The earliest horse riders

« previous post | next post »

The implications of horse domestication — above all its consequent equine chariotry and horseback riding — for the spread of Indo-European are topics we have addressed on numerous occasions before.  A paper that was published just two days ago has made a stunning, convincing breakthrough concerning when and where humans began to ride horses:

"First bioanthropological evidence for Yamnaya horsemanship"

Martin Trautmann, Alin Frnculeasa, Bianca Preda-Blnic, Marta Petruneac, Marin Focneanu, Stefan Alexandrov, Nadezhda Atanassova, Piotr Wodarczak, Micha Podsiado, Jnos Dani, Zsolt Bereczki, Tams Hajdu, Radu Bjenaru, Adrian Ioni, Andrei Mgureanu, Despina Mgureanu, Anca-Diana Popescu, Dorin Srbu, Gabriel Vasile, David Anthony, and Volker Heyd

Science Advances, 9 (9), eade2451. DOI: 10.1126/sciadv.ade2451

View the article online

Yamnaya (c. 3300-2600 BC)

As can be seen from the abstract and several paragraphs from the introduction and the main text quoted below, the research presented in this paper provides conclusive evidence of changes in bone morphology and bone pathology associated with human horse riding.


The origins of horseback riding remain elusive. Scientific studies show that horses were kept for their milk ~3500 to 3000 BCE, widely accepted as indicating domestication. However, this does not confirm them to be ridden. Equipment used by early riders is rarely preserved, and the reliability of equine dental and mandibular pathologies remains contested. However, horsemanship has two interacting components: the horse as mount and the human as rider. Alterations associated with riding in human skeletons therefore possibly provide the best source of information. Here, we report five Yamnaya individuals well-dated to 3021 to 2501 calibrated BCE from kurgans in Romania, Bulgaria, and Hungary, displaying changes in bone morphology and distinct pathologies associated with horseback riding. These are the oldest humans identified as riders so far.


Multidisciplinary sources of evidence for earliest horsemanship

Using horses for transport was a decisive step in human cultural development. Trade and cultural exchange as well as conflicts and migrations leapt with the increase in speed and range provided by horsemanship. Archeological, archeozoological, and paleogenetic research into the beginnings of horse domestication and the initial expansion of domesticated horses (Equus caballus) has recently seen much progress (1, 2), as has our understanding of the appearance of horse-drawn fast chariots with spoked wheels ~2000 BCE (3).
However, information for earliest horseback riding so far is sparse (see section S1 for a detailed review). Possible bit wear in premolar teeth of horses from Botai (Kazakhstan) dating to <3500 BCE were extensively debated during the past three decades (46). Information from the Botai site such as horse demography, horse dung finds, potential paddock fences, or horse milk traces in pot shards (7, 8), as well as horse milk peptides in the calculus of Yamnaya individuals from Krivyanskiy 9 (Russia; ~3000 BCE) (9), suggests that domestication became widely established during the second half of the fourth millennium BCE. However, these do not provide direct evidence for riding.

Drawing on extensive osteological and figurative evidence, the authors are able to pinpoint the time and place of the rise of horseback riding. 

This time span [between mid-fourth and early second millennium BCE in the Pontic-Caspian steppe and the Middle East] … sees the first horse dispersals to the west and south (1, 2, 10, 12), the origins of modern horse breeds (1), the widespread introduction of cattle-pulled wheeled carts and wagons (13), and the Yamnaya (~3200 to 2500 BCE) expansions eastward to the Altai and Mongolia in the form of the Afanasievo culture (14) and westward into the southeast of Europe, coming to a hold at the Tisza river in eastern Hungary (15). Latest research into this event (Fig. 1) indicates its rapid accomplishment within one or two centuries just before and after 3000 BCE. Considering the vast geographical distances of 4500 km between the Tisza river and the Altai mountains, the absence of roads, and the small overall population sizes, it is difficult to envision how this expansion could have taken place without improved means of transport.

I believe that this sort of hard, physical evidence is more reliable than that derived from genetics, which is susceptible to statistical manipulation and liable to speculative interpretation.  This is a trend that has been particularly noticeable in recent years (e.g., with respect to the origins of the Tarim mummies and the Turkic / "Altaic" peoples and their languages in studies emanating from certain genetics research labs.

Although this paper from Trautmann, et al. does not directly address language issues, when plugged into the numerous Language Log posts on the relationship between archeology and linguistics, it makes a significant, solid breakthrough in IE dispersals.

The Yamnaya culture is identified with the late Proto-Indo-Europeans, and the Pontic-Caspian steppe is the strongest candidate for the Urheimat (original homeland) of the Proto-Indo-European language.

(source; passim)


Selected readings


[Thanks to Dan Waugh]


  1. Taylor, Philip said,

    March 5, 2023 @ 10:52 am

    Purely tangential, but where have the diacritics gone in the authors' names ? Just a glance suggested that some may have been omitted, and a quick check reveals the following differences from the names as they appear above —

    Marin Focşǎneanu
    Piotr Włodarczak
    Michał Podsiadło
    János Dani
    Tamás Hajdu
    Radu Băjenaru
    Adrian Ioniță
    Andrei Măgureanu
    Despina Măgureanu
    Dorin Sârbu

    I am used to having to complain when The Guardian omits diacritics, but they are not normally omitted here …

  2. David Marjanović said,

    March 5, 2023 @ 3:36 pm

    I believe that this sort of hard, physical evidence is more reliable than that derived from genetics, which is susceptible to statistical manipulation and liable to speculative interpretation.

    It is fortunate, then, that nobody has ever claimed that a cultural practice like riding would somehow show up in a gene.

    I do apologize if I've misunderstood your point somehow; but if so, please explain what you meant.

    Although this paper from Trautmann, et al. does not directly address language issues, when plugged into the numerous Language Log posts on the relationship between archeology and linguistics, it makes a significant, solid breakthrough in IE dispersals.

    No, why? What was necessary to make these dispersals possible was the wheeled cart, not riding.

  3. Victor Mair said,

    March 5, 2023 @ 6:00 pm

    From Melinda Takeuchi, professor of Japanese art history at Stanford, who for years regularly competed with horse and carriage in combined driving events (see "'The old man at the pass loses his horse'" (5/2/20):

    when i read that article i wondered about the assumption that controlling a horse necessarily involved a metal bit. Native Americans used a simple loop thru the horse's mouth and lower jaw. and there are all sorts of ways of manipulating a horse's tender nose (hackamore-like tack) that don't go into the mouth. many people can ride can control a horse with just a rope around the lower neck ( i cannot, of course, judge the quality of the argument involving human skeletons. i can say from long experience, however, that anyone who deals with horses inevitably gets stepped on (as i did recently by my 17.1hh 1400 lb. horse), and records of broken toes and foot bones should be factored into the question of horse-taming. mimi yiengprksawan tells me that xrays of her feet reveal multiple breaks from her days as a wrangler, and i'm sure mine would too if we had any pix.

    btw, we recently adopted a 3-y.o. wild mustang culled from BLM herds in nevada and sent to the carson city prisoner training program. it's wonderful: save a horse/give a prisoner a skill. the guy who trained him was in for burglary and arson! he was so engaging that we named the horse after him! this is one amazing little horse.

    (click to embiggen)

    my vet sent off a piece of mane for dna testing, with the most remarkable results: first: quarter horse (to be expected). second: lippizaner (spanish riding school flunk-outs?). third: TURKOMAN!!!!!! this horse is sensible, talented, sure-footed, and a total blast to ride on trails.

  4. Victor Mair said,

    March 5, 2023 @ 8:12 pm

    From Robert Drews (Early Riders: The Beginnings of Mounted Warfare in Asia and Europe [2004]; Militarism and the Indo-Europeanizing of Europe [2017]; and many other works on the impact of the horse upon military affairs):

    Interesting article, osteologically, and thanks for calling it to my attention. But I am still convinced that what all three of us would call "riding" began ca. 2000 BCE, and not in the early third, the fourth or even the fifth millennium (in the case of one individual cited who had "riders' bones"). Surely horses were domesticated by the fourth millennium, and used as pack animals, with some kind of rudimentary control by the person leading the pack horse. And the "pack" atop the horse was probably often another person. But actual riding, no. We have enough evidence of what riding looked like ca. 2000 BCE to see that ca. 2000 BCE it was still a very new and very risky sport. And it didn't get much better until ca. 1000 BCE.

  5. Chester Draws said,

    March 5, 2023 @ 11:12 pm

    I would expect changes in dress to follow changes to significant riding. Horse culture always comes with that, even today.

    Also the rise of effective riding is accompanied by the demise of the chariot as a weapon of war.

    So I also vote for much later dates for effective riding.

  6. Taylor, Philip said,

    March 6, 2023 @ 2:38 am

    Victor wrote "there are all sorts of ways of manipulating a horse's tender nose (hackamore-like tack) that don't go into the mouth". I can second that. For the entire time that I had Jingo (18 years, from age 9 to his death at age 27) I wrode him in a scawbrig, and I also used the scawbrig (with considerable success) on other horses that resented a bit in their mouths. A scawbrig does not give the same control as a bit (I was carted, on one near-fatal occasion) but in general it works very well and is a far kinder control aid.

  7. Victor Mair said,

    March 6, 2023 @ 6:06 am


    "What a prehistoric pair of pretty pants can tell us about the spread of early languages" (4/3/21)

    The world's oldest known pair of trousers, late 2nd millennium BC, from Turfan in Eastern Central Asia (ECA).

    We also find men wearing trousers in graves from the first half and middle of the 1st millennium BC in the same part of the world (ECA).

    "Cherchen Man" (

  8. Victor Mair said,

    March 6, 2023 @ 10:53 am

    "Humans Started Riding Horses 5,000 Years Ago, New Evidence Suggests"

    Archaeologists have found a handful of human skeletons with characteristics that have been linked to horseback riding and are a millennium older than early depictions of humans riding horses


    By Meghan Bartels on March 3, 2023

    Scientific American

    Humans Started Riding Horses 5,000 Years Ago, New Evidence Suggests


  9. Chris Button said,

    March 6, 2023 @ 2:12 pm

    Linguistics and archaeology certainly seems a far more natural pairing to me than anything to do with genetics.

  10. KevinM said,

    March 7, 2023 @ 4:46 pm

    @ChrisB: Agreed. So far as I know, cowboys did not pass along bowlegs or saddle sores to their offspring.

  11. Taylor, Philip said,

    March 7, 2023 @ 5:11 pm

    "cowboys did not pass along bowlegs or saddle sores to their offspring" — shurely shome confushion — it is the horse that suffers from saddle sores, not the rider …

  12. Jonathan Smith said,

    March 7, 2023 @ 5:16 pm

    Genetics + archaeology are a natural pair given the rather important overlap at animal (inclusive of human) + plant remains. Whereas outside of cases where recovered writing is involved, associating linguistics with either of the two is a highly problematic undertaking — probably fundamentally misguided in some sense.

  13. ASmith said,

    March 7, 2023 @ 7:19 pm

    @PTaylor: The saddle sores on both sides.

  14. Peter Grubtal said,

    March 8, 2023 @ 12:50 am

    " cowboys did not pass along bowlegs or saddle sores to their offspring"

    Epigenetics has been seized on by some as showing that Darwinism is incorrect.

  15. Keith said,

    March 8, 2023 @ 1:35 am

    @KevinM Saddle sores?

    A person is "saddle sore" from riding when not accustomed to it. A person doesn't get "saddle sores" like a bed-ridden person gets "bed sores". A cowboy who spends hours in the saddle, day after day, will not feel saddle sore.

    One the other hand, a horse can suffer from saddle sores if a saddle fits badly or if a blanket between the saddle and the horse's back ruffles up and creases, creating pressure spots.

  16. Victor Mair said,

    March 8, 2023 @ 7:36 am

    Woman and horse

    Stacy Westfall's Championship Run 2006
    Stacy Westfall riding horse with no tack.

    She's from Ohio.

    At the root of her technique is the principle to "think-like-a-horse”

    Brought tears to my eyes.

  17. Victor Mair said,

    March 8, 2023 @ 7:49 am

    Horse and man

    Further to the point that bits and bridles are not essential to controlling horses, to see what is possible by training, horsemanship, and an understanding between horse and man, watch the following video. (You will, if you look closely, notice that a signal to turn is sometimes given by a tap on the horse's shoulder.)

    "A bit does not stop the horse. It tells the horse that we wish to stop."

    Gala D'oro "Meraviglia" – Fieracavalli 2022 – ACTION & EMOTIONS by Lorenzo

  18. Chris Button said,

    March 8, 2023 @ 8:05 am

    I think it goes without saying that history, and by extension archaeology, is useful for "historical linguistics". Genetics is then sometimes useful for archaeology, which can then indirectly benefit historical linguistics if we want to force the point. But genetics and "general linguistics" bear no connection since languages are not biological entities.

  19. Taylor, Philip said,

    March 8, 2023 @ 9:58 am

    "Stacy Westfall's Championship Run 2006 […] brought tears to my eyes" — and to mine, especially when she demonstrated the full halt from gallop. The crowd were knowledgeable, too, detecting and applauding the changes of canter lead.

    But if I may digress and speak of bringing tears to one's eyes, I attended a funeral of a good friend yesterday, at which the celebrant did three readings. The first was of The Dash, which I have now heard so many times that it has lost all effect on me, but the third was of The Old Rabbit, written by 8-year-old Lenny T. I found it unbelievably moving, and since then can neither read it nor recount it to others without again bursting into tears …

  20. James Wimberley said,

    March 8, 2023 @ 10:47 am

    You would expect domestication to have a much larger and faster impact on the genetics of the horses than of their riders. Zebras are notoriously untameable. Domestication of wild horses in Eurasia must have involved very strong selective pressure from the humans for traits of docility. Once they started riding tamed horses for war, the process would very likely have gone into reverse, as a cavalryman wants an aggressive and brave mount. IIRC camels will tolerate a rider as taxi, but will not take part in a cavalry charge. How did the Australians cope in Palestine in 1918?

  21. Taylor, Philip said,

    March 8, 2023 @ 11:10 am

    Not convinced that a cavalryman (given a choice) would look for aggressiveness (qua aggressiveness) in a potential re-mount. Braveness, yes, but also unswerving discipline and obedience, the last perhaps the most important trait of all.

  22. Gail Brownrigg said,

    March 8, 2023 @ 11:28 am

    Thank you for posting this thought-provoking paper.

    A summary of the findings as described to journalists at the media briefing on March 3, 2023 at the annual meeting of American Association for the Advancement of Science (which publishes Science) in Washington, D.C., is available here and some reactions by other archaeologists are cited here.

    The authors report five Yamnaya individuals dated to 3021 to 2501 calibrated BCE from kurgans in Romania, Bulgaria, and Hungary, displaying changes in bone morphology and distinct pathologies which can be associated with horseback riding. These were among 217 mostly “steppe” individuals from 39 sites dated to the fifth and second millennium BCE, of which some 150 are archaeologically assigned to the early Bronze Age Yamnaya culture. They recorded traits which are widely used as indicators of more than occasional horseback riding. Some skeletons showed compression of the vertebrae, which can result from time spent absorbing jarring bumps while seated – perhaps on horseback, or driving a wagon. Several also had thickened spots on the thigh bone consistent with time spent in a crouched position. Various healed injuries matched those suffered by riders today. Recognising evidence of horseback riding was not the intention of the study: finding these traits was incidental and unexpected.

    Whilst the pathologies seem compatible with an activity such as riding on a horse or other animal putting stresses on the back and limbs, one “pre-Yamnaya” skeleton from Romania and another from Hungary showing a similar set of bone changes were dated to 3331-2927 cal BCE and 4442-4223 cal BCE respectively, the latter long before the proposed Yamnaya “horse riders”. This raises the question as to whether there could have been other causes of the osteological markers interpreted as “horsemanship syndrome”. Various healed injuries matched the kind of damage seen today in riders thrown from their horses, but could also have been caused by falling off a wagon or out of a tree. A fractured foot bone might result from being run over by a wheel or trodden on by the draught cattle. Zooarchaologist William Taylor from the University of Colorado said “They’re strongly overinterpreting an interesting pattern…. In isolation, human skeletal data doesn’t have the power to distinguish horse riding from other activity patterns.”

    Riding bareback on a small horse must have been extremely uncomfortable for a man before the invention of trousers, so early rider may have been tempted to sit further back from the protruding withers (see fig. 5 in the original paper, all from the Near East during the Bronze Age, 2100-1200 BCE). In that position it is not easy to grip with the knees – I tried it for a film with a scene about Assyrian archers. Riders would have relied on balance, especially if spending a long time mounted. Gripping may be a more recent phenomenon associated with activities such as travelling at faster paces or fighting from horseback, and perhaps the use of saddles (but see Molleson 2007 [pdf]), especially p.19. The American Indians and the Mongolian nomads certainly would not ride with tight knees all day, though they do have a tendency to become bow-legged. But that is perhaps a discussion for another time?

    The question of whether an animal is “wild”, “tamed” or “domesticated” is also outside the scope of this discussion, but does need to be borne in mind when considering the implications of the conclusion of this study. The authors acknowledge that horses towards the end of the third millennium BCE would still have been very wild and hard to handle – evidence of a domesticated population with DOM2 (domesticated) lineage analysed by Librado et al was pinpointed in the Western Eurasian steppe, from where they expanded rapidly across Eurasia from about 2000 BC. We do not know if the bones found at Yamnaya sites were from wild animals possibly hunted for their meat. Did some Yamnaya individuals perhaps ride on their cattle?

    This interesting study raises a number of questions, and paves the way for further research and discussion.

    Gail Brownrigg

    P.S. J Lawrence The History and Delineation of the Horse in all his Varieties, 1809, says: “We learn from Hippocrates that the Scythians, and those nations in the habit of being much on horseback, were afflicted with inflammation, and painful tumours on their legs, occasioned by their dependent posture, and the want of support for their legs. Galen also affirms the same of the Roman soldiers.” Would this show up in pathological analysis?

  23. David Marjanović said,

    March 8, 2023 @ 3:19 pm

    Brought tears to my eyes.

    It scared me instead – it's scary to think of how many times she must have fallen off in the process of learning not to fall off. :-S

    […] “They’re strongly overinterpreting an interesting pattern…. In isolation, human skeletal data doesn’t have the power to distinguish horse riding from other activity patterns.”


  24. George said,

    March 9, 2023 @ 11:02 am

    @Taylor, Philip

    For an 8-year-old, Lenny T writes beautifully. Thanks for that.

  25. Ben Zimmer said,

    March 9, 2023 @ 2:20 pm

    Gail Brownwigg sent the following image to accompany her comment above.

  26. Pavel Kuznetsov said,

    March 22, 2023 @ 6:39 am

    The issues about the time when horse riding first appeared, confirmed by the reliable facts based on human bones, and about the characteristic features of early domestic horses are discussed in detail in 2010 in the book "Horses, Chariots and Charioteers". In chapter 1, P.A.Kosintsev gives evidence of the specific conformation of the first domestic horses. These horses were up to 130-140 cm high at the withers and had relatively slender limbs (Tables 3, 4, 10-17). Just imagine a present-day pony with the legs which are twice as thin in diameter. Riding a pony is easy for children, though it is impossible even for kids to ride a slender limbed pony. It is no mere chance that the early domestic horses were harnessed to chariots in pairs, although it would have been much easier to control one horse. In Chapter 2, A.P.Buzhilova examines the skeletal features of the Bronze Age men and the men of the Early Iron Age. She concludes that the features typical of a rider become obvious only in Early Iron Age, 10th-11th centuries BC.

    Getting back to the publication, I would venture a guess that such features of human skeletal injuries can be a result of falling down from an ox. Riding oxen is quite common in places where modern tractors have not yet become widespread.

RSS feed for comments on this post