Māyā Before It Meant Illusion

Three Words in Ten Thousand Verses

Ralph Griffith translated the entire Rigveda into English in 1896, all ten books and every one of its 10,000-odd verses. In that whole translation the word “illusion” appears three times.

That is a strange number. Māyā (माया) is, for most readers today, the single most famous idea to come out of Sanskrit: the veil, the cosmic mirage, the world as a shimmer over an underlying unity. It is the word that reached English by way of Schopenhauer and stayed. And the word māyā is not rare in the Rigveda. It and its relatives turn up in roughly eighty verses, spread across nine of the ten books. If māyā meant illusion, Griffith would have written “illusion” eighty times. He wrote it three.

What he wrote instead was: magic, art, power, wile, charm, device, guile, skill, wisdom, enchantment, sorcery, deceit, magician, enchanter. Fourteen different English words for one Sanskrit noun, chosen more or less afresh each time, because the thing the word was doing in each verse kept refusing to sit still.

Here is one of the verses where he did reach for “illusion,” and it is worth reading closely, because it is the single verse in the Rigveda that ended up carrying the word into philosophy:

rūpaṃ-rūpaṃ pratirūpo babhūva tad asya rūpaṃ praticakṣaṇāya

indro māyābhiḥ pururūpa īyate yuktā hy asya harayaḥ śatā daśa

“Form after form, he has become the counter-form of it; that form of his is for being looked at. Indra moves about many-formed by his māyās, for his bay horses are yoked, ten hundreds of them.”

Rigveda 6.47.18, with diacritics standardised; compare Griffith (1896), who renders the third line “Indra moves multiform by his illusions.”

Read it without the later tradition in your ear. Indra is not producing a false appearance of something that is not there. He is producing an actual second thing, a pratirūpa, a counter-form, and the poet says explicitly what it is for: praticakṣaṇāya, “to be looked at.” The chariot horses in the fourth line are not decorative. They are the reason the verse is in this hymn at all. The god is arriving, in force, in many shapes at once.

The argument of this piece is narrow and, I think, checkable. In the Rigveda, māyā is a technical word for the power to produce form, to lay something out, to make a thing that was not there before. It carries no moral charge of its own. Gods have it. Their enemies have it. Poets have it, and can have a bad one. The meaning “illusion” is not lurking in the Vedic usage waiting to be discovered; it is the outcome of a later and quite traceable decision about what forms are worth.

Methods note. The counts above are my own, made by scanning the romanised Saṃhitā text verse by verse for the stem and resolving sandhi-joined tokens by hand (so that māyā́bhir indra at 1.11.7 is counted, and ṛcīṣamā́ya ádhrigo at 1.61.1, which merely looks like it contains the word, is not). Round them generously. Grassmann’s 1873 Wörterbuch zum Rig-Veda remains the reference for anyone who wants the forms itemised properly. [1]

What the Word Actually Does

The fastest way to feel the Vedic sense is to collect the jobs the word is given and look at them together.

Nothing on that list is a hallucination. The sky stays up. The tree opens. The man gets out. The rain falls, and the poet who watched it fall wrote the verse. When the Aśvins split the trunk to free Saptavadhri, the hymn says they did it māyābhiḥ, by their māyās, in exactly the way another hymn would say a smith did something with his tools.

The last line of that table is the one I find hardest to forget. In RV 10.71, a hymn about the making and judging of speech, there is a portrait of a bad poet:

adhenvā carati māyayaiṣa vācaṃ śuśruvān aphalām apuṣpām

“This one goes about with a milkless māyā: he has heard a speech that bears no fruit and no flower.”

Rigveda 10.71.5, my rendering; compare Griffith (1896), “He wanders on in profitless illusion.”

A milkless māyā. The metaphor is agricultural and the judgement is professional: this man’s technique is barren. You cannot get that sentence out of “illusion.” You get it easily out of a word that means something like the craft by which a thing is brought into being, because craft is precisely the kind of thing that can be good or bad, productive or sterile. The Rigvedic poets thought about their own composition in the vocabulary of joinery and wheelwrighting, as the hymns collected in The Poet as Carpenter show; māyā sits comfortably in that same workshop.

The Verb Hiding Inside the Noun

Now the philology, which in this case is unusually rewarding, because the Rigvedic poets seem to have heard an etymology in the word and to have played on it.

Sanskrit has a root mā- (मा), “to measure, to mete out, to lay out, and by extension to construct.” It gives māna, “a measure, a measuring rod,” and mātrā, “a measure,” the word that stands behind Latin materia and so behind English “matter.” The old and obvious guess is that māyā́ is built from this root: the noun of the measuring-out.

The guess is not provable from morphology alone, and etymological dictionaries have been careful about it. But the Rigveda itself keeps putting the two next to each other, which is a different kind of evidence and in some ways a better one. In RV 5.85, a hymn to Varuṇa, the poet announces a great māyā and then says what it consists of:

imām ū ṣv asurasya śrutasya mahīṃ māyāṃ varuṇasya pra vocam

māneneva tasthivān antarikṣe vi yo mame pṛthivīṃ sūryeṇa

“Let me now proclaim this great māyā of the famed lord Varuṇa: that, standing in the midspace, he measured out the earth with the sun, as with a measuring rod.”

Rigveda 5.85.5, my rendering; compare Griffith (1896), “this mighty deed of magic.”

Three words from the same root in two lines: māyā, māna (“measuring rod”), vi mame (“he measured out”). Whatever a modern etymologist concludes, the poet of this hymn plainly thought māyā was what you did with a māna. And Varuṇa’s great feat here is surveying. He stands in the middle air and uses the sun as a rod to lay out the earth. That is not a conjuring trick. That is geodesy, performed by the god who is elsewhere the keeper of measured order; the same god whose watchfulness is the subject of Varuṇa, the Watchful God.

The pun is not a one-off. Three separate passages make māyín- the subject of the verb mimā-/mamire, “they measured out, they constructed”:

The first of these is the clearest statement of the Vedic idea that I know. Its full sense is that the māyins, “clothing themselves in one lordly appearance after another” (anyad-anyad asuryaṃ vasānā), measured out a form in him. Māyā produces rūpa, form. That is its output. The relation of a māyin to a form is the relation of a carpenter to a chariot.

Jan Gonda devoted the fourth of his Four Studies in the Language of the Veda (1959) to exactly this question and arrived at a definition worth keeping in front of you while reading any of these verses: māyā is wisdom and power that enables its possessor to create, devise, contrive, effect, or do something. [2] Teun Goudriaan’s later book-length study reached compatible ground across a much wider body of texts. [3] Neither scholar found the Vedic word meaning “illusion,” and neither had to strain to avoid it.

Aside. The strongest counter-consideration is that an etymology can be a poetic reconstruction rather than a historical fact. Vedic poets loved to build a verse around a sound-echo, and a poet writing māyā beside mamé may be doing what the Nirukta tradition would later do systematically: proposing a derivation rather than reporting one. So take the mā- connection as well motivated but not closed. What the pun does establish beyond argument is how the composers themselves understood the word, which for reading the hymns is the more useful fact.

Both Sides Have It

Here is where the Vedic usage becomes genuinely strange to a modern ear.

The single most common thing that happens to māyā in the Rigveda is that Indra destroys somebody else’s. Śuṣṇa has māyās. Pipru has them, and is called ahimāya, “having the māyā of a snake.” Vṛtra is a māyāvin. The Dānava has them. The Dāsas and Dasyus have them. Namuci is a māyin. And the formula the poets reach for again and again is not that Indra sees through them or exposes them as unreal. It is that he beats them with his own:

māyābhir indra māyinaṃ tvaṃ śuṣṇam avātiraḥ

“By your māyās, Indra, you brought down the māyin Śuṣṇa.”

Rigveda 1.11.7, my rendering; Griffith (1896), “The wily Śuṣṇa, Indra, thou o’erthrewest with thy wondrous powers.”

The construction is a chiasmus of the same word against itself, and the poets clearly enjoyed it, because they use it repeatedly.

Notice what that last line does. Being a māyin is not a slur; it is something two of the most benevolent gods in the pantheon achieve, on the day they win Sūrya’s daughter. And in a praise list at RV 6.48.14, a poet piles up comparisons for Indra: skilled like Indra, māyin like Varuṇa, delightful like Aryaman, and so on. To be called a māyin there is to be told you are like the sovereign of the cosmos.

The decisive grammatical evidence, though, is a modifier. Twice the Rigveda speaks of adevī māyā, “godless māyā,” and prays for it to be overcome: Agni “overcomes the godless, ill-going māyās” at RV 5.2.9, and at RV 7.1.10 the poet asks that his men prevail over “all the godless māyās.” An adjective like that is only worth adding if the noun does not already carry the charge. Nobody says “evil murder.” The Rigvedic poets needed to specify godless māyā precisely because māyā by itself was neutral equipment.

One more datum settles it. In RV 10.124, the extraordinary hymn in which Agni, Soma and Varuṇa change sides and abandon the Asuras for the gods, the defeat of the losing party is described this way:

nirmāyā u tye asurā abhūvan

“Those Asuras have become without māyā.”

Rigveda 10.124.5, my rendering; Griffith (1896), “These Asuras have lost their powers of magic.”

Losing is being stripped of your māyā. If māyā were deception, losing would mean having your deception uncovered, and the hymn would say so. Instead the word behaves like a capacity, a stock of power that can be possessed, exercised, and confiscated.

Aside. It is worth being clear that Vedic Sanskrit has perfectly good words for malicious magic, and māyā is not one of them. The Rigveda’s sorcery vocabulary is yātú- and yātudhā́na- (the two occur several dozen times between them) and kṛtyā́, the fashioned spell. Those are the words of the curse-and-counter-curse hymns of Mandala 7, and māyā is not among them. Māyā keeps different company: it appears beside rūpa (form), varpas (shape), dhī (insight), and the verbs of measuring and constructing.

The Day the Sun Went Out

If you want a single episode that shows how physical the Rigvedic idea is, take the eclipse.

In RV 5.40 the demon Svarbhānu strikes the sun with darkness, and the sage Atri restores it. The hymn describes the darkness itself as Svarbhānu’s māyās, spreading out below the sky, and describes Indra striking them down and Atri finding the hidden sun “by the fourth formulation.” The event is public, dateable, and visible to everyone standing in the field. It is not a private error of perception that a wiser observer would have avoided.

That is the point I would put to anyone who has met māyā first through Vedanta. A demon’s māyā in the Rigveda is not a false belief in the observer’s mind. It is a change in the world, which is why the response to it is not insight but counter-technique: another formulation, another power, another māyā. The astronomical afterlife of this particular hymn, and the arguments about whether it can be used to date anything, are a separate story told in The Sky Does Not Fall.

What the Translators Did

Now the reception, which is where the modern misunderstanding was manufactured.

Griffith’s fourteen words are not evidence that he was careless. They are evidence that he was working without a theory, taking each passage on its merits and reaching for whichever English noun made the local sense come out. That is honest translating, and the scatter of his choices is a fair picture of the Sanskrit.

The interesting entry is the tenth. Griffith writes “illusions” at 6.47.18 and essentially nowhere else, and H. H. Wilson, translating half a century earlier under the guidance of Sāyaṇa’s fourteenth-century commentary, wrote “multiform by his illusion” at the same spot. Two translators, working from different bases, both switch registers at the one verse that the later tradition had already claimed.

They switch because they knew the verse from somewhere else. It had left the Rigveda long before either of them opened it.

Aside. Two Rigvedic passages do lean toward something like deception, and honesty requires flagging them. At RV 10.54.2 a poet tells Indra that the battles people talk about were māyā of his, since he never actually found an opponent, which can be read as “those fights were a display” or, more strongly, “they were not real.” And the milkless māyā of 10.71.5 shades toward empty pretence. Both belong to the latest layer of the collection. Neither requires the sense “illusion” (a staged display is still a made thing, and a barren technique is still a technique), but both show the semantic door beginning to open, in the same books that produced the Rigveda’s most sceptical hymns, including Nāsadīya.

The Verse That Left Home

Follow RV 6.47.18 forward and you can watch a word change its meaning without changing a syllable.

The verse is quoted, entire, at Bṛhadāraṇyaka Upaniṣad 2.5.19, in a passage arguing that a single reality appears as many. In that setting indro māyābhiḥ pururūpa īyate stops being a description of a god arriving in force and becomes a proof text: the one appears as the many by māyā. [4] Śaṅkara, commenting on the Upaniṣad several centuries later, reads the māyās as the superimpositions of ignorance. The Sanskrit is untouched. The frame around it has been replaced.

The decisive step is in the Śvetāśvatara Upaniṣad, and what makes it beautiful is that it keeps the Rigvedic grammar exactly:

māyāṃ tu prakṛtiṃ vidyān māyinaṃ tu maheśvaram

“Know māyā to be primal nature, and the māyin to be the Great Lord.”

Śvetāśvatara Upaniṣad 4.10; translation after Olivelle (1998). [5]

The pair māyā / māyin is the Rigveda’s own pair, the one Indra and Śuṣṇa fight with. The Upaniṣad has not invented new vocabulary. It has taken the old wielder-and-power relation and identified the power with prakṛti, the material world itself. Once you say that, the world becomes a thing the Lord makes by māyā, and only a small further move makes the world a thing to be seen through. The Bhagavad Gītā takes that move at 7.14, where the god’s māyā is duratyayā, “hard to cross over”: something with a far shore.

graph TD
    A["Rigveda: māyā = form-making power"] --> B["Late RV 10.54, 10.71: display, empty craft"]
    A --> C["RV 6.47.18 quoted at BAU 2.5.19"]
    C --> D["Shvetashvatara 4.10: māyā = prakriti"]
    D --> E["Gita 7.14: māyā hard to cross"]
    E --> F["Shankara: māyā = superimposed ignorance"]
    F --> G["Anquetil-Duperron, Oupnek'hat 1801"]
    G --> H["Schopenhauer: Schleier der Maja"]
    H --> I["English 'maya' = illusion"]

The last three boxes are how the word reached us. Schopenhauer read the Upaniṣads in Anquetil-Duperron’s Latin Oupnek’hat of 1801 to 1802, a translation made from a Persian version prepared in Dārā Shukoh’s circle and thoroughly steeped in the translator’s own Neoplatonism. Out of that reading came der Schleier der Maja, the veil of Māyā, which in The World as Will and Representation names the principle of individuation that makes one will look like many separate things. [6] [7] Schopenhauer is where English-language readers met the word, and the meaning they met was the meaning of the eighth-century commentator, filtered through a seventeenth-century Persian court and an eighteenth-century Latinist.

Nothing in that chain involves anyone misreading anything: each stage is a competent reading of the stage before. It is simply that the far end is nineteen centuries and four languages away from the near end, and the word we inherited comes from the far end.

Why the Meaning Moved

The usual way of describing what happened is that māyā “degenerated,” from creative power to mere illusion. I think that gets the causation backwards, and the Rigvedic evidence suggests a better account.

The word did not change. The valuation of rūpa changed.

In the Rigveda, having a form is what it is to be real and effective. Indra takes form after form because arriving in force requires arriving as something in particular. The gods measure out forms the way an artisan measures out a chariot. Form is the achievement, not the problem. And so the power that produces form is straightforwardly a good power to have, and the only question worth asking about a given instance of it is whose side it is on. That is why the poets had to say adevī māyā when they meant the bad kind.

In the philosophy that develops from the middle Upaniṣads onward, form becomes the problem. If what is ultimately real is one and without distinguishing marks, then the production of many forms is precisely what stands between the seeker and the real. The power that makes forms has not altered its job description. It has been reassigned from the credit column to the debit column. Māyā still means what it always meant: the capacity to bring appearances into being. It is now being run in a system where appearances are what you are trying to get past.

Once you see the change that way, some otherwise puzzling continuities snap into focus. The Śvetāśvatara keeps the māyin, because the power still has a wielder and the wielder is still worth praising. The Gītā calls māyā divine, daivī, in the same breath as calling it hard to cross. Śaṅkara’s whole difficulty in specifying the ontological status of māyā, the problem Paul Hacker traced through his terminology in painstaking detail, is the difficulty of a philosopher inheriting a sturdy old word for real efficacy and needing it to do a job that requires the thing not to be quite real. [8]

Reading It Back

Take a single afternoon and read the māyā verses in sequence, ignoring the translations’ vocabulary and watching only what happens. Start with RV 5.85, where Varuṇa surveys the earth with the sun for a measuring rod. Then RV 5.63, where Mitra and Varuṇa make the sky rain and the Maruts dress themselves in cloud. Then RV 3.38, where the māyins measure out a form and put on one lordly appearance after another. Then RV 1.11 and RV 10.147, where the same power is turned against itself in battle. Finish with RV 6.47.18, and notice that after the other five it reads as the least mysterious verse of the set.

What you are looking at, I think, is a culture that had a single word for the thing an engineer does, the thing a poet does, the thing a god does, and the thing an enemy does, and did not feel the need to split it. Its neutrality is not vagueness. It is a specific position: that the ability to make forms is one ability, morally undetermined, and that everything interesting depends on who holds it and to what end. The Rigveda applies the same logic to ṛta, to dakṣiṇā, to the abstractions catalogued in Manyu, Śraddhā and the Rigveda’s Abstract Deities. Powers are powers. The question is always whose.

That is not a lesser idea than the one that replaced it. It is a different one, older, and mostly unread. The chain that runs from a Punjab poet’s word for measuring out the earth to a German philosopher’s veil is one of the longest continuous arguments in the history of thought, and it is worth knowing that the first person in it was talking about a surveyor.

For how these texts relate to one another, see The Vedas, Upaniṣads and Bhagavad Gītā; for the reputation this word did so much to build, Indra’s 250 Hymns; and for the Rigveda’s own vocabulary of hiding, which is genuinely about concealment in a way that māyā is not, What the Rigveda Hides.

Appendix: Distribution Across the Collection

One pattern in the counts is worth recording, because it cuts against an easy assumption.

The word is not concentrated in the late books. It is spread through the family books, with heavy use in V and VI, which are among the older strata. If “illusion” were a Vedic meaning at all we would expect it to cluster at the end, in the books that produced the speculative hymns. It does not cluster anywhere. Mandala IX, the nine-hundred-odd verses to Soma being pressed and filtered, uses the word once in total, which tells you something about what kind of word it was: a word for what agents do to the world, not for what a ritual substance does to a mind.