The Fallow Strip: The Rigveda's Apocrypha and the Hymn to Lakṣmī That Is Not in the Rigveda

A Hymn Nobody Can Find

On the night of Diwali, in a few million households, someone lights a lamp and recites a hymn that begins hiraṇyavarṇāṃ hariṇīṃ suvarṇarajatasrajām, “the gold-coloured one, the tawny one, garlanded in silver and gold.” It is the Śrī Sūkta, the hymn to Lakṣmī. Ask almost anyone reciting it where it comes from and the answer is immediate: the Rigveda.

Now open a printed Rigveda and look for it. It is not there. Not in Maṇḍala 5, where tradition attaches it. Not anywhere in the 1,028 hymns. Max Müller’s six volumes do not carry it in the main text. Aufrecht’s two do not. Jamison and Brereton’s three-volume translation of 2014, the standard English Rigveda, does not translate it, because the collection they were translating does not contain it.

The Śrī Sūkta is a khila. So are roughly three dozen other hymns, depending on whose count you trust, which have travelled alongside the Rigveda for something like two and a half thousand years without ever being admitted into it. They are copied in the same manuscripts. They are recited by the same families. Some of them are more widely used today than most of the hymns that made the cut. And they sit on the far side of a line that somebody, at some point, drew.

This piece is about that line: where it runs, who drew it, why eleven hymns to Indra ended up on the inside of it while a hymn to Lakṣmī ended up on the outside, and why the most basic question you can ask about the shadow collection, namely how many hymns are in it, still has no agreed answer.

The Word Means Fallow Ground

Sanskrit khila is not a bookish term. Monier-Williams gives its primary sense as “a piece of waste or uncultivated land situated between cultivated fields, desert, bare soil,” and only after that “a space not filled up, gap,” and only after that “supplement (of a book etc.).” Apte’s dictionary preserves the same order, ending with the technical meaning: “an additional hymn appended to the regular collection.”

The metaphor is agricultural and it is exact. A khila is the strip nobody ploughed. It lies between the worked fields, it belongs to the same landscape, it is walked across daily, and it is not part of the harvest. Vedic tradition chose, for its supplementary hymns, a word that says both “this is ours” and “this is not counted.”

That double message runs through the entire history of the material. The khilas were preserved with care by people who did not regard them as scripture in the full sense. They were copied by scribes who knew they were copying an appendix. They acquired ritual uses, commentaries, and eventually enormous popularity, and none of that moved them across the line.

Aside. The Rigveda’s own vocabulary of textual boundaries is thin. There is no Vedic Sanskrit word for “canon” and no ancient list titled “hymns we rejected.” What survives instead is a set of indexes, the anukramaṇīs, which say what is in. Everything they do not enumerate is out by silence rather than by verdict. Reconstructing the edge of the Rigveda therefore means reading an inventory, not a debate.

Eleven Hymns That Got In

Before the hymns outside the wall, consider the ones that were smuggled inside it. Maṇḍala 8 of the Rigveda contains 103 hymns. Hymns 8.49 through 8.59, eleven in all, form a block called the Vālakhilya, and essentially nobody who has studied the text believes they belong where they stand.

The evidence is structural rather than doctrinal. Hermann Oldenberg’s Metrische und textgeschichtliche Prolegomena of 1888 established that the Rigveda’s arrangement is not random: within a maṇḍala, hymns are grouped by deity, and within each deity group they run in order of decreasing verse count. Once you know the rule, violations of it become visible the way a missing stair is visible. The Vālakhilya block does not obey it. It sits between 8.48 and 8.60 like a folded sheet slipped into a bound book.

Editors have handled the intrusion in different ways, and the result is a citation trap that still catches readers. Griffith’s translation, still the most widely reproduced English Rigveda online, prints Maṇḍala 8 as 92 hymns and puts the Vālakhilya in a separately numbered appendix. That means Griffith’s “Book 8, Hymn 49” is not RV 8.49 at all. It is RV 8.60, a hymn to Agni. From 8.49 onward, Griffith and the standard numbering run eleven apart.

What is actually in these eleven hymns is worth a look, because the reason they were treated as marginal is not that they are obscure. Vālakhilya 1, that is RV 8.49, opens as conventionally as any Indra hymn in the collection:

To you will I sing Indra’s praise who gives good gifts as well we know; the praise of Maghavan who, rich in treasure, aids his singers with wealth thousandfold. (RV 8.49.1, after Griffith)

Nothing there marks it as an outsider. The oddity arrives later in the block, in the gift-praise verses attached to a patron named Dasyave-vṛka:

A hundred oxen white of hue are shining like the stars in heaven, so tall, they seem to prop the sky. Bamboos a hundred, a hundred dogs, a hundred skins of beasts well-tanned, a hundred tufts of balbaja grass, four hundred red-hued mares are mine. (RV 8.55.2 and 8.55.3, after Griffith)

A hundred dogs. That inventory, with its tanned hides and its bundles of coarse balbaja grass, reads less like liturgy than like a receipt. It is exactly the kind of material a redactor might file under “keep, but not in the main sequence.”

Here the honest position has to be stated plainly. The Vālakhilya’s late or intrusive status rests on the arrangement argument and on the fact that separate traditions counted them differently. It does not rest on anything the hymns themselves say. They contain no confession of lateness. A reader who wanted to argue that the redactors displaced genuinely old material for liturgical reasons rather than chronological ones would not be contradicted by the text. What can be concluded is narrower than what is often asserted: the Vālakhilya were handled as a unit, that unit sat awkwardly in the collection’s own ordering logic, and different schools resolved the awkwardness differently.

The Redaction, Drawn

flowchart TD
  A[Oral family repertoires] --> B[Family books 2 to 7]
  B --> C[Sakala redaction]
  B --> D[Baskala redaction]
  C --> E[1017 regular hymns]
  C --> F[11 Valakhilya inserted]
  D --> G[1025 regular hymns]
  D --> H[Khila appendix]
  H --> I[Sri Sukta]
  H --> J[Sivasamkalpa Sukta]
  E --> K[Sarvanukramani index]
  F --> K
  K --> L[Canon fixed at 1028]

Five Schools, One Survivor

Patañjali’s Mahābhāṣya says that the Rigvedic tradition had twenty-one branches. The Caraṇavyūha attributed to Śaunaka names five: Śākala, Bāṣkala, Āśvalāyana, Śāṅkhāyana and Māṇḍūkāyana. What we can actually read today is essentially one, the Śākala, with fragmentary knowledge of the Bāṣkala and traces of the Āśvalāyana.

This matters enormously for the khila question, because the khilas are not floating free of the schools. They are the residue of the disagreement between them. Where the Śākala tradition drew its line at 1,028, the Bāṣkala drew it elsewhere: reference works following that tradition report 1,025 regular hymns, with eight Vālakhilya distributed among the regular sequence rather than blocked together, plus a substantial appendix. The number usually given for that appendix is 98.

Michael Witzel’s study of the formation of the Vedic canon, published in Inside the Texts, Beyond the Texts in 1997, is the standard modern treatment of how these schools related to one another and to the political geography of the late Vedic period. The picture that emerges is not of a single moment of canonisation but of parallel transmissions that hardened at different times and in different places, with the Śākala version eventually achieving something close to monopoly for reasons that were institutional as much as textual.

Methods note. When a source says “the Bāṣkala recension contains 98 khilas,” it is worth asking what that sentence is based on. The Bāṣkala saṃhitā does not survive complete. Statements about its contents are reconstructions from citations, from the Caraṇavyūha, and from manuscripts that mix traditions. The figure 98 circulates widely in reference works, and it is repeated here with that provenance rather than as an independently verified count.

Kashmir, Birch Bark, a German Editor

The khilas reached modern scholarship through Kashmir. The Kashmirian tradition wrote Sanskrit in the Śāradā script on birch bark, and its Rigveda manuscripts carried the supplementary hymns after the tenth maṇḍala. Georg Bühler’s Detailed Report of a Tour in Search of Sanskrit MSS. Made in Kaśmīr, Rajputana, and Central India, published in Bombay and London in 1877, is the document that brought this material into European view. Bühler spent 1875 buying and copying manuscripts in Kashmir; he later drowned in Lake Constance in 1898, in circumstances that were never fully explained.

Thirty-one years after Bühler’s report, Isidor Scheftelowitz published Die Apokryphen des Ṛgveda (Khilāni) in Breslau in 1906, as the first volume of the series Indische Forschungen. It remains the edition. Scheftelowitz, who was born in 1875 and died in 1934, gave the khilas what the Rigveda proper had received sixty years earlier from Müller: a text, an apparatus, and a name. The title is doing real work. Apokryphen is a Christian word for a Jewish and Christian problem, and applying it to Sanskrit material imported a whole European argument about canons into a tradition that had never framed the matter that way. Scheftelowitz, a rabbi as well as an Indologist, would have known exactly what he was borrowing.

The electronic text of Scheftelowitz prepared by Muneo Tokunaga in 1995, which is what most researchers now consult, carries thirty-five khila hymns arranged in three adhyāyas: twelve, then sixteen, then seven.

Nobody Agrees How Many There Are

Here is the fact that ought to be better known. There is no settled count of the khilas.

The figures in circulation include 25, 32, 35, 86 and 98. They are not competing measurements of the same object. They are measurements of different objects, produced by editors who made different decisions about what a khila is.

Consider what has to be decided before you can count. Does a hymn that appears in one manuscript tradition and not another count once, twice or not at all? Do the Vālakhilya belong in the khila collection, given that the Śākala tradition swallowed them and the Bāṣkala tradition did not? What about the Śivasaṃkalpa Sūkta, whose verses also appear in the Vājasaneyi Saṃhitā of the Yajurveda, so that including it in a Rigvedic appendix means classifying a text by the company it keeps rather than by its content? Do variant recensions of the same khila count separately? Scheftelowitz, working from the Kashmirian material, produced one answer. The Pune edition of the Rigveda with Sāyaṇa’s commentary, edited by N. S. Sontakke and C. G. Kashikar between 1933 and 1951, produced another. Usha R. Bhise’s monograph The Khila-Sūktas of the Ṛgveda: A Study, published by the Bhandarkar Oriental Research Institute in 1995, is the fullest modern treatment and does not close the question either.

If you want the proportion the khilas represent, the arithmetic is easy and the answer depends entirely on which count you feed it:

$$\frac{35}{1028} \approx 3.4\% \qquad \frac{98}{1028} \approx 9.5\%$$

Somewhere between one hymn in thirty and one hymn in ten. That is the resolution currently available. A discipline that can date the Rigveda’s internal strata by counting syllables cannot say how many hymns are in its appendix, and the reason is not carelessness. It is that the appendix was never a single thing.

Reading the Hymn That Started This

The Śrī Sūkta is khila 2.6 in Scheftelowitz’s arrangement, attached in the manuscript tradition to the end of the fifth maṇḍala. Its opening verses, in the transmitted text, run:

Gold-coloured, tawny, garlanded in silver and gold, moon-bright, made of gold: bring me that Lakṣmī, O Jātavedas. Bring me that Lakṣmī who does not go away, in whom I may find gold, cattle, horses, men. (Ṛgveda-Khila 2.6.1 and 2.6.2, from the Sanskrit in Scheftelowitz’s edition; translation mine)

Read that against the Rigveda and the distance is immediate. Start with the name. The word lakṣmī does occur in the Rigveda proper, but it is not anybody’s name there. At RV 10.71.2, in a hymn about the making of speech, it means a mark or token of good fortune:

Where the wise fashioned speech with the mind, winnowing it as grain is winnowed with a sieve, there companions know their companionship: an auspicious lakṣmī is set upon their speech. (RV 10.71.2, translation mine, following the Sanskrit)

An auspicious sign laid on words. Not a goddess who arrives when invoked and is asked not to leave. The khila has taken a common noun for a favourable mark and turned it into a person with a biography.

The rest of the vocabulary tells the same story. The hymn’s Lakṣmī stands on a lotus and is lotus-coloured, padme sthitāṃ padmavarṇām. She is aśvapūrvāṃ rathamadhyāṃ, preceded by horses and set in the middle of a chariot, and hastināda-pramodinīm, delighting in the trumpeting of elephants. She has a sister, alakṣmī, misfortune personified as a negation, whom the reciter asks her to destroy. She is associated with the bilva tree, bilva, which is not a Rigvedic plant of any prominence.

This is why nobody serious dates the Śrī Sūkta to the Rigvedic period. The usual scholarly placement is later, with most verses assigned to roughly the Brāhmaṇa period and a few later still. David Kinsley, in Hindu Goddesses (1986), treats it as the first text in which Śrī and Lakṣmī, originally distinct conceptions, are drawn together into a single figure. Thomas Coburn’s work on the crystallisation of goddess traditions places it in the same broad process.

Where scholars disagree is on how much of the hymn is a unity. The transmitted text has overlapping strata with verses that repeat and renumber, and different manuscript traditions carry different extents. Reference treatments distinguish a first block of roughly verses 1 to 19, a second overlapping block, and a third beginning later still, which means that “the Śrī Sūkta” recited at Diwali is a selection from something that was already a compilation. What can be concluded is that the hymn is a stratified composition assembled over an extended period, that its language and imagery are post-Rigvedic, and that its manuscript history is messier than any recitation implies.

The sculptural record is a useful check on the textual one. The goddess on a lotus lustrated by elephants appears on the Bharhut railing in the second century BCE, on a coin of the Indo-Scythian ruler Azilises in the first century BCE, and on a third-century CE coin from Kausambi. The oldest securely dated occurrences of the image are not in Vedic contexts. A hymn describing that iconography in fixed epithets is unlikely to be much older than the objects.

What Happens to a Hymn After It Is Left Out

Exclusion did not diminish this material. It redirected it.

The Śrī Sūkta became one of the most recited pieces of Sanskrit in the world, its liturgical reach exceeding that of all but a handful of hymns from inside the canon. The Śivasaṃkalpa Sūkta, twenty-eight verses asking that the mind be well disposed, migrated into the Yajurveda tradition and is used to this day.

The strangest afterlife belongs to the Vālakhilya, and it is a case of a bibliographic term breeding a myth. Vālakhilya was the name of a section. In the epic and Purāṇic literature it becomes the name of a class of beings: sixty thousand ascetics the size of a thumb, who appear in the Mahābhārata in the story of Garuḍa’s birth. Indra, carrying a whole mountain of firewood to Kaśyapa’s sacrifice, sees the vālakhilyas struggling collectively with a single twig and laughs at them. They curse him, and out of that curse comes a being born to be Indra’s enemy and later his ally. A textual appendix, in other words, grew legs, shrank to thumb size, and acquired a grudge.

Aside. This is not an isolated pattern. Ka, “Who?”, the interrogative that ends each verse of a famous Rigvedic hymn, was later read as a proper name and produced a god called Who. Indian tradition has a persistent habit of turning grammatical and editorial artefacts into persons. The vālakhilyas are the same mechanism working on a table of contents.

The wider point is that the canon’s boundary was a filing decision, not a judgement about power. The material that fell outside it kept working. It acquired ritual functions, iconography, commentaries and, in the case of the Vālakhilya, a mythology. What it did not acquire was the protection of the Rigveda’s extraordinary preservation machinery. Hymns inside the line were secured by the padapāṭha and the woven recitations that turn memorisation into error correction. Hymns outside it were copied, which is a far weaker guarantee. That asymmetry is why we have variant recensions of the Śrī Sūkta and a single fixed text of every hymn beside it.

The Edge Is the Interesting Part

The number 1,028 has the feel of a fact about the world, like the height of a mountain. It is not. It is the outcome of a series of decisions made by people whose names, in most cases, we do not know: to fold eleven hymns into Maṇḍala 8 rather than leave them outside, to compile an index that enumerated what was in and thereby defined what was out, to keep copying the leftover strip because it was useful even though it was not counted.

Kātyāyana’s Sarvānukramaṇī, edited by Arthur Macdonell at the Clarendon Press in 1886, is the instrument that made the boundary operational. It lists, for every hymn, its poet, its deity and its metre. An index of that kind does not argue. It simply stops, and where it stops is the edge of the text. Everything the khilas suffered follows from not being on the list.

What the khilas offer, then, is not a lost Rigveda. Their language shows they are later, mostly considerably later, and no serious reading recovers Bronze Age material from the Śrī Sūkta. What they offer is a view of the canon from outside it, at the one place where the machinery is visible. Look at the strip of unploughed ground and you can see the shape of the field: what the redactors thought a Rigvedic hymn was, what they were prepared to lose, what they could not bring themselves to throw away, and how much later religious life would be conducted in the margin they left.

Every Diwali, that margin gets the loudest recitation of the year.