What a prehistoric pair of pretty pants can tell us about the spread of early languages

« previous post | next post »

The following is a photograph of the world's oldest known pair of trousers:


Scientific study:

Ulrike Beck, Mayke Wagner, Xiao Li, Desmond Durkin-Meisterernst, Pavel E.Tarasov, "The invention of trousers and its likely affiliation with horseback riding and mobility: A case study of late 2nd millennium BC finds from Turfan in eastern Central Asia," Quaternary International, Volume 348 (20 October 2014), pages 224-235.



Here, we present the first report on the design and manufacturing process of trousers excavated at Yanghai cemetery (42°48′–42°49′N, 89°39′–89°40′E) near the Turfan oasis, western China. In tombs M21 and M157 fragments of woollen trousers were discovered which have been radiocarbon dated to the time interval between the 13th and the 10th century BC. Their age corresponds to the spread of mobile pastoralism in eastern Central Asia and predates the widely known Scythian finds. Using methods of fashion design, the cut of both trousers was studied in detail. The trousers were made of three independently woven pieces of fabric, one nearly rectangular for each side spanning the whole length from waistband to hemline at the ankle and one stepped cross-shaped crotch-piece which bridged the gap between the two side-pieces. The tailoring process did not involve cutting the cloth: instead the parts were shaped on the loom, and they were shaped in the correct size to fit a specific person. The yarns of the three fabrics and threads for final sewing match in color and quality, which implies that the weaver and the tailor was the same person or that both cooperated in a highly coordinated way. The design of the trousers from Yanghai with straight-fitting legs and a wide crotch-piece seems to be a predecessor of modern riding trousers. Together with horse gear and weapons as grave goods in both tombs our results specify former assumptions that the invention of bifurcated lower body garments is related to the new epoch of horseback riding, mounted warfare and greater mobility. Trousers are essential part of the tool kit with which humans improve their physical qualities.

Here are my initial reactions to this major publication, which I had intended to circulate years ago, but got caught up in too many other things:

These trousers are astonishing in many respects.  First of all, they are exquisitely designed, woven, cut, sewn, and decorated.  Second is their early date of late 2nd millennium BC, which makes them among the earliest known trousers on earth; if not, they are the very earliest archeologically attested, woven trousers.  The next oldest known trousers are the burgundy colored pair worn by Cherchen Man / Chärchän Man / Ur-David (ca. 1000-800 BC), who was discovered in the cemetery of the village of Zaghunluq near the town of Qiemo (Chärchän), southeast Tarim Basin, about one thousand km SSW from the Yanghai burial ground at the edge of the Turfan Basin.  Third is the sheer fact that these are trousers.  Trousers / pants are hard to make, because you have to cut the fabric in irregular shapes, and then you have to sew them up into a crotch, which is a complicated business.  But if you don't sew them right, that completely defeats the purpose of having a crotch after all, which is to make it easier to ride a horse.

Now, riding a horse is a revolutionary development in the history of humanity.  We've talked about it a number of times on Language Log, so I don't need to explain in detail here and now how the horse expanded the power and reach of human beings.  Why go to the trouble of cutting and sewing a crotch when it's so much easier to wrap a bolt of fabric round your waist — unless you want to straddle the back of a horse, and why go to the trouble of cutting and sewing a crotch if you make it in such a way that chafes the rider's sensitive nether parts unless you're serious about your horse riding.

The people who were most serious about and adept at riding horses were the same people who came sweeping across the Eurasian grasslands from the Pontic-Caspian Steppe and southern Urals, the Scythians and their Iranic forerunners, rounded the eastern edge of the Heavenly Moutains (Tian Shan / Tängri Tagh / Tengri Tagh / Tengir-Too), and poured down into the northeast corner of the Tarim Basin and its extension to the northeast, the Turfan Basin.  So far as we know, the earliest waves of settlers entering the Turfan and Tarim basins were Indo-Europeans, including Tocharians and Iranians, and the overall composition from the 2nd millennium BC to nearly the end of the 1st millennium BC continued to be primarily Indo-Europeans.

Because of the nature of its settlement, it is no wonder that the Turfan Basin has such a rich assemblage of languages and scripts up to medieval times, as documented by Doug Hitch in this article:

Doug Hitch, "The Special Status of Turfan," Sino-Platonic Papers, 186 (March, 2009), 1-61.

The same holds for Eastern Central Asia as a whole, as is apparent from this table:

Scripts and Languages in Pre-Islamic Central Asia

Source:  "Turfan Studies", Berlin-Brandenburg Academy of Sciences and Humanities (brochure, 2007), p. 9.

Taking into consideration the archeological, anthropological, cultural, and linguistic evidence for the early inhabitation of the Turfan and Tarim basins, it is likely that the word for "trousers" would have been Indo-European of some sort.  Below, I shall put forward some data pointing toward avenues for further research concerning the terminology for trousers and other horse-related equipment.

The MSM term for "trousers; pants" is kùzi 裤子, where the second syllable is a disyllabicizing nominative suffix.  For etymological purposes, we can focus on the first syllable.

  • Middle Sinitic: /kʰuoH/
  • Old Sinitic (Zhengzhang): /*kʰʷaːs/

One thing we have to bear in mind is that China did not have trousers until the fourth century BC.  Since domesticated horses, chariots, and wheels came from the northwest, it is hard to deny that horse riding, especially for military purposes, did as well.  Indeed, in the late 4th c. BC, King Wuling (r. 325-299) of Zhao implemented as his most important (and very famous) reform of Hú fú qíshè 胡服騎射 ("wearing Hu [style] attire [i.e., pants, belt, boots, fur caps, and fur clothes] and shooting [bows] from horseback [in battle]).  "Hu" is an umbrella term for the so-called "Five Barbarians" — Xiongnu (a Hunnic confederation), Jie, (see "An early fourth century AD historical puzzle involving a Caucasian people in North China "[1/25/19]"), Xianbei (Särbi), Di (of indeterminate ethnicity, but culturally related to the Qiang), and Qiang (Tibetic).  (Quoting myself from here)

So we wouldn't expect to find the Sinograph for kù 裤 on the oracle bone inscriptions, bronze inscriptions, seal characters, or any other early forms, and that is indeed the case (see here).

The instability of the orthographic form of kù 裤 (e.g., 袴, 跨, as given in the ca. 200 AD Shìmíng
釋名 [Explanation of Names])
 dictionary of paronomastic glosses, is an indication that it is likely to be a borrowing from some non-Sinitic language.

Axel Schuessler, ABC Etymological Dictionary of Old Chinese (Honolulu:  University of Hawai'i Press, 2007), p. 337 gives *khwâh as the Old Sinitic reconstruction of 裤 and *khwrâh as the Old Sinitic reconstruction of 跨, which he glosses as "'To step over, pass over' [Zuo]".  On p. 338, Schuessler lists a number of similarly sounding areal words meaning "forked; branching; crotch".

Fāngyán 方言 (Topolects), the dictionary of regionalisms by Yang Xiong (53 BC–18 AD), has an interesting note on dà kù 大袴 ("big trousers"), for which he cites the regional word dǎodùn 倒頓.

  • Middle Sinitic: /tɑuX//tuənH/
  • Old Sinitic (Baxter–Sagart): /*tˤawʔ/
    (Zhengzhang): /*taːwʔ//*tuːns/
    (sources:  herehere, and here)

Even if Sinitic borrowed the word for "trousers" from some neighboring people, it is still probable that the latter picked it up from IE speakers who invented this horse-related item of clothing.  So we need systematically to go through the terminology for "trousers, pants" in IE.

Here's a beginning, kindly submitted by Hiroshi Kumamoto:

It may not be the oldest, but Khot. kaumadai "trousers" comes to mind. Bailey has an article with the title "vāsta" on the words of clothing in Khotanese in Acta Orientalia 30 (= Iranian Studies presented to Kaj Barr), 1966, 25-43, and the word is discussed on p. 26. Although in this article too many words are given the generic meaning of "clothing" and the like, the "trousers" word seems to be a good one.

To tell the truth, what caught my eye about the fancy pants pictured at the beginning of the post above all is the pattern on the band around the knees.  It would be an understatement to say that I was absolutely stunned when I saw it.  The reason I was captivated by that interlocking, angular pattern on the knee bands is because it is so intricate and distinctive and because for decades I have been noticing it on a variety of artifacts from the first millennium BC.  Here are two:

Western Zhou bronze from Royal Ontario Museum.

Zaghunluq, fragment for a sleeve or pant-leg; long-hop twill, slit tapestry; courtesy of Zhang He, who has collected dozens of examples from various locations and will eventually write a paper on this topic.  Many of these instances of the interlocking pattern from the southeastern and northeastern rim of the Tarim Basin are associated with figures from the first half of the first millennium BC who are smothered in cannabis and have harps, jingling bells on their legs, and other attributes of Scythian shamans.

Westward from Zaghunluq (Cherchen) along the southern rim of the Tarim Basin, we find many instances of this interlocking pattern on textiles at Sampul, with its enormous cemetery that was active from around 217 BC to 283 AD.  Sampul also yielded the spectacular 1st-century AD tapestry in Hellenistic style.

We also find this pattern on textiles, weapons, sculpture from Sanxingdui (in recent weeks, sensational new discoveries at this 12th-11th c. BC Sichuan site have been announced), artifacts from Pazyryk (6th-3rd c. BC site in the Altay region), and so on (more locations mentioned below).

John Huntington observes that the design is very similar to the motif on the platform of lady Dai in the Mawangdui (early 2nd c. BC site in Changsha, Hunan) funeral banner (see one third of the way down the page).  There is also a band of a very similar design around the lid of the "red" coffin at Mawangdui. The strategic placement of the angular interlocking pattern on Lady Dai's banner and red coffin attests to its numinous power.

The bronze specialist, Robert Bagley, refers to this pattern at "interlocked T's".  That comports with my supposition that it may be related to the famous TLV pattern on bronze mirrors of the Han Dynasty (2nd c. BC-2nd c. AD).

It has been my hypothesis all along that this intricate pattern was first developed in textiles and that it was subsequently adopted for application on other materials.  The prehistoric textile specialist, Elizabeth J. W. Barber, supports this thesis:

It certainly looks like a weaving pattern: just the kind of diagonal lines that are easy to weave.   This has 2 colors alternating (dark brown and off-white?), such that neighboring scrolls in opposite colors interlock: so it must have been done in a double-weave (what's white on the top is brown on the other side and vice versa: Yingpan Man's outfit was double-weave in red and yellow).  Whether it is also long-hop twill, I can't see — not sure that it needs to be.  It could be simple plain-weave (tabby) binding in two layers (double-weave).   But it does seem to be a development of the same family of patterns we saw at Zaghunluq and that I saw from Pazyryk, 500 miles north, same general era.  The Zaghunluk patterns we saw in 1995 and the Pazyryk ones, however, are ROUNDED interlocking spirals, rather than angular.  So I'm really interested in the new (? 2013?) Zaghunluk piece you show here: the curves have become angles– MUCH easier to weave, in many techniques.  They were already headed towards angles on the turquoise shirt and pants; but this piece is much more so, much closer to the Yanghai pants.

[What to call this pattern]

The rounded ones (as at Pazyryk, and the ones we saw in 1995 from Xinjiang) are usually called scroll patterns:  in this case, interlocking scrolls.  "Scroll" implies rounded lines; "interlock" tells you something about how they are laid out with respect to each other.

I think the best term for the one on the pants and on the W. Zhou vessel is fret pattern — that implies made from straight lines with sharp angles, including right angles.  So here, interlocking fretwork.

Note that the pattern on the pants is FAR more sophisticated in its interlock than the pattern on the vessel, which is actually rather simple.  (If puzzled by that, note that the fretwork curls in different directions as you move from one bit of the pattern to the next–or from one color to the next.  Not so for the bronze, or the Zaghunluk piece.)   It's a real tour de force!


We find the interlocking fretwork pattern on textiles, bronzes, and other types of artifacts in a polygon marked roughly by Sampul in the southwest Tarim Basin, Cherchen in the southeast Tarim Basin, the Turfan Basin, the Chengdu Plain of Sichuan, the area around Changsha (Hunan), the Central Plains of the Yellow River Valley, and the Altay region.  This space within East Central Asia and East Asia is essentially coterminous with the range of expansion of the Scythians and their predecessors and congeners in that part of Eurasia.

Update (6/16/21), documentary:

"The Invention of the Trousers"

I was very happy to see my friend, Lü Enguo (Xinjiang Institute of Archeology), chief archeologist at the site, and my colleague, Mayke Wagner (Deutsches Archäologisches Institut), lead scientist on the project to study the trousers, in this documentary.


Selected readings


  1. Chris Button said,

    April 3, 2021 @ 1:14 pm

    If we can reconstruct 褲 via 袴 back to Old Chinese then I would reconstruct it with a uvular onset. That would mean the rounding (where Schuessler has -w-) would have emerged later, and the source of the loan would not need to have a labial component. But that does depend on the time depth.

    How far does "hose" (as in in Lederhosen) go back with the sense of trousers? Does that sense start in Germanic? The original Proto-Indo-European *(s)keu- source means "cover".

  2. David Marjanović said,

    April 3, 2021 @ 1:50 pm

    That's limited to Germanic, if not less than that. "Breeches" is more widespread because it's a loanword from Celtic in Germanic and Latin.

  3. Victor Mair said,

    April 3, 2021 @ 2:20 pm

    From Juha Janhunen:

    Without trying to take any etymological shortcut, which in any case is likely to be wrong, I come to think of the similarity in form between OChinese khwaas and German Hose etc. (borrowed also to Finnish as PL housu-t 'trousers').

  4. Victor Mair said,

    April 3, 2021 @ 2:34 pm

    At first I thought the band around the knees ((just above the picture of the bronze pot) might have been embroidered, but Elizabeth Barber explains that that is not the case:

    1) Embroidery is done with a needle and thread AFTER the cloth is woven. These knee-strips, however, are MASTERFUL examples of creating the pattern WHILE weaving — that is, the warp and weft (both) which form the cloth itself also form the pattern — a much more difficult thing to do, as it takes very precise and careful planning and attention with every pass of the weft. Embroidery is easy as rolling off a log, by contrast. (As I said to you before, those knee-strips, with that complex pattern, were a real tour-de-force to create!!!) (Technical terms: "inwoven pattern" vs "embroidered pattern")

    2) "Zaghunluq, fragment for a sleeve or pant-leg; long-hop twill, slit tapestry" (just below color picture of fragment)

    It does appear to be long-hop twill (note, however, that this is a term Irene and I made up, for lack of ever having heard of any other, and we needed to call it SOMEthing!). But I see no evidence that it is slit tapestry, although it is indeed some type of tapestry. I'd leave it at that. The Egyptians waffled for some time between slit and dovetail tapestry, for forming color-edges that run parallel to the warp; but they also did as here, and simply DODGED the problem by making designs with no color borders that had to run parallel to the warp! Such tapestry is therefore neither slit nor dovetail–just, well, just tapestry.

    Otherwise… WOW. REALLY interesting! I'm still just astonished at the intricacy of the "fretwork interlock", and I'll keep an eye out for it farther west, especially among Iranian herders (of whose early apparel I have learned a great deal more, in my current quest!)

  5. David Marjanović said,

    April 3, 2021 @ 4:26 pm

    How far does "hose" (as in in Lederhosen) go back with the sense of trousers? Does that sense start in Germanic? The original Proto-Indo-European *(s)keu- source means "cover".

    Wiktionary says it starts in Germanic, and it doesn't know a root *(s)kew-, but it does know a root (s)kewH- with an unidentifiable "laryngeal" that is required by the various vowel lengths and also by the pitch accent of the Lithuanian descendant.

    Lithuanian also shows that the plosive must really have been *k, which should be preserved unchanged even in an Iranian descendant if one ever existed. If we can somehow get the zero-grade *kwH- (i.e. *[kuH]-, most likely!) to metathesize to **kHw-, then that should give Proto-Iranian *xw-, and that might explain the aspiration on the Chinese side… but that's a rather long chain of inference.

  6. Michael Broughton said,

    April 4, 2021 @ 12:38 pm

    Some observations on the Chinese 袴 and 绔

    In Qian Yi’s (钱绎) 方言笺疏, he makes an interesting note when comparing the characters 袴 and 绔: 中分之名,兩脛之衣謂之袴,猶兩足所越謂之跨,兩股之間謂之胯,剖物使分亦謂之刳,義並同也。Rough translation (The name for something divided up the middle. 袴 is clothing that covers both calves, just like 跨 is two feet stepping across, 胯 is the place between two thighs and 刳 is the dissection of an object into halves).

    This idea of something “split up the middle” seems well connected to Schuessler’s etymology for 绔 and 裤, ‘be forked, branch…crotch, branching’. I wonder if the 衩 in 裤衩 is trying to get at the same idea.

    Another interesting point is that in his 同源字典,Wang Li connects 跨 and 胯 with 骑. This seems to make a lot of sense (stride and straddle come from the same root in English). Regardless of whether one is 骑ing a horse or a bike, the nature of 骑 is, it seems, quite similar to what Qian Yi was talking about with the ‘divide up the middle’.


    P.S.: [Happy] Easter and Qingming. (I feel like the connection between these two festivals is more than just a date. Easter < PIE aus "to shine" vs. míng 明 ["bright"].) https://www.etymonline.com/search?q=easter

    Qingming, lit., "clear-bright"; "tomb-sweeping [day / festival]" https://en.wikipedia.org/wiki/Qingming_Festival

  7. Brian Morehouse said,

    April 4, 2021 @ 1:27 pm

    A brief comment: Elizbeth has laid out the likely weaving techniques that would account for the design under discussion. Importantly for one who is interested in visual languages my interest lies, and as she has rightly pointed out, in the angularity of the interlocking “latch hooks” as apposed to a rolling device that is often used as a similar decorative feature. The later often seen as features on classical Greco-Roman tile work; however, there is a third variant which in “carpet parlance is referred to as a “running dog border”. One might think of that form as lying between both the angular and rolling forms. One early example (3rd-4th CE) can be seen in “Pre-Islamic Carpets and Textiles from Eastern Lands”, pg.32 (one can also clearly see the rolling feature in the same publication, pg.57). This “running dog style” has become a prominent design feature especially in Caucasian carpets, whereas the more angular latch hook device indicated on the bronze and certain textiles is commonly seen in Anatolia, both in the designs found on carpets and flatweaves. It is clear that these features come about not only through a preference in design and long held traditions but also as a result of preferences and adaptations to weaving techniques. The angularity of this device appears to be more conducive to weavings using both symmetrical and asymmetrical knotting techniques, but this is not necessarily the case on early rugs woven on a single warp. The rolling form appears more suitable for such things as felts, etc. Given some time and focus one can follow the use of these features both within the repertoire of the design vocabulary of material cultures spread throughout Asia and also their likely entry into that language pool.

  8. ulr said,

    April 4, 2021 @ 2:34 pm

    According to the German etymological and historical dictionaries (Kluge, Pfeifer, Paul) Hose acquired its modern meaning "trousers" only in the 16th century, before that it referred to some kind of stocking ("eine Art weit hinaufgehender Strumpf", according to Paul).

  9. Victor Mair said,

    April 4, 2021 @ 7:23 pm

    From Chau Wu:

    Was the pair of trousers made in Hallstatt?

    Did the archaeologists find the manufacturer's label of the trousers indicating they were made, not in China, but in Hallstatt, or Halle, Galatia, Galicia, Bohemia, the British Isles, or Gaul?

    The pattern on the trousers look quite similar to the so-called "key pattern" (or variations thereof) of Celtic art. Check out the pattern shown in the Websites below:

    (Scroll down to the bottom.)

    (The one below Sutton Hoo.)

    Celtic Pattern – Key Pattern from Aberlady, Haddingtonshire


    It's such a coincidence! Today you show that pair of trousers showing the Celtic-like pattern. Just yesterday, I found an Old Celtic word that could potentially be the precursor of the Old Sinitic word for 'wheat' *mrǝk (> 麥). Because, so far, most of the comparanda I have found corresponding to Sinitic words are from Germanic, Greek, Latin, and Vulgar Latin, and only a few can be related to Celtic, I hesitated to take this one in. But then I remember Elizabeth Barber's excellent book, The Mummies of Urumchi, which I read two decades ago, in which she mentions textile similarity between the Tarim fabrics and Celtic tartan.

  10. Victor Mair said,

    April 4, 2021 @ 8:40 pm

    From Tony Phillips and Jack Morava come these images of Peruvian textiles.

  11. Rodger C said,

    April 5, 2021 @ 6:51 am

    Hose acquired its modern meaning "trousers" only in the 16th century, before that it referred to some kind of stocking

    Which is still its meaning in English, of course.

  12. Chris Button said,

    April 5, 2021 @ 9:14 am

    Khot. kaumadai "trousers" comes to mind.

    Yutaka Yoshida also notes Skt. kaupīna.

  13. gds555 said,

    April 5, 2021 @ 9:47 am

    The Indo-Europeans’ cleverness at sewing trousers makes it all the more remarkable that in the long run, with mass conversions to Christianity and Islam and whatnot, they weren’t able to keep their pantheon.

  14. Chris Button said,

    April 5, 2021 @ 10:43 pm

    Another interesting point is that in his 同源字典,Wang Li connects 跨 and 胯 with 骑. This seems to make a lot of sense (stride and straddle come from the same root in English)

    The phonology is off though. I'd reconstruct 騎 as *gàl (or more specifically *ᵑgàl).


  15. Tim Caldwell said,

    April 6, 2021 @ 1:02 pm

    Stylistically the design on the trousers looks reminiscent of designs on Andronovo ceramics:

  16. David Noble said,

    April 7, 2021 @ 7:25 pm

    As I'm also watching internet sites on Blackadder series, I have to note that these trousers would be high fashion for Baldrick

  17. Wastrel said,

    April 9, 2021 @ 5:04 am

    To a layman, these angular interlocking spirals look an awful lot like the usual Greco-Roman angular interlocking spirals that they obsessively decorated everything with (the meander). (simpler meanders can also take the 'interlocking T' form). It's not identical – you're talking about planes, and the meander is a strip, and some of your designs link the spirals 'across' the strips rather than 'along' them as in Greece – but one could easily be formed from the other. If the meander's spirals are not angular, they become the Vitruvian wave; if you join the curved spirals in threes instead of in lines, they become the Neolithic (Cardial?) triskelion, coincidentally reused by the Celts (so Newgrange, despite far pre-dating the Celts, has this 'Celtic' pattern on it).

    And of course if you cross the lines of a meander (or to put it another way, make the tips of the interlocking angular spirals touch), you invent the swastika, which is why it's on so many Roman artifacts.

    Apparently the meander originates from the ancient middle east, and has indeed been theorised to have been imported to China during the Han dynasty, but it's entirely possible it's convergent evolution (see Peru).

    In the end, I think the point is that there's only so many simple repeating patterns you can use to fill spaces, and meanders (etc) do a good job of being interesting enough to please while being simple enough to easily reproduce.

  18. Chris Button said,

    April 13, 2021 @ 8:06 am

    Further to my earlier post about a uvular onset with secondary rounding, if 袴 can be better reconstructed as *qʰáɣs, then I wonder if there might be any Iranian connection here:


    The association of shoes with trousers seems to find support cross linguistically.

    Alternatively, we are simply dealing with late internal coinage from the words for "stride" etc noted above. The broader word family connections of the phonetic 于 do have a sense of "go" after all.

RSS feed for comments on this post