Nobody Was Fasting: What Vratá Meant While It Still Belonged to the Gods

Seventy-Nine Percent

In 2021 the Pew Research Center asked 29,999 Indian adults, 22,975 of them Hindus, a long list of questions about religious practice. One of the answers is a number worth holding on to: 79 percent of Hindus said they fast. Across all Indian respondents it was 77 percent, which puts religious fasting among the most widely shared practices the survey measured.[1]

The word almost all of those people would use for what they are doing is vrat, from Sanskrit vratá (व्रत). It is one of the oldest words in the language still in daily use, and it is in the Rigveda roughly two hundred times. So the obvious thing to do is open the Rigveda and find out what a Bronze Age vratá looked like: what you gave up, for how long, and what you expected in return.

The search fails. Not a single Rigvedic verse describes anyone abstaining from food under the heading of vratá. What the word does instead is govern the sky. Night falls because Savitṛ’s vratá says so. The waters run in their channels by Soma’s vratá. Heaven and earth hold their positions, mountains included, because they are standing inside somebody’s vratá. Agni, not Varuṇa, is the god most often described as its guardian. And in one late hymn in the ninth book, the vratā of human beings turn out to be their trades: the carpenter’s, the physician’s, the priest’s.

Between that and a woman in Delhi skipping water until moonrise on Karva Chauth lies a semantic shift of unusual size, and an unusually clear mechanism. The pivot is documented, step by step, in prose texts composed a few centuries after the Rigveda, where the word stops meaning “the order a god keeps” and starts meaning “the milk you are allowed to drink”. What follows is that trail, read from the Saṃhitā outward.

A Word With No English

Start with the translators, because their disagreements are evidence. Ralph Griffith’s 1896 version is still the most widely circulated English Rigveda, and he had to render vratá on more than two hundred occasions.[2] He never settled on anything.

Law, dominion, commandment, way, sway, duties, devotion. A Victorian vocabulary straining at something that English divides into separate boxes and Vedic does not. The common element is not moral obligation and not ritual abstinence. It is this: a vratá is a laid-down course that something or someone keeps to. Whether the keeper is the sun, a god, a river, a loyal wife or a wheelwright is a matter of which verse you are in.

Once that is the working definition, the Rigvedic usage stops looking scattered.

The Sun Keeps Its Appointments

The second maṇḍala has a pair of hymns to Savitṛ, the god who drives things into motion and, at dusk, out of it. Verses 7 and 8 of 2.38 are a catalogue of evening: the wild animals go to the water, the woods are handed over to the birds, every bird finds its nest and every beast its lodging. The frame around that catalogue is the word we are tracking.

samāvavarti viṣṭhito jighīṣur viśveṣāṃ kāmaś caratām amā abhūt

śaśvān apo vikṛtaṃ hitvy āgād anu vrataṃ savitur daivyasya

“He has turned back, spread wide, eager to win; the desire of all that moves has come home. Every man has come in, leaving his work unfinished, following the commandment of divine Savitṛ.”

Rigveda 2.38, verse 6. My rendering. Griffith renders apo vikṛtaṃ hitvā as “leaving his evil doings”, which is hard to square with verse 4 of the same hymn, where the craftsman likewise leaves his labour half-completed at dusk.

Three verses earlier the same hymn reports that Savitṛ “will unyoke even the swift horses” and that “night closely followed Savitar’s dominion” (RV 2.38.3). The god is not issuing an edict about virtue. He is running a timetable, and the timetable is his vratá. Human beings appear in the hymn as the people who knock off work when the light goes.

Collect the other passages and the pattern holds with almost mechanical consistency. Mitra and Varuṇa “go according to the vratá” when they raise the sun into the sky (RV 4.13.2). Savitṛ, looking over all creatures, “guards the vratāni” (RV 4.53.4). The dawns are multiplied for Varuṇa by those who follow his vratá (RV 8.41.3); in the next verses of the same hymn all the gods follow it, and he wraps the dark regions in light “according to his vratā” (RV 8.41.7, 8.41.10). Soma, in the filtering hymns of the ninth book, is told that “the waters attend upon your vratá” (RV 9.82.5) and that the peoples should “stand in your vratá” (RV 9.86.37). The flattest statement of the idea is addressed to Indra:

tava dyāvāpṛthivī parvatāso ‘nu vratāya nimiteva tasthuḥ

“Heaven and earth and the mountains stand as if pegged down, in keeping with your commandment.”

Rigveda 3.30, verse 4. My rendering. Griffith construes the dative differently, “for him who followeth thy Law”.

Nimitá is a word for something driven in like a tent peg. The image is of a world held in place by compliance with an order, which is why the natural English for vratá in these verses keeps sliding between “law” and “course”.

Aside. This is close to ṛtá, the Rigveda’s great word for the way things truly fit together, but it is not the same thing and the difference is instructive. Ṛtá is impersonal and singular: the truth of the arrangement. A vratá belongs to somebody. It is always your vratá or his, which is why the word takes possessives and ṛtá mostly does not. Varuṇa does not own ṛtá; he watches over it, as the article on his cosmic bookkeeping sets out. But he does own vratā, and other beings can be inside them or outside them. For the neighbouring word dhárman, the prop or foundation, see the piece on dharman before dharma.

Agni Holds the Register, Varuṇa Holds the Epithet

Here the counting produces a surprise. If you had to guess which god the Rigveda associates most with vratá, Varuṇa is the obvious answer: he is the god of the binding word, the noose, the moral audit. But sort the verses by the deity of the hymn they sit in, and Agni wins by a wide margin.

Counts are of verses in the Śākala text containing any vrat- form, with hymn dedications taken from Griffith’s headings.

Even adding every hymn where Varuṇa is a dedicatee, alone or paired, he reaches the mid twenties and stays behind Agni. The reason is visible once you read the Agni verses: Agni is not the author of the commandments, he is their keeper. The standing title is vratapā́, “guardian of the vratá”.

sa jāyamānaḥ parame vyomani vratāny agnir vratapā arakṣata

“Being born in the highest heaven, Agni the guardian of commandments guarded the commandments.”

Rigveda 6.8, verse 2. My rendering, checked against Griffith (1896).

The same office turns up in a stranger place. Twice the Rigveda preserves a near-identical verse in which a priest explains how he found Agni hiding in the waters: “the guardian of the gods’ vratā told me” (RV 5.2.8, RV 10.32.6). Someone in this system is tasked with knowing where everybody is supposed to be, and that someone speaks.

Varuṇa’s connection to the word is real but it runs through a different grammatical door. The compound dhṛtávrata, “whose commandment is held firm”, is his standing epithet: RV 1.25.6, 1.25.8, 1.25.10, 1.44.14, 2.1.4, 6.68.10, 8.25.2, 10.66.5, and more. Agni carries the bare noun; Varuṇa carries the adjective of fixity. Joel Brereton’s study of the Ādityas is the place to follow what that division of labour does to the theology of the group Varuṇa heads.[3] For a reader of the hymns the practical upshot is simpler. Two of the three things the word needs in order to work are already named: an order, and a god who does not let it slip.

Methods note. The counts above come from running regular expressions over a romanised Śākala Saṃhitā and reading every hit. Two cautions. First, the text is transmitted with sandhi, so words are often fused across boundaries and a token count is approximate where a verse count is not. Second, string matching is not lemmatisation: vratater iva, “like a creeper”, at RV 8.40.6 belongs to vratatí, a climbing plant, and has been discarded. Normalising per maṇḍala with

$$r_m = \frac{1000\,n_m}{V_m}$$

where $n_m$ is matching verses and $V_m$ total verses, gives rates from about 5 per thousand in Maṇḍala 4 to about 35 in Maṇḍala 2 against a corpus average near 19. With three hits in one book and fifteen in another, that spread is mostly noise and should not be made to carry an argument about strata.

The Carpenter Wants Something Broken

The human end of the word is where it gets interesting, and the best single verse is in the ragbag of short hymns that closes Maṇḍala 9.

nānānaṃ vā u no dhiyo vi vratāni janānām

takṣā riṣṭaṃ rutaṃ bhiṣag brahmā sunvantam ichati

“Our thoughts go different ways; the vratā of people are various. The carpenter wants something broken, the physician a fracture, the priest someone pressing soma.”

Rigveda 9.112, verse 1. My rendering; Griffith has “we all have various thoughts and plans, and diverse are the ways of men”.

No god appears in that line. Vratā here are what people are set up to do and therefore what they want to see in the world: the joiner hopes for broken furniture, the bone-setter for a bad fall, the officiant for a patron with soma to press. The hymn is a little masterpiece of unsentimental observation, and it tells us the word was not reserved for the sacred register. A vratá could be a trade.

The compounds fill in the rest of the range.

Two of these deserve a second look. Vívrata is used of the pair of bay horses that pull Indra’s chariot when they strain in different directions, and that usage is the purest available evidence for what the stem means: not “vow”, which cannot diverge, but “course held to”, which can. Sávrata, at the other end, describes Heaven and Earth as a couple keeping the same schedule, and in RV 6.70.3 their products are “various in form but alike in vratá”, which is roughly how a modern biologist might describe a shared developmental program.

And then there is the negative. To be avratá is to be outside the arrangement altogether, and in the Rigveda the charge falls on enemies: the Dasyu of RV 10.22.8 is “without rites, without sense, of another vratá, not human”. That verse has a long and contested history in modern readings of Vedic conflict, discussed in the article on the Dāsas and Dasyus. For the semantics, note only what is being alleged: not vice, but non-compliance with a regime. The Rigvedic insult for an enemy is that he keeps no timetable.

Where the Word Came From

The etymology is unsettled in an informative way. Older Sanskrit lexicography derived vratá from the root vṛ, which covers choosing and willing, giving something like “what has been willed”, hence a command. A second derivation connects it instead to an Indo-European root for solemn speech, which would make a vratá something uttered and thereby made binding; that is the line the Iranian cognate has tended to encourage, and the recent overview of the vrata- and vrāta- words sets out the competing proposals.[15] Both readings have the advantage of explaining why the word behaves as a possession: an order has an author.

The decisive comparative fact sits in Iran. Avestan has urvāta-, glossed in teaching editions of the Gāthās as “rule, order, commandment” and appearing in Yasna 30, the poem about choice and consequence that is the centre of Zarathustra’s surviving work.[4][5] Vedic vratá and Avestan urvāta- are the same inherited word, which puts the “commandment” sense back before the two traditions parted, somewhere in the second millennium BCE. Hanns-Peter Schmidt devoted an entire monograph to the pair in 1958, reading the Indo-Iranian term as a binding formula close to a promissory oath, which is the strongest case anyone has made for a “vow” component in the inherited meaning.[6] His book and P. V. Kane’s enormous survey of the later Hindu vratas appeared in the same year and were noticed together in the same review, so both halves of the word’s history reached print at once.[6][7]

graph TD
    R["Inherited root: to will, or to utter"] --> V["Vedic vratá"]
    R --> U["Avestan urvāta: commandment"]
    V --> A["A god's standing order"]
    V --> B["The course a thing keeps"]
    V --> C["A person's trade or calling"]
    A --> D["Agni as vratapā, its keeper"]
    A --> E["avratá: the outsider"]
    B --> F["Brāhmaṇa: the rite's rule of life"]
    F --> G["vrata as the consecrated diet itself"]
    G --> H["Purāṇic vrata: fast, worship, gift"]
    H --> I["Modern vrat observed by most Hindus"]
    H --> J["Jain mahāvrata; Gandhi's eleven vratas"]

Aside. Cognates constrain a reconstruction without settling a translation. Avestan urvāta- tells us the Indo-Iranians had a word for an authoritative stipulation; it cannot tell us whether a Rigvedic poet in the Punjab heard “commandment”, “covenant” or “the way it is done”. The honest position is that the Vedic evidence rules out “fast” and makes “order kept to” the best available fit, and that the Iranian branch keeps the word on the lawgiving side of that range.

The Milk That Was Called a Vow

So where does abstinence come in? The Rigveda does hand us one hint, and it is in a comic hymn. The frogs of RV 7.103, silent all year in the dried mud and suddenly loud when the rains break, are compared to brahmins:

saṃvatsaraṃ śaśayānā brāhmaṇā vratacāriṇaḥ

vācaṃ parjanyajinvitāṃ pra maṇḍūkā avādiṣuḥ

“Lying still for a year like brahmins performing an observance, the frogs have spoken out a voice quickened by Parjanya.”

Rigveda 7.103, verse 1. My rendering, checked against Griffith (1896). The hymn’s humour is unpacked in the piece on the frog hymn.

Here vratacārin means a man under a temporary rule of conduct that includes keeping quiet, which is the first clear sight of the later idea. It is also, on the standard stratigraphy of the collection, late material rather than early.[8] One verse does not make a doctrine, and the joke only works if the audience already knew about year-long priestly silence, so the practice was running before the word settled on it.

The settling happens in the Brāhmaṇas, and the mechanism is almost comically concrete. In the Śrauta ritual the sacrificer is first consecrated, and from that point he lives on milk. Julius Eggeling, translating the Śatapatha Brāhmaṇa in the 1880s, had to find an English word for what the texts call the consecrated man’s vrata, and settled on “fast-milk”: “hereafter his food consists of fast-milk (vrata) only”.[9] If the vrata-cow will not give milk, the text explains, he may make his vrata from cereals instead. The word for a god’s standing order has become the name of the only thing in the house the sacrificer is permitted to swallow.

The same text supplies the bridge. The householder entering on the rite “enters on the vrata” by giving up certain foods and certain pleasures, by keeping silent at the right moments, and above all by not speaking untruth, since untruth leaves a man foul within. And the god addressed when he takes it up is the one the Rigveda had already made its keeper: “Agni is Lord of Vows to the gods, and it is to him therefore that he addresses these words.”[9] Vratapā́ of the hymns has become vratapati of the prose, with the content of the office quietly transferred from the courses of the sun to the diet of a man in a hut.

Fasting for a Reason

The last move in that table is the one that changes the word’s moral shape. In the Purāṇic material a vrata is no longer something you are inside because the universe is arranged that way. It is something you take on, for a stated period, aimed at a named deity, and undertaken for merit, for a specific favour, or to expiate a specific fault. Anne Pearson’s survey of Purāṇic vrata texts traces exactly that compound of fasting, worship and gift-giving, and the categories under which the texts sort them.[10] Kane’s volume documents the scale of the later literature: hundreds of named observances, each with its calendar day, its deity, its permitted foods and its promised return.[7]

The modern ethnography adds the part the Sanskrit texts leave out. Pearson’s later fieldwork among women in Banaras and Mary McGee’s study of the vrata tradition both describe a practice that is overwhelmingly women’s religion, largely outside priestly control, transmitted between households rather than from manuals, and explained by the people who keep it in terms that have little to do with merit accounting.[11][12] The title of Pearson’s book comes from that register of explanation rather than from the texts: because it gives me peace of mind.

That is a long way from Savitṛ unyoking the horses at dusk, and the distance is worth stating precisely rather than dramatising. Three things changed. The keeper moved from the god to the human. The timescale moved from perpetual to a day or a fortnight. And the direction of benefit reversed: in the Rigveda you stay inside a vratá because that is what the world is like and because stepping outside it is what enemies do, whereas in the later practice you adopt one in order to get something. The word for the reliability of the cosmos became the word for a transaction with it.

Eleven Vows

The ascetic traditions took the same stem in a third direction, which is where its modern ethical sense comes from. The Jain monastic rule is built on five mahāvrata, great vows, with lay versions called aṇuvrata, lesser vows; the whole architecture of Jain ethics is a hierarchy of self-imposed vratas.[13] The sense here is neither cosmic order nor fasting for a favour. It is self-legislation.

Gandhi inherited that usage and extended it. The rule he set for residents of his ashrams, explained in a series of letters written from Yeravda prison in 1930, was an ekādaśa vrata, eleven vows: truth, non-violence, celibacy, non-stealing, non-possession, bread labour, control of the palate, fearlessness, equal respect for all religions, use of home-produced goods, and the removal of untouchability.[14] The first five are the Jain great vows. The sixth is a man deciding that spinning counts as a religious obligation.

Notice what sits at the head of the list. The Śatapatha’s sacrificer, taking up his vrata, undertook before anything else not to speak untruth, on the stated ground that untruth leaves a man foul within.[9] Three thousand years later the first of the eleven vratas is satya. I would not press that into a claim of unbroken transmission, because the intervening literature is vast and the resemblance may be the kind that recurs wherever people write rules for themselves. But it is a tidy illustration of what the word had become: a slot into which a tradition puts whatever it thinks a person should hold themselves to.

What the Older Sense Was For

The Rigvedic vratá is not a better idea than the modern one, and reading the hymns in the hope of recovering a purer original meaning is a mistake the texts themselves do not invite; they are perfectly happy to let a word do several jobs. What the older sense offers is a different location for the sacred. In the hymns, holiness is not in self-denial and not in intensity. It is in dependability: the fact that the sun comes up on schedule, that the rivers stay in their channels, that heaven and earth keep the same rule, and that somebody is watching to see that none of it slips. The religious emotion that goes with that is not awe at the extraordinary. It is relief at the ordinary, which is rarer in religious poetry than you would expect and is the real discovery in these two hundred verses.

It also tells you what the Rigvedic poets thought a person was. Not primarily a sinner or a seeker, but somebody with a course to keep: a carpenter who wants broken wood, a physician who wants a fracture, a priest who wants a patron with soma. The vocabulary of modern devotion asks what you have given up. The older vocabulary asks what you are keeping to, which is a question anyone can answer, including the frogs.

Read RV 2.38 at actual dusk, outdoors, and watch how much of the hymn is just a list of things stopping. Then read RV 9.112, which takes four verses to say that everybody wants different things and that this is fine. For the neighbouring words that travelled similar distances, there are companion pieces on māyā before it meant illusion, on rasa before it meant aesthetics, on tanú, the self before ātman, and on kárman when it was still plain work.