SELF SELECTION OF ATHLETES INTO SPORTS VIA SKELETAL RATIOS

Abstract

This study examined differences in physique and skeletal dimensions between 1,146 elite Caucasian athletes classified into endurance, strength, speed, upper body or combined sports categories and controls, to test the hypothesis that athletes of certain types of sport would exhibit a distinct morphology commensurate with their task. Participants were measured by experienced anthropometrists using an internationally-recognised protocol to describe physique. Significant inter-group differences in indices prevailed after adjustment for age. Endurance athletes had a high crural index and low brachial index, relative to controls. A low skelic index appears characteristic of female strength athletes while a low brachial index is typical of female endurance and strength athletes. Bayesian cluster analysis has shown the crural index in particular is a discriminant in phenotypic categorisation, in addition to the primary somatotype components of endomorphy, mesomorphy and ectomorphy. These findings are congruent with biomechanical imperatives to maximise force and/or minimise energy expenditure offering sports-specific advantage.

Headnote

ABSTRACT

This study examined differences in physique and skeletal dimensions between 1146 elite Caucasian athletes classified into endurance, strength, speed, upper body or combined sports categories and controls, to test the hypothesis that athletes of certain types of sport would exhibit a distinct morphology commensurate with their task. Participants were measured by experienced anthropometrists using an internationally-recognised protocol to describe physique. This involved calculating the anthropometric somatotype in terms of fatness, musculoskeletal robustness and linearity. Further skeletal measures were acquired sufficient to describe an array of morphological indices: crural index (tibial height / femur length), skelic index (leg length / sitting height), androgyny index (biacromial breadth / bicristal breadth) and brachial index (forearm length / upper arm length). Significant inter-group differences in indices prevailed after adjustment for age. Endurance athletes had a high crural index and low brachial index, relative to controls. A low skelic index appears characteristic of female strength athletes while a low brachial index is typical of female endurance and strength athletes. Bayesian cluster analysis has shown the crural index in particular is a discriminant in phenotypic categorisation, in addition to the primary somatotype components of endomorphy, mesomorphy and ectomorphy. These findings are congruent with biomechanical imperatives to maximise force and/or minimise energy expenditure offering sports-specific advantage. Because the skeletal relationships do not respond to the conditioning stimulus in the same way adipose and muscle tissue do, the observed inter-group differences suggest a self-selection of athletes into sports in which they are likely to excel.

Keywords: Caucasian, athlete, skeletal proportions

Drivers for physique specialisation in humans and our precursors have hitherto included factors which are thermoregulatory (where body surface area and proportions appear to alter with climatic adaptation), or biomechanical (where certain proportions may favour agility in difficult terrain), load carrying, speed of locomotion or protection against injury (Bramble and Lieberman, 2004; Porter, 1999; Trinkaus, 1986). Commensurate with subsequent advances in technology and medicine, in the Western world, is the erosion of the link between morphology and survival. In such an urbanised environment, a persuasive argument can be marshalled to suggest that, in the 21st century, the evolutionary forces that shaped human morphology have largely disappeared. While these forces of natural selection may be redundant, elite athletes display morphological adaptations via self-selection by choosing sports to which they are anatomically suited and conditioning their bodies appropriately with increasingly sophisticated and specific training methods.

Although world-class athletes' morphology has attracted scientific enquiry since the 1928 Amsterdam Olympics, it was 32 years later at the Rome Olympics before the first systematic morphological study of track and field athletes was widely publicised (Tanner, 1964). This drew attention to physique differences between different ethnic groups, and between athletes from different sports. Discriminant analysis showed differences between groups in terms of proportions and somatotype, rated photoscopically for endomorphy (roundness or fatness), mesomorphy (musculo-skeletal robustness) and ectomorphy (linearity). Somatotype profiles have formed a rich literature, and large diverse data sets have been comprehensively catalogued, providing a range of evidence for physique variation and specialisation (Carter and Heath, 1990). Notwithstanding this, more recently, ethnicity has proved less robust as a criterion for differentiation of physique, due to hybridisation of populations.

Elite athletes' skeletal dimensions have also proved helpful in predicting body mass in hominid skeletal remains, because their extreme fitness more closely approximates the physical conditioning of earlier humans (Ruff, 2000a). Ruff's approach pooled data from different ethnic groups and postulated a physique which optimised power and speed as being most adaptive, resembling those of athletes from sporting events such as decathlon in men and pentathlon in women.

Physique specialisation is one of several factors determining performance in competitive sport, where the most successful athletes could be expected to acquire morphology commensurate with their performance task (Carter, 1985). The rewards of competitive sport over the last century or so - in terms of personal acclaim, financial inducement and national prestige - have provided sufficient incentive to drive what has been described as a ?morphological arms race' (Norton and Olds, 2001). In some sports, the rate of increase in size of participants is 2-6 times greater than those of the populations from which they are drawn (Olds, 2001). More is worthy of consideration, however, than size itself. Body tissues can be theoretically grouped as ?active' or ?ballast' whether they contribute to or retard force or power production respectively, while relative strength has been shown to scale to mass raised to the power 0.69, if composition is assumed to be constant (Carter, 1985). While soft tissue can be readily modified by adaptation to appropriate training stimuli, the same is not true of skeletal proportions; these are largely genetically determined and remain largely constant in adults. Although body shape variation is understood to show continuous variation, skeletal proportions appear to exhibit a ?race specificity' (Feldesman and Fountain, 1996; Holliday and Ruff, 2001; Ruff, 2000b) when groups from different geographical areas are compared. For instance, this has been graphically depicted in terms of longer limbs in Afro-Caribbean compared to Caucasian athletes in the same track and field events (Tanner, 1964). The rise to global dominance by East African endurance runners has been partly explained by preferential physique in terms of optimised force production and movement. Differing leg morphology and inertial properties, between Senegalese and Italian sprinters, has also been reported (Rahmani, Locatelli and Lacour, 2004). The longer, lighter legs of the Senegalese incurred a lower level of inertial work. By comparison, the Italians needed to generate more force in order to run at the same speed. This approach was taken further by the calculation of ground support forces by the world's fastest runners across eight distances (Weyand and Davis, 2005). They noted that early studies had overestimated the mechanical power requirement for lifting and accelerating the body and limbs during running. However, the larger forces during faster running combined with the constant force-area relationship of skeletal muscle demand more muscle to generate and transmit such forces to the ground. Weyand and Davis argued that for those of the same stature and composition, more tissue may be biologically necessary for attaining faster speeds.

While morphology may explain performance differences between groups, to date, study samples have either focused on specific ethnic group comparisons or involved experimentation of biomechanical parameters with small samples. Given the wealth of data available on elite athletes, it is surprising that relatively little is known of sports involving more complex movements. A further difficulty is introduced by the extreme specialisation of certain sports which focus on certain aspects of physique. This requires a categorisation system which can be generated to group similar influences together, so that observations can be made with larger samples. Therefore, this study aimed to compare skeletal proportions of elite athletes from a range of different sports, while focusing on a large and relatively-homogeneous sample. Specifically, it was hypothesised that a) proportional differences would be observed between different sporting categories and that b) these differences would be in addition to those of sexual dimorphism. If found, such differences might give some indication of criteria for selectivity, representing a competitive advantage for certain sports.

METHODS

Participants (829 male and 317 female adult competitive athletes) were assessed for mass, skinfolds, girths, lengths and breadths by standardised procedures in a 39-measurement protocol (Marfell-Jones, Olds, Stewart and Carter, 2006) by experienced criterion anthropometrists and anthropometric somatotype calculated according to the method of Carter and Health (1990). Local ethical permission was obtained from participating universities and informed consent was obtained from all subjects. Athletes from the United Kingdom, Australia and New Zealand were included if they were of national or international standard. Due to the predominant ethnicities of the three host countries, the sample was almost exclusively Caucasian.

Sporting categories were developed in order to produce meaningful group sizes for comparative purposes. Specific criteria for conditioning which would be likely to influence physique were developed from primary movement imperatives and assigned according to a decision chart summarised in Figure 1.

Endurance (n=235) - where locomotor efficiency emphasising the major postural muscles is dominant - included running, cycling (time trial and road), orienteering, cross country skiing, and rowing. Speed (n=338) - where sprinting and/or very rapid body segment movement is involved - included sprint running (1500m and less), hurdling, javelin, diving, rugby (backs), touch rugby, badminton, track cycling, Australian rules football and martial arts. Combined (n=231) - where basic locomotion was accompanied by additional speed or strength requirements - included gymnastics (female) dance, softball, handball, fencing, hockey, soccer, volleyball, tennis. Strength (n=132) - where absolute strength and / or power requirement is more important than relative strength, relative power or locomotion - included bodybuilding, rugby forwards, powerlifting, gymnastics (male), wrestling, shot put and hammer throwing. Upper Body (n=102) - where the sport requirement favours the development of upper body muscles and/or where the body weight is supported so the postural muscles are de-emphasised - included kayaking, canoeing, dragon boat racing, rock climbing, swimming. Control (n=108) - subjects who were healthy, and possibly physically active, but not competitive in sport. Athletes whose sport required no clear specific physique specialisation such as cricket, golf, yachting and shooting were excluded. Categorical assignation was agreed between members of the research team using these criteria prior to analysis.

Segment length measurements which summarised the major relationships of the skeletal development were made on the right side of the body using a segmometer or wide spreading calliper (Rosscraft Innovations Inc., Canada) to 0.1 cm. Stature and sitting height were measured to 0.1 cm with the head in the Frankfort plane and following inspiration. Data were converted into ratios defined as follows: The crural index is a measure of tibia length in relation to femur length, defined as the height of the tibiale laterale (a) with the subject standing, divided by the trochanterion - tibiale laterale length (b). The brachial index is a measure of the length of the forearm in relation to the upper arm and is defined as the radiale - stylion length (c) divided by the acromiale - radiale length (d) . The androgyny index is essentially shoulder breadth in relation to pelvic breadth, defined as the biacromial breadth (e) divided by the bicristal breadth (f). The skelic index describes leg length in relation to torso length and is defined as the derived total leg length (stature minus sitting height) divided by sitting height. These ratios are depicted in Figure 2.

All landmarks and definitions were measured using the protocol of the International Society for the Advancement of Kinanthropometry (Marfell-Jones, Olds, Stewart and Carter, 2006) using anatomically-defined landmarks and systematic procedures, together with exam-based quality assurance of precision and accuracy. Technicians assembling the current data had intra-tester technical error of measurement below 1% and averaging 0.4% for skeletal data. The widespread use of this protocol has enabled data pooling which in turn allowed comparison of large groups of specialist competitive athletes.

Skeletal indices were entered into a general linear model using SPSS Version 14 (Chicago, IL) with gender and sport category as fixed factors and identifying any interaction. Age was entered as a covariate because controls were significantly older than sporting groups and the effect of a secular trend in height could mean potential differences arising from the effect of stature. As a further check on patterns of physique, all anthropometric somatotype and skeletal data were entered into unsupervised cluster analysis. Bayesian decision theory provided a maximum posterior probability classification of individuals that was maximally probable with respect to the data and model (Cheeseman and Stutz, 1996; Wallace and Dowe, 1999). Sport category and sex were not entered as factors. The resulting class membership thus determined whether the skeletal ratios contributed to phenotypic variation beyond that associated with the three somatotype categories of endomorphy, mesomorphy and ectomorphy.

RESULTS

There were significant differences in skeletal ratios between sporting categories and by sex as shown in Figures 3 to 6. For the crural index, 9.1% variation was explained by sport category, age and gender. Age was negatively associated with crural index, indicating that, in younger athletes, crural index was greater (P<0.001). For the androgyny index, 17.3% of the variation was explained by sport category and gender (P<0.001), but age was not significant (P=0.83). In all sports categories, males had greater values. For the skelic index, 4% of variation was explained by sport category (P<0.01) and gender (P<0.001) but not age (P = 0.45). For the brachial index, 5.3% of variation was explained by age (P<0.05), sport and gender (P<0.001). Interaction between sex and sport (all P < 0.05 except brachial index, P = 0.08) suggested that the observed sporting differences did not follow the same pattern for males and female athletic groups.

Of the skeletal ratios described, only the crural index was selected in the Bayesian cluster analysis as a discriminant of the larger sample, in addition to the established somatotype categories of endomorphy, mesomorphy and ectomorphy. Significant class attributes are depicted in Figure 7 and Table 1. (Note that the crural index here is not age-adjusted).

DISCUSSION

All four ratios remained significantly different after adjusting for age, suggesting that sporting prowess benefits from certain skeletal characteristics that are proportional. Taken together, these show signs of a preferred -morphotype" for the sporting c ategories described. The phenomenon of differential skeletal ratios has not been reported hitherto in a large athletic sample. It is reasonable to assume that athletes will select to compete in sports for which they are anatomically suited. In doing so, they may excel by fulfilling a biomechanical imperative of energy efficiency or power production, via considerations of segmental mass and lever arms, which have different determinants in different sports. For instance, while running at a given speed, an advantage may be offered by a higher crural index which concentrates the mass more proximally on the moving limb thus decreasing its resistance arm. The low crural index of Neanderthals compared with modern humans has been attributed to the imperatives of locomotion and load carrying over difficult terrain, safeguarding knee ligaments against injury (Porter, 1999). In a similar fashion, male and female strength athletes appear to combine similar attributes of relatively short and distally-abbreviated legs with wide shoulder girdles which may favour force production and load bearing at the expense of speed.

Consideration of the possible contribution of endurance running to human evolution has identified leg morphology among many factors affecting locomotion where greater leg length increases sustainable speed and contact time, reducing energy expenditure over a distance as a result of lower cadence and greater stride length (Bramble and Lieberman, 2004). This is supported by comparison of different animal species, where effective limb length explained 98% of the energy cost of locomotion (Pontzer, 2007). However our data suggest the effect may not be due to leg length per se, but also the relative position of the knee joint and the concomitant alteration in both the segmental (thigh and shank) and overall lower limb Moment of Inertia (MI). In the present study, endurance athletes had similar leg proportions to controls (quantified by the skelic index). However, their crural index was greater (P<0.001). This has the effect of moving the knee joint proximally in endurance athletes relative to controls. As the greater balance of the mass of the lower limb lies in the thigh segment, this change may explain the greater speed and efficiency of running achieved through greater segmental angular acceleration. Greater angular acceleration is facilitated by reduced lower limb MI without compromising proximal muscle power (torque), as given by the angular motion version of Newton's second law T = I.α, where T is torque in Nm, MI is moment of inertia in kg.m -2, and α is angular acceleration in rad.s-2. [MI is dependent on the square of the distance between the centre of mass of the rotating limb segment and the axis of rotation]. As the main bulk of the primary muscles of running (hamstrings and quadriceps) and hence lower limb mass, lies proximal to the knee, the effect of any increase in crural index is mechanically profound in the lower limb. The effect of altering mass distribution in the lower limb and hence segmental I, has been previously observed by runners strapping 3.6 kg of additional weight to the waist, upper thigh, shank or ankle (Myers and Steudel, 1985). Moving this additional weight from the ankle to the hip reduced MI and also reduced running energy expenditure by 15%. However, a more recent experimental study of kinematics and moments of inertia of 27 healthy adults found no effect of the crural index on moments during walking, suggesting that postural adjustments might moderate locomotor forces (Gruss, 2007). Running, especially for the maximal performance sought by the athletes of the present study may appear to introduce different criteria which become significant once the forces and velocity become great enough.

The predicted muscle mass exponent (the power to which the muscle circumference scales to mass) from thigh measurements has been shown to be 0.96 in endurance and 1.49 in speed athletes (Nevill, Stewart, Olds and Holder, 2004), suggesting endurance athletes rely less on muscular and more on gravitational force to accelerate their limbs than speed athletes. The present study provides skeletal data suggesting a further energy saving by a more proximal location of their knee joint. Using group mean limb circumferences and lengths with a fixed-density cylindrical model, endurance athletes effectively have the centre of mass of their extended limb moved proximally by about 1.7 cm, enabling a lighter distal segment to be accelerated more rapidly with less energy expenditure. The significance of age in the model suggests that more recent generations of athletes appear to have longer distal lower limb segments, possibly as a consequence of alterations in environmental influences such as nutrition. Paradoxically, the effect of age on the brachial index is positive, suggesting younger athletes have distally abbreviated upper limb segments. Endurance athletes also showed distal abbreviation relative to controls, effectively moving the centre of mass of the extended arm 0.4 cm distally. The reason for this is unclear. A counterbalance effect of opposing pendulum movement of arms and legs may confer a small energy saving, although style variation, via differential arm flexion and torso rotation, could negate this.

It has been suggested that the androgyny index could discriminate successful athletes in bodybuilding for aesthetic reasons and a capacity to increase muscle mass on the torso (Fry, Ryan, Schwab, Powell and Kraemer, 1991). Beyond this, a broad shoulder girdle may enable greater leverage or reach in some upper-body sports such as rock climbing or canoeing. Conversely, competition rules of the International Powerlifting Federation may discriminate against larger athletes whose grip may be required to be narrower than the optimal functional position (Gilbert and Lees, 2003). In addition, a smaller brachial index might be an advantage in strength sports requiring full elbow extension by reducing the work required, as demonstrated in the female, but not male, strength athletes.

Previously, we described mass exponents for athlete data sets for girths (Nevill, Stewart, Olds and Holder, 2004) and skinfolds (Nevill, Stewart, Olds and Holder, 2006). Such surrogates of adipose and muscle tissue respond directly to the stressors of specialist training and high energy expenditure. By contrast, adult skeletal size and proportions are relatively unaffected by such training, but have considerable effects on the biomechanics of movement and work. Since an individual is not at liberty to alter these dimensions, (unlike those relating to soft tissue masses and proportions in an adult), selective processes are driven by them in elite sport. The data specifically illustrate that a high or low crural index is responsible for distinct phenotypes. High values are typically represented in male and female endurance, female speed and male combined sports physiques, while low values are typical of controls and female combined sports athletes. While the skeletal ratios show highly significant differences between groups (figures 2 - 9), the degree of correspondence between crural index in phenotype (figure 10 and table 1) and sporting categorisation is modest. There are two explanations for this. One is that skeletal proportion limitation can, to an extent, be negated by other factors in the sports performance such as physiological fitness, psychological attributes and skill. The other is that our principle-based categorisation of sports into groups inevitably has limitations. If our criteria for sports categorisation were to be changed, the findings would likely alter. Nevertheless, the differences found here could be logically explained by our categorisation. We further concede modelling the limbs on the basis of cylindrical or conical models of an isotropic body tissue to estimate the effect on centre of mass may be too simplistic for suggesting biomechanical models, since adaptive remodelling of bone and adaptations of connective tissue alter tissue density non-uniformly. However, mathematical modelling of athletic performance combining actual athlete dimensions was beyond the scope of the present work.

This notwithstanding, our findings which effectively locate relative joint positions differently on a generic skeletal template for different sporting categories may well have possible explanations in biomechanical theory. This suggests a reduction in energetic cost of transport (i.e. per unit distance) with increasing leg length in other animals, but available data are equivocal (Steudel, 1996). Estimating muscle masses required to maintain equilibrium in selected extant taxa (Hutchinson, 2004) found none with extensor muscles exceeding 7% body mass acting across any joint, suggesting a limit for how muscle is apportioned within a limb. While the present study did not address this, we would surmise that the relative development of primary force-producing muscles in human athletes is more variable, as a possible means of exacerbating or compensating for the effect of skeletal proportions.

The fact that our sporting categories, relative to controls show different effects on skeletal proportions by gender, as well as gender interactions for sport (when controls are treated as a sports category) is of specific interest. If the hypothesis that an exercise stimulus is identical for males and females would result in a broadly similar driver for selectivity, and broadly similar differences by sport from controls, then our observations require a different explanation. Some findings are undoubtedly the result of different exercise adaptation arising from different activities (e.g. rhythmic gymnastics only in females), or differences in the relative representation within a sporting category (e.g. rugby being more prevalent in males than females) or alternatively differences in the extent of the selective process necessary for international representation. However, the data set is sufficiently large to be able to suggest, despite the shortcomings of differing gender prevalence in some sports, there appears to be a strong influence of selectivity based on the skeletal proportions described here.

CONCLUSION

Accepting the limitations of categorisation, our data suggest two conclusions. Firstly, there appears strong evidence of self-selection of athletes into sports to which they are anatomically suited because the skeletal proportions vary significantly between groups. Secondly, the crural index in particular is influential in phenotypic variation within an athletic sample. While the genotypic variation of the sample was limited to predominantly Caucasian athletes, a future study may consider athletes of any ethnic group in order to acquire an even larger sample. Future biomechanical research may align these findings with lab-derived movement data and virtual sports performances. Given the value placed on sporting success, understanding the nature of any competitive advantage offered by morphology will become an important focus for future research effort. Considering the fact that it requires perhaps 10 years to produce athletes of Olympic calibre, a further priority will be to recognise potentially successful adult morphology from the skeleton of the growing athlete.

References

REFERENCES

Ackland, TR; Schreiner, AB; Kerr, DA. Absolute size and proportionality characteristics of World Championship female basketball players. Journal of Sports Sciences, 1997. 15, 485-490

Alexander, RM. Optimum take-off techniques for high and long jumps. Philosophical Transactions of the Royal Society of London Biological Sciences, 1990. 329, 3-10.

Bramble, DM; Lieberman, DE. Endurance running and the evolution of Homo. Nature, 2004. 432, 345-352.

Carter, JEL. Morphological factors limiting human performance. In: Clarke DH, Eckert HM, editors. Limits of Human Performance. American Academy of Physical Education Papers No 18, Champaign: Human Kinetics; 1985; p106-117.

Carter, JEL, Heath BH. Somatotyping: Development and applications. Cambridge UK: Cambridge University Press; 1990.

Cheeseman, P; Stutz, J. In: Fayyad UM, Piatetsky-Shapiro G, Smyth P, Uthurusamy R, editors. Advances in Knowledge Discovery and Data Mining Cambridge. MA: AAAI Press/MIT Press; 1996; p61-83.

Davenport, CB. The crural index. American Journal of Physical Anthropology, 1933. 17, 333-353.

Feldesman, MR; Fountain, RL. -Race" specificity and the femur/stature ratio. American Journal of Physical Anthropology, 1996. 100, 207-224.

Fry, AC; Ryan, AJ; Schwab, RJ; Powell, DR; Kraemer, WJ. Anthropometric characteristics of body-building success. Journal of Sports Sciences, 1991. 9, 23-32.

Gilbert, G; Lees, A. Maximum grip width regulations in powerlifting discriminate against larger athletes. In: Reilly T, Marfell-Jones M, editors. Kinanthropometry VIII London: Routeledge; 2003; p175- 180.

Gruss, LT. Limb length and locomotor biomechanics in the genus homo: An experimental study. American Journal of Physical Anthropology, 2007. 134, 106-116.

Holliday, TW; Ruff, CB. Relative variation in human proximal and distal limb segment lengths. American Journal of Physical Anthropology, 2001. 116, 26-33.

Hutchinson, JR. Biomechanical modelling and sensitivity analysis of bipedal running ability. Extant Taxa. Journal of Morphology, 2004. 262, 421-440.

Marfell-Jones, M; Olds, T; Stewart, A; Carter, JEL. International Standards for Anthropometric Assessment. Potchefstroom, South Africa. International Society for the Advancement of Kinanthropometry; 2006.

Myers, MJ; Steudel, K. Effect of limb mass on the energetic cost of running. Journal of Experimental Biology, 1985. 116, 363-373.

Nevill, AM; Stewart, AD; Olds, T; Holder, R. Are adult physiques geometrically similar?: the dangers of allometric scaling using body mass power laws. American Journal of Physical Anthropology, 2004. 124, 177-182.

Nevill, AM; Stewart, AD; Olds, T; Holder, R. Modelling the relationship between adiposity and body size reveals the limitation of BMI. American Journal of Physical Anthropology, 2006. 129, 151-156.

Norton, K; Olds, T. Morphological evolution of athletes over the 20th century: causes and consequences. Sports Medicine, 2001. 31, 763-783.

Olds, T. The evolution of physique in rugby union players in the twentieth century. Journal of Sports Sciences, 2001. 19, 253-262.

Pontzer, H. Effective limb length and the scaling of locomotor cost in terrestrial animals. Journal of Experimental Biology, 2007. 210, 1752-1761.

Porter, AMW. Modern human, early modern human and Neanderthal limb proportions. International Journal of Osteoarchaeology, 1999. 9, 54-67.

Rahmani, A; Locatelli, E; Lacour, J. Differences in morphology and force/velocity relationship between Senegalese and Italian sprinters. European Journal of Applied Physiology, 2004. 91, 399-405.

Ruff, CB. Body mass prediction from skeletal frame size in elite athletes. American Journal of Physical Anthropology, 2000a. 113, 507-517.

Ruff, CB. Body size, body shape, and long bone strength in modern humans. Journal of Human Evolution, 2000b. 38, 269-290.

Saltin, B; Larsen, H; Terrados, N; Bangsbo, J; Bak, T; Kim, CK; Svedenhag, J; Rolf, CJ. Aerobic exercise capacity at sea level and at altitude in Kenyan boys, junior and senior runners compared with Scandinavian runners. Scandinavian Journal of Medicine and Science in Sports, 1995. 5, 209-221.

Steudel, K. Limb morphology, bipedal gait, and the energetics of hominid locomotion. American Journal of Physical Anthropology, 1996. 99, 345-355.

Tanner, JM. The physique of the Olympic athlete. London: George Allen & Unwin; 1964.

Trinkaus, E. The Neandertals and modern human origins. Annual Review of Anthropology, 1986. 15, 193-218.

Wallace, CS; Dowe, DL. Minimum message length and Kolmogorov complexity. The Computer Journal, 1999. 42, 270-283.

Weyand, PG; Davis, JA. Running performance has a structural basis. Journal of Experimental Biology, 2005. 208, 2625-2631.