Gradlization of the ModernHuman Skeleton

The latent strength in our slender bones teaches lessonsabout human lives, current and past

Christopher B. Ruff

People often think of the human skel-eton as a symbol of death. In one

sense this is true: Bone resists decompo-sition better than flesh, so it has a greaterchance of being preserved after death.However, bone is also a living tissue.The skeleton is remarkably dynamicduring life—even in adults—and it re-sponds to metabolic needs and mechan-ical requirements. When muscles growstronger, the underlying bone adaptsby changing its physical shape to bearthe increased stress. Likewise, atrophiedmuscles lead to weakened bones. In thisway, our bones tell the story of our liveslong after we're gone.

Because most of the archaeologicaland paleontological record consists ofbones, skeletal remains form the basisfor most of what we know about hu-man ancestors and our evolution. Mycolleagues and I read the stories ofthese ancient peoples through the bonesthey've left behind. This work buildson a record of controlled laboratory ex-

Christopher B. Ruff is a professor and director ofthe Center for Functional Anntotni/ and Evohdionnt tile folms Hopkins Unirersiti/ Sclinol of Medi-cine in Baltimore, Maryland. He receiivd his Ph.D.in biological antliropologij at the University ofPennsylvania in 1981 and carried out postdoctoralwork in the Orthopaedic Biomechanics Labora-tory at Beth Israel Hospital and Harvard MedicalSchool. Ruff joined the faculty of Johns Hopkinsill 1983. His major research interest is the relationbetween mechanical forces and bone structure inliving and fossil animals. Most recently he has beenstudying the effects of changes in subsistence strat-egy on skeletal form in Hobceiw Europe and West-em Asia. Address: Center for functional Anatomyand Evolution. Johns Hopkins Unii'ersit]/ Schoolof Medicine, 1830 E. Monument Street, Baltinwre,MD 21205. Internet: cbruff@jhmi.edii

periments that help connect the specificgeometr)' of a bone to a certain patternof behavior (and vice versa).

One of the ctmclusions we've reachedis that the skeletons of human beingshave changed over the past 2 millionyears, becoming less robust, or moregracile. Our explanation for this phe-nomenon provides new insight into themodem problem of osteoporosis andconfirms that our bones retain their an-cient capacity to grow strong.

Sticks and Stones and SidewalksVertebrate skeletons must be both rigidand strong, but animals have to balajicethese needs against the cost of pmducing,maintaining and maneuvering a heavi-er skeleton. Consequently, skeletal sizetends to match mechanical requirementsclosely. There are disadvantages to hav-ing grossly under- or ox'erbuilt bones.

Engineers use a similar concept ofbuilding a structure to meet a specificneed. This so-called factor of safety isthe ratio of actual strength to requiredstrength under maximum load. Biome-chanical engineers such as R. McNeillAlexander at the University of Leedsestimate that factors of safety for verte-brate limb bones generally range fromtwo to four. Indeed, limb-bone fracturesare relatively rare. Scientists estimatethat an individual bone has a one tothree percent lifetime risk of fracture,based on data from a variety of species.

There is one condition, however, thatleads to far higher rates of bone failure:osteoporosis, in which bone becomesmore porous and brittle. This condi-tion is particularly prevalent amongolder women. In the United States, theestimated lifetime risk of osteoporotic

hip fracture is 17 percent among whitewomen and 6 percent among whitemen. The great majority of these frac-tures occur in adults over 50 and resultfrom minimal to moderate trauma—usually a fall from standing height orless. Broken vertebrae and wrists arealso common in this age group.

The risk of fracture among the elderlyisn't uniformly high in every popula-tion, howe\'er. Northern Europeans andpeople of European ancestry in otherparts of the world (North America, Aus-tralia, New Zealand and South Africa)have Hgher rates than African, African-American and some Asian and Pacificgroups. During the second half of the20th century, the fractvire rates amonghigh-risk European populations greweven higher, but this increase was mod-est compared with the spike in fracturesamong residents of Hong Kong, Singa-pore and other rapidly urbanizing popu-lations in Southeast Asia. In these areas,the low incidence of hip fracture in the1960s quickly gave way to a rate similarto that of Europeans by the 1980s.

Older people suffer more brokenbones because the mass and strength ofbone decrease with age. There is no sin-gle reason why this occurs, or why someindividuals and populations are morevulnerable than others. Like other com-plex traits, age-related changes in boneresult from interactions between envi-rcjnmental and genetic factors. Scientistshave linked changes in bone strength tovariations in physical activity, the lev-els of dietary calcium and vitamin D,and alcohol and tobacco use. However,among these, physical activity is the vari-able most likely to account for the get>graphic heterogeneity in the incidence of

508 American Scientist, Volume 94

;. Inc.

Figure 1. The bones of anatomically modem humans are, on average, more slender and less strong than those of our ancestors. Although this trendof gracilization has heen progressing for more than a million years, the pace of skeletal weakening has accelerated over the past few millennia.However, individuals who perform rigorous exercise develop much more rigid bones, nearly equal in strength to the skeletons of our ancestors.This computer-generated artwork is based on individual x-ray images.

www.americanscientist.org 2006 November-December 509

1,600

1,200 -

8 0 0 -

4 0 0 -

50-59 60-69 70-79

age (years)

80-89

Figure 2. The rate of hip fractures among Hong Kong women grew dramatically between1966 and 1985, a period of increasing industrialization. The incidence now equals or exceedsthe fracture rate found in some high-risk European populations. Bone fractures among HongKong men show a similar trend. A decrease in physical activity, coupled with traditionally lowcalcium intakes, probably explains the elevated risk. (Adapted from Lau ct«/. 1990.)

fractures. In Hong Kong, for example, in-creasing urbanization and mechanizationhave led to a reduction in weight-bearingactivities. Together with a low-calciumdiet, this shift to a more sedentary life-style is the most likely explanation forthe recent increase in fracture rates. JohnChalmers and K. C. Ho at the Universi-ties of Edinburgh and Hong Kong, whoauthored the 1970 paper on this subject,actually predicted this trend.

From an evolutionary perspective, thehigh incidence of broken bones late in

life is a recent development for Hoinosapiens. Elderly members of our ances-tral populations had many fewer hipfractures than senior citizens do today,even after controlling for their shorterlife spans. In fact, out of many thousandsof excavated archaeological specimens,only a handful of hip fractures have everbeen described. Yet the same specimensdo show age-related loss of bone massor density—and at a frequency and se-verity similar to modem levels. How canthese seemingly contradictory obser\'a-

+115%

o\o

Figure 3. The thickness of the dense outer cortex is not the hest index of bone strength. Some phy-sicians diagnose osteoporosis when the bone cortex gets thinner. This conclusion may not be cor-rect if the bone's diameter increases, a change that can accompany aging. This cartoon comparestwo hypothetical cross sections of bone: baseline (a) and increased-diameter (IJ). Compared withthe blue cortex of a, the green cortex of (' makes up a smaller fraction of its cross section (expressedas percent cortical thickness or percent cortical area). However, the bone with the expanded crosssection (b) would be more rigid (second moments of area) and stronger (section moduli).

tions be reconciled? The answer comesfrom basic principles of engineering.

Predicting Bone StrengthWlien an engineer analyzes a structureto see how strong it is, he or she takesinto account not only the design, butalso the properties of the constructionmaterials and the size of the stnicture.This analysis is a little simpler for bonebiomechanics because the material prop-erties of bone—at least the kind foundin most parts of the skeleton—are fairlyconstant within and between species.On this basis it appears that bone tissueevolved only once. From giant whalesto tiny shrews, the many skeletons thathave existed during vertebrate historyare largely made of the same substance.As a result, those of us who study oldbones can compare samples that havebeen buried for milleiinia. The materialproperties of these bones may changewith time, becoming friable or fossil-ized, but their size and shape are gener-ally well preserved.

Because the bones themselves aren'tsuitable for testing, we use their dimen-sions in a computer simulation, or mod-el, to predict their original strength. Wecan represent the long bones of the limbsfairly precisely using the same type ofmodel that an engineer would use tojudge the strength of a staictural beam.The most important properties in thissimulation are quantities that describeits cross-sectional size and shape. Thecross-sectional area of bone determines itsaxial rigidity {in other words, how resis-tant the bone is to deformation undercompression or ter\sion). Other proper-ties are called the second moments of areaand section moduli, which measure thebone's resistance to bending in differentplanes as well as to torsion (twisting).These last two variables depend on theamount of material in the cross section,but they depend even more on how farfrom the center of the cross section thatmaterial is distributed. Section modulivary as a product of the third power ofthe distance from the central axis, andsecond moments of area vary as a prod-uct of the fourth power.

The vast majority of bone-agingstudies, including those of archaeo-logical samples, have concentrated onthe bone's mass, volume, density or acombination of these, However, theseparameters yield an incomplete pictureof skeletal biomechanics. Based on engi-neering principles, animal models andhuman observations, the architectural

510 American Scientist, Volume 94

properties described above are probablymore important in determining overallstrength and the likelihood of fracture.

In fact, without an engineering per-spective, some commonly evaluatedmeasurements can easily be nusinter-preted. For example, the cortex—thedense outer shell that makes up theshaft and ends of long bones—gets thin-ner with age in nearly everyone. Thischange is most often expressed as adecrease in percent cortical thickness,and most clinicians view it as a sign ofincreasing fragility. However, this in-terpretation is only valid if the outer di-ameter of the bone stays the same. If thebone itself gets wider with age— a docu-mented phenomenon—then its strengthand rigidity can increase even as thecortex gets to be a smaller percentageof the diameter Some scientists hypoth-esize that the age-related widening oflong bones compensates for the loss ofoverall bone mass, although the effectseems to vary between human popula-tions. Anthropologists rarely considerthis factor in their studies of humanskeletal remains.

More Brain, Less BrawnBiomechanical analysis can tell usmuch about past human popula-tions and skeletal changes over time.With my colleagues Erik Trinkaus atWashington University in St. Louisand Brigitte Holt at the University ofMassachusetts, Amherst, I have inves-tigated the changes in relative bonestrength among modern humans andtheir ancestors from the past severalmillion years. I refer to "relative" bonestrength because we find that it is crit-ical to control for variation in bodysize: Larger bodies have longer, stron-ger bones—a rule that is particularlytrue for weight-bearing bones. Thisrelation applies to adults within andbetween hominid species, and also toindividuals as they grow.

Although we can easily measure thelength of an archaeological or fossil bonespecimen, body mass is more diificult tocalculate. Our methods for doing so arebased on the size of the femoral head(the ball that fits into the hip socket), andestimates of the individual's original to-tal body height and breadth (from longbone lengths and pelvic width, usually).We avoid the possibility of circular rea-soning because the femora! head showspatterns of growth that are largely inde-pendent of die shaft, where biomechani-cal properties are measured.

In one set of experiments, we ob-tained optical cross sections from thelong bones of more than 100 humanspecimens that were 5,000 to 1.9 mil-lion years old. All were members of thegenus Homo, either direct ancestors orclose relatives of modem humans. Theremains came from all over Africa andEurasia, although the earliest speci-mens (older than 600,000 years) wereAfrican, and most of the later ones wereEuropean. We took a few of the mea-surements from photographs of brokenfossils, but most of the data came fromx-ray scans combined with detailedmolds of the originals. (Computed to-mography would have been preferable,but this technology was unavailable atmost of the museums that housed theseremains.) Because of differences in thecondition of the samples, the best datacame from the diaphysis, or middle re-gion of the femur (about mid-thigh). Tocontrol for differences in body size, wenormalized the section modulus at thissite by dividing by the product of esti-mated body mass and femoral length.

In a plot of these relative strengthsversus sample age, the best fit is anexponential decrease, that is, a decline

2 0 0 -

Figure 4. Outwardly similar to its moderncounteqiart (right), the femur of a 1.9-inillion-year-old Homo species (left) has a much thick-er cortex. (Adapted from Ruff et al. 1993.)

£ 1 5 0 -

CO

£ 100

50

t >

105 104 10years before the present (log scale)

102

Figure 5. The average bone strength of human beings and human ancestors has fallen duringthe past 2 million to 3 million years. This graph shows temporal changes in the strength (sectionmoduli) of the femur relative to body size. Time is expressed in logarithmic units because therelation is exponential. The solid line shows a regression drawn through 104 individual fossils(purple circles) attributed to the genus Homo; the dotted line is the theoretical extension of thisline to the near-present The blue triangle is "Lucy," an earlier human relation laustralopith-ecine) not included in the regression. The blue circles indicate mean values for three popula-tions of anatomically modem humans: an archaeological sample of Native Americans from theAmerican Southwest who lived about 900 years ago, and East Africans and U.S. whites fromthe early to mid-20th century. The error bars indicate plus or minus two standard deviations forthe Native Americans and East Africans; individual body-size data were unavailable for U.S.whites. (Adapted from Ruff 2005.)

www.americanscientist.org 2006 November-December 511

that gets progressively steeper. There isa lot of scatter in the data, but the trendis statistically significant. From roughly2 million years ago to about 5,000 yearsago, human bones became almost 15percent weaker.

P

I

Figure 6. Australopitheais aftirensis was oneof the earliest bipedal hominids and prob-ably an ancestor of the genus Homo. Thisspecimen, "Lucy," is unusually complete for a3.1 million-year-old skeleton, which enablesan accurate estimate of overall body size (shestood about 1.1 meter or 3 feet, 8 inches tall).Taking body size into account, Lucy's femuris considerably stronger than those of morerecent human ancestors, despite the fact thather species spent less time walking than didtheir descendants.

We also compared these values tomore recent human remains from lessthan 1,000 years ago, using three di-verse populations from North Americaand East Africa. The bones from all ofthese specimens were, on average, 15percent weaker than those from 5,000years ago—in fact, they lay below theextended regression line from the maindata set. Thus, relative bone strengthdecreased even faster during the past5,000 years than it did over the previous2 million years.

The Penalty for Tool UseBecause the remains are seldom com-plete, it becomes more difficult to re-construct body size for fossils of humanancestors older than 2 million years. Theexception is "Lucy," the famous 3.1 mil-lion-year-old skeleton from Ethiopia,which is complete enough that we canestimate her original body size with afair degree of confidence. Judging byour calculation of Lucy's relative femoralstrength, her bones were even strongerthan those of the early Homo specimer\sand almost twice as strong as an averagehuman from several hundred years ago.

Lucy was an australopithecine—amember of a very early group on ornear the lineage leading to modem hu-mans. Although Lucy and her relativeswalked bipedally, they most likely keptto the trees more than later Homo andprobably weren't long-distance travel-ers. The arm bones from other mem-bers of this group appear to be verystrong, which may reflect this behav-ior. (Cross sections from Lucy's armbones were unfortunately not avail-able for study.) If true, this hypothesismakes Lucy's relative femoral strengtheven more remarkable, since she prob-ably walked less than do many mod-em humans.

We see similar results among modemnonhuman primates, such as chimpan-zees, gorillas and baboons. Relative tobody size, their arm and leg bones aremuch sfronger than those of humans.(The difference is greater for the armthan the leg, of course, since these spe-cies kx:omote using both sets of limbs.)Not surprisingly, nonhuman primatesalso appear to have much strongermuscles than humans relative to theirsize. Thus, the bone strength resultsmake sense: Stronger muscles generategreater force, greater force increases themechanical stress on the bones, and thisstress induces the bone to adapt overtime by becoming more rigid.

Figure 7. Adult chimpanzees in the wild weighbetween 30 and 60 kilograms (between 66 and132 pounds), yet primatologists estimate thatchimps are more than twice as strong as hu-man beings, on average. The musculaturethat generates such force is especially visibleon Cinder, a female chimpanzee at the SaintLouis Zoo who lacks body hair as a result ofthe disease aloyecia arcata. Remarkably, Cin-der is somewhat small compared with othermembers of her species. The bones of chim-panzees, like their muscles, are much strongerthan those of modem humans. (Photographcourtesy of Carol Weerts, Saint Louis Zoo.)

The bone-strength results also implythat earlier human ancestors had stron-ger muscles, a hypothesis consistentwith the large muscle-insertion scarsthat anthrtipologists see on many of thespecimens. (The scar observations, how-ever, are far from definitive, as the bonesof some modem nonhuman primatesdon't carry such marks.) If early humanswere indeed more muscular than you orI, they probably got that way (at least inpart) because they were more active andvigorous. This, in tum, is probably re-lated to tool use: The rise of technologythat has accompanied human evolutionhas, in effect, progressively shielded thehuman body from its environment.

Scientists have never found toolsfrom Lucy's time period, and in terms oftechnology, her kin probably interactedwith the natural world much as mod-em chimpanzees and gorillas do. This

512 American Scientist, Volume ''4

absence of technology predicts a veryhigh relative bone strength. Early Homohad simple stone tools, and their boneswere not quite as strong as Lucy's, butstill n^uch more rigid than the averagemodern human's. The comparativelyweak bones of recent humans are aninevitable consequence of a sedentarylifestyle and ubiquitous, sophisticatedtools that make physical strength muchless relevant to survival. For perhaps thesame reason, human bodies as a wholehave gotten smaller on average duringthe past 50,000 years, despite very re-cent increases that are probably tied tobetter nutrition and health care.

Tennis, Anyone?Many animal studies have demonstrat-ed that bones become stronger with ex-ercise. For example, the femurs of youngpigs that ran an hour every day for ayear were 24 percent stronger than thoseof sedentary controls. The increase waspurely a product of geometric chang-es—primarily a thickening of the bonecortex—with no effect on bone mate-rial properties. Other studies have notedsimilar findings, further supporting thepractice of using a bone's geometry toestimate past mechanical loadings.

The human upper limb presents a"natural experiment" of this sort be-cause almost e\ eryone favors one handor the other for most tasks. Thus, scien-tists can compare bone adaptations indominant and nondominant arms of

people with differing activity levels. Un-like most animals, human beings don'tuse their arms for locomotion, leavingthem freer to reflect asymmetrical usageand structure. In fact, human forelimbsshow more left-right asymmetry thanthose of any other mammal. In the nor-mal population, right-handers and left-handers have similar magnitudes—butopposite directions—of bilateral asym-metry in the strength of the secondmetacarpal (a hand bone). Righties arestronger on the right, lefties on the left.One of the advantages of this compari-son is that it inherently controls for non-behavioral factors such as overall bodysi2e, nutrition and hormonal influences,which affect both sides equally.

In the 1970s, Henry H. Jones (whowas one of my undergraduate advi-sors), James D. Priest and their cowork-ers at Stanford University x rayed thearms of professional tennis players.They focused on the bone in the upperarm called the humerus and comparedthe dimensions of the bone cortex in theplaying and nonplaying arms. In thedominant arm, the outer surface of thecortex had gotten bigger and the innersurface had gotten smaller.

Almost 20 years later, we were ableto get Jones's original measurementsand calculate geometric section proper-ties. By our calculations, playing-sidehumeri were more than 40 percentstronger on average than non-playing-side humeri. By contrast, in nonprofes-

sional athletes the average left-rightasymmetry in bone strength is about 5to 10 percent. The tennis players in thestudy were between 14 and 39 yearsold and had played for at least 5 years.All started playing between ages 5 and19. Interestingly, these changes weremore pronounced among subjects whobegan playing earlier in life. Since wepublished our data, other studies havealso noted that bone adaptation is partlyage-dependent. Adult skeletons remainresponsive to increased exercise, butthey respond more slowly and less com-pletely thiin those of children.

Unlike the cross sections from the hu-meral shaft, the si^es of the right and leftelbow joints were more similar in thetennis players. In animal studies, too,the size of the articular surface and thelength of the bone (which depends onarticular growth) are less affected by me-chanical loadings than are diapbysealcross sections. The cortex of the shaft oflong bones appears to be particularly"plastic," or responsive to changes inmechanical loadiiigs during life, so it'shelpful for reconstructing the behav-iors of past populations. The restrainedgrowth and remcxieling of the ends ofthe bones may help to avoid incongruityat the joint surface that predisposes it toproblems such as artliritis.

We also examined the humeri of Ne-andertals (50,000 to 100,000 years old)and anatomically modern humans(10,000 to 30,000 years old) for which

Figure 8. Vigorous exercise can lead lo large increases in bone strength. This diagram shows the average difference in cross-sectional dimen-sions of the humerus between nonplaying ibluc) and playing (blue plus green) arms of professional tennis players measured by Henry Jonesand coworkers at Stanford University. The author and his colleagues re-analyzed the data and calculated average increases in bone rigidity andstrength of 62 percent and 45 percent, respectively, in the playing arms. The changes were most pronounced in players who began training at anearly age, such as Amelie Mauresmo, currently ranked among the top female professionals in the world. Based on data from Ruff ct al., 1994.

www.americanscientist.org 2006 November-December 513

both arms had been preserved. (Suchfossils are somewhat rare.) (3ur ances-tors showed bilateral asymmetry inshaft strength almost as great as that ofthe modem professional tennis players.Based on experimental studies usinghuman volunteers, Daniel Schmitt andSteven E. Churchill at Duke Universityhave suggested that this asymmetry re-flects the use of weapons or tools thatloaded one arm more than the other.Interestingly, all five Neandertals and19 of the 24 early modem humans havestronger right humeri, indicating right-handedness. In the larger group of earlymodem humans, the frequency at whichwe found stronger arm bones on theright side—about 80 percent—is similarto the rate of right-arm-bone asymmetryin a wide range of recent human popu-lations. Not surprisingly, it's also similarto the frequency at which people favortheir right hand. (Many sources statethat 90 percent of the population is right-handed, but they're usually referring towriting preference, wliich can be biasedby various cultural factors. In cross-cultural studies, the frequency of right-arm preference is a little lower for activi-ties such as throwing and hammering.)

The Leg Bone Connected to the ...It is clear that limb bones, at least, re-spond to increased mechanical force bychanging their geometry, adding bonematerial to strengthen the outer cortex.The bilateral asymmetry found in protennis players and pre-agrarian humanssuggests that bt ne strength can increaseby 40 percent or more under the properconditions. Conversely, reduced me-chanical loads lead to the loss of bone.For example, after six months in spaceunder zero-gravity conditions, the legbones of astronauts aboard the Inter-national Space Station were 20 percentweaker on average than before, basedon section moduli derived from bonemineral scans of the femoral neck. Para-lyzed patients experience even greaterlosses over longer time periods.

These examples may represent ex-tremes, but they demonsfrate the poten-tial for bone to adapt when circumstanc-es change. In a few years, the strength ofa person's bone structure can change asmuch as the total average change overthe past 2 million years of human evolu-tion. Although some of this evolution-ary change may reflect nonmechanicalfactors, including genetic changes, themost parsimonious explanation is thatthe human skeleton has simply adapted

to a lesser workload. The gracilizationof the modem human skeleton is prob-ably a direct result of our consistentlyadvancing technology.

This conclusion has implications forunderstanding the etiology of osteopo-rotic fractures. As noted above, brokenhips are more common in urbanized,less physically active populations. Thesignificant increase in skeletal strengththat is gained through physical exer-cise, if maintained throughout a per-son's lifespan, may help to preventsuch fractures.

The bones of our ancestors showthat the human skeleton was oncestronger than it is today. Studies ofmodern athletes, however, demon-strate that we are still able to achievesuch strength. The skeletons in ourevolutionary closet can teach us somevaluable lessons about modern life-styles and their consequences.

BibliographyAlexander, R. M. 1981. Factors of safety in the

staicture of animals. Science Progress, Oxford67:109-130.

Auerbach, B. M, and C. B. Riiff. 21X16. Limb bonebihiteral asymmetry; Variability and com-monality among mtxiem humans, louriwl ofHumm Emiution 50:203-218.

Brandwood, A., A. S. Jayes and R. M. Alexander.1986. Incidence of healed fracture in the skel-etons of birds, mollusks and primates, lotirimlofZoohgi/ 208:55-62.

Chalmers, J., and K. C. Ho. 1970. Cetigraphicalvariations in senile osteoporosis. The associa-tion with physical activi^. journii! of Bone andloiiit Surgery 52B:667-675.

Holt, B. M. 2tX)3. Mobility in Upper Palet>lithicand Mesolithic Europe: Evidence from thelower limb. American Journal of PIn/sical An-thropolagy 122:200-215.

Jones, H. H., J. D. Priest, W. C. Hayes, C. C.Tichenor and D. A. Nagel. 1977. Humeralhypertrophy in response to exercise, journalof Bone and joint Surgery 59A:204-208.

Lang, T, A. LeBlanc, H. E\'ans, Y. Lu, H. Genantand A. Yu. 2004. Cortical and trabecular bonemineral loss hxmi the spine and hip in long-duration spaceflight. joiirmil of Bone and M/H-

19:1006-1012.

t^u, E. M., C. Cooper, C. Wickham, S. Doiinanand D. J. Barker. 1990. Hip fracture in HongKong and Britain. Internationa! journal ofEpi-dttmobgi/19:1119-1121.

Mosekilde, L. 1995. Osteoporosis and exercise.Bone 17:193-195.

Roy, T A., C. B. Ruff and C. C. Plato. 1994. Handdominance and bilateral asymmetry in struc-ture of the second metacarpal. American lour-i!alofPln/i:iailAnthropoiogif94:20?r-2n.

Ruff, C. B. 2000. Btxly size, body shape, and longbont? strength in modem humans, journal ofHuman Evolution 38:269-290.

Ruff, C. B. 2005. Mechanical determinants ofbone form: Insights from skeletal remains.journal ofMusculoskeletal andNeuronal Interac-tions 5:202-2}2.

Ruff, C. B., E. Trinkaus, A. Walker and C. S. Lars-en. 1993. Postcranial robusticity in Homo, I:Temporal trends and mechanical interpretii-tion. American journal of Phydccil Anthropologvi91:21-53.

Ruff, C. B., A. Walker and E. Trinkaus. 1994.Postcranial aibusticity in Homo, 111: Ontog-eny. American Journal of Physical Anthropology93:35-54.

Schmitt, D., and S. E. Churchill. 2003. Experi-mental evidence concerning spear use in Ne-andertals and early modem humans. JourmIof Arcliaeological Science 30:103-114.

For relevant Web links, consult thisissue of American Scientist Online:

http://www.americanscientist,.org/lssueTOC/issue/902

, BOBBY LE5K-0

514 American Scientist, Volume 94