The effects of body proportions on thermoregulation: anexperimental assessment of Allen’s rule

Michael J. Tilkens a,*, Cara Wall-Scheffler a, Timothy D. Weaver b, Karen Steudel-Numbers a

a Department of Zoology, University of Wisconsin, Madison, WI 53706, USAb Department of Anthropology, University of California, Davis, CA 95616, USA

Received 30 January 2006; accepted 6 April 2007

Abstract

Numerous studies have discussed the influence of thermoregulation on hominin body shape concluding, in accordance with Allen’s rule, thatthe presence of relatively short limbs on both extant as well as extinct hominin populations offers an advantage for survival in cold climates byreducing the limb’s surface area to volume ratio. Moreover, it has been suggested that shortening the distal limb segment compared to the prox-imal limb segment may play a larger role in thermoregulation due to a greater relative surface area of the shank. If longer limbs result in greaterheat dissipation, we should see higher resting metabolic rates (RMR) in longer-limbed individuals when temperature conditions fall, since theresting rate will need to replace the lost heat. We collected resting oxygen consumption on volunteer human subjects to assess the correlationbetween RMR and lower limb length in human subjects, as well as to reexamine the prediction that shortening the distal segment would havea larger effect on heat loss and, thus, RMR than the shortening of the proximal segment. Total lower limb length exhibits a statistically significantrelationship with resting metabolic rate (p< 0.001; R2! 0.794). While this supports the hypothesis that as limb length increases, resting met-abolic rate increases, it also appears that thigh length, rather than the length of the shank, drives this relationship. The results of the present studyconfirm the widely-held expectation of Allen’s rule, that short limbs reduce the metabolic cost of maintaining body temperature, while longlimbs result in greater heat dissipation regardless of the effect of mass. The present results suggest that the shorter limbs of Neandertals, despitebeing energetically disadvantageous while walking, would indeed have been advantageous for thermoregulation.Published by Elsevier Ltd.

Keywords: Resting metabolic rate; Climate; Hominin evolution

Introduction

Numerous studies have discussed the influence of cold adap-tation on hominin body shape (e.g., Roberts, 1978; Trinkaus,1981; Beals et al., 1984; Franciscus and Trinkaus, 1988; Holli-day and Trinkaus, 1991; Ruff, 1991, 1993, 1994). It is widelybelieved that numerous physical features, including relativelyshort limbs, offer an advantage for survival in glacialclimates (Coon, 1962; Badoux, 1965; Roberts, 1978). Suchadvantages of shorter limbs include reducing the surface

area to volume ratio (Coon, 1962; Trinkaus, 1981; Wolpoff,1989; Frayer et al., 1993) in compliance with Allen’s rule,an ecogeographical pattern in which individuals from higherlatitudes exhibit shorter appendages than individuals of thesame species living closer to the equator (Allen, 1877). Be-cause the body dissipates heat through the skin, an organ-ism’s surface area is directly proportional to the amount ofheat lost. A homeothermic animal is better able to retainbody heat in cold temperatures when its relative surfacearea is reduced, while an organism living in a warm environ-ment would be better adapted to dissipate heat if the relativesurface area of the body is increased.

Both extinct and extant hominin populations have beenshown to exhibit body proportions which conform to Allen’s ex-pectations. Holliday (1997a) has characterized the European

* Corresponding author. Tel.: "1 608 215 5079.E-mail addresses: mjtilkens@wisc.edu (M.J. Tilkens), cscheffler@wisc.

edu (C. Wall-Scheffler), weaver@eva.mpg.de (T.D. Weaver), ksteudel@facstaff.wisc.edu (K. Steudel-Numbers).

0047-2484/$ - see front matter Published by Elsevier Ltd.doi:10.1016/j.jhevol.2007.04.005

Journal of Human Evolution 53 (2007) 286e291

Neandertal postcranial morphology as hyper-polar; that is, theirshort limbs and broad trunks are similar to those of modern hu-mans living in Arctic environments. However, anatomicallymodern human (AMH) populations living in arctic conditions(Hrdlicka, 1930; Newman, 1953; Coon, 1962; Auger et al.,1980; Holliday and Trinkaus, 1991; Holliday, 1997b) havesomewhat less pronounced body proportions than Neandertals,at least when considered in a multivariate manner, despite theharsher conditions of the Holocene Arctic (Holliday, 1997a).

Additionally, Holliday and Trinkaus (1991) demonstratedthat, while the European Neandertals possess both shortenedproximal and distal limb segments relative to trunk height,the shortening is much more pronounced in the distal limbsegment. As a result, hyper-polarization is most conspicuouslyexpressed in the extreme shortening of the distal segments ofboth the upper and lower limbs. When Trinkaus (1981) re-gressed the length of the radius on the humerus and the lengthof the tibia on the femur, the European Neandertals fell mark-edly below the regression line for recent humans. Holliday(1999) hypothesized that the reduction in the distal lowerlimb segment among Neandertals might improve their abilityto thermoregulate in the cold because the smaller diameterin the distal lower limb segment suggests any change in lengthwould have a dramatic effect on relative surface areadmoreso than a change in the length of the proximal segment. Hol-liday’s (1999) investigation into limb segment changes in earlyto late Upper Paleolithic anatomically modern humans, how-ever, reported a decrease in total leg length with no concurrentchange in the ratio of distal to proximal segment lengths.

Despite some disagreement regarding the theoretical frame-work of Allen’s rule (Scholander, 1955; Irving, 1957), the factremains that the majority of workers accept the validity of thistrend. Nonetheless, Allen’s rule is a generalization that restssolely on the dependability of the empirical pattern (Mayr,1956). While the expected theoretical basis of Allen’s rule(thermoregulation) seems very plausible, observed patternsdo not necessarily have the causal basis that we might expect(Taylor et al., 1974; Heglund et al., 1982).

The question, however, is amenable to direct experimentaltesting. Numerous experiments have investigated the relation-ship between human morphology and climate in both heat(Robinson, 1942; Wyndham et al., 1970; Epstein et al.,1983) and cold conditions (Sloan and Keatinge, 1973; McAr-dle et al., 1984; Toner and McArdle, 1988; Stocks et al.,2004); several studies reporting human responses to cold con-ditions report a reduction in heat loss with a reduction in thesurface area to volume ratio, consistent with Allen’s rule(Sloan and Keatinge, 1973; Kollias et al., 1974; McArdleet al., 1984). Experiments focusing on human reactions tocold adaptations generally use water immersion techniquesdaprocedure not employed in the current study. Additionally,several reoccurring problems in these studies, such as smallsample sizes, unbalanced participant sex ratios, and unnaturaltest conditions, have been discussed in previous reviews(Steegman, 1975; Hanna et al., 1989; Ruff, 1994). The currentstudy investigates the physiological response to cold whilecorrecting for these methodological issues.

If, in fact, longer limbs result in greater heat dissipation, weshould see higher resting metabolic rates (RMR) in longerlimbed individuals when temperature conditions fall substan-tially below that of body temperature, since RMR will needto replace the lost heat. The present study tests this hypothesisby assessing the correlation between RMR and lower limblength in human subjects. Further, we reexamine Holliday’s(1999) theoretical prediction that shortening the distal segmentwill have a larger effect on heat loss and, consequently, RMRthan would shortening the proximal segment.

Methods

Resting metabolic rates for the current study were quanti-fied by determining the average oxygen consumption overa four-minute period after the subject had been sitting for atleast eight minutes, allowing the values to represent steady-state aerobic respiration solely. Previous work from our labo-ratory (Steudel-Numbers and Tilkens, 2004) has found that thetotal of twelve minutes is an adequate length of time toachieve steady-state conditions. We make no claim to be mea-suring basal metabolic rate, which indeed would requirea much more specialized experimental design. Here we aresimply determining whether extremity length is correlatedwith short-term resting metabolic rate in a cool environment.

All 20 subjects included in the study participated in multi-ple sessions, each taking place on separate days (to avoid sameday changes in VO2 values). The data reported were collectedfrom at least three separate sessions for each subject and theresults averaged.

Oxygen consumption was measured using a SensorMedicsVmax 29 c respiratory gas analysis system. The temperature inthe laboratory was controlled and monitored before and aftereach individual session took place. The average temperaturewas 21.9 #C (s.d.! 0.99). Since average human body temper-ature hovers around 37 #C, [humans usually die if their bodytemperature deviates from about 35 to 40 #C (Kormondyand Brown, 1998)] this gives a discrepancy of approximately15 #C. This serves as cold enough to activate a response, butnot so cold that vasoconstriction would be a factor (McArdleet al., 2001). Each subject wore a tee shirt, running shorts,and running shoes, thus exposing the appendages to theroom air temperature. During each trial, the subjects sat com-fortably in a bean bag chair with their knees bent loosely at anobtuse angle, exposing them to the air. All subjects werehealthy and between the ages of 18 and 35. The Human Sub-ject Committee of the University of Wisconsin approved theexperimental procedures. Volunteers for this study completeda written informed consent form after the nature, purpose,and possible risks were carefully explained and prior to the be-ginning of each subject’s first session.

Anthropometric measurements (mass, height, thigh length,and shank length) were measured during each subject’s first ses-sion in the lab. Thigh length was obtained by measuring thedistance between the proximal portion of the greater trochanterto the lateral midpoint of the knee (equal distance betweenthe femoral epicondyles and the tibial plateau), while the

287M.J. Tilkens et al. / Journal of Human Evolution 53 (2007) 286e291

measurements of the shank included the distance between thelateral midpoint of the knee to the most lateral portion of the lat-eral malleolus. Together, thigh length and shank length consti-tuted the total length of the lower limb. All lower limbmeasurements were collected using calipers. At the conclusionof all trials, lean body mass and percent fat mass were deter-mined at the University of Wisconsin Clinics using dual-energyX-ray absorptiometry (DEXA). This method utilizes low-en-ergy X-ray beams to penetrate body tissues, reconstructingthe bone mineral content, lean mass, and fat mass from the at-tenuation of the X-rays as they scan the entire body. Thismethod is quickly becoming the gold standard in body compo-sition analysis (Kohrt, 1995; Prior et al., 1997).

P-values, as well as coefficient of determination values,were determined from regressions between RMR and height,mass, thigh length, shank length, lean mass, and percent fat,as well as total lower limb length (thigh length" shanklength). In addition, because both RMR and limb length arehighly correlated with lean mass, we examined the correlationbetween RMR and lower limb length while controlling forlean mass. Statistical investigations for changes in distal andproximal portions of the lower limb were conducted with

unpaired t tests; p-values and the coefficient of determinationfor both the shank and the thigh were calculated.

Results

Anthropometric measurements of our 20 volunteer subjectsare listed in Table 1. Subjects ranged from 50e95 kg with totallower limb lengths from 71 to nearly 99 cm. Pearson correla-tions as well as significance levels between resting metabolicrate and the collected anthropometric variables (mass, height,total lower limb length, thigh length, shank length, lean mass,and percent fat) are presented in Table 2. As expected, totalmass (p! 0.002) as well as lean mass (p< 0.001) are bothhighly correlated with resting metabolic rate.

Total lower limb length exhibits a statistically significant re-lationship with resting metabolic rate (p< 0.001; R2! 0.794).While this strongly supports the hypothesis that as limblength increases, resting metabolic rate increases (Fig. 1), italso appears that this relationship is driven primarily by thelength of the thigh, rather than the length of the shank (Table 2).When both lean mass and percent fat are controlled for, shank

Table 1Body dimensions and resting metabolic rates for volunteer subjects1

Sex Mean(VO2RMR)

Mass Averagethigh length

Averageshank length

M 0.25 70.59 45.20 47.20M 0.18 61.09 42.90 45.10M 0.18 71.49 41.45 38.40M 0.17 73.30 38.95 39.40M 0.18 69.68 39.55 39.90M 0.26 81.90 48.80 44.90M 0.28 70.59 43.85 46.00M 0.29 82.58 43.30 43.15M 0.25 95.45 47.35 51.60M 0.18 62.26 36.65 41.60M 0.23 90.45 43.73 51.27

Male mean 0.22 75.39 42.88 44.41s.d. 0.05 10.95 3.60 4.48

Sex Mean(VO2RMR)

Mass Averagethigh length

Averageshank length

F 0.14 50.23 36.70 36.80F 0.17 65.45 40.05 38.90F 0.22 63.57 42.97 41.07F 0.21 61.54 42.05 43.10F 0.14 59.50 38.80 36.45F 0.17 65.91 41.50 41.50F 0.16 66.82 38.70 45.60F 0.15 95.45 35.35 36.30F 0.17 52.73 38.13 36.90

Female mean 0.17 64.58 39.36 39.62s.d. 0.03 12.96 2.52 3.36

Total mean 0.20 70.53 41.30 42.26Total s.d. 0.05 11.6 3.57 4.62

Conversions of external to skeletal measures (Porter, 1996) suggest that external thigh measures arew2% shorter than skeletal and that shank arew8% longer thanskeletal.1 Mean and standard deviation values for resting metabolic rates, mass, and thigh and shank lengths between the sexes. Reported thigh and shank lengths are

external segment measures, not bone (femur and tibia, respectively) measurements.

288 M.J. Tilkens et al. / Journal of Human Evolution 53 (2007) 286e291

length has no significant correlation with RMR, while thighlength remains highly correlated (see Table 3).

To determine whether the correlation between RMR andlower limb length might be simply a result of their mutual cor-relation with lean mass, we calculated the standardized resid-uals of the regression of lower limb length on lean mass andthe residuals of the regression of RMR on lean mass. Residuallimb length retained a significant relationship with residualRMR (p! 0.018, R2! 0.274). This significance level, lowerthan for the original variables, is primarily a consequence ofa single subject who has a Cook’s value (i.e., the calculationof a data point’s degree of influence on the regression; Cookand Weisberg, 1999) substantially above all other subjects(nearly 0.5 while others are less than 0.01). This individualhas a relatively high metabolic rate and relatively short limblength (filled circle in Fig. 2). When this individual isremoved, the significance (p! 0.001) as well as the correla-tion (0.711) are comparable to those of the untransformedvariables.

Because females generally have higher fat mass and lowerlean mass than males, one might predict a gender difference in

the effect of lower limb length on RMR due to the greater in-sulation in females. In our sample, females had roughly twicethe percent body fat of males (22% to 11% on average) andtwo-thirds the lean mass (42 kg to 62 kg). Nonetheless,females have higher significance levels and correlation valuesthan do males (males: p! 0.023, R! 0.673; females:p! 0.016, R! 0.764).

Discussion

Thus, the results of the present study confirm the widelyheld expectation of Allen’s rule for cold-adapted populations(Schreider, 1950, 1964, 1975; Trinkaus, 1981; Ruff, 1991,1994): that short limbs reduce the metabolic cost of maintain-ing body temperature, while long limbs result in greater heatdissipation. As mentioned previously, the collection periodsof VO2 values differ between investigations, with some collec-tions continuing for several hours. Extending our collectiontime may, in fact, help to clarify the relationship betweenRMR and lower limb length. Regardless, it should be notedthat despite the shorter, yet physiological relevant, collectionof twelve minutes for resting metabolic rates, we still reporta statistically strong relationship between RMR and limblength, thus suggesting that the relationship can be seen withas little as twelve minutes of collection.

The present results do not reveal any thermal advantage inchanging the length of the shank segment in particular. Whenthe effects of lean mass are removed, shank length has no cor-relation at all with RMR, whereas thigh length retains a high

Table 2Statistical analysis between body dimensions and resting metabolic rate1

Resting metabolicrate (RMR)

Mass(kg)

Height(cm)

Limb length(cm)

Femur length(cm)

Tibia length(cm)

Lean mass(kg)

Mass (kg) 0.643**Height (cm) 0.815** 0.820**Limb length (cm) 0.794** 0.736** 0.954**Thigh length (cm) 0.815** 0.684** 0.904** 0.931**Shank length (cm) 0.697** 0.706** 0.899** 0.955** 0.782**Lean mass 0.728** 0.902** 0.885** 0.783** 0.698** 0.775**Percent fat mass $0.532* $0.408 $0.668** $0.593** $0.470* $0.632** $0.755**

1 Pearson correlations as well as significance levels between body dimensions and resting metabolic rate. Mass (p-value! 0.002) as well as lean mass (p-value< 0.001) are highly correlated with RMR.* Correlation is significant at the 0.05 level (2-tailed).

** Correlation is significant at the 0.01 level (2-tailed).

Fig. 1. Average total lower limb length exhibits a statistically significant rela-tionship with resting metabolic rate (p< 0.001 and R2! 0.749).

Table 3Correlations for selected variables1

RMR Limb length(cm)

Femur length(cm)

Lower limb length (cm) 0.524*Thigh length (cm) 0.626** 0.871**Shank length (cm) 0.313 0.891** 0.554*

1 Partial correlation matrix for selected variables when variation due to leanmass and percent fat are accounted for.* Correlation is significant at the 0.05 level.

** Correlation is significant at the 0.01 level.

289M.J. Tilkens et al. / Journal of Human Evolution 53 (2007) 286e291

relationship (R! 0.625), higher even than total lower limblength. While it remains possible this extreme result is a func-tion of the variation between external and skeletal measures(Porter, 1996), the results do seem to suggest that the particu-lar shortening of the distal limb segments in Neandertals maynot be a consequence of cold thermal considerations.

The shorter overall limbs of Neandertals, despite being dis-advantageous while walking (Weaver and Steudel-Numbers,2005), would indeed have been advantageous for thermoregu-lation. In contrast, the early Upper Paleolithic modern humansuccessors of the Neandertals in Europe retained body propor-tions that would have been disadvantageous for a substantialperiod of time.

Acknowledgements

We would like to thank Randall Clark for conducting theDEXA scans as well as the L.S.B. Leakey Foundation forfinancial support.

References

Allen, J.A., 1877. The influence of physical conditions in the genesis ofspecies. Rad. Rev. 1, 108e140.

Auger, F., Jamison, P.L., Balslev-Jorgensen, J., Lewin, T., DePena, J.F.,Skrobak-Kaczynski, J., 1980. Anthropometry of circumpolar populations.In: Milan, F.A. (Ed.), The Human Biology of Circumpolar Populations.Cambridge University Press, Cambridge, pp. 213e255.

Badoux, D.M., 1965. Some notes on the functional anatomy of Macropusgiganteus Zimm. with general remarks on the mechanics of bipedalleaping. Acta Anat. 62, 418e433.

Beals, K.L., Smith, C.L., Dodd, S.M., 1984. Brain size, cranial morphology,climate, and time machines. Curr. Anthropol. 25, 301e330.

Cook, R.D., Weisberg, S., 1999. Applied Regression including Computing andGraph. John Wiley & Sons, New York.

Coon, C.S., 1962. The Origin of Race. Knopf, New York.

Epstein, Y., Shapiro, Y., Brill, S., 1983. Role of surface area-to-mass ratio andwork efficiency in heat tolerance. J. Appl. Physiol. 54, 831e836.

Franciscus, R.G., Trinkaus, E., 1988. Nasal morphology and the emergence ofHomo erectus. Am. J. Phys. Anthropol. 75, 517e527.

Frayer, D.W., Wolpoff, M.H., Throne, A.G., Smith, F.H., Pope, G.G., 1993.Theories of modern human origins: the paleontological test. Am.Anthropol. 95, 14e50.

Hanna, J.M., Little, M.A., Austin, D.M., 1989. Climatic physiology. In:Little, M.A., Haas, J.D. (Eds.), Human Population Biology: A Transdisci-plinary Science. Oxford University Press, New York, pp. 132e151.

Heglund, N.C., Fedak, M.A., Taylor, C.R., Cavagna, G.A., 1982. Energeticsand mechanics of terrestrial locomotion IV. Total mechanical energychanges as a function of speed and body size in birds and mammals.J. Exp. Biol. 97, 57e66.

Holliday, T.W., 1997a. Body proportions in Late Pleistocene Europe andmodern human origins. J. Hum. Evol. 32, 423e447.

Holliday, T.W., 1997b. Postcranial evidence of cold adaptation in EuropeanNeandertals. Am. J. Phys. Anthropol. 104, 245e258.

Holliday, T.W., 1999. Brachial and crural indices of European Late UpperPaleolithic and Mesolithic humans. J. Hum. Evol. 36, 549e566.

Holliday, T.W., Trinkaus, E., 1991. Limb/trunk proportions in Neandertalsand early anatomically modern humans. Am. J. Phys. Anthropol. 12,93e94.

Hrdlicka, A., 1930. Anthropological Survey in Alaska. 46th Annual Report ofthe Bureau of American Ethnology, Washington, pp. 19e374.

Irving, L., 1957. The usefulness of Scholander’s views on adaptive insulationof animals. Evolution 11, 257e259.

Kohrt, W.M., 1995. Preliminary evidence by DXA: tried and true? Med. Sci.Sports Exerc. 27, 1349e1353.

Kollias, J., Bartlett, L., Bergsteinova, V., Skimmer, J., Buskirk, E.,Nicholas, W., 1974. Metabolic and thermal responses of women duringcooling in water. J. Appl. Physiol. 36, 577e580.

Kormondy, E.J., Brown, D.E., 1998. Fundamentals of Human Ecology.Prentice Hall, Upper Saddle River, NJ.

Mayr, E., 1956. Geographical character gradients and climatic adaptation.Evolution 10, 105e108.

McArdle, W.D., Magel, J., Gergley, T., Spina, R., Toner, M., 1984. Thermaladjustment to cold water exposure in resting men and women. J. Appl.Physiol. 56, 1565e1571.

McArdle, W.D., Katch, F.I., Katch, V.L., 2001. Exercise Physiology: Energy,Nutrition, and Human Performance, fifth ed. Lippiincott Williams &Wilkins, Baltimore, MD.

Newman, M.T., 1953. The application of ecological rules to the racial anthro-pology of the aboriginal new world. Am. Anthropol. 55, 311e327.

Porter, A.M.W., 1996. Physique and the skeleton. Unpublished Ph.D Disserta-tion, University of London.

Prior, B.M., Cureton, K.J., Modlesky, C.M., Evans, E.M., Sloinger, M.A.,Saunders, M., Lewis, R.D., 1997. In vivo validation of whole body compo-sition estimates from dual-energy X-ray absorptiometry. J. Appl. Physiol.83, 623e630.

Roberts, D.F., 1978. Climate and Human Variability, second ed. Cummings,Menlo Park, CA.

Robinson, S., 1942. The effect of body size upon energy exchange in work.Am. J. Physiol. 136, 363e368.

Ruff, C.B., 1991. Climate and body shape in hominid evolution. J. Hum. Evol.21, 81e105.

Ruff, C.B., 1993. Climatic adaptation and hominid evolution: the thermoreg-ulatory imperative. Evol. Anthropol. 2, 53e60.

Ruff, C.B., 1994. Morphological adaptation to climate in modern and fossilhominids. Yearb. Phys. Anthropol. 37, 65e107.

Scholander, P.F., 1955. Evolution of climatic adaptation in homeotherms.Evolution 9, 15e26.

Schreider, E., 1950. Geographical distribution of the body-weight/body-surface ratio. Nature 165, 286.

Schreider, E., 1964. Ecological rules, body-heat regulation, and human evolu-tion. Evolution 18, 1e9.

Schreider, E., 1975. Morphological variations and climatic differences.J. Hum. Evol. 4, 529e539.

Fig. 2. Residuals of the regression of lower limb length on lean mass exhibita statistically significant relationship with the residuals of resting metabolicrate on lean mass (p! 0.018 and R2! 0.274).

290 M.J. Tilkens et al. / Journal of Human Evolution 53 (2007) 286e291

Sloan, R., Keatinge, W., 1973. Cooling rates of young people swimming incold water. J. Appl. Physiol. 35, 371e375.

Steegman Jr., A.T., 1975. Human adaptation to cold. In: Damon, A. (Ed.), Phys-iological Anthropology. Oxford University Press, New York, pp. 130e166.

Stocks, J.M., Taylor, N.A.S., Tipton, M.J., Greenleaf, J.E., 2004. Human physio-logical responses to cold exposure. Aviat. Space Environ.Med. 75, 444e457.

Steudel-Numbers, K.L., Tilkens, M.J., 2004. The effect of lower limb lengthon the energetic cost of locomotion: implications for fossil hominins.J. Hum. Evol. 47, 95e109.

Taylor, C.R.S., Shkolik, A., Dmi’el, R., Baharav, D., Borut, A., 1974. Runningin cheetahs, gazelles, and goats: energy cost and limb configuration. Am. J.Physiol. 227, 848e850.

Toner, M.M., McArdle, W.D., 1988. Physiological adjustments of man to thecold. In: Pandolf, K.B., Sawka, M.N., Gonzalez, R.R. (Eds.), Human

Performance Physiology and Environmental Medicine at TerrestrialExtremes. Cooper Publishing Group, Carmel, IN, pp. 361e399.

Trinkaus, E., 1981. Neanderthal limb proportions and cold adaptation. In:Stringer, C.B. (Ed.), Aspects of Human Evolution. Taylor and Francis,London, pp. 187e224.

Weaver, T., Steudel-Numbers, K., 2005. Does climate or mobility explain thedifferences in body proportions between Neandertals and their UpperPaleolithic successors? Evol. Anthropol. 14, 218e223.

Wolpoff, M.H., 1989. The place of the Neandertals in human evolution. In:Trinkhaus, K. (Ed.), The Emergence of Modern Humans. CambridgeUniversity Press, Cambridge, pp. 97e141.

Wyndham, C., Strydom, N., van Rensburg, A., Benade, A., Heyns, A., 1970.Relation between VO2max and body temperature in hot humid airconditions. J. Appl. Physiol. 29, 45e50.

291M.J. Tilkens et al. / Journal of Human Evolution 53 (2007) 286e291