In working upon a translation of the Iti-vuttaka, sometimes called the Logia-book of Buddha, it occurred to me that a study of the meters of the metrical portions might perhaps yield something of value as regards both the age of the work and the authenticity of some of its doubtful passages. In this hope I was largely disappointed. But although no satisfactory clue as to the date of the work has been given by this metrical analysis, yet it is possible that further similar examination of other books of the Buddhist canon may permit us to assign to each its proper relative date.

More definite and satisfactory results, however, were obtained, when it came to making a threefold comparison of the Pàli meters with those of the Veda, the epics, and the later classical forms. For comparison with the Vedic meters, I have made use of Arnold, Vedic Metre, Cambridge, 1905, and have employed such of his terminology as was needed; for the Sanskrit meters I have relied on Hopkins' Great Epic of India, N. Y., 1901, pp. 191-362; and for Pàli upon the articles of Oldenberg and Simon mentioned below.

The text of the Iti-vuttaka which I have used is that of Windisch, published by the Pàli Text Society of London in 1890, and in comparison with this I have collated the King of Siam's edition of the work in Siamese characters. The Iti-vuttaka is composed of one hundred and twelve sections, each of which consists of a poetical discourse or saying by Buddha (these poetical passages contain from four to thirty verses [Moore uses the term verse here to indicate what he elsewhere refers to as a line]) and of a prose introduction.

Three meters are used, èloka, triùñubh, and jagatã. Of these I shall examine more particularly the first, the èloka, which is the most frequent and most important. The large Arabic numbers used in citing various lines of the work refer to the various sections, and the small letters, a, b, c, etc., to the verses of these sections.

S1oka. There are in Pàli, as in Sanskrit, two kinds of èloka-stanza: first, the èloka proper or distich, of four pàdas (the [318] pàda being octosyllabic), and second, the much less common mahàpaïkti or tristich of six pàdas. As the same laws of caesura and rhythm apply to each, I include the two varieties in the metrical tables below.

Feet. Every distich stanza has a well-marked division or caesura at the end of the second pàda, so that the stanza falls naturally into two halves. Each pàda may be divided into two feet of four syllables each. As no metrical difference between the two halves of the stanza exists, there are really not eight different feet, but four. With regard to the six-line èloka, or mahàpaïkti, the stanza is divided into three equal parts, mutually independent as to rhythm. The opening feet of the first, third and fifth pàdas may be called first feet.

Syntactical union. Although there be this metrical isolation of successive verse-couplets, there is very frequently a syntactical union of each pàda with the one following. In fact, we occasionally find two stanzas forming a single sentence.

Caesura. The cadence of the èloka naturally depends largely on the sense. It is invariable in the Iti-vuttaka that there be a strong caesura at the end of the second pàda, and also caesuras, somewhat weaker, at the close of the first and third pàdas, but still strong enough to prevent the lines being run together. It is found, also, that when the second foot is of the form ÜÜÜÅ there is usually a caesura within the pàda itself, after the fifth syllable. Out of twenty-four instances there are but two exceptions (51 i, 77 a). Hopkins notes the same rule in epic Sanskrit (op. cit., p. 221).

Run-on verses. In Sanskrit one pàda is sometimes merged with the following in such a manner that the two are inseparable at the end of a line. This is usually the case when a long list of objects is cited, as noted by Hopkins, page 196, but in the Iti-vuttaka such a running together of lines never occurs.

Hiatus and Rhyme. Hiatus is found everywhere. This is partly due to the absence of the application of such rules of sandhi as are carried through in Sanskrit; it is partly owing to the structure of the Pàli language, which is characterized by an avoidance of final consonants. Rhyme, which according to Hopkins (p. 200) is not uncommon in epic Sanskrit, is non-existent in the Iti-vuttaka. Alliteration is rare and is probably largely unconscious and accidental. A marked alliteration [319] effect is found in Æ 90, an interesting jingle, where in the course of sixteen lines the word agga is used eleven times, with more than one meaning.

At this particular point in our investigation of the èloka, before going into the varieties of feet, I wish to acknowledge in a more especial way my indebtedness to two articles in ZDMG., the first by Oldenberg, vol. 35, pp. 181-188, entitled Bermerkungen zur Theorie des øloka, and the other by Simon, vol. 44, pp. 83-97, entitled Der øloka in Pàli. While both papers are very suggestive, the latter was especially valuable for my purpose, based as it is upon a quantitative analysis of the 725 verses of the Dhammapada, 2622 from the Thera-Therãgàthà, and 2430 chosen from the Jàtakas. In analyzing the meters of the Iti-vuttaka I have followed Simon's method in all respects, except that I have treated the eighth syllable of each line as anceps, since Simon's tables themselves show an equal frequency of long and short syllables, and since this eighth has no effect on the character of the rhythm.

Odd Pàda. I append herewith an analytical table of the first and second feet of the odd pàdas; in this table hypermetric feet are not included, but will be treated separately. The even pàdas will be tabulated and described later.

[320] Some little explanation is necessary before making comment on this table. Two consonants, as well as a niggahãta (Skt. anusvàra) followed by a consonant, make a syllable heavy (Henry, Grammaire Palie, p. 3) or long by position. Wherever, in lines which at first glance appear hypermetric, a word containing an anaptyctic or svarabhaktic vowel occurs, this vowel is naturally disregarded in the analysis, and the line is treated as regular, e.g. ariya is analysed as a trochee [ÜÛ]. No catalectic lines occur, and hypercatalectic lines will be found treated in another section of this paper.

An examination of the foregoing table of different combinations of syllables in the odd pàdas reveals many things of interest. There are sixteen possible combinations of the four syllables of the first foot, and eight of the second foot. Although there be this large number of possible combinations, it will, nevertheless, be seen that there is a marked preference for certain particular combinations of long and short syllables. Among the more marked of these peculiarities may be cited the following:Þ

(a) If the fourth syllable is short, the fifth must usually be short also. In the Iti-vuttaka and Jàtaka a short fifth is 16 times as common as a long, in the Thera-therã-gàthà, 23 times, and in the Dhammapada, 40 times as common.

(b) A succession of four iambs [ÛÜÛÜÛÜÛÜ] is very rare, there being no instances in Dhammapada, 3 in Iti-vuttaka, 4 in Thera-Therãgàthà, and 6 in Jàtaka.

(c) Even three iambs [ÛÜÛÜÛÜ] are uncommon in proportion to the number of lines, since there is only about 3% of such succession in the Iti-vuttaka, and only 4% in the other works.

(d) A succession of four trochees [ÜÛÜÛÜÛÜÛ] is very rare, there being none in Dhammapada or Thera-Therãgàthà, one in Iti-vuttaka, and two in Jàtaka A succession of three trochees [ÜÛÜÛÜÛ] is almost equally uncommon; the Iti-vuttaka and Thera-Therãgàthà have each one instance; the Dhammapada has none; the Jàtaka eight.

(e) It appears from the table, furthermore, that if the sixth and seventh syllables of a verse are either both long or both short, the fifth syllable is then usually of a different quantity. In this respect the four works stand in a regular sequence as regards the proportionate frequency of a different quantity in the fifth foot. In the Iti-vuttaka the fifth syllable as a different quantity from the sixth and seventh, whether both of those syllables be long or whether they be short, is nine times as frequent; in [321] the other works the proportions are respectively Dhammapada 8-1, Thera-Therãgàthà 7-1, Jàtaka 6-1. The Iti-vuttaka, therefore, has the strongest proneness thus to differentiate the fifth syllable, the Jàtaka the weakest.

Rare Vipulàs. As to the vipulàs, or second feet, we may first dismiss briefly the least common ones of the odd pàdas, viz., the third pæon or ionic a minore, ÛÛÜÅ [savipulà], the second or diiambus, ÛÜÛÅ javipulà], and the ionic a maiore or third epitrite, ÜÜÛÅ [tavipulà]. In Sanskrit, for example, the first of these occurs sporadically in all parts of the Mahàbhàrata, but is not found in the Ràmàyaõa. The same foot forms about 2% of the second feet in the Iti-vuttaka. The next close ÛÜÛÅ does not often occur in the epic èloka, and in Pàli it is very rare. The use of this vipulà more than anything else separates and distinguishes Pàli meter from the anuùñubh of the Veda, and the meter of the later Rig-Veda, which Arnold calls the epic anuùñubh. In anuùñubh it is the most common foot, forming the ending of the first and second pàdas indifferently. Its use as a close to the first (or third) pàda sinks in epic anuùñubh to one-half the frequency, and in Pàli and epic Sanskrit its employment is sporadic. With reference to the next vipulà, ÜÜÛÅ, we find that in the Pàli èloka it has an average occurrence of one-half of a per cent., about the same frequency in anuùñubh, is sporadic in the epic èloka, but in epic anuùñubh it forms 8% of the second feet.

The most common vipulà. The most frequently used second foot in Pàli is ÛÜÜÅ [pathyà]. It forms in the Iti-vuttaka, Dhammapada and Thera-Therãgàthà about 80%, in the Jàtaka about 70% of the endings of the first pàda. In the epic Sanskrit, it is also the prevalent ending, but in anuùñubh has but a frequency of one per cent. In all four Pàli works this vipulà is used oftenest with the first group of openings, or first feet, as shown in the table. The percentages of the use of this particular vipulà with the first group are Iti-vuttaka 40%, Dhammapada 41%, Thera-Therãgàthà 40%, Jàtaka 36%.

Other vipulàs. The next popular vipulà is the form ÜÜÜÅ, which most commonly follows a third epitrite ÜÜÛÜ, in the first foot. It is slightly more than half as common as ÛÜÜÅ after this opening, in the Iti-vuttaka, and slightly less than half as common in Thera-Therãgàthà and Jàtaka, but in the Dhammapada only one-fifth as common. This vipulà forms less than one per cent. of the [322] second feet in anuùñubh, about five per cent. in epic anuùñubh, but in epic èloka is fairly frequent. As to the vipulà ÜÛÛÅ, it is usually preceded by a diiambic or third epitritic opening ÅÜÛÜ. The third epitrite [ÜÜÛÜ] is the more common.

Opening feet. With regard next to the first feet of the odd pàdas, there exists much greater freedom than in the second feet. As said before, there are sixteen variations, and it is noteworthy that at least one example of each is found in the comparatively small compass of the Iti-vuttaka. The foot ÛÛÛÛ does not occur at all in the other works, and the varieties ÜÛÛÛ and ÛÛÛÜ are not in the Dhammapada.

The most common group of first feet in all four works is the first group, ÜÜÛÜ, ÜÛÜÜ, ÜÜÜÜ. Insignificant are differences in the four works as to which one of these is the favorite. Thus in Iti-vuttaka and Thera-Therãgàthà the second epitrite, ÜÛÜÜ, is of slightly greater frequency, while in Dhammapada and Jàtaka the third epitrite, ÜÜÛÜ, is a little in excess.

The second group of opening feet, ÛÜÛÜ, ÛÛÜÜ, ÛÜÜÜ, differs from the former group in having the initial syllable short, and we notice that this difference has a marked effect on the frequency of the opening, as is shown by the figures in the table. This preference for a long first syllable is much stronger in Iti-vuttaka than in the other three works. An interesting contrast may be made here between the Pàli èloka and the Vedic anuùñubh. In the latter the first syllable is anceps, whereas in Pàli, on examining all the sixteen varieties of openings, we find that a long first syllable is about twice as common as a short.

The next two groups of opening feet agree in having a short fourth syllable, and it has already been said that when such is the case, the fifth is usually short also.

Lanman, Sanskrit Reader, p. 300, states that in the Sanskrit epic èloka, the syllables 2, 3 and 4 in odd pàdas may not have the form of an anapæst, ÛÛÜ, or a tribrach, ÛÛÛ. In general this is true also in Pàli, but it is not invariable, since there are 11 examples of ÛÛÅ for the second, third and fourth syllables in the Iti-vuttaka, 13 in Dhammapada, 73 in Thera-Therãgàthà, and 47 in Jàtaka.

Even Pàdas.- Turning now from the odd to the even pàdas, we at once notice a remarkable difference in the character of the second foot, since it is here almost invariably of the form ÛÜÛÅ. In the Iti-vuttaka among 519 feet, only eight have not this [323] diiambic close; these eight verses are 15b, l8d, 20d, 20f, 73b, 75r, 85d, 105b.

The number of hypermetric even pàdas is smaller than in odd pàdas, since there are 12 hypermetric verses in even pàdas in Iti-vuttaka as compared with 24 in odd pàdas. These hypermetric lines will be treated later.

The opening foot of the even pàdas is variable, although not so greatly as the opening foot of the odd pàdas. I subjoin an analysis of the varieties of third foot in the Iti-vuttaka

A comparison of this table with the similar one in Simon's analysis of the other three Pàli works (p. 93), shows that the first group, in which all four feet agree in having a long third and fourth syllable, contains more than one-half the number of third feet in the Pàli èloka. The first syllable is more than twice as often long as short. The second syllable is anceps, with a slight predominance of longs, the longs being proportionately more common in the Iti-vuttaka than in the other three works.

The second group differs from the first in having the last syllable short. There is a preference again here for a long first syllable; the second is anceps, the long quantity being more numerous.

When the third and fourth syllables are short, as in group three, the second syllable is then long; there are but thirty-three exceptions to this rule in all of the four Pàli works combined, a total of 6422 lines.

Certain special rules as regards even pàdas may be discovered from the above table, as for example the absence of a succession [324] of four iambs [ÛÜÛÜÛÜÛÜ] in the Iti-vuttaka There are, however, three instances of this succession of syllables in the Dhammapada; twelve in the Thera-Therãgàthà; and twenty-one in the Jàtaka. Three iambs in succession are also rather uncommon [ÛÜÛÜÛÜ]; of this there are nine instances in the Iti-vuttaka, twelve in the Dhammapada, forty-three in the Thera-Therãgàthà, and seventy-one in the Jàtaka.

In the epic èloka (Lanman, Sanskrit Reader, page 300) syllables 2, 3 and 4 of even pàdas cannot form a tribrach, ÛÛÛ, an anapæst, ÛÛÜ, or amphimacer, ÜÛÜ. Such is not the case in Pàli, since the Iti-vuttaka, has nine examples of the tribrach, the Dhammapada four, Thera-Therãgàthà eleven, and Jàtaka nine; of the anapæst there are five examples in Iti-vuttaka, five in Dhammapada, fourteen in Thera-Therãgàthà, and seventeen in Jàtaka; of the amphimacer there are four in Iti-vuttaka, ten in Dhammapada, forty-one in Thera-Therãgàthà, and seventy five in Jàtaka

Hypermetric Lines.- A number of the èloka lines in the Iti-vuttaka are hypermetric [have extra syllables]. In itself this fact is not surprising, and the same phenomenon is found in Sanskrit. A good treatment of hypermetric verses in Sanskrit is found in Hopkins, op. cit. p. 252-261. None of the pàdas of the Iti-vuttaka are catalectic [have missing syllables], all of the hypermetric verses being, therefore, hypercatalectic. While there is usually one extra syllable in lines of this kind, we find four èloka lines of ten and one of eleven syllables. As stated before, even pàdas are less often hypermetric than odd pàdas; of the former there are thirteen (18f, 28l, 29j, 29l, 32f, 42h, 64h, 70h, 75b, 77f, 81f, 99d 112l) and out of these thirteen, one line is found three times (32f, 64h, 70h); of the latter, the odd pàdas, there are twenty-four hypermetric lines (l6c, 20i, 2li, 29i, 37a, 37g, 61i, 70e, 75e, 75m, 75o, 76y, 76a', 81a, 85a, 91a, 91c, 93k, 95i, 99k, 103i, 103s, 106i, 111g; of these 20i and 21i are the same). There is one instance (27k) in a passage, probably an interpolation, of a hypermetric pàda within a triùñubh-jagatã stanza.

As stated above, a line is not treated as hypermetric where the extra syllable is due to an epenthetic vowel. For example, such a line as

as hypermetric, since the second word is most likely a passive [325] ppl. of sakkaroti, Sanskrit sat kÔ, where the vowel a in Pàli is epenthetic. Similar vowels occur in the Avesta. Another derivative of kÔ is found in 103s, where the extra syllable is an epenthetic vowel.

A few lines are hypermetric, as Windisch has indicated in his introduction, p. viii, because the designation for some particular virtue, perhaps, a word of two or more syllables, is contrasted with the term for the corresponding vice, of three or more syllables. Seven lines (18f, 29i, 29j, 32f, 64h, 70h, 103i) are hypermetric from this cause. Thus the line

Triùñubh and Jagatã.- Of the 112 sections of the Iti-vuttaka mentioned in the opening of this paper, 98 were in èloka verse. The remaining fourteen are with three exceptions either in triùñubh or jagatã. The three exceptions are probably later interpolations or corruptions of the text, and are written in a mixture of èloka, triùñubh and jagatã. These are touched upon below. A pleasing variety is sometimes produced in triùñubh stanzas by the occasional introduction of a jagatã verse [line]. This occurs in 38f, 46d, 69c, 69e and 84l, while a sporadic triùñubh appears occasionally within a jagatã stanza, as in 87c and 98c. Alternation of the two meters is found in one passage, 47i-h, the first and third lines being in triùñubh and the second and fourth in jagatã. In the poetical portion of Æ 100 we find the first and fourth verses in jagatã and the two intervening verses in triùñubh.

A metrical examination of 27i-p shows that there is a rhythmical irregularity in the passage, as well as textual corruption, as indicated by faulty grammar. The stanza in question has seven lines, composed respectively in triùñubh, jagatã, hypermetric èloka, jagatã, triùñubh, jagatã, triùñubh. As will be seen from the notes on this passage in my forthcoming translation of the Iti-vuttaka, several lines are of very questionable genuineness, and for that reason none is included in the following statistics.

Triùñubh.- The commonest triùñubh line in epic Sanskrit, as pointed out by Hopkins, p. 275, has the form ÅÜÅÜ ÅÛÅÜ ÛÜÅ. There are twenty-nine lines of this kind in the Iti-vuttaka The first and last syllables are anceps. No example of a long third syllable is found; only two lines have a short fifth and only three a long seventh. The characteristic scheme of [327] the triùñubh verse in the Iti-vuttaka is, therefore, ÅÜÛÜ ÜÛÛÜ ÛÜÅ. The caesura is after the fourth or fifth syllables, slightly oftener after the former. Certain variations from this norm occur, such as

in which instance the first syllable may be taken as anacrusis [having an extra syllable at the beginning of the line], as the line is normal in other regards, although it may be noted that the second part has the rare form ÛÛÛÜ.

The line yo ca satimà nipako jhàyã (34g) may perhaps be treated best as a cataleptic triùñubh, with the irregular opening of a first , ÜÛÛÛ.

Jagatã.- Turning next to the jagatã, or line of twelve syllables, we find that here, too, the Iti-vuttaka follows a definite metrical scheme, which is represented

Of this type there are twenty-nine. The third and fifth syllables might be represented as common, but there is found a very strong preference for a short third and a long fifth. Twelve lines are different from this norm, or are hypermetric.

Caesura.- The caesura in the jagatã as in the triùñubh falls either after the fourth or fifth syllable, but the jagatã differs [328] from the triùñubh in having a preference rather for the caesura to fall after the fifth syllable.

It is possible also that in jagatã verse, as mentioned before in treating of the triùñubh, a short syllable may receive metrical lengthening if followed by the caesura, as for example in the line

even though it has the regular opening and close of a jagatã, contains in the middle portion a superfluous long syllable. There is no hint of a Ms. corruption, and we have no help from variant readings. Cf. Hopkins, p. 287 and p. 468.

Irregular jagatãs.- Certain lines in jagatã passages are neither normal nor hypercatalectic, as for example,

In both of these cases also we have no assistance from variant readings, and cannot, therefore, allege Ms. corruption as an explanation of the metrical difficulty. [Both of these lines have the Vedic opening ÜÜÜÜ]

Textual corruption.- There do exist, however, two or three stanzas in the Iti-vuttaka in which Ms. corruption is apparent. In these few cases, not only is one line irregular, but a longer succession of bizarre metrical effects is found. In 47f and g, for example, we find

Conclusion.- Pending a comparative study of the meters of the different Pàli works much more far reaching than has yet been made, nothing can be said regarding the relative age of stanzas of the Iti-vuttaka written in èloka, triùñubh, or jagatã.

All three of these meters are much more free than the corresponding rhythms in classical Sanskrit, as is to be expected. The èloka has changed in a marked degree from the Vedic type, yet it has at the same time distinct differences from the epic èloka. It is impossible to say whether the Pàli èloka is a direct outcome of Vedic imitation, and it is likewise impossible to postulate any connection or rapport with the later stages of [330] Sanskrit metrical development, as it might well be true that certain metrical preferences, for example, a long first syllable, are due to the idiosyncrasies of the language.

A more positive result of our analysis is the discovery that the eleven and twelve-syllable meters show less variety than the èloka. Each has in Pà1i a well-fixed form, with fully as much regularity of syllabic quantities as has epic Sanskrit (cf. Hopkins, p. 273-320), if indeed there is not even more. But this statement must not be applied to Pàli in general until many other works have been analysed. The fact that the Iti-vuttaka employs jagatã as often as triùñubh might seem to point to a late date, but whether this equal occurrence of triùñubh and jagatã is the result of chance or intention, no one can say.