He Wrote One Paragraph

SERIES · Herbal Histories. Herbal Histories takes one claim about a drug plant — the kind that circulates in a hundred articles without a citation — and goes to the document it is supposed to rest on. Twenty-five issues, each built on primary sources read in the original: statutes, papyri, newspaper columns, parliamentary papers and manuscripts in Dutch, German, Spanish, Portuguese, Latin, Nahuatl, Sanskrit, Arabic, Chinese and Egyptian. Translations are the author's own and are marked as such. When a claim survives the document the issue says so; when it does not, the failure is the story.

Somebody burned their hand.

That is the first thing to say about the object, because it is the only thing in the whole find that is unmistakably a person doing something rather than a thing lying where it fell. In the western half of the burial chamber of the second great barrow at Pazyryk, in the eastern Altai, under a permanent lens of ice, Sergei Rudenko's team lifted out a small copper cauldron on a pedestal — fourteen point eight centimetres high, six low knobs around the rim, handles out of all proportion to the body. It was full of stones. And its handle had been wound around with strips of birch bark.

Rudenko wrote down the obvious inference: the stones inside had been hot enough that the handle could not be held in bare hands.

Six sticks stood splayed above it, partly broken and knocked over by looters, with a large decorated leather hanging spread across them; a flask of thin leather, with a griffin sinking its claws into a black grouse worked on one face, was tied by a thong to one of the legs. In the south-western corner of the same chamber, a second copper vessel — rectangular, shallow, standing on four ringed legs — also full of stones, also under six rods. Those rods were a hundred and twenty-two and a half centimetres long, about two centimetres thick, wound spirally with birch bark and pierced two centimetres from the top so that a thong could gather them at the apex.

In each vessel, along with the stones, a small quantity of seeds. Some of them charred. More of the same seeds in the griffin flask.

The barrow had been robbed in antiquity. Everything metal the thieves could reach was gone. These two vessels survived because they had frozen into the ice.

That is the find. It is extraordinary, and almost everything said about it since was settled in advance by a paragraph written some two thousand three hundred miles to the west, at least a century before this barrow was built, by a Greek who had never been to the Altai and did not know it was there.

Book IV of the Histories is Herodotus's account of Scythia, written in the 420s BCE. Chapters 71 and 72 describe the burial of a Scythian king: the embalmed body carried by wagon from people to people, the mourners cutting their ears and arms and driving arrows through their left hands, the strangled concubine and cupbearer and cook and groom and messenger laid in the tomb, the fifty strangled young men mounted on fifty strangled horses in a ring around the mound a year later.

Chapter 73 turns to everyone else. Ordinary Scythians are carried around among their friends in wagons for forty days and then buried. And then this:

Having buried them, the Scythians cleanse themselves in the following way. They wash their heads. Then, for the body, they set up three sticks leaning against each other, stretch woollen felts around them, pack the felts as tight as they can, and throw stones red-hot from the fire into a σκάφη standing in the middle of the sticks and the felts.

Chapter 74 is a botanical digression. Cannabis grows in their country, very like flax except that cannabis is far the thicker and taller. It grows wild and it grows sown. The Thracians make garments of it so like linen that nobody who has not handled a lot of it could tell the difference.

Chapter 75:

Taking the seed of this cannabis, they creep under the felts and throw the seed onto the stones glowing from the fire. Thrown on, it smokes, and gives off so much vapour that no Greek vapour-bath could better it. The Scythians, delighted at the vapour-bath, howl. This is instead of washing for them, since they never wash their bodies with water at all.

Then he moves straight on to what the women do with cypress and cedar and frankincense ground on a stone, and how their skin comes up clean and shining when they scrape the paste off the next day. Then to chapter 76, and the Scythians' loathing of foreign customs.

Four things about that passage are worth holding.

It is about the living. The vapour tent is not in the grave and it is not for the dead. It comes after the burial, and the people inside it are mourners, cleaning themselves.

The vessel is a σκάφη. The word means a hollowed thing — a trough, a tub, a basin, a bowl. It is not a censer, not a cauldron, and not a brazier, and Herodotus does not say what it is made of.

The material is the seed, τὸ σπέρμα. He says it twice in one sentence.

And the last clause is the one everybody quotes. οἱ δὲ Σκύθαι ἀγάμενοι τῇ πυρίῃ ὠρύονται. The delight is real: ἀγάμενοι is a participle of ἄγαμαι, to marvel at, to be delighted with, and it takes τῇ πυρίῃ — they are delighted at the steam. But ὠρύονται means howl. It is what wolves and dogs do. It carries no joy of its own, and Herodotus uses it exactly once in nine books. So does πυρίη, the vapour-bath. Both of the load-bearing words in the sentence are, in the Histories, unique.

Herodotus does not say anyone was intoxicated. He has the vocabulary for it, and he uses it elsewhere. In Book I, describing islanders in the Araxes on hearsay, he writes that they sit in a circle around a fire, throw a certain fruit on it, and smelling the fruit as it burns, μεθύσκεσθαι τῇ ὀσμῇ — get drunk on the smell, just as Greeks do on wine — and that the more fruit goes on the drunker they get, until they stand up and dance and start singing. That is what intoxication looks like when Herodotus means to describe it. It is a different people, a different plant, an open fire, and a report he marks as second-hand.

At IV.75, in his own voice, about a purification, he says they are pleased with the steam and that they howl.

The Scythians of Book IV live on the steppe north of the Black Sea, between the Danube and the Don, with the Greek trading city of Olbia at the mouth of the Bug on their southern edge. Herodotus places their royal graves in a district he calls Gerrhos, as far up the Dnieper as the river is navigable. Everything east of the Don he assigns to other peoples with other names.

He was working from a Greek colonial city looking inland. He names his informant when he has one — at IV.76 he tells us that what he knows about the sage Anacharsis he heard from Tymnes, the deputy of the Scythian king Ariapeithes. For the felt tent and the seed and the howling, he names nobody.

"Scythian" was in any case a Greek word for other people. It covered a scatter of mounted, mobile, kurgan-building societies across several thousand kilometres of grassland who spoke related Iranian languages, buried their dead under mounds, and had in common chiefly that they were not Greeks and were not Persians. Archaeologists working in Inner Asia use "Scythian" as a shorthand for a horizon of material culture, or avoid it, or write it in quotation marks. The authors of the 2019 Pamirs study, when they had to name the thing, called it "the mythical 'Scythian' warrior nomads." They meant the phrase to do work as an abstraction and not as an ethnicity, and they were right to flinch.

It did not come from the Greek, and it did not come out of the ground.

George Rawlinson's English Herodotus, the version that formed the reading of the passage in the English-speaking world, appeared in London in 1858–60 and in New York in 1860. He translated the last clause as "the Scyths, delighted, shout for joy." Then he footnoted his own translation:

The evidence for the reading is a nineteenth-century traveller's account of Russia and a nineteenth-century monograph on Scythian geography. Rawlinson's method throughout the page is the same: two footnotes earlier he explains the three-pole tent by reference to Kalmyk and Kyrgyz and Mongol dwellings as observed in the 1780s and the 1840s, and to a Kurdish shelter Pallas saw on the plain of Moghan. Modern nomads are read backwards onto ancient ones, as though the steppe were a place where nothing happens.

Half a century later, How and Wells's Commentary on Herodotus — the standard English commentary, published by the Clarendon Press in 1912 and reprinted by Oxford for most of a century afterwards — tidied the difficulty away:

Rawlinson had at least noticed that the text does not say what he wanted it to say. How and Wells removed the problem by asserting that it does, and did it by importing the Araxes passage — the fruit, the open fire, the dancing — as though it explained a different chapter about a different people in a different book. The cross-reference runs in exactly the wrong direction, and it has been running that way ever since.

There is a further sentence in the same volume that ought to be quoted whenever this commentary is cited. Debating whether the Scythians were "Mongolian" or "Aryan" in origin, How and Wells list under Customs, as evidence for the Mongolian theory, "their use of hemp as an intoxicant" — alongside their worship of a sword and their impaling of horses. To their credit they add that customs may be borrowed and that the evidence is not decisive. But the reading of IV.75 had by then become a piece of racial classification, with the drug as one of the markers, and it is worth knowing that before quoting the note.

By 1912, then, the following had been established without any archaeology at all: that the Scythians got high, that they did it in a tent, that this was characteristic of them as a race, and that Herodotus had been slightly confused about it.

Everything dug up afterwards arrived into that frame.

Rudenko had trial-trenched the Pazyryk group in 1924, and the first of the barrows was opened in 1929. He came back after the war. Barrow 2 was excavated over the seasons of 1947 and 1948, under the Institute of the History of Material Culture, with the Hermitage joining from 1948; everything went to Leningrad, where it remains.

What made Pazyryk what it is was the ice. Water had got into the robbed chambers and frozen permanently, and inside that lens the organic material survived: felt, leather, wood, textiles, horses, tattooed skin. It is one of the great preservation accidents in the archaeology of Eurasia, and it is the reason a copper vessel full of stones is still a copper vessel full of stones.

Rudenko's report — Moscow, 1953; in English as Frozen Tombs of Siberia, translated by M. W. Thompson, in 1970 — sets out the barrow 2 material in a section on purifying rituals. He quotes Herodotus IV.73–75 at length, in his own rendering, and then writes:

The full set of articles. It is a striking phrase, and a revealing one: the objects are not being described, they are being checked off.

Attached to Rudenko's quotation of Herodotus is a footnote that is almost never quoted:

The excavator has gone back to the Greek and adjusted the standard Russian versions — Mishchenko's and Latyshev's, the two an educated Russian reader of the period would have had — in the very sentence he is about to use to identify what he has dug up. He is not wrong about πυρία. He is right that it means a sweating-room. But the effect of the correction is to move the Greek away from a bathtub and toward a heated chamber, which is what he has in the ice.

And then the objects do not fit anyway.

Herodotus's frame has three poles. Rudenko's has six, twice. Herodotus's covering is woollen felt; the one intact covering in barrow 2 is decorated leather, and Rudenko allows that the other was "probably" felt on the strength of large pieces found lying in the same corner. Herodotus's rite happens after the burial, above ground, in the open air, among mourners who are going home afterwards. Rudenko's censers were inside a sealed chamber with the dead. Herodotus does not put a vapour tent in a grave, and Rudenko does not remark that his are in one.

The seeds themselves were identified as Cannabis sativa L. of the variety ruderalis Janisch. — the wild or roadside form. That identification is morphological, made by eye and by shape. No chemical analysis of what was actually burned in those two censers appears to have been published, then or since. There is also a wrinkle about the famous "hempen clothing" of Pazyryk that tends to get repeated: Rudenko's own microscopy of the plant-fibre cloth from barrow 2 concluded that "there are no grounds for considering the fibres of the cloth under discussion as belonging to one of these plants" — flax, hemp or Chinese nettle. The other fibre he found in the barrows is kendyr, which is Apocynum, dogbane, a plant sometimes called Turkestan hemp and related to cannabis not at all.

The dating has moved too. Rudenko placed the Pazyryk barrows in the fifth and fourth centuries BCE. Larch tree-ring sequences from the tomb timbers, wiggle-matched against the radiocarbon calibration curve, have since pushed them later; Irka Hajdas and colleagues, revisiting Pazyryk 2 and Ulandryk 4 in 2020 with cosmic-ray spikes as synchronisation points, concluded that the culture existed no earlier than the third century BCE.

Herodotus died around 425. Barrow 2 was built, on current evidence, at least a century and a half after that, about 3,800 kilometres east of the country he was describing.

None of which stopped the identification from hardening. Thompson's translator's preface to the 1970 English edition lists "the censers for burning hemp, the shaven heads, the scalping, the joint burial of man and woman and so on" as "further vindication of Herodotus' authority." Rudenko himself, in an earlier chapter, went considerably further than his evidence: smoking hemp, he wrote, "like smoking hashish, took place without a doubt not just as a ceremony of purification after burial but in ordinary life; hashish was used as a narcotic." Without a doubt. The doubt was the only thing the material actually supplied.

The Jirzankal cemetery — also recorded as Quman — sits on the west bank of the Tashkurgan river near the village of Qushiman, in the Tashkurgan Tajik Autonomous County of Xinjiang, at 37°50′54″N and 75°12′11″E, between 3,061 and 3,080 metres above sea level. It was excavated in two seasons, 2013 and 2014, by the Xinjiang Archaeological Team of the Institute of Archaeology of the Chinese Academy of Social Sciences, under Wu Xinhua. The cemetery divides into four zones across two river terraces and a yardang platform; zone B alone holds thirty-four tombs. Radiocarbon on wood, with the artefact typology, puts its use between about 2,400 and 2,600 years ago.

The surface there looks nothing like Pazyryk. Long strips of black and white pebbles run across it in rows, which the excavators say have no parallel in the region, among circular mounds ringed with one or two courses of stone. The dead were buried in shaft chambers, some of them after their bodies had been left out in the air. The grave goods included pottery, ironware, glass beads from western Asia, silk from eastern China, angular harps, and wool cloth. There was no hemp textile anywhere on the site.

The excavators' own reading of the place has nothing to do with Herodotus. Writing in 2015, Hong Shen, Xinhua Wu, Zihua Tang and colleagues described the population as Saka, and argued from the small rounded fire-altars of scented juniper and from the secondary burials after exposure that "the influence of Zoroastrianism may have reached the Tashkurgan region." The strontium isotope work on the same skeletons, published the following year, found ten of thirty-four sampled individuals to be non-local — a highly mobile population on a pass between the Tarim, the Iranian world and the Indian subcontinent, several centuries before there was any such thing as a Silk Road.

Then, on 12 June 2019, Meng Ren, Zihua Tang, Xinhua Wu, Robert Spengler, Hongen Jiang, Yimin Yang and Nicole Boivin published, in Science Advances, the chemistry.

Ten wooden braziers, from eight tombs. Blocks of wood hollowed and burnt through the centre, some with the rounded river stones still sitting in the cavity. Of seventy wooden objects analysed from the site, only four were juniper, and three of those four were braziers — a choice, since juniper gives off a resinous perfume when it burns and is still used for that reason in Tibetan ceremony today.

Meng Ren took roughly twenty milligrams of wood from the internal charred surface and the external surface of each of ten brazier fragments, and material from four of the stones, ground it, extracted it twice in chloroform and methanol under sonication, and ran it on gas chromatography–mass spectrometry. Cannabinol appeared on nine of the ten inner surfaces, and on two of the four stones. It appeared on none of the outer surfaces, which is the control that makes the result mean something: the signal is on the side that faced the fire.

Two absences do the rest of the interpretive work. There were no cannabis seeds in the braziers, and there was no cannabidiol in the residue. Cannabinol is what tetrahydrocannabinol turns into when it is exposed to air, light or heat; wild cannabis carries roughly comparable amounts of THC and CBD, so a residue with strong cannabinol peaks and no trace of CBD or its breakdown products points to plants that were making more THC than a wild plant usually does. Whether that is because people were selecting for it, or hybridising populations that had been genetically isolated by mountains, or because cold and altitude and ultraviolet light push the plant into producing more of the compound on its own, the paper leaves open. The authors are careful about it. They are careful about most things.

Including Herodotus. The paper's own discussion says this:

That is the correct reading, made by the people who did the work, printed in the paper on the day the press cycle turned it into the Scythians of Herodotus getting high in a tent. The Max Planck press release that went out with it made the same point about the state of the evidence, noting that arguments placing the origins of cannabis smoking on the Central Asian steppe "rely solely on a passage from a single ancient text."

And on the seeds, the paper is sharper still. Noting that THC content falls as you go from bract to flower to leaf to stem to root to seed, it observes that "the lack of seeds in the burners may suggest that nonfloral plant parts were burned, or it may suggest that seeds were removed from the floral structures because they do not contain the desired secondary compounds."

Read that against IV.75 and the whole edifice tilts. Herodotus says σπέρμα, twice. The seed is the part of the plant with the least in it. The 2019 study, which is routinely described as confirming him, found no seeds at all, and explains their absence by saying that seeds are exactly what you would take out.

One small thing more. Describing the Altai comparison, the paper places the Pazyryk frozen tombs "in the southern Altai Mountains of the Tuva Republic, Russia." They are not in Tuva. They are in the Great Ulagan valley in the Ulagan district of the Altai Republic, about four hundred and fifty kilometres away, as Rudenko's own site description makes plain. It is a small slip in an otherwise meticulous paper, and it is the kind of slip that happens when the sites in a story have stopped being places and become instances of a category.

Set the paragraph aside for a moment and look at what the two excavations have in common on their own terms, without the Greek.

Fire in the grave. At Jirzankal the excavators found burn marks on a great many objects, cremation deposits are known across Central Asia from the late third millennium, and the small juniper vessels look to them like something other than heating equipment. At Pazyryk, two vessels of heated stones stood under frames in a sealed chamber. In both places somebody carried fire into a burial and left it there.

Aromatics. The Jirzankal braziers were disproportionately made of juniper, which is chosen for its smoke. In barrow 2 the hemp seed was not alone: a leather pouch in the same chamber held coriander and yellow sweet clover with it, and Rudenko wondered whether the clover was burned too. Herodotus's Scythian women, in the sentence immediately after the tent, grind cypress and cedar and frankincense into a paste for their skin, and in IV.71 the embalmed king is packed with cut galingale, incense, parsley seed and anise.

Set out like that, what runs through all of it is a technology of smell — applied to the dead, and to the people who have handled them, from the Black Sea to the Pamirs. Which is roughly what one would expect of a practice moving along the same routes as the glass beads and the silk that turned up in the Jirzankal graves.

That would be a decent story on its own, and it is not the one that gets told, because the paragraph is in the way.

The paragraph is in the way because of what it offers. It has a scene in it, and figures, and a noise. It supplies what an excavation can never supply on its own: somebody's face. And it is very short, which means it can be reprinted whole, in a caption, on a museum label, in the second paragraph of a news story, and reprinted whole it looks less like an interpretation than like a photograph.

Under it lie a great many people about whom the record says something more specific and much less quotable. Ten of the thirty-four individuals sampled at Jirzankal grew up somewhere else. Only one of the ten tombs holding those incomers had a brazier in it, which suggests, gently, that burning whatever it was in a wooden bowl was a local thing to do. Some of the skulls have perforations and some of the bones have cuts, and the excavators think human sacrifice; the 2019 paper says plainly that this needs verifying before anyone builds on it. There were harps in those graves. Somebody at that altitude, in that cold, played music at funerals.

At Pazyryk there was a man and a woman in the chamber, and two sets of vessels, and Rudenko drew from that pairing the inference that both men and women used them.

And there is the handle wrapped in birch bark.

There is no photograph of the Pazyryk censers that can be published here. They have been in the Hermitage since 1948; the pictures that exist were made for Rudenko's report and belong to it and to the museum, and the museum does not release them. So the object at the centre of half of this story cannot be shown, and the alternative — a reconstruction, a drawing, a photograph of some other cauldron from some other steppe grave — would be precisely the substitution the article is about. The gap stays.

What survives instead is Rudenko's description of it. A small copper vessel on a pedestal. Six knobs on the rim. Handles out of proportion to the body. Full of stones that had been in a fire. And around the handle, strips of birch bark, wound on by somebody who had picked it up once already.

Herodotus never saw it, and it has nothing to do with him. He was writing about a country some two thousand three hundred miles west, more than a century before that vessel went into the ground, where the mourners came out from under the felts and went home clean. τοῦτό σφι ἀντὶ λουτροῦ ἐστι. This is instead of washing for them.

He wrote one paragraph. Everything else has been us.

Herodotus. The Greek quoted here is the Perseus Digital Library text, which follows Hude's Oxford Classical Text; in the clauses at issue it is identical to the Greek printed opposite A. D. Godley's translation in the Loeb Classical Library edition of 1921, which I collated directly against it. The English renderings above are mine and are given as glosses on the Greek, not as a published translation. Where a published translation is quoted it is named in the text or the caption: Rawlinson (New York, 1860) for "the Scyths, delighted, shout for joy"; G. C. Macaulay (London, 1890) for "howl like wolves"; Godley (Loeb, 1921) for "howl in joy for the vapour-bath."

The σκάφη of IV.73 is worth watching across those three. Rawlinson makes it "a dish… placed upon the ground." Macaulay makes it "a basin placed in the middle." Godley, with σκάφην printed on the facing page, makes it a pit: "they make a pit in the centre beneath the poles and the rugs and throw red-hot stones into it." Rudenko's Russian, in Thompson's English, makes it "a vessel set within this shelter." One Greek noun; four different objects; and the shape of the object is what every archaeological identification since has turned on.

That ὠρύονται and πυρίη each occur once in the Histories was checked directly against the Greek text of all nine books. It means there is no internal comparandum for either word, and readings of the passage that lean on their connotation are leaning on Greek usage outside Herodotus.

Pazyryk. All statements about barrow 2 are from Rudenko, Frozen Tombs of Siberia: The Pazyryk Burials of Iron Age Horsemen, translated by M. W. Thompson (Berkeley and Los Angeles: University of California Press, 1970), from the Russian of 1953 — chiefly the section on purifying rituals at pp. 284–285, the household-goods description of the two censers, the finds list for barrow 2, and the fibre analysis. The translator's preface is Thompson's, not Rudenko's, and is identified as such. Rudenko's footnote correcting Mishchenko and Latyshev is at p. 284; I have not been able to consult those Russian translations directly, so what is said here is that Rudenko reports departing from them, and why he says he did.

The revised chronology is from I. Hajdas and colleagues, whose 2004 chapter on Pazyryk 2 and Ulandryk 4 in Impact of the Environment on Human Migration in Eurasia (NATO Science Series IV, pp. 107–116) and 2020 reassessment using cosmic-ray synchronisation together place the Pazyryk culture no earlier than the third century BCE. I read the 2020 abstract, in Russian, and not the full paper.

No published chemical analysis of the Pazyryk censer residues could be located. That is a negative finding about the literature I was able to reach, not a certainty.

Jirzankal. The site description, excavation history, coordinates, elevations, radiocarbon range, brazier count, GC-MS method and results, cannabinoid table, and every quotation from the discussion are from Ren, Tang, Wu, Spengler, Jiang, Yang and Boivin, "The origins of cannabis smoking: Chemical residue evidence from the first millennium BCE in the Pamirs," Science Advances 5:6 (12 June 2019), eaaw1391, which is open access and which I read in full. The Saka attribution, the juniper fire-altars and the argument for Zoroastrian influence are from Shen, Wu, Tang, Zhou, Sun and Li, "Wood Usage and Fire Veneration in the Pamir, Xinjiang, 2500 yr BP," PLOS ONE 10:8 (2015), e0134847. The strontium isotope results are from Wang and colleagues, Scientific Reports 6 (2016), 35162, as reported in the 2019 paper. The original Chinese excavation report — Acta Archaeologica Sinica 2015 no. 2, pp. 229–252 — I have not read, and everything attributed to the excavators here reaches me through the two open-access papers.

That the Pazyryk barrows lie in the Altai Republic and not in Tuva is established by Rudenko's own site description, which places them in a hanging valley above the Great Ulagan, and by every subsequent Russian publication on the group. The distances are great-circle calculations from the coordinates the 2019 paper gives for Jirzankal, from the Pazyryk valley, and from the lower Dnieper region Herodotus calls Gerrhos; the Gerrhos figure carries the uncertainty of that identification, which is itself disputed, and should be read as an order of magnitude rather than a measurement.

What remains open. Whether burning whole cannabis inflorescences in an enclosed felt tent would in fact produce a psychoactive dose has not, so far as I can find, been tested, and the question is usually assumed rather than answered. Whether Herodotus's σπέρμα means the achene strictly or the seeded flowering head loosely is a genuine philological question with no settled answer; the article treats the strict reading as what he wrote and the loose reading as available. And nothing here bears on whether people on the Eurasian steppe used cannabis for its effects. They may well have. The claim under examination is narrower: that one Greek paragraph describes it, and that two excavations seventeen hundred kilometres apart have confirmed the paragraph. Neither part of that has been shown.

The point of separating these columns is that the date of the rite, the date of the paragraph, the date of the excavation and the date of the reading have been collapsed into one another for a hundred and sixty years.

Translations here are the author’s own and are marked as such. Sources are reproduced rather than linked. Claims that failed are printed with the verdict on them.

Subscribe — complete in the inbox.