Eight Words for a Vanished Band: Musical Instruments in the Rig Veda

A Line That Plays Three Instruments at Once

Somewhere in the middle of a short, excitable hymn to Indra, composed by the Priyamedha family and now sitting as the sixty-ninth hymn of the eighth book, a Rigvedic poet stops praising and starts listening. The verse before it is nothing but an instruction to make noise: arcata prārcata priyamedhāso arcata, sing, sing forth, you Priyamedhas, sing. Then comes this:

ava svarāti gargaro godhā pari saniṣvaṇat |

piṅgā pari caniṣkadad indrāya brahmodyatam

“The gargara sounds down; the godhā resounds all round; the piṅgā leaps about. For Indra the formulation has been raised up.”

Rig Veda 8.69, verse 9. My rendering from the Sanskrit, deliberately leaving the three nouns untranslated; compare Griffith (1896). [1] [2]

Three nouns, three verbs of sound and motion, and one of the shortest crises in Vedic philology. Ralph Griffith, translating in Benares in the 1890s, heard an ensemble and wrote it down as one: “Now loudly let the viol sound, the lute send out its voice with might, shrill be the music of the string.” [2] The standard reference work of Vedic realia, Macdonell and Keith’s Vedic Index, accepts that the three are probably instruments while conceding that godhā, which has the word for cow or ox buried in it, might instead be something made of hide. [3] Other readers have found no orchestra here at all, only the racket of archery equipment, or of animals. Nobody has produced a decisive argument, because each of these words appears in the Rig Veda either once or almost once, and a word that occurs once cannot be triangulated.

That single verse is the densest patch of musical vocabulary in a collection of more than ten thousand verses. Which is the surprise worth sitting with. The Rig Veda is, by a wide margin, the oldest music that human beings still perform: the tradition of Vedic chanting was proclaimed by UNESCO in 2003 and inscribed on its representative list in 2008, and the recitation you can hear in Kerala or Pune today preserves a pitch system and a metrical shape that are demonstrably ancient. [13] And yet on the question of what the Bronze Age poets played, or whether they played anything, the text is nearly mute. Eight words. Sixteen occurrences. One line of three hapaxes. What follows is an inventory of that wreckage, and an argument about why the voice came through intact while everything it might have been accompanied by disappeared without a trace.

The Whole Inventory Fits in One Table

Here is everything. Not a selection, not the highlights: the complete instrument vocabulary of the Rig Veda, with every verse in which each word appears.

Four of those eight are attested exactly once, and two of the remaining four are attested twice. For comparison, the word ghṛta, clarified butter, turns up in something like a hundred and fifty verses; the collection has more to say about skimming fat off boiled milk than about everything anyone ever plucked, blew or struck. See the Rigvedic kitchen inventory for what that comparison looks like from the other side.

Methods note. The occurrence counts here come from a string search of the Devanagari text of the Saṃhitā, checked verse by verse against the romanised text and Griffith’s translation, and they are counts of words, not of instruments. Searching for musical vocabulary in Vedic Sanskrit is a homograph minefield: a naive search for vāṇa returns more than a hundred hits, almost all of them the tail of some unrelated word (anarvāṇam, grāvāṇaḥ, bruvāṇaḥ). The five vāṇa passages listed above are the ones that survive inspection, and even the fifth, RV 4.24.9, is disputed. Where I give a round number I mean a round number.

The One We Are Sure About

The drum is solid ground. Dundubhí is transparently onomatopoeic, it survives into classical Sanskrit with the same meaning, and the Rig Veda uses it in a context that removes all doubt: three verses at the end of RV 6.47, addressed not to a god but to the instrument itself, in the vocative, with imperatives.

ā krandaya balam ojo na ā dhā ni ṣṭanihi duritā bādhamānaḥ |

apa protha dundubhe ducchunā ita indrasya muṣṭir asi vīḷayasva

“Roar out; put strength and power in us; thunder down, driving off hardships. Snort away misfortune from here, O drum. You are the fist of Indra: be hard.”

Rig Veda 6.47, verse 30. My rendering; compare Griffith (1896), who has “Thunder out strength and fill us full of vigour”. [2]

Read that as a piece of military engineering and it makes complete sense. The drum is not accompanying a song. It is a weapon of morale, pointed at the enemy line, and the poet tells it so: indrasya muṣṭir asi, you are Indra’s fist. The verse before it asks the drum to make earth and heaven breathe hard and to push the enemy far off; the verse after it notes that the drum “speaks repeatedly, carrying the signal” (ketumad dundubhir vāvadīti) while the chariot-warriors come together. This is a battlefield signalling device with a theology attached. The same short block also addresses the war chariot and the bowstring, which is why it belongs as much to the history of Vedic objects as to the history of Vedic music; the hymn that talks to the weapons is its closest relative in the collection.

What the drum does not do in the Rig Veda is play in the rite. The one other occurrence is a simile, and a domestic one: in RV 1.28.5 the wooden mortar used for pressing is told to “give out your loudest sound here, like the drum of conquerors”. The mortar is not a drum; it is being flattered by comparison to one, which tells you the drum was the loudest thing the poet could think of. (The acoustics of the pressing operation get their own treatment in the soma workshop.)

Watch what the next generation of texts does with the same instrument. The Atharvaveda gives the war drum two entire hymns, AV 5.20 and 5.21, in Whitney’s translation “the loud-noised drum, warrior-like, of forest tree”, asked to speak “division of hearts, division of minds” among the enemy. [9] The ritual manuals go further and invent a bhūmi-dundubhi, an earth drum: a pit dug in the ground and covered with a hide, beaten at the Mahāvrata rite near the winter solstice to drive off whatever might keep the sun from returning. [3] The drum, in other words, gets steadily more liturgical after the Rig Veda, not before it. In the Rig Veda it is still mostly out on the field with the soldiers.

The Vāṇa Problem

Vāṇa is the interesting one, because we have four or five usable passages and they refuse to converge.

Take these in order and the object keeps changing shape. Sāyaṇa, the fourteenth-century commentator, and H. H. Wilson after him read dhamantaḥ in the Maruts passage as blowing on a pipe, which is the natural sense of the verb. [4] Griffith, at the same verse, quietly declined to name the thing at all and wrote “sending forth their voice”. [2] At RV 10.32.4 he produced “the seven-toned people of the choir”, turning the instrument into an ensemble of singers. Macdonell and Keith’s index leans the other way and treats vāṇa as a stringed instrument, which is where the word certainly ends up: in later Sanskrit vāṇa shades into vīṇā, the plucked instrument that anchors classical Indian music, and post-Rigvedic texts describe a vāṇa of a hundred strings. [3] [6]

Two observations are worth making. The first is that “blown” and “seven-stranded” are only contradictory if we assume a single object. A set of seven reed pipes bound together is blown and is seven-stranded, and so is a chorus of seven voices, which is exactly the phrase the Rig Veda uses elsewhere: sapta vāṇīḥ, seven voices, appears at RV 1.164.24, 3.1.6, 3.7.1 and 8.59.3. The masculine vāṇa and the feminine vāṇī are the same root wearing different endings, and vāṇī unambiguously means voice. At RV 3.30.10 the vāṇīs themselves are described as dhamantīḥ, blowing. If voices can be blown, then a blown vāṇa need not be a flute.

The second observation is more deflationary, and I think it is right: vāṇa in the Rig Veda may not name a type of instrument at all. It may name sounding, the produced noise, whether from throats or from tubes, in the way English “brass” or “strings” names a section rather than an object. That would explain why the poets can blow it, sound it in concert, count it in sevens, and fit it with a rim, all without contradiction, and it would explain why the word later hardens into the name of a specific instrument once a specific instrument needs naming.

Aside. This is what a live philological problem looks like from the inside. There is no manuscript variant to resolve it, no archaeological find to check it against, and no later commentator with independent access to the answer: Sāyaṇa is separated from these hymns by more than two thousand years and is reasoning from the same text we have. When you read that the Vedic people played the lute, ask which verse says so, and then ask what else that verse could mean.

Back to 8.69.9: Lute, Bowstring, or Lizard

Now the crux. Three words, one verse, and three live readings.

The ensemble reading is Griffith’s and, tentatively, the Vedic Index’s: a stringed trio strikes up for Indra. [2] [3] It has the merit of fitting the context, since the previous verse is a command to sing, and the merit of explaining why three obscure nouns would be listed together at all.

The archery reading takes the same three words as the equipment of a bow: the string sounding, the leather arm-guard slapping, the tawny cord leaping. It is not eccentric. Godhā in later Sanskrit covers a bowstring and the leather guard strapped to an archer’s forearm as well as a monitor lizard, and the Rig Veda itself uses the word in a context that has nothing to do with music. At RV 10.28.10, in a riddling hymn, a godhā turns up in a run of creatures caught fast (an eagle that has left its talon behind, a lion in a snare, a thirsty wild buffalo), and Griffith rendered it there as the leather strap that holds an entangled foot. [2] The same translator who made godhā a lute in book 8 made it a leather thong in book 10, and he was not being careless. He was being governed by context, which is all any of us can do.

The animal reading has the verbs on its side. Ava svarāti is a downward sounding; pari saniṣvaṇat is a resounding all around; and pari caniṣkadat is an intensive of the root skand, to leap, so the third thing is hopping about. A bowstring can leap. A lizard can certainly leap, and so can a frog, and the Rig Veda elsewhere shows it is perfectly capable of building a whole hymn out of amphibian noise, as the frog hymn demonstrates. A lute cannot leap, at least not while being played.

What can actually be concluded? Three things, and only three. That the verse is a deliberate sound-scene, built from three different acoustic verbs, placed immediately after a command to make noise. That its three nouns were probably ordinary vocabulary for the audience and were already obscure by the time anyone wrote a commentary. And that the honest translation is the one that leaves them alone, as Jamison and Brereton’s practice of flagging irreducible obscurity implies. [1] The verse is not a puzzle with a lost solution so much as a measurement of how much ordinary life a ritual anthology does not bother to preserve. For the collection’s other deliberate withholdings, see what the Rig Veda hides.

Notice Where the Instruments Are Not

Lay the eight words out by context rather than by type, and a pattern appears that no single verse would reveal.

graph TD
    A[Rigvedic instrument words] --> B[Battlefield]
    A --> C[Wild divine troop]
    A --> D[Forest and birds]
    A --> E[One unexplained verse]
    A --> F[The chanted rite]
    B --> B1["dundubhi, bakura"]
    C --> C1["vana of the Maruts"]
    D --> D1["aghati, karkari"]
    E --> E1["gargara, godha, pinga"]
    F --> F1[No instrument named]

The drum and the bakura are military. At RV 1.117.21 the Aśvins are “blowing at the Dasyu with the bakura”, which puts it somewhere between a trumpet and a weapon; at RV 9.1.8 somebody blows a bākura skin bag, which is why the older literature guesses at a bellows or a bagpipe. The vāṇa belongs to the Maruts, the divine storm troop who are the noisiest collective in the pantheon and who arrive shouting, adorned, and described as dancers; that is their whole character. The āghāṭi and the karkarī appear only inside similes about wild sound: cicadas in the forest goddess hymn chirp “as if with āghāṭis” at RV 10.146.2, and a partridge’s cry in RV 2.43.3 is asked to sound “like a karkarī”, a verse that belongs to the strange world of Vedic bird omens.

Not one of the eight words appears inside a description of the ritual it was composed for. The soma is pressed, the fire is kindled, the offering is poured, the priests are sorted into their seven cups and specialisms, and nobody plays anything. Vedic instruments, on the evidence of the Rig Veda itself, are for fighting, for weather gods, for the wilderness, and for similes. The rite runs on human voices.

That is a conclusion about the text, not about Bronze Age India. Argue from silence with care: the Rig Veda also has no word for rice, wheat or salt, and nobody thinks the Punjab was rice-free forever. What the silence tells us is what the genre wanted to record.

What the Voice Was Doing Instead

Set the eight instrument words next to the vocabulary for producing sound with the body, and the asymmetry is almost comic. The frequencies below are deliberately rough: a raw search for sāman or arka pulls in homographs, so treat the right-hand column as an order of magnitude rather than a census.

These are not synonyms padding out a thesaurus. Uktha and sāman mark a real distinction between recited and sung, the distinction that will shortly organise the whole Vedic priesthood into reciters and chanters and produce the Sāmaveda as a separate collection of Rigvedic verses rearranged for melody. [7] [10] The verb stobhati at RV 1.88.6, used of the way the poem answers the Maruts, is already the technical term for the meaningless syllables that Sāmavedic chant inserts into a verse to carry its tune, the ho-is and bhās that Wayne Howard spent a career transcribing from living schools. [7] The apparatus of a full musical practice is present in the Rig Veda’s vocabulary of the voice while its instrument list is stuck at eight words.

And they danced. This is not an inference from temple sculpture centuries later; the text says so in ordinary language. Dawn “puts on her ornaments like a dancer” (nṛtūr iva) at RV 1.92.4. Indra is addressed in the vocative as nṛto, “O dancer”, at RV 1.130.7 and RV 2.22.4, and compared to an active dancer at RV 6.29.3. The Maruts are called nṛtavaḥ, dancers, at RV 8.20.22. Best of all:

chandastubhaḥ kubhanyava utsam ā kīriṇo nṛtuḥ

“Keeping time to the metre, chanting, the praisers danced around the spring.”

Rig Veda 5.52, verse 12. My rendering; Griffith has “Hymn-singing, seeking water, they, praising, have danced about the spring”. [2]

Chandastubh is built from chandas, metre, and the same root stubh as stobha. The line describes people singing in metre and dancing at a waterhole, and it is offered as a simile for the Maruts. Whatever the Rigvedic musical event was, it had rhythm you could move to, and it needed no gargara to supply it.

Aside. Two things that get confused are worth separating. Whether Rigvedic chant is “music” is a definitional question, and the answer is yes in any useful sense: it is organised pitch and duration, performed. Whether it was accompanied is an empirical question, and the answer, on this evidence, is mostly no. Lewis Rowell’s history of early Indian musical thought treats Vedic chant as the foundation of the whole later theoretical tradition precisely because of how much it did with so little apparatus. [6]

Wood, Hide, Gut, Reed

Why did the band vanish so completely? Partly because it was made of the wrong things.

A Bronze Age drum is a hollowed log and a stretched hide. A pipe is a reed. A plucked string is gut or plant fibre on a wooden frame. In the monsoon belt of northern India these materials return to soil within a few decades, and unlike Mesopotamia or Egypt, the Vedic world produced no tombs stocked with grave goods and no clay tablets listing temple inventories. The material record for music in South Asia is correspondingly thin. What survives from the Bronze Age is what happened to be fired: the terracotta rattles and small bird-shaped whistles that Jonathan Mark Kenoyer notes are common at Indus sites, made for children as much as for ritual, and unconnectable to any text. [11]

That last row is the one that should unsettle anyone who has seen a museum label confidently captioning a harp as “Vedic”. The earliest detailed depictions of Indian instruments are carved on Buddhist monuments, and they are separated from the composition of the Rig Veda by something like a thousand years. They are excellent evidence for the instruments of the early historic period and no evidence at all for the hymns.

The other half of the answer is that the voice was, by a long way, the better storage medium. What the Rigvedic schools built around the chant was a redundant encoding of extraordinary strength: a fixed text, a three-tone accent that is part of the meaning rather than decoration, strict metrical templates, and then a battery of recitational transformations, word-by-word and criss-crossed, designed to make error detectable. Frits Staal’s work on the living tradition, and the Nambudiri performances he documented, showed how much of this machinery survives in practice rather than in theory. [12] [18] The accent system, the metres and the memorisation apparatus each get their own treatment on this site, and the joint effect is a transmission system with no moving parts.

An instrument has moving parts. It needs a maker, a supply of seasoned wood and good gut, a tuning convention, and a player trained in something other than the text. Every one of those is a point of failure, and none of them is required for a chant. The tradition that survived three thousand years is the one that could be carried entirely inside trained human bodies.

The Sound That Was Worth Keeping

There is a temptation, reading an inventory this thin, to conclude that the Rigvedic world was musically poor. The evidence points the other way. A culture with eight instrument words and a hundred words for kinds of singing is not a culture indifferent to sound; it is a culture that had decided which sound mattered. The poets were connoisseurs of noise: they notice the pitch of a mortar, the timbre of a cicada, the difference between a partridge calling on the wing and sitting still, the way a river’s roar rises like a swell of power. They built an entire technical vocabulary for what a human voice can do to a syllable. They simply did not think the objects that made the other noises were worth writing down, because the hymn was not an accompanied song. It was the instrument.

Which leaves the three words in RV 8.69.9 as something better than a philological embarrassment. They are the sound of the room the hymns were performed in, coming through the wall. Somebody in the eighth book heard a gargara and a godhā and a piṅgā going at once, thought the noise worth one line, and assumed anyone listening would know what he meant. He was right for perhaps four hundred years.

Here is the experiment worth running. Read the drum verses at the end of RV 6.47 out loud, then read RV 8.69.9 out loud, and notice which one you can hear. The drum arrives fully formed across three thousand years because the poet told us what it does: roar, thunder, snort, terrify, signal. The three nameless instruments arrive as pure syntax. The difference between the two is not the age of the text. It is that one poet described a function and the other trusted a noun, and functions travel while nouns rot along with the wood they named.