Green Mouths and Singing Stones: Inside the Rigveda's Soma Workshop

Start With the Noise

Before the theology, before the botany, before anyone argued about mushrooms, there was a sound.

Two men kneel on either side of a pair of wooden boards laid over four pits dug into the ground. Between them sits a heap of stalks that arrived by cart, soaked overnight in river water until they went soft. Each man takes a stone in both hands and brings it down. Then again. Then faster, in a rhythm neither of them chose, because the rhythm is prescribed. The pits under the boards are not drainage. They are there to make the hammering louder, and they work: the ground itself starts to boom. A hymn is being chanted over the top of the racket, and the chant is not about the gods. It is addressed to the stones.

That scene is reconstructed, but every element in it is documented. The pits are called uparava. The boards are adhiṣavaṇa-phalaka. The stones are grāvan, and the Rigveda contains an entire hymn sung at them, RV 10.94, in which the poet notices that the stones have gone green around the mouth from the juice, that they snort like yoked horses, and that they get to taste the drink before anyone else does.

This is where most popular writing about soma never goes. The famous questions are botanical (which plant?) and pharmacological (what kind of high?), and both have been chewed over for two centuries with no agreed answer. Meanwhile the Rigveda is unusually forthcoming about something else entirely: the process. It tells you where the stalks came from, what they were crushed with, what the juice was strained through, what it was mixed with, how many times a day the operation ran, and what it sounded like. Read the corpus as a description of work rather than a description of a drug and a different civilisation comes into focus, one that built a small factory, staffed it with specialists, wrote a hundred and fourteen hymns about one stage of the workflow, and then kept the whole thing running, in modified form, for three thousand years.

The Stalks Came Down From a Mountain

Soma is not a local crop. The Rigveda is consistent about this in a way that is easy to miss: the plant comes from high ground, and specifically from a mountain called Mūjavant. The clearest reference sits in an unexpected place, the gambler’s lament of RV 10.34, where a man ruined by dice compares their pull on him to something he assumes every listener already knows:

somasyeva maujavatásya bhakṣó vibhī́dako jāgṛvír máhyam achān

“Like a draught of soma from Mūjavant, the wakeful dice-nut has pleased me.”

Rigveda 10.34, verse 1. Rendered after Wilson’s translation (1866), diacritics standardised. See also the gambler’s hymn.

Two things are worth extracting from a throwaway simile. First, soma from a named mountain was a recognised grade, the way a wine region is a grade; the poet does not say “soma,” he says Maujavata soma, and expects the distinction to land. Second, the adjective attached to the dice, jāgṛvi, means “wakeful, keeping one awake.” It is the same quality the poets ask of the drink. Nobody in this hymn is describing a dream state.

Yāska, the earliest surviving Sanskrit etymologist, took Mūjavant as a mountain name.[1] Where the mountain was is unresolved; the general scholarly direction points north and west, toward the Hindu Kush and the ranges beyond the Rigvedic heartland, which matches the wider Indo-Iranian picture in which the cognate Avestan haoma is also a highland plant.[2] The practical consequence is the interesting part. If the plant grows on mountains and the ritual happens on the plains, then every soma sacrifice depends on transport, and transport depends on someone selling.

The Rigveda does not give us a purchase ritual. The later śrauta manuals do, and what they preserve is one of the strangest transactions in the history of religion. In the classical Agniṣṭoma, the stalks arrive on a cart and are bought with a specially chosen cow, the somakrayaṇī, brown with red-brown eyes, unblemished, a virgin. The priest and the vendor then haggle in five rounds. The priest opens by offering one sixteenth of the cow. The vendor refuses, saying King Soma is worth more than that. The priest works upward: a hoof, a foot, half the cow, the whole cow, then gold and cloth and a she-goat on top. When the deal closes, the sacrificer touches the gold and says he is buying brightness with brightness and immortality with immortality. Then, in at least one manual’s reckoning, the gold is taken back off the vendor, who may be struck with a bamboo stick for good measure.[3]

Aside. The purchase ritual is post-Rigvedic, and it would be sloppy to read it back into the hymns. But it is not invented from nothing. The Rigvedic poets already think in terms of buying divine things: at RV 4.24 a poet asks, apparently seriously, who will buy his Indra for ten cows and hand him back after the enemies are dead. A culture that jokes about the resale value of a god is a culture with a functioning market economy in its head. See also the gift economy of poet and patron.

The haggling scene tells us something the hymns do not: soma had a price, the price was negotiable, and the negotiation was staged as theatre. Which means the plant was, at some point in its history, an actual traded commodity travelling down from actual mountains along actual routes.

The Apparatus

Vedic ritual vocabulary is precise about hardware in a way that rewards attention. Each object has a name, most names are transparent, and several of them turn up in the hymns as characters.

The two boards get a memorable description in RV 1.28, a hymn that stages the whole operation in a domestic register, with a wooden mortar instead of stones and a woman working the pestle. The poet compares the paired planks to a woman’s hips, addresses the mortar directly as vanaspati, “lord of the forest,” and asks the household kit to behave like temple equipment:

“If in truth, O Mortar, you are set to work in house after house, then give out here your clearest sound, loud as a conqueror’s drum.”

Rigveda 1.28, verse 5. Rendered after Griffith (1896) and Wilson (1866).

Hanns-Peter Schmidt devoted an article to this hymn in 2009, reviewing a century of readings, from the Victorian embarrassment at its sexual imagery to Zoroastrian ritual parallels to the modern proposal that the whole thing describes intercourse in the vocabulary of a soma pressing.[4] The interpretation that survives best is the least exotic one: the hymn deliberately maps solemn ritual onto kitchen equipment, which is exactly what you would do if you wanted the sacrificer’s wife inside a rite that otherwise excludes her, and if you wanted an ordinary household to be able to perform a scaled-down version.

Then there are the fingers. Maṇḍala 9 refers to the hands that handle the stalks as dáśa svásāraḥ, “the ten sisters,” dozens of times. It is not a decorative epithet; it is how the workforce is counted.

“The ten sisters, young women, seize him in the contest on the decisive day.”

Rigveda 9.1, verse 7. Rendered after Wilson (1866).

“Sporting in the wood, spurting out beyond the filter, the ten sisters press him as men rub down a strong horse.”

Rigveda 9.6, verse 5. Rendered after Wilson (1866).

Notice what has happened grammatically. The fingers are personified, the plant is personified, and the labour is described with a verb borrowed from horse-grooming. This is not mystification. It is what happens when a culture spends enough hours doing one physical task that the task acquires its own dense idiom, the way sailors or blacksmiths or climbers develop theirs.

graph TD
    A[Mountain stalks, Mujavant] --> B[Bought from vendor by cart]
    B --> C[Soaked overnight in water]
    C --> D[Crushed: stones on boards]
    D --> E[Pits below amplify the sound]
    D --> F[Juice poured through fleece filter]
    F --> G[Caught in vats and bowls]
    G --> H[Mixed with milk, curds or barley]
    H --> I[Poured into fire, drunk by priests]
    I --> J[Indra strengthened; poet's speech released]

The Stones Are the Choir

RV 10.94 is fourteen verses long and it is addressed, from beginning to end, to lumps of rock. There is nothing else quite like it in the collection. The opening verse sets up an exchange, not a prayer:

práité vadantu prá vayám vadāma grā́vabhyo vā́cam vadatā vadádbhyaḥ

“Let these speak out, and let us speak too. Speak your speech to the pressing-stones, to the ones that speak.”

Rigveda 10.94, verse 1. Rendered after Wilson (1866), diacritics standardised.

The verb vad, “to speak,” lands four times in two lines. The stones are not being praised for a divine quality; they are being credited with participating in the liturgy. Their noise is treated as a voice, and the human chant answers it. What follows is a sustained piece of close observation that any modern reader who has watched heavy manual work will recognise immediately:

“They roar like a hundred, like a thousand men; they cry out with green mouths. Getting to work, the skilful stones, by their skill, reach the offering-food before even the Hotṛ priest does.”

Rigveda 10.94, verse 2. Rendered after Wilson (1866).

Haritébhir āsábhiḥ, “with green mouths.” The stones are stained by the juice around the point of contact, and the poet, watching, decides that the stones have been eating. Verse 3 pushes it: they took the sweet stuff into their mouths, they bellow like bulls over meat, they are chewing the branch of the ruddy tree. Verse 6 shifts the metaphor from feeding to draught animals:

“Yoked together like powerful beasts drawing a load, bearing the pole, panting and gulping they roared aloud; the snorting of them is heard, like the snorting of horses.”

Rigveda 10.94, verse 6. Rendered after Wilson (1866).

And verse 4 puts the workforce back in the frame: seized by the sisters, the stones danced, filling the earth with their shouting.

Now the detail that makes all of this more than poetic licence. The śrauta manuals prescribe that the pressing boards be laid over uparava, a set of four pits dug in the ground in front of the pressing shed, roughly a forearm deep, linked by underground channels, and covered by the boards and a hide.[5] Their function is acoustic. They resonate. The pits exist so that the crushing sound will carry, and the stated reason is that the sound is an invitation to Indra.

Put the two facts side by side. A hymn addressed to the noise of the machinery, and a hole in the ground engineered to make that noise bigger. This is a culture that decided the sound of production was itself an offering, and then built infrastructure to improve it. The Rigveda has a well-known theology in which speech, vā́c, is a power in its own right; see the goddess Vāc in RV 10.125. What RV 10.94 shows is that the category was wide enough to include percussion.

Methods note. The uparava prescription comes from the śrauta manuals, not from the Rigveda, and the Rigvedic pressing may have used a different arrangement, possibly stones held in the hands over a hide. What the Rigveda independently confirms is the interest in the sound: the roaring, the snorting, the earth filled with shouting, the demand that the mortar sound like a war drum. The engineering detail and the poetic attention corroborate each other without either having to be read into the other.

Three Times a Day, Forever

The soma rite is not an event, it is a schedule. The classical form runs three pressings in a single day, and the Rigveda already knows the structure. RV 3.28 works through them in order, one short verse each, addressed to Agni at the dawn pressing, at the midday pressing, and at the third.

The third pressing belongs, tellingly, to the Ṛbhus, the craftsmen who worked their way into the company of gods by making things well. A ritual day that ends with the shift handed over to the artisans is a ritual day with a definite idea about who deserves a drink.

The juice was not served neat. Vedic has a technical term, ā́śir, for the admixture, and three compounds naming what went in: gavāśir with milk, dadhyāśir with curds, yavāśir with barley.[6] Anyone who has tasted pure Ephedra infusion, or for that matter any strongly astringent plant extract, will understand why. The unmixed juice was called śukra or śuci, “bright, clear,” and Indra is repeatedly said to prefer it mixed.

[!NOTE] The mixtures are not incidental. Milk added to a plant extract does real chemical work: casein binds tannins and cuts astringency, which is the same reason milk goes into tea. The Vedic poets had no theory of this, but they had the result, and they had three named recipes for it.

What the three pressings really produce, though, is a clock. The morning, noon and afternoon of the sacrificial day are defined by the pressings, not the other way around. Time in the Rigveda is measured by ritual work as much as by the sun, and a household that hosts a soma rite is submitting to an externally imposed timetable of some severity. This is a small point with a large consequence: the soma cult gave Vedic society one of its first shared, repeatable, publicly verifiable schedules.

What the Drink Actually Did

Here the article has to slow down, because the honest answer is that we do not know, and the confident answers on offer are mostly worse than the uncertainty.

The botanical debate is covered at length in The Soma Problem, so this section will not rerun it. The short version: R. Gordon Wasson argued in 1968 for the fly agaric mushroom, John Brough took the argument apart in 1971, and Harry Falk’s 1989 study in the Bulletin of the School of Oriental and African Studies moved the field decisively toward Ephedra.[7][8][9] Falk’s central observation was not botanical but philological. He pointed out that both the visionary theories assumed the ritual was after hallucination, when the texts themselves ask for something quite different: wakefulness, alertness, energy, clear speech. In his phrase, there is nothing shamanistic or visionary in early Vedic or Old Iranian texts.[9]

That is a claim about the texts, and it holds up against the hardware. Look at what the poets actually want from the operation. They want the stones to roar. They want the juice bright. They want Indra to grow large and go and do something violent. They want their own speech to work. Read RV 10.94.9 again: the stones’ output goes into Indra, and Indra “increases, spreads wide, acts the bull.” Nobody in that verse is having a vision.

Aside. The word “psychedelic” was coined in 1957. Applying it to a Bronze Age ritual is not merely anachronistic vocabulary; it imports a whole modern framework in which the point of a substance is interior experience. The Rigvedic framework is different: the point of the drink is that it makes an agent stronger and makes a poet’s words land. The nearest modern analogue is not a retreat but a performance. On the one hymn that comes closest to a first-person report, see RV 8.48.

A rough calculation is worth doing here, with the caveat attached in advance. Ibragic and Sofić measured total ephedrine plus pseudoephedrine across eight Ephedra species and found between 20.8 and 34.7 mg per gram of dry weight.[10] If a pressing uses a mass $M$ of dried stems, at concentration $c$, with an extraction efficiency $\eta$, the alkaloid recovered is:

$$m_{\mathrm{alk}} = M \cdot c \cdot \eta$$

Take $M = 1\ \mathrm{kg}$, $c = 25\ \mathrm{mg/g}$, and a deliberately pessimistic $\eta = 0.3$ for cold crushing and a fleece filter. That yields roughly 7.5 grams of alkaloid, which, shared among sixteen participants across three pressings, is on the order of 150 mg each per pressing. A single pharmacological dose of ephedrine is a few tens of milligrams.

Methods note. This is an order-of-magnitude illustration using a modern plant, not a claim about what the Rigvedic priests drank. Every input is uncertain: we do not know the species, the quantity, the extraction efficiency, or how many people drank. What the arithmetic does establish is a negative result worth having: you do not need an exotic or hallucinogenic plant to explain the reported effects. A common highland shrub, crushed in bulk, gets you comfortably into stimulant territory. Exoticism is not required, and in the history of this debate exoticism has usually been the problem.

The Rigveda’s own descriptions are consistent with this and stubbornly resistant to more. Soma makes Indra vast. It makes the poet’s dhī, his insight or vision-thought, function. It is repeatedly called jāgṛvi, wakeful. It is honey, it is a flood, it is a king, it is a stallion, it is a bird. What it is not, anywhere in the family books, is a doorway to another world.

The Machine Never Fully Stopped

Here is the fact that reframes everything above. This is not an extinct process.

In April 1975, in the village of Panjal in Kerala, Nambudiri brahmins performed the Atirātra Agnicayana, the twelve-day rite that includes the piling of a falcon-shaped altar from 1,005 bricks and, at its heart, the pressing of soma. Frits Staal of Berkeley organised the documentation, with Robert Gardner filming and a team of scholars observing, and published the results in the two volumes of Agni: The Vedic Ritual of the Fire Altar.[11] It was widely assumed at the time to be a last performance. It was not. The rite was performed at Panjal again in April 2011, with Michael Witzel and a group of Harvard students in attendance, conducted by seventeen specially trained priests who had learned their recitations in family schools from childhood.[12]

The plant is the honest complication. Ephedra does not grow in Kerala. Modern performances use a substitute creeper, most often identified as Sarcostemma (a leafless milkweed vine), and the substitution is not a modern lapse: the Brāhmaṇas already legislate for surrogates, which tells you the supply chain was breaking down within a few centuries of the hymns.[2] The plant changed. The apparatus did not. The stones, the boards, the fleece, the vats, the three pressings, the order of drinkers, the chants: all of it carried, transmitted by the same oral machinery that preserved the hymns themselves (see the oral engine of Vedic transmission).

[!TIP] If you want a sense of what RV 10.94 is describing, do not read it silently. Find the footage from Panjal, turn the sound up, and watch the pressing sequence with the hymn open beside you. The “hundred, the thousand” of verse 2 stops being a metaphor within about ten seconds.

Two observations follow.

The first is about what “oldest” means. The Rigveda is routinely called the oldest surviving text of the Indo-European world, and that is true but slightly inert. What is remarkable is not the age of the words. It is that the words describe a procedure, and the procedure is still executable. There is a set of instructions from the second millennium BCE that a group of people can still run.

The second is about what the culture valued. Look at what got preserved with the most care: not a myth, not a doctrine, but a workflow. The Rigveda’s single largest thematic block, all 114 hymns of Maṇḍala 9, is devoted to one stage of that workflow, the moment the juice passes through the wool. A civilisation reveals its priorities by what it refuses to abbreviate.

What the Workshop Was For

Strip the soma rite to its logic and it is a transformation with four steps. Something is brought down from a height. It is broken. It is filtered. It is shared out, and speech comes with it.

That shape recurs everywhere in Vedic thinking. It is the shape of ṛta, the cosmic order, in which things are not created so much as clarified into their proper form. It is the shape of the poet’s own work, described elsewhere in the corpus as carpentry: raw material, skilled violence, a finished object handed over. It is why the deified soma of Maṇḍala 9 is not the plant and not the drink but the participle, pávamāna, “purifying himself,” a god caught permanently mid-process.

The workshop, in other words, is a cosmology you can operate. That is an unusual thing for a religion to offer, and it is worth being precise about why it is unusual. Most traditions ask you to believe something. This one hands you two stones, a fleece, a heap of stalks and a schedule, and lets the meaning accumulate out of the doing. The green stains on the stones were not a symbol of anything. They were the stones getting the first taste, and a poet three thousand years ago thought that was worth writing down.

Three things to read next, in this order. Start with RV 10.94, the whole hymn, out loud, quickly. Then RV 1.28, which is the same operation performed in a kitchen with a wooden mortar and a woman working the pestle. Then open Maṇḍala 9 anywhere at all and read fifty verses without stopping, letting the repetition do what repetition does. By verse thirty you will not be reading about a drink. You will be listening to a machine.