Download The grasshopper or the ant?: cultigen

Journal of
Archaeological
SCIENCE
Journal of Archaeological Science 31 (2004) 753–762
http://www.elsevier.com/locate/jas
The grasshopper or the ant?: cultigen-use strategies in ancient
Nubia from C-13 analyses of human hair
H.P. Schwarcz a*, C.D. White b
a
b
School of Geography and Geology, McMaster University, Hamilton, Ontario L8S 4M1, Canada
Department of Anthropology, University of Western Ontario, London, Ontario N6A 5C2, Canada
Received 28 July 2003; received in revised form 12 November 2003; accepted 17 November 2003
Abstract
Sections of human hair from naturally desiccated Sudanese Nubian mummies representing X-Group (AD 350–550) and
Christian (AD 550–1300) periods in the Wadi Halfa area were analysed by Journal of Archaeological Science, 20 (1993) 657. These
data can be interpreted in terms of a model of annual variation of food consumption that apparently remained stable for more than
1000 years. The diet oscillated annually between consumption of 75% C3 foods (wheat and barley) in winter, to as much as 75% of
C4 foods (millet and sorghum) in the summer. There was little use of stored summer foods in the winter, whereas a small amount
(c. 25%) of winter foods were present in the summer diet; part of this may be a carry-over in the form of a C3 isotopic signal in the
flesh of C3-fed animals. This evidently small use of stored grains suggests that grain storage facilities (granaries) were largely used
as emergency measures. During normal years, the diet in a given season was dominated by freshly harvested crops.
! 2003 Elsevier Ltd. All rights reserved.
1. Introduction
Isotopic analysis of human tissues allows us to reconstruct the nutrient sources of ancient populations
[11,25]. This is possible because, in many regions,
available nutrient sources, such as diverse cultigens,
differ substantially in their natural stable isotopic compositions: !13C, !15N.1 Since these differences are transmitted faithfully to the living tissues of the consumers,
they provide a basis for establishing the relative proportions of the various foods that were actually utilized by
the peoples studied. For example, pre-Columbian populations of North America have been shown to have
shifted from a diet of wild, C3 (low-!13C) plants and
C3-consuming herbivores to a diet containing progressively more of the C4 plant maize, starting around AD
700 and reaching a climax around AD 1500 [13,25].
* Corresponding author. Tel.: +1-905-525-9140; fax:
+1-905-546-0463
E-mail address: [email protected] (H.P. Schwarcz).
1
!13C is defined as follows: !13C=({[13C/12C]sample/[13C/12C]standard}
"1)#1000 in per mil (‰). !15N is similarly defined for 15N/14N.
0305-4403/04/$ - see front matter ! 2003 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jas.2003.11.004
Typically, such isotopic paleodietary studies are
based on the !13C analysis of collagen in human bones.
The residence time of carbon atoms in collagen of living
bone is on the order of 10 years [15]. Therefore isotopic
paleodiet studies of this material give us only mean
consumption patterns averaging over the last ten or
more years of life. It would also be desirable to show
dietary variations on a shorter time scale, especially
where seasonal variations in food availability are suspected to have existed. For this purpose, it would be
necessary to analyse tissues which had a shorter turnover time, either because they regenerate quickly (such
as muscle or organ-tissues) or because they are fast
growing components of the body such as skin, hair or
nails.
In earlier studies [29,31] we showed that naturally
mummified human remains from the hyper-arid Nile
Valley in Nubia contain preserved soft tissues which can
be used to test shorter time-scale dietary variations.
White [29] showed, in particular, that hair segments
revealed large variations in !13C which she attributed to
changes in the relative use of C3 grains (barley and
wheat) vs C4 sorghum and millet through a seasonal
754
H.P. Schwarcz, C.D. White / Journal of Archaeological Science 31 (2004) 753–762
Table 1
Planting and harvesting seasons in the Sudan (modified after US
Department of Agriculture, 1960 [17])
Table 2
Isotopic composition of modern plants used in ancient Nubia and
sampled from the Nile Valley (after White and Schwarcz [31])
Commodity
Planting season
Harvesting season
Scientific name
Common name
Plant part
!13C (‰)
C4 plants
sorghum
millet
July–August
July–September
October–November
October–December
C3 plants
wheat
barley
November–December
November–December
March–April
March–April
C4 plants
Setaria viridis
Digitaria sanguinalis
Pennisitum divisum
Panicum coloratum
Sorghum durra
S. sudanensis
millet
millet
millet
millet
sorghum
sorghum
seed
seed
seed
seed
seed
seed
October–November
October–November
November
July–August
July
January–mid-April
February–March
February–March
March
October
October
mid-May–June
August–November
"10.8
"10.6
"12.1
"13.1
"11.0
"12.4
Mean"11.7
SD 0.9
C-3 plants
Triticum vulgare
Hordeum vulgare
Vigna membranaceae
Cajanus cajan
Acacia nilotica
A. albida
Hibiscus esculentus
Eruca sativa
Raphanus sativus
Allium sepa
Solanum melongena
Corchorus olitorius
Hibiscus sabdariffa
Balanites aegyptiaca
wheat
barley
cowpeas
pigeon peas
acacia beans
acacia beans
okra
garden rocket
radish tops
onion
eggplant
jew’s mallow
rosella
date
seed
seed
seed
seed
seed
seed
leaf
leaf
leaf
skin
leaf
leaf
fruit
fruit
"27.0
"22.1
"25.1
"27.2
"25.0
"27.5
"21.0
"30.3
"30.9
"28.0
"28.4
"28.2
"23.3
"27.1
Mean"26.5
SD 2.9
broad beans
beans, dry
chick peas
groundnuts
sesame
onions
dates
raisins
lemons
oranges
grapefruit
September
September–August
September–December
September–December
cycle. In this paper we shall look further at these data in
order to extract additional information about the seasonal cycle of cultigen use vis a vis the importance of
food storage in this area.
The samples analysed represent members (N=17) of
the so-called X-Group (AD 350–550) and Christian (AD
550–1300) populations, excavated by the University of
Colorado near Wadi Halfa, in northern Sudan. The site
is now submerged beneath the waters of Lake Nasser.
The plant foods available to these people and the
probable times of the year when they were harvested are
shown in Table 1. The isotopic compositions of the plant
foods are determined by the photosynthetic pathway
which they utilize. The winter-harvested grains (barley
and wheat) are C3 plants which today have isotopic
compositions averaging around "26‰, while the
summer-harvested crops (millet and sorghum) are C4
plants whose !13C values are today around "12‰.
These values would have been about 1.5‰ larger (less
negative) in pre-industrial times [16]. Table 2 shows the
isotopic composition of the major cultigens, corrected
for this shift. A sample of sorghum collected by P.
Rowly-Conwy at a nearby Christian site gave a !13C
value of "9.7‰. The isotopic composition of eggs (of
duck and geese) and meat from animals would be close
to the !13C values of the diets of their source animals,
because there is not much fractionation between these
materials and diet. Milk, due to its higher lipid content,
would be up to 2‰ lighter than the total diet. Because
they were produced over limited time spans, the isotopic
compositions of eggs and milk would have varied
seasonally, as a function of the composition of available fodder, whereas the !13C of animal flesh would
presumably represent a longer-term average of food
consumption by the animal.
2. Analytical procedures
Hair was isotopically analysed according to analytical
procedures described by White [29]. Briefly, bundles of
15 or more strands of hair were cut into 2–4 segments of
20 mm length. The normal rate of growth of hair is
about 0.35 mm/day [23], so the segments would represent about 50–60 days of growth. The samples were
cleaned of lipids with methanol, dried, reacted with CuO
at 550 (C in sealed, evacuated pyrex tubes to produce
CO2, and analysed on a VG602 mass spectrometer to
determine !13C, with a precision of about 0.2‰. The
isotopic data are reproduced in Table 3 from White [29].
3. Temporal analysis of the data
The data are shown in Fig. 1, plotted as a function of
distance from the scalp. The data for single bundles of
hair were graphically presented by White [29, p. 663],
who noted that the isotopic trend from hair tip (T) to
scalp (S) was most commonly in the direction of increasing !13C, terminating with values that indicated greater
levels of C4 food consumption. We have subsequently
755
H.P. Schwarcz, C.D. White / Journal of Archaeological Science 31 (2004) 753–762
Table 3
Analyses of hair bundles
Site/burial
Age (y)
Hair segments (scalp to tip, !13C (‰ VPDB))
Sex
th
b
S1
S2
S3
S4
2
4
6
8
$!s1"sx$a
X-Group (AD 350–550)
NAX
26
3
3
20
25
22
9
21
29
18
25
35
F
?
?
M
M
F
?
M
F
M
F
F
"13.2
"18.8
"18.5
"13.8
"18.7
"19.6
"18.0
"14.6
"18.8
"13.5
"15.4
"17.1
"13.4
"19.5
"18.5
"15.7
"17.3
"19.2
"18.3
"14.6
"18.7
"15.6
"15.0
"16.4
Christian (AD 550–1300)
6G8
2
?
9
6
33
?
6B13
7
19
29
1
?
F
M
F
?
"15.5
"11.7
"15.9
"18.5
"17.2
"15.5
"13.0
"17.8
"19.5
"18.1
24I3
a
b
551
552
580
611B
640B
645
659
671A
676A
5
21
43
0.2
0.7
0.0
1.9
2.3
1.4
0.9
3.3
0.1
2.1
0.9
3.8
"16.4
"18.2
"18.9
"17.9
"14.5
"13.3
"15.6
"13.3
"19.2
"15.9
"14.9
"19.3
0.4
3.2
3.4
1.0
0.9
$!s1"sx$=absolute value of difference between first (s1) and last (sx, x=2,3,4) segment.
th=estimated relative age of hair segment (months).
"16‰) statistically have less variable ($!s1"sx$, Table 3)
values along their length than bundles with higher (less
negative) values (Student’s t=2.2, df=15, P>0.05). This
suggests that these samples represent different phases in
the agricultural year at this site. We have therefore
attempted to treat the analyses of each bundle of hair as
a record of the history of change in !13C in actively
growing tissue through time, and we have tried to synthesize these data into a more-or-less complete picture of
changing diet through time for this site. This approach is
based on the following assumptions:
Fig. 1. Isotopic ratios (!13C) of hair segments of Nubian mummies,
plotted as a function of distance from scalp. The open circles are the
analytical value positioned at the centre of the segment. The solid points
at the end of each line represent the ends of the analysed hair segments.
noticed that these data can be organized in a different
fashion than shown by White [29]. Specifically we note
that hair bundles with low !13C values (i.e. below
1. the average !13C of hair growing on any given date
in a solar year was always about the same over long
(multi-century) time periods. Although bone, skin
and muscle register greater consumption of C4 foods
during the X-Group period [31], the means of hair
samples for these periods do not show a significant
difference in the consumption of C3 versus C4 foods
(Student’s t=0.87, df=42).
2. the samples of hair come from individuals who died
throughout the entire year (although the mortality
rate may have been higher in some seasons).
3. the rate of growth of hair was approximately constant in the population;
4. there was a constant isotopic fractionation between
hair and diet (!hd).
756
H.P. Schwarcz, C.D. White / Journal of Archaeological Science 31 (2004) 753–762
2. There are no bundles whose data represent periods
of slow change and high !13C values.
3. The samples with the highest !13C values exhibit
steep positive and negative gradients.
4. Two pairs of hair bundles from different individuals
(24I3 # 21, 6G8 #2) give closely matching data for
their three sample points, or segments, indicating
that these individuals had consumed similar diets for
a similar period of time shortly before death. A third
individual (NAX 551) showed no change in !13C
over two segments, although its average !13C was
relatively high ("13.4‰).
5. During the Christian period (dashed lines) !13C
dominantly increases from T to S (toward the time
of death) whereas values for X-Group samples increase as commonly as they decrease.
6. The total range of !13C values from "12‰ to about
"19.5‰ suggests that the diet varied by as much as
50% in the proportions of C3 and C4 foods through
the year.
Fig. 2. Data from Fig. 1, replotted with an arbitrary origin for
distance, and with hair bundles arranged to form a continuous curve.
This appears to be a unique arrangement; no other method of
arrangement gives a single continuous curve. Segments NAX 551, 580,
676A have been omitted (see text). Solid lines=X-Group; dashed
lines=Christian.
Based on the first assumption, we have arranged the
individual records on a single plot in such a way that, as
closely as possible, they define a single roughly continuous function (Fig. 2). The T-point on each bundle is
plotted to the left, so that the graph shows date of
growth increasing from left to right. The overall pattern
is a “U”-shaped curve which could be interpreted as a
single year’s sample of a continuous periodic function,
representing the long-term variation in !13C in the diet,
offset by !hd. The graph includes the data for the
X-Group and the Christian period. This method of
plotting shows that, although the data are taken from
two temporally and culturally distinct populations, they
seem to form a single homogeneous data set. Inspection
of Fig. 2 reveals certain characteristic features:
1. The majority of the bundles display !13C values that
are low (<"16‰) and slowly varying; a smaller
number of bundles present data from periods of
gradual to rapid change in isotopic composition,
with !13C values both increasing and decreasing.
We believe that the U-shaped trend of Fig. 1 corresponds to a seasonal trend of variation in !13C through
most of a yearly cycle. Since the individual samples
represent deaths occurring over hundreds of years, the
data must represent a persistent, continuous trend
which was repeated every year for centuries. Based on
the foregoing assumptions, we infer that this annual
cyclicity in !13C corresponds to a seasonal shift in the
isotopic composition of the foods consumed by these
people.
If we assume that the rate of growth of hair was
constant throughout the year, then the x-axis scale of
length is proportional to the time of growth. From Fig.
2, the complete cycle from one maximum to the next
takes place during the growth of about 140 mm of hair.
Assuming that the complete series represents approximately one year of growth, the hair grew at 0.38 mm/day
which is comparable to the average rate of 0.35 mm/day
[23].
3.1. Variation in !13C
Hair is made of the protein keratin. Like other
proteins, the isotopic composition of a sample of hair
represents the average !13C of the diet, biased toward
that of the protein component, due to the tendency of
the body to construct proteins from pre-formed dietary
amino acids rather than to synthesize the non-essential
amino acids from other dietary sources, especially carbohydrate and lipid [1,14]. However, the main protein
sources for people at this site, as indicated by !15N
values of human tissues, would likely have been domesticated animals (mammals, birds) which were fed agricultural products [31]. There is very little offset in !13C
H.P. Schwarcz, C.D. White / Journal of Archaeological Science 31 (2004) 753–762
(<1‰) between the flesh (or other edible products) of
animals, and the plant foods which they are fed [26].
Also, we know that these variations in !13C are not in
any sense intrinsic to hair growth. For example, sequential analyses of hair from individuals from Egyptian
oases showed no systematic trends in !13C along hair
bundles, even though both C3 and C4 cultigens were
available [7,30]. We suppose therefore that the variations seen at Wadi Halfa are entirely due to seasonal
variation in the proportion of C3 (wheat, barley, fruits,
vegetables, pulses) and C4 (sorghum, millet) plants consumed by these individuals, as described by White [29].
In order to calibrate the changes in !13C of hair with
respect to the amounts of C3 and C4 foods eaten, we
must know the isotopic fractionation between diet and
hair. O’Connell et al. [20] found that hair was depleted
in 13C by 1.4‰ with respect to bone collagen. Assuming
a diet-collagen fractionation of 5‰ [25], this implies that
hair is enriched by about 3.5‰ with respect to diet.
Other controlled feeding studies show an offset of 1 to
1.4‰ between diet and hair [6,10,18,27]. Estimates of the
hair-diet fractionation (!hd) for humans in noncontrolled feeding contexts range between 2.6 and 3.2‰
[8,12,19,21]. We will assume an average enrichment of
3‰.
Fig. 2 shows that !13C of hair varied between
"19.0%0.5 and "11.5%0.5‰. This corresponds to a
variation in diet between "22 and "14.5‰, which
would correspond to a shift from dominantly C3 (low
!13C) to C4 (high !13C) plant foods, either consumed
directly by humans as bread and other plant-based
foods, or consumed by animals and then transferred to
humans as meat, milk and eggs. If we neglect the
possible minor role of wild grasses in the animal diet,
then most of the cyclic (seasonal) variation in !13C
appears to be a shift from dominant use of C4 plants to
dominant use of C3 plants and then back again. That is,
the maximum and minimum !13C values correspond to
the times of maximum consumption of C4 (13C-rich) and
C3 (13C-poor) crops, respectively, offset by the lag
between the time of consumption of a given food and the
appearance of its carbon atoms into the growing hair.
The actual seasonal chronology of this progression in
!13C values cannot be independently determined since
we do not have the dates of death of these individuals. In
any case, the date of appearance of a given diet-based
isotopic signal will be offset from the date of consumption of that nutrient by a delay corresponding to the
time elapsed between consumption and protein synthesis
[20,29]. The times of harvest of the C4 and C3 dominant
crops are in early summer (June, the seifi crop) and
late-fall to mid-winter (November–January, the shitwi
crop) respectively. These two harvests are almost exactly
six months apart, consistent with the symmetrical form
of the curve in Fig. 2. A third crop (dameira), which
combines seifi (C4) plants with fruits and vegetables (C3)
757
is harvested in October. Assuming a constant lag between consumption of a food and appearance of its
isotopic signal in hair, the dates of appearance of signals
will be temporally spaced in the same way as the dates of
consumption of the respective crops.
We can use the inferred isotopic composition of the
diet at the isotopic extremes of Fig. 2 to calculate the
range in proportion of C3 and C4 foods in the diet with
the formula from Schwarcz et al. [25]:
%C4!
"!c!!3#!d!h$"100
"!4!!3$
where !3 and !4 are the !13C values of the C3 and C4
foods, respectively. This formula assumes that the quality of the respective C3 and C4 foods (i.e., proportions of
various amino acids, lipids, carbohydrates) were similar.
If we assume that the C3 foods (wheat, barley) had !13C
values of about "26‰ while sorghum had a value of
"10‰, then the most 13C-depleted diet consisted of
about 75% C3 foods, while the most enriched diet
consisted of about 25% C3 foods. Referring to the
percent C4 food in the diet as PC4, the hair data suggest
that PC4 varied between a low value of 25% and a
maximum of 75%.
4. Discussion
We have shown that !13C of hair at Wadi Halfa
varied through the course of a year reflecting change in
the average !13C of diet through the year. The same
extent of variation was seen both in the earlier X-Group
and the later Christian population at this site. This
degree of variation was not clearly shown by analyses of
either skin or muscle [31], although both of these tissues
should turn over their carbon atoms on a time scale of
less than one year. Presumably the continuous secretion
of keratin by hair follicles provides a better running
monitor of changing diet, although its variation too
would be somewhat decreased by the buffering effect of
the protein stores in the body. The minimum and
maximum PC4 values which we have inferred are therefore upper and lower limits, respectively, to the actual
seasonally varying PC4 values of the dietary intake.
Even in the absence of stored C4 plants, and at the
point in time farthest removed from the harvest of C4
plants, it seems unlikely that PC4 in this population
would ever have dropped to zero. This is because the
diet would have continued to include the meat of
animals whose year-round diet included some millet and
sorghum, which are major fodder crops today [2,17].
These crops would have imparted more 13C-rich values
to the meat, for which the turnover rate of C atoms is
much slower than in human hair. Consumption of meat
from a C4-consuming animal would result in dietary
758
H.P. Schwarcz, C.D. White / Journal of Archaeological Science 31 (2004) 753–762
!13C values equivalent to that from direct consumption
of some C4 plant foods, and a PC4 value greater than zero.
The seasonal alternation between C3 and C4 foods is
determined by the climatic tolerance of these plants
and their consequent times of planting and harvest.
Although today a variety of C3 foods are planted and
harvested at various times throughout the year, the
staple grains are scheduled as opposite season crops
(Table 1). Sorghum and millet (C4), wheat are planted
between July and September and harvested between
October and December. The C3 staples, wheat and
barley, are planted after the annual flood from
November to December and harvested in March and
April.
Animal-derived foods would track these seasonal
isotopic shifts to the extent that the animals were fed
harvested grains. This would influence !13C of eggs, milk
and the flesh of domestic animals such as ducks, geese,
goats and sheep. The seasonality of the !13C signal in
each of these foods would depend on the size and
turnover rate of the carbon pool that is represented. It
appears, in any case, that the residents of Wadi Halfa
tended dominantly to consume C3 plants and plantderived foods in the season of harvest of these plants,
and tended to consume C4-labelled foods during the
season of harvest of C4 plants, either directly or through
the consumption of rapid-turnover tissues of animals
(milk, eggs and meat of smaller animals).
In constructing Fig. 2 we were able to combine the
data from X-Group and Christian burials into a single
homogeneous curved array, indicating that this pattern
of seasonal consumption persisted over a span of more
than 1000 years. This suggests that the choices of
possible cultigens and the manner in which their harvests
were scheduled were extremely consistent over these
long intervals even though cultural and climatic factors
(extent of Nile flooding, droughts, etc.) may have affected other aspects of life in Nubia. This consistency in
choices and management of cultigens is brought out in
long-term studies of food consumption in Egyptian
populations [5], and continues in the descriptions of life
in Coptic Egypt [4,22].
From Fig. 2 we also can note that there was no
significant inter-populational difference in the most
probable time of death. White [29] has noted the significantly higher mortality rate in individuals with isotopic
values of hair near the rising (late spring) portion of the
cycle. She concludes that the actual times of death would
be a few weeks after the ingestion of these foods, due to
the delay time needed for the body to equilibrate to the
isotopically enriched foods. Given that the 13C-rich, C4
foods are typically harvested in June, there was a
preferential time of death in late July–August, a time of
increased climatic stress due to higher temperatures [17].
These effects were as prevalent in the X-Group as in the
Christian Group.
Three exceptional individuals (one X-Group, the
other two Christian), as noted earlier, appear not to
have shifted to a relatively pure C3 diet in the winter
season, but rather exhibit intermediate !13C values.
Presumably these were two individuals whose diet included significant amounts of both C3 and C4 foods,
possibly because they lived at a time when stored C4
foods were required to supplement recently harvested
winter grains. It is also possible that these individuals
had differential access to foods or perhaps personal
preferences in diet which resulted in their anomalous
isotopic behaviour.
Let us now consider the significance of the alternation
in dominance of food sources through the year. Under
normal conditions, the steady flow of the Nile would
have guaranteed the availability of crops throughout an
annual cycle. However, availability of grains could be
affected by both environmental change, such as fluctuations in the level of the Nile, and political or economic
policies involving trade or taxation. Trade in foods and
other commodities existed along the Nile, from earliest
Pharonic times [5,28], but we assume that imported food
never formed a mainstay of the subsistence base. Storage
technology was probably developed during the preceding Meroitic period (350 BC to AD 350) when this area
was an agricultural hinterland for the Kingdom of
Meröe to the south. The X-Group populations, however, represent a period of disintegration of centralized
authority and were autonomous chiefdoms. As there is
no current evidence of the necessity to store and submit
grain as a tax, most storage at this time was likely an
insurance against famine.
A population with access to varying sources of food
can be assumed to have various options with which to
handle these supplies. Two extreme possibilities would
be as follows:
Immediate consumption (IC): As each seasonal crop
becomes available it is consumed as the major source of
nourishment for humans and animals.
Storage/pooled consumption (S/PC): As each crop is
harvested, the majority of it is stored in granaries, from
which it is drawn upon as needed.
If the latter (S/PC) model describes the most frequent
behaviour of people living at Wadi Halfa, then we would
expect that !13C of diet would remain almost constant
through the year, as was observed at oases [7,30].
Depending on the extent to which the granaries homogenized the returning flow of food back to the consumers, the S/PC model might partially approximate the
IC model. In the case of complete IC behavior, on the
other hand, we would expect a seasonal variation in diet
and therefore in !13C of hair (and other rapid-turnover
H.P. Schwarcz, C.D. White / Journal of Archaeological Science 31 (2004) 753–762
759
Fig. 3. Interpretation of data from Fig. 1 showing possible temporal sequence in relation to possible temporal sequence. The solid bars represent
possible times of harvest of the respective crops (wheat, barley in November to January; millet and sorghum between June and September). Winter
consumption of 13C-depleted C3 grains leads to falling !13C values in growing tissues (i.e., hair), reaching a minimum in late spring. As consumption
of C4 grains begins in summer, !13C of tissues begins to rise. Lack of a period of constant high !13C values is discussed in text. A lag time of a few
weeks between consumption and appearance of an isotopic signal in the hair might have further shifted the position of these curves with respect to
the corresponding dates of harvest.
tissues) that would reflect the range of !13C in the
nutrient sources.
The data shown in Fig. 2 suggest that the Wadi Halfa
population was at least partly following the IC model.
Very few of the hair bundles showed constant !13C
throughout their length, and most of those that did
displayed !13C values close to the minimum of the
population, as would be expected for those samples that
happened to record the diet during the period of least
availability of C4 foods (i.e., late winter to early spring).
The remainder of the samples record transition from
C4-dominant to C3-dominant diets, or the reverse.
We propose therefore that the isotopic records of hair
are transformed representations of the record of harvests of C3 and C4 plants. On Fig. 3 we show schematically how these two series could have been related in
time. The falling limb of the population (decreasing
!13C) would represent people who died in the spring
while data in the rising limb come from people who died
in the late summer. Only a small number of individuals
show high PC4 values. The distribution of data is partly
a function of the mortality trends in this population. The
lack of individuals with high, constant !13C values
presumably reflects the fact that few individuals died
during the period of C3 harvests. This would have been
a period of cooler climate and abundant food, possibly
leading to lessened morbidity and is the time of the year
when fewer people die even in modern times [17].
According to this model it is likely that at some time
during the year the !13C of the hair of a significant
number of people at Wadi Halfa was greater than
"14‰, but they are not represented in these data
because none of the sampled population had died while
their hair had this composition.
The following fable captures what we have found here:
one sunny day in late summer the grasshopper observes his
friend, the ant, busily storing away grain in his underground storage bins. The grasshopper laughs and chides his
friend for such a waste of energy and goes back to nibbling
on tender stalks of grass. But a few months later, when the
grass has withered and cold winds are blowing, the grasshopper wonders where he can find something to eat; meanwhile, the ant is contentedly munching on his stores of
summer produce. In our treatment of the data, the IC
model represents the grasshopper community: all harvested
resources are immediately consumed and none are stored.
Ant communities follow the S/PC model: some fraction of
the available food is always put into storage and drawn
upon throughout the year. This would tend to even out the
!13C values of the hair of consumers. Note that in some of
the individuals we analysed, !13C is almost constant, but
lying at intermediate !13C values of "15.5 to "16.5‰.
These individuals appear to have followed the S/PC model
for unknown reasons. The intermediate position of these
persons’ isotopic data is consistent with the notion that
when grains were stored they consisted of approximately
equal amounts of C3 and C4 foods. In general it seems
reasonable to suppose that, during good times, all individuals would tend to follow the IC model, consuming crops
as they become available. Use of stored crops would tend
to occur during times of crop failure or insufficiency.
The model presented here and the data on which it is
based inevitably leave many matters unknown:
4.1. Relative crop yields
Although we know that both C4 and C3 crops were
harvested at Wadi Halfa, we do not know the relative
760
H.P. Schwarcz, C.D. White / Journal of Archaeological Science 31 (2004) 753–762
sizes of these two crops. If the IC model is a good
description of behaviour at this site, then we suppose
that both crops were of comparable size, or else stored
C3 (or C4) crops would have been required to see the
populace through the agricultural year. That is, some
partial S/PC behaviour would have been entailed by
unequal crop yields. There is no clear isotopic evidence
to support or reject this model. Although no granaries
or other storage facilities have been reported at this
site, the presence of tetracyline-labelled bone in this
sample does indicate that consumption of stored grains
occurred [3]. Tetracyline is produced naturally from
Streptomyces infestation of stored grain and has also
been found in bones at other Nubian sites [9]. We are
currently carrying out further studies of the distribution
of tetracycline labelling in bones from various sites in
relation to other skeletal indicators of life status.
We may make some comparison with crop harvests in
modern Nubia. Today, the main subsistence crop is that
harvested in the winter (sitwi: [4,17]). The spring crop
(seifi) is dominated by C4 staples, whereas the late
summer (dameira) crops combine both C3 and C4
plants. The existing literature on Nubian agronomy does
not make clear the relative sizes of these seasonal crops,
although the sitwi (winter) crop is considered the most
important. Clearly, however, we cannot assume that
these cropping practices have persisted unchanged since
ancient times.
4.2. Buffering of isotopic composition
One of the unknowns in the interpretation of our data
is the extent to which protein and fat stores in the body
were being called on to provide amino acids for keratin
synthesis. If, for example, nutrient levels were generally
low because one of the annual crops (presumably the
summer crop of sorghum and millet) was skimpy, then
keratin molecules would tend to be built from stored
nutrient, and therefore dampen the shift to higher !13C
values. The opposite trend is also possible, although the
data suggest that it did not occur.
4.3. Effects of age and sex
We would expect that juveniles, because of their
smaller body mass and higher growth rates would turn
over their carbon pool faster and therefore show values
of !13C closer to the actual diet (i.e., less buffering).
Consequently, in the summer months their !13C values
would be higher than for adults and would place them
closer to a pure C4 diet. The present data set shows no
such separation, but one juvenile individual (6B13#29)
is probably pre-weaning and its !13C values would track
those of its mother. The other four children (ages
between 3 and 9 years) had !15N values indicating
that they had been weaned and, therefore, would have
had !13C values indistinguishable from the rest of the
population.
The difference between $!S1"Sx$ for males and
females in Table 3 indicates that male diets swing more
widely and more consistently between C3 and C4 foods
than do female diets (x̄$!S1"Sx$ for males=2.6%7‰, for
females=1.5%1.4‰). This difference verges on statistical
significance (Student’s t=1.57, df=10) and could reflect
differential consumption behaviour between the sexes
where males tend to use the preferred IC model and
females rely more on the S/PC model. Differential access
to food by males is also indicated in the larger sample by
"15N values showing male consumption of different
and/or more abundant protein [31].
Other differences in socio-economic status could
not be tested in this sample as there are no archaeological indicators for any status differences in the
population.
4.4. Dominance of C3 foods
A large proportion of individuals exhibited !13C
values less than "16 ‰ This appears to be a consequence of the mortality trends in the population (that is,
most likely season of death). Nevertheless, it is striking
that there are no individuals with constant, high (C4dominated) !13C values. Indeed, all the line segments
on Fig. 1 for individuals with high !13C values are
characterized by steep positive or negative slopes. The
highest !13C values are between "13 and "11 ‰. This
suggests that, although C4 grains were the dominant
food source of humans and animals for some part of the
year, C3 foods continued to comprise up to 25% of
the nutrient base.
4.5. The routing problem
We have indicated above that proteins are preferentially utilized as a source of non-essential amino acids,
even though the body is capable of synthesizing them.
However, Schwarcz [24] has noted that when diets are
deficient in protein, non-essential amino acids will be
synthesized from the global carbon pool (with the
exclusion of some lipids). At Wadi Halfa, we see that the
isotopic composition of a human protein (keratin) is
apparently tracking changes in the abundance of C3 and
C4 plant foods. This does not necessarily mean, however, that the diet was protein deficient, since rapidturnover animal proteins such as dairy products, eggs,
and the flesh of short-lived animals (birds) could be
reflecting a month-scale turnover in sources of feed
grains. These data do, however, indicate that the flesh of
wild (or domesticated) animals (dependent on dominantly C3 wild plants) did not constitute a major source
H.P. Schwarcz, C.D. White / Journal of Archaeological Science 31 (2004) 753–762
of protein in most people’s diets. This is consistent with
modern ethnographic data indicating that domesticated
animals are dominantly fed C4 plants (millet and
sorghum) [2,17]. Indeed, it is possible that these grains
may have been stored for use as fodder.
5. Conclusions
From a re-analysis of the isotopic data on human hair
recovered at Wadi Halfa, we have demonstrated that
although granaries or storage facilities that may have
integrated one or more years of production were probably used, the populace at this site also tended to
consume crops soon after they were harvested. These
practices, which we have referred to as “grasshopper”like behaviour, persisted over more than a thousand
years at this site, and through major transitions in social
organization.
The arguments presented here, based on isotopic
analyses of hair, help to refine our understanding of
food utilization practices, and provide a model for
seasonal food consumption and storage behaviour at
other Sudanese Nilotic sites.
We have noted that none of these snapshot records of
diet indicate either pure C3 or C4 composition. Such
mixed consumption could result from: (1) storage availability of foods with an isotopic composition different
from the dominant foods immediately available; (2)
storage availability of foods that are both C3 and C4; or,
(3) immediately available C3 and C4 foods. The shortterm fluctuations observed between hair segments and
their arrangement in a seasonal cycle strongly suggest a
significant amount of immediate consumption. Possible
evidence for use of stored grain, however, is most
evident in three individuals whose data failed to cohere
to the cyclical pattern of Fig. 2. The !13C values of the
successive segments of these persons’ hair were uniform,
and close to the midpoint of the total !13C-range.
Perhaps these hair samples came from individuals who
had died during years of food shortage, when the
community was subsisting on a homogeneous supply of
grain. The intermediate !13C value of these samples also
suggests that the individuals were consuming a mixture
of summer and winter grains.
As a possible explanation for the paucity of individuals who had homogeneous diets indicative of yearround consumption of stored grains, we can consider the
real significance of the story of the grasshopper and the
ant. Societies may prepare for times of hardship and yet
seldom experience them. As long as society is stable and
undisturbed by war or internal conflict, we may expect it
to have some mechanisms for creating reserves of food,
whether or not crises arise necessitating their use. There
may be long periods in which no shortages are experienced and grain-storage practices fall into disuse. In the
biblical story of Joseph’s interpretation of the Pharaoh’s
761
dream (Genesis, 41:1–57), it appears that the Pharaoh of
that day lacked a plan to deal with years of decreased
food supply (a grasshopper). But when the crisis arose,
he and his vizier were quite capable of constructing
storehouses (executing ant behaviour) to withhold large
parts of the current, overabundant crop for future use.
Sequential analyses of hair are ideal data to test
for ant/grasshopper behaviour because the growth
sequence from tip to scalp has a “clock” built into it
which allows us to reconstruct the temporal variation
in food consumption. Unfortunately sites at which hair
is preserved for studies such as this are extremely rare.
There are no other rapid-turnover tissues which are
commonly preserved in burials. To further our understanding of the nature and consequences of tissue
turnover vis a vis isotopic values, we are currently
reconstructing the presence of rapid, seasonal excursions in !13C within single osteons using surface ionization mass spectrometry (SIMS) to determine total
!13C (collagen+bone mineral carbonate). This will
allow us to see progressive changes in the isotopic
composition of bone through the course of a year that
would track changes in diet. This may allow us to
determine to what extent other populations made use
of seasonally available agricultural commodities as
they appeared rather than pooling them in long-term
storage. In the meantime, the analysis of hair as a
rapid turnover tissue has hopefully contributed some
new insights to the use of stored foods in the history of
agricultural economics.
Acknowledgements
We thank Prof. George Armelagos for providing
access to the samples analysed in this and our previous
studies at Wadi Halfa. Martin Knyf provided invaluable
assistance in the isotopic analyses. The research was
supported by grants to HPS and CDW from the Social
Sciences and Humanities Research Council and the
Natural Sciences and Engineering Research Council of
Canada.
References
[1] S.H. Ambrose, L. Norr, Experimental evidence for the relation of
the carbon isotope ratios of whole diet and dietary protein to
those of collagen and carbonate, in: J.B. Lambert, G. Grupe
(Eds.), Prehistoric Human Bone—Archaeology at the Molecular
Level, Springer, Berlin, 1993, pp. 1–37.
[2] G.H. Bacon, Crops of the Sudan, in: J.D. Tothill (Ed.), Agriculture in the Sudan, Oxford University Press, London, 1948,
pp. 302–400.
[3] E.J. Bassett, M.J. Keith, G.J. Armelagos, D.L. Martin, A.R.
Villaneuve, Tetracycline-labelled human bone from ancient
Sudanese Nubia (A.D. 350), Science 209 (1980) 1532–1534.
762
H.P. Schwarcz, C.D. White / Journal of Archaeological Science 31 (2004) 753–762
[4] H. Dafalla, Land economy of old Halfa, Sudan Notes and
Records 50 (1968) 63–74.
[5] W.J. Darby, P. Ghaliounghui, P. Grivetti, Food: The Gift of
Osiris, Academic Press, London, 1977.
[6] M.J DeNiro, S. Epstein, Influence of diet on the distribution of
carbon isotopes in animals, Geochimica et Cosmochimica Acta 42
(1978) 495–506.
[7] T. Dupras, Dining in the Dakhleh Oasis, Egypt: Determination of
diet using documents and stable isotope analysis. Unpublished
Ph.D. thesis, McMaster University, Hamilton, 1999.
[8] A. Froment, S.H. Ambrose, Analyse tissulaires isotopiques et
reconstruction du régime alimentaire en milieu tropical: implications pour l’archéologie, Bulletin et Memoires de la Societé
d’Anthropologie, Paris 7 (1995) 79–98.
[9] J.R. Hummert, D.P. Van Gerven, Tetracycline-labeled human
bone from a medieval population in Nubia’s Batn El Hajar
(550–1450 A.D.), Human Biology 54 (1982) 355–371.
[10] R.J. Jones, M.M. Ludlow, J.H. Troughton, C.G. Blunt, Changes
in the natural carbon isotope ratios of the hair from steers fed
diets of C4, C3 and C4 species in sequence, Search 12 (1981)
85–87.
[11] M.A. Katzenberg, Stable isotope analysis: a tool for studying past
diet, demography, and life history, in: M.A. Katzenberg, S.R.
Saunders (Eds.), Biological Anthropology of the Human
Skeleton, Wiley-Liss Inc, New York, 2000, pp. 305–327.
[12] M.A. Katzenberg, H.R. Krouse, Application of stable isotope variation in human tissues to problems of identification,
Canadian Society of Forensic Sciences Journal 22 (1989) 7–19.
[13] M.A. Katzenberg, H.P. Schwarcz, M. Knyf, F.J. Melbye, Stable
isotope evidence for maize horticulture and paleodiet in southern
Ontario, Canada, American Antiquity 60 (1995) 335–350.
[14] H.W. Krueger, C.H. Sullivan, Models for carbon isotope fractionation between diet and bone, in: J.R. Turnlund, P.E. Johnson
(Eds), Stable Isotopes in Nutrition. American Chemical Society,
Symposium Series No. 258, 1984, pp. 205–220.
[15] S. Manolagas, Birth and death of bone cells: Basic regulatory
mechanisms and implications for the pathogenesis and treatment
of osteoporosis, Endocrine Reviews 21 (2000) 115–137.
[16] B.D. Marino, M.B. McElroy, Isotopic composition of atmospheric CO2 inferred from carbon in C4 plant cellulose, Nature 349
(1991) 127–131.
[17] J.M. May, D.L. McLennan, The Ecology of Malnutrition in
Eastern Africa and Four Countries of Western Africa, Hafner,
New York, 1970.
[18] D.J. Minson, M.M. Ludlow, J.H. Troughton, Differences in
natural carbon isotope ratios of milk and hair from cattle grazing
in tropical and temperate pastures, Nature 256 (1975) 602.
[19] K. Nakamura, D.A. Schoeller, F.J. Winkler, H.L. Schmidt,
Geographical variation in the carbon isotope composition of
the diet and hair in contemporary man, Biomedical Mass
Spectrometry 9 (1975) 390–394.
[20] T. O’Connell, R. Hedges, M. Healey, Isotopic comparison of
hair, nail and bone: Modern analyses, Journal of Archaeological
Science 28 (2001) 1247–1255.
[21] D.A. Schoeller, M. Minagawa, R. Slater, I.R. Kaplan, Stable
isotopes of carbon, nitrogen and hydrogen in the contemporary
North American food web, Ecology of Food and Nutrition 18
(1986) 159–170.
[22] P. Rowly-Conwy, Nubia AD 0–550 and the “Islamic” agricultural
revolution: Preliminary botanical evidence from Qasr Ibrim,
Egyptian Nubia, Human Biology 55 (1989) 831–844.
[23] H. Saitoh, M. Uzuka, M. Sakamoto, Rate of hair growth, in: W.
Montagna, R.L. Dodgson (Eds.), Advances in Biology of the
Skin, Pergamon Press, Oxford, 1967, pp. 183–201.
[24] H.P. Schwarcz, Some biochemical aspects of carbon isotopic
paleodiet studies, in: S. Ambrose, M.A Katzenberg (Eds.), Biogeochemical Approaches to Paleodietary Analysis, Kluwer
Academic, New York, 2000, pp. 189–210.
[25] H.P. Schwarcz, J. Melbye, M.A. Katzenberg, Stable isotopes in
human skeletons of southern Ontario: reconstructing paleodiet,
Journal of Archaeological Science 12 (1985) 187–206.
[26] H.P. Schwarcz, M. Schoeninger, Stable isotope analyses in human
nutritional ecology, Yearbook of Physical Anthropology 34
(1991) 283–321.
[27] L.L. Tieszen, T.W. Boutton, K.G. Tesdahl, N.A. Slade, Fractionation and turnover of stable carbon isotopes in animal tissues:
implications from !13C analyses of diet, Oecologia 57 (1983)
32–37.
[28] B. Trigger, History and Settlement in Lower Nubia, Yale University Press, New Haven, 1965.
[29] C.D. White, Isotopic determination of seasonality in diet and
death from Nubian mummy hair, Journal of Archaeological
Science 20 (1993) 657–666.
[30] C.D. White, F.J. Longstaffe, K. Law, Seasonal stability and
variation in diet as reflected in human mummy tissues from the
Kharga Oasis and the Nile Valley, Palaeogeography, Palaeoclimatology, Palaeoecology 147 (1999) 209–222.
[31] C.D. White, H.P. Schwarcz, Temporal trends for stable isotopes
in Nubian mummy tissues, American Journal of Physical
Anthropology 93 (1994) 165–187.