The pith of metre, of treatise, of mantra — and the letters themselves

Sāra is one of those Sanskrit words whose literal sense keeps its metaphor honest. It is the heartwood of a tree, the hard core left when the soft outer wood is stripped; from there it becomes the pith, the best part, the strength of anything. A tradition that calls something the sāra of a body of literature is not paying it a vague compliment but making a claim about what remains when everything removable is removed.

Chandas, the first of the three, is the Vedic word for metre and by extension for a metrical hymn, and the study of it is one of the six auxiliary disciplines a Vedic student took up alongside phonetics, grammar, etymology, ritual and astronomy. The principal metres — gāyatrī of twenty-four syllables, then uṣṇih, anuṣṭubh, bṛhatī, paṅkti, triṣṭubh and jagatī, each adding four — are not treated in the ritual literature as containers for content but as powers in their own right, associated with particular gods, particular worlds and particular effects. The ritual texts derive the word from a root meaning to cover, and explain that the gods covered themselves with the metres out of fear of death. Whatever one makes of the etymology, it records how the metres were regarded: as protective structures rather than as prosody. Calling her their sāra places her inside the oldest layer of Indian sacred speech, well behind anything specifically tantric.

Śāstra comes from √śās, to instruct or to rule — the same root as śāsana, an edict — and covers the whole range of systematic teaching, from grammar and law to medicine and erotics. To be the essence of śāstra is a different sort of claim from being the essence of the Vedas: the śāstras are human compositions, argued rather than revealed, and the name asserts that the argument arrives where the revelation does. Mantra, the third, has been characterized already in this hymn; it is worth noting only that the three names together move steadily toward compression. A hymn is long, a treatise is longer, a mantra may be a single syllable, and the verse is heading somewhere by narrowing.

Then, without transition, talodarī: she whose belly is flat as a level surface. Tala is the flat of the palm, a plane, a surface, and the compound makes a conventional claim about beauty in the conventional language of Sanskrit court poetry, where a slender waist is as standard an attribute as a lotus face. Readers new to this literature often find the shift jarring — three names about the essence of scripture, and then a compliment on her figure. It is worth resisting the impulse to allegorize it away. This hymn opened with a head-to-foot description of her body running to some forty verses, and physical epithets surface throughout precisely because the tradition does not treat her body as a metaphor for something more respectable. The commentators do offer symbolic readings of the slender waist, and those readings are worth knowing, but the plain sense is the primary one.

Udāra and uddāma, opening the second line, are both worth taking apart. Udāra means lofty, noble, generous — the word one applies to a great-hearted man as well as to an exalted reputation, so that udārakīrti says her fame is itself noble, not merely large. Uddāma is more physical: dāman is a cord or tether, the rope used to tie a calf, and ud-dāma is what has had the tether taken off. The same noun sits inside the name Dāmodara, given to Kṛṣṇa because his foster mother tied a rope around his middle. To say that her vaibhava is uddāma is to say that her magnificence has been unleashed and is running without restraint, which is a considerably more concrete image than "unbounded glory" suggests.

Varṇarūpiṇī closes the verse and closes the thought. Varṇa means a letter or speech sound; it also means colour, and it is the word for the four classes of society. The primary sense here is phonetic, and it belongs to a doctrine this hymn has already touched two verses earlier when it called her the fifty seats: the phonemes of Sanskrit, treated as a set, are themselves a goddess, Mātṛkā, and everything articulate descends from them. The suffix is the same one used in that earlier name, -rūpiṇī, she who bears the form of. What the two names together assert is that speech is not a means of referring to her. It is a portion of her, and the metres, treatises and mantras named at the head of the verse are arrangements of her own substance.

The three sāra names are the easiest sequence to garble in this verse because they differ only in their first element and share four identical final syllables. The one that needs attention is the first. Chandaḥsārā has an aspirated palatal ch at the head — a single sound, not c followed by h, and noticeably breathier than the c of caṇḍa — and then a visarga in the middle of the compound. In most recitation styles the visarga before s is realized as a light sibilant echo, so that chandaḥsārā runs together almost as chandas-sārā, with the s slightly prolonged. Do not insert a pause there; the visarga is the seam of a compound, not a break between words.

Śāstrasārā gives the cluster str, three consonants with the tongue moving from the palatal ś through a dental s to a dental t before the r taps. Say śās, then trā, and then join them without letting a vowel creep between the s and the t. Mantrasārā repeats the ntr cluster that appears throughout this hymn, dental n and dental t released into r, and the two names side by side are a good drill for the difference between the two sibilants: ś is made with the tongue toward the hard palate, s at the teeth.

Talodarī is a relief after those, four open syllables and no cluster, and it is where the breath naturally goes. Note that its o is long, as every Sanskrit o is, and that the d is dental.

The second line begins with a long compound produced by sandhi: udārakīrtiruddāmavaibhavā is udārakīrtiḥ followed by uddāmavaibhavā, with the visarga of the first turning into r before the vowel that begins the second. Knowing this is what keeps the line from being an unbroken block of sound. The doubled d of uddāma must be held — a geminate consonant in Sanskrit takes real time, and clipping it makes the word into udāma, which is not a word here. Say ud, hold, then dāma.

Varṇarūpiṇī has two retroflex nasals, one in varṇa and one in rūpiṇī, and both are there because of a preceding r. Curl the tongue back and keep it back through the vowel that follows.

For archanā the datives are chandaḥsārāyai, śāstrasārāyai, mantrasārāyai, talodaryai, udārakīrtaye, uddāmavaibhavāyai and varṇarūpiṇyai. Udārakīrtiḥ is the irregular one: kīrti is an i-stem, so the dative is kīrtaye and not the expected -āyai, exactly like prakaṭākṛtaye three verses earlier. Two i-stems in four verses is not many, and both of them are easy to regularize by mistake.

The verse also has something to say about the act being performed while it is recited. A person chanting this line is using metre, is repeating what a treatise has fixed, is producing sounds treated as mantra, and is doing all of it with letters. The verse names the essence of each of those four things and identifies it with the person being addressed, which means that by the time the reciter reaches varṇarūpiṇī there is nothing left in the situation — not the medium, not the method, not the material — that has been left outside her.