Rare birds… wait, rare moraic metres sighted

Inked estampage of a page of the Maliyapūṇḍi grant of Amma II. See the end of the article for details.


This little article is about an interesting discovery of some really unusual verse forms in epigraphic records. Before I can tell you about that, though, it will be best to do a quick recap on the background for these verse forms. If you feel you know as much as you want to about the Sanskrit and Prakrit moraic verse family called āryā or gāthā, feel free to skip that introduction and jump straight to the juicy bits after the Table 1. If not, read the intro. And in case that has whetted your appetite, look up Andrew Ollett’s excellent article (2013) on the subject.

Sanskrit poetry can take a large number of forms, but as far as classical Sanskrit verse is concerned, the overwhelming majority belongs to one of two categories: one based on a fixed number of syllables and one based on a fixed number of rhythmical units. In both types of versification, syllables are reckoned as short or long. As in classical European poetry, a short syllable is defined as one with a short vowel that is followed by no more than one consonant (regardless of whether or not those consonants belong to the syllable in question). Conversely, a long syllable is one that either has a short vowel followed by more than one consonant, or has a long vowel, in which case the number of subsequent consonants is irrelevant.

The first category of verse mentioned above is syllabo-quantitative, which means that a strict metrical template governs not only the number of syllables permitted in a given verse form, but also the length of each of those syllables. Conversely, what matters in the second category, that of moraic verse, is the number of syllabic instants, or “quanta,” technically called morae. This unit is equivalent to the length of a short syllable, and a long syllable is by convention reckoned as exactly twice that length, in other words, two morae. No syllable can be of a different length, for instance three morae or a mora and a half. Verse forms of this second category consist of a set number of feet (gaṇas), with the number of morae per foot prescribed in the template, but the actual instantiation of these feet in the form of syllables is less determined. Some constraints generally apply, but by and large, a long syllable may be substituted for two short syllables and vice versa.

With me so far? Good. We’ve almost got to the preamble now. Your garden variety moraic verse, in both classical Sanskrit and several Middle Indo-Aryan languages (often referred to with blanket term Prakrit), is the form called āryā. An āryā stanza consists of two hemistichs or, to use an easier term without worrying about the fine print, two lines. Each line is made up of eight feet, generally with a caesura (an obligatory word boundary) after the third foot. Most feet consist of four morae. That is to say, they may in principle be composed of two to four syllables in one of the following patterns: ––, –˘˘, ˘˘–, ˘–˘ and ˘˘˘˘ (where – stands for a long syllable and ˘ for a short syllable). As I mentioned above, there are in fact some constraints, so not all of these patterns are permitted in all of the feet. Namely, the pattern ˘–˘ is preferred in the even-numbered feet (and all but mandatory in the sixth), while the same pattern is generally forbidden in the odd-numbered feet.

But back to the āryā. In this verse form, the last (eighth) foot of each line is catalectic, meaning that it is reduced to two morae (that is, to a single syllable, since by convention a syllable at the end of a line is always reckoned as long regardless of its actual length), and the sixth foot of the second line is reduced to one mora, i.e., a single short syllable. Thus, the first line of an āryā consists of 30 morae (7 feet of 4 morae each plus one foot of two morae), and the second half consists of 27 (one of the four-mora feet being reduced to one mora).

Four additional forms of moraic verse are less common than the āryā, but still widely attested in both classical Sanskrit and Prakrit literature. One called gīti consists of two lines of the pattern found in the first line of an āryā (30 morae); the upagīti’s two lines are like the second line of an āryā (27 morae); and the udgīti is like an inverted āryā, with 27 morae in the first line and 30 in the second. Finally, the āryāgīti consists of two lines resembling the first line of an āryā, but differing from it in that the last foot of each line has the full mora count of 4, so that each line consists of 32 morae.

Table 1. line types in moraic verse

morae per line 1 2 3 4 5 6 7 8
30 4 4 4 4 4 4 4 2
27 4 4 4 4 4 1 4 2
32 4 4 4 4 4 4 4 4

We thus have a basic repertoire of three line types, represented in Table 1 showing the number of morae in each of the eight feet. If we take the 30-mora line as default (which is what most Sanskrit poeticians do), then the two other types can be obtained from this template by changing the mora count of the sixth or the eighth foot, as highlighted in the table. These three line types normally appear in two-line stanzas according to one of the five combinations as shown in Table 2 with the number of morae (i.e. line type) in each hemistich.

Table 2. stanza types in moraic verse

first hemistich second hemistich name of verse form
30 30 gīti
30 27 āryā
27 30 udgīti
27 27 upagīti
32 32 āryāgīti

There is one thing in each of the above tables that may strike you as slightly odd. Table 1 lacks a 29-mora line type, even though one might expect to encounter a line type which has both an extended eighth foot and a reduced sixth foot. The second oddity, apparent in Table 2, is that while there are verse forms corresponding to all possible combinations of 30-mora and 27-mora lines, the 32-mora line only occurs on its own, and never in combination with other lengths.

There is of course no universal law prescribing that line and verse forms must exist to correspond to all conceivable combinations of permitted variants. And for that reason I have never found the above quirks any odder than the other conventions of moraic verse. There is, for example, no line form with just two morae in the first foot, or just one mora in the third, and this has never to my knowledge caused a sleepless night for anyone. Likewise, then, there is no line form with four morae in the last foot and one mora in the sixth and no verse form with thirty-two morae in one line and a different number in other.

Thus, when I recently came across a stanza with a weird mora count while encoding a copperplate grant of the Eastern Cālukya dynasty, I dismissed it as bad text after some consideration. This was not unreasonable, since parts of the stanza were barely legible, parts unintelligible (probably involving non-Sanskritic proper names), and the remaining parts had several scribal errors. Even though two great scholars, John Faithfull Fleet and Eugen Hultzsch had both studied this stanza, both apparently regarded it to be prose.

However, shortly afterward I ran into a much better specimen, where scribal mistakes affecting the mora count seemed unlikely, yet the first hemistich had 30 morae, and the second 32. As I proceeded with my encoding of the Eastern Cālukya corpus, my inventory of these odd moraic verses grew slowly but steadily, and came to include specimens of other unusual line combinations.

Becoming curious, I found out that such stanzas are actually not unknown to literary historians. Anthony Kennedy Warder (1967: 143-144) lists them and says he knows of two extant examples of one of these forms, and “only the shadows of” the others. Most of his information on the topic seems to be based on Carl Cappeller (1872: 14, 21-242), who in turn relies largely on Henry Thomas Colebrooke (1811). He in turn seems to have received his information from an unnamed paṇḍit, and does not say what texts it is based on, but Ollett’s recent paper (2013, 353) specifies some texts of traditional poetics that mention the at least combinations of 30-mora and 32-mora lines. One of these texts, the Jayakīrti of Hemacandra, in fact also discusses combinations involving the 29-mora line, whose non-existence I noted above as the first oddity. Cappeller and Warder agree that this line variant probably only ever existed in theory, and I have not found any attestation of it in the Eastern Cālukya inscriptions.

Table 3 presents the full gamut of theoretically possible combinations for this class of verse. The first row of the table shows the number of morae in the first hemistich (A), and the first column shows the number of morae in the second hemistich (B). The cells of the table give the Sanskrit and Prakrit names used in various sources for each of the combinations. The combinations shaded in red are the well-attested and widely known ones introduced above. Those with a grey background involve the 29-mora line and are not known to be attested in poetic practice. The three cells shaded in bright yellow are those that are attested in the Eastern Cālukya corpus, out of which the sugīti is known to be at least sporadically found elsewhere too, according to Warder. The remaining cell, showing the metre anugīti and shaded in dull yellow, ought to exist given the existence of its mirror image the sugīti, but I am not so far aware of any attestation of this form.

Table 3. full range of stanza types in moraic verse

B↓             A→ 27 29 30 32
27 upagīti
pramadā āryā
29 candrikā vigīti
pragīti saṃgīti
30 udgīti
32 anugīti cārugīti lalitā

To this date, I have found in Eastern Cālukya copper plates one instance of the sugīti (32+27 morae), three instances of the vallarī (32+30 morae), and 10 instances of the lalitā (30+32 morae), though out of this latter 10, one is the possibly incorrectly written stanza mentioned above, and another five are the same stanza re-appearing in five separate inscriptions. I thus know of a total of five stanzas certain to have been composed in the lalitā metre. All of these charters have been edited previously, but their editors appear to have had little interest in this metrical quirk. Variously, they have regarded these stanzas as prose, as verse whose metre they did not care to identify, as prosodically faulty āryāgīti, or as a mixture of āryā and āryāgīti.

The image at the top is an inked estampage of a page of the Maliyapūṇḍi grant of Amma II (edited by Butterworth and Venugopaul Chetty, 1905; re-edited by  E. Hultzsch, 1907; the picture is from the former). The highlighted section (lines 52 to 53 of the grant) is a stanza in the vallarī metre. Its text, with some minor emendations, is as follows:

śrīmān diradeva-munis sutapo-nidhir abhavad asya śiṣyo dhīmān
yam prātihāryya-mahimā-saṁppannam ivābhimanyate lokaḥ

His disciple was the intelligent ascetic the Reverend Diradeva, a repository of noble austerity whom the people esteem as one possessed of miraculous power.

I have not found out why and how the Eastern Cālukya court poets came to use these verse forms, or whether they are also attested in inscriptional poetry from other dynasties close in space and time. All of the attestations known to me come from the reign of Amma II (ca. third quarter of the 10th century), though the problematic stanza that may have been intended for a more common metre is from the reign of his predecessor Bhīma II (ca. second quarter of the 10th century). In the same period, grants to Jaina masters make their appearance in the Eastern Cālukya corpus (including the example cited above), so it is tempting to speculate that the use of these verse forms was adopted by the Sanskrit court poets from Jaina Prakrit literature.


Butterworth, Alan, and V. Venugopaul Chetty. 1905. A Collection of the Inscriptions on Copper-Plates and Stones in the Nellore District: Part I. Madras: Superintendent, Government Press.

Cappeller, Carl. 1872. Die Ganachandas. Ein Beitrag Zur Indischen Metrik. Leipzig: Hüthel & Legler.

Colebrooke, Henry Thomas. 1811. ‘On Sanscrĭt and Prácrĭt Poetry’. Asiatic Researches 10: 389–474.

Hultzsch, Eugen Julius Theodor. 1907. ‘Maliyapundi Grant of Ammarāja II’. Epigraphia Indica 9: 47–56.

Ollett, Andrew. 2013. ‘The Gaṇacchandas in the Indian Metrical Tradition’. In Puṣpikā: Tracing Ancient India, through Texts and Traditions. Contributions to Current Research in Indology. Volume 1, Proceedings of the First International Indology Graduate Research Symposium (September 2009, Oxford), edited by Nina Mirnig, Péter-Dániel Szántó, and Michael Williams, 331–64. Oxford, Philadelphia: Oxbow Books.

Warder, Anthony Kennedy. 1967. Pali Metre: A Contribution to the History of Indian Literature. London: The Pali Text Society.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Search OpenEdition Search

You will be redirected to OpenEdition Search