The questions of when, where, and why horses were first
domesticated are still hotly debated. Textbooks have handily
marked the location of initial horse domestication as Dereivka
(Telegin 1986), a Copper Age settlement in Ukraine, dating
to between 4470 and 3530 BC (Rassamakin 1999: 162–163).
However, the so-called cult stallion of Dereivka, with teeth
heavily worn from clutching a metal bit, was recently radio-
carbon dated directly and found to be an Iron Age intrusion
from the first millennium BC (Anthony and Brown 2000).
When the Dereivka stallion was removed from the record as
the first individual classified as a domestic horse, both the
timing and location of this event were suddenly reopened for
Molecular studies (Vila et al. 2001, and Chapter 23, this
volume) now suggest that it is fruitless to seek a single point
of origin, since their data indicate multiple successful efforts
at horse domestication in different regions. During the
Pleistocene and Holocene, both body size and cranial mor-
phology varied in wild horse populations, depending in part
on the region, climate, and ecozone. As a result, variation also
should be expected in early domesticates across the broad zone
where the process probably occurred. If the mitochondrial
DNA studies are correct, the lack of distinctive diagnostic traits
in the early stages of domestication may be related, as well,
to frequent crossbreeding between domestic and wild stallions
Beyond the possibility of multiple independent occur-
rences, there are a number of other reasons why it has proven
so difficult to document the process of horse domestication.
Unlike dogs, cattle, sheep, goats, and pigs, it is unclear
whether the bones of early domestic horses ever developed
reliable and consistent morphological or size differences that
would allow them to be distinguished from their wild pro-
genitors. Horses are not strongly sexually dimorphic and
lack horns, like cattle, and hypertrophied canines, like pigs,
which were reduced through the process of domestication.
Documenting early horse domestication is made even
more difficult by the poor understanding of the geographic
distribution of possible wild progenitors and the complexi-
ties of mortality patterns in domestic horse populations. An
examination of previous research highlights a number of the
problems that make the archaeological documentation of
horse domestication particularly difficult. This review is
followed by a consideration of the evidence for horse domes-
tication in northern Kazakhstan that serves as a model for
Possible Wild Progenitors
The first problem that confronts researchers seeking to docu-
ment domestication in horses is determining the wild pro-
genitor of the domestic horse. The abundant fossil record of
European Pleistocene horses would seem to provide an excel-
lent data set for identifying the ancestor of the domestic
horse. However, there were many types of caballine equids
(horses and their ancestors, as opposed to zebrines, asinines,
and hemionines) in the Pleistocene, including both small and
large forms (Forsten 1988). Some are clearly separate species,
but there is considerable intraspecific and individual varia-
tion as well. For example, great diversity has been noted
among the Late Pleistocene Magdalenian populations
(Skorkowski 1956), and it is unclear whether this simply
expresses individual characteristics or reflects sympatric
subspecies (Meunier 1962).
In the past, many scholars believed that different wild
equids gave rise to different breeds of domestic horses. The
enormous draft horses, for example, were thought to have had
their origins in northern Europe, where species such as Equus
germanicus lived in the Pleistocene. However, it is now thought
that all the Pleistocene heavy horses became extinct by the
end of the Ice Age (Epstein 1971: 417). Littauer (1963) has also
shown that the titans that used to pull heavy farm wagons
were a late development, and that most of the so-called great
horses used to carry the cumbersome medieval armor in the
fourteenth and fifteenth centuries were, although sturdy,
only about 13–15 hands (130–150 cm) tall.
One of the best summaries of Pleistocene and Early
Holocene horses is provided by Forsten (1988), who argued
that, of all the different varieties of horses in the Pleistocene,
only one relatively small species of wild horse survived into
the Holocene. In her opinion, the domestic horse emerged
from that single, small progenitor, which many refer to as
Equus ferus (Nobis 1971). Current researchers generally accept
this thesis and look toward a single species with a broad
geographic distribution as the wild progenitor of the domestic
horse, instead of toward multiple lineages. As noted above,
however, it is likely that domestication occurred independ-
ently more than once within that species’ range. Two relatively
modern species of caballine equids are normally used for
comparison with possible early domesticates: the tarpan, or
wild European horse, and the Przewalski, or wild Asiatic
horse. The tarpan (Equus ferus) was recorded in historic times
in northern Germany, Lithuania, Poland, Ukraine, and
western Russia, but may have been much more widespread
Early Horse Domestication on the Eurasian Steppe
SANDRA L. OLSEN
EARLY HORSE DOMESTICATION ON THE STEPPE 245
chapter17 2005.12.16 9:24 PM Page 245
the domestic horse. Recent mtDNA research supports this con-
clusion (Vila et al. 2001). The tarpan, as problematic as it is,
is still the best candidate for the ancestor of the domestic
There is a long history of efforts to document the domesti-
cation of the horse in the archaeological record, using a wide
range of conventional and unconventional criteria. Some of
the most significant lines of re.search are discussed below.
The classification of the crania of Late Pleistocene and
Holocene wild horses is quite problematic. Different authors
classify the fossil and subfossil remains in different ways, and
each has his or her own interpretation of which group of fossil
horses gave rise to domestic horses (see Epstein 1971 for a
comprehensive summary of the early attempts to classify
Some of the first work on cranial morphology of recent wild
horses was done by Duerst (1908), Ewart (1907), and Antonius
(1913). Whereas these researchers all classified horses into
groups that were assigned environmental terms (e.g., steppe,
desert or plateau, and forest), the basis for grouping was
actually cranial morphology rather than habitat. There was
no general consensus regarding groupings of these horse
remains, and often more than one type could be found in
the same locality (Epstein 1971). Some groups were merely
hypothetical (see Groves 1986) and lacked any actual
specimens. The implication was still quite strong, however,
that local populations diverged because of their unique
environment and climate.
Undoubtedly, difficulties in the systematic classification of
horses arise when comparisons are drawn among the handful
of remains representing an enormous geographic area,
from western Europe to Siberia, that include a wide range of
habitats and a long temporal span. However, even within an
assemblage from a single locality, multiple forms have been
detected, and these are often interpreted as representing
sympatric subspecies (Meunier 1962). When variation
resulting from age, sex, and individual characteristics is added
into the mix, classification becomes rather daunting. Despite
the variety of forms detected by researchers studying cranial
morphology, it is important to remember that, despite the
enormous size range of modern horse breeds (withers heights
ranging from 60 to 200 cm), their cranial morphology is not
as diverse as that seen in dogs or pigs (Epstein 1971; Wayne
The tarpan, according to Lundholm (1949), could be
generally distinguished from the wild Asiatic horse by its
flatter forehead, wavy cranial profile, short and low nose,
supraorbitals vaulted above the forehead, a depression
246 ARCHAEOLOGY AND ANIMAL DOMESTICATION
in the past. It was described in numerous references from the
eighteenth and nineteenth centuries as a small animal,
having a mouse-dun coat with a light underbelly, sooty to
black limbs from the knees and hocks down, a short, frizzled
mane, and a tail with short dock hair. In fact, in most fea-
tures the tarpan was very similar in appearance to the
Przewalski horse (E. przewalskii), except that the coat was
grayer and apparently turned very light in the winter. Russian
zoologist Heptner reported that the last tarpan died in 1918
or 1919 in captivity (Bökönyi 1974a: 71). In the wild, the last
tarpans were killed in Ukraine in 1851 (Zeuner 1963: 305).
The only available skeletal material attributed to the tarpan
consists of one complete skeleton and a cranium lacking a
mandible from another individual, housed in the Zoological
Institute in St. Petersburg, Russia. There are so many accounts
of tarpans stealing domestic mares and forming harems
(Bökönyi 1974a: 72), however, that recent specimens should
not be considered genetically pure. Many populations attri-
buted to the tarpan in historic accounts may simply have
been feral horses or hybrids. The erect mane tends to be the
most reliable “wild” characteristic, but hybrids can also have
fairly short manes.
The wild Asiatic or Przewalski horse is much better known,
with extant populations now under human control (Mohr
1971; Bökönyi 1974a; Boyd and Houpt 1994). Existing
Przewalski horses were originally derived from 12 individuals
from the Djungar Basin, a barren desert of western Mongolia
and China, but interbreeding with domestic horses in cap-
tivity has tainted their gene pool to some degree. Like the
tarpan, Przewalski stallions were notorious for stealing mares
from Mongol horse herds. The modern herds of Mongolian
horses also suggest that hybridization occurred in the other
direction as well, perhaps through the capture and adoption
of Przewalski horses in the past. Horses with dun coats, zebra
striping on the legs, and prominent eel stripes down their
backs are very common today in domestic herds in Mongolia.
Their artificially clipped manes give them an even stronger
resemblance to the wild form.
Today, few would claim that the Przewalski horse is the
ancestor of the domestic horse. Although Przewalski horses
can interbreed with domestic individuals and produce viable
offspring, they have a different number of chromosomes (in
Equus przewalskii, 2n = 66 and in E. caballus, 2n = 64, and in
hybrid E. przewalskii x E. caballus, 2n = 65), reflecting what
is known in genetics as a Robertsonian translocation (Ryder
1994). Under these circumstances, two chromosomes break
and their two long arms fuse to form a new, large chromo-
some containing essential genes, while the two short arms,
which apparently do not contain significant genes, either fuse
with each other or are lost without joining (Mange and
Mange 1980). Although there are examples of this type of
chromosomal variation at the conspecific level in mam-
malian taxa, it usually occurs between two closely related
species. In light of this evidence, most experts would agree
that the Przewalski is a sister species to the ancestor of
chapter17 2005.12.16 9:24 PM Page 246
EARLY HORSE DOMESTICATION ON THE STEPPE 247
between the orbits, and another depression between the
middle and anterior third of the nasal. Bökönyi (1974a: 72)
remarked that the tarpan’s skull was “broad and spacious, with
orbits rising above the plain of the forehead and with a
markedly short facial part.” It is important to note that the
reconstituted “tarpans” found in the forest of Bialowieza, in
present-day Poland, may be derived from domestic Polish
Konik horses and should never be considered reliable for
metric analysis. Those in the St. Petersburg collections that
are used by most scholars, including Lundholm and Bökönyi,
are to date our best representations of true tarpans.
Eisenmann (1996: 28, 30) demonstrated that the Przewalski
horse has a relatively shorter Vomer-Basion distance (a mea-
surement reflecting the length of the basicranium) than
“tarpans,” Arabian horses, wild fossil horses, and subfossil
domestic horses, but resembles the kiang (a wild Asiatic
hemione) in this character. Interestingly, Lundholm (1949)
has shown that there are significant craniometric differences
between the captured wild Przewalski horses and those raised
in zoos. The zoo specimens exhibited a decrease in basal
skull length and tooth row length, as well as in cranial and
snout width, but postorbital width did not change (Forsten
1988). These alterations may provide clues to possible changes
that would have occurred in Equus ferus under captivity, but
also complicate comparisons between modern Przewalski
skulls, which are mostly from zoo specimens, and archaeo-
The ideal comparative cranial material would be the skulls
of Equus ferus from the Mesolithic and Neolithic of the forest
steppe of Ukraine, Russia, and northern Kazakhstan.
Unfortunately, however, there are few reasonably complete
specimens available from these periods (Kosintsev in press).
The ephemeral nature of shallow campsites of the earlier
periods generally does not lend itself to the preservation of
crania. Most complete crania are derived from horse sacrifices
in pits and burials dating to the Eneolithic or later, presumably
after domestication had begun.
In order to understand better the craniomorphological
changes that have occurred through the process of domesti-
cation, a thorough morphometric analysis should be con-
ducted first on the rather large collections of Bronze and
Iron Age horse skulls from kurgan burials in the former Soviet
Union. Some work has been done in this direction already by
Vitt (1952), Bibikova (1986a), Kuzmina (1997), Eisenmann
(1996), and others. Eisenmann (1996) perhaps best points out
the problems with any osteometric analysis of caballine
equids when she says that no distinct phylogenetic or taxo-
nomic patterns can be discerned in Quaternary horses,
whether based on adaptive or nonadaptive characters. Instead,
traits seem to combine in a “reticular pattern, a mosaic of
characters that are combined at different places and times in
different ways” (Eisenmann 1996: 30).
Bökönyi (1974b: 236) maintained that domestication
brought on few changes in the horse’s cranium, but he listed
some features that may be important. He cited a decrease
in cranial capacity, a broadening of the forehead, facial
foreshortening, a narrowing of the muzzle, and a decrease in
tooth dimensions. It is unclear when these cranial changes
first became evident, although many claim to have observed
differences in Iron Age horses (Vitt 1952; Bibikova 1986a).
Bibikova (1986a) performed a detailed study on the
Dereivka cult stallion, which is now known to date to the Iron
Age. Because it is clearly a domesticate based on its date and
well-developed bit wear, it is interesting to see how it com-
pares to other horses of the region, both wild and domestic.
The Dereivka skull showed many of the traits that Bökönyi
maintained were the products of domestication: a short,
broad brain case, a broad forehead, a broad snout, and a
maxillary cheek tooth row shorter than that of Przewalski
horses and the “tarpan.” But while Bökönyi predicted facial
foreshortening, the Dereivka skull had a long narrow face. It
should be pointed out that the face of a gelding tends to be
longer and narrower, and often has a more convex profile than
that of a stallion (Littauer 1971: 294). Geldings normally do
develop the standard, large male canine teeth like those in
the Dereivka skull, so it is entirely possible that the Iron Age
Dereivka skull could have been derived from a castrated indi-
vidual. Thus, except for the long, narrow face, the Dereivka
skull does support Bökönyi’s characters for domestic horses
and is comparable to the other late archaeological horses
and the modern domestic Kyrgyz breed.
The short, but broad, brain case and the broad forehead may
be useful domestic trends, although there is no proof yet that
these characteristics are seen consistently in other domestic
horses. Facial length and muzzle width may vary in domestic
horses as breeds develop, perhaps as early as the Iron Age. It
is obvious, then, that there is a range of variation in both wild
and domestic animals, and, in most characters, there is
Although individual tooth size and cheek tooth row length
may be useful for distinguishing the domestic horse from the
Przewalski horse, it is unclear whether dental metrics can
differentiate early domesticates from their immediate pro-
genitors. Dental size variation is related to age and occlusal
wear patterns, as well as to genetics. The cheek tooth row can
be noticeably short in older individuals, as it is in one of the
two existing tarpan skulls, so it is wise to avoid using this
measurement in senile samples. The enamel patterns of cheek
teeth are generally similar in all populations of caballine
horses (Forsten 1988), and what variation there is often may
be attributed to individual age and wear patterns.
Despite the lack of clear morphological criteria for distin-
guishing wild from domestic horses, many archaeologists
working in Ukraine, Russia, and Kazakhstan claim to be
able to separate domestic horses from their untamed ances-
tors. Sometimes these claims are based on metric analysis,
but sample sizes for any given site are usually very small, and
chapter17 2005.12.16 9:24 PM Page 247
there is little consistency in how body-size variations are
interpreted. Benecke (1994), for example, maintains that an
increase in variation in body size should be considered a
marker of domestication in horses, similar to that seen in the
probable domestic populations from Kazakhstan and Russia
dating to the Bronze and Iron ages (Akhinzhanov et al. 1992;
Kosintsev in press). Uerpmann (1990), on the other hand,
believes that a similar degree of variation in body size among
Neolithic equine populations in fourth millennium Germany
falls within the natural limits of a wild population.
As with skulls, however, postcranial remains of E. ferus
from Mesolithic and Early Neolithic contexts in the Ukraine,
Russia, and Kazakhstan are extremely rare (Kosintsev in press).
Thus, there is only a limited sample of remains from definitely
wild Early to Middle Holocene specimens from the forest
steppe available for comparison with later, possibly domestic
Another complicating factor is that during the initial phases
of domestication, the middens of sites within the natural
geographic range of E. ferus very likely contained remains
of both domestic livestock and wild hunted horses. It would
be difficult to attribute any observed variation in these
small assemblages to a difference between wild and domestic
horses, however, given the wide range of ages likely to be
represented and size differences among stallions, geldings, and
To muddy the waters further, a temporal shift in size can
occur under natural conditions as a result of changes in mean
annual temperature and precipitation, alterations in season
length and intensity, and other factors related to the change
in climatic conditions known to have occurred throughout
the Holocene (see Guthrie 2003). Eisenmann (1984) has sug-
gested that there is a close link between bone proportions,
especially in the feet, and the local environment (e.g., open
vs. closed country, hard vs. soft ground, and dry vs. humid
climate). Bökönyi (1974b: 242) saw a correlation between
climate change in Bronze Age Hungary and a shift from
horses with narrow hoofs to stockier horses with broad hoofs
as the climate became cooler and wetter. Clearly, determining
whether local size changes reflect adaptation of wild popu-
lations to changing environments or the development or
introduction of a new breed of domestic horses has proven
to be complicated.
Because the early domestic horses must have relied either
solely or primarily on natural vegetation, body size in early
domesticates also may have responded to natural differences
in available pasturage across the wide geographic range in
which horses were kept. If the herds were fed domestic grain
as fodder through the winter, however, then it is possible
that they could have grown to greater size. Today, Kazakh
and Mongolian pastoralists continue to graze their herds
on natural vegetation through the winter and to gather
wild grasses for fodder only in the most critical times.
Significant increase in the body size of these animals would
make horses less efficient grazers in the forest-steppe and
would decrease the sustainable herd size. On the other hand,
significant reduction in body size related to nutrition would
hamper endurance and speed in riding and lessen their
capacity to carry heavy loads. Maintaining body volume is
also advantageous given the severe winters of the region.
Mortality patterns, so helpful in discerning early herd man-
agement in other domestic herbivores (Redding 1981, Zeder,
Chapter 14, this volume), are more complicated in horses. For
most livestock, there is a strong incentive for culling young
males for consumption as soon as they have obtained most
of their adult volume in order to reduce the investment in
grazing land and fodder for these animals. In addition, it is
much easier to control herds and to carry out genetic selec-
tion if only a few of the best mature males are kept for
breeding purposes. The adult females (needed to breed, rear
the young, and provide milk) significantly outnumber the
adult males in a typical livestock herd. Under such a
regime, livestock management is signaled in the archaeolo-
gical record by high numbers of juvenile males in the death
assemblage and prolonged female survivorship. There are
several serious methodological impediments to using this
marker to document horse domestication.
Although the bodies of males tend to be slightly bigger than
those of females, the degree of size-linked sexual dimorphism
in horses is less strongly expressed than it is in other livestock
species. Clear bimodality between the sizes of male and
female horses can only be expected in metric analyses con-
ducted on unusually large collections. Determining sex in
horse remains, then, is based on two characters: the presence
of large canines in males and distinctive differences in the
shape of the pelvis in male and female animals. In the stal-
lion, the internal surface of the pubis is convex, whereas it
is concave in the mare and in horses gelded very young
(Riegel and Halola 1996: 152). The presence of robust canines
can be used to distinguish adult stallions and geldings from
females that either have no canines or only very small ves-
tigial canines. This feature, however, is of no use in sexing
juvenile specimens since there are no deciduous canines.
Generally, sexing by dentition is restricted to individuals
over the age of 4 or 5 years, after the canines have erupted
(except that this author has observed that some females may
obtain their diminutive canines as early as 2.5 years). This
means that it is not possible to determine how many juve-
nile males vs. females were culled based on dentition. We can,
however, look at the overall percentage of juveniles in the
assemblage and the ratio of adult males to females.
Ageing horse remains can also be problematic. Most ageing
schemes are based on patterns of dental eruption and wear.
Because most eruption tables have been constructed by
biologists looking at living horses (Silver 1963; Getty 1975;
Habermehl 1975), the ages of tooth eruption given are gene-
rally when the tooth first erupts through the gums, and not
248 ARCHAEOLOGY AND ANIMAL DOMESTICATION
chapter17 2005.12.16 9:24 PM Page 248
when it erupts through the bone, and thus will tend to over-
estimate age at death in archaeological specimens. There are
also problems with some of the common methods for com-
puting age at death based on the crown heights of cheek teeth,
a technique often used to age horses from archaeological
sites (see Levine 1982). First, there are significant questions
about the accuracy and resolution of estimated ages based on
crown-height methods (Brown and Anthony 1998: 338–339).
Moreover, the practice of treating both whole jaws and each
isolated tooth as separate specimens risks counting one indi-
vidual several times. Another inherent problem in dealing
with isolated teeth is that it is extremely difficult to distin-
guish between third and fourth premolars or first and second
molars. At times, it even can be hard to separate any of these
four teeth from the others. If the problematic cheek teeth are
lumped together, then the age determination for each tooth
is quite a large span of years, but if they are excluded from
the sample, then considerable data will be missing.
Calculating crown heights on isolated P3–4 and M1–2 is sub-
ject to errors amounting to three or more years.
It is also difficult to reconstruct reliable mortality profiles
for horses in Eurasian steppe sites because sample sizes are
often quite small. Moreover, even if the assemblages merit
ageing, the results from one site are often not comparable to
the next, either because one zooarchaeologist group ages dif-
ferently from another or the dates and locations of the sites
are quite different. Since domesticated horses seem to have
arrived at different places at different times and may have been
used in different ways with varying economies, it is not easy
to combine results from all sites across the Eurasian steppes
The use of mortality data in identifying horse domestica-
tion is further complicated by the complex and shifting
demographic patterns in wild horse herds and the diverse
exploitation strategies practiced by both horse hunters and
herders. Wild and feral horses form two types of social groups
(Berger 1986). The primary group is a family band composed
of a stallion, his harem of mares, and their offspring. Both sons
and daughters normally leave their natal band between the
ages of one and three years (Berger 1986: 129). The second
social group among wild horses consists of two or more males
that are too old to stay with their families but have not suc-
ceeded in acquiring mares. Bachelor groups are not stable and
vary through the year. The typical size is around 4 individuals,
although bachelor groups have been reported to have as
many as 17 individuals at a time (Berger 1986: 131).
It is expected that a communal hunting strategy normally
would focus on family bands because they contain more
individuals and are more cohesive than the bachelor groups.
Stalking by one or two hunters would be more appropriate
for dispatching single bachelors, or those in small, loosely
affiliated groups. Most communal hunts would be likely
to yield a high percentage of both male and female juveniles,
a relatively large number of mares, and one stallion.
Unfortunately, the typical harvest pattern for most livestock
produces a similar pattern: high numbers of juveniles (mostly
males, which could not be distinguished using archaeologi-
cal remains), relatively high numbers of adult females, and
low numbers of adult males.
On the other hand, there are many reasons why a high fre-
quency of juvenile males, as expected in most livestock
culling strategies, might not be found in a mortality profile
in an assemblage of domestic horses. First, it is clear from the
frozen horse mummies at Pazyryk (Rudenko 1970) that, at
least by the Iron Age, castration was a common practice.
This would eliminate some of the need to cull juvenile males,
especially if geldings were as highly valued as Rudenko indi-
cates. The question that remains is when this practice began
to be employed for horses. If horse domestication started in
the Eurasian steppe after people in this region had received
domestic cattle and sheep from the Near East, then they
would probably have been familiar with the practice of
Second, for horses, adult males take on important roles
beyond that of breeding. In addition to providing meat,
milk, bones, and hides, at some point in their history horses
were used for riding, carrying loads, pulling wheeled vehicles,
and plowing (although the use of horses as draft animals was
probably considerably delayed compared to other uses). Males
would almost certainly be preferred in all of the nonfood uses
of horses. Despite Rudenko’s (1970) comments on the pre-
ference for geldings over stallions, there is evidence that stal-
lions were highly prized. Even a cursory examination of
classical depictions of cavalry horses and chariot horses on
pottery, metal vessels, and the like from Scythian, Greek,
Roman, and Egyptian art collections will reveal a strong pre-
ference for intact stallions. Littauer and Crouwel (1984) have
also remarked on the predominance of stallions in the depic-
tions of ancient Near Eastern chariot teams. Mongolians
today prefer to ride stallions when herding or hunting.
Adult males may also be better represented than expected
in faunal assemblages in the Eurasian steppes after domesti-
cation began because of the practice of ritual horse sacrifice,
which seems to have focused primarily on males in some
regions. This practice is most likely correlated with the value
of males as mounts from an early date onward and later for
pulling chariots and other vehicles. Sacrificial remains are
common in human burials from the Copper Age through the
Iron Age and in deep domestic pits within Copper Age settle-
ments. In these contexts, preservation is reasonably good
compared to open-air middens containing animals slaugh-
tered for food. For the Bronze and Iron ages, excavators have
concentrated heavily on human burials containing animal
sacrifices in preference over middens, artificially boosting
the ratio of adult males in the faunal collections. The discus-
sion of the Copper Age Botai culture below will demonstrate
that even near the inception of horse domestication, there
was a preference for the use of stallions in ritual deposits.
Kosintsev (in press) has recently gathered fairly comprehen-
sive data on mortality patterns from Bronze and Iron Age sites
EARLY HORSE DOMESTICATION ON THE STEPPE 249
chapter17 2005.12.16 9:24 PM Page 249
on the Eurasian steppe (Table 17.1). Since horse domestica-
tion was certainly accomplished by this time, these mortality
patterns provide a useful point of comparison with earlier sites
where the status of the horses as wild or domestic is still open
to question. Early Bronze Age (2500–2920 BC calibrated)
mortality patterns from the site of Sergeevka, Kazakhstan
(Levine and Kislenko 1997), indicate a relatively low mortality
rate for animals in their first year of life (around 6%).
Slaughter there focuses instead on animals in the 1–5-year
range and the 5–12-year range. The mortality rate for horses
below 1 year of age increases dramatically in later sites to a
mean figure of about 29%. At the same time, there is a rela-
tive decrease (from 29%down to 15%) in mortality for ani-
mals in the 1–5-year range in later sites. The frequencies of
animals in the older 5–12 and 12+ age groups remain about
the same through time.
The high mortality rate of very young animals in the later
Eurasian sites is not consistent with natural mortality rates
in wild and feral herds. The annual mortality rates for feral
horses under one year of age in the United States range from
about 8% to 14%, with an unusually high mortality of 25%
reported for only one locality (Berger 1986: 82, Keiper and
Houpt 1984, Feist and McCullough 1975, Welsh 1975).
The high mortality rate of horses from the Middle Bronze
Age through the Iron Age is also not consistent with modern
pastoral slaughter strategies where the ideal is to slaughter
2.5–year-old animals in the late autumn or early winter.
While adult weight is reached at about 5 years of age in
horses, there is little difference between the weight of a 3-year-
old animal and a 5 year old. At 2.5 years, a horse on the
Eurasian steppe has reached most of its adult weight (about
90%), and in the autumn its fat content is relatively high, a
crucial factor for people dwelling in regions with harsh, pro-
longed winters. The meat of animals slaughtered at this time
can also be readily frozen and kept through early spring.
Horses killed before their first birthday, in contrast, have
only reached 12–50% of their adult meat weight.
250 ARCHAEOLOGY AND ANIMAL DOMESTICATION
Mortality Profiles for Bronze and Iron Age Horses
Early Bronze Age Middle-Late Bronze/
Iron Age Sites
% Killed % Killed
Age % Killed Mean Range
0–1 yr 6 29 13–35
1–5 yr 29 15 8–20
5–12 yr 58 47 35–60
12+ yr 7 9 2–17
SOURCE: Kosintsev (in press).
Sintashta, Srubnaya, Alakul Fedorovo, and western Siberia.
It is possible that the high mortality of neonates and year-
lings in the Bronze and Iron ages is a signal of a distressed
herd management system. It may reflect the harvesting of
half-grown horses out of necessity or the loss of young horses
in the winter because of severe ice storms, heavy snows,
droughts, and epidemics, all known to cause catastrophic
deaths of young animals in modern Kazakh herds. Rice (1963)
has reported that ice storms recur every 10–12 years in
northern and central Kazakhstan, killing nearly half of all live-
stock. In the particularly harsh winter of 1891–1892, for
example, 47% of the horses, 32% of the cattle, sheep, and
goats, and 22% of the camels were lost. Young animals are
especially vulnerable to harsh weather, although they are
frequently taken indoors for protection. But it seems unlikely
that catastrophic loss of young animals can account for the
persistent and widespread pattern of young animals in sites
dating to the Middle-Late Bronze Age and Iron Age in the
Alternatively, this pattern may be attributable to an increase
in the importance of mare’s milk during this later period.
Fermented mare’s milk, also known as koumiss (Kazakh) or
airag (Mongolian), has very high vitamin content and may
well have been greatly prized by Bronze and Iron age peoples.
Production of mare’s milk is most plentiful in the summer
months from May to August but diminishes in the autumn.
If mares were bred to maximize milk production, foals would
have become expendable when milk production decreased
in the autumn. Foals may not have reached full adult body
weight, but their slaughter at this time would reduce the
problems of keeping more horses than necessary through the
winter. Early slaughter of young animals might also reduce
birth spacing so that a second foal could be conceived soon
afterward. This would maximize koumiss production.
It is interesting to compare these patterns to that reported
by Levine (1999a) for the Copper Age site of Dereivka
(Table 17.2). Levine has argued that the mortality pattern for
these horses shows an emphasis on older animals that is
consistent with hunting wild horses, particularly a stalking
strategy that focuses on animals of prime age. However,
judging from the graphical representation of horse mortality
patterns at Dereivka provided by Levine (1993: Figure 12), it
would seem that about 24–25% of the horses were killed
before five years of age, and another 50% were killed between
Mortality Profile for Copper Age Botai Culture Horses
Age % Killed at Dereivka
0–5 yr 25
5–8 yr 50
8–16 yr 20
16+ yr 5
SOURCE: Levine (1993, 1999a).
chapter17 2005.12.16 9:24 PM Page 250
five and eight years of age. This pattern is consistent with that
observed in an earlier study of a larger sample of the Dereivka
material by Bibikova (1986b), who reported that young horses
constitute about 23% of the Dereivka horses. Noting the
similarity between the Dereivka assemblage and modern
Mongol herds, where individuals in the one-to-two-year age
range constitute about 26% of the total domestic horse stock,
Bibikova felt the mortality pattern at this Copper Age settle-
ment was consistent with a domestic herd.
It is difficult to directly compare the Dereivka mortality
patterns with those reported by Kosintsev for later sites since
different age classes were used to construct these profiles.
However, there do appear to be strong similarities between
the mortality profiles at all these sites. At all sites there is an
emphasis on animals below 5 years of age, with only 25% of
the horses surviving beyond 8 years of age at Dereivka and
fewer than 10% surviving beyond 12 years at the Bronze and
Iron Age sites.
Levine also bases her argument for the wild status of the
Dereivka horses on the predominance of males in the sexable
jaws at the site. Nine out of 10 sexable jaws studied by Levine
were males, a figure similar to that found by Bibikova in her
study of a larger sample from the site (15 of 17 sexable jaws).
This emphasis on males is interpreted by Levine as evidence
for the hunting of bachelor herds or the hunting of “inex-
perienced” stallions defending harems. It is quite possible,
however, that the majority of these jaws belonged to well-
preserved crania from sacrificial stallions or geldings purpose-
fully buried in ritual pits, similar to ritual practices seen in
Copper Age contexts in Kazakhstan (Olsen 2003). Crania of
horses slaughtered for subsistence probably would not be as
well represented as these carefully buried specimens, since
they might have been broken open to extract brains or dis-
carded in open-air middens. Thus, the predominance of
males in the Dereivka assemblage observed by Levine may
actually be a reflection of ritual deposits and not of general
Geographic Range and the Spread of Domestic Horses
One promising avenue that will require additional work in
the future is the reconstruction of the geographic distribu-
tion of the probable wild progenitor, Equus ferus, during the
Holocene (Bökönyi 1974b: 236; Grigson 1993). If it can be
shown convincingly that wild horses became extinct in
certain regions of Europe and the Near East by the Middle
Holocene, then the reappearance of horses in these regions
could signal the diffusion of domestic horses outward from
their place (or places) of origin in the Eurasian steppe.
Tracking the distribution of horse remains in these marginal
areas, then, could help bracket the period of initial horse
There are, however, some serious impediments to using
zoogeographic evidence in this way. First, there is consider-
able disagreement about the boundaries of the range of wild
EARLY HORSE DOMESTICATION ON THE STEPPE 251
horses in the Early to Middle Holocene. The data needed to
reconstruct these boundaries have yet to be assembled, and
there are many gaps in our knowledge. Second, many of the
horse remains attributed to Mesolithic and Neolithic contexts
come from very early excavations in which there is a strong
possibility of modern intrusions. Recent efforts to directly date
some of these remains using small sample radiocarbon dating
techniques certainly will clarify the temporal placement of
these specimens. However, even in well-dated assemblages,
it is hard to assess whether the sudden reappearance of horses
in some regions after a significant hiatus reflects the intro-
duction of domestic stock or the expansion of small, relict
wild populations resulting from climate change.
Nevertheless, it is useful to summarize the information
available on some of the key finds in the Near East and
Europe (Table 17.3). In the Near East, a single possible wild
horse specimen was dated to between 7400 and 5000 BC at
the site of Abu Hureyra in northern Syria (Legge and Rowley-
Conwy 2000: 429; Moore 2000: 259). Horse remains were also
found in the somewhat later site of Tappeh Zageh in north
central Iran at about 5000 BC (Mashkour et al. 1999). The
likely domestic horse does not appear in Mesopotamia until
about the mid-third millennium BC (Grigson 1993) and
about 500 years later in southern Iran (Zeder 1986).
Horse remains have been recovered from a number of
sites in the Caucasus dated between 5500 and 4000 BC.
(Anthony 1996: 72) but do not appear in Anatolia until the
late Chalcolithic period (4000–3000 BC) (Anthony 1996;
Boessneck and von den Driesch 1976). Although Boessneck
and von den Driesch (1976) maintained that these animals
belonged to relict wild populations, Bökönyi (1991) argued
that they represented domestic horses. Equally problematic
is the status of the isolated but indisputable horse remains
found in a number of Chalcolithic and Early Bronze Age
contexts in the southern Levant (Grigson 1993; Davis 1976;
Lernau 1978; Zeder, conversation with Olsen, September
There are similar problems with the attribution of either
wild or domestic status to the horse remains from western
Europe and western Central Europe. Horse remains are rare
in this region from the end of the Pleistocene until around
3500 BC. However, there seem to be sparse relict populations
of wild horses in the Early Holocene from Hungary to
England, as shown in Table 17.3. Lundholm (1949: 171) had
no record of them in the Mesolithic and Neolithic in Italy
or in the Balkan Peninsula, and this is still true today.
Greenfield’s (in press) work also supports a lack of horses
before the Bronze Age in the Balkans. It is particularly diffi-
cult to know how to interpret the status of horses found in
small numbers in Mesolithic and Neolithic sites in Central
and western Europe. In many, if not most, cases, attribution
of either wild or domestic status to these specimens is based
solely on their temporal placement (rarely fixed through
direct dating) and the preconceived ideas of the analyst
about the likely date of horse domestication. The domestic
chapter17 2005.12.16 9:24 PM Page 251
horse is thought to have reached western Europe surprisingly
slowly, as attested by directly dated horse remains appearing
in England at about 1700 BC (Clutton-Brock and Burleigh
1991a) and the first arrival of the horse in Ireland at about
1900 BC (Wijngaarden-Bakker 1974).
The gradual succession of forests in Europe with the
warming trend in the Early to Middle Holocene (Flint 1957:
286; Butzer 1971: 532) may have been a key factor in the
reduction of horse populations there, as areas previously
252 ARCHAEOLOGY AND ANIMAL DOMESTICATION
consisting of open steppe began to close up. Similarly, some
have suggested that the later return of horses into these areas
was the result of climate change rather than the diffusion of
domestic horses (Steppan 1998). Recent pollen evidence
from western Siberia to eastern Europe supports the idea
that the period of initial domestication and diffusion of
horses was marked by considerable climatic change
(Kremenetski et al. 1997). These data indicate climatic con-
ditions from about 4000–2500 BC, a time period that likely
Selection of Early Horse Remains from Sites in the Near East and Europe
Region Sites BC BC References
Syria Abu Hureyra 7400–5000 Legge and Rowley-Conwy 2000
Tell Brak 2300 Clutton Brock and Davis 1993
Negev Shiqmim 3240 ± 75 Grigson 1993
Central Iran Tappeh Zagheh 5000 Mashkour et al. 1999
Southern Iran Malyan 1800 Zeder 1991
Anatolia Malatya, Elazig, 3300–3000 Anthony 1996
Norsun Tepe 3000–2500 Boessneck and von den Driesch
1976; Bökönyi 1991
Egypt Buhen 1675 Clutton-Brock 1974
Caucasus Arukhlo, Tsopi, Alimeklek 5500–4000 Anthony 1996
Hungary Many Neolithic Sites 6000–5000 Vörös 1981
Denmark Ertebölle/Ellerbek 5500–4000 Davidsen 1978
Braband So 3550 ± 75 Davidsen 1978
Lindskov 2570 ± 65 Davidsen 1978
Netherlands Swifterbant 3525–3280 Clason 1991
No. Germany/ Poland Potsdam-Schlaatz 9390–9250 Benecke in press
Gaukönigshofen 7960–7760 Benecke in press
Hohen Viecheln 5500 Gehl 1961
Rüde, Bondebrück 4000–3500 Lüttschwager 1967
Dąbki 9 4940–4800 Benecke in press
Eilsleben 4830–4720 Benecke in press
Rosenhof 4720–4540 Benecke in press
Siniarzewo 1 4430–4250 Benecke in press
Southern Germany Bruchsal-Aue 4236 ± 64— Steppan in press
3529 ± 74
France Charcognier 7000–5500 Bignon, O. e-mail to Olsen 5/16/2003
Bercy 3900 Bignon, O. e-mail to Olsen 5/16/2003
Italy Le Cerquiete-Fianello 2510 ± 110 Curci and Tagliacozzo 1995
Querciola-Sesto Fiorentino 2130 ± 150, Corridi and Sarti 1989/90
1690 ± 200
England Seamer Carr 7750 ± 180 Clutton-Brock and Burleigh 1991b
Grimes Graves 1690 ± 210 Clutton-Brock and Burleigh 1991a
Ireland Newgrange 1900 Wijngaarden-Bakker 1974
chapter17 2005.12.16 9:24 PM Page 252
witnessed horse domestication in Eurasia, were both wetter
and less continental than either preceding or subsequent
periods. The climate became colder and drier from about 2500
to 1500 BC, coinciding with the likely development of
nomadic pastoralism on the Eurasian steppe. It is difficult,
then, to distinguish the diffusion of domestic horses out of
the Eurasian steppe and into Central and western Europe from
climate-induced shifts in the zoogeography of wild popula-
tions. It is also possible that climatic change played a role in
the spread of equine husbandry.
Increase in Frequencies of Horse Remains
Bibikova (1986b) has proposed that domestication of horses
will be signaled by a marked increase in the ratio of horse ele-
ments to the overall number of ungulate remains, whether
domestic or wild. The use of abundance data to indicate
domestication in any species must be undertaken with care,
however. Natural changes in climate or local environment can
lead to shifts in wild species’ frequencies in a region, inde-
pendent of any advances in animal control. It is important
to remember that even Paleolithic sites can have a high con-
centration of one species as a result of selective hunting
practices or seasonal herd migration patterns, as at the Upper
Paleolithic kill site of Solutré (Olsen 1989, 1995), where most
of the faunal remains are from horses. On its own, then, a
sizable ratio of a particular animal in the assemblage does not
constitute proof of domestication. Instead, this information
can provide support for other more direct lines of evidence
or contribute to the amassed secondary evidence. This is
especially the case when abundance patterns are examined
across larger geographic and temporal ranges.
Bibikova used Neolithic or Copper Age (Eneolithic) sites in
Romania, Moldova, Ukraine, and western Russia to examine
trends in the abundance of horse remains as a possible indi-
cator of domestication. Although sample sizes for most of the
sites are small, the overall pattern is very informative. In her
Western Zone (from Bucharest to the Dnepr River), prior to
the Late Tripolye period (ca. 2500 BC calibrated) the ratio of
horses was generally well below 15%. In her Eastern Zone
(from the Dnepr to the southern Ural Mountains), the per-
centages were consistently higher. Most significantly, at
EARLY HORSE DOMESTICATION ON THE STEPPE 253
Dereivka, horses made up 63.27%, and at the Early Yamnaya
(Early Bronze Age, ca. 3500 BC) settlement of Repin, on the
Don River, the percentage soared to 79.8% (Figure 17.1). The
sample sizes are relatively good for these two sites. At
Dereivka, the total number of ungulate bones is 3,564, 2,255
of which are horse bones (Bibikova 1986b: 169). The mini-
mum number of horses is given as 44 individuals (Bibikova
1986b: 174), however, the methods for calculating MNI may
differ from the standard procedure used in the West. At
Repin, 652 of the 817 ungulate bone fragments are from
horses. As we shall see, even farther to the east, in northern
Kazakhstan, the numbers reach 99% at the large settlement
of Botai, 3700–3100 BC (calibrated). There are no domestic
sheep or cattle in Copper Age deposits at Botai sites, which
represent the greatest dependence on horsemeat of any
culture through time. These temporal (fourth millennium)
and geographic trends (eastward to Kazakhstan) toward an
increase in proportions of horses almost certainly reflect the
presence of equine livestock. Once nomadic pastoralism
involving cattle and sheep takes hold in the Middle Bronze
Age, however, the percentages of horses diminish.
Artifactual and Architectural Evidence
Evidence for early horse domestication might also include
horse tackle, corrals, and other cultural indicators of horse
control. The most cited early evidence of possible horse
tackle consists of six perforated antler tines from Dereivka,
in Ukraine (Telegin 1986: 82–83). Five of these pieces have
one hole and one has two holes, resembling later bridle
cheek pieces from European Bronze Age sites (Clark 1941;
Britnell 1976; Longley 1980). Late TRB (Funnel Beaker) sites
in Germany produced single-holed antler tine artifacts
similar to those from Dereivka (Telegin 1986: 85). The TRB
culture extended from Ukraine to the Netherlands and dates
from 4350 to 2670 BC (Milisauskas 1978). Antler cheek
pieces are found with Iron Age horse mummies at Pazyryk
(ca. fourth century BC) that are similar in dimension to the
Dereivka perforated antler tines, although they are somewhat
thinner. Dereivka also produced a number of perforated
antler artifacts made from the beam (or base) of the antler
that were clearly hafted to wooden shafts and used as tools,
so it is difficult to say that the perforated antler tine artifacts
from the site were bridle cheek pieces. Unlike these other
antler artifacts, however, the perforations in the antler tine
pieces were only about .4 to 1 cm in diameter and appear to
have been worn or polished as though continually rubbed by
some soft material such as leather or hemp cord. Dereivka did
not produce any artifacts that might be identified as mouth-
pieces. However, here it is important to remember that the
cult stallion with clear signs of bit wear is now known to be
an Iron Age intrusion and that early mouthpieces may well
have been made of soft material that would leave no trace
in the archaeological record (see below).
More secure evidence of cultural control comes in the
form of disk-shaped cheek pieces fashioned from antler that
FIGURE 17.1 Map showing major sites in the Eurasian
chapter17 2005.12.16 9:24 PM Page 253
are found directly associated with horse skulls in Russian
“chariot burials” dating to the Middle Bronze Age (ca.
2000–1700 BC) (Gening et al. 1992). The lack of any preserved
bits in this burial context confirms the suggestion that early
mouthpieces were made of rope or leather (Littauer and
Crouwel 2001). Comparable metal-studded disk cheek pieces
and bits date to the second half of the second millennium
BC in the Near East, Egypt, and Greece (Littauer and Crouwel
1986, 1988, 2001). A donkey with a metal bridle dated to
about 1700 BC was recovered at Tel Haror in the northern
Negev (Littauer and Crouwel 2001).
The absence of parts of tackle in the archaeological record,
however, is not proof of the absence of domesticated horses,
since horses can be ridden without a bridle or with a simple
leather or hemp bridle that would unlikely be preserved.
The Greeks are nearly always shown riding bareback, often
with no bridle (Markman 1969), and the Plains Indians of
North America often used a rawhide loop around the
diastema, or bar, between the incisors and cheek teeth on the
lower jaw (Wilson 1978). The earliest evidence for the use of
saddles comes from Pazyryk (fourth century BC); the earliest
horseshoes come from the Roman era (Clutton-Brock 1996);
and the earliest stirrup is depicted on a bas relief at the Indian
site of Mathura dating to about 50 BC (Littauer 1981).
The search for horse corrals in the archaeological record is
only beginning but should be marked by postholes that out-
line a relatively large enclosure. There should also be chemi-
cal, botanical, and micromorphological evidence of horse
manure in the soil in these enclosures (see below).
Finally, Gheorghiu (1994) claims harnesses can be seen on
some of the stone horse-head scepters dating to the Eneolithic
(ca. 4500 BC) that are found in sites in western Russia, the
Ukraine, and southeastern Europe. However, this interpreta-
tion is highly speculative and based on a generous reading
of highly stylized carvings. It is interesting to note, however,
that these scepters seem to mark the spread of Pontic-Caspian
steppe peoples into southeastern Europe (Mallory 1996) and
may in fact signal the diffusion of domestic horses out of the
region of initial domestication.
Pathologies Associated with Domestication
Levine (1999a) has investigated the relationship between
riding and pathologies in the horse skeleton. Although it has
been documented that weight-bearing and saddles cause
pathologies of the back, many modern riders scoff at the
notion that riding bareback or with a soft pad or blanket
would create significant problems for the horse’s skeleton.
Depending on the conformation of the horse’s back, as well
as the ability and weight of the rider, riding need not result
in any pathologies in either the horse or its rider. With bare-
back riding, the rider shifts his or her weight from one side
of the horse’s back to the other, so circulation is not inhibited
in any part of the horse. The padding in human buttocks
acts as a hydraulic cushion that protects his or her ischial
254 ARCHAEOLOGY AND ANIMAL DOMESTICATION
tuberosities (the part of the pelvis that humans sit on) as well.
A flexible leather pad also helps reduce impact for both rider
and horse. Skeletal pathologies resulting from riding should
be rather rare in horses prior to the invention of the true
saddle with a wooden frame. It might be more profitable to
look for anomalies in the human skeleton that signal muscle
hypertrophy caused by habitual riding or an increase in
mended fractures occurring as a result of falls from horseback
(Molleson and Blondiaux 1994).
On the other hand, it is well established that a saddle
with a wooden tree can cause pathologies. Preventing injuries
through precise saddle fitting and padding is an enormous
industry today. The distribution of the rider’s weight on a rigid
frame creates pressure points, which can eventually cause
muscle atrophy in the horse’s back. A poor saddle can lead
to problems in the regions of the horse’s shoulders, back,
pelvis, hock, stifle, and leg. Baker and Brothwell (1980: 131)
report that riding and workhorses are sometimes afflicted with
spondylosis deformans of the vertebrae, but they do not specify
whether this damage is caused by a saddle or not. These
lesions typically occur in the lumbar region in domestic
horses, whereas they are more common in the thoracic area
in wild horses.
Saddles with wooden frameworks are not known from
contexts dating before the Iron Age and Roman period, how-
ever. Saddles found in the frozen tombs of Pazyryk, in Siberia,
had wooden spacers consisting of flat boards with tongues
at either end that were inserted into leather saddle cushions
(Rudenko 1970: 132–133). These were found elsewhere
among the Scythians and seem to have been widespread. It
would not be surprising if these boards applied pressure that
could lead to muscle atrophy and perhaps eventually have
impact on the skeleton itself. The Romans were the first
known to have used a treed saddle with horns or pommels
(Hyland 1990; Clutton-Brock 1996: 89).
Traction is likely to lead to increased rugosity of muscle
attachments or arthritis of the horse’s limbs. Equids were used
to pull wagons, carts, and chariots by the middle of the third
millennium BC. Yokes initially developed for use on oxen are
quite unsuited to equid anatomy and almost certainly would
have caused the development of pathologies in the neck
and shoulder region. The Greeks, Romans, and Chinese
(second century AD) initiated experimentation with various
means of harnessing draft horses, but this was long after
Perhaps the most widely known pathology used as a marker
of domestication in horses is the wear on the lower second
premolar caused by the use of a bit. This form of attrition
is caused by a horse habitually clenching a bit between its
teeth for brief intervals. After metal bits were developed,
this pattern becomes fairly visible. Evidence of bit wear is
usually seen on the mesial half of the lower second pre-
molar, but occasionally occurs over most of the occlusal
surface of this tooth and often on the upper second premolar.
The most distinctive aspect of the wear caused by a metal bit
chapter17 2005.12.16 9:24 PM Page 254
is the absence or reduction of the “Greaves’ Effect” (Greaves
1973). The Greaves’ Effect is a natural process caused by
food abrasion on the occlusal surfaces of artiodactyl and
perissodactyl teeth that results in differential erosion of the
softer dentine between the more resistant enamel ridges.
Frequent contact with a metal bit causes much faster erosion
of the rigid enamel plates so that they wear at a rate compa-
rable to dentine. The result is a facet with a very smooth, even
surface, at least initially. Older horses that have been bitted
for long periods sometimes develop a cupping of the second
premolar or a steep step down in either the middle part or
near the mesial end of the tooth. This is the kind of wear seen
on a horse specimen from the highland Iranian urban site of
Tal-e Malyan dating to about 2400–1800 BC (Figure 17.2;
Zeder 1986, 1991) and on a premolar of a stallion from
Buhen Egypt dated to 1675 BC (Clutton-Brock 1974, 1992:
83). The blue-green staining and bit wear on the premolars
of donkeys from Tell Brak in Syria (ca. 2300 BC) indicate the
use of copper or bronze bits (Clutton-Brock and Davis 1993).
Bit wear can also be seen on the so-called cult stallion
that until recently was held to be the most convincing piece
of evidence for claiming early horse domestication at Dereivka
in the Ukraine (Anthony and Brown 1991). This specimen
was originally dated by stratigraphic association to around
4000 BC, well before there was any artifactual evidence for
metal bits. Recent direct dating of the specimen reveals it to
be an Iron Age intrusion from between 790 to 200 BC (cali-
brated) (Anthony and Brown 2000). The first solid evidence
for the use of metal bits in Russia is seen in the ninth cen-
tury BC, so it is likely that the wear pattern on the Dereivka
cult stallion was caused by a metal bit.
The question remains whether soft bits made from leather
or hemp, likely used before the invention of the metal bit,
leave similar wear patterns that could be used to detect the
riding of horses. In the late 1980s, Anthony and Brown
(1989) began a series of longitudinal actualistic experiments
comparing the effects of a range of materials for bits on
living horses, including leather, hemp, horsehair, and bone.
According to their experiments, nonmetallic bits leave a
reduced bevel on the mesial portion of the lower second
premolar that lacks the characteristic abrasion pattern left by
metal bits. A few of the Copper Age horse teeth from Botai
(3700–3100 BC calibrated) studied by Anthony and Brown
EARLY HORSE DOMESTICATION ON THE STEPPE 255
show this kind of wear (Brown and Anthony 1998), which
was interpreted as a likely indictor of early horseback riding.
As discussed below, subsequent work calls these results into
Ritual Horse Sacrifice
There is a long tradition of burying horses and other domestic
livestock with humans as funerary offerings all across Ukraine,
Russia, and Kazakhstan stretching back to the Neolithic. The
earliest evidence of horse sacrifice performed to accompany
human interments comes from Late Neolithic graves at
Varfolomievka (5570–4840 BC calibrated), in the Volga-Ural
steppes north of the Caspian (Telegin and Potekhina 1987),
where horse bones were placed in ochre-stained pits above
the human remains. The same was true at the Khvalynsk I
cemetery (5242–4580 BC calibrated), in the Middle Volga
region of Russia (Agapov et al. 1990; Anthony and Brown
2000); the early Eneolithic site of S’yezzhe, in the Samara River
valley (Vasiliev and Matveeva 1979); and Nikol’skoe
(4950–4000 BC calibrated) on the Dnieper River (Anthony
and Brown 2000) (Figure 17.1). Wild animals are unusual in
human graves in this region, and it is clear that the sheep and
cattle remains interred with humans were domesticated. The
horses ritually buried with humans were treated in exactly
the same manner as were cattle and sheep, often with skulls
and foot bones being placed near the human body in what
Piggott (1962) referred to as “head and hoof” burials. The
treatment of horses and obviously domesticated bovid live-
stock in a similar manner is generally taken as a signal of the
horse’s domestic status, although this association should be
considered only circumstantial evidence. This would push the
date for the earliest domesticated horses back to the middle
of the sixth millennium BC, which would be very early
The Process of Horse Domestication
While evidence of riding might be assumed to be a good indi-
cator of horse domestication, Levine (1999b) has argued that
people might have captured wild individuals, broken, tamed,
and ridden them without maintaining a breeding population.
In this way, highly nomadic hunters would have been able
to have the benefit of horse transport without the trouble
of breeding horses and caring for newborns and suckling
young. Thus, the argument goes, while horseback riding
may indicate the taming of horses, it is not necessarily a sign
of their domestication. There are two serious problems with
First, the resolution of the archaeological record is unlikely
to allow us to locate a boundary between taming and early
domestication. It is, after all, a process that, as zooarchaeolog-
ists recognize, flows through our arbitrary classes of wild, cap-
tive, tame, and domesticated. The archaeological record
exacerbates the problem further through deterioration of
FIGURE 17.2 Bit wear on lower second premolar from
Malyan: (a) right-side lingual aspect; (b) left-side buccal
chapter17 2005.12.16 9:24 PM Page 255
perishables related to horse control (rawhide or hemp tackle)
and the more delicate neonatal and juvenile bones expected
in domestic herds.
The second more serious problem with this thesis is that
it seems unlikely that wild mares and stallions would be cap-
tured, tamed, and even perhaps trained for riding but pre-
vented from breeding with each other. This scenario requires
thorough isolation of the two sexes in separate corrals to avoid
any contact during estrus. One mistake and the tame popu-
lation would suddenly become a breeding one. Overcoming
the powerful natural urges of horses to breed is especially dif-
ficult given that, even today, domestic horses in Kazakhstan,
Mongolia, and Russia are released from the corrals in the
evening and driven overland to graze through the night on
natural pasture. Only a couple of horse herders may drive as
many as 250 horses at a time, and even with the aid of
persistent herding dogs, control is never absolute.
This debate over whether horses were tamed or truly
domesticated raises the deeper questions of what constitutes
a domestic population and how it can be detected in the
archaeological record. Different scientists use different criteria
for the recognition of a domestic population. Some require
evidence of total genetic isolation from wild herds, but in
reality, this degree of isolation is rarely achieved until the wild
population is totally extinct. Dogs still breed with wolves, for
example, but none would deny that dogs are domesticated.
For the early domesticates, this standard is much too rigid.
Prior to extirpation from their natural habitat, wild Asiatic
stallions, Equus przewalskii, frequently stole the Mongols’
domestic Equus caballus mares and bred with them. The same
was true for the tarpan, Equus ferus, the likely wild progeni-
tor of the domestic horse, in its natural range until its extinc-
tion. Realistically, without extinction, isolation can occur only
if the domestic herds are moved outside the natural geo-
graphic range of the wild progenitors. Although domestic
herds did eventually spread out to areas not inhabited by wild
equids in the Middle to Late Holocene, they also thrived in
the heart of the wild horse’s habitat, Ukraine, Russia, and
Some zooarchaeologists demand either morphological or
metric distinctions on the skeleton as proof that domestica-
tion has taken place, but such changes may not always be
present. There are many cases of separate species within a
family of wild animals that cannot be distinguished on either
of these grounds, and yet they do not interbreed and are reco-
gnized as valid taxa by biologists. As demonstrated above,
despite over a century of morphometrics on equid skele-
tons, there is still no consensus on which criteria are reliable
for distinguishing between wild and domestic horses, and
there is considerable variation even in the Pleistocene among
caballine equids. Archaeologists place unnecessary restrictions
on themselves if they limit themselves to metric or mor-
phological markers to detect domestication in animals.
Reliance on morphological and size change as the only cri-
teria used to document domestication may delay recognition
256 ARCHAEOLOGY AND ANIMAL DOMESTICATION
of horse domestication by as much as two or even three
millennia after the initial event took place.
Culling is a useful trait of livestock domestication that
holds more promise for recognizing incipient horse breeding,
but the keeping of highly valued stallions and geldings for
riding may have happened fairly soon after domestication
began. Controlling herds of domestic horses without being
able to ride would be difficult, although theoretically possible.
Herds had to be released to graze far and wide over the
natural landscape. To round the animals up in open steppe
with nothing more than pedestrians and dogs would have led
to chaos and the loss of many individuals from the herd. Feral
animals would have found it easy to strike out on their own,
especially in the case of young males, or to join a band of wild
horses in the case of females. Ridden domestic stallions
would have been excellent at rounding up the herd, since it
is their natural instinct to keep control of their family band.
Although there are problems using bit wear from a soft
bridle as a reliable line of evidence for early horse domesti-
cation (see discussion below), there is still a wide range of
secondary data that can be useful in documenting horse
domestication. These data are described in the discussion of
research in northern Kazakhstan that follows.
As seen in the review of previous efforts of researchers
working on this problem across the Eurasian continent,
clearly documenting where and when horses were first domes-
ticated is not an easy task. At present, there is no smoking
gun, no single definitive criterion for demarcating the domes-
tication of the horse in the archaeological record. The solu-
tion to documenting horse domestication lies in attacking the
problem with a holistic strategy that looks at a wide range
of secondary, more circumstantial evidence for the keeping
of horses. When the secondary, indirect evidence reaches crit-
ical mass, then it becomes more and more likely that horse
domestication had, indeed, taken place. The multidisciplinary
approach to the study of horse exploitation of the Botai-age
peoples in the central Eurasian steppe of modern day
Kazakhstan discussed below demonstrates the utility of
applying multiple lines of investigation in order to document
early horse domestication.
Integrated Multidisciplinary Research at Botai
The Geometric Impressed Pottery Province (GIPP) of the
eastern Urals and northern Kazakhstan (Olsen et al. in press)
shows real promise for providing valuable insight into the
process of horse domestication. The Botai culture of northern
Kazakhstan (Zaibert 1993) is now seen as one of the most cru-
cial sources of information for documenting this landmark
in human history. This is not, however, because the Botai
region necessarily represents the first or only site of horse
domestication. Instead, the Botai culture provides the optimal
case study for this difficult task because Botai sites are located
in the heart of the native geographic range of the wild horse
chapter17 2005.12.16 9:24 PM Page 256
and date to the fourth millennium, sometime soon after it
is thought horse domestication began. Moreover, the Botai
based their whole economy on the horse, and their large, per-
manent settlements have yielded enormous collections of
horse remains. As a result, Botai sites provide an ideal oppor-
tunity for developing a multidisciplinary, holistic approach
to the problem.
The Botai culture is represented by four known settle-
ments: Botai, Krasnyi Yar, Vasilkovka, and Roshchinskoe
(Figure 17.1). Of these, most of the work has been done on
the largest site, Botai, which has been radiocarbon dated to
3700–3100 BC (calibrated) (Levine and Kislenko 1997).
Excavations have been conducted regularly at Botai since the
1980s by the University of North Kazakhstan, under the
direction of Victor Zaibert (1993). The nine-hectare settle-
ment, located on a small tributary of the Ishim River, has at
least 158 houses, but many more have been washed away by
the encroaching Iman-Burluk River. Over half of the site has
Two of the other sites, both located near the modern city
of Kokshetau, have been investigated to a lesser extent. Two
pithouses at Krasnyi Yar and one at Vasilkovka were excavated
in the 1980s (Kislenko and Tatarintseva 1999; Zaibert 1993).
Their assemblages compare very closely with those of Botai.
Recently, a joint team of archaeologists from the Carnegie
Museum of Natural History and the Presidential Cultural
Center of Kazakhstan has excavated one house, plus adjacent
areas at each of these settlements. Based on detailed remote
sensing, it is possible to determine that Krasnyi Yar had at least
54 houses and Vasilkovka had 44 (Olsen et al. in press). The
fourth site, Roshchinskoe, is the least known and has had very
little testing or mapping conducted.
The faunal assemblage of the eponymous site of Botai has
never been studied in its entirety, but it is estimated to total
over 300,000 bone fragments. Given that most faunal assemb-
lages from the steppe consist of a few hundred to a few thou-
sand elements, this collection is unique. Only a tiny portion
of the Botai assemblage is from animals other than the horse,
so it is safe to say that 99% of the identifiable remains are
derived from horses. This is a significantly higher percentage
than at Dereivka, where 63% of the faunal assemblage is
horse (Telegin 1986). Dereivka itself has a notably greater focus
on horses than most other contemporaneous sites in the
Ukrainian and Russian steppe.
Previous work on the Botai horse remains includes the
metric analysis of 10,000 postcranial bones by Akhinzhanov
et al. (1992) and numerous partial and complete crania by
Kuzmina (1997), inspection of 36 lower second premolars for
possible bit wear by Anthony and Brown (1998), mortality
patterns and pathological investigation by Levine (1999 a,
1999b), and studies of the taphonomy, butchery, ritual treat-
ment, mortality patterns, dog burials, and bone artifacts by
Olsen (1996, 2000a, 2000b, 2001, 2003). Since 1993, my
research has focused on understanding the horse-based
lifestyle and economy of the Botai people using a variety of
EARLY HORSE DOMESTICATION ON THE STEPPE 257
methods, including remote sensing of whole villages; exca-
vation of houses, middens, and pits; faunal analysis; and
studies of related artifacts in an effort to pursue multiple
lines of evidence in the documentation of early horse domes-
tication (Olsen, et al. in press).
Mortality Patterns at Botai
Levine (1999a) was the first to perform a detailed examina-
tion of the dentition from Botai using the crown-height
method discussed above. She concluded that there were few
fatalities before 2.5 years in the sample of 529 cheek teeth
(MNI 29) that she studied. According to her findings, most
individuals were slaughtered between the ages of 3 and 8
years, with about a quarter of the harvested animals dying
between the ages of 8 and 16. Based on mandibles, the ratio
of males to females was 7:6; based on pelves, it was 17:20
(Table 17.4). Because this profile resembled that in a living
wild herd, she interpreted these mortality data as the result
of a hunting strategy using communal drives. Thus Levine
(1999a) concluded that the horses at Botai, in Kazakhstan,
were wild animals hunted by means of stalking, just as she
had proposed to be the case at the slightly earlier site of
Derievka, in Ukraine.
In 1996, I studied all the Botai maxillae and mandibles
excavated to date, as well as the isolated teeth. Estimated
dental ages in my study were based on dental eruption and
occlusal wear (Hillson 1986: 215; Kainer and McCracken
1994: Plate 48, 49). Minimum numbers of individuals for the
whole dental assemblage were obtained by recording all
intact, identifiable teeth (both those in jaws and isolated
teeth), a total sample of 748 teeth. For isolated teeth, as
mentioned above, it is not always possible to distinguish
among third and fourth premolars and first and second
molars. In such cases, they were grouped and the sums were
divided appropriately to yield estimated MNI figures, but
these never exceeded those for specifically identifiable teeth
in the overall MNI calculations.
Sex ratios obtained for mandibles in my study resemble
those reported by Levine based on mandibles and pelves
(Table 17.4) and show a fairly even proportion of adult males
and females in the sample. The maxillary data differ in
exhibiting a very high ratio of males to females. The high
Comparison of Proportions of Adult Female and Male
Horses at Botai
Sex Indicator Analyst MNI Female MNI Male % Male
Mandibles Levine 6 7 54%
Pelves Levine 20 17 46%
Mandibles Olsen 12 13 52%
Maxilla Olsen 1 8 88%
chapter17 2005.12.16 9:25 PM Page 257
proportion of adult male maxillae, however, is likely a reflec-
tion of the differential deposition and preservation of stal-
lion crania in ritual pits. Mandibles are known to have been
included on occasion in these ceremonial contexts, but with
less frequency than crania. Thus, the lower jaw data are more
likely to represent overall slaughter patterns both for subsis-
tence and ritual, while the maxillary data show the influence
of ritual practices on the assemblage.
The age profile that I have compiled for the Botai horses
based on dental eruption and occlusal wear (Table 17.5)
shows that most of the horses died as mature adults, rather
than through culling of juveniles. In a domestic herd, this
pattern would imply an emphasis on the use of secondary
products or services supplied by adults. In the case of horses,
the females were probably maintained for breeding and
milking, while the males would have been kept for riding and
as packhorses. Samples are small, but there is some slight
(probably not significant) tendency in these data for males
to be killed at younger ages than females. This might repre-
sent a minor amount of culling of young males. In modern
times, only a small number of stallions are ridden, primarily
by the horse herders, in comparison to the much greater num-
bers of mares that are maintained for breeding and milking.
In prehistoric times, however, riding would have been an
important form of transportation for hunting, herding, and
traveling in general.
My mortality pattern for Botai differs from Levine’s based
on crown heights partly because different methods were
employed to determine age. Levine compensates for per-
ceived differential destruction of deciduous teeth by artificially
inflating the numbers of immature individuals, but deciduous
teeth are not significantly different from permanent teeth in
their mineral content and are therefore not more susceptible
to taphonomic destruction. Levine’s method of dealing with
258 ARCHAEOLOGY AND ANIMAL DOMESTICATION
the ambiguities of isolated second and third premolars and
first and second molars also differs from my method of cal-
culating MNIs on teeth whose positions can be more confi-
dently identified. Finally, the use of crown heights (in contrast
to the use of eruption and occlusal wear) does make it pos-
sible to estimate the ages of older adults beyond the age of
eight, which is why Levine has a broader range of years in
her data. In large measure, however, my age profile generally
agrees with Levine’s. Both indicate that there is an emphasis
on mature males and females between 3 and 8 years of age.
It is also useful to examine the ratio of juvenile teeth to
adult teeth for both incisors and the more durable and reco-
verable second premolar (P2) (Table 17.6). Based on the P2
alone, the proportion of juveniles under the age of 2.5 years
in the sample would seem relatively high—at about 35%.
However, based on the I1, which erupts about the same
time, this figure is significantly lower—at about 22%. Because
incisors are easily lost during excavations, they are less likely
to be truly representative of actual patterns and are more sus-
ceptible to random variation than are the larger, more durable
premolars. This is illustrated clearly by the complete absence
of deciduous second and third incisors. The MNIs for jaws
are lower than for P2s because there are so many more iso-
lated teeth, so in this case, the readily identifiable second pre-
molars provide extremely useful and reliable assessments of
proportions of very young vs. animals over the age of 2.5
years. Mandibles were regularly broken and the teeth were
removed in order to make thong-smoothers. To a large extent,
this practice certainly contributes to the paucity of ageable
mandibles in this collection. Thus the P2 estimate of 35% for
juveniles under 2.5 years of age at Botai should be considered
a more accurate number.
However it is computed, the percentage of young in the
Botai sample is not high enough to indicate a strong emphasis
on the culling of juveniles. The adult sex profile also does not
reflect a prevalence of adult females, as is expected in an
assemblage derived from a focus on breeding and milking.
However, as discussed above, the Botai horse assemblage is
likely to contain both domestic and hunted wild horses.
Distribution of Botai Horses by Sex/Age Categories, Based
on Mandible and Maxilla MNIs.
Age Indeterminate, MNI Female MNI Male MNI
5–14 mo 1
9 mo–2.5 yr 1
2–3 yr 3
3.5 yr 1
4 yr 1 2
5 yr 1 1
6 yr 4 5
7 yr 5 3
8 yr 1 1
>8 yr 1
Totals 6 12 13
Comparison of Proportions of Juvenile and Adult Horses at
Age Separating MNI MNI %
Age Indicator Juvenile and Adult Juvenile Adult Juvenile
Mandible 4 yr 6 25 19%
Maxilla 4 yr 0 9 0%
P2 2.5 yr 8 15 35%
I1 2.5 yr 8 28 22%
I2 3.5 yr 0 29 0%
I3 4.5 yr 0 23 0%
chapter17 2005.12.16 9:25 PM Page 258
Moreover, if domesticates were important for riding and as
pack animals, in addition to meat and milk, then more adult
males would be expected in the assemblage. Finally, the
intentional burial of skulls of ritually sacrificed adult males
leads to better preservation of these animals than juveniles
and adult females, so it is not surprising that the sex ratio for
maxillae is biased toward males at Botai. Although an
emphasis on adults, as detected by Levine and myself, might
be consistent with hunting, it could equally reflect a mixed
population of wild and domestic animals. Also, the retention
of adult females for breeding and milking and adult males for
riding could lead to higher numbers of adults in domestic
horse populations than would be typical for other livestock
that are not used for milk and riding. The differential preser-
vation of stallion skulls in ritual pits is also clearly a factor
in increasing the proportion of adults in the record.
Studies of the faunal material from Botai included a detailed
examination of bones in order to understand the distribution
and nature of wounds incurred during slaughter (Olsen
2003). Based on the shape and size of their apertures, three
wounds in horse bones (Figure 17.3) and one in an aurochs
bone appear to have been made by harpoons, rather than by
stone projectile points. This evidence suggests that wild
horses were still in the region and were occasionally hunted,
as they were in the preceding Neolithic.
Richard Marlar, a blood pathologist at the University of
Oklahoma, performed blood residue analysis on 12 stone pro-
jectile points from Krasnyi Yar and found hemoglobin residues
on three specimens. All three tested positive for human
blood, which is rather unusual. One also tested positive for
suid (most likely wild boar) and wild (most likely) or domestic
sheep. The Argali sheep, Ovis ammon, exists in the mountains
of Kazakhstan, although no remains have been found in
Botai sites to date. The third point tested positive for human,
EARLY HORSE DOMESTICATION ON THE STEPPE 259
suid, canine (fox, wolf, or dog), ibex or goat, and bovine
(aurochs, bison, or domestic cattle). Ibexes, like Argali sheep,
still can be found in the mountains of Kazakhstan but have
not been identified in the faunal remains. Surprisingly, no
horse blood was identified on the stone points, but perhaps
harpoons were preferred weapons for hunting wild equines.
One common method of slaughtering animals in the
steppe, pole-axing, was most likely reserved for domestic
livestock. Whereas a single hunter can use a bow and arrow
to bring down a wild beast, pole-axing usually involves three
persons or at least one person and two posts firmly set in the
ground. Normally, two people hold the ends of lassos that are
wrapped around the neck of the horse and pull them taut on
either side of the standing beast. In this position, it is not easy
for the horse to step backward or move out of harm’s way. A
third person then approaches and strikes the animal between
the eyes with a pole-ax, ideally killing it in one stroke. Pole-
axed cattle and horse skulls are common in kurgans (burial
tumuli) from the Bronze Age across Russia and Kazakhstan.
The weapon typically leaves a large, depressed fracture in the
frontal bones, like the one shown on a modern skull of a horse
from Mongolia that was killed in this manner (Figure 17.4).
One cranium in the Botai collection (Figure 17.5) shows a
round, depressed fracture with the broken tissue still in situ;
however, the injury is more lateral than normal and is located
on the maxilla, instead of the frontal. It is possible that the
blow was struck just at the moment the animal averted its
head. Thus, although this cranium seems to bear the kind of
fracture caused by pole-axing, the placement of the fracture
is not entirely consistent with this practice. Therefore, this
skull cannot be considered solid evidence for a method of
FIGURE 17.3 Botai hunting tools and traces: (a) wound in
horse rib from Botai; (b) wound in ?? (c) Botai harpoon.
FIGURE 17.4 Modern horse cranium in Mongolia with
FIGURE 17.5 Horse cranium with possible pole-axing
wound from Botai.
What is b?
chapter17 2005.12.16 9:25 PM Page 259
slaughter typically associated with domestic or at least tame
Neither Levine’s (1999b: 53) nor my own study of the Botai
material discovered any pathologies associated with riding
or the use of horses as beasts of burden or for traction. As dis-
cussed above, however, riding horses without a saddle or with
only a soft pad is unlikely to have a detectable impact on a
horse’s back. The wheel was unknown in this region in the
fourth millennium, so horses probably would not have been
used for draft; however, it is possible that they were used to
haul heavy loads of stone or meat on their backs as pack ani-
mals. The dentition indicates that most horses were killed
before their ninth year, so there was little time for arthritis
and similar work-related maladies to develop.
As noted above, an earlier study of Botai horse dentition
conducted by Brown and Anthony in 1992 detected some
wear that they associated with the use of a soft bit (1998: 344).
This has been presented as evidence of horseback riding at
Botai. They reported examining 36 lower second premolars
from a minimum of 20 individuals. Of these, at least 19
were older than three years of age, the minimum age for reli-
able detection of bit wear. Five of the premolars (26%) showed
a bevel on the mesial part of the occlusal surface measuring
3 mm or more. None of these teeth exhibited any sign of
either macro- or microscopic abrasion or other alteration
indicative of bit wear. Based on Anthony and Brown’s actu-
alistic experiments with different types of soft bridles, they
attributed these wear patterns to the use of a soft bit, prob-
ably made of hemp.
My own examination of 23 lower second premolars from
Botai (including the 5 Brown and Anthony felt showed
bit wear) found only 4 with beveling sufficient enough to
qualify as soft bit wear using Brown and Anthony’s criteria
(Figure 17.6; Table 17.7). Seven of these teeth had bevels that
measured 2 mm at their deepest point, which is less than the
3 mm minimum required by Brown and Anthony to qualify
260 ARCHAEOLOGY AND ANIMAL DOMESTICATION
as bit wear. Five had no bevel or actually curled upward
instead of dipping down. One tooth with an extreme 6-mm-
deep bevel (Brown and Anthony’s specimen no. 37) also had
slightly exaggerated lingual-buccal wear along the entire
tooth, suggesting some kind of malocclusion with the upper
jaw. My observations of the “beveled” teeth show that the
normal Greaves’ Effect, of uneven wear in the dentine and
enamel of the occlusal surface, was still found on all of them.
Recent investigations have identified a potential problem
with restricting the identification of soft bit wear to the
single characteristic of a slope on the mesial part of the
occlusal surface. Upon examining lower second premolars of
horses from four North American Pleistocene localities dating
to many millennia before the arrival of humans, I found sev-
eral examples of Brown and Anthony’s “soft bit wear.” Equus
lambei (Specimen USNM 8426, curated in the Department of
Paleobiology, National Museum of Natural History), from
Gold Run Creek, Yukon Territory, in the Klondike region,
dating to approximately 24,000 BC, has what appears to be
a sharp decline (8.5 mm) that is obvious in the lingual view
(Figure 17.7). The mandible has full dentition with normal
wear along the whole tooth row. Another mandible of this
species from near Ruby, Alaska (USNM 11705) has a bevel on
the P2 of 4.5 mm that includes most of the mesial third of
the occlusal surface. Both of these specimens exhibited the
Botai Premolar Bevels
Bevel in mm
FIGURE 17.6 Graph of Botai P2 bevel measurements taken
Botai P2 Bevel Measurements Taken by Olsen
Tooth Number Side Bevel (mm)
3579 L 0
B82 L 0
491 L 0
33-P-37 R 0
32-2C 75b R 0
29564 L 0.5
2025 L 1
800 L 1
549 L 1
B84-88-HM R 1
32-P-37 R 1.5
518 L 1.75
589 L 2
2361 L 2
3583 L 2
32 no spec. # R 2
532 R 2
32-2ZH-75b R 2
A & B 14 L 2
A & B 21 L 3
A & B 2 R 4
A & B 7 R 5
A & B 37 L 6
chapter17 2005.12.16 9:25 PM Page 260
normal Greaves’ Effect. Examination of a collection of horse
remains from the Big Horn Basin of Wyoming dating from
18,170 to 15,620 BC (curated by the University of Kansas and
studied by Martin and Gilbert 1978) showed a high degree
of variation in the wear of the P
under totally natural con-
ditions. This means that the presence of a beveled second pre-
molar alone can no longer be taken as a reliable marker of
bit wear, whether from a soft bit or from any other kind. For
now, the soft bit wear evidence for early horseback riding at
Botai should be put aside.
At present, the “bit wear” on teeth from Botai identified by
Brown and Anthony (1998) is at the very least ambiguous and
is likely to be simply natural dietary wear. However, there is
clear evidence that people there were manufacturing rawhide
thongs that might well have been used in a wide range of
activities relating to both riding and capturing horses. The
most common bone implement at the site of Botai is a thong-
smoother (Figure 17.8), an implement used to straighten and
stretch a strip of rawhide (Olsen 2001). Made from horse
mandibles, thong-smoothers constituted 32 percent of the
assemblage of bone artifacts from Botai, with at least 135
horses required to produce them. The high polish and fine
striations seen in notches on these tools (Figure 17.9) indi-
cate that narrow strips of soft, pliable material were pulled
back and forth across the edge of the notch. The widths
of the notches range from 1.1–6.5 cm, with an average of
EARLY HORSE DOMESTICATION ON THE STEPPE 261
4.3 cm. Thongs made with these smoothers could have
served many functions, but thongs are essential in horse
control as bridles, whips, lassos, hobbles, and pole snares.
Although thongs also would have been useful as harpoon
lines, the small number of harpoons found at Botai (N = 13)
in comparison to the large number of thong-smoothers
(N = 270) suggests that there was a much more significant
purpose for rawhide thongs.
Skeletal-part distribution data may also shed light on the
status of horses from Botai culture sites. It has long been
known that hunters of large game often leave parts of their
prey back at the kill site (Reitz and Wing 1999; Outram and
Rowley-Conwy 1998). Perkins and Daly (1968) coined the
term Schlepp Effect for this condition, because it was the
result of hunters on foot reducing the load they had to
schlepp, or drag, back home by field-dressing the carcass. To
lessen the transport load, hunters will abandon bones of
their prey that are of low utility for either meat or marrow.
or those that can be easily stripped of meat. However, hunters
may choose to transport certain bones back to their base
camps for later manufacture into bone tools or for ceremo-
nial purposes. Given that horses fall into the large game
category, it seems likely that, without some form of transport,
horse hunters would seek to reduce transport weight by field
butchering wild horses and abandoning skeletal elements not
needed for food, manufacture, or ritual at the kill site. In fact,
as long ago as 1947, Arcikhovski predicted that an assemblage
of domesticated horses should include relatively complete
skeletons, while that of wild horses would be lacking certain
elements, particularly vertebrae and scapulae (cited in
Bökönyi 1974b: 237).
A butchering experiment conducted by Bruce Bradley and
myself on a horse carcass, using replicas of Botai stone tools,
indicated that, with the exception of the intercostal muscles
FIGURE 17.8 Thong-smoother made on a horse mandible
from the Botai culture settlement of Krasnyi Yar.
FIGURE 17.9 Scanning electron micrograph of the notch
of a thong-smoother from Botai.
FIGURE 17.7 Lingual view of lower second premolar of
chapter17 2005.12.16 9:25 PM Page 261
262 ARCHAEOLOGY AND ANIMAL DOMESTICATION
of the rib cage, most of the meat could be removed easily from
the carcass in the field. The most efficient treatment of the
ribs was to cook the meat on the bone and eat it directly off
the ribs, as we do in restaurants today. So, unless the ribs were
cooked and eaten by hunters in the field, the only high
meat yield elements that would necessitate being carried
back to the home base would be the ribs. The use of bone
marrow, however, also played a role in determining which
bones were transported back from a kill site. Outram and
Rowley-Conwy’s (1998) study of meat and marrow content
of horse bones indicated that the most important marrow-
yielding bones in horses are the mandible, humerus, radius,
femur, and tibia. The metapodials and proximal phalanges
also produce small amounts of marrow.
Artifacts made from horse bones at Botai culture sites
include mandible thong-smoothers, notched rib stamps
for decorating pottery, scapula paddles for smoothing
ceramic vessels, decorated first and second phalanges, and
gouges, leather punches, cylinder stamps, and harpoons
made from metapodials (Table 17.8). Other elements with
nonfood-related value would have been crania used in ritual
Taking all these factors into consideration, the elements
most likely to be left behind at the kill site of wild horses
would be the vertebrae. It is relatively easy to remove the large
muscles of the back like the longissimus dorsi, which runs from
the sacrum and the ilium all the way to the neck on both sides
of the vertebral column. The vertebral column also has low
marrow content, has little utility in tool manufacture, and
is weighty and difficult to separate into portions that would
be less awkward to carry. The large, cumbersome pelvis,
which is tightly connected to the vertebral column, would
also seem a likely element to leave behind at the kill site after
the meat was removed.
Analysis of cutmarks and breakage patterns on the Botai
horse remains (Olsen 2003) indicates that Botai people were
Artifact Raw Material from Horse Elements at Botai
Element Number of Artifacts MNI by Element
Mandible 270 135
Hyoid 7 4
Scapula 10 5
Rib 158 5
Innominate 6 3
Radius 1 1
Metapodial (3/4) 108 27
Metapodial splint 64 8
First phalanx 44 11
Second phalanx 2 1
Astragalus 1 1
Tot al 671 135
intensively using most parts of the horse, including brains,
tongues, meat, marrow, sinew, and skins. Ribs were snapped
at the necks in a way that might facilitate transport. However,
fully articulated vertebral columns are found frequently at
both the sites of Botai and Krasnyi Yar, as are pelves. The high
proportion of these parts in the Botai settlements suggests
either that some horses were killed and butchered at or near
the village (perhaps as domesticates) or that alternative
means of transportation, such as packhorses, were available
to hunters. Related to this latter idea, it is interesting to note
that the crania of wild aurochs, with horn cores up to a
meter in length, were found at Krasnyi Yar, along with ver-
tebrae and other large bones that might have been expected
to have been left at the kill site. Carrying these heavy, bulky
elements clearly would have been difficult without the aid
of some form of transport, suggesting that perhaps domestic
horses aided humans in bringing back carcasses of hunted
large game, including wild horses and aurochs.
Evidence for Lithic Transport
There is also an interesting change in lithic sources and use
in Botai culture sites compared to earlier Neolithic sites in the
region that supports the notion that domestic horses may
have been used in transport. The Neolithic site of Zhusan,
just 600 m from Krasnyi Yar, produced finished blades and
blade tools made from local rhyolite available at a quarry 9
km from the two sites. Bradley found 30 shallow pits sur-
rounded by flaking debris at the Zhartas quarry site (Olsen
et al. in press). Blade precores and waste products also found
there indicate that blades, probably of Neolithic age, were pre-
pared at the Zhartas quarry site. Small bifaces, typical of the
later Botai lithic assemblages, were also present at Zhartas,
and may indicate continued use of the quarry.
In addition to rhyolite tools, the Botai culture sites of
Krasnyi Yar and Vasilkovka produced a much wider array of
raw materials from a number of unknown sources. Jasper,
flint, and especially a fine-grained quartzite are plentiful in
Botai lithic assemblages, despite the fact that several sur-
veys have failed to locate the sources of these materials.
Detailed surface geology maps do not indicate potential local
outcrops either. Jasper is common, however, in the Ural
Mountains far to the west (Matyushin 2003), raising the
possibility that some or all of these materials were derived
from more distant sources.
Bradley’s analysis of the Krasnyi Yar assemblage (Olsen
et al. in press) suggests that some of the stone was transported
to the village as large, heavy pieces that were then knapped
into bifaces, scrapers, and other flake tools. A cache full of
large flint flakes and knapping stations with high concentra-
tions of debitage, for example, were found just outside a
house excavated at Krasnyi Yar. Moreover, the biface tech-
nology characteristic of the Botai culture results in heavier
tools and is clearly a less efficient use of raw material than
the earlier Neolithic pressure-blade technology. The diversi-
fication of stone sources used by the Botai (some of which
chapter17 2005.12.16 9:25 PM Page 262
may be quite distant from their homeland), the importation
of large pieces of lithic raw material for on-site manufacture,
and the shift to a technology that used more stone all sug-
gest a significant change in transport costs for raw materials.
The use of domestic horses in the transport of stone would
certainly have had an impact on this industry. Thus, the
changes in lithic technology and raw material use provide
additional circumstantial evidence that might be marshaled
to support the case for the use of domestic horses by the Botai.
A close connection between humans and horses is also shown
in the prominent use of horses in burial and ritual contexts.
The only human burial from Botai (and, for that matter, the
only burial reported from any of the Botai sites) contained
four humans (two men, one woman, and a 10–11-year-old
child) surrounded by at least 14 horses laid out in an arc
(Rikushina and Zaibert 1984). The horse remains consisted
of skulls (N = 14), pelves (at least nine), articulated vertebral
columns (at least five), some ribs, and a few limb elements.
While it is conceivable that this burial represents the cere-
monial sacrifice of 14 wild horses, as noted above, the gen-
eral pattern of animal sacrifice on the Eurasian steppes has
focused on domesticated animals (Jones-Bley 1997; Mallory
More commonly found in Botai sites are ritual deposits of
horse skulls and articulated cervical vertebra in pits around
FIGURE 17.10 Horse cranium and cervical vertebrae in a
pit outside a house at Krasnyi Yar.
EARLY HORSE DOMESTICATION ON THE STEPPE 263
the outsides of houses (Figure 17.10). At the site of Krasnyi
Yar, for example, we found five pits containing horse skulls
modified into partial plates made up of the nasals, frontals,
and parietal bones. The function of these modified skulls is
not clear. Another common Botai ritual involved the place-
ment of whole dogs or dog skulls next to horse skulls, necks,
pelves, or feet in pits on the west sides of houses (Olsen
2000a, 2000b). The ritual association of dogs with horses
evokes the close connection between these species among
steppe peoples today, where dogs are used alongside horses
to hunt game and also to help herd domestic horses. This
horse/dog association in Botai ritual contexts is also seen in
other contemporaneous and later cultures of the steppe
(Telegin 1986: 31–33), as well as in Bronze and Iron Age sites
across Europe (Grant 1984: 222; Merrifield 1987; Green 1992)
and East Asia (Mair 1998), where there is no question of the
domestic status of the horse.
Botai Settlement Patterns
The size and layout of Botai settlements lend support to the
thesis that these people were early horse pastoralists.
Settlements of the preceding Neolithic hunters consist pri-
marily of short-term camps marked by lithic scatters, or
small settlements consisting of a couple of above-ground
houses (Kislenko and Tatarintseva 1999; Shorin 1999; Zaibert
1992). In contrast, Copper Age Botai settlements are large vil-
lages, consisting of 44 to 158 or more houses, usually arrayed
in rows running along a northwest-southeast axis (Olsen et
al. in press). Maps made using remote sensing at Krasnyi Yar
and Vasilkovka clearly show carefully planned layouts of
rows of square houses, with their corners pointing to the car-
dinal directions (Figures 17.11 and 17.12). Possible rectan-
gular plazas evidenced in the northern portions of both sites
may reflect the importance of communal activities, and the
numerous series of postholes demarcating enclosures are
suggestive of corrals. Conservative population estimates
project that from at least 110 to over 400 people lived at each
of the villages of Botai, Krasnyi Yar, and Vasilkovka.
Radiocarbon dates and artifact types suggest that these sites
were occupied contemporaneously. If the fourth village of
Roshchinskoe is included, then perhaps as many as 1,000
people lived in the relatively small Botai region.
The stark contrast between the previous Neolithic camps
and the permanent, well-planned Botai villages points
strongly to a change in subsistence strategies. In fact, it is dif-
ficult to imagine how such large concentrations of sedentary
people could have survived the protracted and severe win-
ters of the Siberian Plain if they were reliant solely on hunting
wild horses, without any other domestic livestock or agricul-
tural resources. The region immediately around the village
would have become quickly depleted of wild horses and
other game, and villagers would have been forced to venture
far from their homes to obtain critical meat resources.
Domestic horse herds could have provided a reliable supply
chapter17 2005.12.16 9:25 PM Page 263
of meat and milk, as well as aided in hunting and transporting
people, meat, and stone. All of these assets would contribute
to a significant degree in making these large, permanent vil-
lages both feasible and stable.
264 ARCHAEOLOGY AND ANIMAL DOMESTICATION
Presence of Horse Manure
Micromorphological analysis of soils in the fill of a pithouse
excavated at Botai in 1995 revealed large quantities of horse
manure lying on the collapsed roof (French and Kousoulakou
2003). The researchers who analyzed the soil column from
our excavations interpreted these remains as the results of
stable or corral cleaning. However, another possibility is that
the Botai people used manure for insulating their houses. The
Kazakhs still use horse and cow manure insulation on the
roofs of their outbuildings to keep animals warm in the
winter and cool in the summer. Although it is conceivable
that this manure represents the droppings of wild horses col-
lected on the steppe, the thick concentration of manure
seen in the house from Botai would seem more likely to
have been generated by managed herds.
Manure may also be detected through fecal biomarkers like
sterols and bile acid, which can persist long after all morpho-
logical remnants have disappeared (Bull et al. 1999; Evershed
et al. 1997). Future micromorphological and chemical analysis
of soil samples will explore the use of manure as a building
material at these sites and as evidence for the possible use of
enclosures as corrals.
Evidence of Milking
One of the more promising lines of future investigation is the
analysis of lipid residues on the insides of potsherds from
Krasnyi Yar and Vasilkovka to determine whether horse milk
was stored in vessels. Unfermented mare’s milk is generally
not drunk, because it is a strong laxative and purgative
(Toomre 1994). Fermented mare’s milk, on the other hand,
is an important source of vitamins and other nutrients for
the modern Kazakh, Kyrgyz, Bashkir, Mongol, Yakut, and
other Eurasian steppe peoples. Koumiss, as it is known in
Kazakh, or airag, in Mongolian, is reported to have been
consumed on the Eurasian steppe at least since Herodotus’s
time (fifth century BC).
Koumiss contains five times more vitamin C than does
cow’s milk (Kosikowski 1982: 43), but also yields vitamins A,
B1, B2, B12, D, and E and contains between 1% and 2.5%
ethyl alcohol (Kosikowski 1982: 43). Mare’s milk is extremely
lean (1–2% fat) and is therefore normally not used to make
yogurt, kefir, cheese, dried curds, or other dairy products.
Mares provide milk over a roughly six-month period from
April through September. The modern, high-yielding Yakut
breed from central Siberia produces 1,200–1,700 kg of mar-
ketable milk per mare over one lactation period (Dmitriez and
Ernst 1989), and today Kazakhstan is the leading nation in
koumiss production. Fermented mare’s milk is generally con-
sumed shortly after it is made, unless it is allowed to convert
into a more potent alcoholic beverage.
Dudd and Evershed (1998) have shown that lipids from
both adipose tissue and milk can be preserved for thousands
of years in pottery and that the products of different
species can be identified using high temperature gas
FIGURE 17.11 Plan of the Botai culture settlement of
FIGURE 17.12 Plan of the Botai culture settlement of
chapter17 2005.12.16 9:25 PM Page 264
chromatography (HTGC) and HTGC/mass spectrometry
(HTGC/MS). As part of the continuing work on Botai horse
pastoralism, Alan Outram of Exeter University has collected
samples of modern horse adipose tissue and milk from the
vicinity of the site of Krasnyi Yar. Using these modern sam-
ples, Richard Evershed, at the University of Bristol, found that
it is possible to distinguish between the two products. Dudd
et al. (2003) have already identified horse adipose lipids in
potsherds from Botai. Continued analysis will attempt to
find out if lipids from mare’s milk are present in Botai ves-
sels as well. The finding of milk fat residues would be one of
the strongest pieces of evidence for horse domestication,
since it is highly unlikely that wild mares would tolerate
The multidisciplinary, holistic investigation performed on the
Botai culture settlements in northern Kazakhstan provides
substantial support for early horse domestication in this
region during the Copper Age (3700–3100 BC calibrated). The
equo-centric, or horse-centered, economy, establishment of
large, permanent settlements with possible corrals and live-
stock enclosures, presence of horse manure in house fills,
horse/dog ritual associations, horse slaughter patterns, horse
sacrifice, abundance of thong-making tools, presence of low-
utility skeletal elements with high transport costs, and
changes in lithic sources and manufacturing practices all
coalesce to produce a compelling argument for the existence
of domestic horses in the Botai sites. Future efforts at finding
residues of milk lipids on pottery and manure in enclosures
may provide even more compelling proof for horse domes-
tication. When all these lines of evidence are brought
together, a strong argument can be made that the Botai
were sedentary horse pastoralists who used domestic horses
as sources of meat, and perhaps of milk. A good case can
also be made that the Botai rode horses to aid in managing
herds of domestic horses and to enhance their success in
hunting wild horses and some aurochs, elk, red deer, and saiga
The probable mixed use of horses for both food and riding
makes sorting out the economic focus of domestic horse
exploitation a complex matter. The likely mixing of the
remains of domestic and hunted wild horses in Botai mid-
dens further complicates efforts to document early horse
domestication using traditional methods like size, mor-
phology, and mortality patterns. Clearly, a combination of
traditional direct, as well as indirect, circumstantial evidence
must be employed to build the case for horse domestication.
This is not to say that the Botai were the first to develop
horse domestication. In fact, early indications are that either
people from the Urals moved into this region in the Copper
Age, bringing domestic horses with them, or that the indige-
nous Neolithic people adopted horse domestication from
neighbors to the west. Moreover, as mentioned above and
discussed in detail in Chapter 23 of this volume, recent
EARLY HORSE DOMESTICATION ON THE STEPPE 265
DNA studies (Vila et al. 2001) suggest that horse domestica-
tion occurred more than once and perhaps many times along
the Eurasian steppe. Therefore, it would not be surprising if
other steppe cultures produce similar evidence in the near
future. Ongoing work with the Botai culture of northern
Kazakhstan provides important direction for those seeking
to document this complex process across this broad region.
I would like to offer my gratitude to the National Science
Foundation (Grants #BS 9816476 and #BCS 0415441),
National Geographic Society, and Carnegie Museum of
Natural History for their tremendous support for this research.
My special thanks go out to the staff of the Presidential
Cultural Center of Kazakhstan and the Kokshetau History
Museum, Bruce Bradley, Alan Outram, our field team, all
the students from Kazakhstan and America who have helped
gather field data, and to Mary Littauer for the many years of
advice she has given me. Finally, and most especially, I would
like to thank Maral Ghabdulina and honor the memory of
her husband, Kimal Akishev, for sharing their exhaustive
knowledge of the prehistory of Kazakhstan and for making
it possible for our project in Kazakhstan to happen.
Agapov, S. A., I. B. Vasiliev, and V. I. Pestrikova. 1990. Khvalynskii
eneoliticheskii mogil’nik. Kuibyshev: Saratovskogo Universiteta,
Akhinzhanov, S. M., L. A. Makarova, and T. N. Nurumev. 1992.
K istorii Skotovodstva i okhoty v Kazakhstane. Almaty: Akademia
Nauk Kazakhskoi CCP.
Anthony, D. 1996. Bridling horse power: The domestication of
the horse. In Horses through time, S. L. Olsen (ed.), pp. 57–82.
Boulder, CO: Roberts Rinehart Publishers.
Anthony, D. and D. Brown 1989. Looking a gift horse in the
mouth: Identification of the earliest bitted equids and micro-
scopic analysis of bit wear. In Early animal domestication and
its cultural context, P. J. Crabtree, D. Campana, and K. Ryan
(eds.), pp. 98–116. MASCA Research Papers in Science and
Archaeology, Volume 6, special supplement. Philadelphia:
MASCA, The University Museum.
______. 1991. The origins of horseback riding. Antiquity 65: 22–38.
______. 1998. Bit wear, horseback riding and the Botai site in
Kazakstan. Journal of Archaeological Science 25: 331–347.
______. 2000. Eneolithic horse exploitation in the Eurasian
steppes: Diet, ritual, and riding. Antiquity 74: 75–86.
______. in press. Early horseback riding and warfare in the steppes.
In Horses and humans: The evolution of equine-human relation-
ships, S. L. Olsen, L. Bartosiewicz, and A. Choyke (eds.). BAR
International Series (number unassigned), Oxford.
Antonius, O. 1913. Equus abeli nov. spec.: Ein Beitrag zur
Genaueren Kenntnis unserer Quartärpferde. Beiträge zur
Paläontologie und Geologie Österreich-Ungarns und des Orients
Baker, J. and D. Brothwell. 1980. Animal diseases in archaeology.
London: Academic Press.
No Anthony and
Brown (in press)
cited in chapter.
Grant is listed as
an editor under
If Grant to be
change all other
entries for this
chapter17 2005.12.16 9:25 PM Page 265
Benecke, N. 1994. Zur Domestikation des Pferdes in Mittel- und
Osteuropa. Einige neue archäozoologische Befunde. In Die
Indogermanen und das pferd, B. Hänsel and S. Zimmer (eds.),
pp. 123–144. Budapest: Archaeolingua Alapítvány.
_____. in press. Late prehistoric exploitation of horses in Central
Germany and neighboring areas—the archaeozoological record.
In Horses and humans: The evolution of the human-equine rela-
tionship. S. L. Olsen, S. Grant, A. Choyke, and L. Bartosiewicz
(eds.). BAR International Series (number unassigned), Oxford.
Berger, J. 1986. Wild horses of the Great Basin: Social competition
and population size. Chicago: University of Chicago Press.
Bibikova, V. I. 1986a. A study of the earliest domestic horses of
eastern Europe. Appendix 2, Part 1. In Dereivka: A settlement
and cemetery of Copper Age horse keepers on the Middle Dnieper,
D. Telegin (ed.), pp. 135–149. BAR International Series, vol. 287.
______. 1986b. On the history of horse domestication in south-
east Europe. Appendix 3. In Dereivka: A settlement and cemetery
of Copper Age horse keepers on the Middle Dnieper, D. Telegin (ed.),
pp. 163–182. BAR International Series, vol. 287. Oxford: BAR.
Boessneck, J. and A. von den Driesch. 1976. Pferde in 4./3
Jahrtausend v. Chr. In Ostanatolien. Saügetierkundliche
Mitteilungen 31: 89–104.
Bökönyi, S. 1974a. The Przevalsky horse. Plymouth: Souvenir
______. 1974b. History of domestic mammals in Central and eastern
Europe. Budapest: Akadémiai Kiadô.
______. 1991. Late Chalcolithic horses in Anatolia. In Equids in
the ancient world, Volume 2, R. Meadow and H.-P. Uerpmann
(eds.), pp. 123–131. Wiesbaden: Dr. Ludwig Reichert Verlag.
Boyd, L. and K. A. Houpt (eds.). 1994. Przewalski’s horse: The
history and biology of an endangered species. Albany: State
University of New York Press.
Britnell, W. J. 1976. Antler cheekpieces of the British Late Bronze
Age. Antiquaries Journal 56: 24–34.
Brown, D. and D. Anthony. 1998. Bit wear, horseback riding
and the Botai site in Kazakstan. Journal of Archaeological Science
Bull, I. D., I. A. Simpson, P. F. van Bergen, and R. P. Evershed. 1999.
Muck ‘n’ molecules: Organic geochemical methods for
detecting ancient manuring. Antiquity 73: 86–96.
Butzer, K. W. 1971. Environment and archeology: An ecological
approach to prehistory, 2nd edition. Chicago: Aldine. .
Clark, G. 1941. Horses and battle-axes. Antiquity 15: 50–70.
Clason, A. T. 1991. Horse remains from Swifterbant, the
Netherlands. In Equids in the ancient world, Volume II,
R. Meadow and H.-P. Uerpmann (eds.), pp. 226–232.
Wiesbaden: Dr. Ludwig Reichert Verlag.
Clutton-Brock, J. 1974. The Buhen horse. Journal of Archaeological
Science 1: 89–100.
______. 1992. Horse power. Cambridge, MA: Harvard University
______. 1996. Horses in history. In Horses through time, S. L. Olsen
(ed.), pp. 83–102. Boulder, CO: Roberts Rinehart, Boulder.
Clutton-Brock, J. and R. Burleigh. 1991a. The skull of a Neolithic
horse from Grime’s Graves, Norfolk, England. In Equids in the
ancient world, Volume 2, R. Meadow and H.-P. Uerpmann
(eds.), pp. 242–249. Wiesbaden: Dr. Ludwig Reichert Verlag.
______. 1991b. The mandible of a Mesolithic horse from Seamer
Carr, Yorkshire, England. In Equids in the ancient world, Volume
266 ARCHAEOLOGY AND ANIMAL DOMESTICATION
2, R. Meadow and H.-P. Uerpmann (eds.), pp. 238–241.
Wiesbaden: Dr. Ludwig Reichert Verlag.
Clutton-Brock, J. and S. Davis. 1993. More donkeys from Tell Brak.
Iraq 55: 209–221.
Corridi, C. and L. Sarti. 1989–1990. Sulla presenza di Equus
nell’Eneolitico italiano. Rivista di Scienze Preistoriche 42:
Curci, A. and A. Tagliacozzo. 1995. Il pozzetto rituale con scheletro
di cavallo dell’abitato eneolitico di Le Cerquete-Fianello
(Maccarese-RM). Origini: Preistoria e Protostoria delle Civiltà
Antiche 18: 297–350.
Davidsen, K. 1978. The final TRB culture in Denmark. Arkæologiske
Studier V. Copenhagen: Akademisk Forlag.
Davis, S. 1976. Mammal bones from the Early Bronze Age city
of Arad, northern Negev, Israel: Some implications concerning
human exploitation. Journal of Archaeological Science 3: 153–164.
Dmitriez, N. G. and L. K. Ernst. 1989. Animal genetic resources of
the USSR. Animal Production and Health Paper. Rome: FAO.
Dudd, S. N. and R. P. Evershed. 1998. Direct demonstration
of milk as an element of archaeological economies. Science
Duerst, U. J. 1908. The horse of Anau in its relation to history
and the races of domestic horses. In Explorations in Turkestan,
expedition of 1904, publication 73, R. Pumpelly (ed.),
pp. 339–442. Washington, D.C.: Carnegie Institution of
Eisenmann, V. 1984. Sur quelques caractères adaptatifs du squelette
d’Equus (Mammalia, Perissodactyla) et leurs implications paléoé-
cologiques. Bulletin du Muséum National d’Histoire Naturelle,
série, 6, section C (2): 185–195.
______. 1996. Quaternary horses: Possible candidates to domes-
tication. In Proceedings of the XII Congress, International Union
of Prehistoric and Protohistoric Sciences. 6: 27–36.
Epstein, H. 1971. The origin of domestic animals of Africa, Volume
II. New York: Africana Publishing Corporation.
Evershed, R. P., P. H. Bethell, P. J. Reynolds, and N. J. Walsh.
1997. 5ß-Stigmastanol and related 5ß-stanols as biomarkers of
manuring: Analysis of modern experimental material and
assessment of the archaeological potential. Journal of
Archaeological Science 24: 485–495.
Ewart, J. C. 1907. The multiple origins of horses and ponies.
Transactions of the Highland and Agricultural Society of Scotland,
Feist, J. and D. McCullough. 1975. Reproduction in feral horses.
Journal of Reproduction and Fertility, Supplement 23: 13–18.
Flint, R. F. 1957. Glacial and Pleistocene geology. New York: Wiley
Forsten, A. 1988. The small caballoid horse of the Upper
Pleistocene and Holocene. Journal Animal Breeding and Genetics
French, C. and M. Kousoulakou. 2003. Geomorphological and
micromorphological investigations of palaeosols, valley sed-
iments and a sunken floored dwelling at Botai, Kazakhstan. In
Prehistoric steppe adaptation and the horse, M. Levine, A. C.
Renfrew, and K. Boyle (eds.), pp. 105–114. Cambridge,
UK: McDonald Institute for Archaeological Research.
Gehl, O. 1961. Die Säugetiere. In Hohen viecheln: Ein mittel-
steinzeitlicher wohnplatz in Mecklenburg, E. Schuldt (ed.),
pp. 40–63. Berlin: Akademie-Verlag.
Gening, V. F. 1979. The cemetery at Sintashta and the early Indo-
Iranian peoples. Journal of Indo-European Studies 7:1–30.
chapter17 2005.12.16 9:25 PM Page 266
Gening, V. F., G. B. Zdanovich, and V. V. Gening. 1992. Sintashta.
Chelyabinsk: Iuzhno-Ural’skoe Knizhnoe Izdatel’stvo.
Getty, R. (ed.). 1975. Sisson and Grossman’s the anatomy of the
domestic animals. Philadelphia: W. B. Saunders.
Gheorghiu, D. 1994. Horse head scepters: First images of yoked
horses. Journal of Indo-European Studies 22: 221–249.
Grant, A. 1984. Survival or sacrifice? A critical appraisal of animal
burial in Britain in the Iron Age. In Animals and archaeology:
4. Husbandry in Europe, C. Grigson and J. Clutton-Brock (eds.),
pp. 221–227. BAR International Series 227. Oxford: BAR.
Greaves, W. S. 1973. The inference of jaw motion from tooth wear
facets. Journal of Paleontology 47: 1000–1001.
Green, M. 1992. Animals in Celtic life. London: Routledge.
Greenfield, H. in press. The social and economic context for
domestic horse origins in southeastern Europe: A view
from Ljuljaci in the central Balkans. In Horses and humans:
The evolution of equine-human relationships, S. L. Olsen,
L., Bartosiewicz, and A. Choyke (eds.). Oxford: BAR.
Grigson, C. 1993. The earliest domestic horses in the Levant? New
finds from the fourth millennium of the Negev. Journal of
Archaeological Science 20: 645–655.
Groves, C. 1986. The taxonomy, distribution, and adaptations of
recent equids. In Equids in the ancient world, R. H. Meadow and
H.-P. Uerpmann (eds.), pp. 11–66. Wiesbaden: Dr. Ludwig
Guthrie, R. D. 2003. Rapid body size decline in Alaskan Pleistocene
horses before extinction. Nature 426: 169–171
Habermehl, K. H. 1975. Die alterbestimmung bei haus und labotieren.
Berlin: Verlag Paul Parey.
Hillson, S. 1986. Teeth. Cambridge Manuals in Archaeology.
Cambridge: Cambridge University Press.
Hyland, A. 1990. Equus: The horse in the Roman world. London:
Jones-Bley, K. 1997. Defining Indo-European burial. In Varia on
the Indo-European past: Papers in memory of Marija Gimbutas.
Journal of Indo-European Studies Monograph 19: 194–221.
Kainer, R. A. and T. O. McCracken. 1994. The coloring atlas of horse
anatomy. Loveland, CO: Alpine Publications.
Kalieva, S. S. and V. N. Logvin. 1997. Skotovodi turgaya v tretem
tësacheletii do nashey eri. Kazakhstan: Institut Arkheologii Im.
A. Margulana, Kustanai.
Keiper, R. and K. Houpt. 1984. Reproduction in feral horses:
An eight-year study. American Journal of Veterinary Research
Kislenko, A. and N. Tatarentseva. 1999. The eastern Ural
steppe at the end of the Stone Age. In Late prehistoric exploita-
tion of the Eurasian steppe, M. Levine, Y. A. Kislenko, and N.
Tatarintseva (eds.), pp. 183–216. Cambridge: McDonald
Kosikowski, F. 1982. Cheese and fermented milk foods. Brooktondale,
NY: F. V. Kosikowski & Associates.
Kosintsev, P. in press. The horse and man in Neolithic to Early
Iron Age in the territory from the Volga to the Irtysh Rivers.
In Horses and humans: The evolution of equine-human relation-
ships, S. L. Olsen, L. Bartosiewicz, and A. Choyke (eds.). Oxford:
Kremenetski, C., P. Tarasov, and A. Cherkinsky. 1997. Postglacial
development of Kazakhstan pine forests. Geographie physique
et Quaternaire, 51: 391–404.
Kuzmina, I. E. 1997. Loshadi Severnoi Evrazii ot Pliotsena do
Sovremennosti. St. Petersburgh, Russia: Rossiiskaya Akademiya
Nauk, Trudy Zoologicheskovo Institututa, Tom 273.
EARLY HORSE DOMESTICATION ON THE STEPPE 267
Legge, A. and P. Rowley-Conwy. 2000. The exploitation of ani-
mals. In Village on the Euphrates, A. Moore, G. Hillman, and
A. Legge (eds.), pp. 423–471. Oxford: Oxford University Press.
Lernau, H. 1978. Faunal remains, strata III-I. In Early Arad,
R. Amiran (ed.), pp. 83–113. Jerusalem: Israel Exploration
Levine, M. 1982. The use of crown height measurements and
eruption-wear sequences to age horse teeth. In Ageing and
sexing of animal bones from archaeological sites, B. Wilson,
C. Grigson, and S. Payne (eds.), pp. 223–250. BAR British
Series 109. Oxford: BAR.
______. 1993. Social evolution and horse domestication. In Trade
and exchange in prehistoric Europe, C. Scarre and F. Healy (eds.),
pp. 135–141. New York: Oxford University Press.
______. 1999a. Botai and the origins of horse domestication.
Journal of Anthropological Archaeology 18: 29–78.
______. 1999b. The origins of horse husbandry on the Eurasian
steppe. In Late prehistoric exploitation of the Eurasian steppe,
M. Levine, Y. Rassamakin, A. Kislenko, and N. Tatarintseva
(eds.), pp. 5–58. Cambridge: McDonald Institute Monographs.
Levine, M. and A. M. Kislenko. 1997. New Eneolithic and early
Bronze Age radiocarbon dates for north Kazakhstan and south
Siberia. Cambridge Archaeological Journal 7: 297–300.
Littauer, M. A. 1963. How great was the “Great Horse”? A reassess-
ment of the evidence. Light Horse 13: 350–352.
______. 1971. V. O. Vitt and the horses of Pazyryk, Antiquity
______. 1981. Early stirrups. Antiquity 55: 99–105.
Littauer, M. and J. Crouwel. 1984. Ancient Iranian horse helmets.
Iranica Antiqua 19: 41–51.
______. 1986. A Near Eastern bridle bit of the second millennium
BC in New York. Levant 18: 163–167.
______. 1988. A pair of horse bits of the second millennium BC
from Iraq. Iraq 50: 169–171.
______. 2001. The earliest evidence for metal bridle bits. Oxford
Journal of Archaeology 20: 329–338.
Longley, D. 1980. Runneymede Bridge 1976: Excavations on the site
of a Late Bronze Age settlement. Research Volume of the Surrey
Archaeological Society No. 6. Castle Arch. Guildford, England:
Surrey Archaeological Society.
Lundholm, B. 1949. Abstammung und Domestikation des
Hauspferdes. Zoologiska Bidrag från Uppsala 27: 3–292.
Mair, V. 1998. Canine conundrums: Eurasian dog ancestor myths in
historical and ethnic perspective. University of Pennsylvania
Sino-Platonic Papers 87. Philadelphia: University of
Mallory, J. P. 1981. The ritual treatment of the horse in the early
Kurgan tradition. Journal of Indo-European Studies 9: 205–226.
______. 1996. In search of the Indo-Europeans: Language, archaeology,
and myth. London: Thames and Hudson.
Mange, A. P. and E. J. Mange, 1980. Genetics: Human aspects.
Philadelphia: W. B. Saunders.
Markman, S. 1969. The horse in Greek art. New York: Biblo and
Martin, L. and B. M. Gilbert. 1978. Excavations at Natural
Trap Cave. Transactions of the Nebraska Academy of Sciences
Mashcour, M., M. Fontugne, and C. Hatte. 1999. Investigations
on the evolution of subsistence economy in the Qazvin Plain
(Iran) from the Neolithic to the Iron Age. Antiquity 73: 65–76.
Not cited in text.
chapter17 2005.12.16 9:25 PM Page 267
Matyushin, G. 1986. The Mesolithic and Neolithic in the southern
Urals and Central Asia. In Hunters in transition, M. Zvelebil (ed.),
pp. 133–150. Cambridge: Cambridge University Press.
______. 2003. Exploitation of the steppes of Central Eurasia in the
Mesolithic-Eneolithic. In Prehistoric steppe adaptation and the
horse, M. Levine, C. Renfrew, and K. Boyle (eds.), pp. 367–393.
Cambridge, UK: McDonald Institute for Archaeological
Merrifield, R. 1987. The archaeology of ritual and magic. New York:
New Amsterdam Books.
Meunier, K. 1962. Zur diskussion über die typologie des
hauspferdes und deren zoologisch-systematische bedeutung.
Zeitschrift für Tierzüchtung und Züchtungsbiologie 76: 225–237.
Milisauskas, S. 1978. European prehistory. New York: Academic
Mohr, E. 1971. The Asiatic wild horse. London: J. A. Allen and Co.,
Molleson, T. and J. Blondiaux. 1994. Riders’ bones from Kish, Iraq.
Cambridge Archaeological Journal 4: 312–316.
Moore, A. 2000. The excavation of Abu Hureyra 2. In Village on
the Euphrates, A. Moore, G. Hillman, and A. Legge (eds.),
pp. 189–259. Oxford: Oxford University Press.
Nobis, G. 1971. Vom wildpferd zum hauspferd. Fundamenta, Reihe
B, Band 6. Köln: Bohlau Verlag.
Olsen, S. L. 1989. Solutré: A theoretical approach to the recon-
struction of Upper Palaeolithic hunting strategies. Journal of
Human Evolution 18: 295–327.
______. 1995. Pleistocene horse-hunting at Solutré: Why bison
jump analogies fail. In Ancient peoples and landscapes, E. Johnson
(ed.), pp. 65–75. Lubbock: Texas Tech University Press.
______. 1996. Prehistoric adaptation to the Kazak steppes. In The
colloquia of the XIII International Congress of Prehistoric and
Protohistoric Sciences, Volume 16: The prehistory of Asia and
Oceania, G. Afanas’ev, S. Cleuziou, J. Lukacs, and M. Tosi (eds.),
pp. 49–60. Forlì, Italy: A.B.A.C.O. Edizioni.
______. 2000a. Expressions of ritual behavior at Botai, Kazakhstan.
In Proceedings of the eleventh annual UCLA Indo-European con-
ference, K. Jones-Bley, M. Huld, and D. Volpe (eds.), pp.183–207.
Journal of Indo-European Studies Monograph Series No. 35.
Washington, D.C.: Institute for the Study of Man.
______. 2000b. The sacred and secular roles of dogs at Botai,
north Kazakhstan. In Dogs through time: An archaeological per-
spective, S. Crockford (ed.), pp. 71–92. BAR International Series
889. Oxford: BAR.
______. 2001. The importance of thong-smoothers at Botai,
Kazakhstan. In Crafting bone: Skeletal technologies through time
and space, A. Choyke and L. Bartosiewicz (eds.), pp. 197–206.
BAR International Series 937. Oxford: BAR.
______. 2003. The exploitation of horses at Botai, Kazakhstan. In
Prehistoric steppe adaptation and the horse, M. Levine, C. Renfrew,
and K. Boyle (eds.), pp. 83–104. Cambridge: McDonald Institute
Olsen, S. L., B. Bradley, D. Maki, and A. Outram. in press. Copper
Age community organization in northern Kazakhstan. In
Proceedings of the University of Chicago conference on Eurasian
archaeology, D. Peterson (ed.). Monograph Series of Colloquia
Pontica. Leiden: Brill Academic Publishers.
Outram, A. and P. Rowley-Conwy. 1998. Meat and marrow utility
indices for horse (Equus). Journal of Archaeological Science
268 ARCHAEOLOGY AND ANIMAL DOMESTICATION
Perkins, D. and P. Daly. 1968. The potential of faunal analysis:
An investigation of the faunal remains from Suberde, Turkey.
Scientific American 219: 96–106.
Piggott, S. 1962. Head and hoofs. Antiquity 36: 110–118.
Rassamakin, Y. 1999. The Eneolithic of the Black Sea steppe:
Dynamics of cultural and economic development 4500–2300
BC. In Late prehistoric exploitation of the Eurasian steppe, M.
Levine, Y. Rassamakin, A. Kislenko, and N. Tatarintseva (eds.),
pp. 59–182. Cambridge: McDonald Institute Monographs.
Redding, R. 1981. The faunal remains. In An early town on the Deh
Luran Plain, excavations at Tepe Farukhabad, H. T. Wright (ed.),
pp. 233–261. Memoirs of the Museum of Anthropology,
University of Michigan, 13. Ann Arbor: University of Michigan
Reitz, E. J. and E. S. Wing. 1999. Zooarchaeology. Cambridge:
Cambridge University Press.
Rice, P. 1963. Changes in the Kazak pastoral economy: History and
analysis. MA thesis, Ohio State University, Columbus.
Riegel, R. J. and S. E. Hakola. 1996. Illustrated atlas of clinical
equine anatomy and common disorders of the horse, Volume 1.
Marysville, OH: Equistar Publications.
Rikushina, G. B. and V. F. Zaibert. 1984. Predvaritel’noe soob-
shchenie o skeletnykh ostatkakh lyudei s eneoliticheskovo
poceleniya Botai. In Bronzovyi Vek Uralo-Irtyshskovo mezh-
durech’ya, pp. 121–134. Chelyabinsk, Russia: Ministerstvo
Bysshevo I Srednevo Spetsial’novo Obrazovaniya RSFSR.
Rudenko, S. 1970. Frozen tombs of Siberia: The Pazyryk burials of
Iron-Age horsemen. M. W. Thompson (trans.). Berkeley:
University of California Press.
Ryder, O. 1994. Genetic studies of Przewalski’s horses and their
impact on conservation. In Przewalski’s horse: The history and
biology of an endangered species, L. Boyd and K. A. Houpt (eds.).
Albany: State University of New York Press.
Shorin, A. F. 1999. Eneolit Urala i Copredel’nykh Territoriii: Problemy
kul’turogeneza. Yekaterinburg, Russia: Rossiiskaya Akademiya
Silver, I. A. 1963. The ageing of domestic animals. In Science and
archaeology, D. Brothwell and E. Higgs (eds.), pp. 250–268.
New York: Basic Books.
Skorkowski, E. 1956. Systematik und abstammung des pferdes.
Zeitschrift für Tierzüchtung und Züchtungs-biologie 68: 42–74.
Steppan, K. 1998. Climatic fluctuations and Neolithic adaptations
in the 4th millennium BC: A case study from south-west
Germany. In Papers from the EAA third annual meeting at Ravenna
1997, Volume I: Pre-and protohistory, M. Pearce and M. Tosi
(eds.), pp. 38–45. BAR International Series 717. Oxford: BAR.
______. in press. Neolithic human impact and wild horses in
Central Europe. In Horses and humans: The evolution of human-
horse relationships, S. Olsen, L. Bartosiewicz, and A. Choyke
(eds.). Oxford: BAR.
Telegin, D. 1986. Dereivka: A settlement and cemetery of Copper
Age horse keepers on the Middle Dnieper. BAR International
Series, Vol. 287. Oxford: BAR.
Telegin, D. Y. and I. D. Potekhina. 1987. Neolithic cemeteries
and populations in the Dnieper basin. BAR International Series,
Vol. 383. Oxford: BAR.
Toomre, J. S. 1994. Koumiss in Mongol culture: Past and present.
In Milk and milk products from medieval to modern times,
P. L ysaght (ed.), pp. 130–139. Edinburgh: Canongate Academic
chapter17 2005.12.16 9:25 PM Page 268
Uerpmann, H.-P. 1990. Die domestikation des pferdes im
Chalkolithikum West- und Mitteleuropas. Madrider Mitteilungen
Vasilev, I. and G. Matveeva. 1979. Mogil’nik u s. S’ezzhee na
R. Samara. Sovietskaya Arkheologiia 4: 147–166.
Vila, C., J. Leonard, A. Götherström, S. Marklund, J. Sandberg,
K. Lidén, R. Wayne, and H. Ellegren. 2001. Widespread origins
of domestic horse lineages. Science 291: 474–477.
Vitt, V. O. 1952. Loshad Pazyrykskikh kurganov. Sovietskaya
Arkheologiia 16: 163–205.
von den Driesch, A. 1976. A guide to the measurement of animal
bones from archaeological sites. Peabody Museum of Archaeology
and Ethnology, Bulletin No. 1. Cambridge, MA: Harvard
Vörös, I. 1981. Wild equids from the Early Holocene in the
Carpathian Basin. Folia Archaeologica 32: 37–68.
Wayne, R. K. 1986. Cranial morphology of domestic and wild
canids: The influence of development on morphological
change. Evolution 40: 243–261.
Welsh, D. 1975. Population, behavioural, and grazing ecology of the
horses of Sable Island, Nova Scotia. PhD dissertation. Dalhousie
EARLY HORSE DOMESTICATION ON THE STEPPE 269
Wijngaarden-Bakker, L. 1974. The animal remains from the
Beaker settlement at Newgrange, Co. Meath: First report.
Proceedings of the Royal Irish Academy. 74: 313–383.
Wilson, G. L. 1978. The horse and dog in Hidatsa culture. Lincoln,
Nebraska: J and L Reprint Co, Reprints in Anthropology 10.
Reprinted from Anthropological Papers of the American
Museum of Natural History 15, 2, 1924.
Zaibert, V. F. 1992. Atbasarskaya Cultura. Chelyabinsk: Rossiiskaya
______. 1993. Eneolit Uralo-Irtyshskovo. Almaty, Kazakhstan:
Academiya Nauk Respubliki Kazakhstan.
Zeder, M. A. 1986. The equid remains from Tal-e Malyan. In
Equids of the Old World, Volume 1, R. Meadow and H.-P.
Uerpmann (eds.), pp. 367–412. Beihefte zum Tubinger Atlas
des Vorderern Orients. Tubingen, Germany: University of
______. 1991. Feeding cities: Specialized animal economy in the
ancient Near East. Washington, D.C.: Smithsonian Institution
Zeuner, F. E. 1963. A history of domesticated animals. New York:
Harper and Row, New York.
Not cited in text.
chapter17 2005.12.16 9:25 PM Page 269
chapter17 2005.12.16 9:32 PM Page 270