Study 2: Description of the Metres


In this section I have attempted to collate all the relevant information on the metres used in the Udānavarga, and have compared the result with the figures for the Pāḷi Dhammapada.

It has to borne in mind that the Dhammapada is, of course, a much smaller collection than the Udānavarga and so the figures given here, except for the Śloka, are only indicative of the situation, but taken altogether they do, I believe, give a good overview of the relationship between the Pāḷi and BHS metres. It would have been better to compare all the verses with their Pāḷi parallels, of course, and this was my original plan, but as most of the verse texts published in Pāḷi, in both Eastern and Western editions, are unreliable, it would have required establishing the text of approximately 700 more verses to have come up with more reliable figures, a task which is beyond me at the present.01

The text of Udānavarga has 1070 verses, This number includes 25 verses of which there are 2 versions printed; and excludes the 5 uddāna verses, which have not been included in the analysis.02 having 4262 lines. They are written in the following metres:

Śloka (3,476 lines = 81%)
Triṣṭubh (329 lines = 8%)
Jagatī (73 lines = 2%)
Vaitālīya (232 lines = 5%)
Aupacchandasaka (170 lines = 4%) These are exactly the same 5 metres that are employed in the Pāḷi Dhammapada.03

Below I present an analysis of these metres as they are found in Udānavarga, together with a comparison of the verses in the Pāḷi Dhammapada.


1: Śloka

In the descriptions of the Śloka which occur in the prosodies See Piṅgala 5.10; Vttaratnākara 43; & Vuttodaya 116.04 it is stated that 2 light syllables in 2nd & 3rd position of the pādas should be avoided. However, as I have shown elsewhere, See the Introduction to the Prosody of the Dhammapada.05 in certain texts in Pāḷi this rule seems not to have been employed. Also occasionally in the Udānavarga it appears that the rule has been ignored, probably because the tradition from which the collection has been compiled also was willing to ignore it.

In the text established by Dr. Bernhard the opening occurs in the following places: 13b; 250d; 352a; 429c; 437d; 474ia; 566c. For a discussion of these readings, not all of which may be well based, see the notes in the commentary.

A sure sign of classical influence is in the avoidance of the opening ⏓−⏑−⏓ in the posterior line. This is in accordance with the rule found in the prosodies to avoid ragaṇa in this position. See Piṅgala 5.11.06 In this text there is not even one occurrence of this opening, though it is quite frequently found in Pāḷi.

If we bear these matters in mind we can then give an outline description of the Śloka. The most common form of the metre is the Pathyā, the pattern for which is:

odd line:


posterior line:



a: Variations

In this text I count 1,484 pādayuga-s as showing the Pathyā form of the metre.

There are then 7 variations (vipulā) in the odd lines found in this text, besides the normal (pathyā) structure, they are:



2a; 58a; 107c; 121a; 136c; 137c; 187a; 205a; 280iic; 302c; 332a; 341a; 342a; 343a; 361c; 399a; 415a; 451a; 458a; 477c; 482a; 483a; 543c; 579a; 590a; 591a; 651c; 658a; 660a; 665c; 667c; 678c; 693a; 740a; 746c; 758a; 762a; 771c; 774a; 776c; 810a; 859a; 860a; 878e; 882e; 907a; 909a; 964a; 965a; 991a; 1014.ic; 1026a; 1029c; 1031c (total = 55)



6c; 13a; 24a; 76c; 77a; 78c; 84a; 86a; 104a; 116a; 128a; 131a; 151c; 159c; 162c; 175a; 176a; 177a; 211c; 224a; 225c; 240a; This variation has resolution at the 5th syllable.07 241a; 244c; 245a; 246a; 247c; 272c; 282a; 286c; 291a; 333a; 337a; 338a; 418a; 431c; 434a; 442a; 446c; 468a; 561a; 565c; 572a; 580c; 585a; 587e; 651a; 705.iia; 711.iia; 749a; 812c; 818a; 880a; 882a; 891c; 895a; 898c; 904c; 906c; 960c; 961c; 972a; 1016a; 1017.iac; 1017.iia; 1038c (total = 67)



5c; 11a; 26c; 43a; 44a; 68a; 86c; 146c; 152c; 157c; 158c; 159a; 161c; 163a; 169a; 170a; 171a; 192a; 203e; 203e; 622a; 623a; 632a; & 633a (the last 4 lines being repetitions) all show resolution after the break at the 5th syllable.08 209a; 222c; 226c; 250a; 251c; 258a; 271a; 274a; 334a; 356c; 359e; 435c; 443c; 450a; 458c; 460a; 540c; 547a; 591c; 622a; 623a; 632a; 633a; 660c; 662a; 671c; 697a; 699c; 720c; 723c; 743a; 768a; 775a; 821a; 822a; 823a; 824a; 825a; 826a; 827a; 828a; 829a; 830a; 831a; 832a; 839c; 851c; 853c; 892a; 893a; 894a; 960a; 961a; 978a; 992a; 993e; 995a; 999c; 1008a; 1021.ic (total = 78)



10a; 15a; 54c; 61a; 84c; 120a; 160c; 230c; 231c; 235c; 236a; 344a; 357a; 358a; 478a; 541a; 579c; 584a; 722c; 748a; 781a; 790a; 849e; 862a; 978c; 986c; 987c; 988c; 989c; 994a; 996c; 997c; 1000c; 1003a; 1004a; 1005a; 1014.ia (total = 37)



102a; 237a; 607c; 718c; 970c; 974a (total = 6)



2c (total = 1)



56a; 79a; 162a; 283c; 345c; 474a; 578c; 598a; 602e; 733c; 1019.iia (total = 11)

There are also a number of 9 syllable lines as they stand in this text. They are found at 126c; 136c; 268c; 336c; 338c; 340c; 445c; 676c (total = 8).

There are therefore 1,742 pādayuga-s in all.


b: Comparison with the Pāḷi Dhammapada

We can extract the information found in the tables above and present it in terms of percentages, and then compare it with the figures for the Pāḷi Dhammapada:






86 % The percentages here and in the tables which follow are (for the most part) rounded up to the nearest integer.09

82 %



3 %

2 %



4 %

3 %



5 %

7 %



2 %

1 %



0.3 %

3 %



0.1 %




2 %

The differences between the Pāḷi and BHS form of the Śloka metre can then be enumerated as follows: Udānavarga excludes ragaṇa from the opening of the posterior line. There is much less prominence of savipulā; and a corresponding higher percentage of the pathyā form of the metre.

All of these changes bring the metre much closer to its Classical form, but still there are clear remants of the earlier form of the metre, most noticeably in the possibility of 2 shorts in 2nd & 3rd positions of the opening, the existence of Anuṣṭubh, and the high percentage of variation still allowed for. This falls away to about 5% in the Medieval period in both Pāḷi and Sanskrit.10 All of this points to a somewhat later form of the metre, but one that is far short of its Classical profile, and which still reflects quite closely what must have been the original form of the verses.


2: Triṣṭubh & Jagatī

The Triṣṭubh verses in this collection show a similar tendency as the Śloka - they are moving more towards the Classical form of the metre, while retaining certain early features.

The profile for the 11 syllable Triṣṭubh looks like this:


The 12 syllable Jagatī is basically the same metre with an extra light syllable in penultimate position, but in this collection at least there is no resolution at the 1st syllable, and no Vedic opening, so that it's profile can be described thus:



a: Mixing

One of the characteristics of this text is the high percentage of mixing of these two metres, which is a sign of the earliness of a lot of these verses.

There are verses in Triṣṭubh metre only at the following places: 27; 30; 33; 50; 51; 52; 53; 65; 97; 105; 106; 124; 125; 164; 181; 191; 216; 217; 218; 260; 266; 297; 298; 299; 306; 307; 397; 417; 475; 537; 563; 596.i; 596.ii; 637; 689; 690; 695; 696; 701; 702; 703; 765; 766; 767; 791; 795; 796; 808; 809; 842; 843; 844; 845; 855; 864; 877; 955; 956; 968; 969; 1040; 1041; 1042; 1043; 1044; 1045; 1046 (67 verses = 64%).

Jagatī only verses are found at: 45; 80; 81; 82; 259; 264; 267; 876 (8 verses = 8%).

Mixed Triṣṭubh & Jagatī verses: 28; 29; 98.i; 98.ii; 139; 180; 182; 185; 261; 262; 263; 265; 303; 331; 396; 455; 457; 476; 604; 659; 763; 764; 814; 1039 (24 verses = 23%)

Mixed Triṣṭubh & Śloka verses: 58; 60; 221; 541 (4 verses = 3%)

Mixed Triṣṭubh & Vaitālya: 797; 798 (2 verses = 2%)


b: Triṣṭubh openings

Nearly all the openings show the following form −⏑−, but there are a number which have the Vedic opening: ⏓−−− 52a; 58d; 216d; 218d; 331b; 417c; 596ia; 596iia; 637a; 956b; 968bd (12 lines = 3.5%), which is a sure sign of their early composition.


c. Triṣṭubh breaks I have seperated the statisics for the Triṣṭubh from the statisics for the Jagatī, as the latter is much more Classical in its profile.11

I count 329 lines in Triṣṭubh metre in this collection.

The most common form of the metre in this collection can be defined as conforming to the Classical Upajāti, having the bhagaṇa break:

−⏑⏑27abcd; 28bcd; 29cd; 30abcd; 33bcd; 50abcd; 51bd; 52bd; 53ac; 58c; 60a; 65ad; 98.icd; 98.iic; 105bc; 106bc; 124abcd; 125bcd; 164acd; 180b; 181abcd; 185abc; 191a; 216abcd; 217cd; 218abc; 221a; 260abcd; 261bc; 262ab; 263c; 265e; 266abc; 297abcd; 298abcd; 299abc; 303bcde; 306d; 307acd; 331a; 396bcdf; 397bcd; 417abcd; 455bcd; 457bcdef; 475abd; 476b; 537acd; 541cd; 563abcd; 596.ibcd; 596.iibcd; 604bcd; 637bcd; 659bc; 689abcd; 690abcd; This verse is in perfect Upendravajrā metre, having the break −,⏑⏑ throughout.12 695abcd; 696abc; 701abd; 702abcd; 703abcd; 763d; 764bd; 765ad; 766abcd (Indravajirā); 767abcd; 791abcd (Indravajirā); 795abcd; 796abcd; 797cd; 798cd; 808abc; 809abcd; 814d; 842abcd; 843abc; 844abcd; 843abd; 855acd; 864abcd; 877abcd; 955abcd; 956ac; 968ac; 969abcd; 1039abd; 1040acd; 1041acd; 1042ac; 1043ac; 1044ac; 1045acd; 1046acd; 1047acd (270 lines = 82%).

A further 66 lines have a variant break, the most popular being ragaṇa:

−⏑− 51ac; 52ac; 53bd; 58d; 191c; 217b; 218d; 263b; 265d; 266d; 696d; 331bd; 596.ia; 596.iia; 637a; (there seems to be a connection between the Vedic opening and this break); 701c; 763b; 765c; 814b; 843d; 968bd; 1040b; 1041b; 1042bd; 1043bd; 1044bd; 1045b; 1046b; 1047b (37 lines = 11%)

⏑⏑− 106a; 125a; 139c; 164b; 299d; 303f; 306ab; 307b; 659d; 855b; 956b (12 lines = 3.5%)

⏑⏑⏑ 33a; 65b; 397a; 537b; 956d (5 lines = 1.5%)

−−− 105d; 191bd (3 lines = 1%) These breaks are very unusual as having a heavy syllable in 6th position. This sometimes occurs in Pāḷi, but normally after a break at the 5th syllable, which doesn't occur in these lines.13

−,− 306c; 763c; 764c; 808d; 843c (4 lines = 1.5%) In line with the rule of replacement (see An Outline of the Metres in the Pāḷi Canon 1.16) there is always a word-break before the presumed replacement of 2 light syllables at the 6th in this break.14

−,⏓⏓⏓ (Extended Triṣṭubh, pausing at the 5th, and restarting from the same position): 53b; 106d; 217a (3 lines = 1%).

−−⏑ 475c (1 line = .3%)


d: Jagatī breaks

The Jagatī lines, of which I count 73, are in all respects much more Classical than the Triṣṭubh lines - there are no Vedic openings; resolution is uncommon; But not unknown, see 139ab, for resolution at the 5th.15 and all but one of the breaks are bhagaṇa:

−⏑⏑ 29ab; 45abcd; 80abcd; 81abcd; 82abcd; 98.iab; 98.iiabd; 139abd; 180acd; 185d; 259abcd; 261ad; 262cd; 263ad; 264abcd; 265abcf; 267abcd; 303a; 331c; 396ae; 455a; 457a; 476acd; 604a; 659a; 763a; 764a; 814ac; 876abcd; 1039c (72 lines)

−⏑− 28a.


e: Comparison with the Pāḷi Dhammapada

If we add the figures for Triṣṭubh and Jagatī together we can compare the figures with the Pāḷi Dhammapada, and present it in a table thus:










Triṣṭubh & Jagatī



Triṣṭubh & Śloka



Triṣṭubh & Vaitālya





Vedic Opening






−⏑⏑ break



−⏑− break



⏑⏑− break



⏑⏑⏑ break



other breaks




We have to bear in mind that the Pāḷi Dhammapada has only 37 verses in these metres, so that the statistics are not so reliable, however we can roughly say that as with Śloka, that while the Triṣṭubh lines retain certain early characteristics like the Vedic opening and a willingness to allow Jagatī lines into the verse; the very much greater proportion of the bhagaṇa break points to a later stage of composition than the Pāḷi.


3: Vaitālīya & Aupacchandasaka

The Śloka and Triṣṭubh / Jagatī metres are syllabic metres, that is, their fundamental structure is defined by how many syllables there are in the line. The last 2 metres organise their lines according to the amount of measures (mātrā) there are in the line. A light syllable has one mātrā; a heavy one two.

In the first of these metres Vaitālīya, there are 14 mātrā in the odd line and 16 in the posterior, and the line has the following distinctive cadence ¦−⏑−⏑×. Aupacchandasaka is similar but has 16 mātrā is the odd line and 18 in the posterior, owing to the cadence including an extra heavy syllable ¦−⏑−⏑−×.

The openings of these metres is the same ⏔⏔⏔ in the odd line, and ⏔⏔⏔⏔ in the posterior.


a: Composition

There are Vaitālīya lines in the following verses: 66abcd; 67abcd; 69abcd; 70abcd; 71abcd; 89abcd; 90abcd; 183abcd; 184abcd; 186c; 215abcd; 284abcd; 285abcd; 305abcdef; 333ef; 364abcd; 366abcd; 367abcd; 368abcd; 370abcd; 371abcd; 595abcd; 638abcd; 646abcd; 647abcd; 648abcd; 649abd; 650abcd; 652abcd; 653abcd; 680abcd; 681abcd; 682abcd; 683abcd; 684abcd; 685abcd; 686abcd; 687abcd; 688abcd; 691abcd; 692abd; 704abcd; 752acd; 753abcd; 755abcd; 756a; 757abcd; 797ab; 798ab; 840abcd; 863abcd; 867abcd; 868abcd; 869abcd; 870abcd; 871abcd; 872abcd; 873abcd; 874abcd; 875abcd; 879abc; 926a; 949a (= 232 lines).

Aupacchandasaka lines: 186abd; 354abcd; 388abcd; 389abcd; 390abcd; 391abcd; 392abcd; 393abcd; 394abcd; 542abcd; 564abcd; 649c; 692c; 751abcd; 752b; 756bcd; 794abcd; 841abcd; 879d; 926bcd; 927abcd; 928abcd; 929abcd; 930abcd; 931abcd; 932abcd; 933abcd; 934abcd; 935abcd; 936abcd; 937abcd; 938abcd; 939abcd; 940abcd; 941abcd; 942abcd; 943abcd; 944abcd; 945abcd; 946abcd; 947abcd; 948abcd; 949bcd; 950abcd; 951abcd; 952abcd; 954ad ( = 170 lines).

Vaitālīya and Aupacchandasaka lines occur in 99 verses

Pure Vaitālīya = 51 verses

Pure Aupacchandasaka = 38

Vaitālīya mixed with Aupacchandsaka = 8 (186; 649; 692; 752; 756; 879; 926; 949).

Vaitālīya mixed with Śloka = 1 (333).

Vaitālīya mixed with Triṣṭubh = 2 (797; 798).


b: Openings

We can take the openings for the two metres together for the purposes of analysis. In the odd lines the following openings are found:


i. odd lines

There are 194 odd lines in the Vaitālīya and Aupacchandsaka verses, 16 of which are irregular as they stand in the text. Many of the irregular lines could easily be regularised by following the suggestions in the notes.16 The various openings are analysed in the tables below.

a. regular (92%):


70c; 71a; 89a; 90c; 215c; 285a; 333c; 354a; 364c; 367c; 368a; 371a; 388a; 389a; 390a; 391a; 392a; 393a; 394a; 595c; 564a; 638ac; 648c; 649ac; 650c; 652c; 680a; 681a; 682a; 683a; 686a; 687a; 688a; 691c; 692c; 751c; 752c; 753c; 755c; 757c; 794a; 841c; 863c; 867ac; 868ac; 869ac; 870ac; 871ac; 872ac; 873c; 874c; 879c; 926c; 927a; 928a; 929a; 930a; 931a; 932a; 933a; 934a; 935a; 936a; 937a; 938a; 939a; 940a; 941a; 942a; 943a; 944a; 945a; 946a; 947a; 948a; 951a; 951a (85 lines = 44%)


69ac; 90a; 183c; 184c; 186ac; 215a; 305e; 364a; 366a; 367a; 368c; 370c; 388c; 389c; 390c; 391c; 392c; 393c; 394c; 564c; 646ac; 647a; 653c; 684ac; 685ac; 691a; 704ac; 756c; 794c; 797a; 798a; 875c; 927c; 928c; 929c; 930c; 931c; 932c; 933c; 934c; 935c; 936c; 937c; 938c; 939c; 940c; 941c; 942c; 943c; 944c; 945c; 946c; 947c; 948c; 949c; 950c; 951c; 952c (64 lines = 33%)


595a; 752a; 756a; 863a; 926a; 949a (6 lines = 3%)


284a; 354c; 757a; 542ac; 950c (6 lines = 3%)


753a; 755a; 873a; 874a; 751a (5 lines = 2.5%)


66c; 67c; 285c; 305ac (5 lines = 2.5%)


366c; 840ac; 954a (4 lines = 2%)


370a; 371c (2 lines)


652a (1 line)

b. irregular (8%)


89c; 875a (2 lines)


647c; 653a (2 lines)


183a; 650a (2 lines)


680c; 681c; 682c; 683c; 686c; 687c; 688c (7 lines)


184a (1 line)


648a (1 line)


70a (1 line)


ii. even lines

There are 203 odd lines in the Vaitālīya and Aupacchandsaka verses, The discrepency between the number for the odd lines (194) and even lines (203) is accounted for by even lines occuring where we would expect to find odd lines.17 18 of which are irregular as they stand in the text.

a. regular (91%):


67b; 69d; 71b; 89bd; 90bd; 183d; 184b; 186b; 215b; 305b; 333d; 354bd; 364b; 367bd; 368bd; 388d; 389d; 390b; 390b; 392b; 393b; 394b; 564d; 595bd; 647bd; 649d; 650d; 685b; 704d; 751d; 753d; 755d; 757d; 840d; 863d; 873d; 874d; 926d; 927d; 928d; 929d; 930d; 931d; 932d; 933d; 934bd; 935bd; 936bd; 937bd; 938bd; 939bd; 940d; 941d; 942d; 943d; 944d; 945d; 946bd; 947d; 948bd; 949d; 950d; 951d; 952b (79 lines = 38%)


66d; 69b; 70d; 284bcd; 285b; 305f; 364d; 366b; 370d; 371bd; 388b; 542d; 564b; 638d; 646bd; 652b; 653bd; 684d; 686d; 687d; 688d; 691bd; 692ab; 704b; 751b; 753b; 755b; 756bd; 757b; 794bd; 841abd; 863b; 867bd; 868b; 869b; 870bd; 871b; 872b; 875bd; 879d; 926b; 947b; 949b; 950b; 951b; 952b (60 lines = 30%)


305d; 370b; 389b; 390b; 391b; 392b; 393b; 394b; 928b; 929b; 930b; 931b; 932b; 933b; 940b; 941b; 942b; 943b; 944b; 945b (20 lines = 10%)


183b; 184d; 186d; 366d; 638b; 797b; 798b (7 lines = 3.5%)


650b; 868d; 869d; 871d; 872d (5 lines = 2.5%)


67d; 840d (2 lines)


649b (1 line)


542b (1 line)


927b (1 line)


879a (1 line)

−⏑−⏑⏑− This and the next line are included with the regular openings because, as I have shown in the Introduction to the Prosody of the Dhammapada, they seem to have been tolerated by the bhāṇaka-s, and the 1st syllable must be counted as light, a phenomena I suggested calling pādādilahu.18

680d; 681d; 682d; 683d (4 lines = 2%)


66b; 70b; 215d; 684b (4 lines = 2%)

b. irregular (9%):


680b; 681b; 682b; 683b; 686b; 687b; 688b (7 lines = 3.5%)


873b; 874b (2 lines)


66a; 67a (2 lines)


648b (1 line)


648d (1 line)


685d (1 line)


879b (1 line)



285d (1 line)


652d (1 line)


692d (1 line)


As can be seen from this there are more problems in the metre in these verses than in the syllabic metres, probably because the metre quickly fell into disuse. The metre was not properly understood in the Pāḷi tradition either.


c: Comparison with the Pāḷi Dhammapada

We can once again extract this information and compare it with the statistics for the Pāḷi Dhammapada: Once again the statistics are very low for the Pāḷi form of the metre as there are only 32 verses to compare. In the table I only show the most common forms and the irregular forms.19


Odd lines





















even lines:



⏑⏑−−⏑⏑ It is, I believe, significant that the secondary form of this line −⏑−−⏑⏑ occurs in both texts.20















After a study of the metres in the Pāḷi Canon See my An Outline of the Metres in the Pāḷi Canon 2.11.21 I have stated that it is a sign of the more recent age of the Matrācchandas texts that the opening −−⏑⏑ in the odd lines gradually declines and the opening ⏑⏑−⏑⏑ becomes more frequent in the later texts, the figures given here would seem to indicate once again that the Udānavarga is later than the Pāḷi, though as with the other metres they still retain early forms, which would later be excluded from the Classical Sanskrit forms of the metres.


d: Conclusions

From the study of the metres made in this Introduction it would appear that all the evidence goes to show that the metres contained in the Udānavarga as established by Dr. Bernhard are substantially the same as the metres in the Pāḷi canon itself, Which itself looks back, of course, to what must be an even earlier form of the metres.22 though at a rather later stage of development.

The verses where they have been rewritten seem to have been Sanskritised, but the bulk of the work often contains remnants of the earlier features of the metres concerned. In this respect therefore the metres closely parallel the Sanskritisation of the language itself.

Edgerton writing about the phonology and morphology of BHS stated that: BHS works, especially the oldest, retain in all parts clear evidences of being based on some form of Middle Indic, only partially, and it seems haphazardly, Sanskritized (BHSG 1.34). This study would seem to suggest that exactly the same sentiments could be applied to the prosody also.