<< . .

. 11
( : 35)

. . >>

of death of the organism based on tooth eruption and wear, and we can control for
sexual dimorphism by determining the sex of the individual represented by various
animal remains. Certainly we can determine the age at death and the season of
death of at least some deer and some wapiti represented in a collection based on
tooth eruption and wear. In many instances with the deer and wapiti remains from
Meier and from Cathlapotle, however, estimates of ontogenetic age were coarse “
each estimate of age at death included a range of ±several months “ particularly
for individuals older than about 4 years. Regardless of the age estimation process,
quantitative paleozoology

figure 3.1. Ontogenetic, seasonal, and sexual variation in live weight of one male and one
female Columbian black-tailed deer. Points are for one individual, vertical lines through
points indicate observed range across multiple individuals. Modi¬ed from Brown (1961 ).

in no case were multiple lower teeth “ the skeletal elements used to determine the
age of deer and wapiti “ the most common element and thus in no case did teeth
provide both ontogenetic data and the MNI (Table 2.11 ). There is no way, then, to
determine how many 6-month-old deer and wapiti, how many 12-month-old deer
and wapiti, how many 18-month-old deer and wapiti, and so on, are represented
without inadvertently omitting some of the forty-one deer (MNImax) and thirty-six
wapiti (MNImax) represented (Table 3.1 ). Examine Figure 3.1 again and consider
how such data might be used paleozoologically. Were the same kind of data available
for wapiti, a pair of curves not unlike those in Figure 3.1 would likely be produced.

Solving Some Problems in Biomass Measurement

Signi¬cant problems with Guthrie™s protocol involve his use of an average adult live
weight and simple MNI measures (no accounting for age, sex, or size variation) of
estimating taxonomic abundances: other methods 89

taxonomic abundance. More recently, paleobiologists who have estimated prehistoric
biomass have distinguished taxa with high abundance in the fossil record from those
with low abundance, and also distinguished taxa that included mostly large individ-
uals from those that included mostly small individuals (Bambach 1993). A relatively
abundant taxon made up of large individuals represents more biomass than a rela-
tively rare taxon made up of small individuals. This analytical procedure solves many
problems because it avoids measurement error. It does so by sacri¬cing resolution;
taxonomic abundances are ordinal scale estimates as are estimates of modal body size
of individuals in a population of a taxon. Biomass estimates based on these variables
cannot be better than ordinal scale. Other forms of corrections have been proposed
in the context of estimating meat weight, the measurement we turn to next.

Measuring Meat Weight

White™s (1953a) seminal procedure for measuring the amount of meat or consumable
tissue represented by a collection of archaeological faunal remains was similar to
Guthrie™s for estimating biomass, plus one additional analytical step. First, determine
the MNI (per taxon). Second, multiply the MNI (for a taxon) by the amount of
meat one average individual (of the taxon) would provide. Third, an additional step
(actually performed second in order according to White), involves multiplying the
total live weight per average individual by the proportion of that weight thought to
be edible. The mathematical equivalent would be to calculate the biomass of each
taxon as Guthrie (1968) did, and then multiply that value by the proportion thought
to be edible (White 1953a). In this case, for example, the biomass of deer and that of
wapiti at Cathlapotle would, according to White (1953a), each be multiplied by 0.5 to
obtain the amount of edible tissue each taxon provided. White provided “pounds of
usable meat” for nearly three dozen species of mammals and more than two dozen
species of birds.
White™s procedure shares features with Guthrie™s. For example, the biomass of
a single individual (what he called “average live weight”) that he used to derive
usable meat amounts is an average of both sexes and all age classes, except for several
sexually dimorphic species. Smith (1975) suggested that because many animal taxa
are sexually dimorphic, such as deer and wapiti (but not considered as such by White),
the analyst should establish the sex ratio in a collection and add that to the procedure
for estimation of biomass. If some individuals were males and some were females,
then step 2 of White™s protocol had to be performed twice for a taxon, once for each
sex. For example, in the postcontact assemblage at Cathlapotle, the sex ratio of wapiti
quantitative paleozoology

Table 3.2. Meat weight for deer and wapiti at Cathlapotle, postcontact

Average live Postcontact Postcontact Percent Meat
weight (kg) MNI biomass edible weight
Wapiti male 400 10 4,000 50 2,000
Wapiti female 300 14 4,200 50 2,100
Deer male 100 8 800 50 400
Deer female 70 19 1,330 50 665

is three males to four females, and the sex ratio for deer is three males to seven females
(both based on anatomical features of the innominate). If those sex ratios are used to
estimate the number of males and females among the total MNI, and then biomass
and meat weight are recalculated using the estimated number of males and the
estimated number of females (based on the observed sex ratio), the calculated values
are different than when sexual dimorphism is not considered (Table 3.2). The ratio
of deer to wapiti biomass increases from 1 :3.6 without sexual dimorphism included
to 1 :3.8 when it is included. Because White™s conversion factor to derive meat weight
is 50 percent for both taxa and both sexes, the deer to wapiti meat weight ratio is also
1 :3.8. We have gained resolution as to taxonomic abundances only if MNI values are
ratio scale values and only if the estimated sex ratio is accurate. The latter may not
be given that the total number of specimens that could be sexed is much less than
twenty-four for wapiti and much less than twenty-seven for deer.
Smith (1975) also suggested that the analyst should determine the age structure or
demography of each taxon represented by the collection. This could be done using
species speci¬c schedules of mandibular tooth eruption and wear. The proportion
of the total MNI represented by each age class in the collection is then be added into
the conversion of MNI into biomass. Estimating the number of individual deer and
wapiti in each of several age categories at Cathlapotle is not possible. To get a feel
for what is involved here, consider yet again Figure 3.1 , and think about the fact that
similar data, although not as ¬ne-grained, exist for wapiti (Hudson et al. 2002).
Problems that attend White™s (1953a) measurement protocol include the fact that
the values for converting the biomass of one individual into mass of edible tissue are
debatable (Stewart and Stahl 1977). White (1953a:397) derived his conversion values
from (1) the gross morphology of the taxon (heavy-bodied, short-legged taxa vs. light-
bodied, long-legged); (2) the percentage of live weight that professional butchers and
meat packers estimated was usable meat; and (3) the presumed level of ef¬ciency
of primitive butchers. He thought his percentages would “give reasonably accurate
estimating taxonomic abundances: other methods 91

Table 3.3. Comparison of White™s (1953a) conversion values (percentage
of live weight) to derive usable meat with Stewart and Stahl™s (1977)
conversion values (percentages of live weight) to derive usable meat

Taxon White Stewart and Stahl
Mole (Condylura cristata) 70
Rabbit (Oryctolagus cuniculus) 50 40.7
Chipmunk (Tamias striatus) 70 39.7
Squirrel (Sciurus carolinensis) 70 26.0
Beaver (Castor canadensis) 70 32.2
Muskrat (Ondatra zibethicus) 70 51.9
Dog (Canis familiaris) 50 80.8
Black bear (Ursus americanus) 70 64.8
Fisher (Martes pennanti) 70 64.8
Lynx (Lynx sp.) 50 42.5
Seal (Phoca hispida) 70 30
57 b
Deer (Odocoileus sp.) 50

based on two individuals.
from Smith (1975).

results” and that any error would be “relatively constant,” likely believing that errors
were randomly distributed within and between taxa. Smith (1975) suggested that
White™s conversion factor for deer of 50 percent was too low and opted for 57 percent.
Stewart and Stahl (1977:267) measured the body weight of a dozen carcasses of various
taxa, then weighed the “edible tissues and organs,” and calculated the latter as a
percentage of the former. Overall, their values for the percentage of edible tissue of a
carcass differ from White™s (Table 3.3). Stewart and Stahl indicated that their data did
not conclusively provide a set of conversion values that gave more accurate measures
of edible tissue than White™s. They hoped such an alternative set of conversion values
could be developed, but no other set of conversion values has been proposed. Recent
researchers have used White™s (1953a) original values (e.g., Stiner 2005).
The problem identi¬ed by Stewart and Stahl (1977) may be greater than they
imagined. It is likely that no set of conversion values that are consistently ordinal scale
values, let alone ratio scale values, can be developed. Based on modern butchering
practices of ri¬‚e-equipped hunters, data have been compiled on the amount of tissue
that might be consumed. The percentage of live weight that comprises usable tissue
of male and female wapiti of different ages varies considerably (Table 3.4). Were one
to develop a set of conversion values such as Stewart and Stahl (1977) proposed, that
set would need to include not just a conversion value for every species, it would need
quantitative paleozoology

Table 3.4. Variation by age and sex of wapiti butchered
weight (eviscerated, skinned, lower legs removed) as a
percentage of live weight. From Hudson et al. (2002:250)

Age Male Female
2 years 45 48
2 to 4 years 44 46
≥ 5 years 42 51

multiple conversion values for each species. A different conversion value would be
necessary for each age class of each sex (Table 3.4). Figure 3.1 makes it clear that
numerous conversion values are required for each taxon if one hopes to have ratio
scale measures of usable meat. And this is not the only signi¬cant methodological
hurdle to perfecting White™s protocol.
Table 3.5 demonstrates that depending on how a carcass was butchered, and what
the butcher thought was edible, the conversion values may vary considerably. For a
mature male wapiti with a live weight of 350 kilograms, the proportion of live weight
that comprises usable tissue varies more than 40 percent over several different stages
of butchering. Not only that, whatever conversion value is used, that value assumes
complete consumption “ from nose through tail, as one anonymous commentator
put it “ of the carcass. As Binford (1978) demonstrated in an ethnoarchaeological
context and Lyman (1979) argued from the perspective of historic zooarchaeological
data, such an assumption is unwarranted (see also Schulz and Gust 1983). A single
deer femur in an archaeological site does not necessarily represent the meat of an
entire animal, regardless of the conversion value one uses to transform that bone
specimen into amount of usable meat (or biomass). Lyman (1979) suggested deter-
mining the minimum number of butchering units, say, hindquarters of a species
of mammal, and applying a conversion value to that quantitative unit. This sim-
ply shifts the problem of deriving a minimum number of carcasses “ the standard
MNI “ to deriving a minimum number of each of several distinct butchering units
(assuming such can be identi¬ed archaeologically), and it retains the problem of
developing conversion values for multiple age“sex categories that may not be visible
paleozoologically. The same problems attend similar suggestions by Schulz and Gust
(1983) who used historical documents to estimate the rank order economic value of
different butchering units (see also Huelsbeck 1989; Lyman 1987b).
Along these lines, Betts (2000:30) suggested that in an historic archaeological con-
text the zooarchaeologist would do well to base estimates of meat amounts “on
consumer units with a known relationship to the [faunal] material being analyzed.”
He suggested that rather than calculate meat weight per taxon based on the MNI
estimating taxonomic abundances: other methods 93

Table 3.5. Weight of a 350-kilogram male wapiti in various
stages of butchering. From Hudson et al. (2002:250)

Weight of Proportion of
Butchering stage carcass live weight
Live weight 350 1.0
Bled weight 340 0.97
Eviscerated 228 0.65
Hide and lower legs 189 0.54

per taxon, the analyst should determine the frequency of consumer units per taxon.
Consumer units might be standard quarters, wholesale units, or retail-like units of
an animal, or something else. This procedure merely relocates the problem from
converting an MNI value to meat weight to converting a (minimum?) number of
consumer units to meat weight. Betts (2000) uses a conversion procedure not unlike
White™s (1953a), complete with all its attendant problems and weaknesses. Thus Betts
(2000:30) correctly emphasizes that his suggested procedure “merely provides esti-
mates of the actual meat contributions indicated by the remains.” Those contribution
amounts are at best ordinal scale.
We are left with a grim picture of measurements of meat weight using protocols
such as Guthrie™s and White™s. Are there better techniques for measuring taxonomic
abundances based on biomass or meat weight or both? Indeed, there are other ones,
though these too have some serious weaknesses that make them rather tenuous.

The Weight Method (Skeletal Mass Allometry)

A variable that one might measure is the weight of skeletal material per taxon.
This quantitative unit was suggested by several zooarchaeologists in the 1960s and
1970s as one that could be used to measure either biomass or edible meat (Reed
1963; Uerpmann 1973; Ziegler 1973). Despite signi¬cant criticisms (Casteel 1978;
Chaplin 1971 ; Jackson 1989; Lyman 1979), zooarchaeologists the world over continue
to measure and interpret the weight of osseous material (e.g., Dechert 1995; van Es
1995; Jackson and Scott 2002; Landon 1996; McClure 2004; Prummel 2003; Tuma
2004; Weinstock 1995). One zooarchaeologist has provided a detailed review of the
nuances of this method, and argues that it should be studied further and perfected
because of its value (Barrett 1993). It is worthwhile, then, to consider this quantitative
variable in some detail.
quantitative paleozoology

First, the bone weight or, simply, weight method is based on the biological property
of allometry. Allometry concerns the relationship of the size of one property of a
body to the size of another. The size of a particular organ to the size of another in a
body, the size of the head relative to the rest of the body, or the size of a limb relative
to the size of the body are all allometric relationships. Allometry concerns the study
of such relationships and if and how those relationships change during the ontogeny
of an organism. The weight method developed by zooarchaeologists cashes in on
the allometric relationship between bone weight and the total body weight of an
organism. As biologists have noted, “animal skeletons scale allometrically with body
mass, so that skeletons of large animals are proportionately more massive than those
of small animals” (Prange et al. 1979:103). The ratio of bone weight to live weight
per individual is greater for large individuals (whether of the same or different taxa)
than that ratio is for small individuals (Needs-Howarth 1995). The presumption is
that if the statistical nature of the relationship between these two variables can be
determined, then that relationship can be used to convert bone weight observed in
an archaeological collection into biomass or usable meat weight.
As summarized by Casteel (1978), many of the original analytical protocols for
using the weight method involved two steps once the specimens comprising a collec-
tion of faunal remains had been identi¬ed to taxon. First, weigh all remains of each
taxon separately. Then, convert the bone weight of each taxon to either a measure
of biomass or of usable meat weight. As might be expected given the comments on
White™s (1953a) conversion values presented earlier, the conversion values proposed
by those advocating the weight method varied considerably from author to author.
Cook and Treganza (1950:245), for example, estimated that dry bone represented
about 6 percent of the original live weight of a mammal and of a bird. Adopting
this estimate, the weight of dry bone would be converted to biomass by multiplying
that weight by 16.67 (Casteel 1978). In contrast, Reed (1963) estimated that dry bone
weight comprised about 7.5 percent of the original live weight of a mammal. Were
one to adopt this estimate, the weight of dry bone would be converted to biomass by
multiplying that weight by 13.33 (Casteel 1978). As Casteel (1978) pointed out, empir-
ical studies (as opposed to Cook and Treganza™s and Reed™s estimates) suggested that
anywhere from about 8.5 to 13 percent of the body weight of mammals constituted
bone weight. Ignoring the slippery issue of choosing which percentage to use, once
you have biomass per taxon, you may want to convert that to usable meat, which
introduces problems like those associated with White™s (1953a) analytical protocol.
Shortly after the weight method began to see some use, Chaplin (1971 :68) noted that
anyone using it had to assume that the relationship between bone weight and biomass
(or usable meat weight, whichever was sought) was constant. This was so because
estimating taxonomic abundances: other methods 95

the conversion value chosen was a constant; it did not vary. Thus, if 5 kilograms of
bone represented 60 kilograms of biomass, then 10 kilograms of bone represented
120 kilograms of biomass, 15 kilograms of bone represented 180 kilograms of biomass,
and so on. In Chaplin™s eyes, such an assumption was not warranted. Furthermore,
in some ways anticipating later criticisms of White™s (1953a) MNI-based method,
Chaplin (1971 ) noted that there was some controversy over which conversion value
to use given individual variation within a taxon as well as the fact that an individual™s
weight would vary over time (recall Figure 3.1 ). Chaplin (1971 :68, 69) concluded
that the “relationship of bone weight and body weight is not an exact one” and he
recommended “the meat/bone ratio must be established for each bone, at different
ages of the animal and for each sex.” The meat’bone ratio constitutes recognition
that people might not have consumed an entire carcass, that 5 kilograms of phalanges
do not represent the same amount of biomass and usable meat as do 5 kilograms
of femora of the same taxon, and that 5 kilograms of humeri from 6-month-old
females do not represent the same amount of biomass as do 5 kilograms of tibiae
from 36-month-old males of the same taxon.
Those who have used bone weight as a measure of taxonomic abundance over the
last 15“20 years usually give reasons as to why it is a measure that should be used.
They also often state that it is a better measure of taxonomic abundance than either
NISP or MNI. Those arguments can be summarized as follows:

1 Measures of bone weight allow the summation or merger of abundance data into
more general taxonomic categories such as a taxonomic family or “large
2 Measures of bone weight are not in¬‚uenced by fragmentation.
3 Measures of bone weight circumvent interanalyst variation in identi¬cation skills
that bias measures of taxonomic abundance.
4 Because of the statistical relationship between bone weight and body weight,
bone weight provides proxy measures of usable meat and of the importance or
contribution of a taxon to diet.

This list may seem impressive if not overly long. Critical scrutiny of the arguments
indicates, however, that each is a justi¬cation to not use other measures of taxonomic
abundance (particularly NISP and MNI) rather than a warrant to use bone weight.
This is so because all four statements are easily shown to be variously false, of minimal
signi¬cance, or to apply to bone weight as well as to other measures of taxonomic
The ¬rst reason given to use bone weight is certainly true because one can sum the
weights of specimens identi¬ed as, say, deer with the weight of specimens identi¬ed
quantitative paleozoology

as deer size. The signi¬cance of the ¬rst statement resides, however, in the unspoken
underpinning assumption that were one to use NISP or MNI to quantify taxonomic
abundances, one would not determine the MNI for categories like a taxonomic family
or large mammal or deer size. That is false; these same categories have been used by
paleozoologists. A well-known example is the ¬ve size categories of African bovid
that zooarchaeologists use because of dif¬culties with identifying the genus or species
of bovid represented by bones and teeth (Brain 1981 ; Bunn and Kroll 1986).
The second reason given to use bone weight is also true, but it ignores the fact
that increasingly intensive fragmentation makes identi¬cation progressively more
dif¬cult. A fragment of medium- or small-size mammal long bone diaphysis might
be confused with a similar fragment from a large- to medium-size bird, for exam-
ple (Driver 1992). Thus fragmentation can in¬‚uence which bones are weighed “
those that are identi¬able are weighed, and intensively fragmented specimens will be
unidenti¬able. The second reason also relates to the ¬rst in the sense that fragments
that cannot be identi¬ed to species but that can be identi¬ed to, say, taxonomic family
can indeed be weighed (just as they can be tallied for the NISP of a taxonomic fam-
ily), but at the cost of losing ¬ne-scale taxonomic resolution (just as when fragments
identi¬able to genus or species are included in NISP tallies of a taxonomic family).
The third reason given to measure bone weight identi¬es a real problem regard-
less of whether one quanti¬es taxonomic abundances with NISP, MNI, or bone
weight (Gobalet 2001; Lyman 2005a). Specimens that are not identi¬ed “ regardless
of whether the categories of identi¬cation are species, genera, families, large, medium,
and small mammal, or whatever “ are simply not tallied nor are they weighed. There-
fore, this reason, like the preceding two, is not a valid warrant to measure bone weight.
The fourth reason contends with the fact that neither an NISP of one mouse and
of one elephant, nor an MNI of one mouse and of one elephant provides an accurate
measure of the contribution of those two taxa to diet. But the fourth reason is not
a good warrant to use bone weight to get at usable meat weight. The problem was
originally identi¬ed by Chaplin (1971 ). If one converts bone weight to biomass or
weight of usable meat, one must assume the relationship between the two variables is
constant and linear. That is, one assumes a single conversion value such as 7.5 percent
of an (average?) individual™s total weight represents the weight of the skeleton, and
the remainder is soft tissue. Chaplin (1971 ) argued that the relationship is in fact
not constant, and Casteel (1978) showed that the relationship is neither constant nor
Casteel (1978) used previously published data (McMeekan 1940) on the relation-
ship between bone weight and biomass of a domestic pig (Sus scrofa) to show that
the relationship between the two variables is curvilinear. Casteel (1978) examined
estimating taxonomic abundances: other methods 97

Table 3.6. Descriptive data on animal age (weeks), bone weight per individual,
and soft-tissue weight per individual domestic pig. All weights are grams. From
McMeekan (1940)

Bone Muscle Muscle + fat Muscle + fat +
Age weight weight weight hide weight
Birth 242.7 388 439 545
4 767 1,901 2,885 3,240
8 1,730 4,182 6,156 6,990
16 3,962 12,669 19,796 21,534
20 5,214 17,718 30,904 33,198
24 6,438 23,015 43,904 46,979
28 7,396 31,647 66,160 69,602

the relationship between the bone weight of a single individual pig and the weight
of muscle tissue of that individual, and the relationship between the bone weight of
a single individual and the total soft-tissue weight of an individual (muscle + fat +
hide) because he was unsure how completely the soft tissues of a pig might be used by
prehistoric consumers. He found that the relationships between both variable pairs
varied with the ontogenetic age of the pig; the ratio of bone weight to soft-tissue
weight increased as ontogenetic age increased.
I used the same data (Table 3.6) and expanded Casteel™s analysis to include another

<< . .

. 11
( : 35)

. . >>