The Prop Under the Sky: What Dharma Meant Before It Meant Duty

Two slabs and the gap between them

There is a verse in the sixth maṇḍala that reads like an inspection report.

Heaven and Earth are addressed as a pair, which is normal. They are called rich in ghee, beautifully formed, milking honey, which is also normal. Then the poet says something that is not decorative at all:

dyā́vāpṛthivī́ váruṇasya dhármaṇā viṣkabhité ajáre bhū́riretasā

“Heaven and Earth, by the dhárman of Varuṇa, are propped apart, unaging, rich in seed.”

Rigveda 6.70, verse 1. My rendering, following Jamison and Brereton.[5]

Viṣkabhité. From √skabh, to prop, to strut, to jam something in place so it cannot move. It is the verb you would use for a post under a sagging beam. The two great slabs of the world are not floating in harmony; they are being held apart, under load, by something. And the something has a name: dhárman.

That word is the ancestor of dharma. Three thousand years later it will carry the weight of an entire civilisation’s moral vocabulary: caste duty, righteousness, the law, the Buddha’s teaching, the wheel at the centre of the Indian flag. But here, in what is probably its oldest surviving habitat, it is doing structural engineering. Nobody in this verse is being told how to behave. Something is being kept from falling down.

This essay is about the gap between those two senses, and about how much smaller and stranger the Rigvedic word is than the one we inherited. Three things are worth establishing up front, because each of them cuts against the way dharma is usually presented. The word is rare. It is late. And in the oldest layers of the text it almost never applies to a human being.

Methods note. The count of sixty-three is Joel Brereton’s, from his 2004 study Dhárman in the Ṛgveda, and it folds in the derivatives dharmakṛ́t (once), satyádharman (six times) and dharmavant (once).[1] Patrick Olivelle gives sixty-seven, counting slightly differently.[7] Both papers, along with Horsch’s, are collected in Olivelle’s 2009 volume, which is the single most useful thing to have on a desk while reading this.[23] My own distribution figures below come from a stem-level pass over the electronic Saṃhitā text of Aufrecht’s edition, excluding the separate word vidhárman; that pass returns sixty-eight tokens, which is the same order and a useful independent check but not a rival number.[3][12] The three counts differ because sandhi hides some forms inside the preceding word and because scholars disagree about which compounds belong. None of this changes the shape of the distribution, which is the part that matters.

A noun built out of a verb for holding

Start with the machinery. Dhárman is a neuter -man stem formed on the root √dhṛ (Sanskrit: धृ), “to hold, to bear, to keep in place, to sustain”. The -man suffix in Vedic typically makes either the instrument of an action or its result: the thing that does the holding, or the arrangement that holding produces. So dhárman is, quite literally, “a hold”, “a support”, “a prop”, “the way something is held up”.

The Rigvedic poets knew this and played with it. In RV 1.22.18, Viṣṇu takes his three world-spanning strides:

trī́ṇi padā́ ví cakrame víṣṇur gopā́ ádābhyaḥ áto dhármāṇi dhāráyan

“Viṣṇu strode out three steps, the herdsman who cannot be deceived; from there, holding fast the dhármans.”

Rigveda 1.22, verse 18. My rendering, following Jamison and Brereton.[5]

Dhármāṇi dhāráyan. The noun and the participle are the same root, placed side by side. English cannot reproduce it without sounding clumsy: he holds the holds, he supports the supports. The figure is not decoration. It tells you that the poet still heard the verb inside the noun, the way an English speaker hears “build” inside “building”.

The Indo-European background confirms the reading, and it produces one of the more satisfying etymological coincidences in the ancient world.

Darius the Great and the word dharma are built from the same root.[2] The Achaemenid king’s name is a claim about load-bearing: he is the one who holds the good in place. Whatever else has changed in three millennia, that is still roughly what a functioning state is for.

Aside. Vedic also distinguishes the neuter dhárman from a rarer masculine dharmán, “the holder, the sustainer”, by accent alone; Grassmann’s 1873 dictionary already separates the two entries.[11] Brereton counts four instances of the masculine.[1] The distinction is invisible in the unaccented editions most readers meet, which is one small reason the word’s early history got flattened. If you are working from a text without the Vedic accents, you cannot tell the prop from the propper.

Who does the holding, and what gets held

Collect the passages and a pattern appears immediately. In the Rigveda, dhárman is almost always something a god has, or does, or is the guarantor of. Humans have vratá, commitments; they perform kárman, work (a story told in detail in Karma Was a Verb First). What they mostly do with dhárman is stay inside it.

Two of these deserve a closer look before we go on.

RV 10.170.2 describes the sun as dhárman diváḥ dharúṇe satyám árpitam, “fixed in truth, in the support, in the dhárman of heaven”. Three words from the same root in one line, plus satyá, “the real”. The sun is not merely bright. It is seated, the way a jewel is seated in a setting or an axle in its bearing. If you want a single mental image for the Rigvedic dhárman, use that one: a socket cut to hold something so precisely that it cannot come loose. Macdonell’s old handbook of Vedic mythology, still the quickest way into the deities involved, collects the parallel passages in which Viṣṇu’s strides and Savitṛ’s arms do comparable work.[9]

And RV 4.53.3 says of Savitṛ, the god who impels, that he raises his call svā́ya dhármaṇe, “for his own dhárman”. Not for his duty in the moralised sense. For the specific piece of cosmic work that is his to do, the way a keystone’s dhárman is holding the arch. Savitṛ’s other job, the impelling one, is the subject of The Impeller and the Gāyatrī.

Notice what is missing from that whole table. There is no verse in which a Rigvedic poet tells a warrior, a wife, a merchant or a priest what his or her dhárman is. The vocabulary of personal obligation that later Sanskrit will build on this word (svadharma, varṇāśramadharma, the entire Dharmaśāstra library) has no Rigvedic foothold. The word exists. The idea attached to it does not.

The distribution is the argument

Here is the part that a word-count cannot fake. If you plot where dhárman occurs, it is concentrated in exactly the places a philologist would predict for a term that is gaining ground rather than fading.

The Rigveda is not a single document. Its ten maṇḍalas stratify: books 2 through 7, the “family books”, are generally taken as the oldest core; books 1 and 8 are later collections; book 9, the Soma cycle, and book 10 are later still, with book 10 containing the collection’s most obviously young material, including the cosmological hymns and the funeral liturgy. The stratification goes back to Oldenberg and has held up through a century of refinement.[10][19][22] Normalising occurrences against verse count gives a density, which is the only fair comparison given how differently sized the books are:

$$d_M \;=\; \frac{1000 \, n_M}{v_M}$$

where $n_M$ is the token count in maṇḍala $M$ and $v_M$ its number of verses.

Maṇḍala 7 is the largest of the family books and one of the most theologically ambitious in the collection. It contains the great Varuṇa hymns, the Vasiṣṭha material, the battle of the ten kings. In 843 verses it uses dhárman once.

Maṇḍala 9, the Soma book, uses it fourteen times in 1,108 verses. Maṇḍala 10 uses it twenty times. Between them, the two youngest books hold half of every occurrence in the Rigveda, at roughly three times the density of the old core.

Books 3 and 5 are the interesting outliers among the family books, and they are outliers for a reason: book 3 is the Viśvāmitra Agni collection and book 5 is thick with Mitra and Varuṇa, which are precisely the two theological neighbourhoods where the word lives. That is a pattern, not noise.

Methods note. With totals this small, individual densities are statistically fragile. Moving three tokens between books would visibly change the middle of that table. What is not fragile is the aggregate contrast, 18 tokens in nearly 4,000 old verses against 34 in fewer than 3,000 late ones, and the single occurrence in the whole of Maṇḍala 7. My counts run over the sacred-texts romanisation of the Saṃhitā and will slightly undercount forms swallowed by sandhi; treat them as lower bounds and reproduce them yourself if you doubt them. Brereton’s chart of the same distribution, built on a properly edited text, reaches the same conclusion by a better road.[1]

The philological upshot is straightforward. Dhárman is not a primitive of Vedic religion. It is a term that grew, and it grew fastest in the layers where the Vedic poets became most self-conscious about ritual as a technology for holding the world together.

Three words for order, and dharma is the junior one

Modern writing on Hinduism tends to treat ṛtá and dharma as the same idea wearing different clothes, the second quietly replacing the first. The Rigveda does not support that. It uses all three of its order-words in a single verse, and it uses them for different things:

dhármaṇā mitrāvaruṇā vipaścitā vratā́ rakṣethe asurasya māyáyā ṛténa víśvaṃ bhúvanaṃ ví rājathaḥ

“By dhárman, O Mitra and Varuṇa, you two who perceive the inspired word, you guard the vratás through the lordly power of the asura; by ṛtá you rule over the whole world.”

Rigveda 5.63, verse 7. My rendering, following Jamison and Brereton.[5]

Read the syntax rather than the theology. Dhárman is in the instrumental: it is the means. Vratá is the object: the commitments are what get guarded. Ṛtá is the instrumental of the ruling: it is what makes their sovereignty legitimate. The three are not synonyms stacked for emphasis. They are a subject, a tool and a warrant.

Vratá is the god’s standing order, the thing Mitra and Varuṇa above all are said to keep, and the older literature on it runs from Macdonell and Keith’s Vedic Index through Paul Thieme’s reconstruction of the Mitra complex.[13][21] Ṛtá outnumbers dhárman by roughly six to one and saturates the oldest hymns; it has been called the single concept that pervades Rigvedic thought, and Heinrich Lüders devoted two posthumous volumes to it.[8] Dhárman is the specialist term. Ṛtá is what is true. Dhárman is what is holding it up. The relationship between them is the relationship between a fact and a bracket, and the Varuṇa material explores it in more depth than any other part of the collection (see The Watchful God).

graph TD
    R["Root dhṛ: to hold"] --> A["dhárman: a prop"]
    R --> B["dharúṇa: a base"]
    R --> C["dhartṛ: the holder"]
    R --> D["dhāráyati: holds fast"]
    A --> E["Physical: props the sky"]
    A --> F["Ritual: the first rites"]
    A --> G["Royal: seats a king"]
    F --> H["Precedent"]
    H --> I["Norm"]
    G --> I
    I --> J["Classical dharma"]

“Those were the first dharmans”

The hinge between prop and precept is a single formulaic line, and it survives in three places.

The famous one closes the Puruṣa Sūkta, the hymn in which the gods dismember a cosmic man to make the world (treated at length in The Cosmic Sacrifice):

yajñéna yajñám ayajanta devā́s tā́ni dhármāṇi prathamā́ny āsan

“With the sacrifice the gods sacrificed to the sacrifice. Those were the first dhármans.”

Rigveda 10.90, verse 16. My rendering, following Jamison and Brereton.[5]

The same line closes RV 1.164.50, and the Riddle Hymn also gives us a stranger cousin at verse 43 of the same hymn, where the poet reports seeing smoke from a dung fire, and men cooking a speckled ox, and then says it again: tā́ni dhármāṇi prathamā́ny āsan. Those were the first dhármans.

Something has shifted here, and it is worth being precise about what. In RV 6.70.1 a dhárman is a physical support. In RV 10.90.16 it is an event that establishes how things are done. Not a rule, not yet; a first instance. The logic is the logic of common law rather than statute. The gods did it this way, once, at the beginning; the doing laid down a foundation; and every subsequent performance is standing on that foundation the way a house stands on a footing.

That is the missing link. Get from “prop” to “precedent” and you are most of the way to “norm”, because a precedent is a prop that holds up a practice instead of a sky. Paul Horsch’s 1967 study, the paper that opened the modern discussion, traced exactly this trajectory: dharma begins inside a creation myth and ends as world law, and the ritual passages are where the transfer happens.[6] Brereton’s more cautious analysis reaches a compatible place by insisting on the concrete meaning throughout: dhárman describes the ritual as the foundation of the world, which is what allows a foundation-word to become a normative one without ever quite stopping being a foundation-word.[1]

The two verses where it starts to bite

Two passages show the ethical charge arriving, and neither is a sermon.

The first is a confession. In the great Varuṇa hymn at the end of Maṇḍala 7, the poet says:

ácittī yát táva dhármā yuyopimá mā́ nas tásmād énaso deva rīriṣaḥ

“If through thoughtlessness we have disturbed your dhármans, do not injure us, god, for that offence.”

Rigveda 7.89, verse 5. My rendering, following Jamison and Brereton.[5]

Yuyopimá, “we have disturbed, displaced, thrown out of true”. You can dislodge a dhárman. Not break a law: unseat a support. And the word the poet uses for what happens next is énas, the Rigveda’s heaviest term for guilt, the thing Varuṇa’s nooses are for. The anxiety in that verse is not the anxiety of a man who has broken a rule. It is the anxiety of a man who has felt something move under his feet.

The second is the strangest place in the entire collection to find the word, and it is my favourite. In the Gambler’s Lament, a ruined dice player watches the dice themselves:

tripañcāśáḥ krīḷati vrā́ta eṣāṃ devá iva savitā́ satyádharmā ugrásya cin mányave nā́ namante rā́jā cid ebhyo náma ít kṛṇoti

“Their troop of fifty-three plays on, like the god Savitṛ whose dhárman is real. They do not bow to the anger of even the powerful; even a king makes obeisance to them.”

Rigveda 10.34, verse 8. My rendering, following Jamison and Brereton.[5]

Satyádharman, “of true foundation”, is an epithet of the sun god. The gambler applies it to a handful of nuts on a board. He has lost his wife, his house and his brothers’ respect, and what he says about the dice is that they behave like the sunrise: they are indifferent, they are exact, and a king bows to them because rank buys you nothing from a thing that is simply holding to its own nature. The hymn is discussed in full in The Gambler’s Lament; Wendy Doniger’s anthology carries a readable version if you want the whole thing in one sitting.[20]

There is no moral instruction anywhere in that verse. There does not need to be. Anyone who has ever tried to argue with a load-bearing wall knows what it is describing.

Seven words for one word

Here is a small experiment anyone can run. Take Ralph Griffith’s 1896 translation, still the most widely reproduced English Rigveda on the internet, and look at how he renders dhárman in eight consecutive encounters.[4]

Decree, law, law, law, way, ordinance, custom, merit. One Sanskrit word, seven English ones, one translator, one book.

The last line is the one to sit with. RV 10.16.3 is from the cremation liturgy: the dead man’s eye goes to the sun, his breath to the wind, and he himself goes to heaven and earth dhármaṇā, by the dhárman, which is to say by the established arrangement, along the channel that has been built for exactly this. Griffith renders it “as thy merit is”, which imports the entire later doctrine of moral desert into a verse that says nothing of the kind. It is a small mistranslation and a perfectly understandable one, because by 1896 everyone knew what dharma meant. That is precisely the problem. The funeral liturgy is discussed on its own terms in The Road the Dead Take.

[!NOTE] Griffith is not being careless. He is doing what every translator of an ancient text does, which is to reach for the target-language word that carries the same weight. The trouble is that English has no word that means “the structural arrangement by which a thing is held in position”, so every choice he makes is a small act of interpretation, and the accumulated choices produce a Rigveda in which dharma already means what it will only come to mean centuries later.

The long silence, and the word that came back

If dharma were the master concept of Indian religion from the beginning, you would expect its frequency to climb steadily after the Rigveda. It does the opposite.

Patrick Olivelle worked through the middle and late Vedic corpus and found the term going quiet. The Atharvaveda has around thirteen occurrences. The four Yajurveda Saṃhitās between them yield roughly twenty-two or twenty-three passages. Three major Brāhmaṇas, enormous texts, produce eleven independent occurrences combined. His conclusion is blunt: dharma “was at best a marginal term and concept within the vocabulary of these texts, and it did not play a central role in the religious world depicted in them.”[7]

Where it does survive, it survives in one specialised setting: the royal consecration. The rājasūya formulae make dharma the thing a king is anointed into, the thing that makes a sovereign’s power legitimate rather than merely large. The load-bearing sense has migrated from the sky to the throne.

Then something external happens. The Buddha’s first sermon sets a wheel in motion, the dharmacakra, borrowing the imagery of an emperor’s conquering chariot wheel and turning it into a metaphor for teaching. Two and a half centuries later an actual emperor picks the word up and carves it into rock and pillar across the subcontinent, something on the order of 111 times across his edicts, glossing it in his Greek inscription at Kandahar as eusebeia, piety.[7][18] Olivelle’s argument, which has been contested but not dislodged, is that Brahmanical writers then took a term that had gone marginal in their own tradition and re-semantised it under this pressure, and that the classical Dharmaśāstra apparatus is downstream of that.[7][14][15]

timeline
    title The Career of One Word
    section Old family books (RV 2-7)
        Prop under the sky : 18 occurrences in ~4,000 verses
    section Late Rigveda (RV 9-10)
        The first rites are dharmans : 34 occurrences in ~2,900 verses
    section Brahmanas (~900-600 BCE)
        Marginal term : Survives mainly in royal consecration
    section Early Buddhism (~450 BCE onward)
        Dharmacakra : Teaching framed as a sovereign wheel
    section Ashoka (~250 BCE)
        Dhamma carved in stone : Roughly 111 uses across the edicts
    section Dharmasastra (~200 BCE onward)
        Total normative order : Duty, law, caste, stage of life

By the time of the Bṛhadāraṇyaka Upaniṣad, the reversal is complete. In a famous passage the text says that the ruling power created dharma, and then that dharma is the ruling power of the ruling power, kṣatrásya kṣatrám, so that there is nothing higher than dharma; a weak man can prevail over a strong one by dharma, as by a king.[17] The word that once described the strut under the sky now describes the thing a sovereign cannot overrule.

And it never lost the metaphor. When the Mahābhārata wants to define the term, it does not offer a list of duties. It offers an etymology: dhāraṇād dharma ity āhuḥ, “they call it dharma because it holds”, and the line goes on to say that dharma is what holds the people together.[14] Twenty-five hundred years after a poet in the Punjab looked up and asked what was keeping the sky off the ground, the answer is still a verb about holding.

Reading it forward

The Rigvedic evidence will not support the claim that the Vedic poets had a theory of moral duty. It supports something narrower and, I think, more interesting.

Three observations to take away. First, the Rigveda’s ethical vocabulary is architectural before it is judicial: things are held, propped, seated, kept from slipping, and the language of command arrives late and mostly through the ritual. Second, the shift from prop to precept runs through the idea of precedent, through the gods doing something for the first time and thereby laying a footing, which is why the transition happens in the youngest hymns and in Maṇḍala 9 rather than in the old family books. Third, the modern reader who wants to know what dharma “really means” is asking a question the Rigveda cannot answer, because in the Rigveda the word has not yet acquired the meaning the question presupposes. Wilhelm Halbfass made the sharper version of this point: the search for a single English equivalent for dharma is itself an artefact of the European encounter with India, and it has been distorting translations since the eighteenth century.[16]

What it does offer is a different default question. Not “what am I supposed to do”, which is where every later tradition takes the word, but the one that comes first and that the later question quietly depends on: what is holding this up? Ask it of a bridge and you get engineering. Ask it of a household, a market, a court, a language, a friendship, and you get something the Vedic poets would have recognised immediately, which is that the answer is never nothing, that the supports are usually invisible until they move, and that a civilisation is a set of props whose names most of its inhabitants have forgotten.

Read RV 6.70 somewhere with a good view of the horizon, where the two slabs meet. Then read RV 10.34.8, the dice that a king bows to. The distance between those two verses is the whole history of the word, and both of them are about the same thing: what does not negotiate, and what happens to you when you lean on it wrongly.

For related reading, see The Watchful God on ṛtá and cosmic guilt, Karma Was a Verb First on the parallel career of kárman, and The Sky Does Not Fall on the Vedic mechanics of a universe that stays up.

[!TIP] If you read only one piece of secondary literature on this, make it Brereton’s 2004 article. It is technical, but it is the study that established the concrete “foundation” reading against a century of translators who reached for “law”, and its verse-by-verse tables let you check every claim in this essay against the Sanskrit.