The origins of the àryà metre
by
Prof. K.R. Norman 01
[Ed: I have expanded most of the abbreviations used in the original for ease of reading]
1. Introduction
[p. 203] 1.1. As is well known, the àryà metre is very common in Jaina Prakrit texts, and it is perhaps not far off the mark to call it the `favourite metre' of early Jaina authors. When European scholars first began to study Jaina texts they were perplexed by the metre, with its distinctive `irregular' sixth gaõa (bar) in the second line, and much time and effort was devoted, especially by Jacobi, 02 to examining the way in which the metre had developed from the earlier tradition of Indian metrics.
1.2. In his study of Tamil poetry, 03 aimed at defining the cultural milieu in which it was composed and the relationship with equivalent types of Sanskrit and Middle Indo-Aryan, George L. Hart has made an interesting and new suggestion about the way in which the àryà metre came into being, completely rejecting the generally accepted idea that the àryà metre was developed, 04 together with other gaõacchandas metres from earlier màtràchandas metres. He points out that when the àryà metre is recited in modern-day Mahàràùñra, the usual pattern of 4, 4, 4/4, 4, 4, 4, 2//4, 4, 4/4, 4, 1, 4, 2// is replaced by 4, 4, 4/4, 4, 6, 4//4, 4, 4/4, 4, 6, 1//, i.e. ßthe sixth caturmàtra of the first line comes together with the first two màtràs of the seventh to make two triplet rhythms that are together given the time of a regular caturmàtra, while the last caturmàtragaõa of the second half absorbs the short syllable before it and one màtrà of the long syllable after it to produce a similar unit with two triplet rhythms.û 05 This analysis enables Hart to suggest a connection between the àryà metre and a number of Tamil metres which are made up of feet of either four or six màtràs (metrical instants).
1.3. A number of objections may be levelled against Hart's suggestion. The type of àryà schematized above is the classical form of the metre with a long vowel of two màtràs at the end of each line. In the form in which the àryà is found in Pàli and Prakrit, the final syllable of each line is anceps, i.e. either long or short, so that the final gaõa often has only one màtrà, not two. 06 This means that when the `borrowing' from gaõas, which Hart describes, takes place, there will be only three màtràs left in the last gaõa in the first line, and the final single màtrà at the end of the second line would disappear altogether.
1.4. Another objection to Hart's suggestion arises from the relative chronology of the texts in which the àryà metre occurs. Hart was considering only the àryà verses of Hàla's Sattasài, which he dated to the 2nd or 3rd centuries A.D., 07 while [p. 204] he dated the Tamil texts to the 1st to 3rd centuries A.D. 08 Although he mentions the existence of standard àryà verses in Pàli, he rejects Warder's dating for the texts in which they occur, 09 preferring the 3rd or the 2nd centuries B.C., and he maintains that this is late enough for them to have been influenced by Tamil.6 He makes no reference to àryà verses in Jaina texts, and in particular he makes no mention of the earlier form of the àryà metre called old gãti, 10 which occurs in a handful of very early Prakrit and Pàli texts. He does not discuss the occurrence of the veóha metre 11 in early prose texts in Prakrit, Pàli, and Buddhist Hybrid Sanskrit.
1.5. Already in the old gãti and veóha metres the acceptance of the amphibrach /ÛÜÛ/ 12 in certain gaõas had been standardized. This is in contrast to the Tamil metres, where according to Hart the rhythm ßis not commonly allowedû. 13 Although he quotes Marr as saying that the rhythm is possible, 14 he does not make it clear whether it occurs more frequently than the example(s) he quotes. In any case, it is clear that this rhythm is as rare in Tamil poetry as it is common in Prakrit.
1.6. It seems clear that we must conclude that the simple hypothesis that the àryà metre in Middle Indo-Aryan is borrowed from Tamil cannot be upheld, and its origin must be sought elsewhere. I am convinced that Hart was able to make his suggestion, ill-founded as I believe it to be, because (despite the large amount of work which has been done on the problem of the origin of the àryà metre) there is no single modern study devoted to the question. The work already done tends to be scattered through the pages of journals, and books devoted to the study of metre have only a portion devoted to the àryà metre, and are moreover restricted to its occurrence in a single language or dialect. 15
1.7. In this paper, offered in honour of Professor N.A. Jayawickrama, I wish to state briefly the facts relating to the àryà metre in Middle Indo-Aryan, to put together the views of earlier scholars on the matter, and to give further evidence for rejecting Hart's suggestion. I hope that a treatment which aims at referring to all the major evidence, albeit very briefly, will help to fill the gap I have just mentioned.
2. The màtràchandas metres
2.1. The Vedic metres seem to be chants, with /Ü/ contrasted with /Û/. In the màtràchandas and gaõacchandas metres however, /Ü/ is contrasted with /ÛÛ/, which seems to imply a musical basis to the metres. 16 In the absence of any other explanation for this change which took place in IA metres at some date after the compilation of the »gveda, it seems reasonable to assume that it was due to the indigenous peoples of North India, who combined their music with the IA metres, and produced what we may call `folk metres'.
2.2. The màtràchandas metres seem to show a halfway position between the Vedic metres and the gaõacchandas metres, because the licence to substitute /ÛÛ/ for /Ü/ is restricted to the first half of the pàda (= the opening). In the second half (= the cadence) the form is fixed to /ÜÛÜÛÅ/ (vaitàlãya), /ÜÛÜÛÜÅ/ (aupacchandasaka), or /ÜÛÛÜÅ/ (vegavatã). 17
[p. 205] 2.3. Many authorities agree 18 that the màtràchandas metres are derived from metres found in the Vedas, with the alternation between /Ü/ and /ÛÛ/ introduced, e.g. an anuùñubh pàda /ÜÜÜÜÛÜÛÅ/ with resolution of the third long syllable into two shorts gives the vaitàlãya prior pàda /ÜÜÛÛÜÛÜÛÅ/. The variation in the length of the opening between the prior and posterior pàdas of the màtràchandas metres can also be paralleled in the mixed metre verses of the »gveda. 19
2.4. The màtràchandas metres in Pàli and Prakrit are found only in texts 20 which, for various reasons, are thought to be old and composed at an early state of the Buddhist and Jaina religions, when they were still confined to the Magadha region. Warder suggests that the name màgadhikà, which is sometimes given to the vaitàlãya metre, probably implies that the metre was invented in Magadha. 21
2.5. The Pàli tradition responsible for transmitting texts in màtràchandas metres seems to have been ignorant of those metres, and consequently mutilated them badly. 22 The Jaina knowledge of the metres does not seem to have been much better. 23 This is presumably because in both traditions the metres dropped out of use at an early date, and knowledge about them disappeared. Knowledge of the màtràchandas metres did, however, remain in North India after Buddhism and Jainism moved further afield, and we find texts containing màtràchandas metres in Buddhist Hybrid Sanskrit and classical Sanskrit, although there they have fixed forms, and lack the flexibility which we find in Pàli and Prakrit. 24
3. The old gãti metre
3.1. If a vaitàlãya prior pàda is sung in four-beats-to-a-bar time, then there are bars (gaõas) of /ÜÜ/,/ÛÛÜ/, and /ÛÜÛ/, with a syllable left over. If a vegavatã prior pàda is sung in a similar way, then there are bars of /ÜÜ/,/ÛÛÜ/, and/ÛÛÜ/, with a syllable left over. If an aupacchandasaka prior pàda is sung in this way, then the extra syllable in the cadence gives another bar of four beats /ÜÜ/.
3.2. If the second gaõa in a vegavatã prior pàda, sung in this way, is `syncopated' from /ÛÛÜ/, to /ÛÜÛ/, to make a contrast with the third gaõa, which is also /ÛÛÜ/, then we can see how the gaõacchandas metres evolved from the màtràchandas metres, 25 with the amphibrach /ÛÜÛ/ already in the second gaõa.
3.3. If a vegavatã prior pàda is followed by a vegavatã posterior pàda, then the additional long syllable (or two short syllables) at the beginning of the posterior pàda combine(s) with the syllable left over at the end of the prior pàda (as seen in Æ 3.1) to form a fourth gaõa, while the remainder of the pàda falls into a gaõa pattern exactly as in Æ 3.1. 26 A `syncopated' gaõa in the same place in the pàda as described in Æ 3.2 will give another amphibrach in what is now the sixth gaõa: /ÜÜ/ÛÜÛ/ÛÛÜ/*//Ü/ÜÜ/ÛÜÛ/ÛÛÜ/*/.
[p. 206] 3.4. This is precisely the pattern of the old gãti metre. This name, which is the one used by Warder, 27 seems preferable to the name `old àryà' which was used by Jacobi, and following him by Schubring and Alsdorf, 28 because both lines are the same, as in the case of the classical gãti. In this paper I shall follow Warder, even when I am quoting the views of the afore-mentioned German scholars.
3.5. That this is indeed the origin of the old gãti metre is shown by the fact, first pointed out by Jacobi, 29 and echoed by Alsdorf, 30 that in the old gãti the first syllable of the fourth gaõa, i.e. the one before the caesura, is anceps, which is a certain sign that it was originally the last syllable of a pàda.
3.6. When the second half of the line starts with one long or two short syllables, after a short syllable at the end of the first half, then the fourth gaõa contains only three màtràs. When we include the possibility of the second half of the line starting, exceptionally, with a long and a short syllable, or even two long syllables, after a long syllable at the end of the first half, we find that the màtrà count of the fourth gaõa can be anything between three and six màtràs. The recitation of such a line by the modem Mahàràùñra method (described in Æ 1.2) would then, of course, be impossible. This fact helps to confirm the incorrectness of Hart's suggestion.
3.7. Where the first half of the line ends in a short syllable, and the second half starts with a long and a short syllable, then we have another amphibrach /ÛÜÛ/ giving amphibrachs in the second, fourth, and sixth gaõas. As, in the course of time, under the influence of the fact that the other gaõas had four màtràs, the four-màtrà pattern of the fourth gaõa became the standard one, then with a shift of the position of the caesura we have the classical gãti metre. In my examination of the old gãti verses of the eighth chapter of the Uttarajjhàyaõa-sutta, 31 I showed how in several verses changes could be made very simply, metri causa (= m.c.), to give the amphibrach pattern. With the standardization of the amphibrach pattern /ÛÜÛ/ in the second, fourth, and sixth gaõas, came its exclusion from the first, third, fifth, and seventh gaõas.
3.8. The old gãti metre occurs in three suttas in Ardha-Magadhã 32 and three in Pàli, 33 and in a handful of individual verses. These suttas and verses appear in texts which for various reasons are considered to be old, 34 and to belong to the earliest strata of Buddhist and Jaina texts, and we may therefore conclude that like the verses in the màtràchandas metres (Æ 2.4) they were probably composed in the Magadha region.
3.9. The old gãti metre appears not to exist outside these texts. The reason for this must be that it is a transitional stage of the metre. Probably its period of greatest use was at a time even before the composition of these early Middle Indo-Aryan texts, and it was just going out of use when they were composed. The classical gãti metre, into which it evolved, was not greatly used, as far as can be judged from the extant Middle Indo-Aryan literature. 35 It was, perhaps, somewhere other than Magadha or Mahàràùñra, [p. 207] which are the regions from which most of our early Middle Indo-Aryan texts come, where much of the experimentation with the gãti-based metres took place, and where the gãti remained in use until classical Ski times.
4. The veóha metre
4.1. Akin to the gaõacchandas metres, and considered to lie between the old gãti and the classical gãti and àryà metres in date of development, 36 comes the veóha, a type of rhythmical prose found in the very oldest prose texts in Ardha-Màgadhi, Pàli, and Buddhist Hybrid Sanskrit, 37 in the so-called varõakas. 38 Winternitz expressed doubt as to whether the veóha was in fact a metre, 39 but the quotations given by Dr Mette 40 make it clear that the Indian tradition did in fact so regard it.
4.2. Jacobi was the first to discover and examine the veóha, 41 of which he found some hundreds of examples in the Jaina canon. According to his investigations, it would seem that a veóha should have an even number of gaõas, not less than four and extendable to 28, eight being the favourite number. 42 Each gaõa should contain four màtràs. Varõakas which consist of only three gaõas are, according to Jacobi, 43 remains of longer units.
4.3. The veóha, like the old gãti has amphibrachs, but in the odd-numbered gaõas, not the even ones. Warder has pointed out 44 the similarity in this respect to the later gurviõã. If we take an aupacchandasaka prior pàda, we have seen above (Æ 3.1) that sung to a four-beats-to-a-bar time it gives four bars (gaõas). If such a pàda has the opening /ÛÛÜ/ÛÛÜ/, and we assume `syncopation' of the first gaõa from /ÛÛÜ/ to /ÛÜÛ/ to make a contrast with the second gaõa, which is also /ÛÛÜ/, then we have an amphibrach in the first gaõa as well as one in the third, i.e. we have a perfect veóha on Jacobi's pattern: /ÛÜÛ/ÛÛÜ/ÛÜÛ/ÜÜ/. This shows clearly how the basic veóha came into existence. The favourite veóha of eight gaõas is simply the basic veóha repeated twice. 45
4.4. Sen examined the veóhas in the Paõhàvàgaraõàiü, and concluded from their form, and from the fact that the classical form of the àryà metre was beginning to appear in that text, that they represented a later form of the veóha. 46
4.5. The veóha occurs in Pàli in the canonical prose of the Kuõàlajàtaka, 47 and has been investigated by Bollée. 48 He has pointed out the existence of several veóhas there with odd numbers of gaõas, up to 17 in number, which goes against Jacobi's findings that they should consist of an even number of gaõas. Some of the 18 veóhas which occur in that text lack the first gaõa, 49 i.e. the amphibrach /ÛÜÛ/ is in the even gaõas. This may be due to corruption. As Bollée points out, 50 it is no wonder that the varõakas, being in a more obscure metre than the àryà, were liable to corruption and addition. Since the varõaka is used in the Kuõàla-jàtaka for stereotyped descriptions of flowers, trees, etc., it would not be surprising if other names and attributes were inserted 51 into the compounds by those who did not realise that they were metrical.
[p. 208] 4.6. On the other hand, it is not impossible that the Pàli veóhas may represent a genuine variety of the standard veóha with amphibrachs in the even, not the odd, gaõas, and which therefore stands to the standard veóha in just the same way as the àryà stands to the gurviõã. Possibly, however, they represent a later development. Bollée points out 52 other examples of metrical licence not found in the Ardha-Màgadhã veóhas, which supports the view that the Pàli form of the veóha is slightly later than the Ardha-Màgadhã form.
5. The àryà metre in Pàli.
5.1. It is generally agreed 53 that there was an historical development from the old gãti to the classical form of the metre. As Alsdorf states, 54 the classical gãti orginated from the coalescence of the two (originally separate) pàdas of the old gãti, combined with a shift of the caesura. Some trace of the older caesura position is perhaps to be seen in those àryà verses which do not have the caesura at the end of the third gaõa, in which case the fourth gaõa is (almost) invariably /ÛÜÛ/ or /ÛÛÛÛ/. 55 The other gaõacchandas metres, including the àryà with its truncated second line, were derived from the classical gãti by introducing various types of `syncopation' and/or rest towards the end of the line. 56
5.2. Further evidence for the belief that the old gãti, by way of the classical gãti, is the origin of the gaõacchandas metres is given by the fact, pointed out by Warder, 57 that six of the nine classical gaõacchandas metres have gãti as part of their name. The fact that their names are derived from gãti suggests that the metres themselves were derived from the gãti metre.
5.3. Hart, however, suggests 58 that by the time the àryà and gãti metres appeared in Pàli, Pàli literature had been exposed to Southern elements in Central India, and perhaps in Ceylon as well. He is able to make such a suggestion because, as stated above (Æ 1.4), he reject's Warder's dating of the new metres to the 5th and 4th centuries B.C., 59 and favours dates in the 3rd or 2nd centuries B.C. at the earliest. 60
5.4. It has already been stated (Æ 3.8) that the old gãti metre occurs in Pàli only in old texts, which can be dated much earlier than the 3rd or 2nd centuries B.C., and can reasonably be assumed to have been composed in the Magadha region. Even verses in the classical form of the àryà metre occur in texts which can be assumed to have been composed before the time of the Third Council, 61 which took place c. 250 B.C. during the reign of Aèoka. There is, in fact, abundant evidence that the gaõacchandas metres were not in general use in Pàli outside Magadha.
5.5. It is clear that the knowledge and understanding of the old and classical forms of the gãti and àryà metres were lost at a comparatively early date. The total number of verses in gaõacchandas metres in Pàli is about 450, 62 and there is a tendency for them to be corrupted into the èloka metre, which was the more common metre [p. 209] in later Pàli texts. This tendency was aided by the fact that some pàdas can be scanned as both gaõacchandas and èloka, as Alsdorf notes. 63 There was, however, no tendency for èloka pàdas to be corrupted into gaõacchandas pàdas.
5.6. As Alsdorf states, 64 the àryà metre fell into disuse in Pàli after the `emigration' of the language from India to Ceylon. The only examples of àryà verses known to me in pre-sixth century A.D. non-canonical Pàli are in texts which are probably to be regarded as of North Indian origin or authorship, e.g. there are àryà verses in the Netti-pakaraõa, 65 and the introductions and conclusions by Buddhaghosa to his commentaries are also in the àryà metre. 66
6. The àryà metre in Prakrit
6.1. As in the case of Pàli, the development from the old gãti to the later form of the metre must have taken place early on in Prakrit, for the classical form of the àryà metre is already found in the oldest texts of the Jaina canon. 67 The Uttarajjhàyaõa-sutta has over 100 àryà verses, although some may be later interpolations, 68 and the nijjuttis (some of which are regarded as canonical) are almost entirely in the àryà metre. Whereas the use of the àryà metre in Pàli ceased when Buddhism migrated from Magadha, there was no break in the Jaina tradition, and the àryà metre became the favourite metre of Jaina authors.
6.2. The metre was also used in Hàla's Sattasaã, a secular text in Mahàràùñrã. Hart, as stated above (Æ 1.4), dates this text to the 2nd or 3rd centuries A.D., 69 but I see no reason for not dating, at least some portions of it, to an earlier date. I see no cause to reject a theory than an anthology was made by a king called Hàla in the 2nd century A.D., but the date of the collection has no bearing upon the date of the composition of individual verses, except for those which are ascribed to the compiler of the anthology himself. The linguistic evidence which has been used to date the material is by no means as conclusive as some scholars have thought. Keith 70 placed most emphasis upon the weakening of consonants in the Sattasaã (i.e. the development to -y-), and claimed that this could not have happened until A.D. 200. According to Mehendale, 71 however, the change occurs in inscriptions of the 1st century B.C. There is now evidence that it occurred in pre-Pàli Prakrit, 72 and so the linguistic evidence could support an origin as early as the 3rd century B.C.
6.3. The àryà metre was widely used at a later date in Mahàràùñra by Jaina authors, and were it not for the existence of the Sattasaã it would seem probable that the knowledge of the metre was brought with Jainism to that country from Magadha. The earlier, however, we date the Sattasaã the less likely it is that this was so, and we must consider other possibilities. Clearly the metre could have developed independently in the two countries, or it could have originated in Mahàràùñra and thence been taken to Magadha, or it could have originated in a third country and been introduced into both from there.
7. Conclusions
[p. 210] 7.1. We must assume that when the Indo-Aryans entered India from the North-West at some time in the second millenium B.C., they found North India populated by peoples who spoke Dravidian languages. As the Indo-Aryans conquered the country, and imposed their language upon the subject peoples, certain aspects of the Dravidian languages and culture inevitably remained and left their mark upon the invaders. It is to this Northern Dravidian influence 73 that we can ascribe the Dravidian words which we find in early Sanskrit, rather than to borrowing by the Indo-Aryans from the Dravidians whom they found as their southern neighbours in classical times. 74
7.2. Among the cultural influences, which the Indo-Aryans adopted from the Dravidian sub-strate they absorbed within themselves, were music and metre. The imposition of the brahmanical religion upon the subjected peoples led to their knowledge of Vedic metres, but their tendency to sing rather than to chant, and to make use of musical syncopation, led to a flexibility being introduced into the old Vedic metres in a way which completely transformed them.
7.3. Under the influence of the indigenous music which allowed an alternation of /Û/ and /ÛÛ/, the Vedic metres developed into the màtràchandas metres. The same music, with four beats to a bar, led to the division of the newly developed màtràchandas metres into bars with four beats in each (gaõas), some of which could be varied considerably in form, while others (sometimes the odd bars, and sometimes the even ones), were much more restricted in form. Certain patterns of these bars gave the form of the old gãti metre. From this developed the classical gãti, from which came the àryà metre.
7.4. In the gaõacchandas metres which we find in the earliest Pàli and Prakrit texts available to us, which we can confidently date to the earliest period of Buddhism and Jainism, i.e. the 5th and 4th centuries B.C., and locate in Magadha, we find that the use of the amphibrach /ÛÜÛ/ was already established in fixed positions. This must rule out any direct connection with the Tamil metres, where the use of amphibrachs is very limited, as we have seen above (Æ 1.5).
7.5. Although our evidence for the early use of the àryà metre is virtually confined to Magadha (Pàli and Ardha-Màgadhã) and Mahàràùñra (the Sattasaã), it is not necessary to assume that the àryà metre was evolved in one or other of these two countries and then carried to the other. Whichever way the movement went, it is clear that the metre must have been known in the areas between the two regions. This opens up the possibility that the place of origin of the metre could have been in some third place.
7.6. If we assume that the development just described (Æ 7.3) was centred upon an area somewhere between Magadha and Mahàràùñra, then we can assume that the metres were carried eastwards and westwards, probably by travelling folk musicians [p. 211] who used the metres for secular purposes. When the metres reached Magadha and Mahàràùñra, the period of experimentation in their use was nearly over, and in their early forms (old gãti and veóha) they did not find great favour. There are only limited traces of the old gãti metre in texts derivable from Magadha, as stated above (Æ 3.8), and no example of the metre is found in texts from Mahàràùñra. Even the màtràchandas metres upon which the gaõacchandas metres were based did not prove very popular in those areas. There are only limited traces in texts from Magahda, and none in those from Mahàràùñra.
7.7. It seems likely that the experimentation which led to the later forms of the gaõacchandas metres, including the classical gãti, also took place outside Magadha and Mahàràùñra, for as stated above (Æ 3.9), the gãti and allied metres, except for the àryà, are rare in both Pàli and Prakrit. Nevertheless these metres remained in use somewhere outside the two regions, and were used again much later in classical Sanskrit.
7.8. The àryà metre won the battle to be the most popular gaõacchandas metre. It could, by coincidence, have done this independently in both Magadha and Mahàràùñra, but it seems likely that it won the battle outside these two areas, and was carried into both of them. It arrived in Magadha in time to be adopted by both the Buddhists and the Jains, probably because it was a folk metre, which therefore appealed to both religions because it represented a rejection of the brahmanical metres, just as they favoured non-brahmanical Middle Indo-Aryan dialects rather than the brahmanical Sanskrit.
7.9. The Buddhists adopted the àryà metre just before they closed their canon and took it to Ceylon. They had not had sufficient experience of the metre before they left for it to have become part of their literary tradition, and they consequently made no further use of it, with the exception of the texts mentioned above (Æ 5.6). The Jains remained in North India, adopted the àryà metre and made great use of it.
7.10. The àryà metre arrived in Mahàràùñra in time for it to have become the dominant metre for secular poetry long before the Sattasaã was compiled, which accounts for the almost complete exclusion of other metres from that text.
7.11. Although in the first section of this paper I rejected Hart's suggestion about the origin of the àryà metre from Tamil metres as being untenable, it is nevertheless clear that some indigenous (Dravidian) influence played a part in its origin. Perhaps the difference of our views can be summed up by saying that by Hart's thesis the relationship between the gaõacchandas metres and the Tamil metres would be that of mother and daughter (or perhaps of sisters); by my reckoning they are no more than very distant cousins.