AMERICAN JOURNAL OF PHYSICAL ANTHROPOLOGY 120:233–251 (2003)

Estimating the Length of Incomplete Long Bones:

Forensic Standards From Guatemala
Lori E. Wright1* and Mario A. Vásquez2
1
Department of Anthropology, Texas A&M University, College Station, Texas 77843-4352
2
Escuela de Historia, Universidad de San Carlos de Guatemala, Guatemala City, Guatemala

KEY WORDS forensic anthropology; stature; long bone fragments; regression; Maya

ABSTRACT We report on new standards for estimat- r2 ⬎ 0.85 as reliable. Landmarks defined by muscle at-
ing long bone length from incomplete bones for use in tachment sites were more variable in location than land-
forensic and archaeological contexts in Central America. marks on articular architecture; thus we retained few
The measurements we use closely follow those defined by equations that use these landmarks. We tested the male
Steele ([1970] Personal Identification in Mass Disasters; and combined sex equations on 36 males of unknown
Washington, DC: Smithsonian Institution), but we add ethnicity exhumed from a military base in Guatemala,
several new landmarks. We measured the femur, hu- and found that the equations performed satisfactorily. We
merus, tibia, and fibula of 100 Maya skeletons (68 males, also evaluated the performance of equations by Steele
32 females) recovered from forensic exhumations. We de- ([1970] Personal Identification in Mass Disasters; Wash-
rived the equations by regressing bone segment length on ington, DC: Smithsonian Institution) and Jacobs ([1992]
bone length, and solved for bone length to maximize the Am J Phys Anthropol 89:333–345) on the Maya bones, and
utility of the equations for taller populations. We gener- conclude that significant population variation in long bone
ated equations for all segments that were significantly proportions hinders their application in Central America.
correlated with bone length for males, for females, and for Am J Phys Anthropol 120:233–251, 2003.
both sexes combined, but accepted only regressions with © 2003 Wiley-Liss, Inc.

The estimation of standing stature during life is a tions of long bone segments (or conversely, in the
key goal in the forensic analysis of human skeletal relative locations of bony landmarks on the bone)
material, because height is a straightforward biolog- may be due to a diversity of factors, including as-
ical characteristic that may aid the investigator in pects of genetic, climatic, and nutritional environ-
deducing the identity of a skeleton. Likewise, for ments. However, the position of long bone land-
archaeological materials, stature provides an impor- marks may also be determined by functional
tant means to address health and adaptation in the stresses on the limb (Jacobs, 1992). Do landmark
past, as it is a sensitive measure of the growth positions scale allometrically with bone length? Or
status of a population. Hence, the development of are functional environments and activity patterns
techniques for estimating stature from skeletal re- more significant? Here, we develop regression equa-
mains has been a core focus of both forensic anthro- tions to estimate bone length for forensic Maya skel-
pology and bioarchaeology since their inception. In etons from Guatemala. Because of the extremely
both fields, stature estimation is often problematic small stature of Maya populations, we reasoned that
because of deterioration of the long bones during the proportions of long bone segments might vary
burial. Methods to circumvent this problem by esti- substantially from published values, as a function of
mating stature from fragmentary bones have been size alone. In addition to providing standards appro-
devised by several researchers. Prominent among priate to this population, we use the Maya data to
these are the equations developed by Steele (1970);
(also Steele and McKern, 1969) to estimate bone
Grant sponsor: Office of the Vice President for Research and Grad-
length for the fragmentary humerus, femur, and uate Studies, Texas A&M University.
tibia. Criticisms of the method have focused on dif-
ficulty in identifying landmarks used in the equa- *Correspondence to: Lori E. Wright, Department of Anthropology,
tions, and on the possibility that the regressions Texas A&M University, College Station, TX 77843-4352.
may be population-specific. E-mail: [email protected]
In this paper, we revisit Steele’s approach (1970)
Received 26 June 2001; accepted 18 February 2002.
to long bone length estimation. We explore the issue
of population specificity of bone segment proportions DOI 10.1002/ajpa.10119
in greater detail. As with variation in limb bone Published online in Wiley InterScience (www.interscience.wiley.
proportions, interpopulation variation in the propor- com).

© 2003 WILEY-LISS, INC.

234 L.E. WRIGHT AND M.A. VÁSQUEZ

examine the nature of interpopulation variation in neck, and head to estimate stature. Using skeletons
bone segment proportions. from the Terry Collection, they found reasonably
STATURE ESTIMATION FROM FRAGMENTARY strong correlations between these measures and
LONG BONES long bone length (r ⫽ 0.22– 0.68). Similarly, Holland
(1992) used the biarticular breadth of the proximal
The estimation of bone length from incomplete tibia to generate a stature regression equation for
long bones was pioneered by Müller (1935), who skeletons from the Hamann-Todd Collection at the
generated equations for the humerus, radius, and Cleveland Museum of Natural History. Although
tibia from measurements of the skeletal collections both of these approaches obtained statistically sig-
of the Österreiches Beinhaus in Zellerndorf. Müller nificant correlations between the transverse mea-
(1935) defined 5 segments for the humerus, 4 for the surements and maximum bone length on the refer-
radius, and 7 for the tibia, using the margins of ence populations from which they were developed,
articular surfaces and key points of muscle attach- they may find less application than the longitudinal
ment to define the segments. She calculated the measurement approach of Steele (1970). The trans-
percentage of total length represented by each sec- verse measurement methods emphasize parts of
tion, and used this value to estimate the length of bones (epiphyses and metaphyses) that are least
fragmentary material. often preserved in forensic and archaeological con-
Working with Mississippian archaeological mate- texts; their utility will be greatest for recent forensic
rial from northeast Arkansas (Steele and McKern, cases. Moreover, transverse measures vary substan-
1969; Steele and Bramblett, 1988) and with skele- tially among populations, together with the degree
tons of American Blacks and whites from the Smith- of sexual dimorphism. Such transverse measure-
sonian’s Terry Collection (Steele, 1970), Steele gen- ment approaches are in no real sense a “revision” of
erated a series of sex- and population-specific Steele’s method (1970) (contra Simmons et al.,
equations for the humerus, femur, and tibia. He 1990), but should be seen as complementary ap-
defined landmarks to delimit 4 segments on the proaches to bone length estimation.
humerus, 4 on the femur, and 5 on the tibia, in part Steele (1970) had generated regression equations
following the landmarks used by Müller (1935). For independently for several populations because he
each segment, he generated a regression equation to
reasoned that bone segment proportions might dif-
estimate bone length. He also calculated multiple
fer among populations, much as total bone propor-
regression formulae to estimate bone length using a
tions do. In applying the equations to complete fem-
combination of segments, and provided equations
ora and tibiae of Mesolithic and Neolithic European
that relate segment length directly to stature, to
skeletons, Jacobs (1992) demonstrated interpopula-
reduce the large standard error that would pertain
after applying a stature regression formula to esti- tion variation in the proportions of long bone seg-
mated bone length. Standard errors of the formulae ments. He observed that some of the equations of
for estimated bone length on the reference popula- Steele (1970) tended to overestimate bone length,
tion are smaller for the multiple regressions than for while others underestimated length significantly,
individual segment equations, and span 0.20 –2.93 and he concluded that bone length could not be
cm. Stature estimates using Steele’s regressions accurately estimated for these skeletons using these
have standard errors of 3.71– 6.17 cm. equations. However, he demonstrated that segment
Critiques of the method of Steele (1970) have pri- lengths did show significant correlations with bone
marily focused on the difficulty that some osteolo- length in this population, and he generated popula-
gists have had in locating several of the landmarks tion-specific equations, thus confirming the utility of
that define the bone segments (Brooks et al., 1990; Steele’s approach. Jacobs (1992) showed that the
Holland, 1992; Simmons et al., 1990). In a survey of poor performance of the femur equations of Steele
members of the American Academy of Forensic Sci- (1970) on the European sample was primarily due to
ences, Brooks et al. (1990) found that 55% (6 of 11) of differences in the proportions of segments 2 and 3,
respondents who had applied the formulae to re- which are separated by a landmark defined by the
mains for which they had an independent estimate divergence of the medial and lateral supracondylar
of stature found that Steele’s formulae (1970) did lines from the linea aspera. Similarly, he observed a
indeed estimate stature accurately. They also un- different proportion of tibial segments 1 and 2,
derscored the problematic nature of stature data which are divided by the landmark defined by the
obtained from documentary and cadaveral sources, proximal margin of the tibial tuberosity. He sug-
to which skeletal estimates are routinely compared. gested that population differences in muscular ac-
Errors in these “known” data undoubtedly contrib- tivity would explain these differences in proportion,
ute to perceptions that skeletal stature estimates and recommended that researchers search out sam-
are inaccurate. ples with diverse behavioral characteristics and
Two studies proposed stature estimation stan- adaptive strategies in order to generate regression
dards that utilize transverse measurements taken equations that might be more appropriate analogues
on long bones. Simmons et al. (1990) used standard for both forensic and bioarchaeological applications
osteometric measurements of the femoral condyles, (Jacobs, 1992).
 ESTIMATING LENGTH OF INCOMPLETE LONG BONES 235
TABLE 1. Mean stature for males and females in the Guatemalan forensic remains, and in comparative series
used for estimation of long bone length and stature

Skeletal series or Male stature Female stature

reference sample Source Mean (cm) SD (cm) N Mean (cm) SD (cm) N
Forensic, Mayan1 This study 158.25 4.48 66 147.29 5.45 32
Forensic, El Chal1 This study 160.78 4.05 33
Mexican cadavers Genovés, 1967 163.99 5.11 22 152.30 6.71 15
Terry Collection, White Steele, 1970 168.44 8.11 61 157.62 7.96 52
Terry Collection, Black Steele, 1970 172.02 7.84 42 159.88 6.22 57
Mississippian1 Steele and McKern, 1969 165.35 — 72 154.75 — 29
1
Stature is calculated from the measured maximum length of the femur, using the femur equation of Genovés (1967).
2
Unreported.

Although he confirmed the success of Steele’s though interments have also been recovered from
methodological approach, Jacobs (1992) demon- military bases and other contexts, most of the civil-
strated that the standards of (Steele, 1970) may not ian victims of the conflict were rural subsistence
accurately estimate bone length for all skeletal pop- farmers of Maya descent and had been hastily in-
ulations. This is likely to be especially true for pop- terred in small “clandestine” cemeteries in remote
ulations that diverge significantly in completed areas of the Guatemalan highlands. To investigate
adult stature or body proportions from reference the applicability of Steele’s approach to estimating
populations of Steele’s. In both our forensic and bio- bone length from long bone segments, we generate
archaeological research in Guatemala, we are sel- regression equations from measurements of these
dom able to estimate stature from intact bones due forensic Maya remains. We test the regressions on
to the rapid deterioration of bone in this tropical an independent forensic sample, and also evaluate
environment, and due to other perimortem and post- the accuracy of the regression equations of Steele
depositional factors. We began our investigation (1970; also Steele and Bramblett, 1988) and Jacobs
with the suspicion that the existing standards would (1992) on forensic Maya remains.
not be appropriate for Maya skeletal remains be-
MATERIALS AND METHODS
cause of the extremely diminutive stature of Maya
populations. Twentieth century Maya populations The skeletal remains used for this study are
in Guatemala and Mexico experience marked stunt- treated as two separate samples. The largest sam-
ing of stature due to nutritional insufficiency during ple, hereafter referred to as the “forensic Maya sam-
the period of childhood growth (Bogin and MacVean, ple,” consists of some 100 human skeletons of rural
1984; Danforth, 1994; Furbee et al., 1988; Martorell, highland Maya villagers. We use this skeletal sam-
1995; Martorell et al., 1994; McCullough, 1982). Ta- ple to generate regression equations to estimate long
ble 1 illustrates that the forensic Maya skeletons bone length. The forensic Maya sample comprises 76
from Guatemala used in this study fall about 10 cm skeletons recovered by exhumations carried out by
short of the average stature of the skeletal series the Area of Exhumations of the Office of Human
from which Steele (1970) generated his regression Rights of the Archbishop of Guatemala (ODHAG),
equations. Maya skeletal remains are also very together with 24 skeletons recovered by the Foun-
slight; established standards for sex estimation from dation for Forensic Anthropology of Guatemala
bone dimensions typically misclassify many males (FAFG). Although the specific identity of many in-
as females. Therefore, most bioarchaeologists work- dividual skeletons is not known, all of the remains
ing on Maya remains have generated series-specific derive from clandestine cemeteries near remote ru-
discriminant functions for metric sex estimation ral communities. The circumstances of the massa-
(Whittington, 1989; Wright, 1994). Both the stature cres and subsequent interment are now well-docu-
and robusticity of Maya skeletons raise the question mented, and permit identification of the remains as
of whether Steele’s formulae would work on Maya to village and/or ethnicity. Positive identities were
remains. achieved for a sizable proportion of the remains (Ar-
We report on new regression equations derived zobispado de Guatemala, 1998; Equipo de Antropo-
from measurements of forensic skeletons from Gua- logı́a Forense de Guatemala, 1997; unpublished
temala. The skeletons were exhumed as part of the data). We considered only adult skeletons, which
ongoing process of reconciliation following Guate- showed complete fusion of all long bone epiphyses.
mala’s long civil war, which spanned the 1954 CIA- Males outnumbered females almost 2:1 in the sam-
sponsored coup d’etat to a final peace agreement ple because females were less often the targets of
signed in 1996. Many of the remains that have been military action than males. We measured all com-
the subject of forensic investigation derive from a plete adult remains that were available for study
period of brutal “scorched earth campaigns” carried during the period of data collection. The remains
out by the military between 1982–1984 (Arzobis- include skeletons from five major highland Maya
pado de Guatemala, 1998; Carmack, 1988; Equipo ethnic groups (Table 2); there is no ethnic difference
de Antropologı́a Forense de Guatemala, 1997). Al- in skeletal stature (as estimated from femur
236 L.E. WRIGHT AND M.A. VÁSQUEZ
TABLE 2. Maya forensic sample by ethnic group and sex board, with the long axis of the diaphysis parallel to
Ethnic group Male Female Total the length of the osteometric board. We use clear
plastic triangles to extrapolate the location of land-
Achi 1 5 6
Ixil 3 5 8 marks relative to the metric scale on the board. For
Kakchiquel 17 3 20 the humerus, femur, and fibula, we align the bone on
Keqchi 14 16 30 the osteometric board with the most proximal por-
Quiche 33 3 36
Total 68 32 100
tion of the bone that is present at the zero stop; thus,
the landmarks are numbered from proximal to dis-
tal. For the tibia, the points are numbered from
lengths) among males of Kakchiquel, Keqchi, and distal to proximal because the maximum length
Quiche descent in the sample (ANOVA, P ⫽ 0.64). (landmark T7) must be measured with the block of
Unfortunately, cell sizes are too small to test statu- the board on the lateral malleolus in order to ex-
ral differences among ethnic groups for the females. clude the cruciate eminence. For the femur, all mea-
The second forensic sample consists of 36 male surements are taken with the bone in a prone posi-
skeletons exhumed by the FAFG from the grounds of tion on the osteometric board (dorsal surface up).
the military base at El Chal, Petén. No females were The fibula is placed on its dorsolateral surface, such
recovered from this site. These remains are of “dis- that the distal articulation is facing to one side and
appeared” individuals, so their ethnic identity is less slightly upwards. Because some of the landmarks on
well-constrained than in the Maya sample. The the humerus and tibia are located on the ventral
mean stature of the El Chal remains is approxi- aspect of the bone, while others are dorsal, the bone
mately 2.5 cm taller than the mean stature of males must be turned over to measure several landmarks.
in the Maya sample (Table 1); thus it is likely that After turning the bone, we verify the total length of
the sample includes a greater proportion of Ladino the bone (or portion present) in order to ensure
individuals (of mixed Spanish and indigenous Maya accurate alignment.
ancestry). We use this sample to test the applicabil- Rather than the multiple regression approach
ity of regression equations developed from the foren- used by Steele (1970; also Steele and McKern, 1969;
sic Maya sample to forensic remains of unknown Jacobs, 1992), we follow a univariate approach to the
ethnicity in Guatemala and Central America at regression analyses, calculating regression equa-
large. tions against bone length for the distance between
A standard set of measurements was taken on pairs of landmarks on each bone. Therefore, it is
four major long bones for each skeleton. All of the possible to estimate length for bones that suffer from
bones were measured by M.V., using the same os- erosion of intermediate landmarks, which would not
teometric board, between June–December 1998. He be possible using the multiple regression approach.
measured the left bone for each skeleton, but sub- We calculate the regression equation for a given pair
stituted the right side for skeletons in which the of points for each sex separately, and for both sexes
right bone was better preserved than the left. All combined. In addition, we regress fibula segment
measurements were recorded in millimeters. The length against the maximum length of the tibia,
landmarks we measured are described in Table 3 since fibula and tibia length are highly correlated
and illustrated in Figure 1. For the most part, the (r ⫽ 0.982, n ⫽ 76 for Maya males and females,
landmarks measured follow those defined by Steele combined), and standards for stature estimation
(1970). However, we eliminated one landmark on from the fibula are scarce.
the tibia (no. 3) that we could not locate confidently, Instead of regressing bone length (y) on segment
and we added several new landmarks. In addition to length (x), as is commonly done in forensic stature
the humerus, femur, and tibia, we also measured estimation, we regress segment length (y) on bone
the locations of several landmarks on the fibula. length (x) and then solve for bone length. This
Rather than numbering the bone segments (Steele method, known as classical calibration, has been
and McKern, 1969), we refer to the segment lengths applied to age estimation (Konigsberg et al., 1997),
by a code that incorporates a letter representing the and is the approach favored in the allometry litera-
bone and the landmark numbers. For instance, ture. The former “inverse calibration” method ap-
“F2–5” represents the segment between landmarks proaches the problem from the perspective that bone
F2 and F5 on the femur. Likewise, “H0 –2” repre- length is a product of segment length, an essentially
sents the humerus segment from landmarks H0 to Bayesian approach. Classical calibration takes a
H2, and “T3–5” represents the tibia segment from maximum likelihood approach, and sees segment
landmarks T3 to T5. To avoid confusion between the length as constrained by bone length. For recon-
femur and fibula, we designate fibular landmarks structing the length of fragmentary bones the latter
with “P,” from the Spanish, peroné. approach makes logical sense, especially for land-
The equations we derived relate the distances be- marks defined by muscle attachment sites, since
tween these landmarks to the maximum length of these features are determined by muscle size and
each long bone, measured following the standard action in the context of a complete functioning limb.
definitions (Bass, 1987). To measure the locations of Konigsberg et al. (1998); (also Hens et al., 2000)
landmarks, we place each bone on the osteometric have shown that inverse calibration is appropriate
 ESTIMATING LENGTH OF INCOMPLETE LONG BONES 237
TABLE 3. Definitions of bone landmarks used in estimating bone length, with correspondence to landmarks of Steele (1970)
Landmark Landmark Position
Bone (this study) (Steele, 1970) of bone Definition of landmark
Humerus H0 1 Prone Most proximal point on the head
H1 Supine Most proximal point of the greater tuberosity
H2 Supine Most projecting proximal point on the lesser
tuberosity, along its lateral border
H3 2 Prone Most distal point of circumference of the head
H41 Supine Most distal point of the deltoid tuberosity,
where the two deltoid lines join
H5 3 Prone Proximal margin of the olecranon fossa
H6 4 Prone Distal margin of the olecranon fossa
H7 5 Prone Most distal point of the trochlea
Femur F0 1 Prone Most proximal point on the head
F1 Prone Most proximal point on the greater trochanter
F2 2 Prone Midpoint of the lesser trochanter
F32 Prone Distal limit of smooth bone between the union
of the pectineal line and linea aspera, at
which point the intersection of the lines is
filled with rough bone
F41 3 Prone Most proximal extension of the popliteal
surface at point where the medial and lateral
supracondylar lines become parallel below
the linea aspera
F5 4 Prone Most proximal point on margin of the
intercondylar fossa
F6 5 Prone Most distal point of the medial condyle
Tibia T7 1 Supine Most prominent point on the lateral half of the
lateral condyle
T6 2 Supine Most proximal point of the tibial tuberosity
T51 Supine Point at which the anterior crest crosses the
central axis of tibia, as drawn through the
tibial tuberosity
T41 Prone Nutrient foramen
T31 Prone Point on the popliteal line where it crosses over
the medial angle of the diaphysis
T21 4 Supine Point where the anterior crest crosses over to
the medial border of the shaft above the
medial malleolus
T1 5 Supine Proximal margin of the distal articular surface,
at a point opposite the tip of the medial
malleolus
T0 6 Supine Most distal point of the medial malleolus
Fibula P0 Supine Most proximal point of the head
P1 Supine Most laterally projecting point on the head,
opposite the proximal articulation
P22 Supine Nutrient foramen
P3 Supine Proximal border of the distal articular facet
P4 Supine Most distal point on the lateral malleolus
1
Landmark is not used in any of the equations listed in Table 5. However, it is used in equations listed in Appendix.
2
Landmark is not used in any of the equations generated and deemed acceptable in this study.

to contexts where the specimen is known to belong to archaeological Maya remains) which are likely to be
the same statistical distribution as that from which taller than the forensic Maya sample. Thus, classical
the estimator was developed. But if the specimen is calibration is the preferred method (Konigsberg et
not known to have come from a given distribution, al., 1998). We reject regression equations that show
classical calibration is the preferred method. Equa- r2 ⬍ 0.85, and retain only those equations with high
tions generated by classical calibration will have correlations and highly significant F values (P ⬍
smaller standard errors than would those generated 0.0001).
by inverse calibration when applied to skeletons RESULTS
with bone lengths that are substantially different
from the reference sample mean (Konigsberg et al., Table 4 contains Pearson’s correlation coefficients,
1998). Although inverse calibration might be more together with sample sizes, for the correlation of
appropriate for the narrow context of forensic Maya each possible segment with maximum bone length
skeletons in Guatemala, we hope that the equations in the forensic Maya sample, for males, for females,
may find application in other closely related popu- and for the two sexes pooled. The correlations range
lations (e.g., forensic Ladinos in Central America, or from essentially 0 – 0.99. Approximately half of the
238 L.E. WRIGHT AND M.A. VÁSQUEZ

Fig. 1. Landmarks used to estimate bone length from the humerus (a), femur (b), tibia (c), and fibula (d). Left bones are illustrated,
with ventral aspect at left, and dorsal aspect at right.
 ESTIMATING LENGTH OF INCOMPLETE LONG BONES 239
TABLE 4. Pearson’s r for correlation of individual segment lengths with maximum bone length

Bone and Maya males Maya females Combined sex

Segment r N r N r N
Humerus
H0–1 0.256* 54 0.015* 23 0.179* 77
H0–2 0.185* 57 ⫺0.068* 23 0.368 80
H0–3 0.355 56 0.140* 24 0.578 80
H0–4 0.827 57 0.836 23 0.868 80
H0–5 0.981 58 0.976 24 0.989 82
H0–6 0.991 58 0.984 24 0.994 82
H1–2 0.048* 53 ⫺0.080* 23 0.308 76
H1–3 0.150* 52 0.112* 23 0.448 75
H1–4 0.841 53 0.845 23 0.875 76
H1–5 0.975 54 0.970 23 0.986 80
H1–6 0.985 54 0.976 23 0.991 80
H1–7 0.993 54 0.992 23 0.996 77
H2–3 0.159* 55 0.159* 23 0.258* 78
H2–4 0.785 56 0.873 23 0.830 79
H2–5 0.941 57 0.959 23 0.969 80
H2–6 0.966 57 0.965 23 0.980 80
H2–7 0.977 57 0.976 23 0.986 80
H3–4 0.773 55 0.806 23 0.789 78
H3–5 0.948 56 0.952 24 0.967 80
H3–6 0.974 56 0.953 24 0.979 77
H3–7 0.981 56 0.974 24 0.986 80
H4–5 0.621 57 0.012* 23 0.694 76
H4–6 0.679 57 0.177* 23 0.748 82
H4–7 0.731 57 0.265* 23 0.789 80
H5–6 0.219* 58 0.613 24 0.371 80
H5–7 0.390 58 0.569 24 0.513 82
H6–7 0.284* 58 0.132* 24 0.294 77
Femur
F0–1 0.198* 64 0.491 32 0.268 96
F0–2 0.378 64 0.679 31 0.628 95
F0–3 0.255* 59 0.544 32 0.513 91
F0–4 0.798 64 0.839 32 0.862 96
F0–5 0.984 65 0.986 32 0.988 97
F1–2 0.358 62 0.559 31 0.637 93
F1–3 0.213* 57 0.370* 32 0.459 89
F1–4 0.792 62 0.779 32 0.852 94
F1–5 0.963 63 0.970 32 0.976 95
F1–6 0.981 64 0.984 32 0.988 96
F2–3 0.113 58 0.244 31 0.292 89
F2–4 0.743 62 0.761 31 0.789 93
F2–5 0.926 63 0.956 31 0.944 94
F2–6 0.953 64 0.975 31 0.970 95
F3–4 0.529 58 0.556 32 0.524 90
F3–5 0.703 58 0.803 32 0.717 90
F3–6 0.727 59 0.842 32 0.772 91
F4–5 0.241* 63 0.472 32 0.314 95
F4–6 0.305* 64 0.530 32 0.445 96
F5–6 0.216* 65 0.270* 32 0.450 97
Tibia
T0–1 0.167* 65 0.115* 30 0.294 95
T0–2 0.505 63 0.112* 30 0.525 93
T0–3 0.815 64 0.731 29 0.814 93
T0–4 0.761 65 0.832 30 0.845 95
T0–5 0.853 63 0.808 30 0.883 93
T0–6 0.981 60 0.981 29 0.988 89
T1–2 0.494 63 0.095 30 0.497 93
T1–3 0.792 64 0.722 29 0.792 93
T1–4 0.749 65 0.801 30 0.828 95
T1–5 0.846 63 0.816 30 0.877 93
T1–6 0.973 60 0.975 29 0.982 89
T1–7 0.992 65 0.996 30 0.996 95
T2–3 0.499 63 0.580 29 0.510 92
T2–4 0.390 63 0.652 30 0.539 93
T2–5 0.596 63 0.841 30 0.711 93
T2–6 0.734 60 0.849 29 0.822 89
T2–7 0.774 60 0.879 27 0.859 87
T3–4 ⫺0.179* 64 ⫺0.226* 29 ⫺0.080* 93
T3–5 0.048* 63 0.009* 29 0.124* 92
T3–6 0.302* 60 0.105* 28 0.457 88
T3–7 0.424 64 0.298* 29 0.559 93
T4–5 0.223* 63 0.255* 30 0.224* 93
T4–6 0.419 60 0.614 29 0.536 89

(continued)
240 L.E. WRIGHT AND M.A. VÁSQUEZ
TABLE 4. (Continued)

Bone and Maya males Maya females Combined sex

Segment r N r N r N
T4–7 0.580 65 0.728 30 0.692 95
T5–6 0.257* 60 0.302* 29 0.361 89
T5–7 0.407 63 0.364* 30 0.517 93
T6–7 0.434 60 0.356* 29 0.561 89
Fibula
P0–1 0.079* 45 0.046* 23 0.073* 68
P0–2 0.335* 46 0.110* 24 0.353 70
P0–3 0.989 51 0.993 25 0.992 76
P1–2 0.345* 40 0.165* 23 0.398 63
P1–3 0.964 45 0.984 23 0.979 68
P1–4 0.981 45 0.990 23 0.990 68
P2–3 0.444 46 0.449* 24 0.459 70
P2–4 0.460 46 0.453* 24 0.501 70
P3–4 0.322* 51 0.162* 25 0.538 76
Tibia from fibula
P0–1 0.081* 45 0.061* 23 0.076* 68
P0–2 0.376 46 0.090* 24 0.364 70
P0–3 0.972 51 0.984 25 0.982 76
P1–2 0.401 40 0.150* 23 0.416 63
P1–3 0.947 45 0.972 23 0.969 68
P1–4 0.947 45 0.966 23 0.970 68
P2–3 0.392 46 0.462* 24 0.441 70
P2–4 0.396 46 0.459* 24 0.475 70
P3–4 0.228* 51 0.052* 25 0.474 76

* P ⱖ 0.01.

segments show correlations greater than 0.9. In gen- presents the largest number of successful equations,
eral, longer segments are better correlated with providing 11 equations each for males, females, and
bone length than are shorter segments, as might be sexes combined. For the femur, we accepted five
expected. Many shorter segments show nonsignifi- equations for each sex. The tibia and fibula each
cant correlations with bone length. Segments de- resulted in three equations per sex, as was also the
fined by landmarks that lie on articular architecture case for the estimation of tibial length from fibular
or secondary centers of ossification also show higher measurements. Table 5 also includes the standard
correlations than segments defined by muscle mark- errors of the estimated bone lengths. For most equa-
ings or nutrient foramina. For instance, segments tions, the error is quite small, ranging between
defined in part by H4 (the most distal point on the 2.86 –18.43 mm. Standard errors are smaller for the
deltoid tuberosity) show a lower correlation than combined sex equations, because of the larger sam-
several other humeral segments. Likewise, correla- ple size on which the regressions were based. The
tions are low for segments defined by femoral land- Appendix contains 19 additional equations that
marks F3 and F4, as well as tibial segments T5, T3, have r2 between 0.6 – 0.85. Although these have sig-
and T2. Indeed, segment T3– 4 is negatively (but not nificant F values (P ⬍ 0.0001), they predict bone
significantly) correlated with tibia length. This is length with less success than the equations in Table
largely due to the variable location of landmark T4, 5, as indicated by the lower r2 and larger standard
the nutrient foramen; other tibial segments delim- errors. Because these equations are defined by mus-
ited by the nutrient foramen also show poor corre- cle attachment sites, they should be applied only
lations with bone length. The nutrient foramen on with extreme caution. We rejected 121 equations
the fibula, P2, is also fairly variable in location, as with r2 ⬍ 0.6, and those with standard errors ⬎20
indicated by the low correlations of segments that it mm.
defines. Correlations between fibular segment To evaluate the applicability of the regression
lengths and tibial length are comparable to those for equations in Table 5 on Guatemalan remains of
fibular length. unknown ethnicity, we apply the equations to com-
We generated regression equations for segment plete bones from 36 male skeletons from the military
length vs. total bone length for each of the segments base at El Chal, Petén. Figure 2 illustrates the re-
that showed a statistically significant correlation at lationship between measured bone length and bone
P ⱕ 0.01 (Table 4). In total, we generated some 215 length estimated using a selection of the Maya male
regression equations: 73 for males, 57 for females, regression equations. Estimates derived with all
and 85 for sexes combined. Of these, 75 equations three equations are represented for the tibia and
have r2 ⬎ 0.85, and highly significant F values (P ⬍ fibula. The slopes of regression lines drawn through
0.0001); they are listed in Table 5. Because the ar- these data range from 0.9 –1.1 for all of the equa-
ticular architecture of the humerus lends itself to tions, including those not illustrated in Figure 2.
the definition of several landmarks, the humerus Following Feldesman and Fountain (1996), we use
 ESTIMATING LENGTH OF INCOMPLETE LONG BONES 241
TABLE 5. Regression lines for estimation of total bone length for forensic Maya skeletons1
Regression line r2 N F SE
A. Humerus
Sexes combined
(H0–5) ⫽ ⫺7.561 ⫹ 0.918 * Humerus length 0.978 82 3,567.4 4.46
(H0–6) ⫽ ⫺4.805 ⫹ 0.968 * Humerus length 0.989 82 7,036.4 3.35
(H1–5) ⫽ ⫺8.145 ⫹ 0.903 * Humerus length 0.973 77 2,709.7 5.03
(H1–6) ⫽ ⫺4.803 ⫹ 0.951 * Humerus length 0.982 77 4,180.6 4.27
(H1–7) ⫽ ⫺0.361 ⫹ 0.984 * Humerus length 0.993 77 9,972.8 2.86
(H2–5) ⫽ ⫺4.255 ⫹ 0.848 * Humerus length 0.939 80 1,191.7 7.12
(H2–6) ⫽ ⫺3.072 ⫹ 0.904 * Humerus length 0.961 80 1,916.2 5.98
(H2–7) ⫽ 1.328 ⫹ 0.937 * Humerus length 0.972 80 2,714.7 5.21
(H3–5) ⫽ ⫺7.229 ⫹ 0.808 * Humerus length 0.935 80 1,126.1 6.98
(H3–6) ⫽ ⫺4.192 ⫹ 0.857 * Humerus length 0.958 80 1,760.9 5.92
(H3–7) ⫽ ⫺0.191 ⫹ 0.892 * Humerus length 0.971 80 2,653.0 5.02
Males
(H0–5) ⫽ ⫺8.976 ⫹ 0.923 * Humerus length 0.963 58 1,441.2 7.24
(H0–6) ⫽ ⫺3.008 ⫹ 0.962 * Humerus length 0.983 58 3,200.8 5.06
(H1–5) ⫽ ⫺7.817 ⫹ 0.902 * Humerus length 0.951 54 1,000.6 8.49
(H1–6) ⫽ 1.191 ⫹ 0.931 * Humerus length 0.970 54 1,668.2 6.79
(H1–7) ⫽ 4.032 ⫹ 0.97 * Humerus length 0.986 54 3,783.4 4.70
(H2–5) ⫽ ⫺10.958 ⫹ 0.87 * Humerus length 0.886 57 426.2 12.51
(H2–6) ⫽ ⫺7.981 ⫹ 0.92 * Humerus length 0.933 57 765.8 9.87
(H2–7) ⫽ ⫺5.936 ⫹ 0.961 * Humerus length 0.955 57 1,162.1 8.37
(H3–5) ⫽ ⫺20.641 ⫹ 0.851 * Humerus length 0.899 56 481.2 11.55
(H3–6) ⫽ ⫺14.351 ⫹ 0.89 * Humerus length 0.949 56 999.3 8.37
(H3–7) ⫽ ⫺12.205 ⫹ 0.931 * Humerus length 0.963 56 1,401.0 7.40
Females
(H0–5) ⫽ 6.396 ⫹ 0.866 * Humerus length 0.952 24 437.1 11.24
(H0–6) ⫽ ⫺7.124 ⫹ 0.977 * Humerus length 0.969 24 682.7 10.14
(H1–5) ⫽ 5.387 ⫹ 0.852 * Humerus length 0.941 23 333.7 12.61
(H1–6) ⫽ ⫺4.765 ⫹ 0.95 * Humerus length 0.953 23 426.4 12.44
(H1–7) ⫽ ⫺4.223 ⫹ 0.998 * Humerus length 0.983 23 1,222.4 7.72
(H2–5) ⫽ ⫺9.534 ⫹ 0.869 * Humerus length 0.919 23 238.2 15.24
(H2–6) ⫽ ⫺19.686 ⫹ 0.967 * Humerus length 0.932 23 286.0 15.47
(H2–7) ⫽ ⫺19.144 ⫹ 1.016 * Humerus length 0.952 23 414.3 13.50
(H3–5) ⫽ ⫺13.503 ⫹ 0.834 * Humerus length 0.905 24 210.7 15.59
(H3–6) ⫽ ⫺27.023 ⫹ 0.945 * Humerus length 0.907 24 215.4 17.48
(H3–7) ⫽ ⫺19.899 ⫹ 0.968 * Humerus length 0.949 24 407.4 13.02
B. Femur
Sexes combined
(F0–5) ⫽ ⫺4.716 ⫹ 0.932 * Femur max. length 0.977 97 3,999.0 6.01
(F1–5) ⫽ ⫺2.152 ⫹ 0.890 * Femur max. length 0.952 95 1,852.7 8.44
(F1–6) ⫽ 0.138 ⫹ 0.960 * Femur max. length 0.976 96 3,893.0 6.28
(F2–5) ⫽ ⫺4.139 ⫹ 0.757 * Femur max. length 0.890 94 751.8 11.25
(F2–6) ⫽ ⫺3.780 ⫹ 0.832 * Femur max. length 0.941 95 1,484.7 8.81
Males
(F0–5) ⫽ ⫺18.842 ⫹ 0.964 * Femur max. length 0.968 65 1,899.8 9.24
(F1–5) ⫽ ⫺17.956 ⫹ 0.928 * Femur max. length 0.928 63 789.7 13.80
(F1–6) ⫽ ⫺1.231 ⫹ 0.965 * Femur max. length 0.963 64 1,624.7 10.01
(F2–5) ⫽ ⫺42.396 ⫹ 0.846 * Femur max. length 0.858 63 367.6 18.43
(F2–6) ⫽ ⫺27.202 ⫹ 0.887 * Femur max. length 0.909 64 617.9 14.91
Females
(F0–5) ⫽ ⫺13.195 ⫹ 0.957 * Femur max. length 0.973 32 1,076.5 11.28
(F1–5) ⫽ 9.809 ⫹ 0.861 * Femur max. length 0.941 32 478.7 15.22
(F1–6) ⫽ 22.378 ⫹ 0.901 * Femur max. length 0.968 32 911.9 11.55
(F2–5) ⫽ ⫺6.179 ⫹ 0.769 * Femur max. length 0.915 31 311.5 16.78
(F2–6) ⫽ 2.260 ⫹ 0.820 * Femur max. length 0.948 31 557.2 13.39
C. Tibia
Sexes combined
(T0–6) ⫽ 10.807 ⫹ 0.904 * Tibia length 0.976 89 3,551.3 5.04
(T1–6) ⫽ 8.847 ⫹ 0.873 * Tibia length 0.965 89 2,402.1 5.92
(T1–7) ⫽ ⫺2.770 ⫹ 0.972 * Tibia length 0.991 95 10,473.2 3.16
Males
(T0–6) ⫽ 7.407 ⫹ 0.914 * Tibia length 0.963 60 1,504.2 8.04
(T1–6) ⫽ 2.803 ⫹ 0.890 * Tibia length 0.948 60 1,047.4 9.39
(T1–7) ⫽ ⫺5.478 ⫹ 0.979 * Tibia length 0.985 65 4,042.0 5.26
Females
(T0–6) ⫽ 3.281 ⫹ 0.930 * Tibia length 0.962 29 684.6 11.09
(T1–6) ⫽ ⫺5.925 ⫹ 0.923 * Tibia length 0.951 29 521.1 12.62
(T1–7) ⫽ ⫺8.034 ⫹ 0.990 * Tibia length 0.992 30 3,488.0 5.24
D. Fibula
Sexes combined
(P0–3) ⫽ 2.877 ⫹ 0.924 * Fibula length 0.984 76 4,463.6 4.55
(P1–3) ⫽ ⫺5.000 ⫹ 0.909 * Fibula length 0.958 68 1,504.4 7.73
(P1–4) ⫽ ⫺9.305 ⫹ 0.990 * Fibula length 0.980 68 3,173.4 5.79

(continued)
242 L.E. WRIGHT AND M.A. VÁSQUEZ
TABLE 5. (Continued)
Regression line r2 N F SE
Males
(P0–3) ⫽ ⫺6.583 ⫹ 0.951 * Fibula length 0.977 51 2,097.4 7.01
(P1–3) ⫽ ⫺13.376 ⫹ 0.933 * Fibula length 0.930 45 573.4 13.21
(P1–4) ⫽ ⫺7.512 ⫹ 0.984 * Fibula length 0.962 45 1,089.5 10.11
Females
(P0–3) ⫽ ⫺13.445 ⫹ 0.981 * Fibula length 0.986 25 1,569.9 7.66
(P1–3) ⫽ ⫺23.848 ⫹ 0.974 * Fibula length 0.968 23 626.6 12.09
(P1–4) ⫽ ⫺10.529 ⫹ 0.993 * Fibula length 0.979 23 998.0 9.77
E. Tibia from fibula
Sexes combined
(P0–3) ⫽ ⫺6.216 ⫹ 0.941 * Tibia length 0.965 76 2,033.5 6.94
(P1–3) ⫽ ⫺11.450 ⫹ 0.918 * Tibia length 0.938 68 1,003.4 9.67
(P1–4) ⫽ ⫺13.239 ⫹ 0.990 * Tibia length 0.942 68 1,067.6 10.11
Males
(P0–3) ⫽ ⫺20.435 ⫹ 0.981 * Tibia length 0.944 51 826.5 11.65
(P1–3) ⫽ ⫺24.067 ⫹ 0.954 * Tibia length 0.896 45 370.3 16.99
(P1–4) ⫽ ⫺13.214 ⫹ 0.990 * Tibia length 0.897 45 373.7 17.56
Females
(P0–3) ⫽ ⫺9.256 ⫹ 0.953 * Tibia length 0.967 25 684.1 11.45
(P1–3) ⫽ ⫺15.667 ⫹ 0.934 * Tibia length 0.945 23 362.2 15.47
(P1–4) ⫽ 1.646 ⫹ 0.941 * Tibia length 0.932 23 290.0 17.42
1
All measurements, estimated bone lengths, and standard errors are in mm. max., maximum.

the mean absolute deviation (MAD)1 and the mean the femoral equations show statistically significant
squared error (MSE)2 for the difference between mean differences, and larger values of MAD and
measured and estimated bone length to evaluate the MSE than the humeral equations, the MAD are
accuracy of the Maya male and combined sex equa- substantially smaller than the standard error of the
tions on the El Chal remains. Smaller MAD and estimating equations. The Maya equations perform
MSE values indicate better accuracy of the equa- less well on El Chal tibiae, and tend to overestimate
tions. Table 6 contains the r2, sample size, MAD, tibia length. Tibial length estimates using the T1– 6
and MSE, as well as the mean differences and prob- equations differ significantly from measured tibial
ability for a paired t-test between estimated and lengths. For the fibula, none of the equations pro-
measured lengths for each equation. Positive mean duced estimates that differed significantly from
differences indicate a tendency to underestimate measured fibular length; MAD and MSE are very
bone length using a given equation, while a negative small. Estimates of tibial length from fibular mea-
mean difference indicates that the regression equa- surements at El Chal appear to work better than
tion overestimates bone length. Given the large estimates based on tibial measurements. Overall,
number of equations for which we calculated t-tests, the absolute differences between measured and es-
we consider these differences to be significant only timated lengths of the El Chal long bones are quite
when P ⬍ 0.001, to minimize the likelihood of type I small, the largest MAD being only 6.4 mm. Despite
errors. the significance of some mean differences, we believe
For all bones, estimated length is highly corre- that the forensic Maya equations estimate bone
lated with measured length in the El Chal remains, length with sufficient accuracy for the El Chal sam-
with values of r2 above 0.84 for all of the equations in ple, and can be applied in diverse forensic contexts
Table 5. For the humerus, mean absolute deviations in Guatemala with confidence.
are less than 3 mm and are well within the standard We used the equations of Steele (1970), Steele and
error of the estimating equations, and estimated McKern (1969), also Steele and Bramblett (1988),
humeral lengths do not differ significantly from and Jacobs (1992) to examine the applicability of
measured length for any equation. However, all other standards for long bone length estimation on
mean differences are positive, indicating slight un- Maya remains. We limit our comparison to Steele’s
derestimation of humerus length. Although a few of equations based on segment combinations that were
shown to be highly correlated (r ⱖ 0.9) with bone
1
length in the original studies, in order to assess
Mean absolute deviation is calculated as follows (after Feldesman
equations of similar accuracy to those generated
冘
and Fountain, 1996):
兩共estimated ⫺ measured兲兩
here. Because we did not measure Steele’s tibia
MAD ⫽ landmark 3, direct comparison with many of the
n
tibia equations of Steele (1970) and of Jacobs (1992)
2
Mean squared error is calculated as follows (after Feldesman and is not possible. Single segment regressions for the
tibia have P ⬍ 0.9, so we compare only equations for
冘
Fountain, 1996):
共estimated ⫺ measured兲2 humerus length and femur length. Although it may
MSE ⫽
n seem obvious that these equations should be less
 ESTIMATING LENGTH OF INCOMPLETE LONG BONES 243

Fig. 2. Comparison of measured and estimated bone length for El Chal male skeletons, with lengths estimated using male forensic
Maya regression equations.

accurate on Maya remains than the Maya equations lengths. Yet several equations give estimated
are, this exercise is a useful demonstration of the lengths that are strongly correlated with measured
magnitude of population differences. length, for which the regression lines pass through
For the forensic Maya males and females, Figure zero and have slopes close to one.
3 illustrates the relationship of measured humerus For the humerus, the white male segment 1 ⫹ 2 ⫹
length to humerus length estimated using the “best” 3 equation appears to give a fairly good estimate of
(segment 1 ⫹ 2 ⫹ 3) and “worst” (segment 2) of the Mayan male humerus length (Fig. 3a), but the Black
humerus equations of Steele (1970) as judged by male 1 ⫹ 2 ⫹ 3 equation consistently overestimates
MAD and MSE. Table 7 lists the r2, sample sizes, humerus length, while the Mississippian male 1 ⫹
MAD, MSE, mean differences, and paired t-test 2 ⫹ 3 equation grossly underestimates humerus
probabilities for all the regression lines that we com- length. White and Black male equations for seg-
pared. Figure 4 and Table 8 provide comparable ments 1 ⫹ 2, 2 ⫹ 3, and 2 ⫹ 3 ⫹ 4 overestimate
data for the femur. On the Maya skeletons, the humerus length consistently, while the Mississip-
equations of Steele’s give high MAD and MSE, and pian equations substantially underestimate hu-
almost all of the lengths estimated using published merus length (Table 7). These population differences
equations are significantly different from measured are less evident for Maya females, for whom many of
244 L.E. WRIGHT AND M.A. VÁSQUEZ
TABLE 6. Comparison of measured bone length to estimated bone length for male skeletons from El Chal
Equation r2 N Mean difference P MAD MSE
Humerus
Maya male H0–5 0.97 32 0.9 0.0270 1.8 5.5
Maya male H0–6 0.98 32 0.1 0.6571 1.1 2.1
Maya male H1–5 0.94 30 1.5 0.0172 2.7 11.4
Maya male H1–6 0.95 30 0.6 0.2769 2.1 7.3
Maya male H1–7 0.98 30 0.6 0.0724 1.4 3.6
Maya male H2–5 0.93 30 1.8 0.0076 2.5 9.2
Maya male H2–6 0.95 30 1.3 0.0232 2.1 6.8
Maya male H2–7 0.96 30 1.0 0.0321 2.0 6.4
Maya male H3–5 0.93 31 0.9 0.1367 1.9 5.6
Maya male H3–6 0.96 31 0.4 0.3401 2.9 14.5
Maya male H3–7 0.96 31 0.3 0.4527 2.6 12.0
Maya combined sex H0–5 0.97 32 0.8 0.0486 1.7 5.3
Maya combined sex H0–6 0.98 32 0.1 0.6146 1.1 2.2
Maya combined sex H1–5 0.94 30 1.4 0.0194 2.7 11.3
Maya combined sex H1–6 0.95 30 0.6 0.2475 2.1 7.4
Maya combined sex H1–7 0.98 30 0.5 0.1897 1.4 3.6
Maya combined sex H2–5 0.93 30 1.9 0.0062 3.0 15.6
Maya combined sex H2–6 0.95 30 1.4 0.0155 2.5 9.5
Maya combined sex H2–7 0.96 30 1.1 0.0296 2.2 7.0
Maya combined sex H3–5 0.93 31 1.5 0.0300 2.8 15.4
Maya combined sex H3–6 0.96 31 0.7 0.1623 2.1 7.3
Maya combined sex H3–7 0.96 31 0.6 0.2011 2.2 6.8
Femur
Maya male F0–5 0.99 34 1.3 0.0006 1.9 5.6
Maya male F1–5 0.95 32 0.1 0.9253 3.2 16.0
Maya male F1–6 0.97 32 ⫺1.3 0.0332 5.4 38.0
Maya male F2–5 0.84 34 ⫺1.4 0.2774 5.9 52.7
Maya male F2–6 0.88 34 ⫺2.5 0.0238 5.2 42.0
Maya combined sex F0–5 0.99 34 1.8 ⬍0.0001 2.1 7.0
Maya combined sex F1–5 0.95 32 ⫺0.4 0.5610 3.2 16.6
Maya combined sex F1–6 0.97 32 ⫺2.1 0.0011 3.1 14.6
Maya combined sex F2–5 0.84 34 ⫺1.3 0.3275 6.4 60.8
Maya combined sex F2–6 0.88 34 ⫺2.8 0.0148 5.5 46.4
Tibia
Maya male T0–6 0.95 29 ⫺2.4 0.0017 3.7 18.8
Maya male T1–6 0.93 29 ⫺4.0 ⬍0.0001 5.0 36.4
Maya male T1–7 0.98 29 ⫺1.2 0.0067 2.0 6.1
Maya combined sex T0–6 0.95 29 ⫺2.4 0.0021 3.7 19.1
Maya combined sex T1–6 0.93 29 ⫺4.0 0.0001 5.1 37.3
Maya combined sex T1–7 0.98 29 ⫺1.1 0.0128 1.9 6.0
Fibula
Maya male P0–3 0.99 15 ⫺1.8 0.0025 1.8 6.3
Maya male P1–3 0.95 12 ⫺1.3 0.2112 3.2 12.5
Maya male P1–4 0.96 12 0.5 0.5565 2.6 8.4
Maya combined sex P0–3 0.99 15 ⫺1.7 0.0066 1.9 6.8
Maya combined sex P1–3 0.95 12 ⫺1.2 0.2480 2.6 10.2
Maya combined sex P1–4 0.96 12 0.8 0.3756 2.7 8.8
Tibia from fibula
Maya male P0–3 tibia 0.94 13 ⫺3.6 0.0067 4.8 27.9
Maya male P1–3 tibia 0.91 10 ⫺2.9 0.0830 5.0 29.3
Maya male P1–4 tibia 0.93 10 ⫺1.3 0.3220 3.4 16.5
Maya combined sex P0–3 tibia 0.94 13 ⫺3.4 0.0136 4.8 27.0
Maya combined sex P1–3 tibia 0.91 10 ⫺2.8 0.1021 4.8 27.7
Maya combined sex P1–4 tibia 0.93 10 ⫺1.4 0.3134 3.4 16.5

the equations give comparably good results, but tend ences, and paired t-test probabilities for all of the
toward slightly overestimating humerus length (Fig. femoral regression lines that we compared. The
3b, Table 7). All of the equations for segment 2 equations for segments 1 ⫹ 2 perform very poorly on
overestimate humerus length substantially for both Maya femora. They show very low correlations, low
Maya males and females, especially for individuals slopes, large MAD, and large MSE. The white and
with shorter humeri (Fig. 3c,d). The segment 2 equa- Black 1 ⫹ 2 equations dramatically underestimate
tions stand out as the least successful predictors of maximum length for most femora, while the Missis-
humerus length on the Mayan remains. sippian and European equations overestimate max-
Figure 4 illustrates the comparison of measured imum length. The nonsignificant mean difference
maximum femur length with maximum femur and small MAD and MSE shown for males using the
length, estimated using the 1 ⫹ 2 and 1 ⫹ 2 ⫹ 3 Mississippian 1 ⫹ 2 equation is misleading; it is an
equations of Steele (1970), Steele and Bramblett artifact of the low slope and sample mean. The equa-
(1988), and Jacobs (1992) on the forensic Maya sam- tions based on femoral segments 1 ⫹ 2 ⫹ 3 perform
ple. Table 8 lists the r2, MAD, MSE, mean differ- much better. Although they too show statistically
 ESTIMATING LENGTH OF INCOMPLETE LONG BONES 245

Fig. 3. Comparison of measured and estimated humerus length for forensic Maya skeletons, with lengths estimated using
regression equations of Steele (1970) and Steele and Bramblett (1988, p. 166).

significant mean differences from measured femur deviations than the white, Black, or European equa-
length, these are smaller than for most other equa- tions.
tions. For both Maya males and females, the white, The inaccuracies in estimating bone length using
Black, and European equations are coincident, and the published equations must be due to different
slightly overestimate femoral length, while the Mis- proportions of the bone in the reference populations
sissippian 1 ⫹ 2 ⫹ 3 equations underestimate fem- from which the published equations were derived.
oral length consistently. Femoral length estimates Table 9 contains the mean proportions of each of
based on equations using segments 2 ⫹ 3 and 2 ⫹ Steele’s segments in the Maya forensic sample, to-
3 ⫹ 4 also show high correlations with measured gether with the mean proportions of the segments in
femur length and small MAD. Yet the equations both the reference samples of Steele (1970), Steele
based on segments 2 ⫹ 3 tend to overestimate femur and McKern (1969), and in the prehistoric European
length slightly, while the equations based on seg- sample of Jacobs (1992). For the humerus, segment
ments 2 ⫹ 3 ⫹ 4 tend to underestimate femur proportions are quite similar among the Terry
length. The Mississippian male 2 ⫹ 3 and Mississip- whites, Blacks, and the Maya. The Mississippians
pian female 1 ⫹ 2 ⫹ 3 equations perform well, but in have a slightly longer humerus segment 3. However,
general, the Mississippian equations show larger the generally similar proportions imply that the in-
246 L.E. WRIGHT AND M.A. VÁSQUEZ
TABLE 7. Comparison of measured humerus length to estimated humerus length of Maya skeletons,
using equations of Steele (1970) and Steele and Bramblett (1988, p. 166)
Equation r2 N Mean difference P MAD MSE
Males
White male 2 0.899 56 ⫺10.6 ⬍0.0001 15.5 1,720.2
White male 1 ⫹ 2 0.962 56 ⫺4.8 ⬍0.0001 10.0 1,622.5
White male 2 ⫹ 3 0.943 56 ⫺5.5 ⬍0.0001 10.7 1,633.4
White male 1 ⫹ 2 ⫹ 3 0.984 56 ⫺0.9 0.0003 6.7 1,597.0
White male 2 ⫹ 3 ⫹ 4 0.956 56 ⫺2.9 ⬍0.0001 8.2 1,609.4
Black male 2 0.899 56 ⫺12.4 ⬍0.0001 17.2 1,760.1
Black male 1 ⫹ 2 0.962 56 ⫺6.0 ⬍0.0001 11.1 1,635.7
Black male 2 ⫹ 3 0.941 56 ⫺7.0 ⬍0.0001 12.0 1,650.8
Black male 1 ⫹ 2 ⫹ 3 0.984 56 ⫺4.4 ⬍0.0001 9.5 1,615.5
Black male 2 ⫹ 3 ⫹ 4 0.955 56 ⫺5.8 ⬍0.0001 10.8 1,633.5
Mississippian male 2 0.899 56 ⫺10.8 ⬍0.0001 15.6 1,723.5
Mississippian male 1 ⫹ 2 0.964 56 10.0 ⬍0.0001 14.8 1,695.4
Mississippian male 2 ⫹ 3 0.949 56 16.4 ⬍0.0001 21.1 1,862.7
Mississippian male 1 ⫹ 2 ⫹ 3 0.984 56 7.2 ⬍0.0001 12.2 1,647.0
Mississippian male 2 ⫹ 3 ⫹ 4 0.968 56 7.4 ⬍0.0001 12.6 1,652.3
Females
White female 2 0.906 24 ⫺10.8 ⬍0.0001 10.3 125.2
White female 1 ⫹ 2 0.924 24 0.7 0.3160 2.4 10.5
White female 2 ⫹ 3 0.906 24 ⫺3.2 0.0004 4.1 23.1
White female 1 ⫹ 2 ⫹ 3 0.980 24 2.1 0.0003 2.2 9.5
White female 2 ⫹ 3 ⫹ 4 0.968 24 ⫺3.6 0.0015 2.6 13.3
Black female 2 0.906 24 ⫺6.5 ⬍0.0001 6.4 53.9
Black female 1 ⫹ 2 0.939 24 ⫺1.1 0.0820 2.3 9.1
Black female 2 ⫹ 3 0.906 24 ⫺5.3 ⬍0.0001 5.8 41.5
Black female 1 ⫹ 2 ⫹ 3 0.968 24 ⫺2.4 ⬍0.0001 2.7 9.8
Black female 2 ⫹ 3 ⫹ 4 0.945 24 ⫺2.5 0.0004 2.9 13.9
Mississippian female 2 0.906 24 ⫺11.8 ⬍0.0001 11.3 149.4
Mississippian female 1 ⫹ 2 0.931 24 ⫺3.7 ⬍0.0001 4.1 22.9
Mississippian female 2 ⫹ 3 0.904 24 0.1 0.9066 2.5 13.2
Mississippian female 1 ⫹ 2 ⫹ 3 0.968 24 1.0 0.0324 1.8 5.0
Mississippian female 2 ⫹ 3 ⫹ 4 0.943 24 1.7 0.0125 2.3 10.9

accuracies of the humerus equations on Maya re- tions would also provide unreliable estimates of
mains are due more to the absolute difference in Maya tibia length.
bone size than in the relative length of the segments.
DISCUSSION
For humerus segment 2 in both sexes, this is sug-
gested by the lower slope of the regression (Fig. The results of this study indicate that long bone
3c,d), which illustrates that the equation overesti- length can be estimated from incomplete bones with
mates the length of shorter humeri more than it considerable accuracy using longitudinal measure-
does for longer humeri. This may simply be due to ments, and provides strong support for Steele and
the fact that the mean length of Maya humeri is McKern’s (1969) methodological approach. Standard
substantially less than the mean length of humeri in errors for the forensic Maya regression equations
any of the reference series. defined here are slightly larger than those of the
Table 9 illustrates that Maya femora have a pro- better-performing equations of Steele (1970), in part
portionately shorter segment 2 and longer segment due to our use of classical calibration rather than
3 than the white, Black, and Mississippian series, inverse regression. The broader standard errors for
but the prehistoric European series have a still the Maya female equations are due to the smaller
shorter segment 2 and longer segment 3 than the sample size of Maya females that we were able to
Maya. These differences undoubtedly account for measure.
the poor performance of the segment 1 ⫹ 2 equations Like Steele (1970), also Steele and McKern (1969),
on Maya remains. Equations that incorporate both we found that landmarks defined by joint architec-
segments 2 and 3 perform better because these dif- ture and secondary centers of ossification were eas-
ferences of proportion are minimized by including ier to identify than those identified by muscle mark-
both in the calculation. ings. Moreover, segments defined by articular
Although it was not possible to estimate Maya landmarks are more highly correlated with bone
tibia length using the equations of Steele (1970) or length than are segments defined by muscle attach-
Jacobs (1992) because we did not measure landmark ment sites. Intrapopulation variation in muscle
3, the proportions of the segments can be compared mass and activity presumably accounts for this
by adding segments 2 and 3 together. For both males lower correlation. Therefore, we believe that many
and females, Maya tibia have shorter segments 1 of the regression equations that use these land-
and 2 ⫹ 3, but a longer segment 4 than in other marks cannot be used with confidence. Where pos-
reference series. This suggests that the tibia equa- sible, we recommend that bone length estimates be
 ESTIMATING LENGTH OF INCOMPLETE LONG BONES 247

Fig. 4. Comparison of measured and estimated femur length for forensic Maya skeletons, with lengths estimated using regression
equations of Steele (1970), Steele and Bramblett (1988, p. 230), and Jacobs (1992).

limited to those equations presented in Table 5. The also found difficult to pinpoint. Because these points
regression equations given in the Appendix should also show greater intrapopulation variation than
be used only with extreme caution; investigators articular points, uncertainty in identifying these
who wish to apply them should first evaluate the points probably contributes to perceptions of inaccu-
accuracy of the equations on more complete skele- racy of the equations overall. Together these factors
tons from the same skeletal series. might contribute to an exaggerated bias against the
Critiques of Steele’s method have focused on the method. For instance, Simmons et al. (1990; p 629)
difficulty that some investigators have had in iden- cite comments by Steele (1970; p 87) regarding a
tifying landmarks, and on inaccuracy in bone length landmark that he discarded,3 to support their claim
estimates (Bass, 1987; Brooks et al., 1990; Simmons
et al., 1990). We discarded only one of Steele’s points
that we could not identify with confidence (tibia 3
This landmark, defined as the point of narrowest anterior-poste-
landmark 3). We suspect that the difficulties other rior diameter of the humerus proximal to the deltoid tuberosity, was
proposed by Steele and McKern (1969), but discarded by Steele (1970).
investigators may have encountered with the Modified versions of the 1969 regression equations for Mississippians
method center on locating those points that are de- that do not use this landmark were published by Steele and Bramblett
fined by muscle attachment sites, some of which we (1988, p. 166).
248 L.E. WRIGHT AND M.A. VÁSQUEZ
TABLE 8. Comparison of measured femur length to estimated femur length on Maya skeletons,
using equations of Steele (1970), Steele and Bramblett (1988, p. 230), and Jacobs (1992)
Equation r2 N Mean difference P MAD MSE
Males
White male 1 ⫹ 2 0.640 62 25.2 ⬍0.0001 25.1 780.3
White male 2 ⫹ 3 0.852 61 ⫺9.2 ⬍0.0001 16.3 2,899.9
White male 1 ⫹ 2 ⫹ 3 0.966 61 ⫺5.5 ⬍0.0001 11.9 2,791.7
White male 2 ⫹ 3 ⫹ 4 0.899 61 0.5 0.5420 11.7 2,791.5
Black male 1 ⫹ 2 0.640 62 39.7 ⬍0.0001 39.2 1,753.2
Black male 2 ⫹ 3 0.856 61 ⫺10.8 ⬍0.0001 17.3 2,924.2
Black male 1 ⫹ 2 ⫹ 3 0.968 61 ⫺4.1 ⬍0.0001 10.7 2,777.9
Black male 2 ⫹ 3 ⫹ 4 0.899 61 3.7 ⬍0.0001 12.5 2,802.9
Mississippian male 1 ⫹ 2 0.623 62 ⫺1.4 0.3767 9.8 146.3
Mississippian male 2 ⫹ 3 0.861 61 ⫺4.0 0.0003 13.4 2,828.8
Mississippian male 1 ⫹ 2 ⫹ 3 0.968 61 13.1 ⬍0.0001 19.1 2,926.6
Mississippian male 2 ⫹ 3 ⫹ 4 0.908 61 14.7 ⬍0.0001 20.7 2,992.5
European male 1 ⫹ 2 0.554 62 ⫺31.5 ⬍0.0001 30.8 1,130.9
European male 2 ⫹ 3 0.861 61 ⫺4.5 0.0001 13.7 2,837.8
European male 1 ⫹ 2 ⫹ 3 0.968 61 ⫺6.1 ⬍0.0001 12.4 2,798.2
European male 2 ⫹ 3 ⫹ 4 0.909 61 5.9 ⬍0.0001 13.5 2,824.6
Females
White female 1 ⫹ 2 0.736 31 30.1 ⬍0.0001 30.1 1,008.8
White female 2 ⫹ 3 0.908 31 ⫺0.7 0.5505 4.6 36.2
White female 1 ⫹ 2 ⫹ 3 0.968 31 ⫺4.2 ⬍0.0001 4.7 29.3
White female 2 ⫹ 3 ⫹ 4 0.939 31 3.5 0.0006 5.0 36.6
Black female 1 ⫹ 2 0.769 31 24.6 ⬍0.0001 24.6 693.5
Black female 2 ⫹ 3 0.914 31 ⫺10.9 ⬍0.0001 10.9 153.0
Black female 1 ⫹ 2 ⫹ 3 0.966 31 ⫺7.1 ⬍0.0001 7.2 64.1
Black female 2 ⫹ 3 ⫹ 4 0.924 31 7.9 ⬍0.0001 8.1 91.2
Mississippian female 1 ⫹ 2 0.783 31 ⫺13.8 ⬍0.0001 14.7 281.3
Mississippian female 2 ⫹ 3 0.914 31 ⫺10.0 ⬍0.0001 10.2 140.9
Mississippian female 1 ⫹ 2 ⫹ 3 0.968 31 3.4 ⬍0.0001 3.9 24.0
Mississippian female 2 ⫹ 3 ⫹ 4 0.943 31 1.0 0.3702 4.8 33.9
European female 1 ⫹ 2 0.787 31 ⫺31.8 ⬍0.0001 31.8 1,100.5
European female 2 ⫹ 3 0.912 31 ⫺2.6 0.0237 4.9 43.6
European female 1 ⫹ 2 ⫹ 3 0.969 31 ⫺3.1 0.0001 4.2 24.7
European female 2 ⫹ 3 ⫹ 4 0.939 31 4.6 ⬍0.0001 5.5 45.2

that the landmarks he retained were impossible to marginal articular bone be present in order to apply
locate. Although Simmons et al. (1990) suggested the equations. Thus, they will not permit estimation
that the method is “plagued” by difficulty and there- of extremely fragmentary remains. Nonetheless, the
fore is seldom used, recall that Brooks et al. (1990) equations can be used to estimate the length of
surveyed a small number of forensic investigators bones with broken or eroded articular surfaces, such
who do use the method and who found it to have as femora that lack the head and neck or condyles.
considerable accuracy. Standard errors of the equations described in this
In contrast, we found the points defined by artic- study are smaller than those obtained by Simmons
ular architecture and secondary centers of ossifica- et al. (1990) for transverse measurements of small
tion to be extremely easy to identify. Greater reli- fragments of the femur. Holland (1992) generated
ance on these articular landmarks, coupled with the equations that relate proximal tibial measures to
linear regression approach used here (vs. Steele’s stature. Standard errors for his equations are com-
multiple regression approach) may improve the ease parable to the regressions of Simmons et al. (1990)
of use and accuracy of longitudinal measurements against stature for femora, which indicates that re-
for bone length estimation. By using the distance gression of proximal tibia measures on tibia length
between pairs of landmarks in linear regressions on would also be larger than the standard errors ob-
bone length, we eliminate error that would be intro- tained here for longitudinal measurements. Thus, if
duced by imprecision in locating intermediate land- sufficient length of the diaphysis is available, esti-
marks, as well as error due to inter- and intrapopu- mates of bone length from longitudinal measure-
lation variation in the location of intermediate ments should be superior to those from transverse
landmarks. That we found estimates of Maya bone measurements. Of course, the longitudinal and
length using the equations of Steele (1970) to be transverse measurement regressions should be seen
more accurate for equations based on longer seg- as complementary methods.
ments than those based on shorter segments pro- We observed that estimates of bone length using
vides some support for this argument. the equations of Steele (1970), also Steele and Bram-
By discarding landmarks defined by muscle at- blett (1988), and Jacobs (1992) for the Maya remains
tachment sites, however, the length of bone required often differed significantly from measured bone
to obtain an estimate becomes larger. Our equations lengths. This finding confirms the observation by
demand that the entire diaphysis as well as some Jacobs (1992) that significant interpopulation differ-
 ESTIMATING LENGTH OF INCOMPLETE LONG BONES 249
TABLE 9. Proportions of long bone segments in forensic Maya sample, and in comparative series1

Segment Segment Forensic Maya Proportion of total bone length

Bone (Steele, 1970) (this study) Mean SD N Maya Terry White Terry Black Mississippian European
Males
Humerus Total length Total length 297.4 14.0 58 1.00 1.00 1.00 1.00
1 H0–3 32.8 2.7 56 0.11 0.11 0.11 0.10
2 H3–5 232.4 12.0 56 0.78 0.77 0.78 0.77
3 H5–6 17.7 2.4 58 0.06 0.06 0.06 0.08
4 H6–7 14.2 1.8 58 0.05 0.05 0.05 0.04
Femur Max length Max length 417.6 20.0 66 1.00 1.00 1.00 1.00 1.00
1 F0–2 72.7 6.2 64 0.17 0.17 0.17 0.16 0.16
2 F2–4 207.8 19.0 62 0.50 0.59 0.59 0.56 0.47
3 F4–5 102.2 13.0 63 0.24 0.15 0.16 0.19 0.28
4 F5–6 32.3 3.6 65 0.08 0.08 0.08 0.08 0.09
Tibia Total length Max length 340.9 18.0 65 1.00 1.00 1.00 1.00 1.00
1 T6–7 22.1 3.5 60 0.06 0.09 0.08 0.07 0.07
2⫹3 T2–6 202.2 14.0 60 0.59 0.62 0.62 0.65 0.62
4 T1–2 104.7 11.0 63 0.31 0.27 0.28 0.25 0.26
5 T0–1 12.5 2.2 65 0.04 0.04 0.04 0.04 0.04
Females
Humerus Total length Total length 271.2 12.0 24 1.00 1.00 1.00 1.00
1 H0–3 28.5 2.7 24 0.11 0.11 0.11 0.10
2 H3–5 212.6 10.0 24 0.78 0.78 0.78 0.77
3 H5–6 16.5 2.2 24 0.06 0.06 0.06 0.08
4 H6–7 13.5 2.1 24 0.05 0.05 0.05 0.04
Femur Max length Max length 386.3 21.0 32 1.00 1.00 1.00 1.00 1.00
1 F0–2 65.3 5.2 31 0.17 0.17 0.16 0.16 0.16
2 F2–4 191.3 15.0 31 0.50 0.60 0.60 0.57 0.46
3 F4–5 99.4 11.0 32 0.26 0.15 0.16 0.20 0.30
4 F5–6 28.2 3.2 32 0.07 0.08 0.08 0.08 0.08
Tibia Max length Max length 312.3 16.0 30 1.00 1.00 1.00 1.00 1.00
1 T6–7 18.6 3.2 29 0.06 0.08 0.08 0.07 0.07
2⫹3 T2–6 185.1 17.0 29 0.59 0.64 0.65 0.66 0.62
4 T1–2 97.2 8.6 30 0.31 0.25 0.26 0.25 0.26
5 T0–1 11.2 1.5 30 0.04 0.04 0.04 0.04 0.04
1
Data for comparative series are from Steele (1970), Steele and McKern (1969), and Jacobs (1992). All measurements are in mm. Max,
maximum.

ences exist in the proportions of segments. Much as far removed from the mean stature of the series on
he found that several of Steele’s (1970) White equa- which these equations were derived to permit an
tions performed poorly on Mesolithic Europeans, we unbiased estimate of bone length. We suggest that
observed little correlation between the success of bone proportions are also significantly constrained
Steele’s equations and any a priori expectations by body size. Regardless of whether the segment
about affinity between the reference populations proportion differences are due to activity differences
and our Maya sample. The Mississippian equations and/or allometric considerations, it is clear that es-
often estimate Maya bone length with greater error timates of bone length of Maya remains cannot be
than the Black and white equations, despite com- made with confidence using the majority of the pre-
mon Native American ancestry, shared maize agri- viously published equations.
cultural subsistence practices, and a similarly warm The equations we have provided here are best
climate. Jacobs (1992) proposed that differences in suited to the specific forensic context of Guatemala,
muscle activity may explain the variation, and sug- and presumably adjacent Central American nations.
gested that data be collected on populations with The success of the El Chal tests demonstrates that
diverse behavioral adaptations. Indeed, we note that the equations perform adequately on the modern
the most evident differences in segment proportion Guatemalan population, even when Maya ethnicity
illustrated in Table 9 are those determined by the is not known. We hope that they will aid stature
position of landmarks defined by muscle attachment estimation and the identification of forensic remains
points. However, if we limit bone length estimation in the ongoing human rights investigations in this
to those equations defined in Table 5, which exclude region. We also expect that the equations will per-
most points defined by muscle action, these popula- form well on archaeological human remains of the
tion differences in proportion may be less problem- Maya and other Mesoamerican cultures, many of
atic. whom were taller than are the modern Maya. Addi-
While activity differences likely contribute to tional research might be directed at collecting mea-
these interpopulation differences, it may be signifi- surements from other populations, in order to deter-
cant to note that most of the equations overestimate mine whether interpopulation differences in bone
bone length when applied to Maya remains. The proportions hinder estimation of bone length using
diminutive stature of Maya skeletons is perhaps too the forensic Maya equations, and whether it would
250 L.E. WRIGHT AND M.A. VÁSQUEZ

be necessary to generate regression equations for Feldesman MR, Fountain RL. 1996. “Race” specificity and the
each population to be studied. Before applying our femur/stature ratio. Am J Phys Anthropol 100:207–225.
Furbee L, Thomas JS, Lynch HK, Benfer RA. 1988. Tojolabal
equations to distant populations, and especially to Maya population response to stress. Geosci Man 26:17–27.
significantly taller populations, we recommend that Genovés S. 1967. Porportionality of the long bones and their
osteologists verify the results with any complete relation to stature among Mesoamericans. Am J Phys An-
long bones that might be available for the popula- thropol 26:67–78.
Hens SM, Konigsberg LW, Jungers WL. 2000. Estimating stature
tion in question. Complementary research on skele- in fossil hominids: which regression model and reference sam-
tal series with differing behavioral adaptations and ple to use? J Hum Evol 38:767–784.
mean statures is needed to provide researchers with Holland TD. 1992. Estimation of adult stature from fragmentary
a variety of options for estimating the length of tibias. J Forensic Sci 37:1223–1229.
fragmentary long bones. Jacobs K. 1992. Estimating femur and tibia length from fragmen-
tary bones: an evaluation of Steele’s (1970) method using a
prehistoric European sample. Am J Phys Anthropol 89:333–
ACKNOWLEDGMENTS 345.
Konigsberg LW, Frankenberg SR, Walker RB. 1997. Regress
We thank the members of the Área de Exumacio- what on what? Paleodemographic age estimation as a calibra-
nes and the Fundación de Antropologı́a Forense de tion problem. In: Paine RR, editor. Integrating archaeological
Guatemala for their kind collaboration in our study. demography: multidisciplinary approaches to prehistoric pop-
This research would have been impossible without ulation. Carbondale, IL: Southern Illinois University. p 64 – 88.
Konigsberg LW, Hens SM, Jantz LM, Jungers WL. 1998. Stature
the various institutions that have supported their estimation and calibration: Bayesian and maximum likelihood
forensic work, both financially and politically. We perspectives in physical anthropology. Yrbk Phys Anthropol
thank Miguel Angel Morales for assistance with 41:65–92.
data recording. Last, but not least, we thank Gentry Martorell R. 1995. Results and implications of the INCAP fol-
Steele for inspiration, for his helpful suggestions low-up study. J Nutr 125:1127–1138.
Martorell R, Kettel Khan L, Schroeder DG. 1994. Reversibility of
during the research, and for his comments on the stunting: epidemiological findings in children from developing
manuscript. countries. Eur J Clin Nutr 48:45–57.
McCullough JM. 1982. Secular trend for stature in adult male
LITERATURE CITED Yucatec Maya to 1968. Am J Phys Anthropol 58:221–225.
Müller G. 1935. Zur bestimmung der Länge beschädigter Ex-
Arzobispado de Guatemala, Oficina de Derechos Humanos. 1998. tremitatenknochen. Anthropol Anzeig 12:70 –72.
Guatemala: nunca más. Informe Proyecto Interdiocesano de Simmons T, Jantz RL, Bass WM. 1990. Stature estimation from
Recuperación de la Memoria Historica. Guatemala: ODHAG. fragmentary femora: a revision of the Steele method. J Forensic
Bass WM. 1987. Human osteology: a laboratory and field manual Sci 35:628 – 636.
of the human skeleton. Columbia, MO: Missouri Archaeological Steele DG. 1970. Estimation of stature from fragments of long
Society. limb bones. In: Stewart TD, editor. Personal identification in
Bogin B, MacVean RB. 1984. Growth status of non-agrarian, mass disasters. Washington, DC: Smithsonian Institution. p
semi-urban living indians in Guatemala. Hum Biol 56:527–538. 85–97.
Brooks S, Steele DG, Brooks RH. 1990. Formulae for stature Steele DG, Bramblett CA. 1988. The anatomy and biology of the
estimation on incomplete long bones: a survey of their reliabil- human skeleton. College Station, TX: Texas A&M University
ity. J Forensic Med 6:167–170. Press.
Carmack RM. 1988. Harvest of violence. The Maya Indians and Steele DG, McKern TW. 1969. A method for assessment of max-
the Guatemalan crisis. Norman, OK: University of Oklahoma imum long bone length and living stature from fragmentary
Press. long bones. Am J Phys Anthropol 31:215–228.
Danforth ME. 1994. Stature change in prehistoric Maya of the Whittington SL. 1989. Characteristics of demography and disease
Southern Lowlands. Latin Am Antiq 5:206 –211. in low status Maya from Classic Period Copan, Honduras.
Equipo de Antropologı́a Forense de Guatemala. 1997. Las masa- Ph.D. dissertation, Pennsylvania State University.
cres en Rabinal: estudio histórico antropológico de las masacres Wright LE. 1994. The sacrifice of the earth? Diet, health, and
de Plan de Sánchez, Chichupac y Rı́o Negro. Guatemala City: inequality in the Pasión Maya lowlands. Ph.D. dissertation,
Editorial Serviprensa. University of Chicago.
 ESTIMATING LENGTH OF INCOMPLETE LONG BONES 251
APPENDIX. Additional regression lines for estimation of total bone length, derived from forensic Maya skeletons1
Regression line r2 N F SE
A. Humerus (sexes combined)
(H0–4) ⫽ ⫺16.558 ⫹ 0.577 * Humerus length 0.754 80 239.1 10.83
(H1–4) ⫽ ⫺17.334 ⫹ 0.562 * Humerus length 0.766 76 242.3 10.48
(H2–4) ⫽ ⫺18.046 ⫹ 0.524 * Humerus length 0.688 79 170.1 11.64
(H3–4) ⫽ ⫺17.743 ⫹ 0.472 * Humerus length 0.623 78 125.4 12.23
(H4–7) ⫽ 16.558 ⫹ 0.423 * Humerus length 0.622 80 128.4 10.83
B. Humerus (males)
(H0–4) ⫽ ⫺17.746 ⫹ 0.579 * Humerus length 0.685 57 119.5 15.78
(H1–4) ⫽ ⫺17.525 ⫹ 0.561 * Humerus length 0.707 57 122.8 15.09
(H2–4) ⫽ ⫺28.425 ⫹ 0.556 * Humerus length 0.616 56 86.6 17.77
(H3–4) ⫽ ⫺30.899 ⫹ 0.513 * Humerus length 0.598 55 78.0 17.20
C. Femur (sexes combined)
(F0–4) ⫽ ⫺42.021 ⫹ 0.774 * Femur max. length 0.743 96 271.2 19.16
(F1–4) ⫽ ⫺39.495 ⫹ 0.734 * Femur max. length 0.726 94 243.6 19.16
(F2–4) ⫽ ⫺43.683 ⫹ 0.606 * Femur max. length 0.623 93 150.2 20.11
D. Tibia (sexes combined)
(T0–3) ⫽ ⫺21.602 ⫹ 0.675 * Tibia length 0.663 93 178.8 16.80
(T0–4) ⫽ 17.393 ⫹ 0.623 * Tibia length 0.714 95 232.2 13.58
(T0–5) ⫽ 1.196 ⫹ 0.757 * Tibia length 0.780 93 322.7 14.01
(T1–4) ⫽ 14.623 ⫹ 0.594 * Tibia length 0.686 95 202.9 13.87
(T1–5) ⫽ ⫺1.561 ⫹ 0.729 * Tibia length 0.769 93 303.8 13.91
(T2–6) ⫽ ⫺16.913 ⫹ 0.644 * Tibia length 0.675 89 181.1 15.90
(T2–7) ⫽ ⫺25.335 ⫹ 0.732 * Tibia length 0.738 87 239.5 15.76
1
All measurements, estimated bone lengths, and standard errors are in mm. Max., maximum.