Domestication of the Horse

Horse-Riding and Mounted Warfare

Two of the oldest images of humans riding horses, dated about 2100-1700 BC, were discovered in the ruins of a Sumerian city (top) and in the cemetery of a fortified Central Asian oasis of the Bactria-Margiana Archaeological Complex (BMAC) (bottom). The seal scenes, one of which was the personal seal of the Animal Disburser for the Sumerian King Shu-Sin, seem to draw on a similar iconography, which might have spread with the first wave of horses to enter the ancient Near East, happening at the time the seals were carved.

Sherratt described how the domestication of the horse was articulated with the SPR in three publications (1983, 1997a, 2003). Although his views changed, he consistently saw the principal significance of horses in military activities, associated first (about 2000 BC) with chariots and later (after 800 BC) with cavalry. He accepted a date in the fourth millennium BC for the earliest domestication of the horse, but like many he felt that horses were not used in war that early (Sherratt 2003:242). The earliest widespread evidence for the use of horses in warfare coincided with the invention of the metal bit and the chariot about 2000 BC (Littauer and Crouwel 2001), leading Sherratt to speculate that fourth-millennium horses might have been too small to ride comfortably, or that bits made of rope or leather might have been inadequate to control them in battle (Sherratt1997a: 217). Doubts about the size of steppe wild horses and the adequacy of organic bits were repeated by an influential group of archaeologists (Renfrew 2002:5-6; Kuzmina 2003:213; Drews 2004:20). They interpreted the spread of domesticated horses across Europe in the late fourth millennium BC as a trade in prestige symbols used as pack animals, and perhaps ridden with some difficulty, but whose potential as mounts in war remained unrealized (Sherratt 2003:242).

Sherratt connected the domestication of the horse with the domestication of the camel and the ass, all of which, he suggested, were responses to the need for heavy-burden transport during the long-distance trade boom caused by the Uruk expansion, 3800-3100 BC (Rothman 2001). He placed the domestication of the horse in the arid valley of the lower Kura River in what is now eastern Georgia and Azerbaijan, where local wild horses could have been domesticated to serve as pack animals for Uruk metal-traders who, in this scenario, brought pack-trains of domesticated asses to the copper mines of the Caucasus (Sherratt 2003:240). The Asian pack ass could have been the model for the domesticated pack-horse, a concept Sherratt borrowed from Uerpmann (1990). Donkey-riding could have inspired the earliest horse-riding, and that could explain why, in Near Eastern art, many early horse-riders are shown sitting on the rear of the horse, a seat suitable for a donkey but uncomfortable and ineffective on a horse (Drews 2004:40-55; Littauer and Crouwel 1979). This early (ineffective) style of horse riding could have been a product of the initial context of domestication on the northern frontier of the Uruk expansion.

We question most of this scenario, including its emphasis on warfare, its interpretation of horse sizes and capacities, and its location for horse domestication. (also see The domestication of the Horse below). On the subject of horse sizes, the Eneolithic (5200-3300 BC) horses of the Eurasian steppes were big enough to ride into battle. More than 70% of the Late Eneolithic horses at Dereivka, Ukraine (4200-3700 BC) and Botai, Kazakhstan (3600-3100 BC) stood 136-144cm at the withers (shoulders), or about 13-14 hands high (Benecke and von den Driesch 2003; Bibikova 1970). The horses ridden into battle by Roman cavalrymen commonly measured 120-150cm at the withers (Hyland 1990:68), and those of the American Plains Indians stood about 130-140 cm, or “a little under 14 hands” (Ewers 1955:33). Eneolithic steppe horses were about the same size as Roman and American Indian cavalry horses. On the question of rope bits, the authors conducted a riding experiment in which two expert riders rode never-bitted horses with rope and leather bits (Brown and Anthony 1998; Anthony, Brown and George 2006). Our riders had “no problem” controlling their horses. The American Plains Indians, regarded in the 19th century as among the finest light cavalry in the world, used a ‘war bridle’ that was just a rope looped around the lower jaw (Ewers 1955:76). History and experiment both show that horses the same size as Eneolithic steppe horses can be ridden effectively at a gallop, even in warfare, with a rope bit.

The notion that the seat used in early horseback riding was ineffective (Drews 2004:40-55; Sherratt 1997a:217) is derived from Near Eastern art. Horses were not native to the Near East and remained rare there until after about 1800 BC. The native Near Eastern equids were onagers (Equus hemionus), behaviorally resistant to domestication, distributed in the steppes of Iran, Syria and Iraq; and asses (Equus asinus), amenable to domestication, and distributed in the south, in Egypt and the southern Levant. Asses were domesticated as pack animals in Egypt perhaps around 4000 BC and were used for transport throughout the Near East by 3500 BC, long before horses appeared (Rossel et al. 2008). Asses have low withers (shoulders) and a high, broad rump. If a rider sits forward on an ass and the animal lowers its head the rider can easily fall forward. Donkey-riders wishing to retain their dignity therefore usually sit back on the rump and hang on to a belly band. Horses are built differently: they have high withers, so horse-riders sit forward, just behind the shoulders/withers, a seat that also permits the rider to grasp the mane. Bronze Age Near Eastern artistic images that show a rider on an animal that clearly is a horse, but sitting on the rump and grasping a belly band, probably indicate only that some Near Eastern artists before 1000 BCE were more familiar with donkey-riding than horse-riding. Other images correctly placed horse-riders in a forward seat, even among the earliest depictions of horse-riding, Akkadian and Ur III images dated 2300-1900 BC (Owen 1991; Oates 2003). Steppe horse-riders would not have used a seat suited only to the body of a donkey, an animal most of them had never seen.

Near Eastern art is an odd lens through which to examine the domestication of an animal that did not live in the Near East. Questions about the domestication of the horse are answered best in the region where wild horses were most important in human economies prior to their domestication—where humans relied on wild horse hunting for the majority of their meat diet. That region was the western Eurasian steppes, from the Dnieper to the Ural Rivers. Many settlements here dated long before 3500 BC contain more than 40% horse bones (Anthony 2007:198). The earlier settlements are interpreted as the camps of pedestrian wild horse hunters, but the later sites, after about 4500 BC, probably contain the bones of domesticated horses.

The Domestication of the Horse

The standard zoological indicators of domestication have proven uncertain guides to the identification of the earliest domesticated horses. The wild horses of the mid-Holocene varied naturally in size. The horses of the central Eurasian steppes in Kazakhstan (#s 7-9 in the figure on the right) were somewhat larger than those of the western steppes in central Ukraine (#6), which were larger than those of the steppe/forest-steppe border in western Ukraine and Romania (#s 4-5), and and all the horses of the steppes were significantly larger than the pony-sized wild horses of central and western Europe (#s 1-3). Steppe horses were about the same size as later Roman cavalry horses, but in the Eneolithic (Copper Age) this made them the largest horses in the world. Any east-west movement of wild horse populations might complicate the detection of size changes caused by domestication.

The increase in variation that began about 2500 BCE, indicated in the graph to the right by the horses of Csepel-Haros in eastern Hungary (#10), continued thereafter. This is often taken as an indicator of domestication beginning at about 2500 BCE (Benecke and von den Dreisch 2003), but increased variation is sensitive to sample size, and in any case the underlying assumptions about controlled breeding for particular types and sizes do not necessarily apply to the earliest stage of horse management.

Levine (1990, 1999) documented different mortality profiles for the horses at Dereivka (Ukraine) and Botai (Kazakhstan), two Eneolithic steppe sites critical for the understanding of early horse domestication. She interpreted both as wild populations, although the Dereivka profile was almost identical to that of the Roman site of Kesteren, where the horses certainly were domesticated. Given these problems with the consistency and even the significance of the standard zoological methods, studies of early horse domestication have fallen back on other indicators.


Pit Grave Culture

The Earliest Possible Indicator is the inclusion of horses in sacrificial rituals with domesticated cattle and sheep. Domesticated cattle and sheep were adopted across the western steppes between about 5000-4500 BCE (the tinted region in the map on the left).Horses certainly had a new ritual and symbolic importance in Volga-region sites dated between 5000-4500 BCE. They were the principal meat animal, followed by sheep and cattle, at the contemporary settlements of Ivanovskaya and Varfolomievka. If they were domesticated, these were the oldest domesticated horses.

At the cemetery of Khvalynsk on the Volga River, archaeologists led by Igor Vasiliev found 158 human graves dated by radiocarbon to about 4500 BCE with the bones of 52 sacrificed sheep/goat, 23 cattle, and 11 horses (Petrenko 1984; Agapov, Vasiliev, and Pestrikova 1990). Animal leg and head parts were offered in all phases of the funerals: on the grave floor, in the grave fill, at the edge of the grave, and in 12 sacrificial deposits stained with red ochre, found above the graves. Horse leg parts occurred by themselves in eight graves. They were grouped with sheep/goat and cattle bones in one grave and in one sacrificial deposit, in rituals that excluded obviously wild animals.

At a contemporary cemetery, S’yezzhe, two horse heads were found in a similar ochre-stained deposit with carved images of horses.

The Next Indicator of Domestication is the appearance of horse-head maces in graves in the Dnieper steppes and in the lower Danube valley (tinted below) about 4200 BCE. Graves of the Suvorovo type appeared in the drier, more open parts of lower Danube valley just before the tells of the "Old Europe" (Karanovo VI-Gumelnitsa-Varna cultures) were abandoned in neighboring regions. Polished stone zoomorphic maces were typical prestige objects in the steppes going back to Khvalynsk and Varfolomievka, but were absent from earlier Karanovo or Gumelniţa societies (Chapman 1999). Maces shaped into horse-heads probably were made by people for whom the horse was a powerful symbol. Horses averaged less than 6% of animal bones in "Old European" (Tripolye B1, Gumelniţa, and Karanovo) settlements, so were unimportant in Danubian diets. Horse-head maces signaled a new iconic status for horses in the Danube valley just when the Suvorovo graves appeared, 4200-4000 BCE, and just before the abandonment of hundreds of long-established tell settlements. Mounted raiding could have contributed to the collapse of "Old Europe."

Suvorovo Culture 4200 BC




By about 3500 BCE the bones of large horses, probably from the steppes, began to appear outside the steppes, in the Danube valley (Bokonyi 1979), in central and western Europe (Benecke 2006), and in the North Caucasus, Transcaucasia, and eastern Anatolia (Bokonyi 1991). At the same time the Botai culture appeared in the steppes of northern Kazakhstan, where several more lines of evidence suggest that horses were domesticated and ridden.


Botai culture 3600 BC

BOTAI: artist's reconstruction of settlement, excavated settlement plan, profile of house floor (looks suspiciously close to yurts, with a hearth in the center, a vent louver in the roof, and a storage of dry dung, which was used as hearth fuel, and any unused dung was left behind as having no value. Same footprint with dung is left by modern yurts. Yurts may have been covered with bark, vs. felt, as was recorded in Chinese ethnological description for Yanto tribe )


Evidence for Riding in the Fourth Millennium BC

The Botai culture in northern Kazakhstan is named after the site of Botai, where 99% of 300,000 recovered animal bones were from horses. Botai was a culture of foragers that rode horses to hunt horses, a peculiar adaptation found only here and only between about 3600-3000 BCE. 

Sandra Olsen’s (2003, 2006) excavations at Botai and Krasnyi Yar found layers of horse dung in house pits, suggesting either stable-cleaning or possibly a roof sealed with horse dung, clear indicators of domestication (likely just simple fuel storage). Also, whole horse carcasses were butchered in the settlement as a regular practice extending over hundreds of years. Since the Botai culture had no domesticated cattle, they had no animals big enough to drag carcasses, so at least some horses were kept for meat in or near the settlement.

Recently scientists led by Alan Outram at Exeter University have found traces of horse milk residues in clay pots from Botai. The milking of horses of course indicates that they were domesticated.

But is there direct evidence that they were ridden? Outram dismissed our evidence for bit wear at Botai. Bit wear is a pathology commonly found on the premolars of bitted horses. A bit is used only to ride or drive a horse, and Botai lacked wheeled vehicles, so bit wear would mean riding.

Bit Wear As A Method To Identify Ridden and Driven Horsessee: Brown, D. R., and D. W. Anthony, 1998, “Bit wear, horseback riding, and the Botai site in Kazakstan,” Journal of Archaeological Science 25:331-347. updated in chapter 10 of Anthony, David W., 2007, The Horse, The Wheel, and Language. Princeton University Press


Five lower second premolars from Botai, representing 3-5 horses, exhibited wear facets of a type that we have shown are associated exclusively with the use of a bit. (Anthony, Brown and George 2006; Brown and Anthony 1998). The diagnostic feature was a bevel or wear facet on the mesial or front corner of the premolars that measured 3mm or more. Wear facets of this depth are found, according to our research, exclusively on the teeth of bitted horses. (Very old and very young horses have naturally irregular teeth so can't be included in our studies. Horses fed on grain also develop many precularities in their occlusion. The 3mm bevel is diagnostic only on mature, pasture-fed horses.) Another site, Kozhai 1, dated to the same period, yielded 46,000 horse bones. We examined 12 lower P2s from horses, two of which showed wear facets of 3mm. Could an organic bit of rope or leather, the only kind available at Botai, create such wear facets?

Measurements of the bevel or facet on the front corner of the P2s of four horses ridden with a bone bit (top left), a hemp rope bit (top right), a leather bit (bot. left), and a horsehair rope bit (bot. right) over the course of 150 hours of riding (0-150 hours, horizontal scale; 0-3mm bevel, vertical scale).

Critics (Levine 2004) have charged that soft bits of rope or leather are too soft to create wear facets on horse teeth. We conducted tests on four previously never-bitted horses (Brown and Anthony 1998). They were ridden for 150 hours each with bits of hemp rope, horsehair rope, bone, and leather, and we made casts of their P2s at intervals. All bits, including leather, created wear facets on the lower P2s, as shown by the steady increase in the bevel measurements or facets on the lower P2 after 150 hours. Grit trapped under the bit probably was the agent of wear. Organic bits can cause bit wear.

Could natural malocclusion cause facets of 3mm or more, without human interference? Olsen (2006) found facets greater than 3mm on the premolars of two Pleistocene equids, and suggested that such facets were common in the wild, caused naturally by malocclusion, so do not indicate bitting. Olsen was a co-author of Outram's study noted above, and it was Olsen's criticism that Outram cited in dismissing our findings at Botai.

But Olsen did not screen her two specimens for age or condition, so we don't know if they are eligible for bit wear study. If they were from very old (20+) or very young horses (<3) their premolars could be expected to show many natural irregularities, so they would not be eligible. Olsen also failed to cite our 2006 study of Pleistocene premolars, which included not 2 teeth, but 74, in addition to 31 premolars of modern never-bitted horses, for a total sample of 105 premolars from never-bitted horses. This is the largest assemblage yet studied by anyone.

Christian George worked with us to study Pleistocene equid lower P2s from the Leisey site in Florida. He used our methods to eliminate teeth from immature equids (younger than 3 years) and from very old equids (see Horse tooth crown heights and ageing). This left 74 Pleistocene lower P2s from 44 mature equids.

Among the Leisey P2s George found no bevels greater than 3mm and just one greater than 2.5mm. In contrast, a bevel of 2.5mm or more occurred in the majority (58%) of mature bitted horses, and their maximum was 10mm (shown in the top row on the right with the measurements above the median tinted). George’s study supported a 3mm bevel as a cultural indicator, a threshold we had defined on the basis of a smaller sample of 31 modern never-bitted horse teeth. Facets 3mm or more in depth should be found on less than 1% of mature horse premolars among never-bitted horses. Zoo or stable feeding seems to affect occlusion in equids, so grain-fed domestic horses might show an increase in malocclusion that might complicate the identification of bit wear in domesticated, grain-fed populations. But at Botai, facets of 3mm or more should not have occurred unless the horses were bitted.

Wear facets 3-6mm deep occurred on five P2s from Botai (second row) and two from another Botai site named Kozhai1 (third row). Combined with the other indicators of domestication, bit wear on multiple teeth from two sites suggests that the Botai people rode horses.

The general increase in the importance of horses between 3500-3000 BCE in Kazakhstan, the Caucasus Mountains, eastern Anatolia, the Danube valley, and Germany; and the appearance in all of those regions of larger horses probably derived from the steppes, suggests that something had changed in the relationship between humans and horses. Botai and Kozhai 1 show what that something was. People had started to ride. Riding might well have started earlier, perhaps before 4000 BCE with the Khvalynsk culture. (We were not able to examine P2s from the Khvalynsk cemetery because all of the animal bones were discarded.)


Riding, Chariots and Warfare

Most observers have emphasized the effects of horseback riding on warfare. But we should differentiate between tribal raiding on horseback, which may have begun before 4000 BCE; and cavalry, which appeared only after 1000 BCE, initially as a specialized force of mounted archers.

Bow parts preserved in Bronze Age steppe kurgans suggest that bows were well over a meter long between 2500-1500 BCE. The length of the bow is directly related to its penetrating power, so until recurved bows were invented, there was a good reason to retain a long bow even if it made mounted archery difficult. Eneolithic and Bronze Age arrowheads were made in many sizes and weights. Long bows and non-standardized arrows, not rope bits or an ineffective riding seat, made Eneolithic and Bronze Age mounted warfare clumsy. Riding might have had a greater tactical role in the escape—often the most dangerous part of a pedestrian raid—than in the attack, which might still have been on foot.

Horse-aided tribal raiding tactics in the Eurasian steppes had little influence on the tactics of Near Eastern urban siege armies during the Bronze Age. This is not surprising if we accept Turney-High’s (1981:34; Keeley1996:47) injunction that warfare is social organization. It was only after Eurasian steppe chiefs developed the chariot during a period of intensified internal conflict and increased social stratification in the steppes that steppe tactics began to influence warfare in the Near East. 

In the southeastern Ural steppes not far from Botai, but a millennium later, at least 16 graves of the Sintashta culture contained vehicles with two spoked wheels.They are dated 2100-1700 BCE, older than the oldest chariots known in the Near East. The Sintashta chariot wheels were set in slots in the grave floors. As they rotted they left stains preserving the details of their spokes and diameters. There has been some question about whether the Sintashta chariots could have been used in war, with the doubts based largely on their size. War chariots in the later Near East carried a driver and a warrior, and had to have a gauge (width between the wheels) of at least 1.5m in order to provide sufficient room for two. Some of the Sintashta chariots were narrower than 1.5m, but as the diagram below shows, many were wide enough. And in any case, if the chariot warrior used a javelin, he could throw and drive, making a second man unecessary. For that matter he could loop the reins around his waist and fire a bow, a fighting method shown in later Egyptian tomb paintings.

Most of the Sintashta chariot graves contained weapons. A new kind of long stemmed point, perhaps designed for javelins, appeared in sets of up to 20 in Sintashta chariot graves. Charioteers could stand in a car and hurl a javelin with their entire body while riders could use only their arm, so javelins were more effective from a chariot than from horseback. Bronze Age bows probably were too long for mounted archery.


The steppe war chariot could have been introduced to the Near East through Central Asia, where there was clear and undoubted evidence for contact between the fortified, brick-walled oasis cities of the Bactria-Margiana Archaeological Complex (BMAC) and Sintashta and Petrovka charioteers from the northern steppes about 2100-1800 BCE. The BMAC oasis communities had far-flung trade relationships with cities and citadels across the Iranian Plateau, where the Elamite state and other smaller alliances were locked in combat with Mesopotamian kings during the Ur III period. This was when horses first became common in the Near East. The earliest Near Eastern chariots had appeared by about 1900-1800 BCE, and they soon became a necessary part of any successful urban army. For about a thousand years, chariots were an elite super-weapon.

But around 1200 BCE the short recurved “cupid” bow was developed, perhaps in Shang China. Cast socketed bronze arrowheads had been made sporadically in the Kazakh steppes from Sintashta through the second millennium BC. Around 1000-800 BC, the recurved bow was united with a method for mass-producing socketed bronze arrowheads of standard weights and sizes. The short but powerful bow and the standard-weight arrowhead together might have been the innovations that made mounted archery truly deadly.

Still, technical advances in bows and arrows were meaningless without a matching change in mentality, in the identity of the fighter, from a heroic single warrior to a nameless soldier. The defining feature of cavalry was that it attacked and retreated as a body in which individual riders became anonymous. An ideological model of fighting that was appropriate for a state, under the leadership of a general, was grafted onto tribal horseback riders, making cavalry an effective new weapon.

That shift occurred somewhere in the steppes between about 1000-800 BCE. After it, cavalry swept chariotry from the battlefield and a new era in warfare began.