You are on page 1of 15

Received: 6 November 2022 | Revised: 15 March 2023 | Accepted: 29 May 2023

DOI: 10.1002/evan.21988

REVIEW ARTICLE

The estimation and evolution of hominin body mass

Christopher B. Ruff1 | Bernard A. Wood2

1
Center for Functional Anatomy and
Evolution, Johns Hopkins University School of Abstract
Medicine, Baltimore, Maryland, USA
Body mass is a critical variable in many hominin evolutionary studies, with implications
2
Center for the Advanced Study of Human
Paleobiology, George Washington University,
for reconstructing relative brain size, diet, locomotion, subsistence strategy, and social
Washington, District of Columbia, USA organization. We review methods that have been proposed for estimating body mass
from true and trace fossils, consider their applicability in different contexts, and the
Correspondence
Christopher B. Ruff, Center for Functional appropriateness of different modern reference samples. Recently developed
Anatomy and Evolution, Johns Hopkins techniques based on a wider range of modern populations hold promise for providing
University School of Medicine, 725 N. Wolfe
St., Baltimore, MD 21205, USA. more accurate estimates in earlier hominins, although uncertainties remain,
Email: cbruff@jhmi.edu particularly in non‐Homo taxa. When these methods are applied to almost 300 Late

Funding information Miocene through Late Pleistocene specimens, the resulting body mass estimates fall
Leakey Foundation; Wenner‐Gren within a 25–60 kg range for early non‐Homo taxa, increase in early Homo to about
Foundation, Grant/Award Number: 6084;
National Science Foundation, 50–90 kg, then remain constant until the Terminal Pleistocene, when they decline.
Grant/Award Numbers: SBR‐891955, SBR‐
8919749, BSC‐0642297 KEYWORDS
Australopithecus, body size, Homo, human evolution, ontogeny, scaling, skeleton

1 | INTRODUCTION techniques have been proposed to do this, almost all of them based on
skeletal dimensions, but other approaches, such as use of footprint
The importance of body mass (or weight) as a fundamental biological size,25 are also possible. Here we review and critically examine these
1–5
trait of organisms has been emphasized by many authors. Variation techniques, noting their inherent assumptions, advantages, and limita-
in body mass affects almost all other aspects of an animal's morphology tions, and provide a number of formulae for body mass estimation that
and physiology, including the mechanical strength of its bones for are applicable to human archaeological and paleontological specimens.
support and movement, neurological requirements for control of the We end by presenting a summary of temporal trends in body mass in
body, and food‐processing capabilities to supply metabolic needs. Thus, the hominin lineage, from the late Miocene through living humans.
analyses of characteristics as disparate as locomotor behavior, relative
brain size (or encephalization), and dietary adaptations must take body
mass into account. Furthermore, many critical ecological and life history 2 | HOW T O E STIMATE B ODY M ASS
variables, such as home range size, life span, reproductive rate, and so
forth are highly correlated with body mass.2,6,7 2.1 | Morphometric versus mechanical
Given its significance in so many biological and behavioral contexts,
body mass as a variable has taken on increasing importance in Methods of body size estimation from skeletal remains can be divided
8–27
paleoanthropological research over the past several decades. into two basic approaches: morphometric, in which the body is actually
Because body mass cannot be measured directly in past individuals, it reconstructed from the available material, and mechanical, in which a
must be estimated indirectly from their preserved remains. A variety of functional relationship between a particular skeletal feature and body

This is an open access article under the terms of the Creative Commons Attribution‐NonCommercial‐NoDerivs License, which permits use and distribution in any
medium, provided the original work is properly cited, the use is non‐commercial and no modifications or adaptations are made.
© 2023 The Authors. Evolutionary Anthropology published by Wiley Periodicals LLC.

Evolutionary Anthropology. 2023;1–15. wileyonlinelibrary.com/journal/evan | 1


15206505, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/evan.21988 by Johns Hopkins Medicine, Wiley Online Library on [20/07/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
2 | RUFF and WOOD

mass is used to develop a prediction equation.28 An example of the first tissue.29,30 For estimating body mass, an analogous method using
approach is the so‐called “Fully” or “anatomical” technique for estimated stature and bi‐iliac (maximum pelvic) breadth has been
estimating stature, in which the heights or lengths of skeletal elements developed31,32 (see Figure 1). These two dimensions encapsulate much
from the foot to the head are summed, along with an addition for soft of the variation in body size and shape present among living and earlier
hominins.31,33,34 When combined, stature and bi‐iliac breadth can
predict body mass with an average absolute prediction error of less than
10% in healthy, nonoverweight individuals.20 Because of differences in
body shape between males and females (shoulder to hip breadth
proportions), separate equations are used for the two sexes, or, if sex is
uncertain, an average of male and female estimates is used. A formula to
convert skeletal to living bi‐iliac breadth is available for application in
skeletal samples (see Table 1).20,21 Equations have been developed
based on a worldwide sample (see below), and have been applied in a
variety of archaeological and paleontological contexts.38–41
Another type of morphometric method for estimating body mass
that has not been widely used in human evolutionary studies is based on
so‐called “volumetric” reconstruction of the body performed by digitally
fitting a surface over an articulated skeleton.42 When this method was
applied to A.L. 288‐1 (“Lucy”), one of the most complete australopith
skeletons known, it resulted in a very low body mass estimate compared
to other methods.43 Possible factors contributing to this result included
problems with the cadaveric reference sample used to generate the
predictive equations and inaccuracies in reconstructing the full skeleton
of A.L. 288‐1. Body masses of a test sample of macaques and modern
humans were also generally underestimated, sometimes considerably.
More testing of the method on additional primate skeletons of known
body mass is needed to establish its reliability.44
Mechanical methods rely on predicted and empirically observed
mechanical relationships between skeletal dimensions and body mass.
The most reliable dimensions are those that reflect the size of structures
involved in supporting body mass (e.g., lower limb joint size). Two such
dimensions are the superoinferior breadth of the femoral head and
mediolateral breadth of the proximal articular surface of the tibia (see
Figure 1). These are relatively straightforward to measure and are
preserved in many early (as well as later) hominins.20 Regression
equations including these dimensions have been developed from a
number of modern reference samples,9,16,45–47 but the most recent and
comprehensive samples include many more individuals representing a
greater variety of body sizes and shapes,20,36 and are used here.
Because it has been shown that early, non‐Homo hominins (i.e.,
australopiths) have relatively small femoral heads,16,20 possibly due to
differences in gait biomechanics relative to modern humans,48,49 an
adjustment for this proportional difference is made first for such taxa
before applying the femoral head formula20 (see Table 1). The possibility

F I G U R E 1 Modern human skeleton showing dimensions used in


estimating body mass. Red arrows: stature and bi‐iliac breadth. Green
arrows: long bone articular breadths, including femoral head
superoinferior breadth, proximal tibial mediolateral breadth, humeral
head superoinferior breadth, and distal humeral articular mediolateral
breadth. Purple: femoral midshaft cortical area (which may be
combined with femoral maximum length, not shown). Blue: maximum
foot length and forefoot breadth, measured on footprints.
15206505, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/evan.21988 by Johns Hopkins Medicine, Wiley Online Library on [20/07/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
RUFF and WOOD | 3

TABLE 1 Body mass estimation equations for adults.

Region Predictor(s)a Sex Equationb r %SEEc %PEd Sourcee

Worldwide ST, BI Male BM = 0.422(ST) + 3.126(BI) − 92.9 0.913 6.2 4.7 (1)

Female BM = 0.504(ST) + 1.804(BI) − 72.6 0.819 7.4 5.5 (1)

Europe f
FHSI Male BM = 2.800(FHSI) − 66.7 0.838 10.1 8.0 (2)

Female BM = 2.180(FHSI) – 35.8 0.671 7.8 6.1 (2)

Combined BM = 2.295(FHSI) − 41.7 0.793 9.0 7.2 (2)

TPML Male BM = 1.838(TPML) − 70.4 0.566 11.1 9.0 (3)

Female BM = 1.406(TPML) − 37.8 0.548 9.2 7.2 (3)

Combined BM = 1.564(TPML) − 49.2 0.736 10.2 8.3 (3)

Worldwide FHSIg Combined BM = 2.262(FHSI) − 38.7 0.831 8.7 7.2 (4)

TPML Combined BM = 1.623(TPML) − 52.7 0.755 9.7 8.4 (4)

HHSIh Combined BM = 2.067(HHSI) − 28.5 0.789 10.3 7.8 (5)

HDMLi Combined BM = 2.124(HDML) − 27.6 0.760 10.5 8.4 (5)

Worldwidej FCA Combined BM = 0.0797(FCA) + 27.8 0.782 10.9 8.5 (6)

FCA, FL Combined BM = 0.0448(FCA) + 0.1136(FL) − 5.3 0.823 9.3 7.2 (6)

Worldwide k
FT Combined BM = −0.250(FT) + 0.00099(FT ) + 59.1
2
0.861 13.7 10.1 (7)

FT, BMI Combined BM = 0.155(FT) + 2.48(BMI) − 32.9 0.965 7.0 5.4 (7)
a
ST: stature (cm); BI: living bi‐iliac breadth (cm) (living bi‐iliac = skeletal bi‐iliac × 1.17–3, except when skeletal bi‐iliac <24 cm, where living = skeletal + 1.0);
FHSI: femoral head superoinferior breadth (mm); TPML: tibial plateau mediolateral breadth (mm); HHSI: humeral head superoinferior breadth (mm);
HDML: humeral distal articular mediolateral breadth (mm); FCA: femoral midshaft cortical area (mm2); FL: femoral maximum length (mm); FT: footprint
area (length × breadth) (cm2); BMI: body mass index (body mass in kg/(stature in m)2). Femoral midshaft defined relative to femoral length' dimension.99
b
BM: body mass in kg.
c
%SEE: percent standard error of estimate [(standard error of estimate of y/mean y) × 100)].
d
%PE: absolute value of percent prediction error [|(observed y – predicted y)/predicted y) × 100|].
e
1: Ruff et al.32; 2: Ruff et al.36; 3: calculated from data in Ruff et al.36; 4: Ruff et al.20; 5: Ruff et al.24; 6: Ruff et al.37; 7: Ruff et al.25 (normal walking
condition).
f
Equations applicable to anatomically modern Homo sapiens from Europe.
g
For non‐Homo specimens, adjust FHSI: FHSI = 1.252(FHSI) – 5.7.
h
Applicable to Homo and Australopithecus sediba. For archaic Middle and Late Pleistocene Homo, adjust HHSI: HHSI = HHSI × (HHAP/HHSI)/0.935 ,
where HHAP is anteroposterior breadth of humeral head.
i
Applicable to Homo and Australopithecus sediba. For Late Pleistocene archaic Homo, multiply body mass by 1.14; for Middle/Late Pleistocene anatomically
modern H. sapiens, multiply body mass by 1.08.
j
Applicable to non‐habilis (sensu stricto) Pleistocene Homo.
k
Applicable to non‐habilis (sensu stricto) Homo. A different equation, based on the proportions of the Australopithecus afarensis A.L. 288‐1 skeleton
(“Lucy”) is recommended for A. afarensis: body mass = 0.218 × FT.

that different early hominin taxa may have exhibited somewhat readily than articular surfaces.46,51–53 There is evidence that such
different scaling relationships between articular size and body mass, factors varied systematically between modern humans and earlier
due in part to variation in locomotor behavior, must also be considered, hominins,48,54,55 which could bias body mass estimates if a modern
however. An example of this as applied to upper limb bone articulations human reference sample were used to develop equations. In an attempt
is given below and in Supporting Information: Text S1. Juveniles have to minimize such biases, a method has recently been developed that
different articular‐to‐body mass proportions than adults, so different uses a Pleistocene Homo sample with body masses estimated from
50
formulae must be applied to them (also see Section 2.5). articular size to derive new equations based on femoral diaphyseal
Lower limb diaphyseal (shaft) cross‐sectional dimensions have also cross‐sectional area and bone length.37 The equations are applicable to
9,14–16,19
been used to estimate body mass, which is theoretically other Pleistocene Homo specimens, but not to non‐Homo specimens.
plausible because these dimensions are also involved in support of body Modern human upper limbs normally do not support body mass,
mass. A potential complication with this approach, however, is the yet some modern human upper limb bone dimensions are still
developmental plasticity of long bone diaphyses, which respond to correlated with body mass, albeit to a lesser degree than the lower
mechanical loadings (forces) due to activity level and muscularity more limb.24,56,57 Body mass estimation equations based on proximal and
15206505, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/evan.21988 by Johns Hopkins Medicine, Wiley Online Library on [20/07/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
4 | RUFF and WOOD

distal humeral articular breadth dimensions in a large modern human larger feet. Thus, a different technique based on the ratio of body
sample have been developed (see Figure 1).24 As might be expected, mass to foot size in the most complete A. afarensis, A.L. 288‐1
though, because of their less tight association with body mass, there (“Lucy”), is more appropriate for this taxon.25
are more uncertainties in applying these equations to earlier hominins
than with lower limb bone articular equations. In particular, the
equations cannot be applied to taxa that show evidence for different 2.2 | Which reference sample?
upper limb bone‐to‐body mass proportions. These include all
australopith taxa for which data are available (except Australopithecus All body mass estimation techniques begin with identifying an
sediba),24 and probably also Homo habilis (sensu stricto).58 For some of appropriate reference (or “training”) sample, from which the
these taxa, formulae based on chimpanzees, or a mean of chimpan- methodology to calculate body mass from available dimensions is
24
zees and modern humans, appear to be more appropriate. Here we developed. Because such dimensions are almost always skeletal (with
take a conservative approach and use a mean of modern human and the exception of footprints), this involves finding reference samples
chimpanzee‐based humeral estimates for these taxa (see Section 3 with associated data for both skeletal dimensions and body mass.
and Supporting Information: Text S1). Even in later Homo, some One option employed by a number of past researchers is to use
adjustments should be made to account for the differently shaped osteological collections derived from cadaveric material, where body
humeral head in archaic Middle and Late Pleistocene Homo, and weights were taken before dissection.8,9,11,15,16,19,45 Unfortunately, it
slightly but systematically different distal humeral to body mass is very difficult to control for factors such as wasting before death,
proportions in Late Pleistocene archaic and anatomically modern the different treatments of bodies after death, and differences in the
Homo compared to very recent humans.24 These adjustments are time elapsed between death and when the body is weighed in such
shown in Table 1 and discussed in more detail in Ruff et al.24 samples. A recent comparison indicated that in one sample true living
Morphometric approaches to body mass estimation have the body masses were probably 21%–29% higher than cadaveric body
advantage that actual body size and shape are estimated directly, masses, with large random as well as directional prediction errors
rather than inferred from functional relationships that may vary, as when using skeletal dimensions to estimate body mass.20 Another
illustrated by the complexities in applying upper limb bone mechani- option is to use skeletal dimensions derived from medical imaging
cal equations. However, they have the disadvantage that much more (radiography, computed tomography), or anthropometric measure-
of the skeleton must be preserved to apply them. As a result, only a ments (e.g., stature, bi‐iliac breadth) in living subjects with known
very few early hominin specimens have been analyzed in this way, body masses.46,47,64–67 A common issue in this case is not under-
and even among more recent archaeological samples such specimens estimation of body mass, as with cadaveric samples, but rather
are relatively uncommon (although see Ruff et al.20,36). Mechanical overestimation, due to the generally overweight condition of most
techniques are much more widely applicable, and when used with the recent Western industrialized populations.68 Thus, equations devel-
appropriate reference samples, they can provide relatively accurate oped from these samples may produce upwardly biased estimates,
estimates. unless they are restricted to nonoverweight individuals.20,47
Finally, preserved footprints in hominin trackways offer a Both cadaveric and living reference samples used to develop
completely different approach to body mass estimation. Hominin body mass estimation equations commonly suffer from another
footprint records, which are available worldwide and from time important limitation: they often incorporate only a limited range of
periods from the Pliocene to the present,59,60 provide unique variation in body size and shape. Most such samples have been
opportunities to sample groups of individuals,61–63 and potentially derived from populations in the United States or Europe, or if they
carry out demographic analyses. In upright bipedal hominins, feet are include other regions of the world, are based on only a few
the sole contact with the ground during standing, walking, and individuals or sex/population means.20 There is abundant evidence
running, and are thus critical in weight support. A recent study of 143 that both body size and body shape varied greatly among hominins
unshod or minimally shod individuals from a variety of modern over the past 6 million years.20,34,69,70 Some of this variation may be
populations demonstrated a high correlation between foot “area” attributable to climatic adaptation26,34,37,40,71,72 and some to broader
25
(maximum foot length × foot breadth) and body mass. However, the temporal/phylogenetic trends12,20,22,23,70,73 (see Section 3), but in
relationship varied by body shape, with relatively heavier populations any event it argues for inclusion of as broad a range of body sizes and
(evaluated using the body mass index, BMI = body mass in kg/(stature shapes as possible in any modern reference sample to avoid undue
in m)2) having relatively smaller feet. Thus, if body shape can be extrapolation when methods are applied to earlier hominins.
assessed in the target individual or sample, body mass estimates from One body mass estimation method that does include a wide range of
footprint size can be improved. Juveniles have relatively large feet, so body forms in the reference sample is the morphometric stature/bi‐iliac
special procedures must be used to estimate their body masses. Body breadth technique described in the preceding section. This method is
masses estimated from footprints are similar to those estimated from based on 58 sex/population mean dimensions, measured anthropometri-
skeletal remains in taxonomic/geographic groupings of Pleistocene cally, with a worldwide distribution, ranging from small‐bodied “Pygmies,”
Homo.25 However, equations based on modern humans overestimate including African Mbuti, Aka, and Kivu Twa, to large‐bodied European and
body mass in Australopithecus afarensis, due to the latter's relatively arctic populations.32,34 Body mass ranges from 38.2 to 82.9 kg and
15206505, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/evan.21988 by Johns Hopkins Medicine, Wiley Online Library on [20/07/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
RUFF and WOOD | 5

stature from 137.3 to 182.1 cm. Both narrow and wide‐bodied specimens, but not to non‐Homo specimens, which have larger
populations are represented, with bi‐iliac breadth ranging from 23.4 to diaphyseal cross sections relative to body size than in Homo.48,83
29.8 cm. These ranges encompass the great majority of specimens in the Other skeletal dimensions, including vertebral body
archaeological and paleontological record, the primary exceptions being breadths,11,13,14,16 breadths of the articular surfaces of the tibiotalar
some very small‐bodied australopiths (such as A.L. 288‐1).20,22,34 (ankle) joint,11,16,84 and craniofacial dimensions,85,86 have also been
Furthermore, for the most part the reference data were collected before used to estimate body in fossil hominins. However, in all cases, the
the last 50 years, on younger adults, avoiding recent increases in relative human reference samples were cadaveric and/or included only a very
body weight in many parts of the world74 as well as increases occurring in limited sampling of modern human morphological variation. Some
middle age.75 When tested, the equations work well on average in both also included only species or sex/species means in broader
elite athletes as well as “normal” nonoverweight subjects.20,31 Thus, they comparative analyses, rather than individual data, precluding assess-
should be broadly applicable to younger, healthy adults in the past as well. ment of true individual‐level prediction errors. The potential
The sex‐specific equations are given in Table 1. problems involved in using cranial dimensions to estimate body
As noted previously, though, relatively few archaeological and mass—changing cranial/postcranial and intracranial proportions dur-
paleontological specimens are complete enough to apply the stature/bi‐ ing hominin evolution—have been discussed previously.85
iliac method directly. Mechanical methods that rely only on measurement To develop body mass estimation equations based on footprint
of long bone dimensions are much more widely applicable, but they size, a special reference sample of habitually unshod (barefoot) or
require appropriate reference samples with recorded body masses in minimally shod modern humans was used, because most of the target
which the skeletal dimensions can be measured, which have the samples were also habitually unshod and shoe wearing affects foot
limitations discussed above. To avoid these limitations, a “hybrid” dimensions.25 A number of geographically dispersed populations
approach that combines morphometric and mechanical methods can be from North and South America, Africa, and the Pacific Islands
employed, where body mass in a skeletal sample is first estimated from (Samoa), exhibiting a large range of body sizes and shapes, were
the stature/bi‐iliac method, and then used to derive body mass prediction included in the reference sample. This contrasted with earlier
equations based on other skeletal dimensions in the same sample. Since attempts to develop such equations,87 which were limited geograph-
many recent skeletal samples are available, this allows inclusion of a much ically and/or ethnically. The importance of including a wider range of
wider range of populations in the reference sample. body forms became apparent when footprint size relative to body
Two reference samples have been used in such analyses: a mass was found to be dependent in part on body shape, with
36
Holocene European sample (n = 1145) and a Holocene worldwide relatively wider‐bodied populations having relatively smaller feet.25
20,24
sample (n = 522–682). In each case individual values, not sex/ Thus, earlier equations produced unrealistic body mass estimates for
population means, were used as data points. The worldwide sample is hominins that did not match the body shapes of the particular
smaller because more stringent requirements were applied for reference sample. Equations were developed from anthropometric
inclusion of specimens.20 It incorporates indigenous populations measurements of foot size in the living reference sample, with
from Europe, eastern and southern Africa, the southwestern US, and conversion factors for converting footprint dimensions to foot
US Euroamericans. Body mass estimation equations were derived dimensions provided. (For foot/footprint area in “normal walking”
based on long bone articular breadths (Figure 1). Both sets of conditions, no conversion is necessary, i.e., footprint area = foot area.)
equations have been applied to a variety of target populations or A quadratic equation and a linear multivariate equation that includes
individuals.20,76–82 The European equations are more appropriate a body shape (BMI) variable are given in Table 1. Use of the latter
specifically for recent (Holocene) European samples, while the equation is recommended whenever possible, with BMI estimated
worldwide equations can be applied to any hominin sample or from related populations or individuals (see Ruff et al.25 for
individual, with a correction for femoral head size in non‐Homo examples). An unshod modern juvenile reference sample was used
specimens, as described earlier. The formulae are listed in Table 1, to develop correction factors for the footprints of likely juveniles
including a new equation for the European reference sample based (also see Section 2.5). Other comparisons between habitually unshod
on proximal tibial mediolateral articular breadth. and shod individuals within a single population provided a correction
Because long bone cross‐sectional diaphyseal dimensions are factor for individuals who likely used footwear.25 As noted earlier,
highly developmentally plastic, responding to differences in activity these equations are only appropriate for Homo footprints; a different
level and muscular forces, and there is evidence for systematic equation for A. afarensis footprints is given in the Table 1 footnote.
variation in such factors between modern and earlier Homo, a special
reference sample restricted to Pleistocene Homo specimens was used
to generate body mass prediction equations from these dimen- 2.3 | Statistical issues and general
sions.37 Femoral head size was used to estimate body mass in 69 recommendations
Pleistocene specimens, and these values were then used to generate
body mass estimation equations based on femoral midshaft cortical The information provided in Table 1 also raises several general
area, and on cortical area combined with maximum bone length statistical issues. The equations based on articular dimensions were
(Table 1). The equations are applicable to other Pleistocene Homo all generated using Reduced Major Axis (RMA) regression rather than
15206505, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/evan.21988 by Johns Hopkins Medicine, Wiley Online Library on [20/07/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
6 | RUFF and WOOD

Least Squares (LS) regression. This is because, as noted earlier, some is available. Among “mechanical” techniques, those based on femoral
of the target specimens (e.g., small‐bodied australopiths) fall outside head and proximal tibial breadths will provide the most accurate body
the range of the reference sample, and in this situation RMA provides mass estimates, although femoral head breadth should be adjusted in
more accurate estimates than LS.50,88 Two different measures of non‐Homo taxa. The “Europe” equations are most appropriate for
predictive precision—the percent standard error of estimate (%SEE) Holocene Europeans, while the “Worldwide” equations can be
and absolute percent prediction error (%PE)—are given for each applied to any area of the world. In specimens that do not preserve
equation. These measure the degree of random error around the femoral head or proximal tibia, humeral articular equations can be
estimates in relative terms. For example, the average %PE for a used, but come with more caveats and are generally not applicable to
body mass estimate of 60 kg derived using the worldwide sample non‐Homo specimens. In the absence of any of these articulations,
femoral head equation in Table 1 is ±4.8 kg (8.0%). The %SEE and % body mass in Pleistocene Homo specimens can be estimated from
PE can be used to help choose between equations based on their femoral midshaft cortical area, combined with femoral length,
predictive precision. Note that the correlation coefficient is an although results should be viewed with more caution because of
imperfect indicator in this regard: because the correlation coefficient possible confounding behavioral factors. Footprints are effective as
is also strongly affected by factors such as sample size and range of x body mass predictors, especially when it is possible to incorporate
and y values. a higher correlation does not necessarily equate with a information regarding body shape (BMI).
lower %SEE and %PE.89 All of the equations in Table 1 have average
%PEs of 10% or less, which is better than many equations reported
for intraspecies and interspecies reference groups.90–92 Among 2.4 | The perils of proxies
articular dimensions, the femoral head produces the lowest predic-
tion errors, followed by the proximal tibia; this is not surprising, given An alternative to estimating body mass directly is to use skeletal
that they are both weight‐bearing joints. Combining femoral length proxies such as long bone lengths and breadths to investigate
with femoral midshaft cortical area, which is equivalent to combining intrataxon or intertaxon variation in body size.8,93–96 This approach
morphometric and mechanically‐based approaches, improves predic- has the advantage that no dimensions are estimated, and the various
tion over cortical area alone. The very low apparent prediction errors issues in estimating body mass noted above are avoided. However,
for the stature/bi‐iliac and footprint/BMI equations (%PEs of around use of proxies rather than body mass itself has some significant
5%) are almost certainly underestimates of true errors in individual disadvantages. Most early hominin skeletal remains are fragmentary,
estimates. The stature/bi‐iliac reference sample consists of sex/ preserving different regions of the skeleton; thus, the same
populations means, and variation among individuals would be dimensions cannot always be compared across specimens assigned
expected to be greater. However, tests of these equations among to the same, or a different taxon. However, using appropriate
samples of individuals of known body mass have generally produced equations, body mass can be estimated from a number of different
average %PEs of less than 10%,20 so in this respect they are still dimensions (Table 1), and can thus serve as a “common denominator”
comparable to the other equations in Table 1. The footprint equation for comparisons across individuals represented by different combi-
that includes a measure of body shape, BMI, assumes that BMI is nations of elements (for example, see Section 3). Body mass is also
actually known for the target individual, whereas in practice it must the most common “size” variable measured in ecological and
be estimated itself from other information, thus increasing overall physiological studies of living species (i.e., you can more conveniently
error. However, this technique is still recommended over the weigh a living animal than measure the size of its femoral head); thus,
equation that includes just footprint area, for reasons that are body mass estimates provide access to a much richer comparative
discussed in detail in the original publication.25 All of the equations in context for interpreting variation among fossil taxa than is available
Table 1 are only applicable to adults, since juveniles have different for individual skeletal dimensions.
body proportions (see Section 2.5). But perhaps the most important limitation of skeletal proxies is
Finally, it should be emphasized that the measures of predictive that they assume similar proportionality across, and within, samples
power in Table 1 are purely statistical in nature. True accuracy of the between the particular skeletal dimension(s) employed and body
equations is dependent not only on the strength of the relationship mass, which is often not the case. There are two factors that should
between a predictor variable and body mass in the reference sample be considered here: overall differences in skeletal/body mass
but also to what extent the same relationship applies in the target proportions between groups, and changes in proportions within
sample. For this reason, several of the equations have limitations in groups that are correlated with body size. Both of these phenomena
terms of appropriate target specimens, and in other cases, either the are part of a broader field of study referred to as allometry, or scaling,
skeletal predictor variable or body mass estimate should be adjusted which examines how body size affects morphology.1,5
to reflect proportional differences from the reference sample, as An illustration of how these factors can affect interpretations of
noted in the table footnotes. body size variation is provided by comparisons of femoral head size
In summary, the stature/bi‐iliac method for estimating body mass scaling with body mass in anthropoid primates. Femoral head breadth
makes the fewest assumptions about body shape and skeletal (or acetabular breadth) is one of the most commonly employed
proportions, and thus should be used if sufficient skeletal material proxies for body mass in comparative studies.8,35,93–97 Figure 2 is a
15206505, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/evan.21988 by Johns Hopkins Medicine, Wiley Online Library on [20/07/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
RUFF and WOOD | 7

slope very close to 3.0. In contrast, the slope of the line through
modern humans is highly nonisometric, with femoral head breadth
increasing much faster than body mass. Baboons show the opposite
pattern, with femoral head size increasing more slowly than body
mass. Thus, large and small modern humans, or large and small
baboons, will show different proportions of body mass to femoral
head size purely as a result of general scaling trends within these taxa
(the underlying functional explanations of these different scaling
trends have been discussed elsewhere).20 This also means that
differences in general proportions between taxa will be in part
dependent on the size range considered; for example, small modern
humans have proportions more similar to those of great apes than do
large modern humans.
Thus, it is clear that using femoral head size as a surrogate for
body mass in comparisons both between and within taxonomic
groups will lead to significant bias in many cases, due to different
average proportions between groups and nonisometric scaling within
F I G U R E 2 Scaling of body mass on femoral head breadth in
some groups. (For further discussion of the dangers inherent in using
modern humans, nonhuman great apes (Gorilla, Pan, Pongo), and
baboons (Papio). All data natural log‐transformed. Regression lines fit other proxies for body size, see Hill et al.101 and Smith102). This is true
through the three groups using Reduced Major Axis regression. for comparisons between nonhuman anthropoid primates as well as
between modern humans and nonhuman primates, and even within
taxonomically/functionally similar groups, such as large and small
plot of body mass relative to femoral head breadth in 682 Holocene modern humans. Furthermore, scaling trends differ from one part of
modern humans, 38 nonhuman great apes (Pan, Gorilla, and Pongo, the skeleton to another (e.g., in modern humans knee joint breadth
hereafter “great apes”), and 20 baboons (Papio cynocephalus). All scales differently than femoral head breadth),20 so trends in one
individuals are adults, and all of the nonhuman specimens were from feature cannot be extrapolated to other skeletal regions. The results
wild populations. Body masses were directly measured in the shown in Figure 2 also argue against combining groups such as
nonhuman taxa90,98 and were estimated from stature and bi‐iliac modern humans and nonhuman hominoids to develop taxonomically
breadth in the human sample.20 Data were logarithmically trans- broader body mass estimation equations.11,15,16 Fitting a regression
formed, allowing easier assessment of proportional changes using line through the combined great ape‐modern human sample in
linear regression. In this type of analysis, line elevations reflect overall Figure 2 would result in an uninterpretable scaling relationship not
proportions of a group, while slopes indicate changes (or not) in observed in any modern taxon. Rather, if a morphologically and
proportions over a size range within a group. All lines were fit with functionally intermediate proportion is hypothesized, as for example
69
RMA regression, appropriate for investigating scaling relationships. in some early hominins that combined terrestrial bipedality with some
Because body mass can be assumed to be proportional to body degree of arboreality,48,103 then equations for different groups can
volume, and body volume is proportional to linear dimensions to the be applied and an average or range of estimates provided. This is the
third power, isometry, or geometric similarity, is indicated here by a approach taken here for estimating body mass from upper limb bone
slope of 3.0 in logarithmic space (i.e., body mass is proportional to dimensions in non‐Homo taxa (see Section 3).
3
femoral head breadth ).
It is apparent from Figure 2 that: (a) the three great apes have the
same scaling relationship between femoral head breadth and body 2.5 | Teenagers are all knees and elbows
mass and (b) modern humans, great apes, and baboons have different
scaling relationships between femoral head size and body mass. Juvenile postcranial specimens can provide critical information on the
Modern humans have the relatively largest femoral heads, adaptive timing of developmental events and the relative importance of
for their bipedal locomotion where all of body mass is supported on genetic and environmental effects on adult morphology.50,104–109
the lower limbs. Both great apes and baboons are quadrupedal and Body size is a potentially important component of such analyses, and
have relatively smaller femoral heads than modern humans, but yet until recently there were no equations available for estimating
femoral head size is relatively larger in the great apes than in body mass from juvenile skeletal remains.50,110,111 Consequently,
baboons. The latter is likely related to great apes' style of vertical earlier studies used indirect methods for assessing body size in
climbing, involving increased hip mobility, which is facilitated by a juveniles, such as weight for height relationships or body (pelvic)
relatively larger femoral head.92,99,100 In addition to these general breadth dimensions.107,112 Even today, body mass estimation
proportional differences, the slopes of the lines through the three equations for juvenile skeletal specimens are available from just
groups are significantly different. Great apes are isometric, with a line one reference sample—the Denver Growth Study sample,50,110,111
15206505, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/evan.21988 by Johns Hopkins Medicine, Wiley Online Library on [20/07/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
8 | RUFF and WOOD

which is limited both ethnically and environmentally.113 Given that year‐old boy.50 If age‐appropriate femoral head equations for 11–13
body mass estimation equations are available for adults from a much year‐olds in the Denver sample are applied, body mass estimates
wider range of populations, as discussed above, it is reasonable to ask between about 50 and 53 kg are obtained.50 If the worldwide adult
whether such equations might be applied to juveniles as well, thus femoral head formula is used instead, his estimated body mass is
broadening the distribution of body sizes and shapes represented. about 63 kg, 19%–26% higher, and larger than would be expected
This question was addressed in an investigation of scaling relation- given his reconstructed stature of less than 157 cm.20,50 Thus, it is
ships between skeletal dimensions and body mass in the Denver clear that adult body mass estimation equations based on femoral
Growth Study sample.50 It was found that during ontogeny both head breadth—and likely any long bone articular breadth—should not
femoral length and femoral head breadth “grow ahead” of body mass in be used for juveniles. Furthermore, the amount of adjustment needed
late childhood and early adolescence, that is, they approach young adult for adult estimates is age‐specific, so no constant correction factor
values earlier, and are thus larger relative to body mass than in adults. As can be applied. Obtaining data similar to those available for the
a consequence, body mass estimation equations based on adults Denver Growth Study sample113 for other populations would be
overestimate body mass in juveniles, particularly in this age range. This is useful for further exploring possible effects of body proportions on
illustrated in Figure 3, which plots the percent difference between body juvenile body mass estimation, although fortunately such effects
mass predicted from femoral head breadth using an adult equation (the appear smaller (in adults) than on stature estimation.28,107,115
worldwide equation in Table 1) and true body mass for 20 individuals in Foot size relative to body mass also varies significantly between
the Denver Growth Study sample followed longitudinally from 6 to 20 juveniles and adults, with juveniles having relatively larger feet.25
years of age (there are fewer points in the oldest age groups because of Thus, adult equations for estimating body mass from footprint size
subject dropout from the study). The maximum deviation between cannot be used directly on juvenile footprints. Again, like femoral
values occurs between about 9 and 13 years of age, averaging 20% or head size, the deviation in proportions between juveniles and adults
more, but by late adolescence (18–20 years) there is no average bias in is not constant, although in this case, the greatest deviation occurs in
estimates, as individuals reach adult proportions. early childhood, with a progressive trend towards adult proportions,
The effects of using an inappropriate adult reference sample for which are reached at about 16 years of age. The developmental
a fossil juvenile can be illustrated by applying adult and age‐matched difference in proportions is likely due to a combination of earlier
formulae to KNM‐WT 15000, an early African Homo erectus male maturation of the foot than most of the rest of the skeleton, and
whose skeletal maturity was equivalent to that of a modern 11‐13‐ changes in body shape (increases in BMI) during growth.25 Because
age is very difficult if not impossible to estimate based on footprint
size alone, a technique was developed whereby the size of a juvenile
footprint relative to an average adult from the same population
(footprint sample) is used to calculate a correction factor for body
mass estimated using the adult formulae.25 The difference in foot/
body mass proportions between juveniles and adults appears to be
relatively constant between populations of different body shape, so
the same correction equations should be universally applicable.
When applied to a juvenile Neanderthal sample,61 they produced
reasonable body mass estimates.25

3 | TEM PO RAL TREN DS IN HO MININ


BOD Y MAS S

The methods described in the preceding sections and summarized in


Table 1 were applied to 299 late Miocene (6.0 Ma [million years ago])
to terminal Pleistocene (12,000 BP [years before present]) hominin
postcranial specimens (individuals) from Africa, Europe, and Asia.
When more than one equation was applicable, the worldwide femoral
head and proximal tibial breadth equations were favored because of
F I G U R E 3 Percent difference between body mass predicted their low prediction errors and more direct functional association
from femoral head breadth using an adult equation and true body with body mass (stature/bi‐iliac equations were very rarely applicable
mass in a modern human juvenile longitudinal sample (extracted from
and were not used here). Femoral head breadths were adjusted for all
the Denver Growth Study). Line fit through data using LOWESS
nonparametric regression.114 [Correction added on 19 July 2023, non‐Homo taxa before entry into the equation, as described in
after first online publication: Minor corrections were made to the y Table 1. Equations based on humeral articular dimensions were used
axis label of Figure 3 and in the Figure 3 caption for clarification.] for Homo, with some adjustments for particular groups as described
15206505, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/evan.21988 by Johns Hopkins Medicine, Wiley Online Library on [20/07/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
RUFF and WOOD | 9

in Table 1. For non‐Homo taxa estimated using humeral articular habilis (sensu stricto) are simply designated “early Homo,” and
dimensions, a mean of body mass estimates derived from modern nonanatomically modern Middle and Late Pleistocene Homo, except
human and chimpanzee reference groups was used, with chimpanzee for Homo naledi and Homo floresiensis, are assigned to “archaic Homo.”
equations derived from reference.90 Supporting Information: Text S1 Individuals assigned to non‐Homo taxa (Orrorin, Ardipithecus,
gives more details. In addition to these skeletal samples, 102 sex/ Australopithecus, Paranthropus) vary in body mass from about
population means for living humans distributed among the same 25–60 kg, and show no evidence of any systematic directional
three geographical regions were included for comparison. All change in body mass between 6.0 and 1.0 Ma. The average body
individual data are given in Supporting Information: Table S1. Subsets mass of these taxa over this time period is 43.8 kg (±10.3 SD). Three
of these data have been published previously.20,24,37 H. habilis (s.s.) specimens, dating to 1.9–2.0 Ma, have a similar mean
A plot of individual body mass estimates against time in shown in body mass of 42.2 kg (range 41–45 kg). Thus, as argued previ-
Figure 4; a dual time scale is used to better differentiate data points ously,119,120 in terms of body size, there is no distinction between H.
after 200,000 BP. Non‐Homo are plotted in red and Homo specimens habilis (s.s.) and non‐Homo taxa.
in blue, with specimens of indeterminate or controversial taxonomic In contrast, non‐habilis early Homo (1.8–2.0 Ma) is somewhat
20,24,37
status in purple. Taxonomy follows previous studies, supple- larger‐bodied, with four specimens averaging 58.6 kg (range
mented by recent analyses of humeral morphology116,117 and general 52–67 kg). By 1.4–1.6 Ma there is a substantial increase in body
postcranial variation among early hominins.118 However, it must be mass in H. erectus/ergaster, with six specimens averaging 75.0 kg
recognized that most early hominin taxa are identified and diagnosed (range 65–90 kg). Following this, average body mass in Homo remains
on the basis of cranial and not postcranial morphology.118 There is relatively constant, with individual specimens ranging between about
also the possibility that taxa that are cranially and dentally distinctive 50 and 90 kg, until the Terminal Pleistocene (post‐25,000 BP), when it
may not necessarily be distinguishable on the basis of postcranial declines. There are three exceptions to this general trend, however.
evidence. Several researchers have recently provided helpful reviews The first is the specimen from Gona, Ethiopia (BSN49/P27), assigned
of the latter,117,118 and, where possible, we have used these to by its discoverers to H. erectus.121 The taxonomic status of this
allocate specimens to taxa. Early (>1.7 Ma) Homo not assigned to H. specimen has been questioned, however, based largely on its very

F I G U R E 4 Body mass trends in hominins from the late Miocene to living humans. Note change in temporal scale at 0.2 Ma (200,000 BP). Red
symbols: non‐Homo; blue symbols: Homo; purple symbols: contested or indeterminate taxonomy. Other subdivisions of the sample are given in
the symbol key (A.: Australopithecus; P: Paranthropus; H.: Homo). “Other non‐Homo” includes Orrorin tugenensis, Ardipithecus ramidus,
Australopithecus anamensis (all labeled), Paranthropus aethiopicus, Australopithecus prometheus, and Australopithecus sp. (unlabeled). Early African
H. erectus includes specimens sometimes attributed to H. ergaster. Unlabeled “archaic Homo” includes H. antecessor, H. heidelbergensis, and H.
neanderthalensis (or H. sapiens neanderthalensis). Living humans are sex/population means.
15206505, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/evan.21988 by Johns Hopkins Medicine, Wiley Online Library on [20/07/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
10 | RUFF and WOOD

small estimated body mass (35.1 kg), which is much lower than any geographically, and temporally.20,25 This confirmation is significant
other Early‐Middle Pleistocene Homo specimen known (Figure 4), because skeletal samples, unlike footprints, are rarely limited to single
122
including all other specimens assigned to H. erectus/ergaster. This populations.
specimen is left taxonomically unassigned in Figure 4, as are four These results are broadly consistent with a number of previous
partial tibiae and one partial humerus without taxonomically studies, but there are some important differences as well. Our
identifiable features. The second exception is the H. naledi sample, conclusion that body mass increased in the earliest members of the
with a mean body mass of 46.2 kg (range 41–67 kg), which is genus Homo that were not H. habilis (s.s.), relative to previous and
considerably lower than other broadly contemporary archaic Homo. contemporary non‐Homo, is in agreement with several stud-
The final exception is the tiny H. floresiensis individual, with an ies,18,22,23,120 but disagrees with others.9,12 A major source of this
estimated body mass of only 31.4 kg. The H. naledi and H. floresiensis discrepancy in results is the use of different body mass estimation
specimens demonstrate the existence of temporally and geograph- equations based on different modern reference samples, with
ically restricted groups of smaller‐bodied Homo in the Middle and equations based on cadaveric samples providing systematically
Late Pleistocene. In the case of H. floresiensis its small size has been smaller estimates and less differentiation in body mass between
73
attributed to island endemism, that is, insular dwarfism. non‐Homo and early Homo samples.9,12 All previous studies agree,
The decline in average body mass at the end of the Pleistocene is however, that body mass increased significantly in Homo after
statistically significant whether or not one includes Late Pleistocene 1.7 Ma, although there are some differences in interpretation of
archaic Homo (i.e., Neanderthals). In fact, Late Pleistocene archaic and subsequent trends within archaic Homo. One study concluded that
anatomically modern (AM) H. sapiens, before 25,000 BP, are not consistently high body masses within Homo were not achieved until
significantly different in body mass.20 The average decline within AM about 0.4–0.5 Ma, in part as a function of climatic adaptation to
H. sapiens between 40,000 and 12,000 BP is about 6 kg, or 9%. Living higher latitudes.22 In contrast, our results indicate that average body
humans are smaller on average than Terminal Pleistocene mass was essentially stable within archaic Homo (with the exception
(25,000–12,000 BP) AM humans (p < .001, t test), but this is in part of H. naledi and H. floresiensis) from 1.6 Ma to near the end of the
confounded by geographic differences in body size. If comparisons Pleistocene. Again, part of the divergence in results is due to
are carried out within regions (Africa, Europe, Asia), there are no methodological differences, including the use in Will et al.22 of lower
statistically significant differences in body mass between Terminal estimates for early Middle Pleistocene (0.77–0.9 Ma) specimens
Pleistocene and living humans. The significance of regional differ- based on cadaveric‐derived equations.
ences in body size within Homo throughout the Pleistocene has been Fine‐tuning body mass estimations can have important ramifica-
discussed elsewhere.37 tions regarding body size variation and its consequences throughout
Sexual dimorphism in body mass is another factor influencing the hominin evolution. For example, an increase in body size within Homo
distribution of values within both taxa and temporal periods. Sexual relative to other hominin taxa may have affected important variables
dimorphism in body mass is greater within non‐Homo taxa than such as home range size, age at maturity, lifespan, subsistence
within Homo taxa or temporally/geographically defined groups, strategy, and social organization,18,120 so the timing of this transition
20
adding to the dispersion of values among the former. The lowest is significant. While earliest non‐habilis Homo can be differentiated
levels of sexual dimorphism are found among populations of from non‐Homo in body size, it is clear that the major increase in body
anatomically modern H. sapiens, although their greater geographic mass within Homo did not occur until after the initial expansion of
distribution adds to overall variability in body mass in H. sapiens, due hominins out of Africa, which was before 1.7 Ma.123,124 This has
34
in part to climatic adaptation. These factors may affect apparent implications for interpreting both the biological and behavioral
temporal trends in body mass, particularly where sample sizes within requirements for geographical expansion in Homo as well as patterns
taxa or temporal periods are relatively small, for example, in early of body size variation throughout the Old World during the
Homo and some non‐Homo taxa. Pleistocene.37,125 The discovery of much smaller‐bodied forms of
Analyses of hominin footprints support the body mass trends Homo later in the Pleistocene (H. naledi and H. floresiensis) raises
shown in Figure 4.25 Presumed Australopithecus afarensis footprints interesting questions regarding localized adaptations within the
63
from Laetoli produce estimated body masses between 31 and genus,70,126 including divergent locomotor behavior.127,128 The decline
59 kg, similar to the range for this taxon determined from skeletal in body mass at the end of the Pleistocene among anatomically
remains. Footprints likely made by Homo erectus/ergaster at Ileret,62 modern H. sapiens has been attributed to a number of possible causes,
dated to about 1.52 Ma, produce body mass estimates that average including changes in diet, demographics, and technology.129,130
about 65 kg, within the range of early Pleistocene archaic Homo
(except H. habilis s.s.). Three adult male presumed Neanderthal
footprints from Le Rozel61 (about 80,000 BP) give average estimated 4 | CONCLUSIONS
body masses of 78–90 kg, similar to larger Late Pleistocene archaic
Homo skeletal estimates. Estimates of sexual dimorphism in body Body mass has become an increasingly important component of
mass in the Laetoli and Ileret footprint samples are also similar to many hominin adaptive scenarios. The accuracy of body mass
those calculated from skeletal remains matched taxonomically, estimates will be affected by several factors, including the type of
15206505, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/evan.21988 by Johns Hopkins Medicine, Wiley Online Library on [20/07/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
RUFF and WOOD | 11

material available for estimation and the appropriateness of the Due to probable use of the upper limb in locomotion (i.e., tree
reference sample used for developing the estimation equations. In climbing) in many non‐Homo taxa, increasing upper limb bone
more complete specimens, a morphometric approach may be applied, dimensions relative to body size, equations based on upper limb
where body mass is estimated from direct reconstruction of the body bone dimensions are not applicable to these taxa.24 Chimpanzee‐
using estimated stature and bi‐iliac (maximum pelvic) breadth. In less based formulae appear to work better for at least some non‐Homo
complete specimens a mechanical approach based on the relationship taxa24; here, an average between chimpanzee and modern human‐
of functionally relevant skeletal dimensions to body mass may be based estimates was used. However, such an approach obviously
applied. The most reliable skeletal dimensions are lower limb articular depends on the extent to which locomotion (and body proportions) in
breadths, especially hip (femoral head) and knee (proximal tibial) various non‐Homo taxa differed from modern humans, something
breadths, with an adjustment for the relatively small size of the that is difficult to determine precisely. Long bone diaphyses (of both
femoral head in non‐Homo specimens. Equations using shoulder limbs) are also stronger relative to body size in non‐Homo taxa48,83;
(humeral head) and elbow (distal humeral) articular breadths have also so methods based on femoral cortical area, even in Pleistocene Homo,
been developed, but they require more adjustments in either cannot be applied to these taxa. Australopiths have relatively small
dimensions or resulting body mass estimates depending on the femoral heads,48 so an adjustment to femoral size must be
target sample, and are not applicable to non‐Homo taxa. A technique incorporated into equations based on this dimension. Australopithe-
based on femoral midshaft cortical area and femoral length can be cus afarensis has relatively large feet,25 necessitating a different
applied to specimens that do not preserve articular surfaces, but it is equation for estimating body mass from footprint size. All of these
subject to caveats stemming from behavioral effects on long bone considerations reflect increased uncertainty of body mass estimates
diaphyses, and again is not applicable to non‐Homo specimens. among non‐Homo taxa. Discovery of additional associated skeletons
Footprint size can also be used to estimate body mass, with better with preservation of multiple skeletal regions, such as A.L. 288‐1
estimates obtained when body shape is factored in. (“Lucy”), will help to further address such uncertainties by allowing
In choosing reference samples, a general objective should be to comparisons between different kinds of estimates.
include individuals as representative as possible of those of the target The use of skeletal proxies as substitutes for body mass in
sample. For example, in forensic applications, a modern, more comparative analyses is discouraged. Specific skeletal features show
overweight sample may be most appropriate. However, for earlier different scaling relationships to body mass in different taxa, and
hominins, who it can be assumed were relatively physically fit, such a even within taxa in different body size ranges. Thus, skeletal proxies
sample will likely produce overestimates of body mass. Cadaveric have a high probability of introducing errors into comparisons, unless
samples, which are subject to wasting and loss of soft tissue, may it can be shown that all units of comparison have the same
have the opposite problem, producing underestimates of body mass proportions. A safer alternative is to actually estimate body mass
in normal, healthy individuals. Both types of samples generally have a using appropriate reference samples, and use that as the body size
limited range of variation in body size and shape, while earlier parameter.
hominins varied greatly in these characteristics. To address these When recently‐developed body mass estimation equations are
issues, a worldwide anthropometric sample of non‐overweight applied to Miocene to recent hominins, some broad patterns emerge.
populations with a wide range of body morphologies was used to Body mass in the earliest hominins varies between about 25 and
develop body mass estimation equations based on stature and bi‐iliac 60 kg, with no consistent temporal trend until the appearance of
(pelvic) breadth. In addition to direct application to relatively Homo, at 1.8–2.0 Ma, which is somewhat larger. This does not apply
complete target specimens, these body mass estimates were also to H. habilis (s.s.), however, which remains small‐bodied. Body mass
used to develop new articular equations based on a wide range of increases greatly in Homo after 1.7 Ma and remains in the general
modern humans. Because cross‐sectional dimensions of long bone range of 50–90 kg until the end of the Pleistocene (12,000–25,000
diaphyses are very developmentally sensitive to differences in BP), when it declines slightly to reach modern values. The small‐
behavior, which likely varied between modern and Pleistocene Homo, bodied Middle Pleistocene H. naledi (41–51 kg) and very small‐bodied
a reference sample of Pleistocene Homo with articular‐based body Late Pleistocene H. floresiensis (31 kg) are exceptions. Variation in
mass estimates was used to develop estimation equations for other hominin body mass has important implications for reconstructing and
Pleistocene Homo specimens using femoral cortical area and bone interpreting changes in such fundamental characteristics as locomo-
length. Ontogenetic effects are another example of the importance tion, diet, brain size, subsistence strategy, social organization, and
of matching reference and target samples: juveniles have different many life history variables. Careful consideration of the assumptions
proportions than adults in both relative articular size and foot size, and limitations of any body mass estimation technique is mandatory
necessitating separate formulae based on age‐matched samples. before its application.
One general observation that can be drawn from these analyses
is that applicability of equations based on modern humans will ACKNOWLEDGME NT S
progressively decline when moving further away from modern Many people over the years have contributed to the collection of
humans taxonomically and/or functionally. This is particularly true data used in this study. For assistance in collating hominin skeletal
for non‐Homo specimens, who are the most distant in both respects. data for Middle and Late Pleistocene specimens, we thank in
15206505, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/evan.21988 by Johns Hopkins Medicine, Wiley Online Library on [20/07/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
12 | RUFF and WOOD

particular Erik Trinkaus. Please see the original publica- 16. McHenry HM. 1992. Body size and proportions in early hominids.
tions20,24,25,45,48,54 for additional acknowledgments. Funding for the Am J Phys Anthropol 87:407‐431.
17. Pilbeam D, Gould SJ. 1974. Size and scaling in human evolution.
collection of original data used in this study was provided by the
Science 186:892‐901.
National Science Foundation (SBR‐891955, SBR‐8919749, BSC‐ 18. Pontzer H. 2012. Ecological energetics in early Homo. Curr
0642297), the Wenner‐Gren Foundation for Anthropological Anthropol 53, Supp. 6:S346‐348.
Research (6084), and the Leakey Foundation. 19. Rightmire GP. 1986. Body size and encephalization in Homo
erectus. Anthropos (Brno) 23:139‐149.
20. Ruff CB, Burgess ML, Squyres N, Junno J‐A, Trinkaus E. 2018.
CO NFL I CT OF INTERES T S T ATEME NT Lower limb articular scaling and body mass estimation in Pliocene
The authors declare no conflict of interest. and Pleistocene hominins. J Hum Evol 115:85‐111.
21. Ruff CB, Trinkaus E, Holliday TW. 1997. Body mass and
encephalization in Pleistocene Homo. Nature 387:173‐176.
D A TA A V A I L A B I L I T Y S T A T E M E N T
22. Will M, Pablos A, Stock JT. 2017. Long‐term patterns of body mass
All data for hominin specimens included in this study are provided in
and stature evolution within the hominin lineage. R Soc Open Sci
Supporting Information: Table S1. 4:171339.
23. Will M, Stock JT. 2015. Spatial and temporal variation of body size
ORCID among early Homo. J Hum Evol 82:15‐33.
24. Ruff CB, Squyres N, Junno J‐A. 2020. Body mass estimation in
Christopher B. Ruff http://orcid.org/0000-0002-2932-3634
hominins from humeral articular dimensions. Am J Phys Anthropol
Bernard A. Wood http://orcid.org/0000-0003-0273-7332 173:480‐499.
25. Ruff CB, Wunderlich RE, Hatala KG, et al. 2021. Body mass
REFERENCES estimation from footprint size in hominins. J Hum Evol 156:102997.
26. Will M, Krapp M, Stock JT, Manica A. 2021. Different environ-
1. Brown JH, West GB, eds. 2000. Scaling in Biology. Oxford: Oxford
mental variables predict body and brain size evolution in Homo. Nat
University Press.
Commun 12:4116.
2. Calder III WA. 1984. Size, Function, and Life History. Cambridge:
27. Puschel HP, Bertrand OC, O'Reilly JE, Bobe R, Puschel TA. 2021.
Cambridge University Press.
Divergence‐time estimates for hominins provide insight into
3. Damuth J, MacFadden BJ, eds. 1990. Body Size in Mammalian encephalization and body mass trends in human evolution. Nat
Paleobiology: Estimation and Biological Implications. Cambridge: Ecol Evol 5:808‐819.
Cambridge University Press. 28. Auerbach BM, Ruff CB. 2004. Human body mass estimation: a
4. Jungers WL. 1985. Size and Scaling in Primate Biology. New York: comparison of “morphometric” and “mechanical” methods. Am
Plenum Press. J Phys Anthropol 125:331‐342.
5. Schmidt‐Nielson K. 1984. Scaling: Why is Animal Size So Important? 29. Fully G. 1956. Une nouvelle méthode de détermination de la taille.
Cambridge: Cambridge University Press. Ann Med Legale 35:266‐273.
6. Tucker MA, Ord TJ, Rogers TL. 2014. Evolutionary predictors of 30. Raxter MH, Auerbach BM, Ruff CB. 2006. A revision of the fully
mammalian home range size: body mass, diet and the environment. technique for estimating statures. Am J Phys Anthropol 130:
Global Ecol Biogeogr 23:1105‐1114. 374‐384.
7. White CR, Alton LA, Bywater CL, Lombardi EJ, Marshall DJ. 2022. 31. Ruff CB. 2000. Body mass prediction from skeletal frame size in
Metabolic scaling is the product of life‐history optimization. Science elite athletes. Am J Phys Anthropol 113:507‐517.
377:834‐839. 32. Ruff CB, Niskanen M, Junno JA, Jamison P. 2005. Body mass
8. Churchill SE, Berger LR, Hartstone‐Rose A, Zondo BH. 2012. Body prediction from stature and bi‐iliac breadth in two high latitude
size in African Middle Pleistocene Homo. In: Reynolds SC, populations, with application to earlier higher latitude humans.
Gallagher A, eds. African Genesis: Perspectives on Hominin Evolution. J Hum Evol 48:381‐392.
Cambridge University Press, pp. 319‐346. 33. Ruff CB. 1991. Climate, body size and body shape in hominid
9. Grabowski M, Hatala KG, Jungers WL, Richmond BG. 2015. Body evolution. J Hum Evol 21:81‐105.
mass estimates of hominin fossils and the evolution of human body 34. Ruff CB. 1994. Morphological adaptation to climate in modern and
size. J Hum Evol 85:75‐93. fossil hominids. Yrbk Phys Anthropol 37:65‐107.
10. Holliday TW. 2012. Body size, body shape, and the circumscription 35. Doube M, Klosowski MM, Wiktorowicz‐Conroy AM, Hutchinson JR,
of the genus Homo. Curr Anthropol 53, Supp. 6:S330‐345. Shefelbine SJ. 2011. Trabecular bone scales allometrically in
11. Jungers WL. 1988. New estimates of body size in australopithe- mammals and birds. Proc Roy Soc B 278:3067‐3073.
cines. In: Grine FE, ed. Evolutionary History of the “Robust” 36. Ruff CB, Holt BM, Niskanen M, et al. 2012. Stature and body mass
Australopithecines. New York: Aldine de Gruyter, pp. 115‐125. estimation from skeletal remains in the European Holocene. Am
12. Jungers WL, Grabowski M, Hatala KG, Richmond BG. 2016. The J Phys Anthropol 148:601‐617.
evolution of body size and shape in the human career. Philos Trans 37. Ruff CB, Sylvester AD, Rahmawati NT, et al. 2022. Two late
R Soc Lond B Biol Sci 371:20150247. Pleistocene human femora from Trinil, Indonesia: implications
13. McHenry HM. 1975. Fossil hominid body weight and brain size. for body size and behavior in Southeast Asia. J Hum Evol
Nature 254:686‐688. 172:103252.
14. McHenry HM. 1976. Early hominid body weight and encephaliza- 38. Arsuaga JL, Lorenzo C, Carretero JM, et al. 1999. A complete
tion. Am J Phys Anthropol 45:77‐84. human pelvis from the Middle Pleistocene of Spain. Nature 399:
15. McHenry HM. 1988. New estimates of body weight in early 255‐258.
hominids and their significance to encephalization and megadontia 39. Niskanen M, Ruff CB. 2018. Body size and shape reconstruction.
in “robust” australopithecines. In: Grine FE, ed. Evolutionary History In: Ruff CB, ed. Skeletal Variation and Adaptation in Europeans:
of the “Robust” Australopithecines. New York: Aldine de Gruyter, pp. Upper Paleolithic to the Twentieth Century. New York Wiley‐
133‐148. Blackwell, pp. 15‐37.
15206505, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/evan.21988 by Johns Hopkins Medicine, Wiley Online Library on [20/07/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
RUFF and WOOD | 13

40. Rosenberg KR, Lü Z, Ruff CB. 2006. Body size, body proportions their implications for understanding fossil hominin paleobiology at
and encephalization in a Middle Pleistocene archaic human from 1.5 Ma. J Hum Evol 112:93‐104.
northern China. Proc Natl Acad Sci 103:3552‐3556. 63. Masao FT, Ichumbaki EB, Cherin M, et al. 2016. New footprints
41. Ruff CB, Holt BM, Sladek V, et al. 2006. Body size, body from Laetoli (Tanzania) provide evidence for marked body size
proportions, and mobility in the Tyrolean “Iceman”. J Hum Evol 51: variation in early hominins. Elife 5:e19568.
91‐101. 64. Elliott M, Kurki H, Weston DA, Collard M. 2016. Estimating body
42. Sellers WI, Hepworth‐Bell J, Falkingham PL, et al. 2012. Minimum mass from skeletal material: new predictive equations and
convex hull mass estimations of complete mounted skeletons. Biol methodological insights from analyses of a known‐mass sample
Lett 8:842‐845. of humans. Archaeol Anthropol Sci 8:731‐750.
43. Brassey CA, O'Mahoney TG, Chamberlain AT, Sellers WI. 2018. A 65. Elliott M, Kurki H, Weston DA, Collard M. 2016. Estimating body
volumetric technique for fossil body mass estimation applied to mass from postcranial variables: an evaluation of current equations
Australopithecus afarensis. J Hum Evol 115:47‐64. using a large known‐mass sample of modern humans. Archaeol
44. Ruff CB, Niskanen M. 2018. Introduction to special issue: body Anthropol Sci 8:689‐704.
mass estimation—methodological issues and fossil applications. 66. Lorkiewicz‐Muszynska D, Przystanska A, Kociemba W, et al. 2013.
J Hum Evol 115:1‐7. Body mass estimation in modern population using anthropometric
45. Grine FE, Jungers WL, Tobias PV, Pearson OM. 1995. Fossil Homo measurements from computed tomography. Forensic Sci Int
femur from Berg Aukas, northern Namibia. Am J Phys Anthropol 97: 231:405.
151‐185. 67. Schaffer WC. 2016. Total body mass estimation from anthropometric
46. Ruff CB, Scott WW, Liu AY‐C. 1991. Articular and diaphyseal measurements in modern young adult U.S. populations with healthy
remodeling of the proximal femur with changes in body mass in body fat percentages (NHANES III). J Forensic Sci 61:1431‐1439.
adults. Am J Phys Anthropol 86:397‐413. 68. Ogden CL, Carroll MD, Kit BK, Flegal KM. 2014. Prevalence of
47. Squyres N, Ruff CB. 2015. Body mass estimation from knee childhood and adult obesity in the United States, 2011‐2012.
breadth, with application to early hominins. Am J Phys Anthropol JAMA 311:806‐814.
158:198‐208. 69. Aiello LC. 1992. Allometry and the analysis of size and shape in
48. Ruff CB, Burgess ML, Ketcham RA, Kappelman J. 2016. Limb bone human evolution. J Hum Evol 22:127‐148.
structural proportions and locomotor behavior in A.L. 288‐1 70. Garvin HM, Elliott MC, Delezene LK, et al. 2017. Body size, brain
(“Lucy”). PLoS One 11:e0166095. size, and sexual dimorphism in Homo naledi from the Dinaledi
49. Ruff CB, Higgins R. 2013. Femoral neck structure and function in chamber. J Hum Evol 111:119‐138.
early hominins. Am J Phys Anthropol 150:512‐525. 71. Trinkaus E. 1981. Neanderthal limb proportions and cold adapta-
50. Ruff CB. 2007. Body size prediction from juvenile skeletal remains. tion. In: Stringer CB, ed. Aspects of Human Evolution. London:
Am J Phys Anthropol 133:698‐716. Taylor and Francis, pp. 187‐224.
51. Lieberman DE, Devlin MJ, Pearson OM. 2001. Articular area 72. Weaver TD. 2009. The meaning of neandertal skeletal morphology.
responses to mechanical loading: effects of exercise, age, and Proc Natl Acad Sci USA 106:16028‐16033.
skeletal location. Am J Phys Anthropol 116:266‐277. 73. Brown P, Sutikna T, Morwood MJ, et al. 2004. A new small‐bodied
52. Ruff CB, Holt BH, Trinkaus E. 2006. Who's afraid of the big bad hominin from the Late Pleistocene of Flores, Indonesia. Nature 431:
Wolff? Wolff's Law and bone functional adaptation. Am J Phys 1055‐1061.
Anthropol 129:484‐498. 74. Ng M, Fleming T, Robinson M, et al. 2014. Global, regional, and
53. Trinkaus E, Churchill SE, Ruff CB. 1994. Postcranial robusticity in national prevalence of overweight and obesity in children and
Homo, II: Humeral bilateral asymmetry and bone plasticity. Am adults during 1980‐2013: a systematic analysis for the Global
J Phys Anthropol 93:1‐34. Burden of Disease Study 2013. Lancet 384:766‐781.
54. Ruff CB, Higgins R, Carlson KJ. 2020. Long bone cross‐sectional 75. Junno JA, Niskanen M, Maijanen H, et al. 2018. The effect of age
geometry. In: Zipfel B, Ward CV, Richmond BG, eds. Hominin and body composition on body mass estimation of males using the
Postcranial Remains from Sterkfontein, South Africa, 1936‐1995. stature/bi‐iliac method. J Hum Evol 115:122‐129.
Oxford: Oxford University Press, pp. 307‐320. 76. Fukase H, Wakebe T, Tsurumoto T, Saiki K, Fujita M, Ishida H.
55. Ruff CB, Holt BM, Niskanen M, et al. 2015. Gradual decline in 2012. Geographic variation in body form of prehistoric Jomon
mobility with the adoption of food production in Europe. Proc Natl males in the Japanese archipelago: its ecogeographic implications.
Acad Sci 112:7147‐7152. Am J Phys Anthropol 149:125‐135.
56. Ruff CB. 2003. Growth in bone strength, body size, and muscle size 77. Macintosh AA, Pinhasi R, Stock JT. 2014. Lower limb skeletal
in a juvenile longitudinal sample. Bone 33:317‐329. biomechanics track long‐term decline in mobility across ~6150
57. Ruff CB. 2000. Body size, body shape, and long bone strength in years of agriculture in Central Europe. J Arch Sci 52:376‐390.
modern humans. J Hum Evol 38:269‐290. 78. Ruff CB, ed. 2018. Skeletal Variation and Adaptation in Europeans:
58. Ruff CB. 2009. Relative limb strength and locomotion in Homo Upper Paleolithic to the Twentieth Century. Hoboken: Wiley‐
habilis Am J Phys Anthropol 138:90‐100. Blackwell.
59. Lockley M, Roberts G, Kim JY. 2008. In the footprints of our 79. Sládek V, Macháček J, Makajevová E, Přichystalová R, Hora M.
ancestors: an overview of the hominid track record. Ichnos 15: 2018. Body mass estimation in skeletal samples using the hybrid
106‐125. approach: the effect of population‐specific variations and sexual
60. Hatala KG, Roach NT, Behrensmeyer AK. 2022. Fossil footprints dimorphism. Arch Anthropol Sci 10:833‐847.
and what they mean for hominin paleobiology. Evol Anthropol 32: 80. Trinkaus E. 2018. Epipaleolithic human appendicular remains from
39‐53. Ein Gev I, Israel. C R Palevol 17:616‐627.
61. Duveau J, Berillon G, Verna C, Laisne G, Cliquet D. 2019. The 81. Trinkaus E, Sparacello VS, Xing S, Thibeault A, Villotte S. 2022.
composition of a Neandertal social group revealed by the hominin Describing cro‐magnon: the femora, tibiae and fibulae. J Archaeol
footprints at Le Rozel (Normandy, France). Proc Natl Acad Sci USA Sci Rep 42:103418.
116:19409‐19414. 82. Wallace IJ, Nesbitt A, Mongle C, Gould ES, Grine FE. 2014. Age‐
62. Hatala KG, Roach NT, Ostrofsky KR, et al. 2017. Hominin track related variation in limb bone diaphyseal structure among Inuit
assemblages from Okote Member deposits near Ileret, Kenya, and foragers from Point Hope, northern Alaska. Arch Osteoporos 9:202.
15206505, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/evan.21988 by Johns Hopkins Medicine, Wiley Online Library on [20/07/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
14 | RUFF and WOOD

83. Ruff CB. Changes in relative bone strength through hominin evolution 106. Ruff CB, Trinkaus E, Holliday TW. 2002. Body proportions and size.
in relation to mechanical loadings and metabolic trade‐offs. Interna- In: Zilhão J, Trinkaus E, eds. Portrait of the Artist as a Child: The
tional Congress on Vertebrate Morphology; 2019; Prague. Gravettian Human Skeleton from the Abrigo do Lagar Velho and its
84. Cunningham DL, Rogers MV, Wescott DJ, McCarthy RC. 2019. Archaeological Context. 22. Lisbon: Instituto Português de Arqueo-
Reevaluation of the body mass estimate for the KNM‐ER 5428 logia, pp. 365‐391.
Homo erectus talus. Am J Phys Anthropol 170:148‐155. 107. Ruff CB, Walker A. 1993. Body size and body shape. In: Walker A,
85. Aiello LC, Wood BA. 1994. Cranial variables as predictors of Leakey R, eds. The Nariokotome Homo erectus Skeleton. Cambridge:
hominine body mass. Am J Phys Anthropol 95:409‐426. Harvard University Press, pp. 234‐265.
86. Kappelman J. 1996. The evolution of body mass and relative brain 108. Susman RL, Stern JT. 1982. Functional Morphology of Homo
size in fossil hominids. J Human Evol 30:243‐276. habilis. Science 217:931‐934.
87. Dingwall HL, Hatala KG, Wunderlich RE, Richmond BG. 2013. Hominin 109. Weaver TD, Coqueugniot H, Golovanova LV, Doronichev VB,
stature, body mass, and walking speed estimates based on 1.5 million‐ Maureille B, Hublin JJ. 2016. Neonatal postcrania from Mez-
year‐old fossil footprints at Ileret, Kenya. J Hum Evol 64:556‐568. maiskaya, Russia, and Le Moustier, France, and the development
88. Konigsberg LW, Hens SM, Jantz LM, Jungers WL. 1998. Stature of Neandertal body form. Proc Natl Acad Sci USA 113:
estimation and calibration: Bayesian and maximum likelihood 6472‐6477.
perspectives in physical anthropology. Yrbk Phys Anthopol 41:65‐92. 110. Robbins G, Sciulli PW, Blatt SH. 2010. Estimating body mass
89. Smith RJ. 1984. Allometric scaling in comparative biology: in subadult human skeletons. Am J Phys Anthropol 143:146‐150.
Problems of concept and method. Am J Physiol 246:R152‐R160. 111. Schug GR, Gupta S, Cowgill LW, Sciulli PW, Blatt SH. 2013. Panel
90. Burgess ML, McFarlin SC, Mudakikwa A, Cranfield MR, Ruff CB. regression formulas for estimating stature and body mass from
2018. Body mass estimation in hominoids: Age and locomotor immature human skeletons: a statistical approach without refer-
effects. J Hum Evol 115:36‐46. ence to specific age estimates. J Archaeol Sci 40:3076‐3086.
91. Dagosto M, Gebo D, Ni X, Smith T. 2018. Estimating body size in 112. Kondo O, Dodo Y. 2002. A study of body shape based on skeletal
early primates: the case of Archicebus and Teilhardina. J Hum Evol reconstruction. In: Akazawa T, Muhesen S, eds. Neandertal Burials.
115:8‐19. Excavations of the Dederiyeh Cave, Afrin, Syria. Kyoto: International
92. Perry JMG, Cooke SB, Runestad Connour JA, Burgess ML, Ruff CB. Research Center for Japanese Studies, pp. 339‐343.
2018. Articular scaling and body mass estimation in platyrrhines 113. Ruff CB. 2022. Bone structural data for the Denver longitudinal
and catarrhines: modern variation and application to fossil growth study. Am J Biol Anthropol 178:544‐547.
anthropoids. J Hum Evol 115:20‐35. 114. Cleveland WS. 1979. Robust locally weighted regression and
93. Lorenzo C, Carretero JM, Arsuaga JL, Gracia A, Martínez I. 1998. smoothing scatterplots. J Am Stat Assoc 74:829‐836.
Intrapopulational body size variation and cranial capacity variation 115. Holliday TW, Ruff CB. 1997. Ecogeographic patterning and stature
in Middle Pleistocene humans: the Sima de los Huesos sample prediction in fossil hominids: comment on Feldesman and Fountain.
(Sierra de Atapuerca, Spain). Am J Phys Anthropol 106:19‐33. Am J Phys Anthropol 103:137‐140.
94. Reno PL, Meindl RS, McCollum MA, Lovejoy CO. 2003. Sexual 116. Lague MR. 2015. Taxonomic identification of Lower Pleistocene
dimorphism in Australopithecus afarensis was similar to that of fossil hominins based on distal humeral diaphyseal cross‐sectional
modern humans. Proc Natl Acad Sci USA 100:9404‐9409. shape. PeerJ 3:e1084.
95. Trinkaus E. 1980. Sexual differences in Neanderthal limb bones. 117. Lague MR, Ward CP. in press. The Paranthropus postcranial
J Hum Evol 9:377‐397. puzzle. In: Constantino PJ, Reed KE, Wood BA, eds. The
96. Ward CV, Maddux SD, Middleton ER. 2018. Three‐dimensional Forgotten Lineage(s): Paleobiology of Paranthropus. Cham, Switz-
anatomy of the anthropoid bony pelvis. Am J Phys Anthropol 166: erland: Springer.
3‐25. 118. Grine FE, Mongle CS, Fleagle JG, Hammond AS. 2022. The
97. Ryan TM, Shaw CN. 2013. Trabecular bone microstructure scales taxonomic attribution of African hominin postcrania from the
allometrically in the primate humerus and femur. Proc Biol Sci Miocene through the Pleistocene: Associations and assumptions.
280:20130172. J Hum Evol 173:103255.
98. Ruff CB. 2003. Long bone articular and diaphyseal structure in Old 119. Wood BA, Collard M. 1999. The human genus. Science 284:65‐71.
World monkeys and apes, II: estimation of body mass. Am J Phys 120. McHenry HM. 1994. Behavioral ecological implications of early
Anthropol 120:16‐37. hominid body size. J Hum Evol 27:77‐87.
99. Ruff CB. 2002. Long bone articular and diaphyseal structure in Old 121. Simpson SW, Quade J, Levin NE, et al. 2008. A female Homo
World monkeys and apes, I: Locomotor effects. Am J Phys erectus pelvis from Gona, Ethiopia. Science 322:1089‐1092.
Anthropol 119:305‐342. 122. Ruff CB. 2010. Body size and body shape in early hominins—
100. Ruff CB. 1988. Hindlimb articular surface allometry in Hominoidea implications of the Gona pelvis. J Hum Evol 58:166‐178.
and Macaca, with comparisons to diaphyseal scaling. J Hum Evol 17: 123. Ferring R, Oms O, Agusti J, et al. 2011. Earliest human occupations
687‐714. at Dmanisi (Georgian Caucasus) dated to 1.85–1.78 Ma. Proc Natl
101. Hills M, Wood BA. 1984. Regression lines, size, and allometry. In: Acad Sci USA 108:10432‐10436.
Chivers DJ, Wood BA, Bilsborough A, eds. Food Acquisition and 124. Zhu Z, Dennell R, Huang W, et al. 2018. Hominin occupation of the
Processing in Primates. Boston: Springer, pp. 557‐567. Chinese Loess Plateau since about 2.1 million years ago. Nature
102. Smith RJ. 1981. On the definition of variables in studies of primate 559:608‐612.
dental allometry. Am J Phys Anthropol 55:323‐329. 125. Dennell R, Roebroeks W. 2005. An Asian perspective on early
103. Lovejoy CO, Suwa G, Spurlock L, Asfaw B, White TD. 2009. The human dispersal from Africa. Nature 438:1099‐1104.
pelvis and femur of Ardipithecus ramidus: the emergence of upright 126. Scerri EML, Thomas MG, Manica A, et al. 2018. Did our species
walking. Science 326:71. evolve in subdivided populations across Africa, and why does it
104. Alemseged Z, Spoor F, Kimbel WH, et al. 2006. A juvenile early matter? Trends Ecol Evol 33:582‐594.
hominin skeleton from Dikika, Ethiopia. Nature 443:296‐301. 127. Jungers WL, Harcourt‐Smith WEH, Wunderlich RE, et al. 2009. The
105. Rosas A, Rios L, Estalrrich A, et al. 2017. The growth pattern of foot of Homo floresiensis. Nature 459:81‐84.
Neandertals, reconstructed from a juvenile skeleton from El Sidron 128. Kivell TL, Churchill SE, Kibii JM, Schmid P, Berger LR. 2018. The
(Spain). Science 357:1282‐1287. hand of Australopithecus sediba. Paleoanthropol 282‐333.
15206505, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/evan.21988 by Johns Hopkins Medicine, Wiley Online Library on [20/07/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
RUFF and WOOD | 15

129. Frayer DW. 1984. Biological and cultural change in the European
Bernard A. Wood is University Professor of Human Origins at
Late Pleistocene and Early Holocene. In: Smith FH, Spencer F, eds.
The Origins of Modern Humans: A World Survey of the Fossil George Washington University. Interested in fossil hominins
Evidence. New York: Wiley‐Liss, pp. 211‐250. since joining Richard Leakey's first expedition to Lake Turkana,
130. Niskanen M, Ruff CB, Holt B, Sládek V, Berner M. 2018. Kenya, in 1968, he is best known for his research on the origins
Temporal and geographic variation in body size and shape of
of Homo, hominin systematics, phylogeny reconstruction, and
Europeans from the Late Pleistocene to recent times. In:
Ruff CB, ed. Skeletal Variation and Adaptation in Europeans: comparative morphology.
Upper Paleolithic to the Twentieth Century. Hoboken: Wiley‐
Blackwell, pp. 49‐89.

A U T H O R B I O G R A P H IE S SUPP ORTING INFO RM ATION


Additional supporting information can be found online in the
Christopher B. Ruff is Professor Emeritus and former Supporting Information section at the end of this article.
director of the Center for Functional Anatomy and Evolution,
Johns Hopkins University School of Medicine. His research
interests focus on evolution and environmental adaptation of How to cite this article: Ruff CB, Wood BA. The estimation
the hominoid postcranium, skeletal growth and development, and evolution of hominin body mass. Evolutionary
and behavioral reconstruction through biomechanical Anthropology. 2023;1‐15. doi:10.1002/evan.21988
analyses.

You might also like