ llt s li.

twirl adult. LL\'>l\'l;tlll

IL'

it. in the rear: lll lllx

I“ .. .
Jill h.Jlt(ll‘J.ltllt'\. V

Ill”. Rt“t?llllt“ll. lelt’l ]

\

tl‘
lalso \sish to

RI

tn. '

wing} Departmc
.\lost imp! trltt

'rt'i'i r v i

u nlttlonal low it

DMI ‘ ‘
“Whit-sit" l-

ACKNOWLEDGMENTS

The completion of this dissertation would not have been possible without the
unwavering support and pragmatic advice (I F D1) of my advisor Dr. Norm Sauer. Thank
you Norm for never giving up on me, and for talking me out of quitting (many times). I
thank all of my committee members, both current, Drs. Todd Fenton, Lynne Goldstein,
and Bill Lovis, and past, Drs. Ann Millard, Larry Ross, and Don Straney. Their time
and patience is deeply appreciated.

I also gratefully acknowledge the assistance of Dr. Roger Haut, who was
beseeched at the last minute to read my dissertation.

In addition, I would like to extend my sincere gratitude to a number of friends
and colleagues who have kept my spirits up over the years by offering a listening ear,
good advice, assistance when needed, words of support, and the occasional mandatory
kick in the rear: Holly Andrews andJohn Fitzsimmons, Matt and Rene Greff, Nick
Groch,JudyJames, Wendy Lackey, Alpin Malkan, Dan O’Malley, Anthony Paganini,
Jim Rechtien, andJon Winters.

I also wish to thank Dr.Jim Potchen and the rest of the folks in the MSU
Radiology Department for their wholehearted support of my PhD quest.

Most importantly, I could not have achieved any of my goals without the
unconditional love and support of my family. Thank you so much Mom and Frank, and
Dad and Celeste! I am also extremely grateful for the loving encouragement of my in-
laws Yvonne andJohn Bice, as well as Trish, Sam, Belinda, and Desiree. A special thank
you goes to my sister Sigrid Chimner Resh and her wonderful husband Rod Chimner for

being loving cheerleaders, and for sending me “multiple paths to enlightenment.” Thank

" I
ndmemami

 

mhmhwsnnrl,
jjiir ml t’molit iml C

stthdmniit

., .
husband

 

you also to Sig and Rod for creating my nephew Galen (although he is only 2.5 years old,
and really wasn’t much help). Many, many thanks to my husbandJohn, who not only
rode my emotional rollercoaster for the lO-plus years I was in the PhD program, but who

also helped with innumerable computer issues. John’s beautiful, skeptical mind inspired

this dissertation.

vi

Ilil ()l’ I'll Ll'Rl-‘S

I’llll’llZR l ‘
bllllllll'tllll l.\

Sldlt'lllt‘lll of tl

 

Ol't‘n'lt‘“ of ti
Chapter sumn

Clill’ll‘lll‘l
llllliliRlll'Xll l.\
BlllllfllllIXllLS
ltitroilui iii in
Slit’lt'lttl llli IlIl
limit
Iliiti v.
.\lt'l It
(It lllt

Ittliiiii- ,il . ,\

'li'm

Cilll’IIR 3
mm\Tt'Ri: Rl
Rttoxsi‘itt'ci
lnll’Ix'lut‘lioi
umdlure r.
\Iili
Bio.
.\ ttitiquei
‘ ”mmary

Cilil’l‘f‘R 4
llElHQD (ll’ .\‘
Al’l’lli'atii .,
llltlui' {M a
{\ndli’tlt til
(it hnl‘itlc

PM

TABLE OF CONTENTS

LIST OF FIGURES ix

CHAPTER 1

INTRODUCTION 1
Statement of the problem 2

Overview of the method of analysis 5
Chapter summaries 6
CHAPTER 2
BACKGROUND INFORMATION: SKELETAL BIOLOGY AND
BIOMECHANICS 8
Introduction 8
Skeletal biology and mechanics 8
Basic bone biology 8
Ontogenetic processes in skeletal growth and development 15
Mechanical behavior of bones 19
Concepts in skeletal mechanobiology 27
Technical overview of cross-sectional geometry 37
Terminology and definitions 38
CHAPTER 3
LITERATURE REVIEW: BIOMECHANICAL ANALYSES AND BEHAVIORAL
RECONSTRUCTION IN BIOARCHAEOLOGY 45
Introduction 45
Literature review 45
Methodological and theoretical literature 45
Bioarehaeological studies of cross-sectional geometry 55
A critique of bioarchaeological studies of behavior 8‘2
Summary 85
CHAPTER 4
METHOD OF ANALYSIS 87
Application of critical reasoning 87
Inductive arguments 87
Analytical procedure 91
Techniques for criticizing inductive arguments 93

Begging the question 95

vii

 

Illi-ll’IllR 3
RICIQIXS‘IRIIJIIII
llllfltllllt‘lll in
Rri‘iiiistrui‘tii
Critical .in.il\ .
.-\n.il\ ..
;\n.il\ ~
The pi

Summan

Cliil’ll'lll ti

ll'llllFS l-\\\'. l‘l'l
lflllllllill'b'

Introduction

lIlt' "lam" th.

The or

\\1 .lil‘.

functional ad.

I)“ ll llt'

I‘llllt ti

.\£cti

Nitn-tnrrlnrni

 

(it‘ltcl
Nutrit
Summary

CHAPTER 7

C(lXClI'SlIlX
Rmic“ of In
Determinin :
Citiii‘liisii in

SUSQP‘HIHI\

BIBLIOG R“)! IV

CHAPTER 5
RECONSTRUCTION AND EVALUATION OF THE ARGUMENT
Introduction
Reconstruction of the argument
Critical analysis of the argument
Analysis of the 'theoretical argument'
Analysis of the 'empirical argument'
The problems with premises 5a through 5e
Summary

CHAPTER 6
WOLFF'S LAW, FUNCTIONAL ADAPTATION, AND CON F OUNDING
VARIABLES
Introduction
The "Law" that is not a law
The origin of Wolffs Law
Wolff‘s Errors
Functional adaptation
Problems linking physical activity to cross-sectional geometry
Functional adaptation: an "umbrella term" for a plurality of effects
Age effects on functional adaptation
Non-mechanical confounding variables
Genetic effects on cross-sectional geometry
Nutrition effects on cross-sectional geometry
Summary

CHAPTER 7
CONCLUSION
Review of functional adaptation: a summary
Determining physical activity from cross-sectional geometry: an analogy
Conclusion
Suggestions for future bioarchaeological research

BIBLIOGRAPHY

viii

 

96
96
96
97
100
l l I
1 I6
120

122
122
122
126
129
133
134
141
144
154
156
171
177

179
180
184
186
189

193

llfll‘t‘ f-l. lillitts ol
fl‘ZJ 3-3. External i
hate 2-3. .\\i'.il ll i.i.
figtttt ‘l-l. litiils Ills.
PERI-:1. Structura:

fair? 3-0. The “inn

 

hart“ 9-7. Sm iiiil in
limit“ 3-.l Rm il‘nlrt
llZIlft‘ 3-2 Rei‘olNrt
FIRST" 5-3. Rt‘i‘i inst t‘

I

sort: .‘i-l. .‘\IQlllllt‘l

LIST OF FIGURES

Figure 2-1. Effects of modeling drifts on bone size and shape 18
Figure 2-2. External forces and modes of loading 21
Figure 2-3. Axial loading 22
Figure 2-4. Loads that produce bending of long bones 23
Figure 2-5. Structural and material properties 25
Figure 2-6. The “mechanostat” model 32
Figure 27. Second moments of area 42
Figure 5-]. Reconstruction of the overall argument 98
Figure 5-2. Reconstruction of the ‘theoretical argument’ 100
Figure 5-3. Reconstruction of the ‘empirical argument’ 101
Figure 5-4. Argument for Ix/Iy as an indicator of sex differences in mobility 1 l8

.. , H
lit-mi! often happt ll
f Hut this can be ~u.

Skeletal \ari.it
tr- llt‘lt'mllllt‘ St\ and
hittilldrl} . it lllt larchai
phtsital dctix’it} is otu
path-:1». izit‘al skt'lctal \

iirlrrrnces in Ix me in

 

 

Littler-m plnsir .11 at Il‘

ltiferring lx'lia'
31min skeletal rctn.ti
li-‘T‘Itntilt' 1089; lglrsi
'i-illi'lfi’SNl str‘tlt‘tttre i
hiilirrt‘t means I if [1TH

.rdrr spm‘ified lt iai lit

r'"l¥'f‘t'iles i\

h.\—[)(1[Ilt‘\

ii""l'
all. ‘ product (i l \
I».
manual mm c;
kc’
la]

1"
tit-'i .
t’itT- I» . .
.If,.l>l\-In]( a] mini l

CHAPTER l—INTRODUCTION
Thus it often happens that every step in the scientific method is carried out correctly except the

last, but this can be sufficient to nullify all the previous efforts.
James K. Feibleman (1972)

Skeletal variation provides the empirical foundation for a skeletal biologist’s ability
to determine sex and ancestry or estimate age and stature from human skeletal remains.
Similarly, a bioarchaeologist’s “reconstruction” of past human lifeways, of which habitual
physical activity is one component, is also predicated on the existence of normal or
pathological skeletal variation. The underlying assumption is that certain observable
differences in bone morphology among individuals are the direct result of performing
different physical activities.

Inferring behavior (e.g., habitual physical activity patterns, occupation, etc.) from
human skeletal remains has become a standard component of bioarchaeological research
(Kennedy 1989; Larsen 2000). Cross-sectional geometric analysis of long bone
diaphyseal structure is a common approach. This biomechanical method provides an
indirect means of predicting the mechanical behavior (e.g., strength, rigidity) of bones
under specified loading conditions. Group variation in cross-sectional geometric
properties is hypothesized to result primarily from differences in localized mechanical
loadings produced by differing patterns of physical activity. The application of
biomechanical concepts to archaeological populations is largely due to the pioneering
efforts of Ruff and colleagues (Ruff 1987; Ruff and Hayes l983a, 19831); Ruff and Larsen
1990; Ruff et al. 1984, 1993, 1994; Trinkaus et al. 1994). Their research has generated a
biomechanical model that is the basis for “reconstructing” the behavior of past

populations. The model makes three predictions: 1) Group variation in cross-sectional

mimetn rellects gri e
l’ll'lilllllll in degree is:
liliirrnres in sexual .
totttlmetn' of m ISS-x
unilateral and l)il.itei
tinrrhm )li Mn} “1] n . \
ciiiuatiiin ofthe dl’l‘

lrha itiral reeoiistrizi

 

 

 

Statement of the p

Bioart'lldct ili i!
tiriatiim in long lx in
lfllltlllil’lllt"l1£lllflt‘\' ;:
h the l'nitetl States.

:‘xh ' '
.dLJ’llllT-q 10 nldl/I‘ Ll

hiailht‘asl Brill

Hes l

.(l
Lift. .
.. 1 . \orthi‘m‘l B

Putin.- Rufflflfq; \.

9m -
l

nllfllii

4“ F o
.flrf (“Its

t: it. ,
m“ hitting] (.1

w?- '
‘ «illf *
\‘t‘t

geometry reflects group differences in levels and types of physical activity. 2) Group
variation in degree of sexual dimorphism of cross-sectional geometry is attributable to
differences in sexual division of labor. 3) Group variation in degree of bilateral
asymmetry of cross-sectional geometry reflects behavioral differences related to specific
unilateral and bilateral activities. These predictions have become unquestioned tenets of
bioarchaeological research. This dissertation presents the results of a qualitative critical

evaluation of the application of long bone diaphyseal cross-sectional geometry to

behavioral reconstruction.

Statement of the problem

Bioarehaeological studies are often undertaken to detect temporal or geographic
variation in long bone cross-sectional geometric properties, inferred to result from
behavioral changes associated with adaptive shifts in subsistence and mobility strategies.
In the United States, many studies have focused on the transition from hunting and
gathering to maize agriculture in the Midwest (Barondess 1998; Bridges et al. 2000),
Southeast (Bridges 1989, 1991; Larsen et al. 1996; Ruff and Larsen 1990; Ruff et al.
1984), Northeast (Barondess 1998), Southwest (Brock and Ruff 1988), and the Great
Plains (Ruff 1994; Wescott 2000). Studies have also investigated differences in bone
structure between pre- and post-European contact Native American populations on the
Georgia and northern Florida coast (Ruff and Larsen 1990), and in Michigan (Barondess
1998)

A core construct of this research is the premise that bone has the ability to adapt
to its mechanical environment. This construct, known as VVolfI’s Law, proposes a direct

relationship between bone architecture and mechanical forces; bone is added where it is

ntttlt‘tl to maintain -
pmllered as the print
diaphyseal structure
although virtualh tt‘
lIitilIS IA“ is not \\ '-.
Currey l‘lt‘il7. 2i ll i3; 1

The notion ll
lmm tlflt‘tl skeletal I't
"ll-"W haw qiiestii mi

1”“‘1‘ 'I‘ 6‘t al. fit II N-

 

 

iinibt that lx inc cut
triumstances \I In.
. . l

fil‘il"[i~
“ “man I
. )t the
111

tsr -
W“) m atlult' -
-‘. 1:

cm. ~
‘* V204“) .
I if' \\iIIII

grill “hlt‘h {he

Sllltl'

Emilie '- -
.l‘lsk Ul Idllm

I‘hlP (II\\‘¢
“ -r!xII

firmer-
ll .
Ihf Ihp, ’rI‘I.
1.

needed to maintain structural integrity, and removed from where it is not needed. It is
proffered as the principal theoretical foundation for inferring behavior from long bone
diaphyseal structure (e.g., Barondess 1998; Bridges 1989; Ruff and Hayes l983a).
Although virtually axiomatic, and often cited by skeletal biologists and biomechanists,
WVolfl’s Law is not without its detractors (Bertram and Swartz 1991; Cowin 2001b;
Currey 1997, 2002; Dibbets 1992; Roesler 1982, 1987).

The notion that biological anthropologists are capable of reconstructing behavior
from dried skeletal remains has been recently discredited (lurmain 1999). In addition,
others have questioned the fruitfulness of the biomechanical approach specifically
(Lovejoy et al. 2002; Ohman and Lovejoy 2001; Wescott 2001). While there can be little
doubt that bone can and does respond to its mechanical environment under certain
circumstances (Martin et al. 1998), the relationship of habitual physical activity to non-
evolutionary bone modification, particularly with regard to cross-sectional geometry, and
especially in adults, is not fully understood (Bertram and Swartz 1991; Cowin 2001a;
Currey 2002). Without a thorough reexamination of the most fundamental principles
upon which the study of human behavioral adaptation is based, biological anthropologists
run the risk of failing to adequately consider alternative explanations for their data.

This dissertation project was undertaken because of an apparent discordance
between the theoretical underpinnings of the bioarchaeological research and the
hypotheses emerging from non-anthropological research on the response of bone to its
mechanical environment. An important goal of this dissertation is to bring into the forum
of anthropology the ongoing debate among skeletal biologists in non-anthropological
fields regarding Wolff's Law and the relationship of bone architecture to mechanical

loading. This dissertation demonstrates that VVolff’s Law has been the subject of much

trititism by skeletal l
hmrrtiral lramew .:
llai'rs 19.0.33 . \\ hi is
’ I
held. suggest a liin
llates' interpretatii i:
explanation for l)t )llt
than (lies in ll pret
lt’JZtht‘S pri ilial ily n.
lliiiirsm" P 1'," .

Subsequent t

it: the early-1‘18! ls. tl

 

 

hint adaptation in r:

.h - .
an» appeared in nu

International. I turn

lime and Mineral R
Pt

Vduhi ' I:
, 2}. ()sltfiijiii

Lit-“in 19:8 1 a ‘)i It I I

IV:
If”.

While the Iii.

Ffiflals
.Uhsnp '
-, l‘litl1\‘

Mn;
arithrr,p.,|(,,,i(._i.
Self-Haj ~
‘ '- lml
")g\
, rvse‘
l

if,m,_ ” .
(hdnltal int j
~N i
élt-‘l-Qii

”i I“ f‘drlt ]

criticism by skeletal biologists, and suggests that it does not provide an adequate
theoretical framework for inferring behavior from bone structure. Interestingly, Ruffand
Hayes (l983a), whose pioneering studies led to the development of the biomechanical
model, suggest a limitation of Wolff’s Law in a footnote, “Note that while [Ruff and
Hayes’ interpretation of] Wolff’s Law may appear to invoke a purely environmental
explanation for bone form (i.e., response to in-vivo mechanical stresses and strains), the
theory does not preclude heritable genetic effects. In fact, most bone morphological
features probably result from a combination of both genetic and environmental
factors...” (p. 371).

Subsequent to the first bioarchaeological application of cross-sectional geometry
in the early-19805, there has been a tremendous amount of research on the biology of
bone adaptation in response to mechanical stimuli (mechanobiology). Published studies
have appeared in numerous preeminent journals including Bone, Calcified Tissue
International,Joumal of Biomechanics,Journal of Bone andJoint Surgery,Journal of
Bone and Mineral Research,Journal of Experimental Biology,Journal of Applied
Physiology, Osteoporosis International, and in many books (e.g., Currey 1984, 2002;
Cowin 1981a, 2001a; Martin and Burr 1989; Martin et al. 1998; Odgaard and VVeinans
1995)

While the bioarchaeological study of cross-sectional geometry has grown into a
virtual subspecialty, anthropologists have for the most part remained uninformed by the
non-anthropological literature. This is inferred from the relative lack of citations for
skeletal biology research conducted since the mid-19805, and the fact that the basic
biomechanical model has not changed. A handful of experiments conducted from the

late-19705 to early-19808 (e.g., Goodship et al. 1979; Lanyon and Rubin 1984; Woo et al.

 

lilil arc n'pcatctlh

 
  
  

Biiuri‘haet llt igists ha
pnrrsses that pri 'itll .
the nun-anthri ipt lll
tilt? .111th ‘l” ill )Qlt‘dl

Utrsntrd l’,‘ Rull. l

 

Overview of the n

 

The anah sis
trimmed from in I} i
presented for inli‘rr
liliKlt‘lt’d {is a Scrics
spits-fin of the it till
PitmlSes. and 2 \\

mluatrtl in this (.1

a:imlf‘m. the ITII!

tr? i~ .
l) ”133. BI’I’dllx.

$0941: .. .
.iipril‘l "5 I‘ldll

If , .
.o. prfml‘(k\ (

Biyi‘l (in

fact a, ‘
\\("‘Ii .\ I

it‘ll,
‘ M“ from th
I‘ l

I)
‘ t

fifirrfi
“S lite r,

1981) are repeatedly cited as providing empirical support for “’0le Law.
Bioarehaeologists have focused on skeletal mechanics while all but ignoring the biological
processes that produce, maintain, and adjust the mechanical properties of bone. Much of
the non-anthropological research on skeletal biology suggests alternative explanations for
the anthropological findings, and casts doubt on the premises, inferences, and conclusions

presented by Ruff, Larsen, Bridges, and others.

Overview of the method of ana_.lysis

The analysis presented in this dissertation was accomplished utilizing principles
borrowed from informal logic to reconstruct and evaluate the strength of the argument
presented for inferring behavior from cross-sectional geometry. In logic, an argument is
modeled as a series of propositions; one is a conclusion, the rest are premises offered in
support of the conclusion. The success of an argument is dependent on 1) the truth of the
premises, and 2) whether the conclusion follows from the premises. The argument
evaluated in this dissertation is inductive, rather than deductive. Unlike a valid deductive
argument, the truth of an inductive conclusion is not guaranteed by the truth of the
premises. Because of this, an inductive argument must use all available relevant evidence
to support its claims. If it does not, it can be shown that the introduction of new evidence
(i.e., premises) can change the conclusion.

Based on a review of non-anthropological literature on skeletal biology, it is shown
that: l) Wolff’s Law is not strongly supported, 2) The conclusions do not necessarily
follow from the premises because confounding variables provide alternative explanations
for the data, and 3) The argument fails to make use of all available relevant evidence

regarding the relationship of physical activity to bone structure. In addition, many

   
 

I I i' ‘ L.“
omttusmns an l) t
khaxior-hasetl U m.

Chapter ‘2 p1
presented in this div
finds ufskrletal bit tl

Chapter 3 p
Efitmt‘lrlt‘ \‘dridllt it
tuntexts. 'l‘hrnugl‘
‘bin.‘»art"haet ilt )qit gt‘

Chapter 4
ii\t;il\'t‘s apply lllfl
armments. This

nutlines the pr! K‘t

inductive argumr

Chapter I

i; h
«It dt‘t )lt mid

and

fi‘uluates tl‘n

Chaptt‘rt

«1‘55. This is M

~ ‘.
[tin]

f , g “ .
“l" ”i (‘Yltlt‘n

9mm"?

lllf l

Conclusions are based on statistically non—significant group differences. Therefore,
behaviO r— based conclusions resulting from bioarchaeological studies ofcross-scctional

geometry are not borne out by scientific evidence, and must be rejected as conjecture.

ghnter summaries

Chapter 2 provides the requisite background information for the critical analysis
presented in this dissertation, and introduces relevant concepts and terminology from the
fields of skeletal biology and biomechanics.

Chapter 3 presents a review of the anthropological literature on cross-sectional
geometric variation, and its application to behavioral reconstruction in archaeological
Contexts. Throughout the dissertation this body of literature is referred to as
“bioarchaeological.”

Chapter 4 describes the method of critical analysis used in this dissertation, which

i hvolves applying concepts borrowed from informal logic to critically evaluate inductive
arguments. This chapter provides an operational definition of an inductive argument,
() utlines the process of reconstructing an argument, and discusses the ways in which an
i hdUCtiVC argument can be critiqued.

Chapter 5 reconstructs and critically analyzes the argument contained in the
t)i0211'Chaeological literature reviewed in chapter 3. This chapter identifies the premises,
E34nd evaluates the strength of the argument.

Chapter 6 provides a brief history of W olfl’s Law and presents a summary of its
flaws. This is followed by a discussion of the concept of functional adaptation, as well as a
rGView of evidence for the effects of genetics and nutrition on long bone diaphyseal

S trUCture. The information presented in this chapter was obtained from non-

antitmptiltigit'al tht .
funt‘tiunal .ttlaptati

Chapter 7 p
mnt‘ludts ht sue!“
lfllli gt'nt‘mll} unstt]
age. genetics. and l’;
gettmtiallr t‘umplt-x
irxlirrnt‘t's. or [est 1:.
tinnut he lflt‘llllllt‘t;
p-Ttpulatinns ran Pt m

r~'lpltttatit ins,

 

anthropological theoretical, experimental, and clinical literature on skeletal biology and
functional adaptation.

Chapter 7 presents the conclusion of the critical analysis. This dissertation
concludes by suggesting that the bioarchaeological studies are based on poorly defined
and generally unsupported assumptions about the relative importance of physical activity,
age, genetics, and nutrition to cross-sectional geometric variation. Because of its
potentially complex etiology, long bone cross-sectional geometry cannot be used to make
inferences, or test hypotheses regarding behavior. Furthermore, while specific “causes”
cannot b6 identified, cross-sectional geometric variation among archaeological

populations can potentially be explained without resorting to behavior-based

explanations.

l
mm 2"”

Tm- Itiming anti knt .

tit dft‘ ignurmt.

WWI—$592

This chapter

 

in Ihh (liN‘rttttit in. .
I‘tt't‘nit’tt' (if kt‘t tt 4 ii
the relationship l)(‘i‘
TDt‘t‘hltnulntiltiQ ,

diaphws. and intrt

Skeletal biology

TlllS St‘t'ilt i'

mnrr-pts relating t

at,
.s. 1. [dl grim th an

'nw' ‘ ‘
um. hen: tint )lt iQV.

w skeletal tr
L?

._i‘t l

.- i") I ..~ , ~
5‘} Uinttunx i

Blsi
C .

Th

“ ("fltlt

l
53‘; 7*

”an“? tissm.

CHAPTER 2—BACKGROUND INFORMATION: SKELETAL BIOLOGY AND
BIOMECHANICS

The learn ing and knowledge that we have, is, at the most, but little compared with that ofwhich

we are ign orant. Pl
ato

W

This chapter presents background information for the critical analysis presented
in this dissertation, and is divided into two main sections. The first section provides an
overview of key topics in skeletal biology and establishes a foundation for understanding
the relationship between skeletal structure and the biology of mechanical adaptation
(mechano biology). The second section focuses on cross-sectional geometry of long bone

diaphyses, and introduces and defines the terminology used throughout the dissertation.

Skeletal biology and mechanics

This section of the chapter presents important terminology and qualitative
COHC€PtS relating to four subjects: l) basic bone biology, 2) ontogenetic processes in
Skfilfital growth and development 3) the mechanical behavior of bones, and 4) skeletal
l"INT-him()l)iology. Jee (2001) and Martin et a]. (1998) provide excellent up-to-date reviews
Of basic Skeletal biology. Unless otherwise indicated, the following primer of skeletal

EDiology Contains information drawn from these sources.

Basic hone biology

The enduring quality of dried skeletal remains belies the fact that living bone is a

dYnamic tissue, which undergoes dramatic transformations throughout the life of an

individual. An 11 n.

be relationship lu-

[mam Obe'l
“it" skr‘lt‘td
function «ifsnme lit
Hi’Mt‘Vt‘r. a5 (‘1. umpt
hers. providing at
mint e rm ix'emen t
rid strung i.e.. Llill‘
additional \ital futtt
pru‘xiuees blrx xi (‘t'lls
ralrium. \‘t'l‘tit‘h an" (
mmplementart' fir],
iunt‘tion Oi‘lxtnf‘ is t“
biliifiz‘ts bone

ph}‘s

Currey Etitt 12' \l'ir‘tit

 

 

individu a]. An understanding of basic bone biology is a prerequisite to any discussion of

the relationship between cross-sectional geometry and physical activity.

Functions of bone

The skeletal system is important biomechanically and metabolically. A critical
function of some bones is to enclose and protect the visceral organs of the body.
However, as components of the musculoskeletal system, most bones serve as struts and
levers, providing attachments for the skeletal muscles that maintain body posture and
produce movement. For this, bones need to be both rigid (i.e., able to resist deformation)
and strong (i.e., able to withstand loads without breaking). Physiologically, bones have
additional vital functions. The marrow that resides within most bones of the body
produces blood cells, and the bone tissue itself serves as a reservoir for minerals, such as
Calcium, which are critical in muscle contraction and nerve conduction. In the
Complementary fields of biomechanics and skeletal mechanobiology the mechanical
Function of bone is emphasized. Indeed, in the view of most contemporary skeletal
biologists bone physiology is primarily regulated by mechanical usage (Cowin 2001a;

Currey 2002; Martin et a]. 1998; Rubinacci et a1. 2002).

(30mp08ition of bone
Bone is a type of connective tissue in which the extracellular matter consists of
i norganic mineral deposited into an organic protein and water matrix. Thus, at a
tjlolfit‘lllatr level, bone is a composite material. The mineral component of bone
distinguishes it from the other structural connective tissues (e.g., tendons, ligaments, and

Cartilage), and characterizes it as a “hard” tissue. Sixty-five to 70% of bone is inorganic,

LO

ft‘tliSlSIing 0f hid“ "
SmspthlMt‘ liir 11‘
rrmaining 3(1-34')“(~. .
Eartnet‘ells. The ma
celled npe l mllau'.
mnsists of numen it.
trite-rinm‘tin. the fur:
tertiirm imp« urtant

rumple. ()Sit‘t )pr mt
\nmura et al. L’t H H
b'é’nt‘lumm‘er. 'I'hz

tunrttnning. partit'u

the

m1, matrix.
I-("urty'1)(_‘; "

~17]. ' . - .
d lxlnf‘-lln]ng (“.12

 

consisting of hydroxyapatite crystals (Ca|o(PO4)5(OH)2). The mineral component of bone

is responsible for its rigidity and material stiffness, and its strength in compression. The
remaining 30-35% of bone is made up of proteins and water (the organic matrix), and
bone Cells. The majority of protein in bone, roughly 90%, is the fibrous structural protein
called type I collagen, which provides bone with tensile strength. The remaining 10%
consists of numerous non-collagenous proteins, including osteocalcin, osteopontin, and
osteonectin, the function of which is the subject of speculation. These proteins may
perform important roles in the regulation of skeletal adaptation to mechanical usage. For
example, osteopontin is thought to mediate osteoclast attachment to bone surfaces
(Nomura et al. 2000). The presence of some of these proteins in urine is used to assess
bone turn over. The water component of bone is also essential to its mechanical

Functioning, particularly with regard to the flow of extracellular (interstitial) fluid through

the bony matrix.

Cell Wes

F()ur types of bone cells have been identified: osteoclasts, osteoblasts, osteocytes,
and bOTIS-lining cells. Osteoclasts are large, multinucleated cells derived from
hCmOPOietic bone marrow and formed by fused cells of the monocyte/phagocytic lineage.
T‘Ihe Primary function of osteoclasts is to resorb bone. An active osteoclast has a ruffled
border adjacent to the bone surface, which secretes acids and enzymes to dissolve the

mineral and collagen components of bone, respectively. The resorption spaces created by
(DStCOClasts, called Howship’s lacunae, represent transient porosity in bone tissue. The
activity of osteoclasts is regulated by many factors including numerous hormones. The

Signal for osteoclasts to begin resorption at a specific site likely involves the activity of

10

 

 

   
   
  

tunrolining tells .\.
()stenhlasts

ggjr brine SUTlilt 1‘."
mmrlnntal tells
0i uStt‘t tl‘)l‘.‘t.<i$ \S \N it t
partiripate in the in
lft’mS knms'n It) at“
mt‘tlhttlites. stert in
llfltt‘tx‘guu

that herame trap

"EVUPF are caller
fr‘llular prt it (\‘st‘
ifllf‘t‘t‘t'mnp‘f‘ “H
i'i'r‘lgif‘tn'hx I,

112mg cells. ant

)lll‘
- ‘11,. ‘Xllllt ‘u

rmmlfili’int‘e (
fl??? ‘
. flit“ 1
’n ”fa;
intildtt“ is me

c

3,.."1‘2

tuna,“ P _
‘ .2.‘
thine

fllnlnt‘r (.

lh‘ij‘ii'
lip‘
“Jilin

s Fem 21*

bone—lit! i ng cells (Majeska 2001; Martin 2000).

Osteoblasts are cuboidal cells derived from multipotent mesenchymal cells located
near bone surfaces and in marrow stromal tissue. Differentiation of osteoblasts from
mesenchymal cells is thought to be dependent on mechanical stimulation. The function
ofosteoblaStS is bone formation; they secrete osteoid (the unmineralized bone matrix) and

participate in the mineralization process. Active osteoblasts possess receptors for many
agents known to affect bone metabolism, e.g., parathyroid hormone, vitamin D
metabolites, steroid hormones, and growth factors.

Osteocytes, the most abundant cell type in mature bone, are inactive osteoblasts
that became trapped within the bony matrix during bone formation. The spaces they
occupy are called lacunae. Osteocytes are stellate-shaped cells with numerous slender
cellular processes housed within tiny canals called canaliculi, which extend from and
i nterconrlect the lacunae. Gapjunctions between the cellular processes provide a means
F01” OStCOCytes to communicate and exchange substances with other osteocytes, bone-
lining cells, and osteoblasts in an extensive “connected cell network” (Cowin and Moss
2001). Although not thoroughly understood, the functions of osteocytes likely include: 1)
Maintenance of the local ionic environment, 2) detection of microdamage, and 3)
dCtCCtiOU of alterations in their local strain environment and transmission of the signal to

i nitiate bone adaptation processes.

Bone-lining cells are flattened, elongated cells, which cover all quiescent bone
surfaces, e.g., periosteal, endosteal, trabeeular, and vascular canals. Like osteocytes,
1”DOM-lining cells are inactive osteoblasts, which maintain cellular connections with

osteocytes and osteoblasts via the lacunocanalicular network. The functions of bone-

1 ining cells are still poorly understood (Ice 2001), but are thought to include bone

11

limitation and mint

 

impirmnt mle in n.

 
  
  
  

Classification 0!
At the mat !‘
t‘mt‘ellt tttS. (114111!“
tx'iiume of l)t ine tisst'
with bone mam m .
“funerary rest )rpti
Compare-d
ssiid~ its pormitt
temp-act hone. m

shafts of long in n'

Li'ttfr‘lltius in ")IR‘

\- *:1-~ ‘
L. .434) lint)“ n 1b
(41,.“
L1 tl ilfllk‘t‘ul‘u

'vs'it' '
Ill“ ill? \‘prud

glf' : '
mdn]. hip in]

.4-
are 91»
-» h

Upmenl .1

BI )llf‘ {'1‘

L
ill" ‘
"“9 l1<=
t ill
in». .
, I} .. é .
CUT ()rli‘l‘]

formation and mineral homeostasis. In addition, bone-lining cells likely play an

important role in initiating mechanical adaptation.

Glasgflcah‘on of bone tissue

At the macroscopic level, bone tissue is categorized as either compact or
canceIIOUS, distinguished primarily by its porosity. Porosity refers to the fraction of a
volume of bone tissue consisting of voids in the bony matrix, e.g., trabecular spaces filled

with bone marrow, vascular canals (Haversian and Volkmann’s canals), and the

temporary resorption spaces formed by osteoclasts.

Compared to cancellous bone, healthy compact bone is dense and appears
solid~its porosity is very low (33-10%). Approximately 80% of the mass of the skeleton is
Compact bone, the majority of which is contained in the cortices (or outer shells) of the
Shafts of long bones; hence, compact bone is often referred to as cortical bone.
Cancellous bone is often referred to as spongy bone because it is very porous (75-95%). It
i S also known as trabecular bone because it consists of a lattice of bony struts and plates
Callttd trabeculae. The majority of cancellous bone is located in the ends of long bones,
Within the vertebral bodies, and sandwiched between the cortices of flat bones (skull,

S ternum, hip bones). Compact and cancellous bone differ with regard to function,

development, architecture, material properties, rate of turnover, and age-related changes.

Bone tissue is also categorized based on its microstructure. Woven and lamellar
IZION? tiSSue differ with regard to collagen fiber organization. In woven bone, the collagen
fibers are oriented somewhat randomly making it mechanically inferior to lamellar bone.
Th6 Primary biological advantage of woven bone is that it can be formed rapidly. In

h umans, woven bone forms early in development, and later as part of the normal

12

   
  
  
   

pfipxtfit‘ in ltartmx
hunt- is subsequent
ftu'illrl-lilwml or
mull animals such

lamellar l).

strurturr is nltvn «ll

.mentation in 'd‘ll‘“
lL'nellar lxmt‘ is U
dam mun h nu un‘
pttduminam t_\ p
Plexili »m

men and lanu
damn and the a
were rapidly tl
ml deer: it is I
Conic;
Primary hm“.
flfi‘\1n1lercn [i
10 Lhe lmm- S,
{it"lf-iilm‘tl \x"
1

“‘61“ .
d1 ( (1)1](1

W". ‘
i‘f‘f’rallS’ rm
In ht

SF' 1
' ”MAW l)

response? in fracture repair and adaptation to extreme mechanical overloading. W'ovcn
bone is subsequently replaced by lamellar bone. A special type of woven bone, called
paralle ltfibered (or fine-fibered) bone, has more organized collagen fibers, and is found in
small animals such as mice and rats.

Lamellar bone, as the name implies, is organized in layers, i.e., lamellae. Its
structu re. is often described as being analogous to plywood, with alternating collagen fiber
orientati on in adjacent layers; however, the microstructural organization and variation of
lamellar bone is complex, and still largely unresolved (Currey 2002). Lamellar bone is laid

down much more slowly, and is ultimately less mineralized than woven bone. This is the
predominant type of bone found in the human adult skeleton.

Plexiform bone, also called fibrolamellar or laminar bone, is a combination of

Woven and lamellar bone. It is formed when an initial scaffolding of woven bone is laid
down and the gaps are filled in with lamellar bone. Plexiform bone can be formed much
more rapidly than lamellar bone, and is common in large, fast-growing animals like cattle
811d (1661‘; it is not characteristic of adult human bone.

Cortical bone tissue may be further classified as either primary or secondary bone.
Primary bone is either plexiform or circumferential lamellar bone. The lamellae of
Circumferential lamellar bone are oriented around the circumference of the bone parallel
‘20 the bone surface. Blood vessels within a region of circumferential lamellae are
contained within primary osteons, which consist of a central vascular canal surrounded by

several Concentric lamellae. In adult human long bones, circumferential lamellae are
generally restricted to several layers immediately subjacent to the periosteum.
In humans and many other vertebrates, primary bone is gradually replaced by

secondary bone via remodeling, a process whereby existing bone is removed by

[3

usmlasts. leltl tlu
stmmldr)‘ 051m I” t
mnsist ol‘mnrem l"
canal . l'nlikr pri:
mullan' ustmn t

fi‘krill‘dl lxmr has I.

 

(mammal) "i“ m.
lmt‘. The tralm l
ualrrulac are 1_\ P
mmin rrrsu-nt s
Brnx't‘t‘n \

use “intrnm‘tll
E9731. lilifih. l‘
‘funt'titinally im
intramnit‘al ll.

the net bone ha

mum formrt

fmri‘lupes. ()\

lit. librmatit in

2%. 0r slightl
{lilmdilhn rpld

a5“ '
Ajrl} “NIH-d

1') frijmliililt in l

osteoclas ts, and the resorption space filled in with new lamellar bone by osteoblasts. The
second ary osteon or Haversian system is one outcome of remodeling. Secondary osteons
consist of concentric lamellae surrounding a central neurovascular canal (Haversian
canal). Unlike primary osteons, a distinct boundary called the cement line separates each
secondary osteon from the surrounding bone tissue. In skeletally mature humans, most
cortical bone has been remodeled and is composed of numerous complete and

fragmen tary secondary osteons. Haversian bone is mechanically weaker than primary
bone. The trabecular bone of adults is also secondary bone; however, individual
trabeculae are typically too small to contain whole secondary osteons, and may instead
contain crescent shaped osteons called hemiosteons.

Between the early-19605 and mid-19805 Harold M. Frost developed the concept
of the “intermediary organization” (10) of bone (i.e., tissue level organization) (Frost
197321, 1 973b, 1986). As part of the skeletal IO, Frost (1986: 150) conceptualized four

“function ally independent bone surfaces” which he called bone envelopes: periosteal,
intracorti cal (Haversian), endosteal (endocortical), and trabecular. According to Frost,
the “Ct bone balance (the difference between the amount of bone resorbed and the
amount formed), and the biological processes that affect bone balance, differ among the
QHVClOPeS. Over a lifetime, the periosteal envelope exhibits a net positive bone balance,
i.e., forttuition is greater than resorption. The intracortical envelope shows close to a
zero, 01‘ Slightly negative, balance resulting from increased porosity due to reduced
fOYmafiOn relative to resorption. In contrast, the endosteal and trabecular envelopes show

a fairly substantial net negative bone balance resulting from increased resorption relative

to fonnation on bone surfaces adjacent to bone marrow.

l4

W6C pro

 

Must of flu

mwnthtmal (‘t‘lh

  
  
  

:ephring the (uni).
hints of the appvii
Lht‘t‘mnldl \"giult .ii:
iin‘r-li'ip 1h mi film“
at. initial cartilauii‘
undergo IX 1111 ts pt
lilll'glfmnt‘.Q Eff)“ l
n’phcemem of e:

(mm‘Lh in than“

1

mt“ appositit m

lit illi ml 1‘;
undergo an (‘Xlt
éhleral maturit
maintenance a
rszfigpmia. fr

de-

tf‘ml [figs Il]{

Ontogenetic processes in skeletal growth and development

Most of the skeleton is preformed in cartilage that develops from condensations of
mesenchymal cells during early embryonic life. Endochondral ossification involves
re placing the cartilaginous anlage with osseous tissue. This process produces most of the
bones of the appendicular skeleton, as well as the cranial base. In contrast, most bones of
t he cranial vault and facial skeleton, as well as portions of the mandible and clavicle
d evelop from membrane-like sheets of mesenchymal tissue. Ossiflcation in the absence of
an initial cartilaginous model is termed intramembranous. Most bones ultimately
u ndergo both types of ossification during the course of their development. For example,
long bones grow in length through a process of endochondral ossification involving the
replacement of existing cartilage by osseous tissue at the epiphyseal plate (physis).
G rowth in diameter occurs through an intramembranous ossification process involving
bone apposition by cells located in the osteogenic layer of the periosteum.

Following the initial development and ossification of the skeleton in ulero, bones
undergo an extended period of postnatal growth and shape modification, which ceases at
Skeletal maturity. Thereafter, bones experience less dramatic modifications from

maintenance and repair functions, as well as age-related degenerative processes, such as
osteopenia. Frost (1986: l l l) succinctly describes the processes involved, “Growth
determines the size. Modeling molds the growing shape. Remodeling then maintains
functional competence.”
Growth involves an increase in size resulting from a proliferation of cells and an
increase in extracellular matter. Unlike many tissues in the body, bones cannot grow by
expansion (interstitial growth), only by apposition of osseous tissue onto an existing

surface (cartilage or bone). Simply growing larger would ultimately result in grossly

l5

d}'sfunctiniial 1n ”1‘
such as muscles at
retarding gri‘m'th i
”8.). 2H. .\l4)(lt'
actions OleNP‘ ll ’lJ
“mi-we periosteal .1
example. fumiatiu
and iliapht seal ('11!
t0 matrh a grim in:
mass. muscle sin-ii
malt-ling \irtuallt
li’ftikjee 21m] .
ln (‘01)!qu
indixiduul. ln 191‘,
iUlNi‘lUPm n’1)lau
'til‘i‘l‘la-‘Li and 0in

‘r; f ‘
pamum ‘ of bun

at . .. .
.tiiaunnqesorpti

mmplrte. Rem! '
)( It

ume BSl’

of b« m'

 

Rt‘motlt‘ll!‘

r... .
\1_;n_‘. .. '

Ration mm"
if’ilti ‘ l

(‘ tlt‘t‘l‘rw

k'1
”tinf- r.
(mtXlQ-Iin”
-\-‘l

 

dysfunctional bone shapes and altered anatomical relationships with surrounding organs,

5 uch as muscles and other bones. Modeling guides the growth process “by locally

re tarding growth in some locations and directions, and potentiating it in others” (Frost

1 985: 213). Modeling (sometimes called macromodeling) involves the independent
ac tions of osteoblasts and osteoclasts resulting in formation and resorption “drifts” that
“ move periosteal and cortical-endosteal [envelopes] in tissue space” (Frost 1985: 2). For
6 xample, formation and resorption drifts are responsible for changes in cortical thickness '
and diaphyseal curvature in growing long bones (Figure 2-1). One purpose of modeling is
to match a growing bone to changing functional demands incurred by increasing body
mass, muscle strength, and mechanical usage. The dramatic shape altering potential of
modeling virtually ceases following skeletal maturity (Frost 1985, 1990; Martin et al.

1 998;]ee 2001).

In contrast to modeling, remodeling continues throughout the life of an
individual. In 1964, Frost (1997) redefined the term remodeling to mean the removal and
Subsequent replacement of “quantized” packets of bone through the “coupled” actions of
Osteoclasts and osteoblasts. Frost (1986) named the group of cells that replace a

“quantum” of bone the basic multicellular unit (BMU). BMUs replace bone in an
activation-resorption-formation (ARF) sequence that takes roughly four months to
complete. Remodeling of cortical bone produces secondary osteons, the basic structural
units (BSU) ofbone.

Remodeling typically results in a net loss (disuse mode remodeling) or no change

(conservation mode remodeling) in bone quantity, depending on the bone envelope (Frost
1990). The decrease in bone mass that results from remodeling is often referred to as

“bone remodeling-dependent bone loss” (lee 2001: 1-31). Remodeling has both positive

16

and nt‘fltitt‘ “m"
remnalnfnurrfi“
éiiélllatii‘)n ()ftlit" I
rllrt‘tslnt'ludt": l 1
him" balant‘c on ll
artreathin (d TP>'
pn‘gflurtinn Ufmt‘t
lii lllt‘ lit: in
slt'lrtal bioh mists 1.
sequence. bit mm '1
smiint'mous \\ ith '
lmTfiurininran
iddpldlltm [1) lllt‘t f
ftmi‘xleling. or lx ll
Produce dillrrent t'
gamuf0i~and du-

l

+
.uununatelt it i

 

and negative consequences for the skeletal system. The beneficial effects include: 1) the

re rnoval of microscopic fatigue damage (microdamage) and necrotic bone, and 2)
adaptation of the microarchitecture to local mechanical stresses or strains. Deleterious
emcts include: 1) the thinning and eventual loss of trabeculae resulting from the negative
bone balance on the trabecular envelope, 2) an increase in intracortical porosity through
the creation of resorption spaces, 3) decreased cortical thickness with aging, and 4) the

p roduction of mechanically inferior bone (i.e., Haversian bone).

In the literature, use of the term remodeling often interjects ambiguity. While,
skeletal biologists use the term remodeling to refer specifically to the BMU—ARF
sequence, biomechanists often use the pre-l964 definition of “remodeling,” which is
synonymous with “adaptation” (Martin et al. 1998; Odgaard and Weinans 1995). This
difference in meaning can be problematic because, depending on the circumstances, bone
adaptation to mechanical usage may involve the processes of modeling, BMU
1‘ e modeling, or both. Mechanical loading differentially activates these processes; they
produce different effects in terms of bone macromorphology (including cross-sectional

geometry), and they are dominant at different points during the life of an individual.
Unfortunately, it is not always clear from the context in what sense bioarchaeologists are

using the term remodeling when discussing long bone structural variation.

 

+
+
+-
+
+

 

+++++
I

 

lime 2~ F ~
sh . ‘llf‘(I-\
_ dllo 4. ill ..

iii‘ii‘rllf‘leri ‘ ( Hr;

, . 'hr .
”Surplltln' (“1 ‘

 

 

 

+++++
I
I
+++++

 

 

(b)

 

 

+++++
I

+++++
l

 

 

_ H
\

 

 

(C)

Figure 2-1. Effects of modeling drifts on bone size and shape. (a) Metaphyseal cutback.
(b) Changes in cortical thickness, external (periosteal) diameter, and internal (endosteal)
diameter. (c) Changes in bone curvature. (+) indicates bone formation; (-) indicates bone
resorption.

18

litteri—

l

fllfli’tqfs‘ P r

 

Mechanic“ beh
l

l'nless oil;
discussion of the 1'?"
rSDt‘tldll\. Cm‘lll‘n
furt‘t’ is dc

orthr pound ll) .
(haratterized l’.‘ li
external forces i.e.
thmges its origind

mist the (left innati

 

L .
lrnflll 1.e.. AL/ 1 . .

1.5- .. . .
ulrrdllll‘t‘ strain is r

3!? Ernt‘rally rep: .r

borne stretched to 1

therefore. the strait

NH ~
f ‘ lus . Thei

‘1

l_~"“‘i"'?l Stress

Slrst

v.11 ~»
' ””0- Thf‘ uni

The

IleP'd:l\'/
. /]
1M

41 l’ml’t‘Ttt- k
\ Slf‘ujn £5

E- .

Mechanical behavior of bones

Unless otherwise indicated, most of the information contained in the following
discussion of the mechanical behavior of bones was derived from Martin et al. (1998) and,
especially, Cochran (1982).

F oree is defined as the action of one body on another; its units are the Newton (N)
or the pound (lb). Mathematically, forces are represented as vectors, which are
characterized by magnitude, direction, and point of application. The application of
external forces (i.e., a mechanical load) to a structure, such as a bone or a portion thereof,
changes its original dimensions (i.e., causes deformation) and produces internal forces that
resist the deformation. Strain refers to a relative change in some dimension, such as
length (i.e., AL/ L), of a structure due to an external load, and is therefore unit-less. In the
literature strain is represented by the symbol 8. The strains in bone are very small and
are generally reported as microstrain (its; lite = l0’68). As an example, a 10cm piece of
bone stretched to 10.00001cm represents a change in length of 0.00001cm (AL 2 10‘5cm);
therefore, the strain is 0.0000018, and the microstrain is 1.0tts (AL/L = 10'5cm/10cm =
l 0'68 ‘—" lite). The internal forces that develop within a structure to resist deformation are
Called stress. Stress is defined as a force per unit area (F/ A), and is represented by the
Symbol 0. The unit of stress is the Pascal (Pa), which is equal to one Newton per square
meter (lPa = l N/mQ). Stress levels in bone are typically reported in megapascals (Mpa;
l Mpa = 106Pa). Stress and strain are directly related to one another (0 2 E8; E is a
material property known as the elastic modulus). Stress is not measurable phenomenon;

however, strain is.

Externally applied forces are classified as either normal or shear, depending on

19

Lhcir puim of up} WIT
fiifl‘t‘fi are applied
Slftlt‘llt‘tl. or a uni
{muse one plane
Compressive. trn~
:unwntinn. a "111'
pmiiii‘rx'alues. ll.
strength in triisii u
ln the stir:
matrrialnftsliith ;
strui‘tuml pmlm.”I
M" Ifht' mt‘t'lm:
”m“ (lt'prntls n;

and the ”it Klt‘ of 1H

 

lv'uns rxiwrienrrd ll

grid. to a murh 1m

i‘mntinunush' ran 11

1‘ -'
f‘lsl ‘ I
...tntiuti(ins ofstrt‘

ll11h rt‘ll‘rt‘l
"T‘mrr -. ~~ ‘
fllf‘llll‘lt‘.1ll\

flt‘i :'
.ulllf'fl (\‘lintlc‘r‘
o I I \

Illfil" .
"s'm'atluns :1
~ A

{Ugh s.
‘15 lllf‘ ffIanT (.
f

(iii
hran lflfi') in
‘~ \\

their point of application relative to the surfaces of a structure (Figure 2-2a). Normal
forces are applied perpendicular to a surface, and produce tension if the object is
stretched, or compression if the object is shortened. Shear forces, on the other hand, tend
to cause one plane of the material to slide relative to an adjacent parallel plane.
Compressive, tensile, and shear forces produce corresponding strains and stresses. By
convention, compressive strains are reported as negative values, and tensile strains as
positive values. Bone is strongest in compression, weakest in shear, and of intermediate
strength in tension.

In the science of mechanics, a distinction is made between a structure and the
material of which it is composed. In biomechanics therefore, one must distinguish the
structural properties of a bone, such as a femur, from the material properties of the bone
tissue. The mechanical behavior of a bone (i.e., how it responds to a specific loading
regime) depends on a combination of its geometry (size and shape), its material properties,
and the mode of loading (e.g., axial, bending, and torsion; see Figure 2-2b). The typical
loads experienced by bone are primarily produced by muscle contraction (Frost 1997a)
and, to a much lesser extent, by gravity. “[D]uring normal function bone is subjected to
Continuously variable modes of loading” (Cochran 1982: 172), resulting in complex

distributions of stresses and strains.

With reference to an idealized vertical cylinder, axial loads are applied either off-
Center (eccentrically) or aligned with the center axis (Figure 2-3). Because bones are not
idealized cylinders, but rather have anatomically complex shapes and joint
configurations, axial loading in the skeleton is typically eccentric. VVeight-bearing bones

such as the femur commonly experience a type of loading called eccentric column loading

(Cochran 1982) in which a longitudinally oriented force applied eecentrically produces

20

 

 

V

unli nttlt‘t’l

 

 

\J

axial

 

(9*

 

 

 

 

 

 

V
unloaded tension compression shear

t

 

 

 

 

 

 

v
v
axial axial
tension compression bending torsion

0))

Figure 2-2. External forces and modes of loading. (a) Depending on the point of
application, forces can be normal (tension or compression) or shear. (b) Modes of
loading: axial tension, axial compression, bending, and torsion (twisting).

‘21

 

 

 

 

fszure 2-3. .-\.\1.

trailing in L1(l( ll
Etimprrsslt‘t‘ st r.

C5th ("ilitl‘ie (lidp‘

féiimtirrssiw str

Maximum run

time. and dim

IrlnS‘ilit in l‘n m

Hurling result

glam? iJUne is

.aritilmpr bra

“filling in I},

 

 

 

 

 

 

 

\fi/ \/

(a) (b)

 

Figure 2-3. Axial loading. (a) Centric loading. (b) Eccentric loading.

bending in addition to direct compression (Figure 2-4a). Therefore, in addition to axial
Compressive strains, bending superimposes additional compressive strain on the concave
Side of the diaphysis, and places the convex side under tension. Consequently,
Compressive strains in long bones are typically greater in magnitude than tensile strains.
Maximum compressive and tensile strains and stresses are found at the surface of the
bone, and diminish toward its center eventually decreasing to zero in the plane where the
transition from compression to tension occurs. This is referred to as the neutral plane.
Bending results from other loading configurations as well. For example, when one end of
a long bone is fixed and a load is applied to the other end, the bone is said to be loaded as
cantilever beam (Figure 2-4b). In this situation, the free end of the bone is deflected

resulting in the creation of bending moments along the shaft of the bone. The greatest

22

tensile and mmpi

 

guliierted to benii

 

 

u Lamilfwr
1., ,

LL -:
Hui thin in 'i

b ”main”
is

tensile and compressive stresses are located closest to the fixed end of the bone. Bones

subjected to bending also experience shear stresses and strains.

load

 

 

(b)

Figure 2-4. Loads that produce bending of long bones. (a) Eccentric column loading.
(b) Cantilever beam loading.

To empirically test the mechanical behavior of a bone, a static load is applied
until the bone breaks. The resultant deformation can be plotted against the load
producing a load-deformation curve (Figure 2-5a). From this curve, the structural

PrOperties, such as strength and rigidity, can be ascertained. The load-deformation curve

has two distinct regions; the first is called the elastic deformation region, and the second is

t‘lllt‘tl 111“ plasttf l

limit. a point in it

 

ill!“ graph ()l‘lllt‘ ‘
Edim‘tlt' pm pm
tt‘mt‘tVt‘tl in this rt
region represenb
Rigidity varies ltt\
utltrr words. a lull
shutter bone. 1n
Alter the load is rt
grim. and the .‘l‘
Cuntinual lnmliy
l‘ the ultimate 1t
Simply ret‘mt-tt

To deter

:‘jtl gromf‘ln' 111T

mange in le

ne‘
figure 2-3b
Ft, '.
twill. lelttlth
ms '

tenal m e: in

mtutulus. Y! in

t.
«Frey 204 is)

called the plastic deformation region. The regions are separated by the proportional
limit, a point in the curve where the slope becomes non-linear and begins to level off.
The graph of the elastic region is a straight line because the initial increase in deformation
is directly proportional to the load, a phenomenon known as Hooke’s Law. If the load is
removed in this region, the bone will return to its original shape. The slope of the elastic
region represents the rigidity of the structure, a measure of its resistance to deformation.
Rigidity varies inversely with the length and directly with the thickness of a structure. In
other words, a longer bone must be thicker to resist deforming under the same load as a
shorter bone. In the plastic region of the curve, a permanent deformation remains even
after the load is removed. The point in the curve where this occurs is called the yield
point, and the yield strength is the load at which plastic deformation commences.
Continual loading will eventually cause the bone to break. The load at failure is known
as the ultimate load, and it defines the ultimate or breaking strength of the bone, often
simply referred to as the “strength” of a bone.

To determine the mechanical properties of a material, e.g., bone tissue, the effects
of geometry are factored out by converting load (force) to stress (F / A) and deformation
(change in length, AL) to strain (AL/ L). The resultant curve is called a stress-strain curve
(Figure 2-5b). The slope of the stress-strain curve within the elastic region is known as the
elastic modulus or Young’s modulus (E), and is a measure of the intrinsic stiffness of the
material in compression and tension. For shear loading, this property is called the shear
modulus. Young’s modulus varies within a cross-section and along the length of a bone

(Currey 2002). The area under the curve represents the toughness of the material, a

24

mmmohmwn

pewtpw~

 

bunt int‘lUtle (le:

.mmmmtmm

YU)

I ,( ’Elf'l ('0')

 

 

 

 

 

 

Dd}

Emng-Su
Wilt)“ Uf lilt‘ (f
P:t[)la§llt‘ “‘31"
Film :ummc
”WWI“; a“.

Thee
Ftp: '
..\ntl)lL\ (‘I )1“

t
l '1 . ‘
slit 15 L146,

 

measure of the amount of energy absorbed (or work') to failure. Toughness relates to the
ability of a bone to resist crack propagation. Factors that affect the material properties of
bone include degree of mineralization, porosity, microarchitecture, collagen fiber

orientation, fatigue microdamage, and rate of deformation (i.e., strain rate).

 

 

 

 

FAIL A

E: YLD fi

:5 PL 51

«s 33

3 E ii

Toughness m
Deformation (AL) Strain (AL/ L)

(a) (b)
Figure 2-5. Structural and material properties. (a) Load-Deformation curve. E=elastic
region of the curve, the slope of this region represents the rigidity of the structure;
P=plastic region of the curve; PL=proportional limit; YLD=yield point; FAIL=failure
point (=ultimate or failure strength of the structure); the shaded area under the curve
represents the energy absorbed to failure (toughness). (b) Stress-Strain curve.

The combination of mineral and collagen makes bone a composite material that
exhibits complex mechanical behavior. Bone is weakly viscoelastic, which means that its
material strength and elastic properties are somewhat dependent on the rate at which the
bone is loaded. In fact, these properties increase with higher rates of loading. Bone is

also anisotropic, which means that material properties, such as Young’s modulus, are

dependent on the direction of loading. This is because bone, like wood, has a “grain”

 

I Work is defined as force acting over a distance. Its units are the newton-meter (Nm) orJoule (It

25

that is partly tlllt
orientation olset
I

laneitudtnal (llrt‘t
'l‘lte lll(‘t i

Ill-ill? pointed Ht
lmffitt‘tt’tttt; in m ('2
mt defomt Ulltlt'!

Stmnq bone is Ul i

pttttxittitittdl to tl

 

mineralization. A
brittleness: then-ti
still“ Cum-y 2m
diluent function
:3!“ ll)? same li tr e
Lllrt"llg’ltt_tttt the .\‘l
‘JbL‘tt‘t’ed amt ittq

Most of ti

ill: ' ‘ i I
(lt‘l'tomlllltnh

v

lttlttt‘t‘t‘t‘l'. in life

that is partly due to collagen fiber orientation and partly due to the longitudinal
orientation of secondary osteons. Bone is stronger and stiffer when loaded in the
longitudinal direction, than in the transverse direction.

The mechanical properties of bone often represent a compromise. As Currey
(2002) pointed out, one might assume that the ultimate strength of a bone is most
important; however, because the efficiency of muscle contractions requires that bones do
not deform under a load, clearly, for many bones, rigidity is of foremost importance. A
strong bone is of little value if it bends when the muscle contracts. Rigidity is
proportional to the niaterial stiffness (E), which is largely dependent on degree of
mineralization. As bone becomes more mineralized, stiffness increases but so does
brittleness; therefore, toughness decreases—“Bones cannot be both very tough and very
stiff” (Currey 2002: 136). Martin et al. (1998: 140) point out, “Different bones have
different functions, and the best solution to the conflicting mechanical demands need not
be the same for every bone.” The specific evolutionary compromises struck vary
throughout the skeleton, and likely contribute to the differential response to loading
observed among bones.

Most of the foregoing discussion dealt with the mechanical properties of bones
under conditions of static (monotonic) loading (i.e., loads that are continuously applied);
however, in life, bones typically experience dynamic loads, in other words, repetitive
loading cycles (i.e., repeated cycles of loading and unloading; also referred to as cyclical
loading). Repetitive loading generates microscopic fatigue damage (e.g., mierocracks),
which alters the mechanical behavior of bone. Failure resulting from the accumulation of
fatigue damage (fatigue failure) occurs at smaller loads than the ultimate strength of bone.

Studies have shown that dynamic not static loading is osteogenic (Lanyon and Rubin

26

1931; Ruhin and
"designdl" to m t r‘

I
ultimate strength

only respt'intl to tit

l‘illi-tir“ “the cellti

 

thestrain it exper
from high lmmrli
til itijimtal 1x me a
“Junttiunal st mi t 1‘
1983‘; Rubin l‘li‘
potential future

Trtf‘rhanism by \
Simmer 191-13;

mf‘t‘hanisms (a

‘ I‘D—JrilQfiD
‘~ ~ A
fig-Int} ’

<1“ r\ 3

1984; Rubin and Lanyon 1984b), and it has been hypothesized that bones may be
“designed” to minimize fatigue damage (Carter 1984; Frost 2000) rather than maximize
ultimate strength. This hypothesis makes sense when it is considered that bones “can
only respond to the strains to which they are actually subjected” (Lanyon 1987:
lO86)—“the cellular population responsible for functional adaptation can only respond to
the strain it experiences while swinging through the trees, not the potential for falling
from high branches” (Rubin 1984: $17). In fact, it has been observed that some aspects
of normal bone architecture, such as long bone curvature, may actually increase
functional strains, but also “control” the direction of bending (Bertram and Biewener
1988; Rubin 1984). One explanation for this is that because bones cannot foresee
potential future catastrophic strains, high but predictable functional strains provide a
mechanism by which bones can adapt to their typical strain environment (Bertram and
Biewener 1988; Lanyon 1987). This would be advantageous as long as repair

mechanisms can keep up with the microdamage that results from higher strains.

Concepts in skeletal mechanobiology

Mechanobiology is the study of how the biological processes in bone produce
adaptation to mechanical loading. The following are some of the questions
mechanobiology research is attempting to answer (Hart 2001; Rubin et al. 1992; van der
Meulen and Huiskes 2002): 1) What is the mechanosensor? 2) What mechanical loading
variable is being “sensed,” i.e., what is the nature of the mechanical signal? 3) What are
the pathways involved in the transduction of the mechanical signal into a cellular
response? 4) Is the adaptive process the same throughout the skeleton, or does it vary by

bone and location? While there has been significant progress in the field of

27

ntt'rhanoliiult tux

 

. l
21ltll;ll;lt1 21W

  
 

and contempt mt .
The tiliili'
necessitate the (‘\

accepted schenm

 

l‘f’fli'. ()stem‘ylt
llttlt’l through tht'
electric a1 signal

osteoblasts. ill
adapting the lx

<ulliriently add
‘Tl‘JS‘Kl.

Resear
nerts'nrk“ CC
Klein-Xmem
Moss ‘Ztlttl
,. .
killing (ring. 2

’“f‘Ptinn. q

(t

\,
lita

“sthlnttln r‘.

mechanobiology, there are still no definitive answers (Burger 2001; Cowin and Moss
2001; Hart 2001). The following discussion highlights some of the predominant issues
and contemporary hypotheses.

The ability of bone to adapt its architecture to mechanical usage would seem to
necessitate the existence of a self-regulating biological feedback mechanism. A generally
accepted schema of skeletal adaptation is as follows (Cowin and Moss 2001; Lanyon
1993). Osteocytes sense some aspect of their strain environment via the flow of interstitial
fluid through the lacunocanalicular network. The osteocytes then transmit a chemical or
electrical signal to bone-lining cells, which recruit the effector cells (i.e., osteoclasts and
osteoblasts). The effector cells respond by removing or adding bone tissue thereby
adapting the bone’s architecture to suit its mechanical environment. Once the bone is
sufficiently adapted, the signal diminishes to an acceptable level, and the feedback loop is
closed.

Research suggests that the “mechanosensory organ” of bone is the “connected cell
network” (CCN) and its associated lacunocanalicular network (Burger 2001; Burger and
Klein-Nulend 1999; Cowin and Moss 2001; Turner and Forwood 1995). Cowin and
Moss (2001) describe the CCN as a “functional syneytium” consisting of osteocytes, bone-
lining cells, and osteoblasts, which functions as the “site of intracellular stimulus
reception, signal transduction, and intercellular signal transmission” (p. 29-3). Within
the CCN, osteocytes are considered the mechanosensor cells (Lanyon 1993). Their
distribution within the bony matrix and their numerous cellular connections make them
ideally suited to detect local changes in the mechanical environment, and transmit the
information to other cells in the network.

As an interesting side note, it has been hypothesized that rather than stimulating

28

tt‘mtxlt‘llng. the t

 

inhibitory signal!
an inhibitun' siet

This inltihitiun i~
death. or the trttt

the formation of

 

.lrt‘urrliug tn .\ ld
ttinttdtlit'tttt'l' ”11
€jt\'t'flt.t&(l sittttitit

just ht m
Burger 20411: B
lttputhesis is th
arunrx‘analiru
tilt-(ts oflluid t

pmstaglantlins

t. -
“willing B‘thr‘

it
uh), I)“ It]

lf‘l
' ”mimic IT

(Wat
(is ”“811

1
Erin - ..
m” andllt'

1n.-
:f‘ \ ‘
””11" 31V

21v,
anl [H (a

Elli '
“Q Ca , .
n dill“;

”119T. ti d1”
R

f)r

remodeling, the normal response of osteocytes to mechanical usage is to transmit
inhibitory signals to bone-lining cells (Martin 2000). This assumes that in the absence of
an inhibitory signal, the normal behavior of bone-lining cells is to activate remodeling.
This inhibition is released when generation of the signal ceases due to disuse or osteocyte
death, or the transmission pathway of the signal is disrupted, which could occur through
the formation of microcracks. In this view, microdamage mimics localized disuse.
According to Martin (ibid), this hypothesis is consistent with and explains the apparently
contradictory observation that remodeling activity is increased in both disuse and
overload situations.

Just how osteocytes sense a mechanical signal is the subject of much research
(Burger 2001; Burger and Klein-Nulend 1999; Turner et a1. 1994). The predominant
hypothesis is that osteocytes sense mechanical loads via interstitial fluid flow through the
lacunocanalicular network. In vitro studies of cultured osteocytes have shown several
effects of fluid flow on osteocytes including enhanced production of nitric oxide and
prostaglandins, paracrine messenger molecules known to increase following mechanical
loading (Burger and Klein-Nulend 1999; Burger 2001; Nomura and Takano-Yamamoto
2000). The mechanism of interstitial fluid flow involves deformation of the bone matrix
by dynamic mechanical loading, particularly loads which produce bending. Bending
creates pressure gradients, which in turn forces interstitial fluid through the
lacunocanalicular network (Tate 2001) from regions of compression to regions of tension.
Interestingly, there is some evidence that increased venous pressure and the heartbeat
also produce interstitial fluid flow through cortical bone (ibid). The flow of interstitial
fluid can affect osteocytes either by transduction into an electrical signal (e.g., streaming

potentials) or by direct mechanical deformation of the cell (e.g., fluid shear stress).

29

Streaming pt )lt‘1.

int‘ludes piezt N‘lt

 

 

 

 

on the pie/.tx‘leet
charge prntlttt‘t‘tl
it} the separatit 11 ,
piezrrlet‘trit‘it} is

instead turned tli

 

 

 

generated hy the
la‘untx‘analit ulat
met‘hunit'al (lt‘li )f
ul the shear st rt w
shear stress p H A?
min emirt mmt
stimulated by str
unresolved Ct m

“hilt“ tlt
ititrrmiriing am
111 lde‘fltili'ing (h

responsible ft )1‘ t

asttitttttes int‘l

Streaming potentials fall under the category of stress-generated potentials, which also
includes piezoelectricity. Initial interest in electromechanical phenomena in bone focused
on the piezoelectric effect (Currey 2002). The direct piezoelectric effect is an electrical
charge produced by deformation of crystal structures—an electrical potential is created
by the separation of charged particles resulting in polarization of the crystal. However,
piezoelectricity is not thought to play a role in the adaptive process, and researchers have
instead turned their attention to streaming potentials (Pollack 2001)—electrical currents
generated by the flow of an ionized fluid along a charged solid surface (e.g., through the
lacunocanalicular network). The flow of interstitial fluid can also produce direct
mechanical deformation of cells through the effects of fluid shear stress. The magnitude
of the shear stress is related to the strain generated by mechanical loading; therefore,
shear stress provides a direct source of information for osteocytes regarding their local
strain environment (Burger and Klein-Nulend 1999). The issue of whether osteocytes are
stimulated by streaming potentials, fluid shear stress, both, or neither is currently
unresolved (Cowin and Moss 2001).

While there is general agreement that mechanical stimuli are essential in
determining and maintaining the structural integrity of the skeleton, “controversies arise
in identifying those specific parameters of the complex physical milieu which are actually
responsible for modulating or initiating. . .adaptive processes” (Rubin et a1. 1992).
Possibilities include stress, strain, or microdamage (Prendergast and Huiskes 1995). Most
models assume that some time-averaged stress or strain variable (i.e., loading history) is
involved, and most research suggests that bone responds to changes in the strain
environment rather than stress (Burr 1992; Burr et al. 1989; Frost 1990b; Lanyon 1981,

1984, 1987; Martin and Burr 1989; Rubin et al. 1992; Tate 2001). Many possible strain

30

variables have 1”
inrluding: strait!
and strain enerL’“
l9”? and lany
interaetix'e et mil .
Rubin et :11. ll“‘
siznilieant resent
fling
Since [lie
functionally atla'I
Philip-"5 Rneslpr
sietliitit'in ("run1 ll .1
l’f'T‘PtTties Ufa I),
n"T’TTTtiil. Swain; t

rrmrxleling . \\ lllt

 

i.e‘httet‘ture. ther
nteehanostat" m'

mnrlitinns ie “
. .. \

nr’gatit'e fee

(lbark
h. “
‘1'?“ [1191? is a {71
Teal

y .
1 th -~
resholtl \"1ltl
k _ I
all. ) i.
sage \x'

variables have been proposed as the primary stimulus for mechanical adaptation
including: strain mode, strain magnitude, strain rate, strain gradient, strain frequency,
and strain energy density (Burr 1992; Martin and Burr 1989; Rubin et al. 1992). Burr
(1992) and Lanyon (1984, 1987) have suggested that it is likely the actual “signal” is some
interactive combination of variables, such as strain rate, distribution, and magnitude.
Rubin et al. (1992) suggest that bone adaptation is also frequency dependent. In spite of
significant research on the subject, this issue is unresolved for lack of consensus (Currey
2002)

Since the late 19‘h century it has been hypothesized that bones become
functionally adapted to their mechanical environment by way of an error-driven feedback
process (Roesler 1987). In other words, some optimal level of adaptation exists, and
deviation from this optimum initiates a corrective response. Relative to the mechanical
properties of a bone, mechanical usage can generate strains that are too low, too high, or
normal. Strains that are too high or too low initiate biological processes (i.e., modeling or
remodeling), which either increase or decrease bone mass and modify the bone’s
architecture, thereby normalizing the strains experienced by the bone. The
“mechanostat” model was proposed by Frost (1987, 1996) to explain under what
conditions (i.e., “when”) the adaptive mechanisms are activated. The mechanostat is a
negative feedback model analogous to a thermostat—the adaptive system is turned “on”
when there is a mismatch between mechanical usage and bone architecture, and is turned
“off” when they are brought into parity. The mechanostat model employs two concepts:
1) threshold values or set points, which Frost has named minimal effective strains (MES),
and 2) “usage windows,” which are ranges of strain values that entail a specific set of

adaptive responses (Figure 2-6). Because there are probably more strain variables

31

 

E

Z

:3
l
C
D
1

[—fi

l loss

 

limtre 2-6. The
mtleeling and re
tstndow. lies belt

lth'Imild «m

uterloarl \eim lc m

tt‘SlE‘Tl intend to
.lletab 18: 278

int"' '
tilted tn stimtt

1'13? ~
proposed t

-1
mhnmdte

to .\ll
inshohl\

'Iilues .

.l . .
ll.Sr; ~31) - 10"

 

mirror ,

ldman‘ _\I

fray lqrp
.., 1a. 2m

 

 

 

DWI AW ’l‘ MOW ,f POW i

gain .’. T

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

.0. Fracture
E .................. ‘.o'.
2 .o" """"""""
s
O
.0
C
D
i, 0
loss
IVIESr MESm MESp Fx
100 1000 3000 25.000
Strain (its)

Figure 2-6. The “mechanostat” model. The dotted curve represents the net effects of
modeling and remodeling on bone mass within the four usage windows: DW=disuse
window, lies below the MESr; AWZadapted window, lies between the MESr and MESm;
. MOWZmild overuse window, lies between the MESm and MESp; POWZpathological
overload window, lies above the MESp. Adapted from Frost, HM (2000) Does bone
design intend to minimize fatigue failures? A case for the affirmativej Bone Miner

Metab 18: 278-282.

involved in stimulating bone adaptation than simply peak strain magnitude Lanyon
(1987) proposed the acronym MESS (minimum effective strain-related stimulus) as an
alternative to MES. The current version of the mechanostat model includes four
threshold values and four usage windows (Frost 2000). There is an MES for remodeling
(MESr; ~50 - lOOuE), an MES for modeling (MESm; ~ 1000118), an MES for
microdamage (MESp; ~3000pl8), and a threshold value for fracture (Fx; ~25,000u£)

(Frost 1997a, 2000). Below the MESr lies the “disuse window” in which disuse-mode

32

rtmmlt‘ling l5 tit \
rrmrxlellng-(lrpr
range ofs‘train \ .
mnsm'ation-m;
this mmlcl. the “ r;

\xintlms'.” frost

 

lies liens een .\l lill
llLSp ln l‘lHltll
resulting in the a
Xtt'uftllnfl ll) 11'
remrrleling ma}
pathological (in
extremely high
normal \'< )luma
Ffpair itself thn
Lian lamellar l;
“5‘7 3 n'ieral ad
211989.,

The m:
in [he an‘ew

fifty

r‘ .
H|}\ C (‘(1I]\t.

5me ..
k ling l)“

a» . .
if} \s'rrh aw.
‘\ 2

”KT” __ .
x

remodeling is activated, modeling is inhibited, and bone mass is reduced due to
remodeling-dependent bone loss. Between MESr and MESm there is a physiological
range of strain values that do not initiate an adaptive response, and baseline
(conservation-mode) remodeling occurs primarily in response to metabolic demands. In
this model, the “goal” of adaptation is to maintain bone strains within this “adapted
window.” Frost has also defined two overload windows. The “mild overload window”
lies between MESm and MESp, and the “pathological overload window” lies above the
MESp. In conditions of mild overload, formation-mode modeling may be activated
resulting in the apposition of lamellar bone on periosteal or endosteal bone surfaces.
According to Frost, remodeling is inhibited in conditions of mild overload. However,
remodeling may be initiated or redirected to remove microdamage. Within the
pathological overload window the accumulation of microdamage resulting from
extremely high peak strain magnitudes (i.e., greater than those typically generated during
normal voluntary activities in adults) is hypothesized to outpace the bone’s ability to
repair itself through remodeling. This triggers a type of modeling in which woven rather
than lamellar bone is deposited. There is some debate as to whether this is a pathological
or a normal adaptive response to extreme overloading (Bertram and Swartz 1991; Burr et
al. 1989).

The mechanostat model predicts that mechanical usage will have opposite effects
on the processes of modeling and remodeling, which nevertheless produce similar
adaptive consequences. For example, stimulation of modeling and inhibition of
remodeling both tend to increase bone mass. Because modeling and remodeling activities
vary with age and exhibit envelope specificity, the architectural consequences of

mechanical usage differ between skeletally immature and mature individuals (Frost I987).

33

  
 

Prior to Slit‘lt‘tal r
litmmtinn-mr x lt-
is‘hirh conserves
inrreasetl extent.
Ml greater t‘ot‘th

mnsert'es hr )llt' n

 

trabecular t‘th‘lU
”Urinal mt Klt‘lifi‘.
brine can dt'itit‘V
This issue is atlt
r'rt'erload i ng f.t *
experience a fr
r‘ndt‘tste :11 surf.
trmtxieling. l

tunes with sn

mn‘hanical

U

cortices resul

em .

\ t

#5511110? 21 \‘1

.\n ‘

rh-
“iqpl—FFS -

C) .
jli‘lf‘t'mhh.
{'0pr

'1 (11‘1“?

Prior to skeletal maturity, increased mechanical usage involves gains in bone mass 1) by
formation-mode modeling on periosteal surfaces, and ‘2) by suppression of remodeling,
which conserves bone on endosteal surfaces. Architectural consequences include
increased external diameter, decreased medullary cavity diameter, thicker cortical bone,
and greater cortical area. In contrast, increased mechanical usage in adults primarily
conserves bone mass by suppressing remodeling-dependent bone loss on endosteal and
trabecular envelopes. It is not known whether increased mechanical usage produces a
normal modeling response in adults, and it is still an open question as to whether adult
bone can achieve similar increases in external diameter as seen in younger individuals.
This issue is addressed in more detail in chapter 6. Under conditions of extreme
overloading (i.e., within the pathological overload window) adults andjuveniles might
experience a form of modeling involving the addition of woven bone on periosteal or
endosteal surfaces. Decreased mechanical usage inhibits modeling and stimulates
remodeling. In skeletally immature individuals, the decrease in bone formation produces
bones with smaller external dimensions. The architectural consequences of decreased
mechanical usage in adult bones are increased medullary cavity diameter and thinned
cortices resulting from resorption of endosteal bone. Additionally, because the periosteal
envelope tends to experience a net positive bone balance throughout life, increased
remodeling does not tend to reduce the external diameter of a bone, and may in fact
produce a slight increase in diameter (Frost 2000).

An important concept of the mechanostat, which will be further addressed in
chapters 5 and 6, is that changes in the MES setpoints can mimic changes in the strain
environment of the bone—an increase mimics disuse, a decrease mimics overload.

Consequently, changing the setpoints results in alterations of bone architecture that are

34

intlistingIIIFhal-ll‘
furtheminre. tht
nutrition. genetit
19542: Frost 19“.".
terlmnical stint
emphasis added
The lllt"(
Burr I902: .\l.u
lm‘unsistencies l
Tt‘t‘ml}~ pr: th N
«“nncept of mini
mnditions of t'( 1
dt'nSity of zero .
the wrong mt )(l:

threshold v

alue
stmngh' to trait

mm" ‘
ct] no he“

Epid' '
. adaptatm

emrrmment" t

Second we‘ll-m
_ . c:

l . .
1E3 is. the sam

:Jlj'a ‘
ml? bones a
lift
”fill? r‘r‘

- (Mr-
1‘ {\‘1 r».

indistinguishable from those produced by changes in mechanical loading (Frost l996).
Furthermore, the mechanostat model predicts that nonmechanical agents, such as
nutrition, genetics, hormones, and disease are responsible for changing the setpoints (Burr
1992; Frost 1996;Jee 2001; Martin et al. 1998). In other words, “the {fleet of the
mechanical stimulus is dependent on the nonmechanical environment” (lee 2001: 1-37,
emphasis added).

The mechanostat is not the only model of skeletal adaptation to mechanical usage
(Burr 1992; Martin et al. 1998; Turner 1999); however, it is probably the most cited.
Inconsistencies between Frost’s mechanostat and experimental data led Turner (1999) to
recently propose “the principle of cellular accommodation” as an alternative to the
concept of minimal effective strain. Turner claims the mechanostat predicts that under
conditions of complete disuse, bones would be completely resorbed (i.e., reduced to a
density of zero). Because this does not happen, Turner suggests the mechanostat must be
the wrong model. Whereas the mechanostat proposes that bone cells respond to certain
threshold values, the principle of cellular accommodation assumes that “bone cells react
strongly to transients in their environment, but weakly to steady state signals. ..[which
convey] no new information to the cells” (ibid: 467). Turner suggests that the initial,
rapid adaptation response dies out as “bone cells ‘accommodate’ to the new
environment” through “changes in the cell’s sensitivity to external stimuli” (ibid: 467). A
second weakness of the mechanostat, according to Turner, is the assumption that the
MES is the same throughout the skeleton. A variable setpoint explains why weight-
bearing bones are more affected by disuse than non-weight-bearing bones, and why the
bones of the cranial vault do not erode away even though they are routinely subjected to

much lower strains than bones of the appendicular skeleton. Turner claims that the

  
   
 

phdfltufnlh
Sllrtt‘llhln the s
have become at .
site <(X‘t‘lllt‘. am i
an important my

prinriple of t‘t'llll

 

tempural S("l“‘in
(.1th \t'tll'flS. tltt‘
t;t_tntitt'§1t‘t\t‘§l 11 \
Frost‘s “
birth tiahle mt :1
with.” I‘ix'en a
untlerstantlitt‘:
research seem:
Til? significan
ariifferent rel.
mt‘rlel \\ ill im
Knnrding tn
heriu- '

u see ll get
”mini: is lik
’fr Ctjtngpu? (g
..ei‘s‘hbliftin;
iron} Inf K. 1

‘ , . Vt _

principle of cellular accommodation “predicts that the set point. . .will vary from site to
site within the skeleton depending upon the local strain environment to which the cells
have become accommodated” (ibid: 468). Of course, it is also possible that the MES is
site specific, and the question of whether bone cells “know where they are” is considered
an important one in mechanobiology (Currey 2002). A third distinguishing feature of the
principle of cellular accommodation is “that final bone mass will be dependent upon the
temporal sequence of preceding mechanical loading/ hormonal events” (ibid: 469). In
other words, the final structure of a bone can be seen as an example of historical
contingency; it is dependent on a sequence of unique antecedent states.

Frost’s “mechanostat” and Turner’s “principle of cellular accommodation” are
both viable models, but it is likely that neither is, as Frost (1995) would say, “the whole
truth.” Even a relatively casual perusal of the literature demonstrates that our current
understanding of the biology of mechanical adaptation is still extremely incomplete, and
research seems to produce more questions than answers (Cowin 2001a; Currey 2002).
The significance of competing models to the topic of this dissertation is that each implies
a different relationship between physical activity and bone architecture, and the choice of
model will impact the interpretation of group variation in cross-sectional geometry.
According to Frost (1997a, 1997b), long-distance running does not necessitate large bones
because it generates peak strains within the “adapted window.” Therefore, no amount of
running is likely to result in an adaptive modeling response. Running will however help
to conserve existing bone. In contrast, according to Frost, the high peak strains generated
weight-lifting does result in bigger, stronger bones. If the principle of cellular
accommodation is correct, the deciding factor in determining whether running or weight-

lifting will produce an adaptive response does not depend on the peak strains involved,

36

  
  
  
  

but rather on th
l’urthenriore. u:
artitities to hunt
“3'87: “’88 "t1.
tttt stiapth‘t‘ t’esp

differences in lllt

 

L-mss-set‘tit ”1111 2'«

Technical ove

The up;
inms‘n as l)lt mt:
studied by antl‘

lllt‘it’lg Iltdt It‘t'l

st’rftts'are. allt m
pn‘ipulations I.
tulleagues (:2
ifs-tit): Ruff et

r). ,tdrt‘hac‘tfdt t"
g

but rather on the strain environment to which the cells have become accommodated.
Furthermore, until the mechanical stimulus is known the relationship ofspecific physical
activities to bone structural variation is speculative at best. For example, if Lanyon’s
(1987: 1088) “theoretical prediction” is correct that “regardless of peak strain magnitude
no adaptive response will be engendered providing the strain distribution is normal” then
differences in the intensity level of an activity are not likely to explain group variation in

cross-sectional geometry.

Technical overview of cross-sectional geomegy

The application of mechanical engineering principles to biological materials is
known as biomechanics. Biomechanical properties of the human skeleton have been
studied by anthropologists since the mid—20‘h century; however, it was not until the early
19803 that technological innovations, e.g., computed tomography (CT) and computer
software, allowed the application of biomechanical principles to archaeological
populations (Larsen 1997; Ruff and Larsen 1990). Pioneering studies by Ruff and
colleagues (e.g., Brock and Ruff 1988; Ruff and Hayes 1983a, 1983b; Ruff and Larsen
1990; Ruff et a1. 1984, 1987) resulted in, what might be considered, a paradigm shift in
bioarchaeology whereby structural variation of the human postcranial skeleton, and long
bones in particular, became interpreted in a functional, rather than a nutritional context.
“The application of this biomechanical approach in the analysis of archaeological skeletal
remains has represented an important breakthrough in bioarchaeological study, especially
in the elucidation of specific levels and types of activities in now-extinct human groups”
(Larsen et a1. 1996: 96). This shift in focus led to research projects aimed at using cross-

sectional geometry to elucidate changes in human behaviors coincident with

37

. .7 ,
anthn’iiwlngim.

 

   
  
 
 

mt'ibilitt' strateu:

Terminology
This St‘t'i

terminology use.-

Digphyseal st:

By tit-fut
such as the fem:
shafi. or diapht‘
called the meta
thejunctiun of
fipipht‘sis. meta

distinct. T\'[)1(
function as arti

stmtturalls tra

Tl" v
tL-iatlt tapered

if]? diaphVSig

\
‘0‘
\-4’5‘ f

0r”it‘tit‘il

t'i‘tni' ‘ ‘
«1m: yellt m

1 VB 3

J : .
. I‘IHIJht,
gilt

rum oftrd}
)1

:zr. _ _
t Jp.nlf‘5

\- nlr

anthropologically relevant and archaeologically identified transitions in subsistence and

mobility strategies.

Terminology and definitions
This section provides an overview of relevant biomechanical concepts and

terminology used throughout this dissertation.

Diaphyseal structure

By definition, long bones are significantly longer than they are wide. Long bones,
such as the femur, are composed of two ends, called epiphyses, attached to a connecting
shaft, or diaphysis. The transitional region between each epiphysis and the diaphysis is
called the metaphysis. Prior to skeletal maturity, the growth plate or physis is located at
the junction of the epiphysis and the metaphysis. These three regions of a long bone, the
epiphysis, metaphysis, and diaphysis, are to a certain extent structurally and functionally
distinct. Typically the epiphyses of long bones are broader than the diaphysis, and
function as articular regions, forming joints with other bones. The metaphysis, as a
structurally transitional zone between the broad epiphysis and the narrower diaphysis, is
usually tapered, and functions to transmit and mitigate joint forces from the epiphysis to
the diaphysis. The diaphysis of an adult long bone is essentially composed of an outer
shell of cortical bone surrounding a “hollow” center, or medullary canal, which, in life,
contains yellow (fatty) bone marrow and a negligible amount of trabecular bone (Ruff
1983). Epiphyses and metaphyses have a relatively thin cortex and contain a significant
amount of trabecular bone even into adulthood. Interpretation of their mechanical

properties is more complex than for the diaphysis.

38

 

Engfue'iflg

 
  
  
   

\e (lis‘t‘tl
strength. rigitlit‘s.
for the purpnst'
lending patterns
gerimetrt'. and t
Ruffand llas't“
by experiment;
ahhuugh than
1930 . “itilt‘
Ufanthrt )pnlti

Diret'
“i afresh Spc
Effects of c 1 it
ITlfi'i‘Wthftnet

it)“;
. .t.>.ant1 i

B”’i’mfi‘han
midi-t.“ Ufa
5&1 ”final QC

CL
1i". j
J’ S 1x”

Fifi .
LT ham“ ';1

Engineering beam theory

As discussed in the previous section, the structural properties ofa bone (e.g.,
strength, rigidity, toughness) are dependent on both its geometry and material properties.
For the purposes of behavioral interpretation, it is assumed that alterations in mechanical
loading patterns (e.g., physical activities) principally result in adaptive changes to bone
geometry, and that material properties remain unchanged (Larsen 1997; Ruff 2000b;
Ruff and Hayes l983a; Ruff and Larsen 1990; Ruff et al. 1984). This is largely borne out
by experimental research (Haapasalo et al. 2000; Heinonen et al. 2001; Woo et al. 1981),
although changes to material properties have also been reported (e.g., Matsuda et al.
1986). While adaptive changes can theoretically occur in any skeletal element, the focus
of anthropological studies of bone biomechanics has been on long bone diaphyses.

Direct measurement of whole bone strength and rigidity requires empirical testing
of a fresh specimen with uncompromised material properties. However, “Because the
effects of climate, immersion, and time on skeletal tissue are severe, the direct
measurement of strength in archaeological specimens is not possible” (Lovejoy et a1.
1976), and indirect methods that estimate mechanical properties must be used.
Biomechanical analyses employing engineering beam theory allow estimation of the
rigidity of a bone, and more indirectly, its strength2 via analysis of diaphyseal cross-
sectional geometry. The theory behind biomechanical analysis is that the tubular shape
of long bone diaphyses allows them to be modeled as hollow beams. Just as civil or

mechanical engineers determine the mechanical behavior of beams used in construction,

 

2 The terms rigidity and strength are often used interchangeably in the bioarchaeological literature (Ruff
and Hayes l983a). Although they are correlated, as noted earlier, they are in fact distinct properties.

39

hinmt‘t‘hanists t

 

(fries-sectional

 
  
  
  
  

attrtiral hone \\ i
truss-sectiuttai L
taiues. as 10112 J
ffJRii~~ "beam t
but seriously rm
model get \metri
inapprt )j)f1111 t

(irUSS‘S
and second mt
t‘rtiss-sectit m .
1m. CA. '1’
lituntlart- “1‘
my“? CA as
differenfp 1“
ad n’fitfl‘tg
[m‘S‘S‘f‘Ctiru
lemme

. ~ (ltt‘
{Ups Sf‘S ‘ (\j

a??’&

biomechanists can determine the mechanical behavior of long bones (Larsen 1997).
Cross-sectional geometric properties reflect the amount and distribution of diaphyseal
cortical bone within a specific cross-section. Estimates of rigidity and strength based on
cross-sectional geometry correlate with theoretical and experimentally derived (in vitro)
values, as long as the section location is not too near the ends of the bone (Ruff
1983)—“beam theory is adequate for representation of stresses in the femoral diaphysis,
but seriously overestimates stress in the trochanteric and femoral neck region. . .beam
model geometric analyses of metaphyseal and epiphyseal areas of long bones are
inappropriate” (ibid: 143).

Cross-sectional geometric properties fall into two categories, cross-sectional areas
and second moments of area. Cross-sectional areas measure the amount of bone in the
cross-section, and include total subperiosteal area (f A), medullary area (MA), and cortical
area (CA). TA represents the total area contained within the outer (subperiosteal)
boundary of the cross-section, and thus includes both the actual area taken up by cortical
bone (CA) as well as the “empty” space of the medullary canal (MA). Cortical area is the
difference between the total subperiosteal area and the medullary area (CA = TA ~ MA),
and reflects the bone’s resistance to the axial forces of tension and compression. A larger
cross-sectional area is more effective at resisting deformation due to pure axial loads
because the forces are distributed over a larger surface resulting in lower strains and
stresses. Axial rigidity is equal to the product of Young’s modulus (E) and cross-sectional
area.

Second moments of area (SMA; also known as area moments of inertia) are
determined by the amount and distribution of bone within the cross-section. Second

moments of area predict structural behavior in bending and torsion with reference to a

40

.. .. j ,
spttiiht nt utm.

 

~ 1
intlit‘aturs of l). .

   
 
  
  

striking. the ft.
more rigid the 1
rigidity: h m e\' t
are set mm mm
stranger struct it
further and furt
limited \\ hen p

The S}
iter'ilic neu t ra
mtdulus F. tt‘
1"“ Wu‘ntittl .
k‘1‘,.1nta\. (1
medial-latent
mfiilJl-ldlvr,‘
fl'ttgg-g(..\.““ ’11

l- ' i .

(Tau-”i“ the.
l ‘ 1min: aht

maL\iimum —

I

. .—\ l.
Nation 01"

“‘1 v, ‘
“Mimi

specific neutral plane or neutral axis. Larsen (1997: 200) claims they are “more accurate
indicators of bone strength and mechanical function than areas alone.” Generally
speaking, the further the bone tissue is distributed from the neutral plane or axis, the
more rigid the bone. Slight increases in external diameter produce large increases in
rigidity; however, outward distribution of bone tissue is not without cost. “Redistribution
of a set amount of material to produce a more rigid structure will not always produce a
stronger structure in terms of overall load that can be sustained. As the material gets
further and further from the centrum and becomes too thin, strength eventually becomes
limited when part of the structure fails by buckling” (Cochran 1982: 151).

The SMA designated by the letter 1 predicts bending rigidity with reference to a
specific neutral plane of bending. Bending rigidity is equal to the product of Young’s
modulus (E) and I. For each cross-section there are as many possible values of I as there
are potential planes of reference, but for simplicity, biomechanical studies focus on four:
Ix, Iy, Imax, and Imin. Ix and Iy represent bending rigidity in the anterior-posterior and
medial-lateral planes, respectively, and reflect the distribution of bone relative to the
medial-lateral (x-axis) and anterior-posterior (y-axis) planes of the bone. In other words, a
cross-section with bone tissue more outwardly distributed in the anterior-posterior plane
(i.e., exhibits a greater anterior—posterior diameter) would have a greater value for lx,
because the bone tissue is located further from the x-axis (medial-lateral plane) (Figure 2-
7). Imax and Imin, known as the principal second moments of area, are measures of
maximum and minimum bending rigidity relative to two empirically derived orthogonal
planes. A bone’s resistance to torsional loading (i.e., loads that produce twisting or

rotation of material about an axis) is predicted by the polar second moment of area,

symbolized by], and is equal to the sum of any two values of 1 relative to two

41

perpendicular j
“an average 1): -:
overall bending
at‘tumte fur er.

l’intrtm‘ski 198

 

 

perpendicular planes (e.g., Ix+Iy or Imax+Imin). Therefore,J is also seen as representing
“an average bending rigidity about all planes through the section. . .a useful measure of
overall bending/ torsional rigidity” (Ruff 1999: 296). The calculated value of} is most
accurate for cross-sections that do not deviate significantly from circularity (Burr and

Piotrowski 1982; Ruff and Hayes l983a).

Posterior

A

\ .

(a)

 

3

Lateral is Medial
Q a?

A

 

 

f‘

 

J

Anterior

Figure 2-7. Second moments of area. (a) Anterior-posterior expanded cross-section;
Ix>Iy. (b) Medial-lateral expanded cross-section; Iy>Ix. (c) Circular cross-section; Ix'—'Iy.

  
 

Rulf an
identifiet'l cross
shape in this its,
ltflt and lmax

Temporal and t

 

femoral midsha

External dial

Some ii
hnuset'l t‘Xt‘ltis‘i
and anterior-[1
Lirsen 1981 .
etternal diapl
\lant‘ studies
addition I() (T
Wescott 211111
ltdnnxered b_\
("21‘ terhm )lt ,gj
if'anstiming a]

[4111“-

Ruff and colleagues (Ruff 1987; Ruff and Hayes l983a; Ruff ct a1. 1984) have
identified cross-sectional “shape” as an important indicator of behavior. Cross-sectional
shape in this usage refers to ratios or indices of second moments of area (SMA), such as
Ix/Iy and Imax/Imin. Ratios approaching 1.0 have a more circular cross-section.
Temporal and geographic differences in these SMA ratios, particularly Ix/Iy of the

femoral midshaft, are seen as reflecting changes in types of activities, such as mobility.

External diaphyseal dimensions

Some of the earliest bioarchaeological studies of long bone. diaphyseal structure
focused exclusively on external diaphyseal dimensions, such as midshaft medial-lateral
and anterior-posterior diameters, rather than cross-sectional geometric properties (e.g.,
Larsen 198]). In these studies, skeletal “robusticity” is typically represented as a ratio of
external diaphyseal dimensions to a measure of overall long bone size, such as length.
Many studies continue to utilize external dimensions in structural analyses, usually in
addition to cross-sectional geometry (Bridges 1989; Bridges et al. 2000; Ruff 1987;
Wescott 2001), primarily because this standard anthropometric technique is less
hampered by fragmentary or poorly preserved remains (Larsen 1997). Moreover, in spite
of technological improvements, calculation of cross-sectional properties is a time-
consuming and expensive procedure (W escott 2001).

Larsen and Rulf(Larsen 1997, 2002; Larsen and Ruff 1991; Ruff 2000; Ruffet a1.
1984) have maintained that external dimensions, while correlated with cross-sectional
properties, are much less precise in terms of reflecting biomechanical strength because
they do not consider the internal architecture of the bone (such as the medullary and

cortical areas). In a bioarchaeology review article Larsen (2002: 135) stated, “measuring

43

the external (117

bone. \s lllt‘h is

 

Hi'mevt‘r. “if.“

 

 

 

rttinnal geniin
liiunierlmniral .

Sfflli.)ll;il am. ”Hi

 

uti1ize external 1

the external dimensions of long bones. . .does not allow the analysis of distribution of
bone, which is the most meaningful attribute of bone strength for inferring lifestyle.”
However, Wescott (2001) has argued that external dimensions accurately predict all cross-
sectional geometric properties, except cortical area, and provide equivalent results to
biomechanical analyses. For the purposes of this dissertation, the critique of using cross-
sectional geometry to reconstruct behavior can be considered to apply to studies that

utilize external diaphyseal dimensions as well.

44

CHAPTER 3
new,

We are too mm '

lutroduction

This ch.

 

 

 

monstrurtiun 1!
{inside a u mip
:mmt‘try. but i

this tllSSt‘l'ldllt >1

Literature l:

A law
truss-sectit )[l a
literature is (l
rmphasi‘lfmu
'{r‘smetry in
llilpl-t‘ (T! )SS—

.11"; l _, ,
ind‘T-J l( “V‘-
V K

M‘Q‘Odc
x:

L

-6“ _
.. 1'7 -
(“an

CHAPTER 3—LITERATURE REVIEW: BlOlVIECHANICAL ANALYSES AND
BEHAVIORAL RECONSTRUCTION IN BIOARCHAEOLOGY

We are too much accustomed to attribute to a single cause that which is the product of several

Justus von Liebig

Introduction

This chapter reviews the anthropological literature pertaining to behavioral
reconstruction from cross-sectional geometric analyses. The chapter is not intended to
provide a comprehensive summary of all anthropological studies of cross-sectional

geometry, but rather focuses on the conclusions of the research that are most relevant to

this dissertation.

Literature review

A large body of anthropological literature deals with the application of long bone
cross-sectional geometry to behavioral reconstruction. For ease of presentation, this
literature is divided into two broad categories. The first category includes articles
emphasizing the methodology and theory behind the application of cross-sectional
geometry in anthropology. The second category includes bioarchaeological studies that
apply cross-sectional geometric analyses to the reconstruction of behavior in

archaeologically derived populations from the United States, Europe, Asia, and Africa.

Methodological and theoretical literature
Since the mid-19705, anthropologists have investigated a variety of issues

pertaining to the biomechanical structure of long bones, including locational variation in

45

the cross-wet it t
19333 . and gr:
19331): Trinkat.
regard to all: in:
trends in hnmir
ohsen'etl variati

aft’llilf‘tdt rgit‘al .

 

research in biun
relationship bet
by physical at‘ti

pmrides the 1') .1

Bone archite

['tili/ir
an 'haeolr )Q’lt‘d‘
tariazit m in ct

Pmmlnf‘d si x

A} .
1.: diaphyfiif‘u

"v.-.
(xi?
t I'
....,

A 3‘15 .
1“)” 0‘ ‘7

13h )xln\"i\
37me“ 1‘! rd U

iii‘hlihY‘I-S‘ l .

”5'”; him l\

.L ,_
4 if? ' V .2“.

t1

the cross-sectional properties of the femur and tibia (Lovejoy ct al. 1976; Ruff and Hayes
l983a), and group variation with regard to sex, age, and side (Ruff 1987; Ruffand Hayes
1983b; Trinkaus et al. 1994). These studies have also examined long bone structure with
regard to allometry (Ruff 1984, 2000; Ruffet al. 1993), ontogeny (Ruff et a1. 1994), and
trends in hominid evolution (Ruffet al. 1993, 1994; Trinkaus et a1. 1994). By interpreting
observed variation in cross-sectional geometric properties of paleontological,
archaeological, and modern samples within a framework of theoretical and experimental
research in biomechanics, Ruff and colleagues have attempted to define a causal
relationship between long bone structure and the mechanical loading patterns generated
by physical activity. Collectively this research has generated a biomechanical model that

provides the basis for “reconstructing” the behavior of archaeological populations.

Bone architecture and mechanical function

Utilizing a large sample of femora and tibiae from the Pecos Pueblo
archaeological population, Ruff and Hayes (1983a) investigated the general pattern of
variation in cross-sectional properties within the lower limb. To accomplish this they
examined six femoral section locations (5 diaphyseal plus the femoral neck), and five tibial
(all diaphyseal). Diaphyseal sections were located by percentage of bone. length, i.e., 20%,
35%, 50%, 65%, and 80%, with 20% most distal, 50% at midshaft, and 80% most
proximal (e.g., the femoral subtrochanteric location). Results of this investigation
demonstrated that bone geometric properties are not uniformly distributed along the
diaphysis, but rather vary by section location. For example, in the femur, [max is oriented
more mediolaterally in the 80%, 65%, and 20% sections, and more anteroposteriorly in

the 50% and 35% section locations. Additionally, by examining the ratio of the principal

46

semntl momet
midshaft. and

Experi:
stresses of the i
uhsen'ed pattr
til cortical hi it ‘t:
liatlitigs. lion
or to relatin-h
elongated e.g._
limrtion relatit

utilerences in 11

 

 

”1’41“ Pflpulati

 

second moments of area (Imax/ 1min), it was found that the femur is most circular at
midshaft, and least circular proximally.

Experimental and theoretical predictions of in viva mechanical loadings and
stresses of the lower limb “generally correspond” (Ruff and Hayes l983a: 379) to the
observed pattern of variation in cross-sectional geometry suggesting that the distribution
of cortical bone within the diaphysis reflects localized adaptation to specific types of
loadings. For example, a more circular section is best adapted to either torsional loading
or to relatively equal bending forces in two perpendicular planes. A bone section that is
elongated (e.g., ovoid) is better adapted to bending in just one plane. This structure-
function relationship, originally proposed by Lovejoy et a1. (1976) as an explanation for
differences in tibial shape between prehistoric Native American populations and modern
urban populations, forms the core of the hypothesis that long bone cross-sectional shape

can be used to infer specific types of activity (Ruff et al. 1984).

Sexual dimorphism and “the mobility index”

Building on the observations discussed above, studies of sexual dimorphism in
cross-sectional properties by Ruff and Hayes (l983b) and Ruff (1987) led to the
development of “the mobility index,” the ratio of anterior-posterior to medial-lateral
bending strength (Ix/Iy) of the femoral midshaft. In analyzing sex differences in Pecos
Pueblo femora and tibiae, Ruff and Hayes (1983b) observed that the lower limb bones of
males are generally more “adapted” to anterior-posterior bending loads, and females
more medial-lateral. This is most apparent when comparing the cross-sectional shape
ratio Ix/Iy, which is greater in males, and especially pronounced around the knee (distal

femur and proximal tibia). Sex differences in this ratio are virtually non-existent for the

47

pruximal fruit;

 

are minimal.
structure lead ‘
Htmevcr. sint .
ohsen'atiun t1 t .
the authors su;
between males
and Hayes l‘Qt:
Ruff 1
lnrlude a Hit )(3
other samples
Slelegx-_ San
Pt‘tpulatirm u
39mg} (limt tr
Molly". .,\(,(

ti
tme‘ (70mm!

proximal femur and distal tibia, where expected in vivo anterior-posterior bending loads
are minimal. To explain these findings the authors propose that sex differences in pelvic
structure lead to increased medial-lateral bending loads in the proximal femur of females.
However, since this anatomical difference in pelvic structure does not account for the
observation that the greatest amount of sexual dimorphism in Ix/Iy is around the knee,
the authors suggested, “a second factor. . .related to a probable behavioral difference
between males and females of the Pecos Pueblo. . .[such as] long distance running” (Ruff
and Hayes 1983b: 394).

Ruff (1987) expanded this study of sexual dimorphism in cross-sectional shape to
include a modern U.S. sample and published diaphyseal shape data from a variety of
other samples, in order to elucidate temporal trends related to changes in subsistence
strategy. Samples were classified as hunter-gatherer, agricultural, or industrial.
Population comparisons suggested that hunter-gatherers show the greatest degree of
sexual dimorphism in Ix/ 1y, industrialists the least, and agriculturalists an intermediate
amount. According to Ruff (1987), this is due to a decrease in mobility among males over

time, combined with reduction in sexual division of labor (sex-specific behavior).

Age and cross-sectional geometry

Cross-sectional properties not only vary between the sexes, but by age as well.
Loss of cortical bone from the endosteal surface with compensatory subperiosteal
apposition has been documented as an age-related process (reviewed in Ruff and Hayes
1983b). Using the Pecos Pueblo archaeological population, Ruff and Hayes (l983b)
examined the effects of age on long bone cross-sectional geometry. In summary, they

found that total subperiosteal area (TA) increased over all age categories in both sexes.

48

Him-eye r. ft ill
menus? in [WI
decline in (IA
malts. The in"
medullary ant
intreased \‘altt

regions of grr '.

femur and 111)l

Banter-alas)

 

Bildtcr
A; ~ ‘ . l
alcnllfin m an
(7f " . ‘ ‘
139”. Rull 2!"
Strength ditfen
ttpttfally later-

the... (title

rfil‘
l ’ . .- -

Blimmf‘tn~

r1: -
(Sn-IT? («ll l)il»u

[e ‘.

flf‘x- '.

me‘ , .
rimmed, 1(,

I .
”Half

.‘I‘al , q

d..\‘\-

, ll‘

9'.
ulP ‘
- Do

1

However, following a peak in cortical area (CA) during the fourth decade, the subsequent
increase in medullary area (MA) outpaced the increase in TA, resulting in an overall
decline in CA during the last decades of life. This effect was greater in females than
males. The net result of this age-related phenomenon is a thinner cortex with greater
medullary and subperiosteal areas. This more outwardly distributed bone also leads to
increased values for second moments of area (SMA). The authors noted sex-specific
regions of greatest increase in SMAs—males increase most in the midshaft region of the

femur and tibia, and females in the proximal femur.

Bilateral asymmetry

Bilateral asymmetry, particularly of the upper limb, has received a great deal of
attention in anthropological studies (e.g., Bridges 1989; Bridges et al. 2000; Fresia et al.
1990; Ruff 2000b; Trinkaus et al. 1994). Individuals exhibit varying amounts of size and
strength difference between the bones of their right and left sides, with the right side
typically larger and stronger in the upper limb (the opposite is typical of the lower limb).
These differences have most often been interpreted in terms of handedness. In addition,
the relative amount of difference between the two sides (i.e., the degree of bilateral
asymmetry) varies among populations. Trinkaus et al. (1994) have argued that the high
degree of bilateral asymmetry in cross-sectional properties exhibited by professional
tennis players and some Neanderthal specimens strongly supports the idea of extreme
developmental plasticity of bone structure in response to life-long differences in
mechanical loading. Consequently, population variation in degree of upper limb bone
bilateral asymmetry is hypothesized to reflect variation in mechanical loading patterns of

the right and left arm (i.e., differential use of the upper limb).

49

 

Ontogeny

Rul
intreasetl n
ultimate pH
properties 1
bone strut ‘t
classic stutl
hilatcral as
Ms found
suhperir is t:
conical an

To
Ruffet a1.
liiflOf.S {13¢
rlllml‘x‘l’ (if
“Wrt‘l‘dtirm
players by
through (‘51
@fill‘lt‘d] (1
[urslt‘ind] S,
liter-ml“?! n

lll

‘ and I...

“rm,
(halllt Iii

Ontogeny

Ruff et al. (1994) evaluated the influence of developmental age on the effects of
increased mechanical loading on the cross-sectional geometry of long bones. \Nhile, the
ultimate purpose of the study was to provide a basis for interpreting the biomechanical
properties of juvenile fossil hominids, the findings are pertinent to any evaluation of long
bone structural adaptation. As part of this study, the authors reanalyzed the data from a
classic study of professional tennis players who were found to exhibit significant humeral
bilateral asymmetry related to side-dominance of the playing arm (jones et a1. 1977). It
was found that the dominant playing arm of tennis players exhibited increased total
subperiosteal area (T A) and decreased medullary area (MA), resulting in increased
cortical area (CA) and torsional strength (D over the non-playing arm.

To examine the effects of age on the development of extreme bilateral asymmetry,
Ruff et al. (1994) performed correlation analyses between TA, CA, and] and three timing
factors (age of subject at the time of the original study, age at which play began, and
number of years played). Only age at which play began showed a statistically significant
correlation with the cross-sectional properties. Further analysis of three of the tennis
players by Ruff et al. (1994) produced the following observations: 1) From childhood
through early adolescence, the primary effect of increased mechanical loading is
periosteal expansion, leading to relatively greater increases in TA, and consequently
torsional strength (I). 2) Following mid-adolescence, the primary response of bone to
increased mechanical loading is conservation of endosteal bone leading to a reduction in
MA, and less dramatic increases in]. An important overall conclusion of this study is that

“mechanical stimuli may have different effects on diaphyseal modeling/ remodeling

50

depending up

mom"? “1

(irttsg-
3159 Vary in n
film. As (,x

trill [hey-(~be ‘

 

 

 

order to imf‘rlJ
sectional P“ "1’
pmptm‘tl (liVl‘
and Hayes 19
hemeen bod}
stintlardizat in
ofallrimetrjx' a:
£993; Ruff 2t n
Ruff 1
litilizing femoi
isimples. as m
that the relatn
S1Mddl‘tllfli‘ li‘ii

1”. brine lenvvrl

f" Crln(ll](i(‘{i

depending upon the age at which they are applied” (ibid: 52).

Allometvy and size standardization

Cross-sectional properties have been shown to vary by sex, age, and side. They
also vary in relation to body size and body shape (Ruff 1984; Ruff et a1. 1993; Ruff
2000a). As explained by Lovejoy et al. (1976: 497), “Two beams of identical cross section
will therefore withstand equal loads only if they are also the same length.” Therefore, in
order to interpret differences in bone strength among different population samples, cross-
sectional properties must be “size-standardized”. Lovejoy et al. (1976) originally
proposed dividing cross-sectional properties by the square of bone length, however, Ruff
and Hayes (1983a) argued for the importance of studying the allometric3 relationships
between body size and shape and long bone structure, prior to adopting a specific size-
standardization factor. Subsequently, Ruff and colleagues have conducted several studies
of allometry and size-standardization on human skeletal samples (Ruff 1984; Ruff et a1.
1993; Ruff 2000b).

Ruff (1984) examined the dependence of cross-sectional properties on bone length
utilizing femora and tibiae from the Pecos Pueblo and Georgia coast archaeological
samples, as well as modern autopsy material. Based on this original study, Ruff suggested
that the relationship of cross-sectional properties to long bone length is isometric“. To
standardize for differences in body size, he proposed that cross-sectional areas be divided
by bone length squared and second moments of area by bone length to the fourth power.

He concluded that, “Significant differences between populations after this standardization

 

3Allometric refers to the change in proportion or shape of parts of an organism during growth and
development.

4 Isometric refers to equality of proportion; in other words, no effect of size on shape.

   
  
  
 

should he intli
study" ihitl: ll
size-standartli,
Ups 3 (liller
Cii'lr rent ht ttlt

R u 1f s

1993; RulfQ‘ H

 

hotly mass and
1993'. '23 (left
within): 1111‘; n 1".
the skeleton (
size” is 1x xly '
relationship 1
Pmpt'Jsetl a n
length Cubed

Ruff

(Tillt‘al impi-

( \

should be indicative of relative differences in mechanical loadings of the bones under
study” (ibid: 356), and proposed four factors that could produce population differences in
size-standardized cross-sectional properties: 1) different activity levels, 2) different activity
types, 3) different body weight/ height ratios or different limb length proportions, and 4)
different body shape (e.g., short and broad versus tall and slim).

Rufl’s (1984) standardization methodology was subsequently revised (Ruff et al.
1993; Ruff 2000b), based on the hypothesis that cross-sectional properties vary relative to
body mass and body shape, rather than simply bone length (i.e., stature). Ruff et a1.
(1993: 25) defined “skeletal ‘robusticity’. . .as the strength or ngirligi ofa structure relatiz'e to the
mechanically relevant measure tfbmly size,” and proposed that for the weight-bearing portion of
the skeleton (e.g., the lower limb bones), the “mechanically relevant measure of body
size” is body mass. Ruff et a1. (1993) rejected Rufl’s (1984) earlier contention that the
relationship between bone length and cross-sectional properties is isometric, and
proposed a new size-standardization method—dividing cross-sectional areas by bone
length cubed and second moments of area by bone length to the 5.33 power.

Ruff (2000b) expanded the findings of Ruff et al. (1993) further emphasizing the
critical importance of controlling for differences in body shape as well as body size in
biomechanical analyses. “[I]n order to distinguish the effects of specific behavioral use of
the limbs on limb bone structure, it is necessary to first account or control for the effects
of both body size and body shape on diaphyseal morphology” (Ruff 2000b: 270). He
added, “it is potentially misleading to use unadjusted bone length alone as a ‘size’
measure against which to compare cross-sectional diaphyseal dimensions. . .raising bone
length to a different power has no effect...” (ibid: 282). Ruff suggested that a “correction

factor for body shape should be incorporated” (ibid: 269) when there are potentially

52

significant t
QWNm'pn)
data. l'nl}
shape (lilli-r
What (‘tingt
lT'ldlt't’l (‘t in
U‘insitlerztti
Stu
Sfflltindl l)
1116 01’3"”;
l‘it‘tjnm m )r
bhdl ”1353'
diff""Tt‘ntll'
ft‘mur 19113
and lyxh S

”Anddrdiz;

Evolution

'fhr
and hitimm
lillt'ren(~~(ix -
mlflts by l

1.
”-0139; 11;“
7.

12’ -

t

significant differences in body shape among groups. In a separate publication (Ruff
2000a) proposed using articular dimensions, such as femoral head size, to standardize the
data. Unfortunately, while strongly arguing for the importance of controlling for body
shape differences between groups, Ruff does not specify the criteria used to determine
what constitutes “signgicant” body shape variation, thereby casting doubt on the behavior-
related conclusions of any study that does not explicitly take body shape into
consideration.

Studies by both Ruff et a1. (1993) and Ruff (2000b) indicated that the cross-
sectional properties of the humerus as well as the femur scale with body mass, in spite of
the observation that, “The upper limb of bipedal hominids, whose primary function is not
locomotor and therefore not weight-bearing, should probably not be scaled strictly to
body mass” (Ruff et al. 1993: 25). Ruff (2000b) also found that the proximal femur scales
differently from the femoral midshaft, being more dependent on pelvic breadth than
femur length. These results suggest that relationship between cross-sectional properties
and body size and shape may be rather complex, and that the appropriate

standardization method may vary by section location.

Evolutionary trends

The apparent existence of long-term evolutionary trends in long bone robusticity
and biomechanical strength has contributed to the theoretical basis for drawing
inferences about behavior from population differences in cross-sectional properties (see
reviews by Bridges 1995 and Ruff 2000a). Cross-sectional geometric properties of long

bones have been reported for a variety of fossil hominid specimens (Lovejoy and Trinkaus

1980; Ruff et al. 1993, 1994; Trinkaus et al. 1994). These studies have demonstrated a

53

 

r.\mr1d\\'l(lt‘ u

 

continuing in‘
“1111111311011 14
physical emit
siphistit‘atit )1 r
in iinjxirtanu

relative to Rt .(

 

ltfiading of tltt‘

Comp
l’ieisttx‘ene sa
.v‘llf'zrm and 11
groups “as si!
tlernnnstratin
magnitude Iti
Circular .Britl
earlier, this \\
Tf‘t‘t‘nl. 1‘10“
NE‘mflf‘f‘thdlr
1t

iii-1’ '
..i this fit

“worldwide trent ” of decreased postcranial robusticity throughout human evolution and
continuing into recent human prehistory. Ruff (2000b: 79) wrote, “The simplest
explanation for this general trend is that as cultural mechanisms for interacting with the
physical environment elaborated during the last two million years (i.e., technological
sophistication increased), biological mechanismr—in this case, bone strength --—decreased
in importance.” Juvenile fossil hominids also exhibit “increased postcranial robusticity
relative to Recent humans” interpreted as being “consistent with increased mechanical
loading of the skeleton throughout life” (Ruff et al. 1994: 53).

Comparisons of degree of sexual dimorphism of femoral Ix/ 1y ratios between Late
Pleistocene samples (archaic Homo sapiens, i.e., Neanderthals, and early modern Homo
sapiens) and Holocene samples, suggests that sexual division of labor in Late Pleistocene
groups was similar to that of more recent hunter-gatherers. Interestingly, though
demonstrating a degree of sexual dimorphism in femoral midshaft shape similar in
magnitude to Holocene hunter-gatherers, the Neanderthal femoral midshaft is more
circular (Bridges 1995; Ruff 2000b). Based on Rull’s mobility index (Ix/1y) described
earlier, this would suggest that Neanderthals were less mobile than groups that are more
recent. However, the authors reason that because it does not seem probable that
Neanderthals lived a more sedentary lifestyle than more recent hunter-gatherers (Bridges
1995), this finding more likely reflects a Neanderthal body shape “quite different from
that of later humans” (Ruff 2000b: 86). It is worth noting that when the existing
preconception regarding a specific type of behavior, namely Neanderthal mobility, did
not fit the biomechanical model (i.e., Ix/Iy as an indicator of mobility), it was the,
biomechanical model that was modified, not the assumption about Neanderthal mobility.

This would seem to call into question the claim that biomechanical analyses yield

54

inlnmtatinn .

 

behatior M'ltt

l)lOfl‘l(‘t"l1.lnl( .

Bioarehaeo]
Numc
Q’t‘tiundl gr‘t‘ ).'

lldt't‘ (x*(‘L1rn.(

 

literature atld‘
from di lle ri n u
ttt‘lahor asst w
resptftnse to l“.
Y‘fllt',1ndl get )I
approach the:
Erlitmetric pr<

”immmm \x‘

’I‘h("\‘(u

aux
int
“Hf _
”lat

information about behavior, and instead suggests that existing hypotheses regarding

behavior (whether supported by evidence or based on conjecture) are used to interpret

biomechanical data.

Bioarehaeological studies of cross-sectional geometry

Numerous anthropological studies have interpreted population variation in cross-
sectional geometric properties in the context of changes in human adaptive strategies that
have occurred throughout the course of human cultural evolution. Specifically, this
literature addresses changes in long bone structure presumed to have resulted directly
from differing patterns of activities (level and type of physical activity) and sexual division
of labor associated with transitions in subsistence and mobility strategies, as well as, in
response to Eumpean contact. Research has also looked at population variation in cross-
sectional geometry in relation to the physical environment (e.g., terrain). Anthropologists
approach these issues by quantifying variation in external dimensions and cross-sectional
geometric properties, as well as, changes in degree of sexual dimorphism and bilateral
asymmetry within, between, and among sample populations.

These studies have been divided into two broad categories based on geography:
bioarchaeological studies of North American populations and bioarchaeological studies of
non-North American populations. For each study, the results and conclusions most
germane to this dissertation are presented. A major thesis of this dissertation is that the
data from many of these studies have been over-interpreted (see chapter 5). In other
words, the authors’ conclusions are frequently based on statistically non-significant group
differences and non-significant “trends” in sample means. Therefore, it should be noted

that most of the findings summarized here are taken directly from the authors’ data

tables. and in

findings that
inrlurled in tl
lienteen the .
impact of illlt
entital anal) .
(“UntTlUSltms rt

1311311011 in el

 

 

Bioarehaeol
By far
skeletal Cullet‘

mlr’m and v

lit tlit

been cont l tit“
freesia et al.
antllarsen l
t«est—central I
llit'hifqan an:
The r

flilehg‘St‘d] ‘
~\ .1

hmi‘l'd or

tables, and not from the written “results” portion of each publication. Only those
findings that are statistically significant (p < 0.05) or near-significant (p < 0.10) are
included in this review. This is done to facilitate an evaluation of the relationship
between the actual data and the conclusions reached by the authors, and to minimize the
impact of interpretive bias. In keeping with the main focus of this dissertation, which is a
critical analysis of the application of biomechanical analyses to behavioral interpretations,
conclusions related to changes in cross-sectional geometric properties (as opposed to

variation in external dimensions) are emphasized.

Bioarehaeology of North American populations
By far, the majority of research has been conducted on archaeologically derived
skeletal collections representing a variety of prehistoric and early historic groups from the

eastern and western United States.

Eastern United States

In the eastern United States, studies of variation in cross-sectional geometry have
been conducted on populations primarily from three main regions, the southeast coast
(Freesia et al. 1990; Larsen 1981; Larsen and Ruff 1991; Larsen et a1. 1996, 2001; Ruff
and Larsen 1990, 2001; Ruff et al. 1984), northwest Alabama (Bridges 1989, 1991), and
west-central Illinois (Bridges et al. 2000). A single study has examined populations from
Michigan and New York (Barondess 1998).

The most comprehensive investigation of temporal changes in long bone
diaphyseal structure has been conducted on samples from a geographic region known as

La Florida or Spanish Florida (Larsen et al. 2001), which includes archaeological sites

 

lmm tht‘ G"
on the must
roast IAIN)”
includml sari
stud) beltii‘ 1'
This resetm l
biomechanit
predgrnvultut
[arse
femora. tibia'
suumuralad.
agriculturalis‘
diaphyseal Si/
magnitude oi
Ufslzeletal si/r
tFtlUtititirt fliti
While (leqrati
for this reduu

(‘U‘momir ft x 1

FUllt w.

f: r n
“”293 m er.

d'3T1ti'ultu rat I)

aflep ~
~ ”659' ‘
» In 1
fail“, .

 

from the Georgia coast and northern Florida. While the earliest of these studies focused
on the transition from hunting and gathering to maize agriculture along the Georgia
coast (Larsen 1981; Ruff et al. 1984), this long-term research project subsequently
included samples representing populations from the Spanish mission period in order to
study behavioral changes associated with European contact (e.g., Ruff and Larsen 1990).
This research examined temporal changes in long bone size (external dimensions) and
biomechanical properties of femora and humeri from males and females of precontact
preagricultural, precontact agricultural, and postcontact agricultural periods.

Larsen’s (1981) anthropometric study of external diaphyseal dimensions of
femora, tibiae, and humeri laid the groundwork for subsequent research on long bone
structural adaptation in this geographic region. Comparisons of hunter-gatherers to
agriculturalists demonstrated that long bone dimensions, including measures of
diaphyseal size and long bone length, decreased over time in both sexes, but that the
magnitude of the changes was greater in females. Larsen (ibid: 498) concluded, “Analysis
of skeletal size on the prehistoric Georgia coast presents us with a trend of skeletal size
reduction that is probably associated with the adoption of an agricultural lifeway. . ..
While degradation of nutrition may have been a factor, it seems most likely that the cause
for this reduction is centered on change in degree of functional demand with the shift in
economic focus.”

Following this original analysis, Ruff et al. (1984) conducted the first study of
changes in cross-sectional geometric properties of the same hunter-gatherer and
agricultural populations, focusing on the femur. Because Larsen (1981) had documented
a decrease in long bone length for this region, the authors “standardized” the data to

facilitate interpretation of group comparisons.

57

firm

 

analyses 01-91
statistically si
regit’in. 1n n
in the agriru
identified for
signifnanth
TA “ for l)(
agriculturali-
sulitrtx‘hantt
“All titflltt‘x‘t'
rftlur'tion in
Emphasis at
Sub:
tXpand tin t
p'ilpUlflllt'1ng
repomxd 11]:

three tempv

pilsl‘f‘fjnlar

Ruf-
It;

i
ll prf‘Vlr

significant (‘

Group differences were determined by ANOVA and t~tests. Results of statistical
analyses of size-standardized cross-sectional properties demonstrated relatively few
statistically significant differences between sample means, particularly in the midshaft
region. In males at midshaft, only medullary area (MA) and Ix/Iy were significantly lower
in the agricultural group. In contrast, a greater number of significant changes were
identified for the subtrochanteric section; MA, TA, Imax, Imin,J, and Imax/Imin were all
significantly reduced in male agriculturalists. In females, only total subperiosteal area
(TA)—-for both the midshaft and subtrochanteric sections—was significantly lower among
agriculturalists. Female differences in midshaft and subtrochanteric MA and
subtrochanteric torsional strength (I) reached near-significant levels. The authors claim,
“All of these changes in the distribution of bone tissue within cross sections strongly suggest a
reduction in mechanical loading of the femur in the agricultural group” (ibid: 131,
emphasis added).

Subsequent studies of the Georgia coast archaeological populations continued to
expand on these two studies ultimately including samples from post-European contact
populations, and adding the humerus to the structural analysis. Ruff and Larsen (1990)
reported the results of the first biomechanical study to include femora and humeri from
three temporal/ cultural periods (precontact pre-agricultural, precontact agricultural, and
post-contact). This study also expanded on the results of Ruff et al.’s (1984) comparison
of the two precontact populations.

Ruff and Larsen (1990) demonstrated that the statistically significant decrease in
Ix/Iy previously reported for male agriculturalists (Ruff et al. 1984) was due to a near-
significant decrease in Ix. In addition to the decrease in femoral midshaft TA

documented by Ruff et a1. (1984), females also showed significant decreases in both Ix and

58

l\'- Bmu‘cn
in 1mm. 1‘ it.
humeri shim

Bet“
medullary at
significant w
pmpt't‘l}. 1111'
perirxls. lnt
significantly

Cont.
femoral crud
suhtnxhante
the contact 1 )

the precon tat

 

Pmpfny of 11
significant int

Sexual
at all Section 1
sectional shat
Sexual dimr nr
5111 reached r.
SlQTIlllt‘allll}‘ d
wriunal shal

13””.th

Iy. Between the two precontact p0pulations, male humeri showed significant reductions
in Imin, Iy, and], and near-significant reductions in TA, Imax, and Ix; whereas, female
humeri showed no signyfcant dg'flerences.

Between the pre-eontact agricultural and contact populations, an increase in
medullary area (MA) was the only change seen in the femoral midshaft of males. No
significant (or near-significant) change in any other size-standardized midshaft geometric
property, including cross—sectional shape, was observed between the males of these two
periods. In contrast, all size-standardized properties except cortical area (CA) were
significantly larger at the femoral subtrochanteric location in contact period males.

Contact period females exhibited significant increases in all size-standardized
femoral cross-sectional properties except midshaft and subtrochanteric MA and
subtrochanteric Imax. Female subtrochanteric Imax/Imin showed a significant decrease in
the contact period. No signyiamt, or even near-significant, differences were found between
the precontact agricultural and contact periods for any size-standardized cross-sectional
property of the humerus for either sex. The humeral shape ratio Ix/Iy showed a near-
significant increase in males following contact.

Sexual dimorphism was found to be significant for most cross-sectional properties,
at all section locations, and in all temporal periods. The primary exception was cross-
sectional shape ratios, for which male-female differences were generally not signyimnt.
Sexual dimorphism of femoral midshaft Ix/Iy was not significant for any temporal period,
but reached near-significance in the contact period. Subtrochanterie Imax/ 1min was
significantly dimorphic only in the precontact agricultural period, and humeral cross-
seetional shape was significantly sexually dimorphic only in the precontact preagricultural

period.

 

The
conclusions.
in the femur
period. tht‘l‘
reiterated a
sex. was a «l
near-sigmifit
midshaft 1\
disrusx‘tl 11
Hill) 5 of l
in midshaft
ftsults indi
111d incrm;
males and
112. Th“?
Contact )X‘T
pan of I)“.
.‘Ulgtlrm‘hdr
5’72““! zit‘th
hem-Pm] fl
("*‘ima‘..[. Ii
pinlid‘ Slit
”litre

95.11):

f ‘
l”.(-I_lll§\l.\‘lf‘I’

The authors discussed their findings at some length, and reached the following
conclusions. 1) “1n the Georgia coast samples studied here, females decline through time
in the femoral midshaft Ix/Iy ratio, while males first decline in the precontact agricultural
period, then increase in the contact period” (Ruff and Larsen 1990: 1 10). It should be
reiterated at this point that the only statistically significant change in this ratio, for either
sex, was a decrease in males between the two precontact periods. No other significant or
near-significant difference was observed. In fact, the one-way AN OVA of femoral
midshaft Ix/Iy among periods within sex was not statistically significant. The authors also
discussed the finding that among contact period males the 1x/ 1y distribution is bimodal,
with 5 of l 1 males showing an extremely high mean value for this ratio. These “changes”
in midshaft cross-sectional shape were interpreted in terms of mobility, “Thus, these
results indicate that among some of the males in the contact period, long-distance travel
had increased greatly from the average levels documented prior to contact, while in other
males and all females, long-distance travel either stayed about the same or declined” (ibid:
1 12). They suggested that this is consistent with historic records documenting that some
contact period males were forced by the Spanish “to make periodic long-distance trips” as
part of the “repartimiento” labor system. 2) The authors suggested that the decrease in
subtrochanteric Imax/Imin over time in males and females “is indicative of a decrease in
general activity level” (ibid: l 13). In males, the significant decrease in Imax/Imin occurred
between the two precontact periods, with only a near-significant decrease following
contact. In females, the decrease occurred following contact with the two precontact
periods showing almost identical values. 3) The authors noted that an hypothesis of
decreasing general activity levels over time (based on subtrochanteric Imax/ 1min) seems

inconsistent with the finding that femora of contact period males and females show

60

increases in «
fiilluts'ing (‘\ I;
missionizetl t
to increasul -
rarfmhytlratt

the length-51.1

lUWtr at'tix'itt

 

 

 

hftT’JUSt’ weft;

regard to the

humeri in th<
exception. I

strength relat
contact perit.
in any other

hypothesized
missiortized t
olmale hum
the contact I:
Rt‘iults of tftt
‘Wilitfltfl. a
‘lmilit“artt di
94ml fem

ale (1

11") T
' h

13 is

increases in cross-sectional areas and second moments of area. They offered the
following explanation: “it is plausible that body weight/ stature increased in the
missionized Guale relative to previous or nonmissionized contemporary populations due
to increased sedentism. . .confinement. . .and increased consumption of

carbohydrates. . .This in turn would increase the body weight/ bone length ratio, and thus
the length-standardized cross-sectional properties in the lower limb. . .despite a generally
lower activity level” (ibid: 1 15-1 16). Unfortunately, it is impossible to test this hypothesis
because weight (i.e., body mass) cannot be determined from skeletal remains. 4) With
regard to the upper limb the authors stated, “geometric properties of the femora and
humeri in the present study sample tend to follow similar temporal trends, with one major
exception. Unlike femora (or male humeri), female humeri continue to decrease in
strength relative to bone length in the contact period. Thus, it appears that females in the
contact period were placing relatively lower mechanical loads on their upper limbs than
in any other period” (ibid: 1 16). This was interpreted as being “consistent with the
hypothesized changes in general activity level and relative body weight among the
missionized Guale” (ibid: 1 17). The authors also claimed that the biomechanical analysis
of male humeri suggests that males became “more involved in agricultural tasks during
the contact period” (ibid: 1 17). These interpretations are not supported by the data.
Results of the one-way AN OVA among periods within sex for the humerus were not
significant, and t-tests for precontact agricultural-contact period comparisons showed no
significant differences for either sex. In their final comments the authors asserted, “Male
and female differences in use of the upper limb continued to decline [in the contact
period], possibly reflecting more male participation in agricultural responsibilities” (ibid:

120). This is in direct conflict with the data, which show sexual dimorphism of humeral

61

Properties at
mnmut9‘
[wag-if“ 11 [ll
1
pawl" ,1l)ftli

lit a t

temporal ("11.x

humeral lent"

 

ofpertent sitl
near similicat
shape lx/lt’. V
wise com paris
differences in

Sift‘tiottal area
pmmntact pc
second momt‘
(ltifrf‘e ofbilal
liUt decreases

Ll}? m0 prer‘rn

L.
vtmterat stem

immnaanu
St'qtllllt
PVal.

illtl'

properties at its maximum during the contact period, a point the authors noted earlier in
the paper, “sexual dimorphism in humeral properties first declines from precontact
preagricultural to precontact agricultural, then increases to its largest values in the contact
period” (ibid: 108).

In a companion study to Ruff and Larsen (1990), Fresia et al. (1990) examined
temporal changes in bilateral asymmetry (calculated as percent side difference) of
humeral length and mid-distal humeral cross-sectional properties. A one-way ANOVA
of percent side difference among groups within sex was significant for TA, Ix, Iy, and];
near significant for length; and not significant for CA, MA, or humeral cross-sectional
shape Ix/Iy. This study did not include post-hoe t-tests (e.g., Tukey HSD test) for pair-
wise comparison of group means, so it is unclear which periods showed significant
differences in the above properties. However, examining the data reveals that for cross-
sectional areas, bilateral asymmetry decreased most dramatically between the two
precontact periods then remained unchanged into the contact period for both sexes. For
second moments of area (SMA), the results differ somewhat between the sexes. In males,
degree of bilateral asymmetry of Ix and] is similar between the two precontact periods,
but decreases in the contact period, whereas bilateral asymmetry of Iy decreases between
the two precontact periods, and does not change in the contact period. In females,
bilateral asymmetry of all three SMAs decreased dramatically with the transition from
hunting and gathering to agriculture, but was unchanged following Spanish contact.

Significant differences between right and left sides (i.e., bilateral asymmetry) were
evaluated by t-tests for paired comparisons. For statistically significant side differences,
total subperiosteal area (TA) and cortical area (CA) were found to be typically greater on

the right side with medullary area (MA) greater on the left side in both sexes. Very few

62

mmmh

males. no it
tnhersvnrds
'Enengthf'
tlurint: the p

essentially~ st

 

 

 

 

 

1n 5“
Percent sitlt‘
between Ll“
ltilllou'itlLI (if:
pn‘rttllldll l
traflSlLl‘m 1‘

Iht
mfihanit‘d
“Compilrl‘
limh...(th‘d
mfimah
mater (1‘75
asymmetfj
exhibit grr
side" iliitl.
‘li'TtlliCdnl
orni'agrit ul

statistically significant side differences in second moments of area were detected. In
males, no right-left SMA comparisons reached the level of statistical significance. In
other words, males of all three periods were nearly symmetrical in terms of humeral
“strength.” For females, the right humerus was significantly stronger than the left only
during the precontact preagricultural period. Female right and left humeri were
essentially symmetrical during the other two periods.

In summary, male humeri were virtually symmetrical in all periods, but the mean
percent side difference for some properties changed over time; TA asymmetry decreased
between the two precontact periods and second moments of area asymmetry decreased
following contact. Female humeri were significantly asymmetric only during the
precontact preagricultural period, and became bilaterally symmetrical coincident with the
transition to agriculture.

The authors asserted, “changes in bilateral asymmetry. . .reflect alterations in
mechanical loadings, and thus activity patterns” (ibid: 121). They concluded,
“Comparisons of the temporal periods reveal a decline in asymmetry of the upper
limb...[that] reflects activity patterns involving change in the use of the upper limbs in the
shift to a lifeway that involves agriculture. . . [and] females were affected to a relatively
greater degree in the shift to agriculture than males” (ibid: 131). With regard to bilateral
asymmetry of humeral shape the authors asserted, “females of the precontact periods
exhibit greater anterior-posterior bending strength on the left side, males. . .on the right
side” (ibid: 130). However, results of paired t-tests for side differences in shape showed no
significant asymmetry in shape for females of any period, or for males in the precontact
preagricultural and contact periods (only male precontact agriculturalists showed

significant shape asymmetry). In fact, for anterior-posterior bending rigidity (Ix)

63

Spt‘flllt‘all.‘
pnsmntat‘l
significant
,t\'()\'.~\:
Ct
Larsen (‘1
sample: If
t‘t‘tmtit‘l QT
period I)”
Statistical
restricted
lyiitlt mid
total sulip
early and
Stud} the ,
females. 21
limbs. . lil,
Slldnlfih ir‘
Contact P‘
hate mag

Wiles"

I.‘
~3I~tleath (,1

[are

”it Al I];

specifically, females showed statistically significant side dominance only during the
precontact preagricultural period, where the right side (not the left) was greater, and no
significant asymmetry for the other two periods. In addition, as noted above, one-way
ANOVA among periods within sex was not significant for humeral shape.

Continuing the trend of building on earlier research of Georgia coast populations
Larsen et al. (1996) expanded their study with the addition of a late contact period
sample; therefore, in this study the contact period is represented by early contact and late
contact groups. Results of comparisons among the two precontact and early contact
period populations are consistent with those already reviewed (Ruff and Larsen 1990).
Statistically significant differences between the early and late contaetperiods were
restricted to females, who experienced an increase in medullary area (MA) of the femur
(both midshaft and subtrochanteric sections) and humerus, and an increase in humeral
total subperiosteal area (TA). No other statistically significant differences between the
early and late contact periods were detected. Nevertheless, based on the results of this
study the authors stated, “In both Early and Late Contact males and Late Contact
females, an increase in bone strength in the humerus suggests increased use of their upper
limbs. . .likely related to the increase in demands placed on these populations by the
Spanish in labor-related projects. . .that structural properties increase during the Late
Contact period in both females and males suggests that during this time both sexes may
have engaged in similar types of activities, or at least activities involving similar loading
modes” (Larsen et a1. 1996: 1 13). The authors acknowledged that a higher average age-
at-death of the late contact sample could have contributed to observed findings, but
maintained that, “the increases in second moments of area during the Contact period

[are most likely explained by] behavioral factors” (ibid: l 15). Neither sex showed a

64

statistit‘all)

1A and .\1

 

mean 39“ "
I.
Rut (

bone strutt ‘l

Huritla. ltt

 

flot‘itla. rep
Ruffand 1..
ftt’“hl$tt trlt‘
hlission Ya
that some (
Larsen 19‘.
These Sltlt

31.11993 1

statistically significant increase in second moments of area. The significant increases in
TA and MA observed in late contact period females are consistent with an increase in
mean age of this sample, as discussed earlier in this chapter (Ruff and Hayes 1983b).
Ruff and Larsen (2001; see also Larsen et al. 2001) revamped their study of long
bone structural adaptation on the southeast coast to include samples from northern
Florida. In both studies, six samples were compared, three from Georgia and three from
Florida, representing different subsistence strategies and temporal periods. Following
Ruff and Larsen (2001) they are referred to as Early Prehistoric Guale (EPG), Late
Prehistoric Guale (LPG), Early Mission Guale (EMG), Late Mission Guale (LMG), Early
Mission Yamasee (EMY), and Early Mission Timucua (EMT). The authors (ibid.) noted
that some of the samples previously categorized as preagricultural (i.e., EPG) (Ruff and
Larsen 1990) were, in the current study, included with the LPG, a horticultural group.
These studies also employed a different size-standardization method—following Ruff et

\ I

al. (1993) cross-sectional areas are divided by (bone length)3 and SMAs by (bone
length)5-33.

As compared to previous studies, the number of cross-sectional variables analyzed
was reduced to CA and] for the femur (subtrochanteric region and midshaft) and
humerus, Ix/Iy for the femoral midshaft and humerus, and Imax/ 1min for the femoral
subtrochanteric region (Ruff and Larsen 2001). Results of a two-way ANOVA by sex
and group (with age included as a covariate) revealed statistically significant differences
among groups in femoral midshaftJ and Ix/Iy, femoral subtrochanteric] and Imax/Imin,
and humeral CA and]. The following statistically significant differences between

temporally contiguous groups were found. Between the EPG and LPG femoral

 

sulnrt‘u‘lnu:

EMU/EM

sigmilit‘am I
No dilli‘n‘r;
analysis.

On-

 

unlil‘r the nl
significant c
Lilli grt Kl)
LNG. whit
intemretali
Larsen 199
Mission \i;
sample is”
Ihan the n
mug). . (1)
in [le Yd!)
P’F‘Up}? u h
3011p (i011
SUlHrrK‘hm

dblr‘k‘h '1‘

subtrochantericJ, as well as humeral CA andJ decreased. Between the LPG and
EMG/EMY groups femoral subtrochanteric lmax/ 1min decreased. No statistically
significant changes occurred between early mission EMG and late mission LMG groups.
No differences among the Early Mission groups5 (EMG, EMY) were detected in the
analysis.

Overall, these findings are consistent with those already presented. However,
unlike the results reported in Ruff and Larsen (1990) this study failed to detect any
significant changes in the femoral midshaft Ix/Iy between the two precontact (EPG and
LPG) groups (the only significant group difference in this ratio is between the EPG and
LMG, which are not temporally or culturally contiguous groups). Behavioral
interpretations do not deviate from those already reported (Larsen et al. 1996; Ruff and
Larsen 1990) with respect to the EPG, LPG, EMG, and LMG. With regard to the Early
Mission Yamasee, the authors proposed the following: “The Yamasee early mission
sample is. . .in many respects. . .more similar to the late pre-mission Guale group (LPG)
than the mission period Guale or is intermediate between this group and the EMG
group...[i]n particular. . .variation among males in long-range mobility [is] not increased
in the Yamasee” (Ruff and Larsen 2001: l38). They asserted that this is suggestive of a
people who were “less acculturated during the mission period.” Results of pair-wise
group comparisons show that the EMY significantly differ from the EPG in
subtrochanteric [max/Imin, and humeral CA and], and they differ from the LPG in

subtrochanteric Imax/Imin. The EMY do not differ significantly for any property from

 

5 Ruff and Larsen’s (2001) interpretations of the EMT will not be presented in this dissertation because, by
the authors’ own admission, “Results for the Timucuan ossuary sample must be treated cautiously, since the
exact sex composition of this sample is unknown, and sex was seen to be an important influence on
biomechanical properties in the other samples“ (p. 139).

66

 

[llt‘ Ehlfi \
index. the
would seen
groups Sllll l
Ll’fi is ha.»
similar in n

Brill

 

 

 

dimensium
from huntiu
research “0‘
it‘lh'ltles l)t‘
on ethm igm
subsistent'c
increase in l
that the cha
must of the
The
mexrties i1
Panunt‘ed
Film-“idly

.tr- CONT
.‘e. ,‘
’ h

_ ”if.
“'7: t

l.
,l(1r(f

the EMG or LMG. With respect to group means for femoral midshaft Ix/ Iy (the mobility
index), the EMY do not differ significantly from any of the other groups. Therefore it
would seem that, statistically, the EMY are, in most respects, similar to all of the other
groups studied. In fact, the authors’ interpretation of similarity between the EMY and
LPG is based primarily on the coeflicient (fziariation5 for midshaft Ix/Iy in males, which is
similar in magnitude between the two groups.

Bridges (1989, 1991) studied temporal changes in upper and lower limb external
dimensions and cross-sectional geometry (CA, Imax, Imin,J) associated with the transition
from hunting and gathering to maize agriculture in northwestern Alabama. This
research “examine [d] evidence for changes in both the level and the types of subsistence
activities between the Archaic and Mississippian time periods” (Bridges 1991: 89). Based
on ethnographic and ethnohistoric literature, Bridges (1989, l991) hypothesized that the
subsistence transitiOn would have entailed a change in the types of activities and an
increase in the level of activity for Mississippian populations. In addition, she predicted
that the changes would be more pronounced in females because they typically take on
most of the responsibilities associated with food production in agricultural societies.

The results (Bridges 1989, 1991) revealed an increase in femoral cross-sectional
properties in the Mississippian population for both sexes, but with the changes more
pronounced and uniform in male femora. Size-standardized humeral dimensions (i.e.,
robusticity) and cross-sectional properties were not different between hunter-gatherer and
agricultural males. In females, robusticity of both humeri increased in the Mississippian

period, as did biomechanical “strength” of the left mid-distal humerus (cross-sectional

 

6 The coefficient of variation compares the variability between samples with different means. It is the ratio
of the standard deviation to the group mean.

 

Prtipcrtl“S '
Winnie"? I
bilateral “Si
(Llfllle ”it“

Brad

sut’t’t‘Sl 3 Chi

 

 

 

ln leinales. I.
Hi side mm
mm in ll)? l
males. the (l
adoptinn UT
strength in t
Bridges attr
artixities. . .i
and possibl}
preparation
maize attrit-
‘llllt‘rent‘e ii
is attributed

Britl

:‘entral lllinr

rt’r-mal din

iii: 31.3th (‘n

.i.aens1lit“atir-

properties of right humeri were not studied). Both sexes exhibited decreased bilateral
asymmetry in humeral robusticity in the Mississippian sample (no data is available for
bilateral asymmetry of humeral cross-sectional properties), attributed to more equal usage
of the right and left arms.

Bridges (1989: 391) states, “Differing patterns of changes in males and females
suggest a change in the division of labor coinciding with the shift to maize agriculture.”
In females, the combination of increased humeral size and strength, particularly on the
left side near the elbow, and decreased bilateral asymmetry is attributed to “pounding
corn in the traditional manner of southeastern Indian women” (Bridges 1991: 98). For
males, the decrease in humeral bilateral asymmetry was tentatively attributed to the
adoption of the bow and arrow (Bridges 1989, 1991). The increase in femoral size and
strength in males is referred to as “an unexpected result” (Bridges 1991: 98), which
Bridges attributed, not to “long-distance running”, but rather to “a variety of
activities. . .includ[ing] long-distance raiding and hunting, agricultural or building chores,
and possibly increasing involvement in the ball game, a vigorous sport regarded as
preparation for warfare...” (ibid: 99). Bridges (1991: 99) concluded, “the adoption of
maize agriculture involved a major increase in workload” in northwestern Alabama. The
difference in results between this region and the Georgia and Florida coastal populations
is attributed to regional variation in practices associated with the adoption of agriculture.

Bridges et al. (2000) applied a similar research design to populations from west-
central Illinois to examine temporal changes in long bone diaphyseal structure (both
external dimensions and cross-sectional geometry) of upper and lower limbs. However,
this study emphasized skeletal adaptation associated with increasing “horticultural

intensification” in order to “link activity levels with specific subsistence economies, rather

68

 

than with s
samples w

ll'ntx'llantl

 

 

 

ll'mtllantl
\thllantl 3
Mississippi.

Cm

 

lmzts/ lmin-

dillcrf‘nf‘is

statisticalh

the later I.

humeri. on

but only (it

pri;iix*rties .

ll'mlland

llixitlland

statis'ticall}~

L116 IVOtitl
the Middle
L1H) Late 1
lf‘mftles '
t. \\ l1]

lt’ltl Later L.
‘1 h

I";
dl ln‘f‘f‘f‘d‘
. if

than with somewhat simplistic categories such as ‘agriculturalists’” (ibid: 218). The
samples were derived from populations representing four subsistence stages: 1) Middle
Woodlandh-“low-level horticulturalists relying on native seeds”, 2) Early Late
Woodland—“intensive horticulturalists using the same crops,” 3) Later Late
Woodland—intensive horticulturalists following the initial introduction of maize, and 4)
Mississippian— “intensive maize agriculturalists” (ibid: 218).

Cross-sectional properties examined in this study were CA, Imax, Imin,J, and
Imax/Imin. Results of the analysis of male femora and humeri demonstrated no significant
differences in cross-sectional properties among the time-periods examined. The only
statistically significant findings for female femora were greater midshaft CA, Imin, and] in
the Later Late Woodland period as compared to the Middle Woodland period. Female
humeri, on the other hand, showed several significant differences between time-periods,
but only on the left side (the right side showed no differences). The cross-sectional
properties of the left humeri of Later Late Woodland females were greater than Middle
Woodland females. Mississippian females showed a subsequent decline from both Late
Woodland periods in the geometric properties of their left humeri, a reduction to levels
statistically indistinguishable from those of the Middle Woodland period. Only one Early
Late Woodland variable, left midshaft Imax, showed a statistically significant increase over
the Middle Woodland period. In other words, Later Late Woodland females (but not
Early Late Woodland females) had “stronger” left humeri than Middle Woodland
females, while Mississippian females had left humeri that were weaker than both Early
and Later Late Woodland females, but similar in strength to Middle Woodland females.

These results suggest that there was an increase in left humeral strength in females

that preceded the adoption of intensive maize agriculture as the primary mode of

69

 

gulisistcnt‘t‘
llntxlluntl [

was unanllt

 

mm. in ll‘n
Pmessing.
continUt‘ I“

requires “‘11

 

Linn guenull
making in li
for this sub:

'lillt'
male and ti
suggesting
conducted
crops. and
archaeolo-
llr'aely re:
\lississ'im

B.
"(mum in
T‘Wli Sta
(hang-3 .

qln‘lf‘ngi‘

subsistence in Mississippian times, i.e., it occurred at some point during the Late
Woodland period. The subsequent decrease in left arm strength of Mississippian females
was unanticipated by the authorsF-“The increasing reliance on starchy seeds, especially
maize, in the Mississippian period would be expected to coincide with a greater need for
processing, a chore traditionally carried out by females. If so, female arm strength should
continue to increase in the Mississippian period. That it did not, but actually decreases,
requires rethinking this hypothesis” (ibid: 232). The authors suggest the decrease in left
arm strength might reflect a technological improvement in corn processing, such as
soaking in lye or boiling, which ultimately led to reduction in the physical labor necessary
for this subsistence chore.

The results were also interpreted in terms of sexual division of labor. “Overall,
male and female changes in strength occur largely independently of each other,
suggesting that their roles in society differed as well. . .Given that females historically
conducted the majority of agricultural tasks, including both growing and processing
crops, and that variation in female strength in this study fits well with the observed
archaeological sequence of subsistence change, it can be assumed that females were also
largely responsible for growing crops in the Woodland period as well as later in
Mississippian times” (ibid: 235).

Barondess (1998) evaluated humeri and femora from prehistoric and historic
groups in Michigan, and from hunter-gatherer and agricultural groups in western New
York State to detect structural differences “that may have resulted from concomitant
changes in physical activity” (ibid: 93). This study analyzed both external diaphyseal
dimensions and biomechanical strength (cross-sectional properties).

In comparisons of Michigan prehistoric and historic samples, statistical analyses of

70

 

female in
pri‘iptTtit ‘s
exhibited ‘
statistic all“

.‘

mhlshalt r
in both l.

Based on t

 

 

 

 

sectional p
llmwx'er.
lll‘ations. ;
Blt
“ere ”Wm.
Statisticalh
lllt‘ null ll\
for [hp l)”
null hl‘Pnt

midshaft l

female femora demonstrated a statistically significant increase in many cross-sectional
properties of both the proximal and distal femur, but not the midshaft. Male femora
exhibited significant increases only in midshaft and distal MA and TA. Most of the
statistically significant differences between female humeral samples were localized to the
midshaft region. Male humeri showed few significant differences, confined to an increase
in both TA and torsional strength of the distal humerus, and TA only of the midshaft.
Based on these findings, Barondess accepted the null hypothesis (no difference in cross-
sectional properties) for male femora, and both male and female proximal humeri.
However, he rejected the null hypothesis for female femora at proximal and distal section
locations, and for male and female humeral distal and midshaft locations.

Biomechanical data from New York hunter-gatherer and agricultural samples
were restricted to male femora and female humeri. Group comparisons revealed few
statistically significant differences for any biomechanical variable. Barondess accepted
the null hypothesis for male femora. For female humeri, he accepted the null hypothesis
for the proximal humerus, where no significant differences were found, but rejected the
null hypothesis for the distal humerus, where only MA increased significantly, and for the
midshaft humerus where MA increased and CA decreased.

According to Barondess (1998: l 10), “Interpreted within the context of physical
activity, the increases in the cross-sectional area and strength measures provide evidence
for increased biomechanical demand, for both sexes, in the Michigan historic period.”
He observed that the magnitude of the changes was greater for females, and suggested,
“the level of activity changed more dramatically for females than for males between the
two periods” (ibid: 1 10-1 1 l). Barondess concluded that there is no biomechanical

evidence for a change in workload with the adoption of agriculture in New York.

71

 

Western 1

 

populatii (
Plains.

Si
sample is.

1983b:.

 

(‘llitplt‘r (lt
hare pri )\'
limit] the l
Ruffet 3L
Pm’iflncu
Ytprese n t
“grlt‘ulltn

the midsl

ratir'is I",
leSS run“
”Ch as li
al’allalil...

“hllp I}](.

”We in

fl

Western United States

Variation in cross-sectional geometry has also been investigated in archaeological
populations of the western United States including the Southwest, Great Basin, and Great
Plains.

Some of the earliest work on cross-sectional geometry utilizing an archaeological
sample was performed on the Pecos Pueblo of New Mexico (Ruff and Hayes l983a,
1983b). The major findings of this work are presented in the previous section of this
chapter dealing with methodological and theoretical issues. Studies of the Pecos Pueblo
have provided a foundation for all subsequent research in this area. In addition, the data
from the Pecos Pueblo sample has served as a basis for several regional comparisons (e.g.,
Ruff et al. 1984; Larsen and Ruff 1995). Ruff et al. (1984) compared Georgia coast
preagricultural and agricultural femoral samples with the Pecos Pueblo sample, which
represents an agricultural adaptation roughly contemporaneous with the Georgia
agricultural sample. They found that for most cross-sectional properties, particularly of
the midshaft, the Pecos Pueblo femora exhibited strength and robusticity more similar to
the Georgia preagricultural sample, and greater than the Georgia agricultural group.
The most significant exception to this was cross-sectional shape, midshaft Ix/Iy and
subtrochanteric Imax/Imin, where both agricultural groups had lower values for these
ratios (i.e., increased circularity). Ruff et al. (1984: 134) conclude “a possible scenario of
less running and climbing, and more walking and possibly other more sedentary pursuits
such as lifting and carrying in the two agricultural groups, is at least consistent with the
available biomechanical data. The relatively more robust Pecos femora may indicate that
while the basic types of activities were similar in the two agricultural groups, they were

more difficult and mechanically demanding at Pecos than on the Georgia coast.”

72

 

B!

I
Skeletal s‘a

 

Early \il l.
presumal i
hunter-2a
Aggrega It -
typified hx
examined

in order It

 

period ant|

Fol
dillerenccs
lmav' 1min
cross-sectii
lmax / 1min
was lower
“is no sig
Perlfxls fi )1
5k7l‘515-1‘r1r1r x1
P*fi()t‘ls. 8
Simple. me

2513an

tr." 5

Brock and Ruff (1988) studied structural changes in the lower limb of three
skeletal samples representing a temporal-cultural sequence in prehistoric New Mexico:
Early Villages, Abandonments, and Aggregated Villages. These cultural periods
presumably practiced different mobility and subsistence strategies, ranging from mobile
hunter-gatherers in the Early Villages period to sedentary maize agriculturalists in the
Aggregated Villages period, with the Abandonments period categorized as transitional,
typified by “disequilibrium and restructuring of communities” (ibid: 1 16). The authors
examined cross-sectional properties of male and female femora from each of these periods
in order to test the hypothesis that levels of mobility were greatest in the Abandonments
period and lowest in the Aggregated Villages period.

For females, results of statistical analyses of the data revealed no statistical
differences among periods for my; cross-sectional property, including shape (Ix/1y and
Imax/Imin). In males, no statistically significant group differences were found for any
cross-sectional property, except the cross-sectional shape index Imax/Imin. Midshaft
Imax/ [min was lowest in the Aggregated Villages period, and subtrochanteric Imax/ [min
was lower during the Abandonments period than the Aggregated Villages period. There
was no significant difference in Imax/ 1min between the Early Village and Abandonments
periods for either section location (midshaft or subtrochanteric), and there was no
Signyicant difliarence in midshaft Ix/Iy (the mobility index) for males among the three cultural
periods. Somewhat surprisingly, considering the lack of significant differences between
sample means in this study, the authors concluded, “The temporal patterns in femoral
geometry suggest that activity levels in the prehistoric American Southwest increased
between the Early Villages and Abandonments periods, then declined during Aggregated

Villages in both sexes” (Brock and Ruff 1988: 125).

73

 

 

Great Ba
prehistnri
groups at
upland se
ofhumeri
activity, rt
iliitl: l0?

Cr
compared

"tlillirent I

 

Slgnilitanc
authim; St.-

La:
Stillum,.r ‘

Value-‘3 1i ')r q

SUT‘R‘QT: [l]

lhis “NM;

\dlUf‘S “\‘(1-
s _ "I

’V’Ulhth‘M...

Larsen et al. (1995) studied the prehistoric Stillwater Marsh people of the western
Great Basin. According to the authors (ibid: 107), “With regard to Great Basin
prehistoric populations, a debate has centered on the relative degree of mobility of human
groups and the degree of importance of dietary resources extracted from marsh versus
upland settings.” Patterns of osteoarthritis along with cross-sectional geometric properties
of humeri and femora were examined, “in order to shed light on patterns of physical
activity, resource acquisition, and settlement in the prehistoric Stillwater populations”
(ibid: 107). This summary focuses on the results of the biomechanical analysis.

Cross-sectional properties (CAJ, Ix/Iy) of the Stillwater Marsh sample were
compared to those published for a variety of prehistoric population samples representing
“different topographic and subsistence contexts” (ibid: 125). No indication of statistical
significance (i.e., P—value) was provided for any of the group comparisons discussed by the
authors; statistical analyses were either not performed or simply not reported.

Larsen et al. (1995) reported that in comparison to other populations, the
Stillwater Marsh remains have relatively low CA, but high TA, resulting in relatively high
values for second moments of area (e.g.,J). Among males, the authors asserted that
variation in torsional rigidity of the femoral midshaft “closely parallels subsistence
strategy; that is, hunter-gatherers are highest and agriculturalists are lowest” (ibid: 128).
This association was not found in females, however, whose femoral torsional rigidity
values “seem to correspond to degree of ruggedness of terrain, with Great Basin and
Southwestern (Pecos) populations showing the highest values, Georgia coastal populations
the lowest, and Plains populations intermediate between the other two regions” (ibid:
128). The low cortical area in combination with high torsional rigidity of the Stillwater

Marsh femoral sample was interpreted as suggesting “episodic undernutrition” (ibid: 131),

74

 

but with
support! ‘4

cortical a

 

bitimt‘t‘lla

the Stillu .'

.-\

.shaptj
travel

1131’ lt‘nltir,

gatherers i

 

authors (Ll
reconstruc
than fema
females \s':
Wnparm
ln[t*rpn.u,(
lt‘lt‘ling it I'
demanclim

133
Sll 1)
(Tim

Parf’d 1‘

311(1 ”01.11)“.

 

but with a high degree of mechanical loading. The authors claimed this conclusion is
supported because relative to the femora, the Stillwater Marsh humeri also have low
cortical area, reflecting a systemic factor (e.g., nutrition), but are comparatively
biomechanically weak, reflecting localized mechanical loading patterns.

According to Larsen et a1. (1995: 132), relative to the other population samples,
the Stillwater Marsh males have a relatively high femoral midshaft Ix/ 1y value indicating,
“. . .shapes are modified such that they reflect highly mobile behaviors (e.g., long-distance
travel).” The Stillwater Marsh people also exhibited a high degree of sexual dimorphism
for femoral shape, which the authors consider “a striking characteristic of hunter-
gatherers in general...probably the result of a sexual division of labor” (ibid: 129). The
authors claimed, “This study of structural morphology, then, appears to support a
reconstruction of a mobile lifeway for men and women, but with males being more mobile
than females” (ibid: 132, emphasis added). Parenthetically, the Ix/Iy value for Stillwater
females was reported as 1.0. This was the second lowest value among the six populations
compared and indicates a circular cross-section, a result that in other studies has been
interpreted as indicative of low mobility or sedentism. The authors concluded, “Far from
leading a leisurely existence, these populations likely followed a highly physically
demanding lifeway that involved frequent travel over rugged terrain. . .including the
nearby uplands. . .subsistence strategies were not focused entirely on the marsh” (ibid:
133)

Subsequently Ruff (1999) expanded his Great Basin study to include samples from
Stillwater Marsh, Malheur Lake, and the Great Salt Lake regions. Results were again
compared to those published for the Southwest, Georgia coast, the northern Great Plains,

and northwestern Alabama. This study sought to evaluate the relationship of cross-

 

set‘tir )nal
mnrrn tai

The Slillt

 

hunter-fr
by inmrp

“l
strength. ‘

CtlirSS-St‘t' t'

 

 

Group dil
r

her t-tests
femoral cr
Pmpt‘rtier
R.
Statistical}
lit/ll. “(is
suhtmr-h ;
lower tha
Sex (run,
l‘fls'. “‘11

“”1 lilies '.

T .
51 .
”‘1 (mm

‘\

sectional geometric variation of femora and humeri with geographic terrain
(mountainous, plains, coastal), subsistence strategy (pre-agricultural, agricultural), and sex.
The Stillwater Marsh, Malheur Lake, and Great Salt Lake samples are all described as
hurl ter—fisher-gatherers, with the Great Salt Lake population differing from the other two
by incorporating corn agriculture.

“In this study, CA and] are taken as the principal indicators of bone rigidity or
strength, while other second moments of area, considered as ratios, are used to examine
cross-sectional shape differences or relative strengths in different planes” (ibid: 296).
Group differences were analyzed using two-way AN OVA (sex and group) and Tukey post
hoc t-tests for between-group comparisons. The effects of terrain, subsistence, and sex on
femoral cross-sectional properties were tested by three-way ANOVA. Cross-sectional
properties were size-standardized following Ruff et al. (1993).

Results of comparisons among the Great Basin samples demonstrated few
statistically significant differences in cross-sectional properties. In the femur, midshaft
Ix/ Iy was lower in the Malheur Lake sample than either of the other two, and
subtrochanteric CA was lower in Stillwater than Malheur Lake, and near-significantly
lower than Great Salt Lake. Humeral Ix/Iy was lower in Stillwater than Malheur Lake.
Sex differences were confined to cross-sectional shape ratios, namely femoral midshaft
IX/ 1y, where male values are greater than female, and subtrochanteric Imax/Imin, where
females are greater than males.

Femoral midshaft comparisons between the combined Great Basin sample and
two Other combined samples (one representing various other Native American groups,

and another representing other hunter-gatherer populations) show the Great Basin

remol‘a to have the greatest torsional rigidity. Differences in other midshaft properties

76

 

were [it it

 

dernnnsr z:
with sarn

plains L...
(Ll-rt «in a
factor liir‘
While dill!
strategy.

agrirulur r

 

1“.
femoral n
terrain in
broadly xi
S“l’POrts i
that the (
Ampmm
lower lint
prop...” 1t .
mitlghdfi
agrir-Uku
Bibi“ Sat
find: 3 l ‘
SUM? Ix.

Mall“.

UK

were not significant. Results of the three-way AN OVA of all samples combined
demonstrated a statistically significant effect of terrain on femoral midshaft CA and],
with samples from mountainous regions exhibiting greater values than samples from the
plains (GA) and coastal (CA andJ) regions. Subsistence strategy was found to have no
521?th on any midshaft cross-sectional property. Not surprisingly, sex was an important
factor for CA,J and Ix/Iy with males having greater values than females. Interestingly,
while differing between the sexes, Ix/Iy did not vary with either terrain or subsistence
strategy. Sexual dimorphism in Ix/Iy was significant among hunter-gatherers, but not
agriculturalists.

In combination, these results were interpreted as “indicat[ing] that the increase in
femoral robusticity in the Great Basin is most likely related to the ruggedness of the
terrain in this region. . .Hunter-gatherers from other regions, who presumably shared
broadly similar patterns of behavior, have significantly less robust femora. . .[which]
supports this interpretation” (ibid: 312). Ruff added, “Another possible explanation is
that the Great Basin inhabitants. . .had to travel farther during foraging than most other
American Indian groups, thereby imposing more prolonged mechanical loading of the
lower limbs” (ibid: 312). Although, subsistence strategy had no effect on cross-sectional
properties, Ruff maintained that it does have an effect on sexual dimorphism of femoral
midshaft shape, in which hunter-gatherers exhibited greater sex differences than
agriculturalists. The high degree of sexual dimorphism in Ix/Iy exhibited by the Great
Basin samples is “consistent with very different degrees of mobility in males and females”
(ibid: 315). Ruff discussed the finding that the Malheur Lake sample had significantly
lower Ix/Iy than the other two Great Basin samples, “the lower [midshaft Ix/Iy] in the

Malheur sample, particularly among males, could indicate a somewhat more sedentary

77

life—style when compared to the Stillwater and eastern Great Basin samples. This index in
isolation must be interpreted cautiously, however, since population differences in other
factors, such as general body build, can also influence femoral diaphyseal shape. It is
possible that this sample as a whole had an overall physically demanding yet more
geographically limited life-style than the other two Great Basin samples, with males
remaining more mobile than females. . .” (ibid: 317).

Ruff (1994) also studied variation in femoral midshaft geometry from samples
representing a cultural/ temporal sequence in the northern Great Plains, and a single
sample from the southern Great Plains. As in previous studies, Ruff compared the data
from these samples to those published for the Georgia coast pre-contact and Pecos Pueblo
groups. Results demonstrated few differences in size-standardized geometric properties
among the various geographic regions (Georgia, Southwest, Northern Plains, and
Southern Plains), as well as among the Great Plains samples (only MA varied
significantly, and only among females). The primary exception was in the midshaft cross-
sectional shape ratios, where the Southern Plains sample was greater in magnitude than
any other groups, and significantly greater than the other Great Plains samples. Sexual
dimorphism of these ratios was also greatest in the Southern Plains sample. The different
cultural/ temporal groups from the Northern Plains exhibited virtually no variation in
midshaft cross-sectional shape. Furthermore, within the Northern Plains, Ruff found no
significant changes in femoral midshaft structure following the introduction of the horse.

The dramatic anterior-posterior elongation of the midshaft femur and high degree
of sexual dimorphism in midshaft cross-sectional shape of the Southern Plains sample is
described as “an apparent paradox” (ibid: 242) because it placed this agricultural group at

the high end of the range of Ix/Iy values Ruff has reported for hunter-gatherers. Ruff

78

suggested that this “could indicate a very unusual behavior pattern, with generally high
levels of mobility in both males and females, but particularly high levels among males,”
and concluded, “the present data indicate significant variation in mechanical loading and
behavioral use of the lower limb among the Plains Village sites, with a possible North-
South dichotomy” (ibid: 243).

Wescott (2001) applied a biomechanical approach in a large-scale, comprehensive
study of long bone structural variation within and among groups from the American
Great Plains, Great Basin, Southwest, and Texas gulf coast, as well as, early and late
modern industrial populations. Wescott assigned ordinal scores for activity level, mobility
level, and ruggedness of the physical terrain to his sample populations to analyze their
effects on long bone morphology. Results of statistical analyses revealed that significant
group differences did not correspond to degree of mobility, nor did hunter-gatherer male
femora exhibit “greater anteroposterior elongation of the femoral midshaft than
horticulturalists, agriculturalists, or modern industrialists” (ibid: 184). Wescott stated,
“very little of the total variation in the humerus or femur is explained by activity
level. . .[except] between lower-level activity groups. . .and higher-level activity groups”
(ibid: 184-185). Greater activity level was correlated with increased medial-lateral
femoral diameter leading to increased torsional strength. Wescott’s results also showed
that, “males tend to cluster by subsistence technology but females do not” (ibid: 185).
Females clustered by geographic region; however, Wescott did not interpret this as an
effect of physical terrain, as has been suggested (Larsen et al. 1995), because “Females
cluster nearly identically whether femoral or humeral dimensions are used” (ibid: 185).
In fact, this study provided no evidence for a strong relationship between terrain and

femoral robusticity. Similar to what has been observed in other studies, sexual

79

dimorphism of the femur was found to correlate somewhat with subsistence strategy, with
“broad-spectrum hunter-gatherers” most dimorphic at the femoral midshaft. Exactly the
opposite relationship was found for the midshaft humerus. Wescott concluded, “Within a
tightly controlled geographical region, patterns of behavior can probably be reliably
interpreted from long bone structural morphology, but variation in mobility within
subsistence groups, cultural practices, diet, health and other factors probably preclude the

broad generalizations about the type of activity based on long bone morphology” (ibid:

195).

Bioarehaeology of non-North American populations

Fewer studies of cross-sectional geometry have been published for population
samples from outside of North America. However, studies have been conducted on
skeletal populations from Great Britain, Africa, and Asia.

Mays (1999) applied a biomechanical approach to the assessment of differences in
physical activity patterns among three skeletal groups buried in a medieval period
cemetery in York, England. The study samples consisted of humeri from adult female
and male laypeople and a male monastic group. Results showed significant female-male
differences in bilateral asymmetry between the laypeople, but no statistical difference
between the male samples, or between the female layperson and male monastic samples.
In addition, the lay-male humeri had greater torsional strength than monastic male
humeri. Mays concluded that males (but not females) typically participated in activities
requiring differential use of the upper limbs, and that laymen were more involved in
manual labor than monastic males.

Ledger et al. (2000) evaluated the cross-sectional properties of tibiae and humeri

80

from an l8lh century Cobern Street burial site in Cape Town, South Africa for the
purpose of “more accurately quantifyling] the physical demands (type and level of
activity) of this group” (p. 208). This sample, of “slaves or ‘free black’ people of low
socioeconomic standing” (ibid: 207), was compared to two other samples, 1) a modern
cadaver collection representing the likely descendants of the burial population, and 2) a
South African hunter-gatherer population dating to 2000-4000 years BP.

Statistical comparisons of sample means showed no significant differences
between the Cobern Street and modern samples for tibiae of either sex. Almost all of the
significant differences were between the hunter-gatherer sample and one or both of the
other two groups. Male hunter-gatherer tibiae had significantly greater cortical area than
both the Cobern Street and modern samples, and significantly greater torsional strength
than the Cobern Street sample. Female hunter-gatherer tibiae had significantly lower Iy
than both the Cobern Street and modern samples. For the humerus, again no significant
differences were found between the Cobern Street and modern samples for any
biomechanical property (CA, Ix, Iy). Male hunter-gatherer humeri had lower Iy than
both the Cobern Street and modern samples, and lower torsional strength than the
Cobern Street sample. For female humeri, both Ix and Iy were lower in the hunter-
gatherer sample than in the other two. Humeral bilateral asymmetry was greatest in the
hunter-gatherer sample for both sexes, but similar between the Cobern Street and
modern samples.

In spite of the lack of statistically significant differences between the Cobern Street
and modern samples, the authors concluded, “The Cobern Street people appear to have
been manual laborers living a more physically demanding lifestyle than their descendants

have in the modern group” (ibid: 215). The differences between the hunter-gatherer and

81

A”- a...

II»

the other two samples was attributed to “their high mobility. . . [that required] a greater
strength of their lower limbs than of their upper limbs” (ibid: 214).

Stock and Pfeiffer (2001) utilized a slightly different approach in their study of
variation in cross-sectional geometry. They looked for evidence of systemic versus
localized mechanical loading influences by studying skeletal adaptation in the upper limb
(humerus, clavicle) versus the lower limb (femora, tibiae, lSt metatarsals) from two
different foraging populations with “well-documented evidence” (ibid: 338) for different
patterns and modes of mobility. The skeletal samples were derived from two groups: 1) a
Later Stone Age (1 1,000 to 2000 years BP) group from coastal South Africa, and 2) a
group of protohistoric (ca. 19‘h century) Andaman Islanders. According to the authors,
Later Stone Age group subsistence “is characterized by the hunting of small game,
terrestrial foraging, and the intensive exploitation of coastal marine resources...
[requiring] ...negotiation of rugged terrain” (ibid: 339-340). The Andaman Islanders
subsistence was also both terrestrial and marine based, but with frequent use of canoe for
exploiting both marine and terrestrial resources. Swimming was also a common activity
in this group, beginning at an early age. The authors characterized these two groups as
having high terrestrial mobility and high marine mobility, respectively. Based on this
classification of habitual activity, the authors predicted greater lower limb strength for the
Later Stone Age group, and greater upper limb strength for the Andaman Islanders.
This hypothesis was generally confirmed by the statistical analysis of sample means for

most cross-sectional properties analyzed.

A critigne of bioarchaeological studies of begayior

 

In addition to the original research summarized above, proponents of the

82

biomechanical approach in bioarchaeology have published a number of review articles
(Larsen 1995, 1997, 2000, 2002; Bridges 1995; Ruff 2000b) promoting cross-sectional
geometry as a viable approach to inferring behavior from skeletal remains. In stark
contrast,Jurmain (1999) has published a critique of the research calling into question the
conclusions of biomechanical analyses. In his book,Jurmain dubbed the tendency for
anthropologists to focus on behavioral explanations for observed patterns of variation in
cross-sectional geometry “activity-only myopia” (ibid: 262), and pointed out that
interpretations based exclusively on mechanical hypotheses ignore alternative
explanations, and the inherent biological complexity of skeletal adaptation. Furthermore,
he argued for continued research on the effects of known levels and types of physical
activity on bone geometry.

Jurmain also strongly criticized these studies for constructing “just-so” stories (ibid:
254) based on ethnographic and archaeological data, and stated that, “such data, by no
means, provide an adequate test for any functional hypothesis” (ibid: 249). He added,
“What is immediately obvious here is that activity is assumed to explain all the variation
present, regardless of the complexities of the patterns expressed. Moreover, what is being
assumed is exactly what such analyses should be trying to demonstrate” (ibid: 253).

Are biomechanical data used to reconstruct behavior, or are presuppositions
about behavior used to interpret the biomechanical data? Many of the studies reviewed
in this chapter claimed that the purpose of the study was to determine whether physical
activity patterns changed with the transition from hunting and gathering to agriculture,
or following European contact. To what extent has this purpose been fulfilled? In
actuality,just asJurmain has pointed out, the discussion section of most publications

involves using archaeological, ethnohistoric, and ethnographic information to explain the

83

biomechanical data, not the other way round. In other words, the behavioral
reconstruction of a population is considered the “known” variable. It is “known” that
agriculturalists are more sedentary than hunter-gatherers; therefore, decreases in femoral
strength or changes in femoral shape in agricultural groups are the result of decreased
mobility. It is “known” that female agriculturalists spend hours each day pounding com;
therefore, a decrease in humeral bilateral asymmetry in Mississippian females can be
explained by this activity. When humeral strength was found to have decreased in
Mississippian females when it was expected to have increased (Bridges et al. 2000), again,
the explanation was sought in archaeological and ethnohistoric data. When the
biomechanical data revealed opposite findings in Alabama and Georgia, regional variation
in the transition from hunting and gathering to agriculture was the assumed explanation.
Where the archaeological record is not specific enough to generate plausible explanations
for the biomechanical data, the results are considered non-interpretable or simply
attributable to general increases or decreases in activity level (e.g., Larsen et al. 1995; Ruff
and Larsen 1990).

The problem with basing conclusions on the fact that the biomechanical data are
“consistent” with the archaeological record is that the archaeological record is
incomplete, and interpretations of it may be flawed. It is simply not known, and cannot
be estimated in any reliable sense, how labor-intensive one prehistoric lifestyle was
relative to another. This was the promise of biomechanical analyses, yet in order for
biomechanical data to be an independent source of information about such things, the
archaeological record cannot be used to explain the biomechanical findings. Could
bioarchaeologists be correct in their “behavioral reconstructions”? The answer is yes, of

course, they could; however, their conclusions rarely deviate from what is already

84

“known” or inferred from other, more traditional sources of information.

Jurmain refers to this practice of confirming hypotheses regarding physical activity
from equally unsubstantiated sources of evidence as “circular reasoning.” He is
particularly critical of the “mobility index” (Ix/ Iy):

What is perhaps most disturbing is that this very tentative interpretation of the etiology of
altered femoral shaft shape, i.e., A/ P expansion due specifically to long-distance travel, has
been repeated numerous times elsewhere. Given enough repetitions, the hypothesis
now appears to be accepted (uncritically) as established. But, to my knowledge, there is
not one single clinical study which has demonstrated a relationship between activity levels
and consistent changes in the shape of the femoral diaphysis. Given this lack of even a
general functional correlation, the claims that such changes are the result of a speci ‘c
activity are even more speculative. Of course, the hypothesis may eventually be
confirmed, and would prove the suggestion to be a brilliant (and most useful) insight.

Clearly, however, until substantiation is forthcoming, judicious restraint is called for (ibid:

249-250).

Summggy

As the literature review in this chapter illustrates, behavioral reconstructions of
past populations based on biomechanical analyses of long bone structure are numerous.
Although, the biomechanical approach in bioarchaeology has evolved since its inception,
it is still essentially grounded in the initial biomechanical model produced in the early
19805. These early anthropological studies showed that locational variation of cross-
sectional geometric properties within the lower limb bones corresponds to theoretically
predicted levels of mechanical stress generated by activities such as walking and running.
Subsequently, it was hypothesized that group variation in cross-sectional geometric

properties could be explained primarily by differences in mechanical loading patterns

generated by physical activity, as long as sex, age-at-death, side of the body, body size,
and body shape factors are taken into consideration.

When attempting to look for general patterns of variation in cross-sectional
properties, it is worth noting that the results of many of the studies, even those conducted
by the same researchers, are not necessarily comparable. Over the past twenty years
there have been some methodological changes. For example, the method used to size-
standardize the raw cross-sectional data has undergone revision. There are also
methodological differences between studies in terms of the section locations analyzed, the
specific geometric properties measured, and the statistical procedures employed (e.g.,
compare Ruff and Larsen 1990 with Bridges 1989). Moreover, the specific results of the
studies differ (i.e., whether biomechanical variables, sexual dimorphism or bilateral
asymmetry increased or decreased over time).

In spite of the obvious differences in methodology and results, as well as a notable
lack of statistically significant findings in many studies, most authors’ contend that their
initial research hypotheses, usually based on a combination of archaeological,
ethnographic, and ethnohistoric evidence, were supported by the biomechanical data.
While occasionally circumspect and tentative in their conclusions about the specy‘ic
activities that produced the observed variation in cross-sectional properties, most of the
authors in this literature review have not seriously questioned the ability of the

biomechanical model to test behavioral hypotheses.

CHAPTER 4—METHOD OF ANALYSIS

There is a deductive logic, and an inductive logic— -~and there is seductive ‘logic’
Arnold Kamiat (1936)

Application of criticgl reasomg

 

As reviewed in chapter 3, several biological anthropologists, most notably Ruff
and colleagues, have generated a biomechanical model in which long bone cross-sectional
geometric variation among archaeological populations is attributed to differences in
behavior (i.e., patterns of physical activity). Fundamentally, the purpose of this
dissertation is to answer the question: Is this model supported by scientific evidence?
Bioarehaeologists who employ a biomechanical approach have presented their rationale
in the publications reviewed in chapter 3. This dissertation applies concepts and
analytical techniques from the discipline of informal logic7 to reconstruct and critically

evaluate their line of reasoning using the model of an inductive argument.

Inductive meats

In logic, an argument is modeled as a series of statements called propositions; one
is the conclusion, the rest are premises intended to support the conclusion. Arguments
are categorized as either deductive or inductive (Cederblom and Paulsen 2001; Copi and
Cohen 1990). In a deductive argument, the truth of the premises guarantees the truth of

the conclusion. The following is a classic example of a deductive argument (Copi and

 

7 . . . .

Informal logic uses logical pnncnples to analyze natural-language arguments from everyday contexts
(Cederblom and Paulsen 2000); whereas, formal logic employs symbolic language to analyze the formal
structure of arguments.

87

Cohen 1990):

(1) All humans are mortal.

(2) Socrates is human.

Therefore, Socrates is mortal.

In this argument, the conclusion is logically inevitable if the premises are true. Validity is
a concept that refers to the “truth-preserving” form of deductive arguments (\‘Varburton
2000). If an argument is valid, no additional information can change the conclusion of
the argument (Copi and Cohen 1990). A flaw in the structure or form of the argument
results in a formal fallacy, i.e., an invalid argument. The following is an example of a
formal fallacy:

(1) All humans are mortal.

(2) Socrates is mortal.

Therefore, Socrates is human.

Although similar to the previous valid form of the argument, the form of this argument
does not guarantee the truth of the conclusion. Both premises are true, but Socrates
could be a dog. There are several types of formal fallacies. Distinguishing between valid
deductive arguments and formal fallacies is the subject of formal logic. A valid deductive
argument with true premises, and therefore a true conclusion, is a sound argument
(VVarburton 2000).

In inductive arguments, true premises do not guarantee a true conclusion
(Cederblom and Paulsen 2001; Copi and Cohen 1990); all the premises of an inductive
argument can be true and yet the conclusion false. This is because the relationship
between the premises and the conclusion is probabilistic rather than deterministic. The

probability that the conclusion of an inductive argument follows from its premises is “a

88

matter of degree and dependent upon what else may be the case” (Copi and Cohen 1990:
49).

Inductive arguments are never valid. In logic, the concept of validity, as a
criterion for determining whether an argument is “successful,” only applies to arguments
intended to be deductive. In contrast, inductive arguments are evaluated based on the
degree of support they provide for the conclusion (Copi and Cohen 1990). A successful
or “strong” inductive argument is one in which the conclusion is highly probable. The
potential weaknesses of inductive arguments are 1) false or poorly supported premises,
and 2) a conclusion that does not follow from the premises even if the premises are true.

Although truth-preserving, the deductive process adds no new information; it is
said to be “non-ampliative” because the conclusion is essentially contained in the
premises (Bird 1998; Pennock 2000). Logical and mathematical proofs are deductive. In
contrast, the empirical sciences (e.g., chemistry, physics, biology, geology) are
fundamentally based on an inductive process (Bird 1998). Only induction allows
predictions (i.e., conclusions) regarding unknown situations or future events to be made
based on observations of past events. Induction has been criticized as flawed in the sense
that its conclusions are always tentative. Although it would be a strong inductive
argument to suggest that because the sun has always risen in the past it will rise
tomorrow, there is no guarantee that it will.

Induction is often defined as the process of deriving general principles from
specific observations. However, this is not always the case. Induction, in the broadest
sense of the word, simply means non-deductive (Bird 1998). There are several forms of
inductive argument (Cederblom and Paulsen 2001). Three ofthese are addressed below:

1) sampling arguments, 2) arguments with statistical premises, and 3) arguments from

89

Ef‘tter;

partit-i

 

analogy.

Sampling arguments lead to an inductive or empirical generalization; they argue
from particular instances to a general conclusion, and thus fit the standard definition of
induction. The following is an example of a sampling argument:

(1) Observation A showed an increase in bone mass following physical activity.

(2) Observation B showed an increase in bone mass following physical activity.

(3) Observation C showed an increase in bone mass following physical activity.

Probable conclusion: Bone mass increases following physical activity.

Causal arguments are a special form of sampling arguments. If observations A, B, and C
are the results of controlled experiments designed to establish a cause-effect relationship
between physical activity and bone mass, the premises might be statements to the effect
that bone mass directly correlates with physical activity, and the conclusion of the
argument could be a theoretical statement: Physical activity causes increases in bone
mass.

In contrast to sampling arguments, some inductive arguments have a statistical
generalization as a premise. Arguments of this type move from a general premise to a
particular conclusion. The following is an example of this type of argument:

(1) Most people who participate in vigorous physical activities have greater bone mass
than people who do not.
(2) Bob participates in vigorous physical activities.
(3) Tom does not participate in vigorous physical activities.
Probable conclusion: Bob has greater bone mass than Tom.
In this argument, premise l is a generalization, presumably the conclusion of a sampling

argument. If all the premises are true, then the statistical probability represented by the

90

word “most” in premise l, to a large degree, determines the strength of the above
argument; that is, if most equals 99%, then the conclusion is much more likely to be true
than if most equals 60%.

Arguments from analogy are based on a comparison between two things that are
allegedly similar. These arguments rely on the principle that if two things share some
characteristics they are likely to share others. The following represents an argument from
analogy:

(l) Populations A and B are both Native American.

(2) Populations A and B both represent stratified societies.

(3) Populations A and B are both agricultural.

(4) The females of population A are more involved in agricultural activities than the
males.

Probable conclusion: The females of population B are more involved in agricultural

activities than the males.

maul procedure

The critical analysis presented in this dissertation involved three basic steps (based
on Cederblom and Paulsen 2000): 1) Reconstruction of the argument, 2) Evaluation of
the argument, and 3) Determination of whether to accept or reject the conclusion of the
argument.

The argument presented in chapter 5 was reconstructed by identifying the
premises and the conclusion embedded in the prose of the research articles and
dissertations reviewed in chapter 3. The premises of an argument are sometimes

indicated by key words and phrases, including: since, because, as, follows from, as

91

rt‘l

mi

Sum

1135

shown/ indicated by, the reason is, may be inferred/ derived/ deduced from (Copi and
Cohen 1990). Likewise, conclusion statements often have indicator words or phrases: in
conclusion, therefore, so, hence, thus, accordingly, as a result, suggests, implies, for this
reason.

The conclusions were relatively easy to identify because they were found explicitly
stated at the end of the discussion or in a separate conclusion section. The premises were
more difficult to identify. This is partly because much of the text of a scientific article is
exposition intended to provide the reader with background information necessary to
understand the subject matter, and does not actually contribute to the argument. In
order to reconstruct the argument as fairly as possible, it was assumed that if a statement
was necessary to complete the line of reasoning or strengthen the argument being made,
it was included as a premise. Not every premise was articulated in each research article;
however, the majority of the premises were derived from the earlier works of Ruff and
colleagues (Ruff 1987; Ruff and Hayes l983a; Ruff and Larsen 1990; Ruff et a1. 1993,
1994; Trinkaus et al. 1994). Once identified, the premises and conclusion were
summarized and paraphrased, making sure to preserve the author’s original intent. This
was done so that the reconstructed argument would be easier to follow. The argument
was then evaluated to determine whether the conclusion should be accepted or rejected.
This step involved reviewing the non-anthropological theoretical, experimental, and
clinical literature on skeletal biology and mechanobiology to determine how well the
premises were supported, and whether the conclusion followed from the premises based

on all available relevant evidence.

are
(in;
‘llltl
Lilli

1" ”t

‘lt't

3rd,

Techniques for criticizipg inductive mments

Sampling arguments and arguments with statistical premises can both be
criticized by: 1) Casting doubt on the premises, and 2) Demonstrating that the argument
does not use all available relevant evidence (Cederblom and Paulsen 2000; Copi and
Cohen 1990). Obviously if the premises of an argument are false, the conclusion of the
argument has not been supported. It is important to note that this does not prove that
the conclusion is false, only that it does not follow from the premises. Sampling
arguments can be criticized by pointing out problems with the data or the research
design, which could cast doubt on the premises. For example, the sample size could be
small or the group non-representative. Alternatively, the study outcome may have been
affected by confounding variables". Causal arguments can also suffer from post hoc
errors, i.e., confusion of correlation with causation.

Unlike the ease of a valid deductive argument, the addition of relevant new
information to an inductive argument can lead to a different conclusion. For example,
additional information can Show that the supposed regularity of an inductive
generalization does not actually exist, i.e., introducing counterevidence can potentially
refute the generalization. This is demonstrated by adding premises 4 and 5 to the
previous example of a sampling argument:

(1) Observation A showed an increase in bone mass following physical activity.

(2) Observation B showed an increase in bone mass following physical activity.

(3) Observation C showed an increase in bone mass following physical activity.

(4) Observation D showed no change in bone mass following physical activity.

 

8 A confounding variable is defined as “any uncontrolled variable that might affect the outcome of a study”
(Graziano and Raulin 1989: 384).

93

(5) Observation E showed a reduction in bone mass following physical activity.
Probable conclusion: 60% of the time bone mass increases following physical
activity.
It is clear that the previous inductive generalization is not as strongly supported when the
information contained in the new premises is brought to bear on the argument.

Likewise, the introduction of more specific information into an argument with
statistical premises can lead to a different conclusion. This is demonstrated using the
previous example of an argument with statistical premises:

(1) Most people who participate in vigorous physical activities have greater bone

mass than people who do not.

(2) Bob participates in vigorous physical activities.

(3) Tom does not participate in vigorous physical activities.

(4) Bob carries a gene that reduces the sensitivity of his osteocytes to mechanical

strain.

Probable conclusion: Bob and Tom will have similar bone mass.

Arguments from analogy can be criticized by showing that the analogy itself is
flawed. For example, relevant dissimilarities between the two things being compared make
it less likely that they share the additional characteristic inferred in the conclusion.
Furthermore, the conclusion is made less likely by showing that the alleged similarities do
not exist, or that there is no real connection between the shared characteristics and the
additional characteristic in question. In the previous example of an argument from
analogy, the strength of the argument is partly dependent on whether the two populations
in question actually share the characteristics proposed in premises 1 through 3. In other

words, the conclusion that the females of population B are more involved in agricultural

94

activities than the males is much less likely if there is little evidence to show that
population B is agricultural. Likewise, the conclusion is weakened if it can be shown that
there is no consistent relationship between sexual division of labor and the shared
characteristics. An example of a Native American, stratified, or agricultural population in
which the males were more involved in agricultural activities would cast doubt on the

conclusion.

Begging the question

Jurmain (1999) has accused bioarchaeologists of circular reasoning. In logic,
circular reasoning, also known as “begging the question,” is considered an informal
fallacy9 that occurs when the premises presuppose the conclusion (Walton 1989), as when
a premise is either a restatement of the conclusion or a proposition that will be equally
doubtful for reasons similar to those which make the conclusion doubtful. Begging the
question is considered particularly “uninformative” (\N’arburton 2000) because the
premises do not provide a reason to accept the conclusion. Most examples of begging the

question represent arguments with unsupported premises (Copi and Cohen 1990).

 

9 An informal fallacy is any type of faulty reasoning other than a formal fallacy (an invalid argument
resulting from a flawed formal structure).

95

lit;

CHAPTER 5—RECONSTRUCTION AND EVALUATION OF THE
ARGUIWENT

Ifit was so, it might be; and ifit were so, it would be; but as it isn’t, it ain’t. That’s logic.

Lewis Carroll

Introduction

Since the early 19805, numerous bioarchaeological studies have applied a
biomechanical approach to the problem of “reconstructing” physical activity patterns
(i.e., behavior) from archaeological skeletal remains. Ultimately, the written presentations
of these research projects are more than a simple reporting of results; they are an attempt
by the authors to persuade the reader that the conclusions they have drawn from their
data are correct. In other words, they construct an argument. The purpose of this

chapter is to reconstruct and critically analyze the argument.

Reconstruction of the a_rgpment

The argument for inferring physical activity patterns from biomechanical
variables, e.g., cross-sectional geometry of long bone diaphyses, is embedded in the prose
of the bioarchaeological literature reviewed in chapter 3. It is not within the scope of this
dissertation to systematically reconstruct and evaluate the nuances of each author’s
specific argument. For example, it is not my intent to reconstruct Bridges’ (1989)
argument for her conclusion that, in northwestern Alabama, the transition from hunting
and gathering to agriculture entailed an increase in the workload in females, or the
argument presented by Ruff and Larsen (1990) for their conclusion that the exact

opposite occurred following the adoption of agriculture on the Georgia coast. Rather, I

 

limit
and 0

min.

lllt’ Qt
19'th

Wt

present a reconstruction and critique of the overall argument for the ability to draw
behavioral conclusions from long bone diaphyseal structure. This argument is
summarized by the propositions listed in figure 5-1.

The propositions that comprise the argument represent paraphrased statements
from the introductions, discussions, and conclusions of the journal articles, book chapters,
and dissertations reviewed in chapter 3. Not all of the propositions are explicitly
articulated in each manuscript; however, they are all necessary to present the complete
line of reasoning. They are listed here more-or-less in order of increasing specificity, from
the general theoretical statement in the first premise to the specific interpretation
represented by the conclusion. The overall argument is best described as an inductive

argument with statistical premises.

Critical an sis of the ment

The overall argument (Figure 5-1) is analyzed as two separate arguments
connected by a linking proposition. The first argument, represented by premises 1
through 5, provides the explanatory framework for inferring behavior from bone
structure. This will be referred to as the ‘theoretical argument’ (Figure 5-2). Its purpose
is to link long bone cross-sectional geometry with physical activity. That is, it promotes
mechanical loading patterns resulting from different physical activities as a principal
cause of long bone structural variation. Premise 5 is the conclusion of this theoretical
argument; it is inferred from the first four premises. This conclusion then functions as the
main premise in the second argument, represented by premises 5 and 6. This argument
is used by bioarchaeologists when drawing specific conclusions about behavior from

comparisons of cross-sectional properties between two or more samples. It is referred to

97

as the ‘empirical argument’ (Figure 5-3) because the conclusion is based on empirical
evidence generated by a specific bioarchaeological study.

Figure 5-1. Reconstruction of the overall aggument.

 

 

Premise 1 Bones are capable of adapting to alterations in the mechanical loads to which
they are subjected by adding bone tissue where it is needed, and removing it
from where it is not needed, i.e., changing the amount and distribution of bone
tissue to suit functional demands (W olff’ 5 Law).

Premise 2 Physical activity places mechanical loads on the bones of the skeleton.

Premise 3 Bones adapt to physical activity through changes in the amount and

Premise 3a

Premise 4

Premise 5

Premise 5a

Premise 5b

Premise 5c

Premise 5d

Premise 5e

Premise 6

Conclusion

distribution of bone tissue.

Bone material properties are not altered by physical activity.

The cross-sectional geometric properties of a long bone diaphysis measure the
amount and distribution of bone tissue.

Variation in cross-sectional geometric properties reflects differences in patterns
of physical activity.

Larger values for cross-sectional properties (i.e., cortical area and second
moments or area) indicate higher levels of physical activity, while smaller values
indicate lower levels of physical activity.

Group differences in the amount of bone (i.e., measured by cortical area) are
less informative for interpreting behavior than group differences in the
distribution of bone tissue (i.e., measured by second moments of area).

Group differences in second moments of area reflect group differences in the
level of physical activity.

Group differences in cross-sectional shape, represented by ratios of second
moments of area, suggest group differences in apes of physical activities.
Non-mechanical factors do not produce localized skeletal effects, i.e., localized
changes in the distribution of bone.

A difference in cross-sectional properties exists between groups.

It is probable that the type/ level of physical activity differs between groups.

 

In the reconstruction of the argument, premise 5 subsumes a series of additional

 

premises, identified as 5a through 5c. These premises merit special consideration. First,
they are essential to drawing specific conclusions about physical activity from observed
patterns of cross-sectional geometric variation. Premises 5a through 5d attribute specific
meaning to patterns of cross-sectional geometric variation. Premise 5e, on the other
hand, is posited as a means of ruling out the potential confounding effects on skeletal
structure of non-mechanical variables, such as group differences in nutrition and genetics
(Ruff et al. 1984). It asserts that observed variation in cross-sectional geometry,
specifically localized variation in bone tissue distribution and “cross-sectional shape,” is
mechanical in nature, and not the result of other factors. Second, these five premises (5a
— 5e) were originally conclusions inferred from ‘empirical arguments’ based on early
studies of cross-sectional geometric variation among a variety of prehistoric, early historic
and modern groups (Ruff 1987; Ruff and Hayes 1983a; Ruffet al. 1984). One problem
with these propositions is that they represent inductive generalizations of arguments that
used fallacious reasoning; they begged the question. A brief discussion and example of
their derivation is presented in the last section of this chapter; however, a thorough
critique of these premises is beyond the scope of this dissertation because it involves the
reconstruction and evaluation of five specific arguments. Instead, the purpose of this
dissertation is to critically analyze the overall argument for inferring behavior from
variation in long bone diaphyseal structure. Obviously, the primary critique of the
theoretical argument has profound implications regarding the acceptance of premises 5a
through 5e and their subsequent application to behavioral reconstruction in

bioarchaeological research.

99

Analysis of the ‘theoretical argument’

The theoretical argument represents a plausible line of reasoning, and the
conclusion would be as plausible as the least plausible premise (Walton 1989), y” no other
relevant information was brought to bear. However, the latter point is the crux of this
critique. The strength of an inductive argument with statistical premises is primarily
based on the truth (amount of support for) the premises, and whether the argument
makes use of all available relevant information, including information relating to other possible
alternative explanations, such as the effects of confounding variables. Acceptance of the
conclusion of the theoretical argument hinges on two primary issues: 1) The eviclentiary
value and relevance of the first premise, which is the primary theoretical proposition, and
2) Whether there are viable alternative explanations for variation in cross-sectional

geometry, i.e., are there confounding variables?

Figire 5-2. Reconstruction of the ‘theoretical argument’.

 

Premise 1 Bones are capable of adapting to alterations in the mechanical loads to which
they are subjected by adding bone tissue where it is needed, and removing it from
where it is not needed, i.e., optimizing the amount and distribution of bone tissue
to suit functional demands (W olff’s Law).

Premise 2 Physical activity places mechanical loads on the bones of the skeleton.

Premise 3 Bones adapt to physical activity through changes in the amount and distribution
of bone tissue.

Premise 3a Bone material properties are not altered by physical activity.

Premise 4 The cross-sectional geometric properties of a long bone diaphysis measure the
amount and distribution of bone tissue.

Conclusion Variation in cross-sectional geometric properties reflects differences in patterns of

(Premise 5)

 

physical activity.

 

 

100

Fi re 5-3. Reconstruction of the ‘empirical argument’.

 

Premise 5 Variation in cross-sectional geometric properties reflects differences in patterns of
physical activity.

Premise 5a Larger values for cross-sectional properties (i.e., cortical area and second
moments or area) indicate higher levels of physical activity, while smaller values
indicate lower levels of physical activity.

Premise 5b Group differences in the amount of bone (i.e., measured by cortical area) are less
informative for interpreting behavior than group differences in the distribution of
bone tissue (i.e., measured by second moments of area).

Premise 5c Group differences in second moments of area reflect group differences in the level
of physical activity.

Premise 5d Group differences in cross-sectional shape, represented by ratios of second
moments of area, suggest group differences in pipes of physical activities.

Premise 5e Non-mechanical factors do not produce localized skeletal effects, i.e., localized
changes in the distribution of bone.

Premise 6 A difference in cross-sectional properties exists between groups.

Conclusion It is probable that the type/ level of physical activity differs between groups.

 

 

 

The theoretical proposition

The proposition that bone adapts to its mechanical environment is considered a
fundamental principle of skeletal biology, typically referred to as Wolff’s Law. In a sense,
Wolff’s Law is an explanation for the structure of bone, asserting the biological axiom
that form follows function; consequently, changes in function produce changes in form.
Alterations in mechanical loading generate new patterns of forces that necessitate
modifications in the quantity and distribution of bone.

There are two primary problems with the application of Wolff’s Law in
bioarchaeology. The first is that Wolff’s Law is fundamentally flawed in its derivation,
and the second is that it is not, in fact, a “law.” In other words, the implied certainty or

regularity of a scientific law does not exist. W olff’s Law is based on a flawed argument

101

from analogy. Its originatorJulius Wolff compared two things that appeared similar,
namely an engineered crane and the proximal end of a human femur, and concluded that
because they shared some similarities, the femur must, like the crane, follow
mathematical laws of construction. Chapter 6 demonstrates that the analogy is not
supported. However, it is argued that current usage of the phrase “Wolfl’s Law” is simply
a synonym for the concept of functional adaptation (Cowin 2001b).

The concept of functional adaptation, as opposed to Wolff’s Law per se, is an
inductive generalization based on a sampling argument. When evidence from a variety of
studies is examined, the implied regularity of the generalization does not appear to exist.
There are a number of counterexamples in which bone either does not seem to adapt to
increased physical activity, or does so in a manner that is contrary to expectations (see
chapter 6).

The theoretical argument, which relies on “Wolff’s Law” or functional adaptation
as its primary theoretical foundation, fails to make use of all available relevant evidence.
Bioarehaeologists have not kept pace with developments in the field of skeletal
mechanobiology over the past two decades (as evidenced by a lack of citations). While
there is abundant empirical evidence that many biological tissues, including bone, are
capable of adapting to their functional milieu, there are still many critical details about
the process that remain unknown. This precludes the ability to associate cross-sectional
geometry with physical activity at this time.

According to most of the researchers whose studies were reviewed in chapter 3,
Wolff’s Law provides the theoretical basis for reconstructing behavior from

biomechanical analyses, by directly linking physical activity to long bone cross-sectional

geometry (Barondess 1998; Bridges 1989; Larsen 1997, 2000, 2002; Larsen and Ruff

102

1991; Larsen ct al. 1995; Ruff and Hayes 1983a; Ruff and Larsen 1990; Stock and
Pfeiffer 2001; Wescott 2001). Bridges (1989: 387), citing Wolff (1892), stated,
“Reconstructing changes in prehistoric activities from long bone diaphyseal dimensions
and cross-sectional structure is possible because the cross-sectional structure of a bone is
related to the forces acting upon it during life.” This interpretation is reiterated by
Larsen (1997: 195), “Julius Wolff, a leading nineteenth century German anatomist and
orthopedic surgeon, recognized the great sensitivity of bones to mechanical stimuli,
especially with regard to their ability to adjust size and shape in response to external
forces.” One of the strongest statements in support of the importance of Wolff’s Law was
made by Barondess (1998: 8) in his Ph.D. dissertation, “Wolff’s Law provides the
theoretical underpinning for understanding how it is that a skeleton’s final form provides
a window through which the external stimuli (i.e., mechanical forces) to which an
individual’s skeleton is subjected throughout life can be examined and interpreted.”
However, it was the groundbreaking publication by Ruff and Hayes (1983a: 371) that
provided the most thoroughly articulated explanation of the pivotal role played by Wolff’s
Law, “It is commonly stated that living bone alters its structure in accordance with
specific mechanical loadings, a concept first popularized byJulius Wolff. . .and frequently
referred to as ‘Wolff’s Law.’ This implies not only that mechanical stresses and strains
developed under loading will influence bone morphology, but also that preserved bone
morphology may be used to reconstruct past in-vivo bone loadings.” This interpretation
of Wolff’s Law laid the foundation for future studies of the biomechanical adaptation in
archaeological populations.

During the same period of time in which biomechanical analyses were gaining

acceptance in bioarchaeology, non-anthropological skeletal biologists and biomechanists

103

published critiques of Wolff’s Law (e.g., Bertram and Swartz 1991; Dibbcts 1992; Lee
and Taylor 1999; Roesler 1981, 1987; Currey 1997; Cowin 2001b), and have continually
reanalyzed its validity in the study of skeletal mechanobiology (e.g., Cowin 1981a; Cowin
2001a; Frost 1987, 1990a, 1990b, 1995, 1996, 1997a, 1997b, 2000, 2001; Huiskes 1995,
2000; Huiskes et al. 1987; Odgaard and Weinans 1995; Prendergast and Huiskes 1995;
Turner 1992, 1999). Yet, the applicability of Wolff’s Law to the anthropological problem
of behavioral reconstruction from archaeological skeletal remains has never been
questioned, or thoroughly analyzed by the anthropological community; “’0le Law has
simply been repeatedly paraphrased. As Roesler (1987: 1027) put it, Wolff’s Law is
“often quoted, hardly read.”

The majority of the bioarchaeological studies reviewed in chapter 3 cite a small
number of experimental studies of Wolff’s Law, which were conducted in the late 19703
to mid-19805. While these experiments (e.g., Goodship et al. 1979; Woo et al. 1981) were
relatively current for the earliest behavior studies (e.g., Bridges 1989; Brock and Ruff
1988; Ruff and Hayes 1983a, l983b), most recent bioarchaeological publications (i.e.,
those published within the past decade; Bridges et a1. 2000; Larsen et al. 1995, 1996;
Ledger 2000; Mays 1999; Ruff and Larsen 2001) do not cite any contemporaneous
experimental literature dealing with Wolff’s Law or skeletal mechanobiology. Either they
do not include non-anthropological references at all, or they cite the same few studies as
before. Thus, with respect to understanding and applying principles of skeletal
mechanobiology, there seems to be an ever-widening gulf between skeletal biologists and
biomechanists on one side and bioarchaeologists on the other. The reason for this seems
best summarized by Larsen (2000: 52, emphasis added) in a non-technical book written

for a general audience, “The premise of bone changes in response to mechanical forces is

104

an outgrowth of work done in the late 18005 by the eminent German anatomist and
orthopedic surgeon,Julius Wolff. He observed that during life—from infancy through
adulthood—bone tissue is added in areas of the skeleton where it is needed, and is taken
away where it is not needed. Such an overwhelming amount of experimental and other research has
accumulated in support of Wolfl’s conclusions that the phenomenon is iclenly‘ied simply as Wolfl’s Law.”

The View is that Wolff’s Law is axiomatic.

Confounding variables

The existence of potential confounding variables impacts the conclusion of the
theoretical argument in two ways. 1) Mechanical factors other than physical activity
could produce. observed variation in cross-sectional geometric properties. 2) Diachronic
changes (or group differences) in the amount and distribution of bone leading to variation
in cross-sectional geometric properties could occur in the absence of changes in physical
activity or other mechanical loading factors.

Premise 2 is the proposition that physical activity places mechanical loads on the
bones of the skeleton. This fact was hypothesized long before it was empirically
demonstrated using strain gauges applied to living bone (Martin et al. 1998). Since then
this premise has been confirmed by numerous experimental data, and is not disputed.
However, there is a caveat; factors other than physical activity can exert mechanical
forces on the skeleton. So, while it is probably true that all physical activity generates
mechanical loads, the inverse of this statement, all mechanical loads are generated by
physical activity, is false. In addition to physical activity, body size, shape, and mass,
normal muscle tone (i.e., tonus), abnormal muscle tone (e.g., spasticity or paralysis),

trauma, breathing, body posture, clothing and other accoutrements, and prosthetics can

105

all influence the mechanical environment of the skeleton, and potentially result in bony
adaptation. Of these, it is reasonable to assume that only body size (height), shape
(relative body proportions), and mass (weight) are likely to have a confounding effect on
patterns of variation in long bone structure, particularly at the population level.

Ruff has acknowledged the potential mechanical effects of body size, shape, and
mass, especially with regard to the bones of the lower limb (Ruff 1984, 2000a, 2000b;
Ruff and Hayes l983a; Ruff et al. 1993). Standardization methods to control for group
differences in body size using long bone length have been in common use since the mid-
19805; however, as reviewed in chapter 3, these standardization methods have changed
over time and vary among researchers casting some doubt on the accuracy and
comparability of results. Ruff (2000b) has also suggested that it is hypothetically possible
to “control” for body shape differences by limiting comparisons to groups with “similar”
body shape. Determination of body shape is feasible if the skeletal remains are complete
enough to allow the assessment of relevant body proportions. However, at present, there
is no generally accepted method to reliably estimate body mass from skeletal remains"),
because this anthropometric variable is inherently dependent upon both hard and soft
tissues. Ruff (2000a, 2000b) has admitted that conclusions are suspect for any study in
which group differences in body shape and mass are not considered. While in most
circumstances it may be “safe” to assume that ratios of body mass to body size (e.g., body
mass index or BMI) did not differ significantly among prehistoric populations, the issue
becomes contentious when a hypothetical group difference in body mass is invoked as a

potential explanation for unexpected or otherwise non-interpretable findings (e.g., Ruff

 

'0 Ruff (2000b; see also Ruff et al. 1993) has proposed a method of estimating body mass, using estimates of
body breadth and body length.

106

and Larsen 1990). Furthermore, in addition to potential mechanical effects, as a
confounding factor in biomechanical analyses, body mass may have metabolic effects on
cross-seetionaligeometry as well (Reid 2002).

Ruff et al. (1984) maintain that differences in the relative size of bones (i.e.,
robusticity or size-standardized cross-sectional properties) and distribution of bone tissue
(i.e., shape changes) as opposed to increases or decreases in body size “[seem] to suggest
localized (mechanical) rather then systemic (nutritional or other) effects” (p. 135).
Furthermore, Ruff (1987: 409) has stated, “dietary. . .explanations. . .may be relevant to
general size differences. . . [however] it is difficult to see how a systemic factor like diet
could have. . .specific effects on localized bone remodeling.” This general assertion has
appeared in several other publications (Bridges 1989, 1991; Larsen 1997; Larsen et al.
1995; Ruff and Larsen 1990). With regard to differences in long bone structure between
hunter-gatherers and agriculturalists in northwestern Alabama, Bridges (1991: 99,
emphasis added) wrote, “Although cortical area may be affected by a variety of other
factors (including genetics, diet, and age), activity differences seem to be the most
plausible explanation for the changes seen here. Nutritional differences, for example,
would lead to systemic differences in cortical area... Given the patterning of changes seen
here, it is unlikely that they result from a major change in diet, which should ajfk’ct all bones
equally.”

The idea that mechanical loading produces localized structural adaptation is
simply derived from Wolff’s Law, and follows from the idea that bone structure is
optimized: minimal material for maximum strength. Bone adaptation to mechanical
loads is need-based. When confronted with increased mechanical stresses or strains bone

cells do not add bone tissue anywhere, they add new tissue where it is necessary to resist

107

the new forces, and they remove it from where it is no longer necessary, thus
redistributing it. The implication that nutrition, genetics and other non-mechanical
factors do not produce localized effects is an untested hypothesis, not a well—supported
premise (designated as premise 5e, Figures 5-1 and 5-3). Empirical evidence for this
hypothesis is of critical importance to bioarchaeologists because many of the group
comparisons involve populations where dietary (i.e., nutritional) and genetic differences
are highly likely (e.g., comparisons of hunter-gatherers and agriculturalists, or Pecos
Pueblo and Georgia coast populations). Because Ruff et al. (1984) provide no citations
for research supporting their hypothesis (premise 5c), the literature was reviewed for
evidence that might refitte their assertion. The results of this review are included in
chapter 6. While not conclusive, studies suggest that genetics and nutrition can produce
differential changes in cortical geometry either directly, or indirectly, through primary
effects on the amount and distribution of lean and fat body mass, bone quality (i.e.,
material properties), or the processes involved in skeletal adaptation (e.g.,
mechanotransduction).

Individual (and group) variation in the respOnse of bone to mechanical usage, and
by extension, response to physical activity contributes potentially numerous confounding
variables. Ruff and colleagues (Ruff and Hayes l983b; Ruff et al. 1994) have discussed
the influential role of sex and age in skeletal adaptation, and consequently on the
interpretation of variation in skeletal structure. It is recognized that inherent sex-related
differences (e.g., hormonal status) render direct comparison of cross-sectional properties
between males and females impossible. Therefore, studies are limited to evaluating
relative male-female differences in terms of degree of sexual dimorphism. In addition,

due to the normal age-related bone loss, the structural properties of individuals of

108

different ages cannot be directly compared, and bioarchaeologists strive to control for
age-at-death in their samples‘ '. Even so, group differences in mean age are sometimes
overlooked or too readily dismissed as contributing factors in geometric variation (e.g.,
Larsen et al. 1996; Ruff and Larsen 1990).

Furthermore, Ruff and colleagues (Ruff et al. 1994) have pointed out that there
are ontogenetic differences in the response of bone to mechanical usage; the same
physical activity will produce a different response depending upon the point it is
performed within the life cycle. It is well recognized that mechanical loads are most
osteogenic prior to skeletal maturity, and especially during puberty (Duncan et al. 2001;
Frost 1986, 1997a, 1997b; Forwood and Burr 1993; Haapasalo et al. 2000; Heinonen et
a1. 2000;Jee 2001; Kannus et al. 1995; Khan et a1. 2000; Martin et a1. 1998; Ogden
2000; Seeman 2002; Slemenda et a1. 1991). Thereafter, physical activity tends to
maintain bone mass rather than increase it (Forwood and Burr 1993; Frost 1997a).
Research suggests that the cross-sectional properties of adult bones are strongly
influenced by activities performed earlier in life, and differences in cross-sectional
, geometric properties among samples might be due to group differences in the timing of the
activities relative to skeletal maturity, rather than group differences in the activities. This
“timing” factor, as potential confounding variable, cannot be addressed by controlling for
the mean age-at-death of the sample. It must be known in order to interpret the results.
The assertion by bioarchaeologists that bone structure provides a record of activities

performed over a lifetime may be true, but the inability to control for the “timing”

 

'1 Because of methodological difficulties in aging prehistoric skeletal remains (Iackes 2000; Larsen 2002),
and the inherent population specificity of techniques such as functional dental wear (Russell et al. 1990),
there is reason to question whether it is actually possible to control for age-at-death among different
population samples.

109

   

variable precludes the interpretation of this record with a reasonable degree of
confidence.

In addition, research on skeletal mechanobiology has shown
mechanotransduetion to be a complicated phenomenon, influenced by a large number of
regulatory factors (Cowin and Moss 2001), many of which are likely under genetic control
(Akhter et al. 1998; Burr 1992; Drake et al. 2001; Ishijima et al. 2001; Kodama et al.
2000; Mikic et al. 1996; Pedersen et a1. 1999; Robling and Turner 2002; Slemenda et al.
1996; van der Meulen and Huiskes 2002; Young and Xu 2001). It has been hypothesized
that metabolic factors can change the threshold (e.g., MES set-point) necessary to initiate
bone adaptation, thereby increasing or decreasing an individual’s sensitivity to
mechanical loading (Burr 1992; Hernandez 2000). If true, this implies that although
mechanical loading may be the principal stimulus in bone structural adaptation, the
biological processes involved in adaptation may be amplified or attenuated by other non-
mechanical factors; metabolic differences between individuals could lead to differences in
bone structure in the absence of absolute differences in mechanical loads. Metabolic
factors may be under either genetic or environmental control.

In the context of bioarchaeology, regulation of skeletal adaptive processes by non-
mechanical factors provides yet another alternative explanation for observed group
variation in cross-sectional geometric properties. For example, the extreme level of
humeral bilateral asymmetry seen in professional tennis players and certain Neanderthal
specimens, is usually touted as virtually irrefutable evidence for the primary role of the
mechanical environmental in determining skeletal structure (Trinkaus et al. 1994).
Because right-left side differences within an individual are said to provide the ultimate

control for confounding variables such as age, sex, nutrition, and genetics, bilateral

110

asymmetry is usually attributed to differential use of the upper limbs (Bridges 1995; Fresia
et al. 1990). However, differences in mechanosensitivity (e.g., through changes in the
MES setpoints) could also hypothetically explain group differences in degree of humeral
bilateral asymmetry. For example, a hypothetical allele responsible for increasing the
sensitivity of bone to mechanical loading could exist at a high frequency in a population.
For individuals with this allele, a relatively small difference in loading between the right
and left upper limb, such as that which typically occurs in an individual who is right- or
left-handed, could lead to significant bilateral asymmetry as compared to someone who
was less sensitive to mechanical loads. At the population level, genetic differences in
mechanosensitivity could lead to statistically significant differences between group means
even if there were no differences in the type or level of activity performed. Thus, while
the final structure of bone may not be directly controlled by genetics, as has been argued
(Bridges 1995; Trinkaus et al. 1994), it is likely that the sensitivity of bone to mechanical
usage and the processes that regulate bone adaptation on a localized level are genetically
controlled, and potentially influenced by non-genetic metabolic factors (e.g., hormones,
nutritional status, health, age, etc.) as well. Therefore, unknown metabolic and genetic
differences among study samples further limit interpretation of variation in bone

SU‘UCtUI‘C.

Analysis of the ‘empirical argument’

Acceptance of the conclusion of the empirical argument (Figure 5-3) rests on: 1)
The degree of support for premise 5 (including 5a — 5e), and 2) The existence of group
differences in cross-sectional geometric properties. Premise 5 states that variation in

cross-sectional properties reflects differences in patterns of physical activity. Because this

111

is the conclusion of the theoretical argument, which has already been analyzed, it is not
necessary to further discuss the degree of support for this proposition. However, the issue
of the existence of group differences (premise 6) warrants discussion. .While this would
seem to be a simple procedural matter of measuring the cross-sectional geometric
variables in question and statistically analyzing the data; it is actually somewhat
problematic in terms of the empirical argument put forth to support the specific
conclusions of bioarchaeological studies of long bone structural adaptation.

The premise that a difference in cross-sectional properties exists between the
samples under investigation (premise 6) is, in fact, what the individual research projects
were designed to demonstrate; it is empirical. The specific formulation of premise 6 (i.e.,
whether a difference exists and what the specific differences are) is derived solely from the
skeletal data, and it varies from study to study. Therefore, the specific conclusion, based
on premise 6, will also vary among studies.

Most of the bioarchaeological studies reviewed in chapter 3 had specific research
hypotheses that were tested via the analysis of cross-sectional geometric data. The crux of
hypothesis testing is determining whether the observed group difference in the variable of
interest can be attributed to chance or to some potential causal factor. “Ruling out the
null hypothesis is a necessary first step” (Graziano and Raulin 1989: 164) in establishing a
causal link, in this case between physical activity and cross-sectional geometry. To avoid
subjective bias when interpreting the meaning of group differences it is common practice
for researchers to employ statistical methods. Selection of an appropriate statistical
method is considered of paramount importance in any research design intended to
establish a cause or correlation (Graziano and Raulin 1989). For the purposes of this

dissertation, it is assumed that the researchers have selected an appropriate statistical

112

method (e.g., that the data do not violate the assumptions of the statistical test), and that
the reported results are accurate.

The most common statistical methods employed in bioarchaeological studies are
the Student’s t-test and one-way analysis of variance (AN OVA). Some studies employing
ANOVA follow with post hoc Tukey HSD tests for pair-wise group differences. These
tests allow the researcher to determine the statistical significance of the differences
between or among group means, which is reported as a significance level or P-value. It is
generally accepted, and indeed stated in the studies reviewed that a P-value of less than
0.05 is considered statistically significant (when reported, P-values less than 0.10 are
considered “near-significant”).

While most studies report significance levels, a surprising number of researchers
have drawn specific behavioral conclusions even in the near-total absence of statistically
significant (or even near-significant) group differences in the cross-sectional properties
ascertained by the study (e.g., Bridges et al. 2000; Brock and Ruff 1988; Fresia et al. 1990;
Larsen and Ruff 1990; Larsen et al. 1996; Ledger 1999; Ruff and Larsen 2001; Ruff et al.
1984). Much of the alleged support for mechanically based explanations of long bone
structural variation has come from these statistically non-significant group differences. I
would argue that without statistically significant differences between sample means, the
results of a comparative study do not provide support for any activity-related conclusion.
In fact, in these instances the most appropriate tack would seem to be accepting the null
hypothesis. Only Barondess (1998) approached his results in this manner, and ultimately
had to accept the null hypothesis for a large proportion of his comparisons.

There is an apparent tendency to over-interpret data, and draw conclusions where

there are none to be drawn. Statistically non-significant results are discussed as temporal

113

“trends” (e.g., Bridges ct al. 2000: 222; Brock and Ruff 1988: 120) despite the fact that the
inability to assign individuals to anything other than broad temporal categories precludes
the type of correlation analysis that would allow direct evaluation of the hypothesis that
bone structural properties changed over time. Furthermore, the discussion presented in
many papers frequently fails to distinguish between significant and non-significant
findings, which are ultimately given equal weight in arriving at the conclusions (e.g.,
Bridges et al. 2000; Brock and Ruff 1988; Fresia et al. 1990; Ruff and Larsen 1990). In
other words, by simply reading the discussion and conclusion sections of an article, one
would have no idea that most of the results were not statistically significant.

As an example of the insidious nature of this problem, when summarizing Brock
and Ruff’s (1988) publication on femoral structural changes in the prehistoric Southwest
for a review article, Larsen (1997: 217) wrote, “The ratio of Ix/Iy in the femur midshaft
shows a decline in both sexes, which also suggests a reduction in bending
stresses—particularly in the anteroposterior plane. . .These observations are consistent
with archaeological reconstructions of increasing sedentism with the shift to agriculture in
the American Southwest.” The problem with this interpretation is that when the results
of the original study are examined, it is clear that Brock and Ruff found no statistically
significant (or even near-significant) differences among their temporal/ cultural groups for
lx/Iy. In fact, the values for males over the three periods in question were virtually
identical (1.28, 1.27, 1.24) (Brock and Ruff 1988: l 19). It is difficult to understand how
these group differences can be attributed to anything other than chance (i.e., the null
hypothesis should be accepted).

This discussion is concluded with an example to illustrate the importance of

relying on statistically significant differences to interpret the results of a scientific study.

114

This discussion contrasts the conclusions drawn by Ruff et al. (1984), based on their own
statistical analysis, with two other possible interpretations of their data, consistent with the
premises of the overall argument as summarized in this chapter. Ruff et al. calculated the
following cross-sectional properties for groups representing hunter-gatherers and
agriculturalists: cortical area (GA), medullary area (MA), total subperiosteal area (TA),
Imax, Imin,J, Ix/Iy, and lmax/ 1min for femoral midshaft and subtrochanteric section
locations. After size-standardization the following statistically significant group
differences were observed: Male agriculturalists showed decreased midshaft MA and
Ix/Iy, and decreased subtrochanteric MA, TA, Imax,J, and [max/Imin. Females showed
only decreased TA at both femoral locations. The authors concluded, “In sum, our
results indicate that adoption of corn agriculture on the Georgia coast was very likely
associated in both sexes with a decrease in mechanical stress...” (ibid: 132, emphasis
added). Apparently, these authors place a great deal of weight on changes in the
subtrochanteric region of the male femur for interpretation of mechanical stress in both
sexes. There are at least two possible alternative interpretations using the authors’ own
reasoning (Ruff et al. 1984). l) Statistically significant decreases in several male
subtrochanteric cross-sectional properties suggest some highly localized mechanical
variable exerting forces on the hip and proximal femur, but with minimal effect on the
midshaft region. This unknown mechanical variable, possibly activity-related, did not
affect females in the same way. 2) While few statistically significant changes in cross-
sectional properties were observed, all were reduced in the agriculturalists. Moreover,
although not statistically significant, most of the other properties measured also showed
reductions, a few reaching near-significance. Additionally, bone length was decreased in

agriculturalists. Therefore, the pattern that emerges is of a nearly across-the-board

115

decrease in overall size and strength of the femur. These results are consistent with a
change in some systemic factor such as nutrition, which is plausible because dietary

differences are likely to have existed between these two groups.

The problems with premises 5a through 5e

In the brief history of bioarchaeological studies of cross-sectional geometry,
propositions 5a through 5e have first appeared as inductive generalizations, then as
premises, of an empirical argument. In other words, each of these premises is derived
from an empirical study that applied the conclusion of the theoretical argument (i.e.,
premise 5) to group comparisons of cross-sectional geometric variation, and each has
subsequently been utilized as a premise for interpreting cross-sectional geometry. A

critical analysis of proposition 5d is provided below.

Cross-sectional shape as an indicator of mobility

Ruff, Larsen and colleagues (Larsen 1997, 2002; Ruff and Hayes 1983a; Ruff and
Larsen 1990; Ruff et a1. 1984) have argued that the amount of cortical bone within the
cross-section (i.e., CA), is not as informative for interpreting behavior as differences in the
distribution of bone, measured by the second moments of area (i.e., I andj). The
reasoning is twofold. First, “bone curvature and other factors” (Ruff et a1. 1984: 126)
eccentrically load the long bones resulting in bending and torsional stresses (hence
increased bending and torsional rigidity, or I and J), rather than pure compressive or
tensile stresses. While it is true that the majority of strain experienced by long bones is
due to bending, it is not necessarily true that differences in CA are any more or less

informative about physical activity levels. For example, the aforementioned study of

116

humeral bilateral asymmetry (Jones et al. 1977; Ruff et al. 1994) demonstrated significant
differences in CA between the playing and non-playing arms of tennis players. Second,
it is suggested that GA “is more likely to be due to some systemic influence, such as
undernutrition” (Larsen 1997). Moreover, it is suggested that group differences in the level
of physical activity are revealed by variation in second moments of area (SMAs), while
differences in the apes of activities (i.e., specific mechanical loadings) are. reflected by the
shape of the cross-section, represented by ratios of SMAs (Ix/1y and Imax/ 1min). The
latter statement, relating cross-sectional shape to type of physical activity grew out of
some of the earliest biomechanical investigations of archaeological samples (Ruff 1987;
Ruff and Hayes 1983a; Ruff and Larsen 1990; Ruff et al. 1984), and has become,
perhaps, the most pivotal in drawing specific behavioral conclusions from the data. For
example, Ruff et a1. (1984: 134) have used this premise to infer that, “Relative reduction
in A-P bending loads in the midshaft femur could be brought about by relatively less
running and climbing and more walking.” The Ix/Iy ratio has earned the moniker “the
mobility index,” and it is considered an indicator of “long-distance travel” (Larsen 1997;
Ruff and Larsen 1990). The following assertions have been made about the Ix/Iy ratio
(Ruff 1987): 1) It is higher in more mobile groups, such as hunter-gatherers, 2) It is higher
in males than females, presumably because males are typically more mobile than females,
and 3) Sexual dimorphism of this index is greatest in hunter-gatherers where sex
differences in mobility are supposedly greatest. References to this index appear in the
discussion and conclusion of many studies and review articles when addressing cross-
sectional geometry of the femur (Bridges 1991, 1995; Larsen 1997; Larsen et al. 1995:
Ruff 1994; Ruff and Larsen 1990, 2001). However, there is evidence that the

relationships between physical activity and cross-sectional geometry are not as specific as

117

proposed. Some studies of skeletal adaptation to mechanical usage suggest that bones are
not necessarily more reinforced in the specific maximum plane of loading (e.g., Demes et
al. 1998; Demes et al. 2001). Furthermore, as previously mentioned, it is likely that group
differences in genetics and nutrition can produce localized changes in cross-sectional
properties, including shape. Therefore, the “sub” premises of premise 5 are also
confounded.

The hypothesis that group differences in cross-sectional shape reflect group
differences in macs of physical activities is based on the theoretical argument (Figure 5-2)
that physical activities, such as running, impose specific forces on the human skeleton, to
which the bones adapt according to Wolff’s Law. Figure 5-4 presents the propositions for
an argument, derived from Ruff (1987), that group differences in sexual dimorphism of
Ix/Iy reflect sex differences in the level of mobility engaged in by the group.

Figure 5-4. Argument for Ix/Iy as an indicator of sex differences in mobility.
Hypothesis Differences in sexual dimorphism of Ix/Iy among hunter-gatherers,

(Ruff 1987)

 

agriculturalists, and modern industrial populations are due to differences in

sexual division of labor (sex-specific activities) involving mobility.

Premise 1 Variation in Ix/Iy reflects differences in mobility.

(Premise 5)

Premise 2 Sex differences in mobility are greatest among hunter-gatherers.

Premise 3 Sex differences in mobility are less among agriculturalists.

Premise 4 Sex differences in mobility are least among industrial populations.

Premise 5 Sexual dimorphism of Ix/Iy is highest in hunter-gatherers, lowest in industrial

groups, and intermediate in agriculturalists.

Premise 6 Male Ix/Iy decreased over time, while female Ix/Iy did not.
Conclusion It is highly likely that Ix/Iy is a reflection of the relative mobility of males and
(Hypothesis

females in these populations.

 

 

accepted)

 

As with any inductive argument, the confidence one has in its conclusion is largely

118

based on the degree of support for (e.g., the cvidcntiary value of) the premises, and
whether the conclusion follows from the premises. Premise 1 is an assertion made by
Ruff and Hayes (1983a) based on analysis of Pecos Pueblo femora. Reasons to question
premise 1 are based on the analysis of the theoretical argument, and will be further
elaborated in chapter 6. Premises 2 through 4 are assertions made by Ruff, about the
relative mobility levels in three broad “subsistence” categories, hunter-gatherers,
agriculturalists, and modern industrial populations. The assumptions represented by
premises 1 through 4 determine the conclusion of the argument, but they are not themselves
strongly supported. For example, if instead, one were to assume that there were no sex
differences in degree of mobility among the three groups, then the conclusion of the
argument makes no sense. Furthermore, asjurmain (1999) has emphatically pointed out,
what these premises assume are the very questions biomechanical studies are supposed to
be answering. The premises are no better supported than the conclusion itself.
Therefore, the argument begs the question. Anthropologists do not know what the
relative mobility levels were in these three categories of populations, and there are no
independently verifiable means of determining relative mobility levels in prehistoric
populations, apart from presupposing the conclusion that group variation in Ix/Iy is an
indicator of mobility.

As with any research, there are concerns regarding the generalization of results to
other circumstances-"other populations, other time-periods (i.e., the external validity of
the study). Therefore, even if the assumptions about the degree of mobility in these
specific groups (i.e., study samples) are in fact correct, and who is to say they are not, it is
certainly not known to what extent they are representative of the broad categories in

which they have been placed. From this standpoint alone, it is questionable to apply the

119

conclusion of this study to the interpretation of variation in Ix/Iy in any other population.
Lastly, premises 5 and 6 make claims as to observed differences in Ix/ Iy among
hunter-gatherers, agriculturalists, and industrialists. The previous discussion of statistical
significance applies again in this argument. In fact, in this study, there was no
significance level reported. It is therefore unclear whether the observed pattern of
variation is attributable to chance, or to some independent variable, possibly mobility. In
sum, the problems with the premises cast doubt on the specific conclusion that Ix/Iy is
reflective sex differences in degree of mobility, and call into question any subsequent

generalization about the relationship between cross-sectional shape and type of activity.

Snmmm

This critical analysis has identified four problems that cast doubt on the
conclusions of the reconstructed arguments presented in this chapter: 1) Wolfl’s Law has
little cvidcntiary value as the primary theoretical premise in the argument used to infer
behavior from long bone cross-sectional geometry. 2) The argument does not make use
of all available relevant evidence; there is an ever-increasing body of literature dealing
with skeletal mechanobiology. As detailed in chapter 6, this literature points to the
complexities of functional adaptation to mechanical usage, which would seem to preclude
drawing conclusions about physical activity patterns from cross-sectional geometry.
Perhaps, more importantly, this literature calls attention to how much we do not yet
know about skeletal adaptation suggesting that, at this time, any conclusions are
premature. 3) The argument has failed to make a case for the ability to rule out or
adequately control for a number of confounding variables, such as age, body mass,

genetic constitution, timing of activity relative to skeletal maturity, nutrition, and other

120

metabolic factors. Consequently, while the arguments presented in this chapter represent
plausible lines of reasoning, there are several potential alternative explanations for the
data. Furthermore, there is no reason to rule out a priori the hypothesis that every group
comparison represents a unique “case study” for which there is a different “cause” or set
of causes responsible for observed differences in bone geometry. 4) Results of statistical
analyses often failed to reject the null hypothesis for many of the cross-sectional properties
ascertained (i.e., few statistically significant differences were found). In such instances,
conclusions regarding group differences in physical activity are spurious. The first three
of these problems are further developed in the following chapter by reviewing theoretical,
experimental, and clinical literature dealing with Wolff’s Law, functional adaptation, and

the non-mechanical confounding variables, genetics and nutrition.

121

CHAPTER G—WOLFF’S LAW, FUNCTIONAL ADAPTATION, AND
CONFOUNDIN G VARIABLES

It is quite possible for scientists to overlook the fact that observation is of data only: all the rest is
interpretation.

James K. Feibleman (1972)

Introduction

This chapter reviews theoretical, experimental, and clinical literature to evaluate
the problematic issues surrounding Wolff’s Law, functional adaptation, and confounding
variables. The information bears directly on the strength of the reconstructed argument
presented in chapter 5, and therefore on whether conclusions regarding specific
relationships between cross-sectional geometry and types or levels of physical activity

should be accepted or rejected.

The “Law” that is not a law

Wolff’s Law is proffered as the primary theoreticaljustification for inferring
activity patterns from bone geometry. In the previous chapter Larsen (2000: 52) was
quoted as claiming, “Such an overwhelming amount of experimental and other research
has accumulated in support of W olfl’s conclusions that the phenomenon is identified
simply as Wolfl’s Law.” However, examination of the literature on skeletal biology
reveals that rather than being well supported, “’0le Law is contentious. Its derivation is
based on a “false premise” (Cowin 2001b) and a fundamental misunderstanding of bone
biology (Dibbets 1992). Furthermore, the contemporary conception of VVolfI’s Law,
namely that bone is capable of sensing and adapting to mechanical stimuli via a cell-

based, self-regulating feedback mechanism is not what Wolff proposed. As illustrated

122

below, there are skeletal biologists who believe that Wolfl’s Law is a hindrance to the field
of mechanobiology.

Wolff’s Law is by no means universally accepted among skeletal biologists. One of
its harshest critics isJohn D. Currey, a renowned expert in the field of skeletal biology
(Currey 1979, 1981, 1984, 1984a, 1984b, 1987, 1995, 1997,2001, 2002; Currey and
Butler 1975). In a recent publication Currey (2002: 350) wrote, “Studies on bone
modeling have been bedevilled [sic] by Wolfl’s law. . .The unfortunate thing is that, for
many workers, it seems only necessary to show that bone is adapting, invoke Wolfl’s law,
and depart, conscious of a day’s work well done.” In an earlier article titled “Was VV’olfl
Correct?” Currey (1997) expressed his gratitude for Stephen Cowin’sl2 “demolition” of
the fundamental premise of W olfl’s Law, which relates to the alleged orientation of bony
trabeculae along principal stress trajectories. Currey (1997: 265) offered the following
reason for the persistent use of Wolff’s Law in skeletal biology: “. . .Wolff, with splendid
nineteenth century hubris, called his book ‘Das Gesetz der Transformation der
Knochen’, and ‘Gesetz’ means ‘law’ whichever way you look at it. Wolff set the
nomenclatural ground rules and we have obeyed them ever since.”

Currey is not alone in his criticism of Wolff’s Law. Huiskes (2000: 145) recently
wrote, “‘W'olfl’s Law’ is a question rather than a law. . .it is argued that it was the wrong
question, putting [biomechanicians and biologists] on the wrong foot.” Further, “While
the authority of W olfl’ 3 Law made biomechanicians concentrate on bone design, the
‘production technology’ was largely neglected. Yet, it is here that the real question is
found. . .Nature has found its design by trial and error, over time, by creating a metabolic

process responsive to environmental mechanical factors, which inherently dictate the

 

'2 Cowin, SC (1997) The false premise of Wolff’s law of trabecular architecture. Forma 12: 247-262.

123

design requirements. By wondering about what mathematical rules bone architecture
might be the answer to, we do not learn anything useful at all. The key to information is
in the metabolic process of bone production and maintenance” (ibid: 154). Cowin

(2001b) summarizes the status of VVolff’s Law:

There is no general agreement on the form of the rule that is to replace the
‘mathematical’ falsity or ambiguity associated with Wolff’s name (p. 30-1). . .Wolfl‘s law in
its present form is a mantra and not a statement that would pass muster in either the
biological sciences or the physical sciences. The use of the phrase ‘Wolff’s law’ is an
inappropriate suggestion that the concept is a concrete result generally accepted by the
scientific community. It implies a dignity that the concept has not achieved. In reality,
the exact meaning of Wolff 8 law, other than that it is a synonym for the functional

adaptation of osseous tissue, is uncertain. There is no generally accepted physiological

model for Wolff’s law (p. 30-13).

Given the foregoing comments on Wolff’s Law it is difficult to justify its continued
application to bioarchaeological research. However, these views are not entirely new.
“’0le Law has been subjected to a continual barrage of critical analyses by preeminent
skeletal biologists since long before its invocation in the first bioarchaeological study of
cross-sectional geometric variation. For example, in 1922 Triepel cited 20 objections to
W olff’s Law (Cowin 2001b; Evans 1953). In 1968, Enlow made the following
observation, “Wolff’s law of bone transformation has all but defied our best attempts at
meaningful explanation” (p. 803). With regard to experiments designed to verify “/0le
Law, Enlow commented:

Many workers presume that bone growth and biomechanics are virtually synonymous
and that mechanical stresses are the principal (perhaps the sole) factor governing the
course of skeletal morphogenesis (ibid: 804). . .. It would be overextending the
information. . .to presume that, because artificially applied stress (or the removal of
normal stress) can change the normal course of growth, in vivo stress thereby represents

the basic influence that regulates all bone growth and remodeling. Although not

124

precluding such a possibility, the experiment does not in itself verify it, since a multitude

of other variables and possible regulatory factors are not accounted for in its design (ibid:

806).

Just prior to publication of Ruff and Hayes’ major collaborative efforts on the Pecos
Pueblo archaeological population (Ruff and Hayes l983a, l983b), Roesler (1981: 33)
wrote, “These deductions of V‘Volff . . .contain so many objective errors and
misinterpretations ofelastomechanical principles that it is difficult to understand why his
work was not challenged by his contemporaries.” Several years later Carter (1984: S l 9)
wrote, “Although the basic concepts proposed by Roux and Wolff are now universally
accepted, the ‘mathematical laws’ relating bone remodeling to bone stresses or strains
have yet to be formulated. Furthermore, the mechanical/ biological control system or
systems that mediate these processes are unknown.” Huiskes et al. (1987) observed,
“‘Wolff’s Law’. . .as it is commonly understood, is not a scientific law in the traditional
sense, but rather a series ofqualitative observations and expectations. . .it is not a theory
suitable for quantitative predictions...” (p. 1 135). And in 1991 Bertram and Swartz
reviewed numerous experimental and clinical studies (including those cited by Bridges,
Larsen, Ruff and others) of W olff’s Law and concluded, “future progress in the study of
bone structure and function will require the removal of certain conceptual blinders
associated with a well-accepted but weakly supported paradigm [i.e., VVolfl’s Law] . . .the
general assessment of the functional and adaptive behaviour of bone that is often cited in
the introductory paragraphs of new research papers is actually based in part on poorly
supported presumed ‘facts’ and the interpretations of the results of previous work that
have been passed from paper to paper without receiving adequate critical appraisal” (p.

268). Bertram and Swartz’s critical review has subsequently been cited by many skeletal

biologists (Cowin 2001b; Currey 2002; Huiskes 1994; Martin et al. 1998) as providing “a
very useful antidote to a blind acceptance that [bone models adaptively to its mechanical

environment]” (Currey 2002: 341).

The origin of Wolfl’s Law

Julius Wolff was born in 1836 and died in 1902. He was a German orthopaedic
surgeon and anatomist who became interested in the relationship of form and function in
bone while obtaining his medical degree in Berlin during the 18503 (Forward by Peltier,
L. F. in Wolff 1988). Wolff is often depicted as an obsessed and arrogant scientist (Cowin
2001b; Currey 1997; Dibbets 1992; Roesler 1981, 1987) as well as a prolific and skillful
writer (Dibbets 1992). The 1892 publication of Das Gesetz der TranJormation a'er Knochen,
which literally translates as “The Law of Transformation of Bone,” represents Wolff’s
best-known work and the culmination of ideas—some his own, many borrowed—which
he had been developing in a series of publications beginning in 1869 (Roesler 1981). In
his famous monograph, Wolff presented what he considered to be irrefutable
mathematical proof of “the law according to which alterations of the internal architecture
clearly observed and following mathematical rules, as well as secondary alterations of the
external form of the bones following the same mathematical rules, occur as a
consequence of primary changes in the shape and stressing or in the stressing of the
bones” (Wolff 1892: l). Wolff (ibid: 83) wrote of “the action of a determined force of
nature” (i.e., “the transformation force”) that could alter the shape and architecture of
bone and had tremendous potential in orthopaedics for the correction of deformities. In
the 1986 translation of Das Gesetz der Trans/annation a'er Knochen, Maquet and Furlong

modified Wolff’s ori 'nal work to include a “more accurate account of bone rowth”
g1 g

126

(Dibbets 1992: 2) by substituting the word “remodelling” for “transformation”
throughout the book and in the title, which became 'l7ze Law (J Bone Remodelling.

Wolff’s Law has been succinctly summarized as: “Every change in the function of
a bone is followed by certain definite changes in internal architecture and external
conformation in accordance with mathematical laws” (Forward by Peltier, L. F. in Wolff
1988). However, as stated by Enlow (1968: 803), “Definitions or interpretations of
Wolff’s law. . .vary widely according to the nature of the application by different authors.”
One reason for a diversity of interpretations is that the current conception of Wolff’s Law
is an amalgamation of three general principles (Roesler 1981 , 1987), namely that the
internal architecture and external shape of a bone are: 1) based on mathematical laws (the
stress trajectorial theory of trabecular bone architecture), 2) determined by and adapt to
changes in the “function” of the bone (the concept of functional adaptation), and 3)
represent the minimum amount of material required for maximum strength (the principle
of optimization).

Wolff (1892: 6) credits anatomist G. H. Meyer with having “significantly
contributed to our knowledge of the internal architecture of bone,” and German bridge
and railway engineer Carl Culmann for having “discovered the mathematical significance
of this architecture.” In 1866 Culmannl3 attended a talk by Meyer and was supposedly
struck by the similarity between Meyer’s drawings of idealized trabecular architecture
and the lines of stress drawn for engineered constructions. Following Meyer’s talk
Culmann drew a “crane-like curved bar” (Roesler 1981: 28) with “the same contours as

those of the proximal extremity of the human femur...then asked his students to draw

 

‘3 Culmann founded the 19lh century science of graphic statics, which allowed complex systems of forces in
engineered structures to be simplified and rendered graphically as principal stress trajectories.

127

the. . .comprcssion and tension trajectories in the crane” (Wolff 1892: 6). Meyer originally
published these drawings in 1867 with the hypothesis that “the structure of cancellous
bone generally was determined by the direction of the principal stresses” (Roesler 1981:
28). Wolff (1892), however, believed that Meyer had failed to see the true mathematical
significance of Culmann’s drawings and sought to correct this oversight.

With the publication of Das Gesetz der Transformation der Knochen \Nolff intended to
prove that bones were designed using the same mathematical rules of construction that
engineers used for man-made structures. However, \Nolff was neither a mathematician
nor an engineer, and his “proof” merely consisted of series of statements derived from
detailed descriptions of the internal architecture of the proximal end of a normal femur as

seen in a series of longitudinal sections. According to Wolff (1892: 7) simple observation
of the femoral sections “demonstrated that the cancellous trabeculae regularly intersect
each other [and the surface of the bone] at right angles”just as the principal stress
trajectories did in the drawings of Culmann’s crane rendered by graphic statics. This
obse rvation, which is the cornerstone of Wolff’ 8 Law, is known as the “stress trajectorial
theory of trabecular bone architecture.” Wolff also interpreted the presence of the
med ullary cavity as evidence of the mathematical laws of trabecular architecture as “there
are no trajectories in the crane in the area which corresponds to the medullary cavity in
the bone” (1892: 20). Furthermore, W'olff believed these observations supported the idea
that bones are constructed with the minimal amount of material necessary to perform

the i1‘ function.

The medullary cavity as well as the spongy structure of the bone ends means saving
material while the bone presents sufficient size for the insertion of powerful muscles.
However, only after we have learnt from the mathematicians where material is necessary

and where it is superfluous, and only after Culmann. . .can we see that material is absent

128

not in a somewhat arbitrary way but in areas where it would be superfluous. . .. Nature
has built the bone as the engineer builds his bridges so that, as Culmann says, ‘the most

appropriate shape is obtained with a minimum expenditure of material’ (Wolff 1892: 2 1).

Though Wolff based his “law” on the apparent similarity of trabecular
orientations to stress trajectories in an engineered structure, he believed the external
shape of a bone followed mathematical rules of construction as well. Just as the final
curve of the crane (i.e., its external contour) represents a line connecting the end points of
all the other curves, “. . .the surface of the bone. . .has to be considered as the last or

limiting trabeculum of the whole trabecular system. . .. hence its shape, is derived from the

internal bony architecture” (Wolff 1988: 5).

Wolfl’s Errors

Criticisms of Wolff’s Law center around three issues: 1) Verification of the stress
trajectorial theory of trabecular bone architecture (Cowin 2001b; Roesler 1981, 1987), 2)

VVOlff’s notion of function (Dibbets 1992; Lee and Taylor 1999), and 3) \Nolff’s

misunderstanding of bone biology (Dibbets 1992).

me stress trajectorial theory of trabecular bone architecture

W olfl’s Law is predicated on the supposed “perfect mathematical
Correspondence” between the alignment of trabeculae in the proximal femur and the
StFess trajectories depicted in Culmann’s crane (Martin et al. 1998; Roesler 1981), which
to \’Volff implied that bones follow the same mathematical rules of construction as
engineered structures. Wolff believed bony trabeculae to be the physical embodiment of

S t regs trajectories (Cowin 20011)). His evidence was that trabeculae, like stress trajectories,

129

 

cross each other “at right angles in the most obvious way everywhere” (Wolff 1892: 19).
However, quantitative verification of this relationship has been elusive (Currey 2002;
Martin et a1. 1998). While there is some experimental support for the trajectorial theory
(e.g., Hayes and Snyder 1981; Huiskes 2000), Huiskes (2000: 154) claims, “The
correspondence between trabecular architecture and stress trajectories. . .is circumstantial,
not causal.” Cowin (2001b) refers to trajectorial theory as the “false premise” of Wolff’s
Law—“the law compares things that appear similar but are not” (p. 30-1).
Skeletal biologists have noted five problems with the trajectorial theory (Cowin

200 l b; Dibbets 1992; Evans 1953; Martin et a1. 1998; Roesler 1981). 1) Trabeculae do
not always cross at right angles as was asserted by Wolff, whereas principal stress
trajectories must, by definition. Cowin (2001b) points out, it is actually irrelevant whether
trabeculae cross each other at right angles or not, because regardless of the angle at which
they cross, the stress trajectories associated with each trabecula do cross at right angles.
Whereas the trajectorial theory assumes a one-to-one correspondence between a

tr at>ecula and a stress trajectory, each trabecula is actually associated with an infinite
nuInber of stress trajectories (ibid). 2) The principal stress trajectories in Culmann’s crane
Were, based on the assumption of an elastically isotropic, homogenous, and more
importantly, according to Cowin (2001b), a continuous (containing no gaps) material,
none of which is true of cancellous bone. Stress trajectories do not exist in a
diSContinuous material (Cowin 2001b; Martin et al. 1998). 3) The trajectories

re presented in Culmann’s crane were based on “only one static loading. . .while it is
QbVious that the living femur is subjected to many types of loading” (Roesler 1981: 38).
Fu rthermore, the loads placed on the femur are unknown, and therefore can only be

ambiguously or unrealistically specified (Cowin 2001b). 4) Culmann’s crane depicted a

130

m

 

finite number of stress trajectories, whereas in reality, “there are an infinite number of
stress trajectories between any two points in the model” (Cowin 2001b: 30-10). The
trajectories that were depicted in Culmann’s crane, being “solutions of a differential
equation” were based on an arbitrary choice of initial values; “There is no obvious reason
to draw only those trajectories that Culmann did” (Roesler 1981: 38). If all stress
trajectories were graphically depicted, the resultant figure would not contain individual
curves; it would instead be completely filled in. In other words, if bone really were the
embodiment of stress trajectories there would be no cancellous tissue, only solid compact
bone. 5) By reconstructing Culmann’s crane, Roesler (1981) demonstrated that the stress
trajectories in the crane were likely based on a straight rectangular cantilever, which was
subsequently curved to approximate the shape of the proximal femur without altering the
stress distribution. These problems led Roesler (1981: 38) to conclude, “Whatever the
proof for a trajectorial structure of the cancellous bone in the proximal part of the human

femur may be, it is not Culmann’s crane.”

Wow; Roux, and the concept of function

Roesler (1987: 1026) wrote, “The most surprising observation which can be made
after reading the old literature on bones. . .is that there never has been the least doubt
about the interrelation between structure and function of bone.” In fact Galileo is
credited with having first recognized the relationship between bone shape and
mechanical function (Martin et al. 1998). Wolff believed that his observations of
trabecular patterns in both normal and pathological specimens proved that the internal
architecture and the external shape of bones were a result of their function. However,

according to Dibbets (1992: 8), Wolff’s concept of function was of“static. . .constraints to

131

be met” rather than “our present day notion of a dynamic process involving action.” Lee
and Taylor (1999) echoed this view. According to Wolff (1892: 83; emphasis added),
“The shape of bone is determined only by the static stressing for which bone is pre-
programmed or, in other words, by its function. . .. Only static use/illness and necessity or
static superfluig) determine the existence and location of every bony element and,
consequently, of the overall shape of the bone.” The equation of function with static
forces or static stressing occurs repeatedly throughout the text, and at one point Wolff
asserted, “. . .nothing much would be changed. . .by speaking of static shape instead of
functional shape” (ibid: 82). Wolff’s belief that the normal architecture of bone is related
to “function” yet at the same time, inherited (“it preexists in the foetus”; 1892: 14) further
distinguishes his view of function from the current one, and also suggests a certain
ambivalence toward the meaning of the word, e.g., “the femur presents its functional
architecture long before the child’s first attempts at standing and walking” (1892: 71).
Wolff failed to understand that it was the dynamic process of loading and
unloading that stimulated bone adaptation. This led Lee and Taylor (1999) to propose
that Wolff’s Law be renamed “Roux’s Law” in recognition of the fact that it was Wilhelm
Roux, notJulius Wolff, who first introduced the modern concept of function. It was also
Roux, not Wolfl, who first described functional adaptation as a cell-based, self-regulating
biological process involving both bone resorption and formation in response to a
mechanical stimulus (Roesler 1987: 1030). Nevertheless, as stated by Roesler (1987:
1027), “Wolff’s doctrine. . .became nearly inextricably involved with Roux’s principle of

functional adaptation.”

132

Wolfl’s misunderstanding of bone biology

According to Dibbets (1992), Wolff was mistaken on two key points with regard to
bone biology. First, Wolff (1892) maintained that cortical bone was merely “compressed”
cancellous bone. He believed that normal trabecular architecture was latent in the
cartilaginous anlage, and that the normal trabecular pattern “unfolded itself” from its
compressed state in compact bone through the process of expansion (Dibbets 1992: 8).
Hence, the second of his misconceptions was that bones grow by expansion, i.e.,
interstitially. He steadfastly rejected the position held by most of his contemporaries_“l
am almost the only one so far to struggle against. ..[the] theory of growth by apposition
and resorption only” (Wolff 1892: 91), which, in an earlier publication, he referred to as
“‘a theory of permanent structural upheaval’” (quoted in Roesler 1981: 35). Clearly, this
state of “permanent structural upheaval” closely depicts what Frost has termed
remodeling. Regarding this point, Dibbets (199217) wrote, “Wolff was so successful in
attacking his opponents that he alone may be held responsible for carrying the incorrect
notion of interstitial bone growth through the latter quarter of the nineteenth century into
the twentieth century. Thus, it is obvious that the translation of the title”. . .should not
include the term ‘remodeling’.” Although to be fair, translators Maquet and Furlong were
probably using “remodeling” in the general sense of adaptation, not as the activation-

resorption-formation sequence proposed by Frost (1986).

Functional adaptation

The preceding discussion leads to the conclusion that W olff’ 5 Law is not a

 

'4 Referring to Maquet and Furlong’s 1986 translation of Das Gesetz der Trans/finnation der Knochen as 77w Law
of Bone Remodelling.

133

scientific law based on mathematical rules, but rather a synonym for the general concept
of functional adaptation. At this point it would be reasonable to ask, if current usage of
the phrase “Wolff’s Law” implies functional adaptation, then is the problem simply with
the name “Wolff” or the word “Law”? The answer is, as eloquently stated by Cowin
(2001b: 30-14) that, “the problem is not the lack of a reasonable terminology, but the lack

of an acceptable theory for the functional adaptation of bone.”

Problems linking physical activity to cross-sectional geometry

The concept that bone functionally adapts to mechanical loading is vague,
particularly with respect to establishing a causal link between physical activity and specific
patterns of cross-sectional geometric variation. Rather than specifies, the literature on
functional adaptation is replete with conflicting and equivocal information (Bouvier 1985;
Burr 1992; Currey 2002; Martin et al. 1998). Furthermore, many details regarding the
processes involved are not known (van der Meulen and Huiskes 2002). Currey (2002:
353) cautions that there is a difference between theoretical predictions and reality, and
states that, “In real life, the situation is not very clear cut.” In an attempt to “(escape) the
vagueness and confusion associated with the term ‘Wolff’s law’” Martin et al. (1998: 230)
proposed the “mechanical adaptability hypothesis,” which “states that bone structure is
regulated so as to minimize fracture risk and bone mass while simultaneously optimizing
stiffness.” However, there is no consensus on what property of bone, if any, is Optimized
by an adaptive response (Biewener and Bertram 1986; Currey 2002; Hart 2001; Huiskes
2000; Lanyon 1984; Loitz and Zernicke 1992; Martin et al. 1998; Rubin 1984).
Reflecting a critical departure from the implied certainty of “W olff’s Law” the authors

emphasized that they were calling their idea “a hypothesis because we are not sure it is

134

true or how it works” (ibid: 230). In a recent survey article, van dcr Meulen and Huiskes
(2002) summarized the status of functional adaptation research:

Adaptation experiments examining skeletal mechanobiology have been performed for
more than a century and have examined the influence of both increasing and decreasing
the habitual loads applied to the skeleton. Yet, despite the multitude of studies that have
been completed, there are still many unanswered questions regarding the skeletal
response to normal, perturbed and pathological mechanical loading (p. 403)....
Answering these questions will require carefully controlled experiments with well-

characterized mechanical environments (p. 405).

As discussed in chapter 2 it is still not known which mechanical loading variables,
or combination thereof, are osteogenic (e.g., strain magnitude, strain rate, strain
distribution, strain frequency) (F ritton et al. 2000). Judex and Zernicke (2000: 158) point
out that, “Few studies. . .have associated the exercise-related milieu with induced
morphological changes or a lack of them.” Bone does not adapt to physical activity per
se, it adapts to some quality of the mechanical load generated by physical activity.
Therefore, without an understanding of the stimulusresponsible for an adaptive response,
any attempt at associating specific cross-sectional geometries with types or levels of
physical activity is purely speculative and not grounded in solid scientific evidence. For
example, it could be that the type of activity, not the intensity of the activity determines
whether an adaptive response will occur. Frost (1997a, 1997b) suggests the reason
marathon runners have smaller, weaker bones than weight lifters is because weight-lifting
increases muscle strength, which results in a much greater peak strain magnitude. In
contrast, he suggests, running, no matter the distance, does not produce strain
magnitudes that reach the MES for modeling, particularly in adults. Muscle mass is

strongly correlated with bone mineral density (BMD) (Arden and Spector 1997). Weight

lifters have bigger, stronger muscles than runners, and muscle contractions produce the
greatest, non-traumatic forces experienced by bones (Burr 1997; Frost 1997a, 1997b).
Parenthetically, an additional complicating factor in the attribution of bone geometry to
physical activity lies in the estimation that up to 50% of the variability in muscle mass is
genetically determined (Arden and Spector 1997). Frost distinguishes between activities
that require endurance and those that require increased strength, pointing out that the
latter, and not the former are more likely to result in skeletal structural adaptations. An
activity may be repetitive and it may be physically exhausting, but this does not mean that
it will increase bone mass or alter geometry.

Research supports strain magnitude as a critical factor for inducing functional
adaptation, but only once a certain threshold is reached (Fritton et al. 2000; Frost 1997a,
1997b; Hsieh et al. 2001; Robling and Turner 2002; Rubin and Lanyon 1985). Some
suggest this threshold may vary within and among bones, and possibly between species
(Carter 1984; Currey 2002; Fritton et al. 2000; Turner 1999; Hseih et al. 2001). The
threshold may also vary among individuals (Robling and Turner 2002). Others suggest
(e.g., Frost 1997a) much greater uniformity of the strain threshold. Although the typical
peak strains experienced by bones during normal activities tend to be similar across a
variety of animal species (Burr et al. 1996; Currey 2002; Fritton and Rubin 2001; Rubin
and Lanyon 1984a), during a specific activity such as running, the values for the strain
variables experienced by the skeleton are likely to be different for different bones, and are
non-uniformly distributed throughout the cortex, varying within a cross-section, as well as
along the length of the shaft (Bertram and Swartz 1991; Carter 1984; Fritton et al. 2000;

Judex et al. 1997; Lanyon and Baggott 1976; Loitz and Zernicke 1992). Strains resulting

from the same activity also vary among individuals (e.g., Demes et al. 1998; Milgrom et

136

al. 2000). Consequently, strain measurements obtained from one location on one bone
cannot necessarily be used to predict specific adaptive responses in other bones or other
subjects (Burr et a1. 1996; Milgrom et al. 2000).

It is generally accepted that there is a range of activities unlikely to initiate an
adaptive response because the peak strain magnitudes they generate fall within the
“adapted window” (Bertram and Swartz 1991; Burr 1992, 1997; Carter 1984; Carter et
a1. 1980; Currey 2002; Frost 1996, 1997a, 1997b; Martin et al. 1998). However, in
addition to strain magnitude, many studies also indicate that strain rate is a key variable
(Burr 1992; Hseih et al. 2001; Lanyon 1981; Milgrom et a1. 2000). Other research
suggests that only mechanical loadings that produce an atypical strain distribution are
osteogenic (e.g., Lanyon et al. 1982; Milgrom et a1. 2000; Rubin and Lanyon 1985). A
study byJudex and Zemicke (2000: 153) supports the hypothesis that the activity in
question must generate a “mechanical milieu” that differs “substantially from the habitual
milieu to induce significant adaptations.” In addition, Rubin et al. (1992: 89) have noted
that, “Although the predominant responsibility of the skeleton may be to withstand the
extremes of physical activity, it does not necessarily follow that the strains generated
during this activity are what drive the skeleton’s morphology.” Fritton et al. (2000) found
that the peak strains resulting from normal locomotor activities occur rather infrequently;
however, during the same period of time bone experienced numerous low magnitude
strains. These workers found that bone sensitivity to strain seems to be frequency-
dependent, and suggest that low-magnitude high-frequency signals might play a
significant role in bone adaptation (F ritton et al. 2000; Rubin et al. 1992).

Another complication in linking physical activity to specific long bone cross-

sectional geometries is that the external loadings applied to bones and the consequent

137

stress and strain magnitudes, rates, distributions, and frequencies associated with specific
activities are, for the most part, uncharacterized (Lanyon and Baggott 1976; Lanyon et al.
1979). Ruff and Hayes (1983a: 372) have noted “the importance of considering the total
loading configuration when attempting to interpret structural variation.” Ruff and Hayes
attempted to relate locational cross-sectional geometric variation in the femur and tibia to
predictions of in vivo loading based on theoretical and in vitro loading models. However,
the loads placed on long bones during normal activities are extremely complex and time
variable (Burr et al. 1996; Fritton and Rubin 2001; Lanyon and Rubin 1984; Rubin et a1.
1990; van der Meulen et al. 1993). Consequently, theoretical estimates of in vivo loading
may have limited usefulness in predicting bone strain (Carter et a1. 1980; Currey 2002;
Demes et al. 1998, 2001). Bouvier (1985: 241) warns that, “relating stress patterns to
bone morphology without first examining in vivo strain distributions may lead to
erroneous conclusions.” Because measurement of strain parameters during activity
requires attaching strain gauges directly to living bone, this type of invasive experiment is
not often performed on humans. To date, in vivo strain of a long bone diaphysis has only
been recorded for a single mid-diaphyseal location on the medial tibia in seven human
subjects (e.g., Burr et al. 1996; Milgrom et al. 2000). Of the major long bones, the tibia is
the most amenable to strain measurement (Milgrom et al. 2000) because its anteromedial
surface is relatively superficial while the shafts of the femur and the humerus are
surrounded by muscles.

Principal compressive and tensile strains were recorded from the midshaft tibia of
a single, middle-aged male subject during a variety of “vigorous activities” including
walking with and without a weighted pack, jogging and sprinting on level and inclined

surfaces, and “zigzag” running uphill and downhill (Burr et al. 1996). For the most part,

138

strain magnitude and strain rate varied significantly between walking and running;
however, the overall patterns of variability were complex. The highest strains
experienced by the tibia were shear strains (calculated as the difference between the peak
compressive and tensile strains) generated during uphill and downhill zigzag running;
these approached 2000ue. Shear strains were also high during level sprinting and uphill
running. During most activities peak compressive and tensile strains were relatively low,
ranging between 350 and 1000u8; only peak compressive strain exceeded 1000]“: and,
even then, only during zigzag running. Carrying a pack during walking made little
difference in peak strain magnitude or rate. Strain rates were significantly higher during
running than walking; this was particularly true for shear strain rates during sprinting on
a level surface and zigzag running downhill. The experiment by Burr et al. (1996)
suggests that the more vigorous the locomotor activity, the greater the strain magnitude
and rate. This general conclusion was supported by the findings of Milgrom et al. (2000).
The results also generally support Frost’s (1997a, 1997b) hypothesis that, in adult human
bone, running does not typically generate peak compressive and tensile strains that
exceed the modeling threshold of 1000u€ required for bone formation. However, as
pointed out by Frost (1997b), the role of shear strain in initiating an osteogenic response
requires further investigation.

Ruff and Hayes (1983a: 371) have claimed that, “the optimum cross-sectional
‘shape’ of a bone subjected only to bending in one plane would be to place as much bone
as far as possible from the neutral axis of bending.” They have also asserted that,
“differences in bone ‘shape’ can be viewed largely as adaptations which reduce stress
and/ or strain in bones under in-vivo loading conditions” (ibid: 373). These statements

reflect a confounding of two separate concepts, the second of which may or may not

139

follow from the first: 1) Bones with a more outward distribution of bone tissue in a single
plane are more resistant to bending in that plane (i.e., they have a greater bending rigidity
in that plane). 2) Increased bending loads within a single plane produce an adaptive
response that results in a greater outward distribution of bone tissue, in order to minimize
bending in that plane. This requires bones to be more reinforced where they are
maximally loaded. Whether optimal or not, studies suggest that this is not how bones are
constructed (Demes et al. 1998, 2001; Ohman and Lovejoy 2001). Demes et al. (1998)
studied ulnar deformation during locomotion in rhesus macaques. From observations of
macaque locomotion, the authors hypothesized that macaque ulnae experience anterior-
posterior bending. In contrast, results of strain gauge recordings indicated that the ulnae
primarily experienced bending in the medial-lateral plane. In addition, the authors found
that macaque ulnae are not reinforced in the plane of maximum bending; they are instead
more expanded in an anterior-posterior direction. This led them to suggest that, “Bone
cross-sectional geometry may not be a simple mirror reflection of functional loads” (ibid:
96). Similar results were reported for macaque tibiae (Demes et al. 2001). The authors
“caution against broad behavioral conclusions derived from long bone cross-sectional
shapes” (ibid: 264).

Other research indicates that specific sites of bone formation do not correspond to
predictions based on the hypothesis that bone adapts to minimize the strains it
experiences during loading. Following a three-week experimental period in which
mature roosters ran on a treadmillJudex et al. (1997) observed activation of periosteal
surfaces in five of the eight exercised animals. However, the sites of periosteal activation
did not correlate with either peak strain magnitude or strain rate, interpreted as

suggesting that bone is deposited “where mechanical integrity is least challenged” (Judex

140

et a1. 1997: 1742). To explain these and other similar findings (Rubin 1984) it has been
hypothesized that the “goal” of bone adaptation, rather than minimizing strain, is to
maintain a certain “type of strain” (Rubin et al. 1990). It is reported that over 80% of
strain in long bones is due to bending (Rubin et al. 1990), and as obvious from the above
quote from Ruff and Hayes (l983a), it is assumed that bone architecture is optimized to
reduce bending by increasing rigidity. However, as discussed by Bertram and Biewener
(1988), rather than minimizing bending, normal longitudinal curvature often accentuates
it in a particular direction. Likewise, an elliptical cross-sectional shape in which bending
rigidity is decreased in one direction can control the direction of bending through
unequal second moments of area. Bertram and Biewener (1988) have suggested that
bone construction “sacrifices strength for load predictability.” This would be
advantageous for bones that experience “a highly variable loading environment” (ibid:
91), as limb bones do during locomotion. This hypothesis suggests an intriguing
possibility: differences in cross-sectional geometry, particularly ratios of second moments
of area (e.g., Ix/Iy), may be related to differences in the amount of relative curvature

among bones from different populations.

Functional adaptation: an “umbrella term” for a plurality of efi'ects

Finding therapies for the prevention or treatment of osteoporosis has been a
driving force behind much of the research conducted on bone adaptation. Functional
adaptation of bone to both increased and decreased mechanical loads has been studied
experimentally and clinically. Experiments designed to study the effects of increased
mechanical loading have utilized surgical (e.g., osteotomy) and non-surgical loading (e.g.,

exercise regimens) techniques in a variety of animal models (Biewener et al. 1983;

141

Bravenboer et al. 2001; Buhl et al. 2001; Carter 1981; Carter et al. 1980; Chamay and
Tschantz 1972; Churches et a1. 1979; Cullen et al. 2000; Forwood and Parker 1987;Jee
et al. 1991; Kimura et al. 1979; Lanyon 1980; Lanyon and Baggott 1976; Lanyon and
Bourn 1979; Lanyon et al. 1979, 1982; Raab et al. 1990; Robling et al. 2000; Steinberg
and Truetta 1981; Tanck et al. 2000, 2001). Studies of decreased loading in animals have
employed immobilization (casting, hind-limb suspension, space flight, simulated
weightlessness) and surgical techniques (neurectomy) (Abram et al. 1988; Amtmann 1974;
Jaworski et al. 1980; Martin 1990; Shaw et al. 1987; Uhthoff andJaworski 1978; Wunder
et al. 1979). Research projects involving human subjects employ either a cross-sectional,
or less frequently, a longitudinal design in which bone mass, density, and occasionally
geometry are compared between exercise and control groups or between athletes and
non-athletes (Davee et a1. 1990; Gleeson et al. 1990; Hasegawa et al. 2001;Jonsson et al.
1992; Kirk et al. 1989; Sandler et a1. 1987). Clinical studies following permanent
paralysis or injury resulting in temporary immobilization have provided information
about decreased loading in humans (Kiratli et al. 2000), as have studies of astronauts
following prolonged weightlessness in space (Bikle and Halloran 1999).

Studies of functional adaptation are far too numerous to review here. Moreover,
the results are extremely varied, and difficult to interpret (Bertram and Swartz 1991;
Currey 2002; Martin et al. 1998). Bertram and Swartz (1991) provide an excellent
critical analysis of numerous studies conducted prior to 1990 in which they suggest that
the various phenomena reported in the literature and attributed to VVolff’s Law actually
represent a “plurality” of effects. These authors suggest that workers have ignored
alternative explanations for their findings, particularly those related to “complications

arising from indirect effects of the investigative procedures on other aspects of the

142

organism’s physiology” (p. 246). For example, bone loss during spaceflight has been
interpreted as a direct response of bone to mechanical unloading, and is frequently cited
as evidence for Wolff’s Law (e.g., Bridges 1989). However, Bertram and Swartz suggest
alternative possibilities including a generalized, systemic response to unloading, or other
physiological effects, such as bone loss due to increased corticosteroid secretion from
physiological stress. In addition, the extreme non-physiological loading imposed by
osteotomy and external bending devices may not produce results that are equivalent to
the effects of normal, voluntary activities on bone (e.g., Bertram and Swartz 1991; Martin
et al. 1998). It seems likely that the studies most relevant to inferring behavior from cross-
sectional geometry would be those involving activities and loads within the normal
physiological repertoire of the animal. Much of the strongest evidence for functional
adaptation comes from comparative studies of athletes with sedentary controls, or studies
of upper limb bilateral asymmetry in people who play racquet sports (e.g.,Jones et al.
1977; Bass et al. 2002). Although professional athletic activities are within the normal
physiological capabilities of humans, they often fall at the upper limit. Seeman (2002:
375) refers to the “Olympian effort” required to generate such extreme structural
variation in the human skeleton, and suggests that these studies tell us “what is possible,
not what is feasible in day-to-day life.” Extrapolation from these studies to prehistoric
societies actually presumes that the behaviors and lifestyles of past peoples reflect this
“Olympian effort,” which again, as pointed out byJurmain (1999), is what needs to be
demonstrated, and should not be assumed.

Other researchers reiterate Bertram and Swartz’s (1991) skepticism: “While bone
‘adaptation’ to loading is a long-standing concept in bone physiology, researchers may

sometimes be too willing to accept this paradigm as an exclusive explanation of in viz'o

143

tissue responses during experiments, while overlooking confounding variables, alternative
(non-mechanical) explanations, and the possibility that different types of bone (e.g., woven
bone, Haversian bone, plexiform bone) may have different sensitivities to loading under
healing vs. quiescent conditions” (Brunski 1999: 99). Loitz and Zemicke (1992: 14), on
the other hand, suggest that the lack of consistent experimental results “may reflect
complex interactions between bone remodelling and exercise intensity, animal species,

and skeletal age.”

Age efl'ects on functional adaptation

In bioarchaeological studies of past human behavior it is frequently inferred that a
larger group mean for certain geometric properties, such as the polar second moment of
area (I; representing torsional rigidity or “average” bending rigidity), implies a greater
level of physical activity, i.e., a more strenuous lifestyle (e.g., Larsen et al. 1995).
However, this dissertation asserts that the effect of the “timing” of physical activity within
an individual’s lifespan confounds the interpretation of cross-sectional geometric
properties. “Timing” is a confounding variable because it may differ among
archaeological population samples, yet it cannot be determined, cannot be controlled for,
cannot be ruled out, and could potentially provide an alternative explanation for the
results. In other words, conclusions regarding greater or lesser levels of activity based on
cross-sectional geometry may be false because group differences in the mean values of
cross-sectional geometric properties could in fact reflect population (e.g., cultural)
differences in the commencement of physically demanding activities. Alternatively, adult
cross-sectional geometry could reflect real differences in the activity levels of younger

members of the population, but have little to do with differences in lifestyle resulting from

144

major transitions in mobility patterns or subsistence strategies. The geometry of an adult
long bone is largely dependent on activities performed earlier in life. Research has shown
that physical activity has age-dependent, site-specific and variable effects on cross-
sectional geometry due to differential bone formation and resorption activity on periosteal

and endocortical surfaces.

Mechanical adaptation in immature versus mature individuals

It is considered an established phenomenon that the bones of children and
adolescents are more responsive to mechanical loading than those of adults (Bass et al.
1998, 2002; Bertram and Swartz 1991; Bradney et al. 1998; Duncan et al. 2002; Frost
1986, 1997a, 1997b; Forwood and Burr 1993; Haapasalo et al. 1998, 2000; Heinonen et
al. 2000;Jee 2001; Kannus et al. 1995; Khan et a1. 2000; Kontulainen et al. 2003;
Lieberman et a1. 2001; Martin et a1. 1998; Milgrom et al. 2000; Mosley and Lanyon
2002; Nordstrom et al. 1998; Ogden 2000; Petit et a1. 2002; Ruff et al. 1994; Seeman
2002; Slemenda et al. 1991; Zanker et al. 2003). “Growth” is said to be “the single most
opportune time to modify the mass and geometry of the skeleton” (Seeman 2002: 374),
and according to Frost (1997b: 183), “‘Vigorous’ voluntary activities help to increase bone
mass and strength in children, but in aging adults they seem to minimize bone losses
instead of increasing bone mass.” This is also the interpretation of Forwood and Burr
(1993). The difference in response to mechanical loading may not relate to a diflerential
sensitivity to increased loading between the bones of young versus old individuals
(Jarvinen et a1. 2003; Klein-Nulend et al. 2002). Jarvinen et al. (2003) compared the
response of the femoral neck in young and old rats to treadmill running exercise and

subsequent deconditioning (cessation of activity). While both age groups showed

145

increases in bone mass and bone mineral density resulting in increased bone strength, the
bones of young rats exhibited dramatic changes in geometric properties, whereas old rats
exhibited an increase in bone mineral density with no corresponding increase in
geometric properties. Both young and old rats lost a similar amount of bone as a result of
deconditioning. The authors concluded that while both age groups are capable of
adapting to mechanical loading (i.e., the sensitivity to mechanical loading was not
reduced in the older rats), the adaptive mechanisms were different. This conclusion is
supported by experimental findings that the sensitivity of cultured human osteoblastic-
lineage cells to mechanical deformation was not related to donor age (Klein-Nuland et al.
2002)

In contrast to the strong support for activity-induced adaptation prior to skeletal
maturity, Bertram and Swartz (1991: 246) claimed that there was no unequivocal
evidence for bone adaptation in the “healthy mature appendicular skeleton.” Forwood
and Burr (1993) reiterated this skepticism and pointed out the failure of studies “to
consider important confounding variables” (p. 97). Seeman (2002: 374) has also recently
expressed similar reservations:

There is little replicated and methodologically sound evidence to suggest that exercise
during young adulthood, peri-menopause, late adulthood, or old age modifies bone size,
prevents bone loss, or restores bone mass, architecture, or strength. Consistency in results
is lacking; some studies suggest bone loss is prevented by exercise and others suggest bone
loss is not prevented or is increased. The increase in aBMD of a few percentage points
reported in some studies is probably due to a reduction in the size of the reversible
remodeling space. There is little, if any, evidence of changes in bone tissue mass beyond
that produced by reducing the remodeling space. There is no evidence that exercise in
adults increases cortical thickness by increasing periosteal apposition, reducing

endocortical resorption, or increasing endocortical bone formation.

146

The reasons for the apparent age-related difference in response to mechanical
loading are not yet known, although there is speculation about the osteogenic role of
growth and sex hormones in older children and adolescents (Bass et al. 1998; Khan et al.
2000; Seeman 2002). These hormones rise during early puberty, and the level of growth
hormone falls immediately after the onset of menarche in females (Khan et al. 2000), thus
providing physiological evidence for an increased osteogenic response around puberty.
There are also fundamental differences in the mechanical properties of rapidly growing
versus mature bone (Currey and Butler 1975; Currey 2001; Martin et a1. 1998). For
example, rapidly growing bones are less mineralized and more compliant due to the
mineralization lag following bone formation. Furthermore, modeling is thought to either
cease or become greatly reduced at skeletal maturity, and as Frost (1997b: 183) stated,
“Modeling, not BMU-based remodeling, determines size and architecture.”

Frost’s (1997a) hypothesis, in keeping with the “mechanostat,” is that during
childhood and adolescence relatively rapid increases in body mass, muscle strength, and
physical activity outpace the “sluggish” ability for bones to adapt. Consequently, strains
experienced by rapidly growing bone in young individuals are greater, and are more
likely to exceed the modeling threshold, thus stimulating formation drifts on periosteal or
endosteal bone envelopes. This would be true regardless of whether there are other age-
related differences in adaptation. Once changes in body mass and muscle strength have
stabilized, which occurs at some point in the late teens and early twenties, the bones
“catch-up,” and strains are reduced below the modeling threshold (i.e., they fall within
the “adapted window”).

Studies of the response ofimmature (i.e., growing) bone to exercise using animal

models, such as pigs (Woo et al. 1981), rats (Forwood and Parker 1987; Li et al. 1991;

147

Robling et al. 2000), and roosters (Biewener and Bertram 1994;Judcx and Zemicke
2000; Matsuda et al. 1986) have produced variable results. Some studies show positive
effects on bone mass and strength, including increases in cross-sectional second moments
of area (Biewener and Bertram 1994; Woo et al. 1981). However, several studies have
demonstrated negative consequences of intense activity on growing bone including
reduced long bone length (Forwood and Parker 1987; Li et al. 1991; Matsuda et al. 1986;
Robling et al. 2000), reduced periosteal diameter (Matsuda et al. 1986), reduced material
properties (Forwood and Parker 1987; Matsuda et al. 1986), and reduced geometric
properties (Li et a1. 1991). Forwood and Burr (1993) have suggested that there may be a
threshold of intensity or duration, above which exercise is detrimental to growing bone.
Still other studies have reported no significant difference between the bones of exercised
animals and controls (Judex and Zemicke 2000; Raab et al. 1990). Studies that show
either a negative response or a lack of response to increased physical activity suggest that
caution should be exercised when interpreting lower values for cross-sectional geometric
properties as evidence for reduced activity.

Woo et al. (1981) is the most frequently cited bone adaptation experiment among
the literature reviewed in chapter 3 (Bridges 1989; Fresia et a1. 1990; Larsen 1997; Ledger
2000; Mays 1999; Ruff and Hayes 1983a; Ruff and Larsen 1990; Ruff et al. 1993, 1994).
In this classic study of “Wolff’s Law”, five pigs were subjected to a regimen of treadmill
running for twelve months. Four additional pigs served as unexercised controls. All pigs
were one-year—old and considered sexually immature at the beginning of the experiment.
At the end of the experiment, there was no difference between exercised and control pigs
in the external (i.e., periosteal) diameter of the femoral midshaft; however, the endosteal

diameter was significantly reduced in exercised pigs. As a result, exercised pigs exhibited

148

significantly greater cortical thickness, cortical area, and second moments of area (Imax
and 1min). It is worth noting that a similar increase in bending rigidity would have been
more efficiently obtained by the addition of a small amount of bone to the periosteal
surface, yet the bones adapted to increased mechanical loading by reducing the size of the
medullary cavity. This is an important point with respect to interpretation of cross-
sectional geometric variation. “All else being equal” (Larsen 1997) bones with a more
outward distribution of bone have greater bending rigidity, consequently it has been
assumed that increased bending loads result in bone deposition on the periosteal surface
where bone strains are greatest. Clearly, bone adaptation does not always result in the
most structurally efficient design, which suggests possible physiological constraints on the
adaptive process that control the ultimate distribution of bone tissue, and therefore the
bone’s geometry. It is also worth noting that in this experiment, measurement of external
dimensions alone would have led to the spurious conclusion that exercise had no effect on
bone mass and distribution. Woo et al. (1981) did not indicate if the increased cortical
thickness was due to the addition of new bone on the endocortical surface (e.g., by
formation-mode modeling); therefore, as interpreted by Martin at al. (1998) the effect of
moderate exercise in these pigs was likely a reduction in the amount of bone resorbed
from the endocortical surface, as would be expected in conservation—mode remodeling.
“Growth in the external size of a long bone, its cortical thickness, and the
distribution of cortical bone about the neutral axis are determined by the absolute and
relative behavior of the periosteal and endocortical bone surfaces along the length of the
bone” (Bass et a1. 2000: 2277). Therefore, interpretation of cross-sectional geometry in
skeletal samples for which activity is unknown necessitates knowing the effects of physical

activity and age on the activity of these bone surfaces. However, it is not known what

149

factors are responsible for whether bone adapts by periosteal expansion versus endosteal
contraction (Bradney et a1. 1998).

Woo et al.’s (1981) finding of reduced endosteal dimensions has been reported for
other animal exercise studies involving mature (e.g., Loitz and Zemicke 1992) and
immature animals (Matsuda et a1. 1986), as well as humans, including children and
adolescents involved in exercise intervention programs (Bass et a1. 1998; Bradney et al.
1998; Petit et al. 2002; Seeman 2002), and professional athletes involved in racquet sports
(Bass et al. 2002; Haapasalo et a1. 1996;Jones et al. 1977; Kannus et a1. 1995).

Studies of the effects of activity on femoral midshaft geometry in children and
adolescents are inconclusive. While reductions in endosteal dimensions are frequently
observed, effects on periosteal dimensions are more variable. Jumping exercise in early-
pubertal girls produced no significant changes in femoral midshaft geometry (Petit et a1.
2002). In this study, increases in periosteal diameter were non-significantly greater in the
control group. In contrast, jumping exercise did produce increases in femoral neck
geometry in the exercise group. Bradney et al. (1998) reported similar findings for the
femoral midshaft periosteal diameter in prepubertal males. However, in this group,
femoral midshaft cortical wall thickness increased in the exercise group due to a decrease
in endosteal diameter. Likewise Bass et al. (1998) found no difference between female
gymnasts and controls for femoral midshaft periosteal diameter; however, cortical area
was greater in gymnasts, again due to a reduction in endosteal diameter. Duncan et al.
(2002) compared femoral midshaft geometry among groups of female swimmers, cyclists,
runners, triathletes, and sedentary controls aged 15 to 18 years. All athletes were state or
national level competitors. Results showed that runners had significantly greater cortical

area and second moment of area than controls, swimmers, and cyclists, but not

triathletes. The cortical area of triathlctcs was significantly greater than that of swimmers
only. Total periosteal area, however, did not differ among groups, again suggesting a lack
of periosteal expansion. Medullary area was greater in swimmers and cyclists than in
runners and triathletes, which could reflect either an increase in endocortical bone
resorption or a decrease in endocortical formation.

It has been stated that “Young bone has a greater potential for periosteal
expansion than aging bone” (Forwood and Burr 1993), and that “the period of
longitudinal growth is the only time in life when bone may be added substantially on both
the inner (endosteal) and outer (periosteal)” surfaces (Haapasalo et al. 2000: 353). Based
on a study of young female tennis players, Bass et al. (2000) suggested that prior to
puberty the periosteal surface is most active, while during puberty both periosteal and
endocortical apposition occur. If the above hypotheses are correct then increased loading
prior to puberty should result in greater increases in cross-sectional second moments of
area and the polar second moment of area than the same loading applied later in life.
However, a tendency for increased periosteal apposition prior to puberty was not
apparent from the studies reviewed above (Bass et al. 1998; Bradney et a1. 1998; Duncan
et al. 2002; Petit et al. 2002), which primarily show a decrease in endosteal dimensions
rather than an increase in periosteal dimensions following increased activity. In contrast,
studies of long-term unilateral loading in competitive tennis do show increased periosteal
dimensions of the humerus (Bass et al. 2000; Haapasalo et al. 2000;Jones et a1. 1977;
Kontulainen et al. 2003). W hether this discrepancy reflects bone-bone (e.g., femur versus
humerus) differences with respect to periosteal and endosteal adaptation, loading

differences from different types of activities, or methodological differences is not known.

151

Another complicating factor in the evaluation of the effects of activity on bone
geometry is that the relative contribution of increased periosteal and endocortical bone to
observed increases in cortical area or cortical thickness has been shown to vary depending
on cortical location (anterior, medial, lateral, posterior) within a given section, and along
the length of the limb bones (Bass et al. 2000; Bradney et al. 1998; Duncan et al. 2002;
Haapasalo et al. 1996; Seeman 2002). Within the same bone the effects of a particular
activity are site-specific (Petit et a1. 2002). Consequently, the selection of bone sites for
analysis bears on the particular analytical outcome, and could be partly responsible for

the seemingly variable response of bone to activity reported in the studies above.

The optimal period for physical activity

Many studies have attempted to determine the optimal or “critical” period in
which physical activity produces maximal gains in bone strength. In this regard, the time
around puberty is often described as a “window of opportunity” to increase bone strength
through adaptations in bone mass and architecture (Khan et al. 2000; Mosley and
Lanyon 2002; Seeman 2002). Whether the optimal period is confined to either the pre-
pubertal or early pubertal years has not been determined.

Probably the most convincing research supporting adolescence as a time during
which activity has the greatest effect on cortical geometry comes from studies of humeral
bilateral asymmetry in long-term players of racquet sports (Bass et al. 2002; Haapasalo et
al. 1996, 1998; Kannus et al. 1995; Kontulainen et al. 2003; Seeman 2002). An
advantage of these studies is that the observed effects of activity on the dominant arm are
not confounded by differences in genetics, nutrition, and other lifestyle factors (Khan et

al. 2000), although differences in any of these factors could account for the large range of

inter-individual variation that is often observed (Kannus ct al. 1995). Activities such as
tennis impose large mechanical loads on the dominant playing arm, while the non-
playing arm experiences loads more typical of the arms of non-players. The arms of
players tend to exhibit significantly greater side-to-side differences in bone mineral
content, bone mineral density, and cortical cross-sectional geometry than those of non-
playing controls, although the nature and degree of the differences vary by location (e.g.,
proximal versus midshaft or distal humerus) (Haapasalo et al. 1996, 1998; Kannus et al.
1995). Haapasalo et al. (1996) also noted male-female differences in the specific location
of greatest side differences; male side-to-side differences were greatest proximally, female’s
distally. Based on a reevaluation of a classic study of humeral bilateral asymmetry in
professional tennis players (Jones et a1. 1977), Ruff et a1. (1994) demonstrated that cross-
seetional properties in the dominant playing arm are inversely correlated with starting
age. These results have been confirmed in studies of other players of tennis and squash
(Haapasalo et al. 1996, 1998; Kannus et a1. 1995; Kontulainen et a1. 2003). In a study of
female tennis players aged 7 to 17 years, Haapasalo et al. (1998) found that players
relative side-to-side differences were not significantly different from controls until the
adolescent growth spurt at puberty. Based on a separate study of young versus old
starters Haapasalo et al. (1996) concluded that, “even intense physical loading of a
mature bone is only marginally better in increasing the bone mass, bone density, and
[cortical wall thickness] of the target bone than the normal daily use of the dominant
extremity” (p. 864). In contrast to the above findings, Nara-Ashizawa et a1. (2002)
observed a reduction in cross-sectional properties (e.g., total area, medullary area, second
moment of area) and strength in the midshaft radius of the dominant playing arm in older

females who began playing recreational tennis on a routine basis after thirty years of age.

153

These findings contrast with side differences reported for the radial shaft in male tennis
players who began playing during childhood (Haapasalo et al. 2000). Nara-Ashizawa et
al. (2002: 621) suggest, “It is therefore conceivable that habitual exercise, after peak bone
mass has been attained, suppresses acceleration of bone loss from the endocortical area,
resulting in suppression of compensatory bone formation at the periosteal surface.” Once
again, the site-specific and sometimes counter-intuitive effects of physical activity on long
bone cross-sectional geometry suggest caution when interpreting group differences in
levels and types of activities based on variation in cross-sectional geometric properties.

In summary, the proposed counterargument to conventional interpretations of
cross-sectional geometry (i.e., those reviewed in chapter 3) is supported by evidence for
the following propositions: l) The bones of younger individuals are more likely than those
of older individuals to respond to their mechanical environment in such a way as to
produce localized differences cross-sectional geometry, and 2) There is a period of time
during development when mechanical loading has a maximal impact, such that activities
performed during this period produce more dramatic effects on cross-sectional geometry

than at other points during an individual’s lifespan.

Non-mechanical confounding variables

Attributing cross-sectional geometric variation to differences in mechanical
loading vis-a-vis physical activity assumes that nonmechanical factors do not influence
cross-sectional geometry or that such factors can be ruled out or controlled for in the
research design. There is an assumption that variation in the distribution of bone “seems
to suggest localized (mechanical) rather than systemic (nutritional or other) effects” (Ruff

et al. 1984: 135). Seemingly at odds with this notion is the fact that researchers who study

the effects of activity on skeletal adaptation in living humans must carefully design their
research protocols to control for a number of potentially confounding variables including
height, body proportions, weight, lean and fat mass, age, sex, maturation stage, genetic
background, socioeconomic status, and diet (Seeman 2002). Furthermore, even when the
amount and type of physical activity are known, as they are in controlled exercise
intervention studies, there is a limited ability to draw strong conclusions due to the
existence of confounding variables, such as selection bias (e.g., individuals who are
physically stronger self-select to participate in athletic activities) and unaccounted for
group differences in nutrition, socioeconomic status, and genetics (Khan et a1. 2000;
Seeman 2002). There is every reason to assume that a similar set of variables may
confound bioarchaeological studies attempting to compare the cross-sectional geometric
properties between two or more groups.

Frost believes that mechanical usage is the driving force in shaping skeletal
architecture, but that nonmechanical “agents” can “optimize or impair skeletal responses
to mechanical usage” (1996: 144) “by making the MES mechanisms either somewhat
deaf, or somewhat overreactive, to the skeleton’s normal mechanical usage” (Frost 1987:
6). Among the nonmechanical “agents” mentioned by Frost are nutrition, genetics,
hormones, and disease. As explained by Frost (1987: 6) “the bone mass effects of such
agents should duplicate those of changing mechanical usage and MES setpoints with
respect to kind and anatomical distribution.” For example, administration of fluoride
may act by lowering the setpoints so that the bone perceives a spurious overload, and
increases its mass appropriately (Frost 1987). “If a genetic factor set the MES setpoints
somewhat lower than the norm for most people and races, the mechanostat should sense

that a normal amount of bone is somewhat inadequate, so it would make affected

155

individuals accumulate somewhat more bone than the norm during growth, and then
retain more bone than the norm throughout life” (Frost 1987: 7). The opposite effect
would occur if a genetic factor resulted in a higher setpoint. This interpretation of the
potentially significant role of nonmechanical agents in bone adaptation is echoed byJee
(200121-37)

Nonmechanical agents influence the mechanical regulation and have a direct influence
on bone cells and their precursors involved with biologic mechanisms that are
independent of the mechanical stimuli. In addition, there are interactions between
mechanical and nonmechanical factors. Such additive or synergistic influences have
shown that mechanical stimuli for bone hypertrophy or atrophy can be altered by
nonmechanical agents. If one maintains the view that mechanical factors dominate bone
regulation, one can consider nonmechanical factors as agents that effectively alter the

level of mechanical stimulus.

The specific effect a mechanical stimulus has on bone may vary depending upon a variety
of factors: age, sex, previous loading history, anatomical location and function, bone
material properties, nutritional status, health status, hormone levels, body composition
(lean versus fat mass), total body mass, height, body proportions, and genetics. There is
no reason to assume that this list is exhaustive, and there is good reason to assume that
there are numerous interactions between these factors. The following sections will discuss

potential genetic and dietary effects on cross-sectional geometry.

Genetic efl’ects on cross-sectional geometry

Historically, phenotypic traits were often characterized as being under either
genetic or environmental control (i.e., nature versus nurture). This is now known to be a
false dichotomy. In general terms, phenotypic plasticity refers to the ability of an

organism to express a different phenotype depending on its environment. At one time it

was thought that such traits were not under genetic control. More recently it has been
accepted that phenotypic plasticity often has a genetic basis (Agrawal 2001). The modern
definition of phenotypic plasticity includes the ability of a single genotype to generate
different phenotypes in disparate environments, reflecting the interaction of genotype and
environment on developmental processes. Depending on the environment, a single
genotype can exhibit different chemistry, physiology, development, morphology, or
behavior. Clearly, the ability of bone to adapt to its mechanical environment provides for
many skeletal properties to be defined as phenotypically plastic (Lieberman 1997). Unlike
many phenotypically plastic traits, which at some point during development become fixed
(e.g., long bone length), some properties of the skeleton may remain phenotypically plastic
throughout the life of the organism (e.g., bone mass, cross-sectional geometry, bone
mineral density).

Ontogenetically, the genetic constitution of an individual is, to some unknown
extent, responsible for the general build of bones, providing “the basic genetic template.”
Evolutionarily, genetic differences account for a great deal of the interspeeies variation in
skeletal form (Goodship and Cunningham 2001). Currey (2002: 339) has stated, “the
form of bones, lying latent in the genes, is the result of natural selection acting in the past
on mechanically functioning skeletons.” Currey also acknowledges that, “the interaction
between the genetic endowment of the cells concerned with remodeling and the strain
imposed on the bone must be complex” partly because “in the mature skeleton, the kinds
of stresses imposed on bones will differ from place to place” throughout the skeleton (ibid:
379)

Obviously, genetics play a fundamental role in the structure and mechanical

properties of the skeleton. The initial development of the skeleton, the formation of the

157

cartilaginous anlage and its subsequent ossification, and the number and general form of
the skeletal elements are unquestioningly under genetic control. In addition, vertebrates
have evolved the physiological machinery necessary for the skeleton to sense and adapt to
its environment, and there can be no doubt that the biological processes involved in this
adaptive response are under the control of multiple, probably hundreds, of polymorphic
genes. Furthermore, it is almost certain that these biological processes differ between
species, and that these differences are genetic in origin. For example, the bones of most
modern fish do not undergo remodeling, yet are “adapted” to the loads they must bear
(Currey 2002: 26).

Ruff and colleagues (Ruff 1987, 2000b; Trinkaus et al. 1994), and Bridges (1995)
have argued that differences in mechanical loading account for the gradual reduction in
robusticity and sexual dimorphism that occurred during the evolution of the Homo
lineage. However, as pointed out by Martin et al. (1998), an alternative explanation
involving genetic differences between Neanderthals (and other archaic Homo species) and
modern Homo sapiens cannot be ruled out. They suggest two possibilities, one being a
genetic change in bone metabolism resulting in a lowering of the setpoint for modeling.
In this situation, “ordinary loads would stimulate more bone formation through modeling
in growing Neanderthals, and suppress remodeling in adults, fitting the observations
equally well” as differences in physical activity (ibid: 270). A second possibility involves
genetic differences in Neanderthal periosteal tissue resulting in “enhanced bone formation
and larger skeletons” (ibid: 270). Other possibilities, such as endocrine diflerences, exist
as well. Ruff, Larsen, and Bridges, while acknowledging that genetic factors play a role in
skeletal structure, have also asserted that temporal and geographic variation in skeletal

robusticity and cross-sectional geometric properties among more recent prehistoric and

early historic Native American groups is better explained by differences in mechanical
loading than genetics. Alternatively, this dissertation suggests that genetic differences
may account for some of the observed variation among recent groups as well. More
definitively, in the absence of strong evidence to the contrary, genetic differences cannot
be ruled out, and preclude the attribution of cross-sectional geometry to physical activity.

Certainly, it is probable that genetic differences existed between geographically
distinct populations such as those from Pecos Pueblo, New Mexico and the Georgia coast.
And while it may be less likely that genetic differences account for diachronic changes in
cross-sectional geometric properties that coincided with the transition to agriculture on
the Georgia coast'5 (Ruff and Larsen 1990; Ruff et al. 1984), it is possible that some of the
differences observed following Spanish contact could be genetic in origin. For example,
during the contact period, Spanish mission populations may have reflected a genetically
heterogeneous group composed of aggregations of various regional native populations
(Larsen 1990). This could potentially explain the bimodal distribution Ruff and Larsen
(1990) observed for male femoral cross-sectional shape during the contact period, which
they interpreted as evidence that certain males were recruited by the Spanish for long
distance travel.

Simply postulating a potential genetic difference between groups does not by itself
create a strong argument for an alternative, genetic explanation for the observed
variation in cross-sectional properties. As discussed in chapter 5, Ruff and colleagues

(Ruff et al. 1984) have suggested that variation, particularly localized variation (as

 

'5 Gene frequency differences between the preagricultural and agricultural precontact populations could
exist (Ruff et a1. 1984). The time frame involved encompasses 40 or more generations. It is unclear
whether the preagricultural group, which included burials from a period of over 1000 - 3000 years
(2200BC - ADl 150, with most post-dating AD500) (Ruff and Larsen 1990; Ruff et al. 1984), demonstrated

any significant within-group variation over time.

opposed to systemic skeletal variation), in long bone cortical cross-sectional geometry is
more likely the result of differences in mechanical loading than other factors such as
genetics or nutrition. The argument for an alternative, genetic explanation is
strengthened by evidence for genetic loci involved in producing normal skeletal variation
in long bone structure, including cross-sectional geometry. The argument would be
further strengthened by evidence that the genetic effects could be site-specific. Therefore,
an important question, within the context of this dissertation, and in response to claims
that only mechanical loading results in localized differences in structure, is this: can
genetic differences produce localized versus systemic effects? \Nhile there are no
definitive answers, preliminary evidence suggests it is reasonable to hypothesize that
genetics could produce localized variations in skeletal structure through at least three
different pathways (adapted from Volkman et a1. 2003): 1) By direct local control over
growth, modeling, and remodeling activities during development, resulting in regional
geometric size and .shape differences among the different long bones of the skeleton,
which persist into adulthood. 2) By indirect effects on local geometry through a primary
effect on muscle strength, body shape, body mass, including distribution of lean and fat
mass, or bone material properties. 3) By altering the sensitivity of bone to its mechanical
environment, which could amplify or attenuate local mechanical stimuli. This could
produce a regional pattern of cross-sectional geometric variation, which would appear to
be due to differential mechanical loading vis-a-vis physical activity.

Skeletal biologists have long acknowledged the potential role of heredity in
contributing to bone architecture (e.g., \Nolff 1892). In 1968 Enlow wrote, “genetic
predisposition must be included in any complete account of the composite, diverse factors

that can contribute to bone morphology, growth, and differentiation. . .. Until the

1 60

dilemma of the local control mechanism itself is resolved, possible contribution by all such
factors [including genetics] must necessarily be considered and taken into account, since
the actual extent of their individual roles is not now known” (p. 810). Enlow’s remarks
are as applicable today as they were thirty-five years ago, as evidenced by van der Meulen
and Huiskes’ (2002: 41 1) recent comment, “The challenge lies in disentangling
environmental modulation and genetic predisposition in the skeleton.. .. Successfully
distinguishing adaptation from genetic programming of cell metabolism will be a great
mechanobiological feat.” With numerous recent breakthroughs in molecular genetics
and genetic engineering techniques (Young and Dieudonné 2001; Young and Xu 2001),
this feat may yet be accomplished.

Studies (Amblard et al. 2003; Beamer et al. 2001; Volkman et al. 2003) estimate
that the heritability of bone mineral density (BMD) is between fifty and ninety percent.
Heritability is the proportion of variance in a phenotypic trait that is accounted for by
genetic variance. In other words, somewhere between 50-90% of normal variation in
BMD is genetically determined. The strong genetic component of BMD was found to be
largely independent of lean muscle mass and muscle strength, which are other factors
strongly associated with BMD (Arden and Spector 1997). Other researchers (e.g.,
Yershov et a1. 2001) have reported that as much as 50—70°/o of bone strength is inherited.
Moreover, studies have shown genetic effects on bone mass, geometry, and mechanical
properties.

A survey of recent volumes of theJoumal of Bone and Mineral Research reveals
that the genetic basis of skeletal properties is the subject of much ongoing research.
There is a growing awareness of the genetic complexity of long bone cortical geometry,

and it is becoming clear that there are numerous genetic loci involved in producing

161

normal variation in skeletal structure and strength. Among these are loci involved in the
regulation of the skeletal response to mechanical loading, as alluded to by Martin et al.
(1998) above, and predicted by Frost (1986). Furthermore, there are often significant
environmentzgene interactions involved in phenotypic expression, whereby external
environmental factors, such as mechanical loading, modify gene expression (Toma et al.
1997; Young and Dieudonné 2001). Environmentally responsive genes are involved in
the biological processes of mechanosensation (the sensing of mechanical signals) and
mechanotransduetion (the processes involved in converting the mechanical signal into a
cellular response). Therefore, genetic differences at both the individual and population
levels could alter both mechanosensitivity and the efliciency of the adaptive processes

(Nomura and Takano—Yamamoto 2000; Robling and Turner 2002; Volkman et al. 2003).

Genetic effects on bone structure in animals

A series of experiments involving two inbred breeds (strains) of mice designated
C3H and B6 have revealed several apparently genetic-related differences in the response
of bone to mechanical loading (Akhter et a1. 1998; Amblard et a1. 2003; Kodama et al.
2000; Pederson et al. 1999; Robling and Turner 2002). The C3H and B6 breeds differ
with respect to bone density, exhibiting high density and low density, respectively (Akhter
et al. 1998; Shultz et al. 2003). A study of bone accumulation in early postnatal, pubertal,
and post-pubertal C3H and B6 mice (Richman et al. 2001) has demonstrated that the
greater BMD of the C3H mice is apparent by 7 days of age. The mice do not differ in
body size or weight and have similar external bone dimensions (Kodama et a1. 2000).
However, they exhibit breed-dependent differences in bone mass and cross-sectional

geometry (Akhter et al. 1998; Kodama et al. 2000; Richman et al. 2001; Turner et al.

162

2000). Both total area and medullary area of lower limb bones are greater in the B6
mice, while cortical area is similar between the two, resulting in thinner cortices in the B6
mice. This explains the observed lower bone density in the B6 mice, but this structural
arrangement also results in higher second moments of area, hence greater bending
rigidity. Turner et a1. (2000) report higher femoral bending strength in C3H mice due to
their more mineralized and thicker cortices. Interestingly, these authors report that there
was no difference in strength between B6 and C3H mice in their lumbar vertebrae.
Within each breed, males exhibit thicker cortical bone than females (Richman et al.
2001). The high bone mass of C3H mice has been attributed to a reduction in bone
cellular activity, while the low bone mass of B6 mice is a result of greater bone cell activity
(i.e., higher levels of bone resorption and formation) (Amblard et a1. 2003).

Several studies have identified differences between B6 and C3H in their response
to mechanical loading and unloading prior to and after skeletal maturity (i.e., 16-weeks of
age). Akhter et al. (1998) tested the effects of in vivo mechanical loading using a four-point
bending device applied to the tibia of 16-week-old mice. Bending induced greater bone
formation on both periosteal and endosteal surfaces in B6 mice but produced little
periosteal bone formation and no endocortical response in the C3H mice prompting the
authors to propose that the B6 mice are more sensitive to loading than the C3H mice.
The authors suggested that the genes that regulate the adaptive response might differ
from those that control peak bone density.

A study by Kodama et a1. (2000) suggests a bone-specific, localized difference in
response to identical mechanical loading between these two genetically distinct strains. A
four-week jumping exercise regimen, begun when the mice were 9-weeks-old, produced

an increase in tibial, but notfemoral, dry weight in B6 mice. Thejumping exercise

163

produced periosteal bone formation in the tibiae of B6 mice, but had no effect on the
endocortical surface or medullary area in either strain. Kodama et al. (2000) attribute the
differences in response to mechanical loading as a decreased sensitivity in the C3H mice.
A study by Robling and Turner (2002) support this view. These authors performed
mechanical loading experiments using the ulna, and observed that, unlike the tibia and
femur for which the B6 mice exhibit a greater total cross-sectional area, the midshaft ulna
of the C3H mice exhibited the larger total area. Thus, there seem to be bone specific
differences in strength and geometry between these genetically distinct strains.

A study of 9-week-old C3H female mice (Pederson et al. 1999) revealed a load
magnitude-dependent periosteal woven bone response in tibiae subjected to bending.
Significant increases were observed in the loaded compared to the unloaded tibiae for
total area, medullary area, and second moment of area. Interestingly, endocortical
formation was opposite that of periosteal, with the nonloaded tibiae showing increased
lamellar bone formation, resulting in a decreased medullary area. The behavior of the
endocortical surface in the C3H strain is different from other animal models, which the
authors interpret as suggesting “potential differences in the genetic control of bone
adaptation” (Pederson et al. 1999).

The bones of B6 and C3H strains also respond differently to unloading (i.e.,
disuse). Kodama et al. (2000) report that immobilization by sciatic neurectomy led to a
greater bone loss in B6 mice. Relative to the C3H mice, the tibiae of B6 mice showed an
increase in endosteal bone resorption, and a decrease in bone formation as would be
expected in disuse. Amblard et al. (2003) obtained similar results in an unloading
experiment employing tail-suspension in 16-week-old B6 and C3H mice. Following 3

weeks of immobilization, B6 mice showed significant cancellous bone loss from the distal

164

femur due to thinning of the trabeculae, while the C3H mice did not. In contrast to the
interpretation of Kodama et a1. (2000) and Robling and T umer (2002), these authors do
not interpret this as reduced sensitivity to unloading in C3H mice. Based on the presence
of biochemical markers of bone activity in both strains, Amblard et al. (2003: 567) instead
proposed that, “C3H mice. . .constitute a unique model in which genetic background
overwhelmed the usual effects of reduced biomechanical usage in bone.” They conclude
that their “results strongly suggested that susceptibility to bone loss is a genetically
determined trait” (ibid: 567).

Findings by Turner et al. (200 1 b) suggest that inbred strains of rats also provide
models with which to study the effects of genetics on skeletal structure and strength.
Based on their evaluation of skeletal variability, these authors propose site-specific control
of properties involved in skeletal strength, implying that, “no single gene regulates skeletal
fragility at all sites” (ibid: 1532). Taken together these experiments on inbred strains of
mice and rats suggest the possibility that individual or group differences in BMD, bone
mass, and cortical bone geometry may be due to genetic differences. Furthermore,
differences in the biological responses to mechanical loading and unloading may also
have a genetic basis, and perhaps more significant in the context of this dissertation,
genetic differences in adaptive response can potentially result in site-specific, localized

skeletal effects.

Identifying the genes
Quantitative traits, such as body size, obesity, bone mineral density, and bone
cortical geometry, are known as complex traits, controlled by multiple genetic loci, i.e.,

they are polygenic (Reis 2003; Robling and Turner 2002; Shultz et al. 2003; V olkman et

165

al. 2003). Complex traits are often analyzed using quantitative trait locus (QTL) analyses
of inbred mice strains (Drake et al. 2001). In QI’L analysis two strains with distinct
phenotypic differences are crossed and successive generations are screened for the
frequency of recombination between the phenotype under study and available markers.
A statistically significant association of the phenotypic trait of interest with a series of
markers at the same chromosomal locus indicates that at least one gene affecting the trait
is likely to be physically linked with those markers, and therefore found at that locus.

Researchers have created a number of inbred strains of mice in order to estimate
the heritability of skeletal phenotypes. QTL analysis of these strains is a first step towards
identifying specific genes involved. Turner et al. (2001a) evaluated the heritability of
“factors associated with bone strength” (e.g., volumetric BMD, cortical geometry, and
microstructure) in femora and lumbar vertebrae from inbred strains of mice representing
the progeny of crosses betWeen B6 and C3H strains. Results of this study suggested site-
specific regulation of bone strength by polymorphic genes. Findings of Yershov et al.
(2001) and Klein et al. (2002) are similar. Both groups identified several chromosomal
loci linked with skeletal properties associated with bone strength, including diaphyseal
diameter and second moments of area.

Using a genetically heterogeneous mouse population derived from the progeny of
crosses involving four inbred strains, including B6 and C3H, Volkman et a1. (2003)
detected significant associations between 14 genetic markers located on 13 different
chromosomes and a number of geometric traits of the femur, including cortical area,
cortical thickness and second moments of area. The authors found that the genetic
markers accounted for only 8.2-21.7°/o of the observed variance in the geometric traits

studied, but added that, “some of the variance could be attributed to genes whose effects

166

are too small to be measured in our survey” (ibid: 1502). In their discussion, the authors
noted that in addition to genetics, bone geometry is influenced by a variety of factors
including body weight, muscle strength, disease status, and nutritional levels. They
pointed out that many of these other “nongenetic” factors, may in fact, “ultimately be
regulated by genetic controls” (ibid: 1503). Volkman et al. concluded that, “the genetic
control of cortical geometry is complex and that femoral size and shape may be
influenced by different, although overlapping, groups of polymorphic loci” (ibid: 1502).
However, they were also careful to state that their study could not distinguish among
three possible mechanisms by which the genes could affect femoral geometry: 1) by
directly influencing bone size and shape, 2) by indirectly influencing bone size and shape
through primary affects on body weight, muscle strength, or activity level, or 3) by
altering bone mechanosensitivity. The authors performed additional statistical analyses
of covariance to account for variation in body weight, and were able to suggest'that it was
unlikely that the geometric effects of the genetic markers were due to a primary effect on
body weight.

Drake et al. (2001) have identified seven chromosomal loci linked to femoral
structural traits, including femoral length and multiple measurements of femoral width
(e.g., width of the femoral head, intertrochanteric, mid-diaphyseal, and supracondylar
regions). Three loci, located on different chromosomes, were found to be associated with
the femoral width measures, suggesting the potential for site-specific effects within the
femur. According to the authors, none of the identified loci controlling femoral width
“influenced body weight or length, excluding these mechanical-related factors as
mechanisms for these local effects” (ibid: 516).

Interestingly, one chromosomal locus involved in controlling femur width includes

167

 

genes for parathyroid hormone (PTH) and calcitonin. Hagino et al. (2001) found that
PI‘H and mechanical loading have a synergistic effect on bone formation in rat tibiae,
and suggest that, “Pf H sensitizes bone cells to mechanical stimulation” (p. 249). In
accordance with Frost’s mechanostat hypothesis, the authors explained their observations:
“hormones act to adjust the skeletal ‘set point’ that regulates bone mass relative to
customarily encountered loading stimuli. The observations in the present study indicate
that intermittent PT H administration lowers this ‘set point’, such that identical loading
stimuli are perceived as more intense. . .” (ibid: 249). The combined findings of Hagino et
a1. and Drake et al. are strongly suggestive that genetic control of hormone levels provides
another source of normal variation in cross-sectional geometry. Perhaps not surprisingly,
the two inbred strains of mice (C3H and B6) discussed earlier also exhibited differences in
serum PTH levels (Akhter et al. 1998).

The creation of transgenic mice has allowed the study of gene function in vivo
(Young and Xu 2001). As described by Young and Xu (2001) there are “two major types
of transgenic mice”: 1) “Conventional” transgenic mice “overexpress” normal genes
resulting in a gain in function. Conventional transgenic mice are created by inserting
selected gene sequences into the host genome. The inserted genes can be transmitted to
oflspring thereby allowing the creation of strains of transgenic mice through selective
breeding. 2) “Knockout” mice are created by “targeted deletions of specific genes”
resulting in a loss of function for a single specific gene of interest. Subsequent breeding
produces lineages that are homozygous for the normal allele, heterozygous, or
homozygous for the mutant allele, i.e., “total knockouts.” Scientists are also working on
creating “designer mice,” in which genes that have lethal consequences when “knocked

out” at conception, can be knocked out at a specific developmental stage or within a

168

specific location, tissue, or cell type (Young and Xu 2001). These authors conclude with
the optimistic proclamation, “anything is possible in creating transgenic mice; one is
limited only by well-characterized genes and promoters and the proper analyses” (ibid: 4-
13).

Both conventional transgenic and knockout mice strains have demonstrated
numerous abnormal skeletal phenotypes, including structural anomalies (Young and Xu
2001). Conventional transgenic mice have revealed alterations in the activities of
osteoblasts and osteoclasts leading to conditions of either too much or too little bone (e.g.,
Gardiner et a1. 1998; van der Meulen and Huiskes 2002). Use of knockout mice has
shown that, “different strains of mice vary in bone mass as a result of intrinsic differences
in rates of bone formation” (Young and Xu 2001: 4-6). For example, in response to hind
limb immobilization by way of tail suspension, osteopontin-knockout mice experienced
no significant resorption of trabeculae in the tibial metaphysis, and no suppression of
bone formation relative to wild-type mice (Ishijima et al. 2001). Osteopontin (OPN) is a
noncollagenous bone protein that may facilitate attachment of osteoclasts during bone
resorption (Ishijima et a1. 2001). OPN is expressed in bone cells in response to
mechanical loading, and is hypothesized to play a role in loading-induced changes in
bone metabolism. Loss of OPN does not affect normal bone development or phenotype;
however, loss of OPN function appears to reduce sensitivity to factors that would
normally lead to bone loss during conditions of disuse. The authors conclude, “OPN may
be involved in the mechanisms sensing the physical force that induces the increase in the
number of osteoclasts. . .OPN directly modulates bone formation in response to

mechanical stress” (ibid: 403).

169

Genetics efl’ects on bone structure in humans

Studies of BMD in humans have revealed a high heritability with involvement of
several chromosomes (Duncan et al. 2003). Evidence suggests that inheritance of BMD is
site and gender specific (ibid). As discussed by Babij et al. (2003), QTL analysis has
identified a specific region of chromosome 1 l as the potential location of genes that “may
contribute to the normal variation in BMD seen in the general population” (ibid: 961).
The low-density lipoprotein 5 gene (LRP5 gene) is one candidate gene for which a variety
of mutations are associated with variation in bone mass in humans (Babij et a1. 2003;
Koller et al. 1998). A particular mutation of this gene has already been identified in
certain families with higher than normal bone mass, and Duncan et al. (2003) propose
that LRP5 allelic variants play a role in normal population variation. A study of
transgenic mice with this LRP5 allelic variant suggests the high bone mass results from
increased numbers of active osteoblasts (Babij et a1. 2003).

Researchers have also studied the genetics of proximal femoral structural
variation in humans (Koller et al. 2001, 2003; Slemenda et al. 1996). Slemenda et al.
(1996) focused on bone mineral content and geometric properties of the femoral neck,
and found evidence for genetic influence on all properties except femoral neck length.
Koller et al. (2001, 2003) have performed linkage analyses on large samples of sister-pairs
to elucidate the location of potential genes involved in controlling femur neck axis length,
midfemur width, femur head width, and pelvic axis length. In their first study of 309
sister-pairs they found linkage of structural variables with human chromosomes 3, 4, 5, 7,
9, l7, and 19. Their second, larger study (437 sister-pairs) confirmed linkage with
chromosomes 3, 7, and 19, and identified a new locus on chromosome 8. Probable genes

controlling femoral shaft width were restricted to chromosome 3 in this analysis, whereas

170

femoral head width showed significant linkage with regions on chromosomes 3, 7, and 8.
In summary, results of genetic analyses are not definitive. Moreover, it remains to
be seen whether the effects seen in inbred and genetically engineered mice strains can be
extrapolated to naturally occurring genetic variation in humans. Nevertheless,
preliminary results are suggestive that genetics versus mechanical loading arguments need
to be revisited. Cortical bone geometry is considered a complex trait under polygenic
control. In mice, genes controlling cortical geometry are located on numerous
chromosomes, and many traits appear to segregate independently. Studies on humans
have also shown genetic control of femoral structure with associations at several loci
located on different chromosomes. Furthermore, genes do not have to directly control
the final form of the bone to play an important role in producing that form. Polymorphic
genes can influence whether and to what degree exercise and mechanical loading will
affect bone geometry. Genes can modify metabolism, developmental timing, and
sensitivity to mechanical stimuli—“all of which can subsequently affect cross-sectional
geometry and produce localized (intraskeletal) variation that might mask or mimic
changes in external loading regimes. It is safe to conclude that, as an alternative
explanation for group variation in cross-sectional geometry, genetic differences should not

be ruled out a priori.

Nutrition efi'ects on cross-sectional geometry

Nutritional status resulting from diet is thought to: 1) affect all bones equally
(Bridges 1991), 2) affect stature via its effects on longitudinal growth (Larsen 1981; Ruff et
al. 1984), 3) possibly affect the amount of bone (e.g., CA), but not its distribution (Ruff et

a1. 1984), and 4) affect bone quality (e.g., material properties) (Barondess 1998; Brock and

171

Ruff 1988). Dietary factors are not thought to influence the distribution of bone tissue, or
have differential effects throughout the skeleton. Consequently, skeletally localized cross-
sectional geometric variation is thought to primarily reflect differences in mechanical
loading patterns (Ruff et a1. 1984).

There are very few studies of the effects of nutrition on cross-sectional geometry.
However, diet is considered a confounding variable in studies of functional adaptation
(F orwood and Burr 1993), and is a risk factor for osteoporosis (Hernandez et al. 2000;
Jiang et a1. 1997; Wohl et al. 2000). Both macronutrients (protein and dietary fats) and
micronutrients (vitamins A, C, and D, and minerals, e.g., calcium, magnesium, copper,
iron) play important roles in skeletal physiology (Frost et al. 1998; Li et al. 1999). Dietary
factors are likely to have indirect effects on skeletal structure, including cross-sectional
geometry, through primary effects on body weight, muscle mass and strength, fat mass,
hormone levels (e.g., estrogen, insulin, leptin, parathyroid hormone, calcitonin,
glucocorticoids), and bone metabolism (Frost 1985, 1997c;Jiang et al. 1997; Li and
Miihlbauer 1999; Orwoll 1992; Zemicke et al. 1995). In addition, the effects of diet,
particularly nutritional deficiencies, on bone growth and maturation are likely to have
consequences for cross-sectional geometry (Frost 1985). Significant changes in bone
geometry are associated with growth spurts, developmental timing, and puberty, which
can all be affected by nutritional status. Mosley and Lanyon (2002) have suggested that
the response of bone to loading in growing animals is related to the rate of growth, i.e.,
increased physical activity during growth spurts is likely to produce a larger response.
Therefore, factors that affect rate of growth, such as protein deficiency (Orwoll 1992), can
affect the acquisition and potentially the distribution of bone mass.

Nutrition can affect body mass including the absolute amount, proportion, and

172

distribution of lean and fat mass. In a recent review article, Reid (2002) discussed the
relationship among soft tissue mass, bone turnover, and bone mass. Studies have
demonstrated a positive correlation between bone mineral density (BMD) and body mass.
Although increased mechanical loading associated with increased body mass is a likely
contributor to this relationship, the correlation exists for both weight-bearing (e.g.,
vertebrae, femora, and tibiae) and non-weight-bearing (e.g., radius and the metacarpals)
skeletal elements (Reid 2002). Reid “cautiously” concluded “that both fat and lean
masses impact positively on bone mass, although their relative impacts may vary across
populations” (ibid: 550, emphasis added). Studies also point to male-female differences in
the effects of soft tissue mass on bone mass, with the effects on males less significant (ibid).
If true, this factor could potentially contribute to group differences in sexual dimorphism
of cross-sectional geometry. Reid proposed that hormones “associated with nutrition are
prime candidates” for the association between fat mass (adipose tissue) and bone mass
citing “biochemical evidence that both food intake and weight change can impact on
bone turnover” (p. 550). For example, fasting has been found to decrease osteoblast
activity, hence reduce bone formation. Furthermore, insulin is osteogenic.
Hyperinsulinemie conditions, such as obesity and type 2-diabetes, are associated with
high bone density. Additionally, positive effects on bone density are linked to hormones
secreted by adipocytes, such as estrogen and leptin.

While it is probable that the primary effects of nutrition are systemic, as has been
asserted by Ruff and colleagues (Ruff et al. 1984) and Bridges (1989), diet and nutrition
can affect bone metabolism, which could have more localized or site—specific effects by
altering mechanical loading thresholds (e.g., MES setpoints). Researchers attempting to

more accurately model bone adaptation (Hernandez et al. 2000) suggest that it is

173

important to consider metabolic factors (nutrition, drugs, disease state) because
“metabolic factors can change the mechanical setpoint thereby modulating the
mechanobiologic response” (p. 237). The authors suggest that metabolic factors, like
mechanical factors, can influence relative osteoclast and osteoblast activity during
remodeling. Resultant changes in bone density “caused by changes in activation
frequency or bone balance ratio” can “ [modify] the mechanical daily stress stimulus even
when the mechanical loading histooi remains constant” (ibid: 242, emphasis added). By altering
setpoints, nutrition and other metabolic factors would be capable of producing localized
effects on cortical cross-sectional geometry, which could mimic the site-specific
adaptations produced by differences in mechanical loading.

There is some experimental evidence to support this hypothesis. The magnitude
of dietary effects on bone geometry and mechanical properties is often site-specific (Jiang
et al. 1997; Parsons et al. 1997; Wohl et al. 2000). Bone-to-bone differences have been
noted between axial and appendicular elements, as well as between bones of the lower
limb. Li et al. (1990) and Zemicke et a1. (1995) studied dietary effects on postnatal bone
development in young rats. Li et al. (1990) fed young, rapidly growing rats a high fat-
sucrose (HFS) diet for 10 weeks. Compared with the tibiae of control rats, fed a low fat,
complex-carbohydrate diet, the tibiae of rats on the HFS diet exhibited no significant
difference in cortical geometry, however, structural strength was lower. In contrast, the
metatarsus bone exhibited increased cross-sectional area and reduced material properties,
but no difference in strength. These authors concluded that, “adaptation of a bone to
changes in diet can be bone specific” (ibid: 312). After a two-year experimental period
Zemicke et al. (1995) found that compared with controls, rats on the HF S diet exhibited

reduced architectural and mechanical properties of the sixth lumbar vertebra and femoral

174

neck. Changes to the femoral neck primarily involved a significant reduction in the
percentage of cortical bone, which was most likely responsible for its reduced mechanical
properties. The negative effects on the mechanical properties of these bones occurred in
spite of significantly increased body mass in the rats on the HFS diet.

Many bone biologists recognize that nutrition plays a role in bone metabolism
(Frost 1985; Frost et al. 1998; Rubin et al. 1990). Rubin et al. (1990) have stated that the
physiology of functional adaptation is “(inextricably linked) to the organism’s systemic
milieu” (p. 52). Furthermore, according to these authors, “Nutritional disorders. . .alter
not only the manner in which the tissue responds, but also attenuate the ability of the
tissue to react to osteogenic stimuli” (ibid: 48). They propose that the interaction of
calcium deficiency and decreased loading is synergistic. Moreover, during systemic
calcium deficiency, dynamic loading is insuflicient to maintain bone mass. Specker and
colleagues (Specker and Binkley 2003; Specker et a1. 1999) evaluated the combined effects
of calcium intake and increased gross motor activity in young children and infants.
Decreased calcium intake during periods of rapid growth led to lower rates of bone
accretion in infants engaged in gross motor activity as opposed to fine motor activity
(Specker et al. 1999). Furthermore, Specker and Binkley (2003) found a significant
interaction between calcium supplementation and physical activity with regard to cortical
thickness and cortical area in young children; increased calcium intake resulted in greater
tibial periosteal and endosteal dimensions. Iuliano-Burns et al. (2003) detected an
interaction effect of physical activity and calcium intake on bone mass in prepubertal
girls. Importantly, the effects were site-specific (i.e., they were not uniform throughout
the skeletal sites evaluated). Among the exercise-loaded lower limb bones, a significant

calcium-activity interaction effect was detected for the femur, but not the tibia-fibula site.

175

In addition, it was found that exercise, but not calcium supplementation, produced an
increase in bone mass at the tibia-fibula site. However, at the non-loaded upper limb sites
(humerus, and ulna-radius) there was no effect of exercise on bone mass; however,
calcium supplementation alone resulted in increased bone mass. The researchers did not
specifically evaluate the effects of calcium supplementation, if any, on cross-sectional
geometry. Because of the interaction between mechanical loading and nutrition, Rubin
et al. (1990: 48) proposed that, “exercise regimes prescribed for one given population may
not be as effective (or generate completely different results) in another.”

Studies of the effects of specific nutrients and foods have demonstrated effects on
bone metabolism, including altered bone cell activity, with consequent alterations in bone
distribution. In a review of the effects of dietary protein on bone metabolism, Orwoll
(1992) noted that protein deficiency not only leads to decreased longitudinal skeletal
growth in children, but can also increase endosteal resorption. There are no reported
negative effects of protein deficiency on periosteal growth. A similar observation was
made for aged male rats; Bourrin et al. (2000) observed an association between dietary
protein deficiency and decreased cortical bone BMD, cross-sectional geometry, and bone
strength. In this study, a low protein diet led to an increased medullary area and
decreased cortical thickness. The authors suggest that reduced dietary protein impairs
bone formation creating a more negative bone balance on endocortical and trabecular
surfaces undergoing remodeling. In addition, certain types of foods are associated with
differences in bone cell activity. Consumption of a diet including onions, vegetables, leafy
salad greens, and herbs was found to inhibit bone resorption and increase bone mass in
rats (Miihlbauer et al. 2002). Sun et al. (2003) reported that the addition of fish oil to the

diet might reduce bone loss through inhibition of osteoclast generation and activation.

176

Another potential indirect effect of nutrition on cross-sectional geometry is via its
effect on bone material properties. For example, calcium deficiency decreases bone
mineralization, which reduces the elastic modulus of bone tissue. Ruff has maintained
that, “the primary response of long bone diaphyses to changes in mechanical loading
during life is through alterations in diaphyseal geometry or structure, rather than material
properties” (Ruff 2000b: 72). While most research on bone adaptation bears this out, it is
also true that nonmechanical factors, including nutrition, affect material properties with
potential secondary effects on geometry. It has been demonstrated that bone can
compensate for reduced material properties by increasing cross-sectional area or
changing cross-sectional shape (Akhter et al. 1999; Burr et a1. 1981; Currey 2001, 2002;
Li et a1. 1990; Martin and Atkinson 1977; van der Meulen et al. 2001). As stated by
Currey (2001: 19-13), “the architecture of the bones is rather precisely adapted to the
loads placed on them and to the mechanical properties of the bone material.” A
compensatory response may have been what occurred in the metatarsus bone of the rats
fed the high-fat high-sucrose diet (Li et al. 1990); in this bone, the structural properties
were similar to those of the control group in spite of reduced material properties, most
likely because of the significant increase in cross-sectional area. Although this study
showed no compensatory geometric response in the tibia, which also exhibited reduced
material properties, it is possible that the time frame of the experiment (10 weeks) was not

sufficient for such a response to occur in this bone.

Summg_ry

A review of Wolff’ 3 Law reveals that its derivation is flawed, and it is largely

unsupported. The contemporary usage of the phrase “W olff’s Law” is generally

177

understood to be synonymous with the concept of functional adaptation. However,
functional adaptation may be an “umbrella term” for a variety of skeletal responses, not
all of which are directly related to mechanical stimuli resulting from physical activity.
Research over the past century, and particularly within the last two decades, has
increased the knowledge base regarding the biological processes involved in adaptation of
the skeleton to mechanical loading; however, many facts remain unknown. To a large
extent, the apparent variability in the results of functional adaptation experiments in both
animals and humans, and among individuals of different ages may reflect the general lack
of understanding of how and under what circumstances functional adaptation occurs, as
well as the difficulty distinguishing the structural changes that result from adaptation to
physical activity from those attributable to methodological variation and confounding
factors. Furthermore, the adaptive response of bone is not always what would be
predicted based on structural optimization hypotheses such as \N olff’s Law, e.g., bone
added to the endocortical rather than periosteal surface. Because of the age-dependent
effects of activity on cross-sectional geometry, as well as the site-specific effects of non-
mechanical factors such as genetics and nutrition, neither the level nor the type of activity
need be different for two groups to exhibit localized differences in their long bone
geometry. Therefore, while functional adaptation of bone may be a widely accepted
phenomenon, the concept does not provide a coherent theory upon which to base

interpretations of physical activity patterns from cross-sectional geometry.

178

CHAPTER 7—CONCLUSION

...plausibility alone does not prove an explanation is correct

Harold M. Frost

This dissertation presented a reconstruction and critical analysis of the argument
underlying the following tenets in bioarchaeology: 1) Group variation in cross-sectional
geometry reflects group differences in levels and types of physical activity. 2) Group
variation in degree of sexual dimorphism of cross-sectional geometry is attributable to
differences in sexual division of labor. 3) Group variation in degree of upper limb
bilateral asymmetry reflects differential usage (e.g., unilateral versus bilateral activities).

It was shown in chapter 5 that the overall argument can be evaluated as two,
separate but linked inductive arguments: 1) A ‘theoretical’ argument, the purpose of
which is to establish a causal relationship between physical activity and cross-sectional
geometric variation, and 2) An ‘empirical’ argument, which uses the conclusion of the
theoretical argument tojustify drawing specific behavioral conclusions from
biomechanical data.

Evaluation of the theoretical argument revealed the following flaws: 1) Wolff’s
Law has little cvidcntiary value as the primary theoretical premise in the argument. 2)
The argument does not make use of all available relevant evidence regarding the ability
of bone to functionally adapt to its mechanical environment. The review of non-
anthropological literature on skeletal mechanobiology presented in chapter 6 suggests
that the concept of functional adaptation is not an adequate substitute for Wolff’s Law as
the primary theoretical premise. 3) The argument has failed to make a case for the ability
to rule out or adequately control for a number of confounding variables, which could

provide alternative explanations for the biomechanical findings. Consequently, the

179

conclusion of the theoretical argument, that group variation in cross-sectional geometry
reflects differences in physical activity, does not follow from the premises.

The empirical argument is fundamentally flawed because its main premise, the
conclusion of the theoretical argument, is not strongly supported. Furthermore, the
second premise of the empirical argument, namely that group differences in cross-
sectional geometric properties exist, is also frequently not supported by the specific
findings of individual studies. Results of statistical analyses often failed to reject the null
hypothesis for many of the cross-sectional properties ascertained (i.e., few statistically
significant differences were found). In such instances, regardless of the strength of the
theoretical argument, conclusions regarding group differences in physical activity are

spurious.

Review of functional adaptation: a summa_ry

 

The theoretical argument reconstructed in this dissertation represents a plausible
line of reasoning relating cross-sectional geometry to physical activity patterns. In other
words, if none of the premises was false, and there was no additional relevant evidence
that could be brought to bear, then the conclusion of the argument would be at least as
plausible as the least plausible premise. However, the review of the literature on skeletal
biology and mechanobiology, presented in chapter 6, revealed a number of “facts,” which
bear directly on the argument. This review revealed that what is missing from the
argument put forth by Ruff and colleagues are all the caveats and contrary or equivocal
evidence, which have the effect of rendering behavioral inferences based on long bone
diaphyseal cross-sectional geometry less tenable. The following is a summary of this

information.

180

“(0le Law is cited as providing the primary theoretical premise linking cross-
sectional geometry to behavior. However, Wolff’s Law is based on a faulty argument
from analogy, which compares two things that appear similar, but which actually differ in
fundamental ways. Thus, the conclusion of the analogy proposed by Wolff, is not
supported. W'olfl’s Law is not a scientific law, and is not generally accepted by skeletal
biologists. The mathematical rules of bone construction predicted to exist by Wolff have
not been identified. Therefore, current usage of the phrase “Wolfl’s Law” is suggested to
be synonymous with the concept of functional adaptation, which asserts that bone has the
ability to adapt to its mechanical environment via a cell-based biological feedback system.

Experimental studies of functional adaptation have revealed numerous problems
that limit one’s ability to directly link long bone cross-sectional geometry with physical
activity. The biological processes involved in skeletal adaptation to mechanical stimuli
are still poorly understood; indeed, the relevant characteristic of the mechanical stimulus
responsible for initiating adaptation has not yet been identified. Research suggests the
likely existence of a range of physical activities, which would not initiate an adaptive
response. Furthermore, the actual in vivo loadings experienced by human bones during
physical activity are mostly unknown, and cannot be accurately predicted using
theoretical models or results of in vitro experiments. Consequently, association of levels
or types of physical activities to specific patterns of cross-sectional variation is based on
conjecture, not scientific evidence.

Experimental studies of functional adaptation in animals and humans have
generated highly variable and equivocal results, which suggest the possibility that
functional adaptation may be an “umbrella term” for a variety of effects, some systemic,

and some due to local mechanical factors. The most dramatic examples of “functional

181

adaptation” have come from studies of extreme mechanical loading imposed by
osteotomy and external loading devices in animals, or from studies of professional athletes
who commenced their training while they were still young and growing (i.e., prior to
skeletal maturity). It is questionable whether these extreme conditions apply to
archaeological populations. To assume that they do presumes the very thing
biomechanical studies are supposed to demonstrate.

Furthermore, studies of animals and humans have shown that the effects of
mechanical usage are both age-dependent and envelope-specific. Prior to skeletal
maturity, and particularly around puberty, bone seems to respond to mechanical loading
by altering its geometry through modeling drifts on both periosteal and endocortical
envelopes. In contrast, it is suggested that mature bone responds to mechanical loading
by: l) conserving existing bone mass via reduction of remodeling-dependent bone loss on
endocortical and trabecular surfaces, or 2) increasing volumetric bone mineral density, a
material property. Functional adaptation in mature bone does not seem to have primary
effects on cross-sectional geometry.

Moderate to moderately intense exercise in animals and adolescent humans tends
to primarily produce changes in endosteal dimensions, either through a reduction in bone
loss or an increase in bone formation. Statistically significant changes to periosteal
dimensions of lower limb bones are reported less frequently. However, changes in
periosteal dimensions are necessary to affect the external shape of bones, e.g., the relative
circularity of the femoral midshaft. Studies of extreme differential loading of upper limb
in long-term players of racquet sports show significant bilateral asymmetry in cross-
sectional properties, including cortical area and second moments of area. Changes

usually involve both periosteal and endocortical envelopes, particularly in younger

182

players. Degree of bilateral asymmetry is indirectly correlated with starting age.
Therefore, while on an individual level, bilateral asymmetry is at least partly due to
differential mechanical loading, differences in degree of bilateral asymmetry among
individuals or groups of individuals can be due to differences in the timing of the activity
relative to puberty.

An additional difficulty in interpreting behavior from long bone cross-sectional
geometry is the finding of counterintuitive effects of activity on these properties. Studies
have demonstrated that reductions in cross-sectional geometric properties can occur
following intense activity in both young animals and older women. Prolonged, intense
activity may have detrimental effects on immature bone, including decreases in cross-
sectional second moments of area. Furthermore, after skeletal maturity, increased
mechanical usage may inhibit remodeling-dependent bone loss on the endocortical
surface, and may therefore forestall the compensatory increase in periosteal diameter,
thus resulting in relatively lower cross-sectional area and second moments of area. In
contrast, reduced mechanical loading after skeletal maturity increases bone loss on the
endocortical envelope, which could produce a compensatory increase in periosteal
dimensions similar to, but potentially greater than that observed with age-related bone
loss. These scenarios contradict standard interpretations of cross-sectional geometry, i.e.,
greater values indicate more activity and lower values indicate less activity. One common
finding of most of the experimental research reviewed above is that activity-related
functional adaptation appears to be both site- and bone-specific. Therefore, the bone
location selected for analysis can affect the results, and therefore the conclusions of the

study.

183

The review of the literature presented in chapter 6 also demonstrated that long
bone cross-sectional geometry is influenced by non-mechanical factors. Consequently,
just as Ruff and colleagues have presented a plausible line of reasoning linking group
variation in cross-sectional geometry to physical activity, one could construct an equally
plausible line of reasoning linking the same variation to group differences in genetics and
diet. Genetic research provides substantial preliminary evidence that bone mass and
cross-sectional geometry are complex traits influenced by multiple, polymorphic genes.
Genes involved in controlling cross-sectional geometry have been identified in both mice
and humans. Furthermore, genes controlling bone architecture appear to have site- and
bone-specific effects. Genetic differences between groups could produce localized
variations in skeletal structure by: 1) direct local control of bone cell activities during
development, resulting in regional geometric size and shape differences among the
different long bones of the skeleton, which persist into adulthood, 2) indirect effects on
local geometry through primary effects on muscle mass and strength, body shape, body
mass, or bone material properties, or 3) altering the sensitivity of bone to its mechanical
environment, which could amplify or attenuate local mechanical stimuli. Studies have
also demonstrated dietary effects, including site- and bone-specific effects, on cross-
sectional geometric properties. The effects of nutrition on cross-sectional geometry are
most likely due to primary effects on body mass and body composition, bone metabolism,

and bone material properties.

Determm’ ’ g physical activity from cross-sectional geometgy: an analogy

 

To better illustrate why the concept of functional adaptation is an inadequate

theoretical foundation for relating physical activity to bone geometry, an analogy to the

184

relationship between exposure to sunlight and skin pigmentation is presented. In this
analogy, amount of time spent in the sun is analogous to level of physical activity, and
skin pigmentation, a measurable quantity, is analogous to cross-sectional geometry. The
theory that skin “functionally adapts” to sunlight exposure by becoming more deeply
pigmented is supported by scientific evidence. The underlying physiological mechanism
involves the production of melanin (pigmentation molecules) by melanocytes, a type of
skin cell, in response to ultraviolet (UV) radiation.

If a group of individuals with known biological characteristics (e.g., age, sex,
nutrition, health status, hormonal status, melanocyte density, degree of initial skin
coloration, genetic constitution, and ancestry) is exposed to a known amount of UV
radiation for a specified amount of time, a researcher could probably predict the group’s
changes in skin pigmentation. On the other hand, given skin samples from a population
with unknown characteristics (apart from age and sex), could a researcher estimate the
amount of time the individuals had spent in the sun? Given two such sample groups, one
significantly darker than the other, would it be reasonable to conclude that the darker-
skin group spent more time in the sun than the lighter-skin group?

Certainly, a plausible argument could be made, but it would be flawed in much
the same way as the reconstructed argument presented in this dissertation. For example,
there could be genetic differences between the groups. Perhaps genetic differences made
the darker-skin group more sensitive to UV radiation resulting in ontogenetic differences
in phenotype. Alternatively, the difference in skin pigmentation between the groups
could reflect a difference in ancestry (phenotypic differences that are phylogenetic in
origin). Perhaps the lighter-skin group used sunscreen, yet spent more time in the sun

than the darker-skin group, or maybe the darker-skin group used a sunless tanning lotion,

185

but actually spent little time in the sun. Perhaps the skin samples were from regions of the
body differentially exposed to sunlight, due to different types of apparel worn by two
groups. There could be physiological diflerences between the groups, such as hormone
levels, biochemical differences related to nutrition, or differences in melanocyte function.
Obviously there are numerous confounding factors that could potentially
contribute to the observed difference in skin pigmentation between the two groups, and it
would be impossible, without knowledge of the confounding factors, to determine with
any degree of confidence which group spent more time in the sun, in spite of the fact that
there is a known causal relationship between skin pigment and UV exposure. This
admittedly imperfect analogy illustrates a significant obstacle to using cross-sectional
properties to infer behavior, even given the general acceptance of the concept of

functional adaptation.

Conclusion

The biomechanical model developed by Ruff and colleagues is based on the
assumption that, because long bone cross-sectional geometric properties predict whole
bone mechanical behavior (e.g., rigidity and strength) under known or approximated
loading conditions (e.g., compression, tension, bending, and torsion), there is a direct
causal relationship between the development of these properties and the mechanical
forces they are capable of withstanding. The existence of the causal link is predicated on
Wolff’s Law. Users of this model suggest that by knowing a bone’s cross-sectional
geometry, one can infer the past mechanical loading patterns (i.e., the loading history),
and sometimes the specific physical activities, which gave rise to that geometry.

In contrast, this dissertation has suggested two counterarguments to the current

186

biomechanical model, which are based on a review of the non-anthropological skeletal
biology and mechanobiology literature spanning the past two decades. The first
counterargument is that research on functional adaptation has shown the relationship
between mechanical loading and bone architecture to be extremely complex and
variable, for reasons that are not yet understood. Currently, there is no generally
accepted model of functional adaptation that would allow predictions about behavior to
be made from long bone cross-sectional geometry. Such predictions would entail answers
to the following questions: 1) How do biological processes bring about site-specific
adaptation to mechanical loads? 2) How are these processes affected by systemic factors?
3) What is the “goal” of adaptation, i.e., is it to minimize some strain-related variable, or
maximize the predictability of the strain? 4) What are the circumstances under which an
adaptive response is initiated, e.g., what is the nature of the stimulus? 5) What is the
relationship between a specific activity and the resultant patterns of strain-related
variables generated in the specific bone in question?

The second counterargument made in this dissertation is that the results of
biomechanical analyses of archaeological skeletal remains can be potentially attributed to
one or more of the following alternative explanations: 1) Group differences in the starting
age of a particular physical activity, 2) Group differences in genes that directly or
indirectly affect cross-sectional geometry, or 3) Group differences in metabolic factors
(nutrition, health) that directly or indirectly affect cross-sectional geometry. It has also
been suggested that there is no reason to assume a priori that all observed group variation
in cross-sectional geometry is due to a single cause (physical activity or genetics or
nutrition). Whole bone mechanical properties, estimated by cross-sectional geometry,

reflect the cumulative effects of all these factors, but mechanical effects may have primacy

187

in one group, while nutritional factors or genetic differences may predominate in another.
In fact, the “cause” of variation could be bone-specific, with some bones predominantly
affected by their mechanical environment, others by nutrition, etc. Furthermore, there is
no reason to assume that there are not complex, antagonistic or synergistic relationships
among the various factors; certain combinations may be “good” for building bone, others
may be particularly detrimental. In addition, observed group differences could be due to
mechanical loading, but not as predicted. Increased mechanical loading can produce
bones with reduced cross-sectional properties, and reduced mechanical loading can
produce bones with larger cross-sectional properties, particularly second moments of
area. In other words, the nature of the relationship between physical activity and bone
structure could also differ among populations. Consequently, the conclusion of the
theoretical argument presented in chapter 5 must be rejected, and one must therefore
reject the three predictions of the biomechanical model: 1) Group variation in cross-
sectional geometry reflects group differences in levels and types of physical activity. 2)
Group variation in degree of sexual dimorphism of cross-sectional geometry is
attributable to differences in sexual division of labor. 3) Group variation in degree of
upper limb bilateral asymmetry reflects differential usage of the upper limb.

In conclusion, the strong likelihood that bones are capable of functionally
adapting to their mechanical environment in no way implies that temporal or geographic
differences in bone structure can be used to make reliable inferences (i.e., not purely
conjectural) about changes in behavior, beyond what is already inferred from other, more
traditional sources of evidence, e.g., archaeology. This is because there is ample
preliminary evidence to suggest that long bone cross-sectional geometry is affected by a

large number of factors, many of which cannot be known, and virtually none of which

188

can be adequately controlled for, particularly in archaeological contexts. It is
hypothetically possible for two populations performing identical activities, in the same
manner, and at the same frequency to demonstrate differences in their cross-sectional
geometry, including skeletally localized differences, because of group differences
unrelated to physical activity. If there is a story told by the skeleton, there is currently no

“Rosetta stone” to decipher its language.

Suggestions for future bioarchaeological research

The method of critical analysis used in this dissertation, namely the application of
techniques of critical reasoning in the analysis of a complex scientific argument, is a
valuable tool for reflexive evaluation in research and academia. Reconstruction of the
arguments used to support the fundamental tenets of any discipline can reveal critical
logical flaws, weaknesses in the underlying theory, as well as dogmatic adherence to
poorly supported assumptions and premises. As an interdisciplinary science, biological
anthropology frequently borrows theories, models, and data from non-anthropological
fields, such as medicine, epidemiology, evolutionary biology, biomechanics, physics,
chemistry, ecology, etc. This is true for a number of related sub-disciplines of biological
anthropology, including paleoanthropology, paleopathology, forensic anthropology, and
bioarchaeology. Consequently, it is incumbent upon bioarchaeologists to revise their
methods over time by continuing to review the non-anthropological literature and
keeping abreast of changes and controversies in the relevant disciplines. The occasional
reevaluation of core constructs can prevent the propagation of untenable conclusions, and
reveal potential alternative explanations for research findings. For example, the critical

analysis presented in this dissertation has shown that unquestioning acceptance of a

189

century old theory, namely Wolff’s Law, resulted in scientifically unsupportable tenets
regarding the primacy of the relationship between long bone cross-sectional geometry
and physical activity.

In spite of the problems with the biomechanical model used to infer behavior,
studies of past populations have contributed valuable data regarding group variation in
cross-sectional geometric properties, and this type of research should continue.
Bioarehaeological research can generate hypotheses, which could potentially be tested under
experimental (i.e., controlled) or quasi-experimental conditions in living populations. For
example, based on the tentative conclusions of the bioarchaeological studies reviewed in
chapter 3, one could construct a single-blind experiment in which the anthropologist-
researcher would attempt to “predict” the behaviors of living people from cross-sectional
geometric properties ascertained through radiological methods, such as peripheral
quantitative computed tomography (pQCT). The experiment would entail recruiting
individuals who had participated in known categories and levels of activity (e.g., running,
swimming, walking, cycling, skiing, or sedentary activities). Many biological
characteristics of the participants could be controlled for: body shape, body size, body
mass, age, sex, ancestry, nutritional and health status, age of commencement of activity,
length of time involved in the activity, etc. The researcher analyzing the cross-sectional
geometric data would be unaware of which activity each individual was involved in, and
would assign the individuals to predetermined groups based purely on their long bone
cross-sectional geometry. Statistical analyses could be performed to determine if the
resultant group assignments were better than what would be predicted by chance. Such
an experiment could potentially assess the efficacy of using cross-sectional geometry to

infer behavior under controlled conditions, in a population with known biological

190

characteristics and known behaviors. If under these conditions, predictive capability of
the method was acceptably high, its use in archaeological populations might bejustilied.

However, being historical scientists, bioarchaeologists are in the particularly
difficult position of never “knowing” whether their conclusions are correct (W aldron
2001) with the same level of confidence as scientists who are able to design replicable,
controlled, prospective experiments in an attempt to falszfi erroneous hypotheses. This is
a problematic issue when studying archaeological populations, where numerous, un-
testable assumptions must be made (e.g., “the osteological paradox” Wood 1992), and
points to the problems inherent in both developing and testing a model, such as the
biomechanical model for inferring behavior, on archaeological skeletal collections. In
bioarchaeology, research findings, even those that are statistically signyicant, should always be
interpreted conservatively and cautiously. Bioarehaeologists must be especially cognizant
of and explicit about the potential alternative explanations for their data, and must often
accept that, at the present time, there may be no way to distinguish among them, i.e., no
means of “differential diagnosis.” Tony Waldron (2001: 143) provided a positive view of
this potential frustration that should serve as an inspiration to all bioarchaeologists:
“There is no shame in not knowing and only when we admit that we don’t know, will we
try to think up means by which we may be able to know in the future.”

Even if long bone cross-sectional geometry cannot be used to reconstruct the
behavior of past populations (at this time or maybe ever), the continued study of group
variation in geometric properties using archaeologically-dcrived skeletal collections can
potentially contribute to an understanding of both skeletal biology and human
adaptation. There is a great deal more to be learned about cross-sectional geometric

variation, and it is important to do so. Understanding as much as possible about the

191

etiology of “strong” bones is a paramount goal of osteoporosis research. The distribution
of bone tissue within the cortex of a long bone is a primary determinant of whole bone
mechanical behavior, including strength. However, future research must be conducted
without the constraints of an activity-only perspective. For example, as discussed in this
dissertation, it has been hypothesized that cross-sectional geometry and longitudinal bone
curvature are architectural features of bone that control the direction of bending, thereby
making loads more predictable (Bertram and Biewener 1988). It could therefore be
informative to determine whether the relationship between cross-sectional geometry and
bone curvature varies among different populations. In addition, Ohman and Lovejoy
(2001) have suggested that the shape of the adult diaphyseal shaft is primarily a
consequence of the shape of the developing physis. It would therefore be informative to
study diaphyseal shape variation among prehistoric populations and its relationship to the
shape of the growth plate.

The most striking result of all the bioarchaeological studies reviewed in this
dissertation was the finding that so many of the differences among populations were non-
significant. This, by itself, is interesting. Given the vast potential for differences in
physical behaviors, nutrition, health, genetics, etc, one would expect there to be greater
differences in long bone structure. It could be useful to reexamine the significant findings
to see if a pattern emerges. Perhaps a meta-analysis of existing data would be
informative. Because cross-sectional shape and cortical area are generally similar among
widely divergent groups, it may be reasonable to ask: what is so different between the
groups where significant differences in cross-sectional geometry are observed? It is
possible that creative use of multivariate statistical methods could point future researchers

in a new direction.

192

BIBLIOGRAPHY

Abram AC, Keller TS, and Spengler DM (1988) The effects of simulated weightlessness
on bone biomechanical and biochemical properties in the maturing rat.Journa1 of

Biomechanics 21:755-767.

Agrawal AA (2001) Phenotypic plasticity in the interactions and evolution of species.
Science 294:321-326.

Akhter MP, Cullen DM, and Pederson EA (1998) Bone response to in vivo mechanical
loading in two breeds of mice. Calcified Tissue International 63:442-449.

Amblard D, Lafage-Proust M-H, and Laib A (2003) Tail suspension induces bone loss in
skeletally mature mice in the C57BL/6J strain but not in the C3H/HeJ strain.
Journal of Bone and Mineral Research 18:561-569.

Amtmann E (1974) An effect of gravity on the postnatal development of the human and
rat femur. Z. Anat. EntwickL-Gesch. 143:150-183.

Arden N K, and Spector TD (1997) Genetic influences on muscle strength, lean body
mass, and bone mineral density: a twin study. Journal of Bone and Mineral

Research 12:2076-2081.

Babij P, Zhao W, and Small C (2003) High bone mass in mice expressing a mutant LRP5
gene.Journal of Bone and Mineral Research 18:960-974.

Barondess DA (1998) Anthropometric and biomechanical assessment of skeletal structural
adaptations in bioarchaeological populations from Michigan and Western New

York. PhD Dissertation, Michigan State University, East Lansing.

Bass S, Pearce G, and Bradney M (1998) Exercise before puberty may confer residual
benefits in bone density in adulthood: studies in active prepubertal and retired
female gymnasts.Journal of Bone and Mineral Research 13:500-507.

Bass SL, Saxon L, and Daly CH (2002) The effect of mechanical loading on the size and
shape of bone in pre-, peri-, and postpubertal girls: a study in tennis players.
Journal of Bone and Mineral Research 17:2274-2280.

Beamer WG, Shultz KL, and Donahue LR (2001) Quantitative trait loci for femoral and
lumbar vertebral bone mineral density in C57BL/6J and C3H/he] inbred strains
of mice.Joumal of Bone and Mineral Research 16:1 195-1206.

BertramJEA, and Biewener AA (1988) Bone curvature: sacrificing strength for load
predictability?Joumal of Theoretical Biology 131:75-92.

193

BertramJEA, and Swartz SM (1991) The 'Law of Bone Transformation': a case of crying
Wolff? Biological Reviews 66:245-273.

Biewener A, and BertramJ (1994) Structural response of growing bone to exercise and

disuse.Joumal of Applied Physiology 76:946-955.

Biewener AA, ThomasonJ, Goodship AE, and Lanyon LE (1983) Bone stress in the
horse forelimb during locomotion at different gaits: a comparison of two
experimental methods.Joumal of Biomechanics 16:565-576.

Bikle DD, and Halloran BP (1999) The response of bone to unloading.Journal of Bone
and Mineral Metabolism 17:233-244.

Bird A (1998) Philosophy of Science. Buffalo: McGill-Queen's University Press.

Bourrin S, Toromanoff A, and Ammann P (2000) Dietary protein deficiency induces
osteoporosis in aged male rats.Journa1 of Bone and Mineral Research 1521555-

1563.

Bouvier M (1985) Application of in vivo bone strain measurement techniques to problems
of skeletal adaptations. Yearbook of Physical Anthropology 28:237-248.

Bradney M, Pearce G, and Naughton G (1998) Moderate exercise during growth in
prepubertal boys: changes in bone mass, size, volumetric density, and bone
strength: a controlled prospective study. Journal of Bone and Mineral Research

13:1814—1821.

Bravenboer N, Engelbregt MJT, Visser NA, Popp-Snijders C, and Lips P (2001) The
effect of exercise on systemic and bone concentrations of growth factors in rats.
Journal of Orthopaedic Research 19:945-949.

Bridges PS (1989) Changes in activities with the shift to agriculture in the southeastern
United States. Current Anthropology 30:385-394.

Bridges PS (1991) Skeletal evidence of changes in subsistence activities between the
Archaic and Mississippian time periods in Northwestern Alabama. In ML Powell,
PS Bridges and AMW Mires (eds.): What Mean These Bones? Studies in
Southeastern Bioarehaeology. Tuscaloosa: The University of Alabama Press, pp.
89-101.

Bridges PS (1995) Skeletal biology and behavior in ancient humans. Evolutionary
Anthropology 4:1 12-120.

Bridges PS, BlitzJH, and Solano MC (2000) Changes in long bone diaphyseal strength

with horticultural intensification in West-Central Illinois. AmericanJournal of

Physical Anthropology l 12:217-238.

194

Brock SL, and Ruff CB (1988) Diachronic patterns of change in structural properties of
the femur in the prehistoric American Southwest. AmericanJournal of Physical

Anthropology 75:1 13-127.

BrunskiJB (1999) In vivo bone response to biomechanical loading at the bone/dental-
implant interface. Adv Dent Res 13:99-1 l9.

Buhl K,Jacobs C, and Turner R (2001) Aged bone displays an increased responsiveness
to low-intensity resistance exercise.Journal of Applied Physiology 90:1359-1364.

Burger EH (2001) Experiments on cell mechanosensitivity: bone cells as mechanical
engineers. In SC Cowin (ed): Bone Mechanics Handbook, 2nd edition. New
York: CRC Press, pp. 28.1-28.16.

Burger EH, and Klein-NulendJ (1999) Responses of bone cells to biomechanical forces in
vitro. Adv Dent Res 13:93-98.

Burr DB (1992) Orthopedic principles of skeletal growth. In DS Carlson and SA
Goldstein (eds.): Bone Biodynamics in Orthodontic and Orthopedic Treatment.
Ann Arbor: University of Michigan Press, pp. 15-49.

Burr DB (1997) Muscle strength, bone mass, and age-related bone loss.Joumal of Bone
and Mineral Research 12:1547-1551.

Burr DB, Milgrom C, Fyrie D, Forwood MR, Nyska M, Finestone A, Hoshaw S, Saiag E,
and Simkin A (1996) In vivo measurement of human tibial strains during vigorous
activity. Bone 18:405-410.

Burr DB, and Piotrowski G (1982) How do trabeculae affect the calculation of structural
properties of bone? American Journal of Physical Anthropology 57:341-352.

Burr DB, Schafller MB, Yang KH, Lukoschek M, Sivaneri N, BlahaJD, and Radin EL
(1989) Skeletal change in response to altered strain environments: is woven bone a
response to elevated strain? Bone 10:223-233.

Carter DR (1981) The mechanical and biological response of cortical bone to in vivo
strain histories. In SC Cowin (ed): Mechanical Properties of Bone. New York:

The American Society of Mechanical Engineers, pp. 81-92.

Carter DR (1984) Mechanical loading histories and cortical bone remodeling. Calcified
Tissue International 36:819-824.

Carter DR, Smith DJ, Spengler DM, Daly CH, and Frankel VH (1980) Measurements
and analysis of in vivo bone strains on the canine radius and ulna.Journal of

Biomechanics 13:27-38.

CederblomJ, and Paulsen DW (2001) Critical Reasoning, 5th edition. Stanford, CT:

195

Wadsworth.

Chamay A, and Tschantz P (1972) Mechanical influences in bone remodeling.
Experimental research on Wolff's Law.Journal of Biomechanics 5: 1 73-180.

Churches AE, Howlett CR, Waldron K], and Ward GW (1979) The response of living
bone to controlled time-varying loading: method and preliminary results.Journal
of Biomechanics 12:35-45.

Cochran GVB (1982) A primer of orthopaedic biomechanics. New York: Churchill

Livingstone.

Copi IM, and Cohen C (1990) Introduction to Logic. New York: Macmillan Publishing
Company.

Cowin SC (1981a) Mechanical Properties of Bone. New York: The American Society of

Mechanical Engineers.

Cowin SC (2001a) Bone Mechanics Handbook, 2nd edition. New York: CRC Press.

Cowin SC (2001b) The False Premise of Wolff’s Law. In SC Cowin (ed.): Bone
Mechanics Handbook, 2nd edition. New York: CRC Press, pp. 30.1-30.15.

Cowin SC, and Moss ML (2001) Mechanosensory mechanisms in bone. In SC Cowin
(ed.): Bone Mechanics Handbook, 2nd edition. New York: CRC Press, pp. 29.1-
29.17.

Cullen DM, Smith RT, and Akhter MP (2000) Time course for bone formation with

long-term external mechanical loading. Journal of Applied Physiology 88: 1943-
1948.

CurreyJD (1979) Mechanical properties of bone tissues with greatly differing functions.
Journal of Biomechanics 12:313-319.

CurreyJD (1981) What is bone for? Property—function relationships in bone. In SC
Cowin (ed.): Mechanical Properties of Bone. New York: The American Society of
Mechanical Engineers, pp. 13-26.

CurreyJD (1984) The Mechanical Adaptations of Bones. Princeton: University Press.

CurreyJD (1984a) What should bones be designed to do? Calcified Tissue International
36:87-810.

CurreyJD (1984b) Can strains give adequate information for adaptive bone remodeling?

Calcified Tissue International 36:S1 18-Sl22.

CurreyJD (1987) The evolution of the mechanical properties of amniote bone.Journal of

196

Biomechanics 20: 1035-1044.

CurreyJD (1995) The validation of algorithms used to explain adaptive remodeling in
bone. In H Weinans and A Odgaard (eds.): Bone Structure and Remodeling. New
Jersey: World Scientific, pp. 9-13.

CurreyJD (1997) Was Wolffcorrect? Forma 12:263-266.

CurreyJD (2001) Ontogenetic changes in compact bone material properties. In SC
Cowin (ed.): Bone Mechanics Handbook, 2nd edition. New York: CRC Press, pp.
19.1-19.16.

CurreyJD (2002) Bones: Structure and Mechanics. Princeton: Princeton University Press.

CurreyJD, and Butler G (1975) The mechanical properties of bone tissue in children.
Journal of Bone andJoint Surgery 57A:810-814.

Davee AM, Rosen CJ, and Adler RA (1990) Exercise patterns and trabecular bone
density in college women.Journal of Bone and Mineral Research 5:245-250.

Demes B, Qin Y-X, StemJTJ, Larson SG, and Rubin CT (2001) Patterns of strain in the
macaque tibia during functional activity. AmericanJoumal of Physical
Anthropology 1 16:257-265.

Demes B, StemJTJ, Hausman MR, Larson SG, McLeod KJ, and Rubin CT (1998)
Patterns of strain in the macaque ulna during functional activity. American

Journal of Physical Anthropology 106:87-100.

DibbetsJMH (1992) One Century of Wolff’s Law. In DS Carlson and SA Goldstein
(eds.): Bone Biodynamics in Orthodontic and Orthopedic Treatment. Ann Arbor:
University of Michigan Press, pp. 1-13.

Drake TA, Hannani K, and KaboJM (2001) Genetic loci influencing natural variations
in femoral bone morphometry in mice.Jouma1 of Orthopaedic Research 19:51 1-

517.

Duncan CS, Blimkie CJR, Kemp A, Higgs W, Cowell CT, Woodhead H, BriodyJN, and
Howman-Giles R (2002) Mid-femur geometry and biomechanical properties in
15- to 18-yr-old female athletes. Medicine and Science in Sports and Exercise
34:673-681.

Duncan EL, Cardon LR, and SinsheimerJS (2003) Site and gender specificity of
inheritance of bone mineral density. Journal of Bone and Mineral Research

18:1531-1538.

Enlow DH (1968) Wolff’s law and the factor of architectonic circumstance. American
Journal of Orthodontics 54:803-822.

197

Evans FG (1953) Methods of studying the biomechanical significance of bone form.
American Journal of Physical Anthropology 1 1:413-435.

F orwood MR, and Burr DB (1993) Physical activity and bone mass: exercises in futility?
Bone and Mineral 21:89-1 12.

Forwood MR, and Parker AW (1987) Effects of exercise on bone growth: mechanical and
physical properties studied in the rat. Clinical Biomechanics 2:185-190.

F resia AE, Ruff CB, and Larsen CS (1990) Temporal decline in bilateral asymmetry of
the upper limb on the Georgia coast. In CS Larsen (ed.): The Archaeology of
Mission Santa Catalina De Guale: 2. Biocultural Interpretations of a Population

in Transition. New York: American Museum of Natural History, pp. 121-132.

Fritton SP, McLeod KJ, and Rubin CT (2000) Quantifying the strain history of bone:
spatial uniformity and self-similarity of low-magnitude strains.Journal of

Biomechanics 33:317-325.

Fritton SP, and Rubin CT (2001) In vivo measurement of bone deformation using strain
gauges. In SC Cowin (ed.): Bone Mechanics Handbook, 2nd edition. New York:
CRC Press, pp. 8.1-8.41.

Frost HM (1973a) Bone modeling and skeletal modeling errors. Springfield, IL: Charles
C Thomas.

Frost HM (1973b) Bone remodeling and its relationship to metabolic bone disease.
Springfield, IL: Charles C Thomas.

Frost HM (1985) The "new bone": some anthropological potentials. Yearbook of Physical
Anthropology 28:21 1-226.

Frost HM (1986) Intermediary Organization of the Skeleton, vol 1. Boca Raton: CRC

Press.

Frost HM (1987) Bone "mass" and the "mechanostat": a proposal. The Anatomical

Record 219:1-9.

Frost HM (1990a) Skeletal structural adaptations to mechanical usage (SATMU): l.
redefining Wolff’s law: the bone modeling problem. The Anatomical Record
226:403-413.

Frost HM (1990b) Skeletal structural adaptations to mechanical usage (SATMU): 2.
redefining Wolff‘s law: the remodeling problem. The Anatomical Record

226:414-422.

Frost HM (1995) Perspectives: on a "paradigm shift" developing in skeletal science.

198

Calcified Tissue International 56: 1-4.

Frost HM (1996) Perspectives: a proposed general model of the "mechanostat". The
Anatomical Record 244: 1 39-147.

Frost HM (1997a) Why do long-distance runners not have more bone? A vital
biomechanical explanation and an estrogen effect.Joumal of Bone and Mineral

Metabolism 15:9-16.

Frost HM (1997b) Why do marathon runners have less bone than weight lifters? A vital-
biomechanical view and explanation. Bone 20: 183-189.

Frost HM (1997c) On our age-related bone loss: insights from a new paradigm.Journal of
Bone and Mineral Research 12: 1539-1546.

Frost HM (2000) Does bone design intend to minimize fatigue failures? A case for the
aflirmative.Journa1 of Bone and Mineral Metabolism 18:278-282.

Frost HM (2001) From Wolff’s law to the Utah paradigm: insights about bone physiology
and its clinical applications. The Anatomical Record 262:398-419.

Frost HM, FerrettiJL, andJee WSS (1998) Perspectives: some roles of mechanical usage,
muscle strength, and the mechanostat in skeletal physiology, disease, and research.
Calcified Tissue International 62:1-7.

Gardiner EM, Sims NA, and Thomas GP (1998) Elevated osteoblastic vitamin D

receptor in transgenic mice yields stronger bones. Bone 2328176.

Gleeson PB, Protas EJ, LeBlanc AD, Schneider VS, and Evans HJ (1990) Effects of
weight lifting on bone mineral density in premenopausal women.Journal of Bone

and Mineral Research 5: 153- 158.

Goodship AE, and CunninghamJL (2001) Pathophysiology of functional adaptation of

bone in remodeling and repair. In SC Cowin (ed.): Bone Mechanics Handbook,
2nd edition. New York: CRC Press, pp. 26.1-26.31.

Goodship AE, Lanyon LE, and McFie H (1979) Functional adaptation of bone to
increased stress.Joumal of Bone andJoint Surgery 61A:539-546.

Graziano AM, and Raulin ML (1989) Research Methods: A process of inquiry. New
York: Harper and Row, Publishers.

Haapasalo H, Kannus P, and Sievanen H (1998) Effect of long-term unilateral activity on

bone mineral density of femalejunior tennis players.Journal of Bone and Mineral
Research 13:310-319.

Haapasalo H, Kontulainen S, Sievanen H, Kannus P,Jarvinen M, and Vuori I (2000)

1 99

Exercise-induced bone gain is due to enlargement in bone size without a change
in volumetric bone density: a peripheral quantitative computed tomography study
of the upper arms of male tennis players. Bone 27:351-357.

Haapasalo H, Sievanen H, Kannus P, and Heinonen A (1996) Dimensions and estimated
mechanical loading characteristics of the humerus after long-term tennis loading.

Journal of Bone and Mineral Research 1 1:864-872.

Hagino H, Okano T, Akhter MP, Enokida M, and Teshima R (2001) Effect of
parathyroid hormone on cortical bone response to in vivo external loading of the
rat tibia.Journal of Bone and Mineral Metabolism 19:244-250.

Hart RT (2001) Bone modeling and remodeling: theories and computation. In SC Cowin
(ed.): Bone Mechanics Handbook, 2nd edition. New York: CRC Press, pp. 31.1-
31.42.

Hasegawa Y, Schneider P, and Reiners C (2001) Age, sex, and grip strength determine
architectural bone parameters assessed by peripheral quantitative computed

tomography (pQCT) at the human radius.Journal of Biomechanics 34:497-503.

Hayes WC, and Snyder D (1981) Toward a quantitative formulation of Wolff‘s law in
trabecular bone. In SC Cowin (ed.): Mechanical Properties of Bone. New York:
The American Society of Mechanical Engineers, pp. 43-68.

Heinonen A, Sievanen H, Kannus P, Oja P, Pasanen M, and Vuori I (2000) High-impact
exercise and bones of growing girls: a 9-month controlled trial. Osteoporosis
International 11:1010—1017.

Heinonen A, Sievanen H, Kyrolainen H, PerttunenJ, and Kannus P (2001) Mineral

mass, size, and estimated mechanical strength of triple jumpers' lower limb. Bone

29:279-285.

Hernandez CJ, Beaupre GS, and Carter DR (2000) A model of mechanobiologic and
metabolic influences on bone adaptation.Journal of Rehabilitation Research and
Development 37:235-244.

Ho S, Chan SG, and Yi Q(2001) Soy intake and the maintenance of peak bone mass in
Hong Kong Chinese women.Journal of Bone and Mineral Research 16:1363-
1369.

Hsieh Y, Robling AG, Ambrosius WT, and Burr DB (2001) Mechanical loading of
diaphyseal bone in vivo: the strain threshold for and osteogenic response varies
with location.Joumal of Bone and Mineral Research 16:2291-2297.

Huiskes R (1995) The law of adaptive bone remodeling: a case for crying Newton? In H
Weinans and A Odgaard (eds.): Bone Structure and Remodeling. NewJersey:
World Scientific, pp. 15-24.

200

Huiskes R (2000) If bone is the answer, then what is the question?Journal of Anatomy
197:145-156.

Huiskes R, Weinans H, Grootenboer HJ, Dalstra M, F udula B, and Slooff TJ (1987)
Adaptive bone-remodeling theory applied to prosthetic-design analysis.Journal of
Biomechanics 20:1 135-1 150.

Hurwitz DE, Sumner DR, Andriacchi TP, and Sugar DA (1998) Dynamic knee loads

during gait predict proximal tibial bone distribution.Journal of Biomechanics
31 :423-430.

Ishijima M, Rittling SR, and Yamashita T (2001) Enhancement of osteoclastic bone
resorption and suppression of osteoblastic bone formation in response to reduced

mechanical stress do not occur in the absence of osteopontin.Joumal of
Experimental Medicine 193:399-404.

Iuliano-Burns S, Saxon L, and Naughton G (2003) Regional specificity of exercise and
calcium during skeletal growth in girls: a randomized controlled trial.Journal of

Bone and Mineral Research 18:156-162.

Jackes M (2000) Building the bases for paleodemographic analysis: adult age
determination. In MA Katzenburg and SR Saunders (eds.): Biological
Anthropology of the Human Skeleton. New York: Wiley-Liss, pp. 417-466.

Jarvinen T, Pajamaki I, and Sievanen H (2003) Femoral neck response to exercise and

subsequent deconditioning in young and adult rats.Joumal of Bone and Mineral
Research 18:1292-1299.

Jaworski ZF G, Liskova-Kiar M, and Uhthoff HK (1980) Effect of long-term
immobilisation on the pattern of bone loss in older dogs.Jouma1 of Bone and
Joint Surgery 62B:104-1 10.

Jee WSS (2001) Integrated Bone Tissue Physiology: Anatomy and Physiology. In SC
Cowin (ed.): Bone Mechanics Handbook, 2nd edition. New York: CRC Press, pp.
1.1-1.68.

Jee WSS, Li XJ, and Schaffler MB (199 1) Adaptation of diaphyseal structure with aging
and increased mechanical usage in the adult rat: a histomorphometrical and

biomechanical study. The Anatomical Record 230:332-338.
Jiang Y, ZhaoJ, and Genant HK (1997) Long-term changes in bone mineral and

biomechanical properties of vertebrae and femur in aging, dietary calcium
restricted, and/ or estrogen-deprived/-replaced rats.Journal of Bone and Mineral

Research 13:820-83 1.

Jones HH, PriestJD, Hayes WC, Tichenor CC, and Nagel DA (1977) Humeral

201

hypertrophy in response to exercise.Journal of Bone andJoint Surgery 59A:204-
208.

Jonsson B, Ringsberg K,Josefsson PO,Johnell O, and Birch-Jensen M (1992) Effects of
physical activity on bone mineral content and muscle strength in women: a cross-
sectional study. Bone 13:191-195.

Judex S, Gross TS, and Zemicke R (1997) Strain gradients correlate with sites of exercise-

induced bone-forming surfaces in the adult skeleton.Joumal of Bone and Mineral
Research 12:1737-1745.

Judex S, and Zemicke RF (2000) Does the mechanical milieu associated with high-speed
running lead to adaptive changes in diaphyseal growing bone? Bone 26:153-159.

Jurmain R (1999) Stories from the skeleton. Canada: Gordon and Breach Publishers.

Kannus P, Haapasalo H, and Sankelo M (1995) Effect of starting age of physical activity
on bone mass in the dominant arm of tennis and squash players. Annals of

Internal Medicine 123:27-31.

Kennedy KAR (1989) Skeletal Markers of Occupational Stress. In MY Iscan and KAR
Kennedy (eds.): Reconstruction of Life from the Skeleton. New York: Wiley-Liss,
pp. 129-160.

Khan K, McKay HA, Haapasalo H, Bennell KL, Forwood MR, Kannus P, and Wark
JD (2000) Does childhood and adolescence provide a unique opportunity for

exercise to strengthen the skeleton?Journal of Science and Medicine in Sport
3: 150-164.

Kimura T, Amtmann E, Doden E, and OyamaJ (1979) Compressive strength of the rat
femur as influenced by hypergravity.Journa1 of Biomechanics 12:361-365.

Kiratli BJ, Smith AE, and Nauenberg T (2000) Bone mineral and geometric changes
through the femur with immobilization due to spinal cord injury.Journal of

Rehabilitation Research and Development 37:225-233.

Kirk 3, Sharp CF, Elbaum N, Endres DB, Simons SM, MohlerJG, and Rude RK (1989)
Effect of long-distance running on bone mass in women.Journal of Bone and

Mineral Research 4:515-522.

Klein RF, Turner RJ, and Skinner LD (2002) Mapping quantitative trait loci that

influence femoral cross-sectional area in mice.Jouma1 of Bone and Mineral

Research 17:1752-1760.
Klein-NulendJ, SterckJGH, Semeins CM, Lips P,Joldersma M, BaartJA, and Burger

EH (2002) Donor age and mechanosensitivity of human bone cells. Osteoporosis
International 13:137-146.

202

Kodama Y, Umemura Y, Nagasawa S, Beamer WG, Donahue LR, Rosen CR, and
Baylink DJ (2000) Exercise and mechanical loading increase periosteal bone
formation and whole bone strength in C57BL/6J mice but not in C3H/Hej mice.
Calcified Tissue International 66:298-306.

Koller DL, Liu C, and Econs MJ (2001) Genome screen for Quantitative trait loci
underlying normal variation in femoral structure.Journal of Bone and Mineral

Research 16:985-991.

Koller DL, Rodriguez LA, ChristianJC, and Slemenda CW (1998) Linkage of a QTL
contributing to normal variation in bone mineral density to chromosome 1 1ql2-
13.Joumal of Bone and Mineral Research 13: 1903-1908.

Koller DL, White KB, and Liu G (2003) Linkage of structure at the proximal femur to
chromosomes 3, 7, 8, and 19.Journal of Bone and Mineral Research 18:1057-
1065.

Kontulainen S, Sievanen H, and Kannus P (2003) Effect of long-term impact-loading on
mass, size, and estimated strength of humerus and radius of female racquet-sports
players: a peripheral quantitative computed tomography study between young
and old starters and controls.Joumal of Bone and Mineral Research 18:352-359.

Lanyon LE (1980) The influence of function on the development of bone curvature. An
experimental study on the rat tibia.Journal of Zoology 192:457-466.

Lanyon LE (1981) The measurement and biological significance of bone strain in vivo. In
SC Cowin (ed.): Mechanical Properties of Bone. New York: The American
Society of Mechanical Engineers, pp. 93-105.

Lanyon LE (1984) Functional strain as a determinant for bone remodeling. Calcified
Tissue International 36:856-861.

Lanyon LE (1987) Functional strain in bone tissue as an objective, and controlling
stimulus for adaptive bone remodelling.Journal of Biomechanics 20: 1083-1093.

Lanyon LE (1993) Osteocytes, strain detection, bone modeling and remodeling. Calcified
Tissue International 53:8102-8107.

Lanyon LE, and Baggott DG (1976) Mechanical function as an influence on the structure
and form of bone.Joumal of Bone andJoint Surgery 58Bz436-443.

Lanyon LE, and Bourn S (1979) The influence of mechanical function on the

development and remodeling of the tibia.Journal of Bone andJoint Surgery
61Az263-273.

Lanyon LE, Goodship AE, Pye CJ, and MacFieJH (1982) Mechanically adaptive bone

203

remodelling.Journal of Biomechanics 15:141-154.

Lanyon LE, Magee P 1', and Baggott DG (1979) The relationship of functional stress and

strain to the processes of bone remodelling. An experimental study on the sheep

radius.Joumal of Biomechanics 12:593-600.

Lanyon LE, and Rubin CT (1984) Static vs dynamic loads as an influence on bone
remodelling.Journal of Biomechanics 17:897-905.

Larsen CS (1981) Functional implications of postcranial size reduction on the prehistoric
Georgia Coast, U.8.A.Joumal of Human Evolution 10:489-502.

Larsen CS, ed. (1990) The Archaeology of Mission Santa Catalina De Guale: 2.
Biocultural Interpretations of a Population in Transition. New York: American
Museum of Natural History.

Larsen C8 (1995) Biological changes in human populations with agriculture. Annual

Reviews in Anthropology 24: 185-2 1 3.

Larsen C8 (1997) Bioarehaeology: interpreting behavior from the human skeleton.
Cambridge, England: Cambridge University Press.

Larsen CS (2000) Skeletons in our closet. Princeton: Princeton University Press.

Larsen C8 (2002) Bioarchaeology: The lives and lifestyles of past people.Journal of
Archaeological Research 10:1 19-166.

Larsen CS, Griffin MC, and Hutchinson DL (2001) Frontiers of Contact: Bioarehaeology
of Spanish Florida.Journal of World Prehistory 15:69-123.

Larsen C8, and Ruff CB (1991) Biomechanical Adaptation and Behavior on the
Prehistoric Georgia Coast. In ML Powell, PS Bridges and AMW Mires (eds.):
What Mean These Bones? Studies in Southeastern Bioarehaeology. Tuscaloosa:
The University ofAlabama Press, pp. 102-1 13.

Larsen C8, Ruff CB, and Griffin MC (1996) Implications of changing biomechanical and
nutritional environments for activity and lifeway in the Eastern Spanish
borderlands. In BJ Baker and L Kealhofer (eds.): Bioarehaeology of Native

American Adaptation in the Spanish Borderlands. Gainesville: University Press of

Florida, pp. 95-125.

Larsen C8, Ruff CB, and Kelly RL (1995) Structural analysis of the Stillwater postcranial
human remains: behavioral implications of articular joint pathology and long
bone diaphyseal morphology. In C8 Larsen and RL Kelly (eds.): Bioarehaeology
of the Stillwater Marsh: Prehistoric Human Adaptation in the Western Great

Basin: Anthropological Papers of the American Museum of Natural History, pp.
107-133.

204

Ledger M, Holtzhausen L-M, and Constant D (2000) Biomechanical beam analysis of
long bones from a late 18th century slave cemetery in Cape Town, South Africa.
AmericanJoumal of Physical Anthropology 112:207-216.

Lee TC, and Taylor D (1999) Bone remodelling: should we cry Wolff. Irish Journal of
Medical Science 168:102-105.

Li K, Zemicke R, and Barnard R (1990) Effects of a High Fat-Sucrose Diet on Cortical
Bone Morphology and Biomechanics. Calcified Tissue International 47:308-313.

Li K, Zemicke R, and Barnard RJ (1991) Differential response of rat limb bones to
strenuous exercise.Journal of Applied Physiology 70:554-560.

Li Y, Seifert MF, and Ney DM (1999) Dietary conjugated linoleic acids alter serum IGF-
1 and IGF binding protein concentrations and reduce bone formation in rats fed
(n-6) or (n-3) fatty acids.Journal of Bone and Mineral Research 14:1 153-1 162.

Lieberman DE (1997) Making behavioral and phylogenetic inferences from hominid
fossils: considering the developmental influence of mechanical forces. Annual
Reviews in Anthropology 26:185-210.

Lieberman DE, Devlin MJ, and Pearson OM (2001) Articular area responses to
mechanical loading: effects of exercise, age, and skeletal location. American
Journal of Physical Anthropology 1 16:266-277.

Loitz BJ, and Zemicke RF (1992) Strenuous exercise-induced remodelling of mature
bone: relationships between in vivo strains and bone mechanics.Joumal of

Experimental Biology 170:1-18.

Lovejoy CO, Burstein AH, and Heiple KG (1976) The biomechanical analysis of bone
strength: a method and its application to platycnemia. AmericanJoumal of
Physical Anthropology 44:489-506.

Lovejoy CO, Meindl RS, and OhmanJC (2002) The Maka femur and its bearing on the
antiquity of human walking: applying contemporary concepts of morphogenesis to
the human fossil record. AmericanJournal of Physical Anthropology 1 19:97-133.

Lovejoy CO, and Trinkaus E (1980) Strength and robusticity of the Neandertal tibia.
AmericanJoumal of Physical Anthropology 53:465-470.

Majeska RJ (2001) Cell Biology of Bone. In SC Cowin (ed.): Bone Mechanics Handbook,
2nd edition. New York: CRC Press, pp. 2.1-2.24.

Martin RB (1990) Effects of simulated weightlessness on bone properties in rats.Journal
of Biomechanics 23:1021- 1029.

205

Martin RB (2000) Toward a unifying theory of bone remodeling. Bone 2621-6.

Martin RB, and Atkinson PJ (197 7) Age and sex-related changes in the structure and
strength of the human femoral shaft.Joumal of Biomechanics 10:223-231.

Martin RB, and Burr DB (1989) The Structure, Function, and Adaptation of Compact
Bone. New York: Raven Press.

Martin RB, Burr DB, and Sharkey NA (1998) Skeletal Tissue Mechanics. New York:
Springer.

MatsudaJ, Zemicke R, and Vailas A (1986) Structural and mechanical adaptation of

immature bone to strenuous exercise.Joumal of Applied Physiology 60:2028—
2034.

Mays S (1999) A biomechanical study of activity patterns in a Medieval human skeletal
assemblage. InternationalJournal of Osteoarchaeology 9:68-73.

Mikic B, van der Meulen MCH, Kingsley DM, and Carter DR (1996) Mechanical and
geometric changes in the growing femora of BMP-5 deficient mice. Bone 18:601-
607.

Milgrom C, F inestone A, Simkin A, Ekenman I, Mendelson S, Millgram M, Nyska M,
Larsson E, and Burr DB (2000) In vivo strain measurements to evaluate the
strengthening potential of exercises on the tibial bone.Joumal of Bone andJoint

Surgery 82Bz591-594.

MosleyJR, and Lanyon LE (2002) Growth rate rather than gender determines the size of
the adaptive response of the growing skeleton to mechanical strain. Bone 30:314-

319.

Muhlbauer R, Lozano A, and Reinli A (2002) Onion and a mixture of vegetables, salads,
and herbs affect bone resorption in the rat by a mechanism independent of their
base excess.Journal of Bone and Mineral Research 17:1230—1236.

N ara-Ashizawa N, Liu LJ, Higuchi T, Tokuyama K, Hayashi K, Shirasaki Y, Amagai H,
and Saitoh S (2002) Paradoxical adaptation of mature radius to unilateral use in

tennis playing. Bone 30:619-623.

Nomura S, and Takano-Yamamoto T (2000) Molecular events caused by mechanical
stress in bone. Matrix Biology 19:91-96.

Nordstrom P, Pettersson U, and Lorentzon R (1998) Type of physical activity, muscle
strength, and pubertal stage as determinants of bone mineral density and bone

area in adolescent boys.Joumal of Bone and Mineral Research 13:1 141-1 148.

Odgaard A, and W einans H (1995) Bone Structure and Remodeling. NewJersey: World

206

Scientific.

OhmanJC, and Lovejoy CO (2001) The shape of a long bone's shaft: bending stress or
growth plate form? American Journal of Physical Anthropology 1 14:1 15.

Orwoll ES (1992) The effects of dietary protein insufficiency and excess on skeletal health.
Bone 13:343-350.

Parsons TJ, Dusseldorp MV, and Van Der Vliet M (1997) Reduced bone mass in Dutch
adolescents fed a macrobiotic diet in early life.Journal of Bone and Mineral
Research 12:1486-1494.

Pederson EA, Akhter MP, and Cullen DM (1999) Bone response to in vivo mechanical
loading in C3H/HeJ mice. Calcified Tissue International 65:41-46.

Pennock RT (2000) Tower of Babel. Cambridge, Massachusetts: MIT Press.

Petit M, McKay H, and MacKelvie K (2002) A randomized school-based jumping
intervention confers site and maturity-specific benefits on bone structural

properties in girls: a hip structural analysis study.Journal of Bone and Mineral
Research 17:363-372. ‘

Pollack SR (2001) Streaming Potentials in Bone. In SC Cowin (ed.): Bone Mechanics
Handbook, 2nd edition. New York: CRC Press, pp. 24.1-24.22.

Prendergast PJ, and Huiskes R (1995) The biomechanics of Wolff's law: recent advances.
IrishJournal of Medical Science 164:152-154.

Raab D, Smith E, and Crenshaw T (1990) Bone mechanical properties after exercise
training in young and old rats.Journal of Applied Physiology 68: 130-134.

Reid IR (2002) Relationships among body mass, its components, and bone. Bone 31:547-
555.

Reis RJS (2003) From QTL mapping to genes: the long and winding road.Journal of
Bone and Mineral Research 18:186-189.

Richman C, Kutilek S, and Miyakoshi N (2001) Postnatal and pubertal skeletal changes
contribute predominantly to the differences in peak bone density between
C3H/HeJ and C57BL/6J mice.Joumal of Bone and Mineral Research 16:386-
397.

Robling AG, Burr DB, and Turner CH (2000) Partitioning a daily mechanical stimulus
into discrete loading bouts improves the osteogenic response to loading. Journal of

Bone and Mineral Research 15: 1596-1602.

Robling AG, Duijvelaar KM, GeeversJV, Ohashi N, and Turner CH (2001) Modulation

207

of appositional and longitudinal bone growth in the rat ulna by applied static and
dynamic force. Bone 29: 105-1 13.

Robling AG, and Turner CH (2002) Mechanotransduction in bone: genetic effects on
mechanosensitivity in mice. Bone 31:562-569.

Roesler H (1981) Some historical remarks on the theory of cancellous bone structure
(W olff’s Law). In SC Cowin (ed.): Mechanical Properties of Bones. New York: The
American Society of Mechanical Engineers, pp. 27-42.

Roesler H (1987) The history of some fundamental concepts in bone biomechanics.

Journal of Biomechanics 20: 1025-1034.

Rubin CT (1984) Skeletal strain and the functional significance of bone architecture.
Calcified Tissue International 36:81 1-818.

Rubin CT, and Lanyon LE (1984a) Dynamic strain similarity in vertebrates; and

alternative to allometric limb bone scaling.Journal of Theoretical Biology
107:321-327.

Rubin CT, and Lanyon LE (1984b) Regulation of bone formation by applied dynamic
loads.Journal of Bone andJoint Surgery 66Az397-402.

Rubin CT, and Lanyon LE (1985) Regulation of bone mass by mechanical strain
magnitude. Calcified Tissue International 37:41 1-417.

Rubin CT, McLeod KJ, and Bain SD (1990) Functional strains and cortical bone
adaptation: epigenetic assurance of skeletal integrity.Journal of Biomechanics

23:43-54.

Rubin CT, McLeod KJ, Gross TS, and Donahue HJ (1992) Physical stimuli as potent
determinants of bone morphology. In D8 Carlson and SA Goldstein (eds.): Bone
Biodynamics in Orthodontic and Orthopedic Treatment. Ann Arbor: University
of Michigan Press, pp. 75-91.

Rubinacci A, Covini M, Bisogni C, Villa 1, Galli M, Palumbo C, Ferretti M, Muglia MA,
and Marott G (2002) Bone as an ion exchange system: evidence for a link between

mechanotransduetion and metabolic needs. AmJ Physiol Endocrinol Metab
2822E851-E864.

Ruff CB (1983) The contribution of cancellous bone to long bone strength and rigidity.
American Journal of Physical Anthropology 61:141-143.

Ruff CB (1984) Allometry between length and cross-sectional dimensions of the femur

and tibia in Homo sapiens sapiens. AmericanJournal of Physical Anthropology
65:347-358.

208

Ruff CB (1987) Sexual dimorphism in human lower limb bone structure: relationship to
subsistence strategy and sexual division of labor.Journal of Human Evolution

16:391-416.

Ruff CB (1994) Biomechanical Analysis of Northern and Southern Plains femora:
behavioral implications. In DW Owsley and RLJantz (eds.): Skeletal Biology in
the Great Plains: migration, warfare, health, and subsistence. Washington D.C.:
Smithsonian Institution, pp. 235-245.

Ruff CB (1999) Skeletal Structure and Behavioral Patterns of Prehistoric Great Basin
Populations. In BE Hemphill and CS Larsen (eds.): Prehistoric Lifeways in the
Great Basin Wetlands. Salt Lake City: University of Utah Press, pp. 290-320.

Ruff CB (2000a) Biomechanical analyses of archaeological human skeletons. In MA
Katzenburg and SR Saunders (eds.): Biological Anthropology of the Human
Skeleton. New York: Wiley-Liss, pp. 71-102.

Ruff CB (2000b) Body size, body shape, and long bone strength in modern humans.
Journal of Human Evolution 38:269-290.

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.
American Journal of Physical AnthrOpology 60:359-381.

Ruff CB, and Hayes WC (1983b) Cross-sectional geometry of Pecos Pueblo femora and
tibiae--a biomechanical investigation: 11. Sex, age, and side differences. American

Journal of Physical Anthropology 60:383-400.

Ruff CB, and Larsen CS (1990) Postcranial biomechanical adaptations to subsistence
strategy changes on the Georgia coast. In CS Larsen (ed.): The Archaeology of
Mission Santa Catalina De Guale: 2. Biocultural Interpretations of a Population
in Transition. New York: Anthropological Papers of the American Museum of
Natural History, pp. 94-120.

Ruff CB, and Larsen CS (2001) Reconstructing behavior in Spanish Florida. In CS

Larsen (ed.): Bioarehaeology of Spanish Florida: the impact of colonialism.
Gainesville, FL: University Press of Florida, pp. 1 13-145.

Ruff CB, Larsen CS, and Hayes WC (1984) Structural changes in the femur with the

transition to agriculture on the Georgia Coast. American Journal of Physical
Anthropology 64:125-136.

Ruff CB, Trinkaus E, Walker A, and Larsen CS (1993) Postcranial robusticity in Homo.
1: Temporal trends and mechanical interpretation. American Journal of Physical

Anthropology 91:21-53.

Ruff CB, Walker A, and Trinkaus E (1994) Postcranial robusticity in Homo. III:

209

Ontogeny. AmericanJournal of Physical Anthropology 93:35-54.

Russell KF, Choi I, and Larsen CS (1990) The paleodemography of Santa Catalina de
Guale. In C8 Larsen (ed.): The Archaeology of Mission Santa Catalina De Guale:
2. Biocultural Interpretations of a Population in Transition. New York:
Anthropological Papers of the American Museum of Natural History, pp. 36-49.

Sandler RB, CauleyJA, Horn DL, Sashin D, and Kriska AM (1987) The effects of
walking on the cross-sectional dimensions of the radius in postmenopausal
women. Calcified Tissue International 41 :65-69.

Seeman E (2002) An exercise in geometry.Journal of Bone and Mineral Research
17:373-380.

Shaw SR, Zemicke RF, Vailas AC, DeLuna D, Thomason DB, and Baldwin KM (1987)
Mechanical, morphological and biochemical adaptations of bone and muscle to
hindlimb suspension and exercise.Joumal of Biomechanics 20:225-234.

Shultz KL, Donahue LR, and Bouxsein ML (2003) Congenic strains of mice for

verification and genetic decomposition of quantitative trait loci for femoral bone
mineral density.Journal of Bone and Mineral Research 18: 175-185.

Slemenda CW, MillerJZ, Hui SL, Reister TK, andJohnston CCJ (1991) Role of physical
activity in the development of skeletal mass in children.Joumal of Bone and
Mineral Research 6:1227-1233.

Slemenda CW, Turner CH, Peacock M, ChristianJC, SorbelJ, Hui SL, andJohnston
CC] (1996) The genetics of proximal femur geometry, distribution of bone mass
and bone mineral density. Osteoporosis International 6:178-182.

Specker BL, and Binkley T (2003) Randomized trial of physical activity and calcium

supplementation on bone mineral content in 3- to 5-year old children.Journal of

Bone and Mineral Research 18:885-892.

Specker BL, Mulligan L, and Ho M (1999) Longitudinal study of calcium intake, physical

activity, and bone mineral content in infants 6-18 months of age.Joumal of Bone

and Mineral Research 14:569-576.

Steinberg ME, and TruetaJ (1981) Effects of activity on bone growth and development in
the rat. Clinical Orthopaedics and Related Research 156:52-60.

StockJ, and Pfeiffer S (2001) Linking structural variability in long bone diaphyses to
habitual behaviors: foragers from the Southern African Later Stone Age and the

Andaman Islands. AmericanJournal of Physical Anthropology 1 15:337-348.

Sun D, Krishnan A, and Zaman K (2003) Dietary n-3 fatty acids decrease

osteoclastogenesis and loss of bone mass in ovariectomized mice.Journal of Bone

210

and Mineral Research 18:1206-1216.

Tanck E, Blankevoort L, Haaijman A, Burger EH, and Huiskes R (2000) Influence of
muscular activity on local mineralization patterns in metatarsals of the embryonic

mouse. Journal of Orthopaedic Research 18:613-619.

Tanck E, HommingaJ, Van Lenthe GH, and Huiskes R (2001) Increase in bone volume

fraction precedes architectural adaptation in growing bone. Bone 28:650-654.

Tate MLK (2001) Interstitial Fluid Flow. In SC Cowin (ed.): Bone Mechanics Handbook,
2nd edition. New York: CRC Press, pp. 22.1-22.29.

Toma CD, Ashkar S, and Gray ML (1997) Signal transduction of mechanical stimuli is
dependent on microfilament integrity: identification of osteopontin as a

mechanically induced gene in osteoblasts.Joumal of Bone and Mineral Research
1 2: 1626- 1636.

Trinkaus E, Churchill SE, and Ruff CB (1994) Postcranial robusticity in Homo. II:
Humeral bilateral asymmetry and bone plasticity. AmericanJoumal of Physical

Anthropology 93: 1-34.

Turner CH (1992) Functional determinants of bone structure: beyond W olff's law of bone
transformation. Bone 13:403-409.

Turner CH (1999) Toward a mathematical description of bone biology: the principle of
cellular accommodation. Calcified Tissue International 65:466-471.

Turner CH, and Forwood MR (1995) What role does the osteocyte network play in bone
adaptation? Bone 16:283-285.

Turner CH, Forwood MR, and Otter MW (1994) Mechanotransduction in bone: do
bone cells act as sensors of fluid flow? FASEB 8:875-878.

Turner CH, Hsieh Y, and Muller R (2000) Genetic regulation of cortical and trabecular
bone strength and microstructure in inbred strains of mice.Journal of Bone and
Mineral Research 15:1 126-1 131.

Turner CH, Hsieh Y, and Muller R (2001a) Variation in bone biomechanical properties,
microstructure, and density in BXH recombinant inbred mice.Journal of Bone

and Mineral Research 16:206-213.
Turner CH, Roeder RK, and Wieczorek A (2001b) Variability in skeletal mass, structure,
and biomechanical properties among inbred strains of rats.Joumal of Bone and

Mineral Research 16: 1532-1539.

Uhthoff HK, andJaworski ZFG (1978) Bone loss in response to long-term
immobilisation.Journal of Bone andJoint Surgery 60B:420-429.

211

van der Meulen MCH, Beaupre GS, and Carter DR (1993) Mechanobiologie influences
in long bone cross-sectional growth. Bone 14:635-642.

van der Meulen MCH, and Huiskes R (2002) Why mechanobiology? A survey article.
Journal of Biomechanics 35:401-414.

van der Meulen MCH,Jepsen KJ, and Mikic B (2001) Understanding bone strength: size
isn't everything. Bone 29:101-104.

V olkman SK, Galecki AT, and Burke DT (2003) Quantitative trait loci for femoral size
and shape in a genetically heterogeneous mouse population.Joumal of Bone and
Mineral Research 18:1497-1505.

Waldron T (2001) Shadows in the Soil: Human Bones and Archaeology. Charleston, SC:
Arcadia Publishing, Inc.

Walton DN (1989) Informal Logic. New York: Cambridge University Press.
W arburton N (2000) Thinking from A to Z, 2nd edition. New York: Routledge.

Wescott DJ (2001) Structural variation in the humerus and femur in the American Great
Plains and adjacent regions: differences in subsistence strategy and physical
terrain. PhD Dissertation, University of Tennessee, Knoxville.

\Vohl GR, Boyd SK,Judex S, and Zemicke RF (2000) Functional adaptation of bone to
exercise and injury.Journal of Science and Medicine in Sport 3:313-324.

W'olff J (1892) The Law of Bone Remodeling (1986 translation by Maquet, P. and R.
Furlong). New York: Springer-Verlag.

Wolff J (1988) The Classic: Concerning the interrelationship between form and function
of the individual parts of the organism (Translation by Max Scheck). Clinical
Orthopaedics and Related Research 228:2-1 1.

“/00 SL-Y, Kuei SC, Amiel D, Gomez MA, Hayes WC, White FC, and Akeson WH
(1981) The effect of prolonged physical training on the properties of long bone: a
study of Wolff's law.Journal of Bone andJoint Surgery 63Az780-787.

VVoodJW, Milner GR, Harpending HC, and Weiss KM (1992) The Osteological
Paradox. Current Anthropology 33:343-370.

W under CC, Welch RC, and Cook KM (1979) Femur strength as influenced by growth,
bone-length and gravity with the male rat.Joumal of Biomechanics 12:501-507.

Yershov Y, Baldini T, and Villagomez S (2001) Bone strength and related traits in

HcB/ Dem recombinant congenic mice.Journal of Bone and Mineral Research

212

16:992- 1003.

Yingling VR, Davies 8, and Silva MJ (2001) The effects of repetitive physiologic loading
on bone turnover and mechanical properties in adult female and male rats.

Calcified Tissue International 68:235-239.

Young MF, and Dieudonne SC (2001) Molecular Biology Techniques to Measure
Skeletal Gene Expression. In SC Cowin (ed.): Bone Mechanics Handbook, 2nd
edition. New York: CRC Press, pp. 3.1-3.l7.

Young MF, and Xu T (2001) Creating transgenic mice to study skeletal function. In SC
Cowin (ed.): Bone Mechanics Handbook, 2nd edition. New York: CRC Press, pp.
4.1-4.19.

Zanker CL, Gannon L, Cooke CB, and Gee KL (2003) Differences in bone density, body
composition, physical activity, and diet between child gymnasts and untrained
children 7-8 years of age.Journal of Bone and Mineral Research 18:1043-1050.

Zemicke R, Salem G, and Barnard R (1995) Long-Term, High-Fat-Sucrose Diet Alters

Rat Femoral Neck and Vertebral Morphology, Bone Mineral Content, and
Mechanical Properties. Bone 16:25-31.

213

l1131111111]11111111111111111111111

2 3 0250