Introduction to the Prosody of the Dhammapada


1: The Text

The text of the Dhammapada as presented here has been established through a comparison of the four standard printed editions:

BJT: Dhammapadapāḷi. Buddha Jayanti Tripitika Series, volume XXIV. Colombo 1960.

PTS: Dhammapada. Edited by O. von Hinüber and K. R. Norman. Pali Text Society, Oxford, 1994.

Thai: Dhammapadagāthā. The Royal Thai Edition, volume 25. Originally published 2469 (i.e 1915), reprinted Bangkok, 2500 (i.e 1956).

ChS: Dhammapadapāḷi. Chaṭṭha Sangāyana Edition, 1956, reprinted Rangoon 1972.

I have also consulted the following texts for comparison of the readings (but have not entered the variants in the notes):

The Dhammapada. A new edition by Sūriyagoḍa Sumaṅgala Thera, Pali Text Society, London 1914.

The Dhammapada. Edited by Nārada Thera (4th Edition). 1993, reprinted Taiwan, 1999.

Dhammapadaṭṭhakathā. Edited by Kahave Siri Ratanasāra Thera & Mahagoḍa Siri Ñāṇissara Thera. Simon Hewavitarne Bequest vol V. 1919; vol XIII, 1922 (= parts I & 2), reprinted Colombo 1991(?)

For the prosody I have consulted:

WD: The Word of the Doctrine (Dhammapada). Translated with an introduction and notes by K. R. Norman. Pali Text Society, Oxford, 1997.I would like to note here that throughout this work I have had the opportunity to consult with Prof. Norman on various points to do with the establishment of the text, and he has always answered most courteously and promptly - I am very grateful to him for all the help he has given. The debt I owe to his written works, of course, should be evident on every page. References to WD are always to the note to the verse concerned, unless otherwise stated.

PM: Pali Metre. A. K. Warder. Pali Text Society, London, 1967.

When I started preparing the Dhammapada for the Sri Lanka Tipitaka Project, I really had no intention of re-establishing the text, but as the work progressed, and I grew more familiar with the Buddha Jayanti Tripitaka edition (BJT), it became clear that there were many problems in the text, including non-standard forms, Sanskritisation, and unmetrical readings. I therefore began a comparison of BJT with the other main editions and started entering the variant readings found therein. During the process of that work it became clear that none of the other texts were quite satisfactory either, and what had started out as a simple re-presentation of the text as it is found in the Sri Lankan tradition, has finished up requiring a complete re-appraisal of the text, especially from the point of view of its metre.

As with any verse text, of course, it is essential to understand the prosody that underlies the composition, otherwise the likelihood is that wrong readings will find their way into the text. This should not be understood as implying that every verse should be rigidly conformed to a standard metre. Indeed it is such unskilful editing that is one of the faults in the main editions available to us. To understand the prosody correctly, one must know not only what is its standard pattern, but also what amount of deviation is allowed from the norms that otherwise prevail. And to understand that we must let the texts themselves be our guide.


2: Preliminaries

1: Resolution

A normal feature of Pāḷi prosody is the resolution of one presumed heavy syllable into two light ones.In this text syllables in resolution are normally coloured green and are underlined for the sake of identification. As I have explained elsewhereSee my Outline of the Metres in the Pāḷi Canon 1.15. there is an underlying rule regarding resolution which states that only the first two syllables in a word may be resolved (including syllables that appear as the second half of a compound, or after a prefix).There is one exception to this, in that it appears from the texts themselves that na, when it precedes the word it modifies may form the first half of a resolved syllable. This is no doubt because of the close syntactic proximity of the negative with the word it is modifying. Examples of resolution including the negative can be seen in the text at 131d; 291d; 302f; 364d. With the help of this rule we are able to identify more accurately the underlying variation, especially in the Siloka prior lines.The general usefulness of this rule may be emphasized by noting that with its help we can certainly identify the variation involved in the following pādas: 8e; 14bd; 21a; 47a; 48a; 182a; 183a; 183b; 185a; 223d; 228a; 231d; 232d; 248a; 283c; 292a; 302a; 333c; 389a; 414a; 418a - whereas otherwise we would have no way of correctly identifying the variation. In the text as here established I count resolution 61 times. 58 of these comply with the rule outlined above, and there are only two examples where the rule may not hold, the first at 27a, where the line in question is the opposite of a line in the preceding verse - something which often disturbs the prosody; and at 137c.

As can be seen from the table below the syllable most liable to resolution is the 1st, and in the Tuṭṭhubha/Jagatī verses that occur in the Dhammapada it is the only resolution found (marked in red in the table - all the rest of the resolutions occur in Siloka lines).Because of the nature of the prosody we do not normally count resolution as occuring in the mattacchandas verses (Vetālīya/Opacchandasaka).



(total: 61 instances)


(23 instances): 8a; 20e; 40b; 51d; 52d; 73a; 99a; 108d; 125d; 126d; 140d; 153c; 183b; 185a; 227e; 271c; 302f; 307d; 328c; 346d; 347d; 411c.


(17 instances): 8e; 9b; 14bd; 74b; 131d; 185a; 223d; 228a; 231d; 232d; 233d; 248a; 291d; 302f; 333d; 364d; 414a.


(2 instances): 21a; 27a.


(11 instances): 47a; 48a; 181c; 182a; 183a; 275a; 283c; 292a; 302a; 333c; 418a.


(3 instances): 246c; 387c; 409c.



We may note here that resolution is found in two places within the same line in the following lines: 185a (1st & 5th); 302f (1st & 4th).


2: Replacement

The compliment to the rule of resolution is the rule of replacement, which states that when 2 presumed light syllables are replaced by a heavy one, it is always the first two syllables in a word that are presumed to be light.

In this edition of the Dhammapada there is replacement in line with the rule at 40c & 125c. As with resolution, the application of the rule of replacement can also help us to identify the underlying structure of the verse, in a way that has not been recognised before. The reading at 19d (together with its repetition at 20f) has a Tuṭṭhubha line with the following structure:

Sa bhāgavā sāmaññassa hoti.

In discussing lines like this it has hitherto been thought that there is replacement of two shorts syllables in 6th & 7th position by one heavy (as is, indeed, normally the case).See Norman’s comments in his notes to these verses in The Word of the Doctrine. However if we accept the rule, we can see that it is in fact the 5th & 6th syllables that have been replaced, and the underlying structure of the break is ,⏑⏑−. This shows once again how helpful the discovery of these rules has been to understanding the correct prosody of the texts.


3: Sarabhatti

We must also include here the sarabhatti vowels (or epenthetic vowels) which have been written in the text, but which were not pronounced; these can sometimes look similar to resolution, but they can be easily identified as they nearly always involve two of the semi-vowels and/or the aspirate.Rarely we find other combinations; in this text vajiraṁ, and elsewhere others like nahāru, kilesa, etc. are found. Sarabhatti is found in the following 49 places:

arahati 9d, 10d, 230b

-arahanto 98c

ariya- 79c, 190c

ariyaṁ 208b

ariyañ- 191c

ariyānaṁ 22d, 164b, 206a

issariyaṁ 73c

kadariyaṁ 223c

kadari 177a

kayirañ 313a,

kayirati 292b

kayi 42a, 43a, 53b, 61c, 105c, 117ab, 118a, 159a, 281b, 330c

kayirātha 25c, 117c, 118b, 211a

garahito 30d

-cariyaṁ 61c, 155a, 156a, 312c

-cariyavā 267b

-cari 141a, 388b

payirupāsati 64b, 65b

-pariyodapanaṁ 183c

pariyodapeyya 88c

pahareyya 389a

rahado 82a, 95c

vajiraṁ 161d

viriyam 112d

viharantaṁ 8a

Note that occasionally in words that have sarabhatti vowels the vowel must have been pronounced and given its full value as in the following words, where it is necessary to count the vowel towards the metre:

arahataṁ 164a, 420c

ariyo 270ad

kayirātha 118c, 313a

viharantaṁ 7a

viharanti 98c

viharāma 197d, 198d, 199d


4: Changes in word form

Because of the need to meet the requirements of the metre, which demands syllables of a specified length in certain places to produce the rhythms which make up a verse, there are certain changes which take place in word form.It should be noted that in the lists that follow an attempt has been made to collect all the words that have been changed in the text metra causi. In the notes to the text itself there is no attempt to be comprehensive (which would only multiply the notes without good reason). There changes in word form are normally only discussed when there is need to explain why I have taken the reading in the text in preference to a variant reading. As can be seen from the following lists it is the lengthening of i > ī, especially at the end of a word, that is by far the most common change that occurs.

long ī at the end of a word (36 instances):

vuṭṭhī 13b; 14b

munī 49d

ramatī 99b; 116d

vijjatī 127c; 128c

khantī 184a

jayatī 193c; 212ab; 213ab; 214ab; 215ab; 216ab; 282a; 283b

sukhī 206d

nayatī 257b

āpajjatī 309b

gatī 310a

ratī 310b; 373b

palavatī 334c

sahatī 335a; 336a

chattiṁsatī 339a

sabbadhī 340a

bhāvayatī 350b

dhammaratī 354c

labhatī 374c

sucī 393d

long ī medially (14 instances):

-vīriyaṁ 7d; 8d

satīmato 24a; 91a

-vīriyo 112b

hirīnisedho 143a

vīriyena 144c

icchatī 162d

sukhī 177d

satīmataṁ 181d

tatīyaṁ 309d

satīmā 328d

khantībalaṁ 399c

nandībhava- 413c

other vowels lengthened medially (6 instances):

kasām 143d

ajjatanām 227b

-sārīro 352e

anūpalitto 353b

-sārīraṁ 400c

nirūpadhiṁ 418b

shortening of vowels (6 instances):

okata (from okato) 34b

va 138d; 139bc; 195b; 409a

attanaṁ 355d

consonants doubled unhistorically (6 instances):

suggatiṁ 18d; 319d

-ppacessati 44d; 45d

kaṭukapphalaṁ 66d

-cchecchati 350d

consonant groups simplified (4 instances):

dukhena 83c

dukhā 186c; 203b; 302b

niggahīta dropped (13 instances):

kata' 74a

yesa' 92d

paṭivadeyyu' 133b

phuseyyu' 133d

lokasmi' 143b

macchāna' 182b

buddhāna' 184d; 185f

addhāna' 207b

devāna' 224d

vissāsa' 272c

vanasmi' 334d


3: The Metres

There are 1733 lines in the Dhammapada, written in five metres, they are:

1. Siloka (Skt: Śloka, a.k.a. AnuṣṭubhIdentified as Anuṭṭhubha (Śloka) by Norman in The Word of the Doctrine (against his normal practice of calling the metre Śloka). However the Anuṭṭhubha is a samavutta metre; the Siloka (or Śloka) is addhasamavutta, and they shouldn't be confused.) 1482 lines, 86%

2. Vetālīya (Vaitālīya) 94 lines, 5%

3. Opacchandasaka (Aupacchandasaka) 11 lines, 0.5%

4. Tuṭṭhubha (Triṣṭubh) 120 lines, 7%

5. Jagatī (Jagatī) 26 lines, 2%



By far the most common metre in the canon, and in the Dhammapada in particular, is the Siloka which accounts for some 86% of the verses found in the collection.I count only one verse (No 330) as being mixed. The following are Siloka verses (giving a total of 352 verses, or 741 pādayugas): 1-14, 21-23, 25-39, 41-43, 47-53, 55-79, 81-82, 85-93, 96-107, 109-124, 126, 129-140, 146-149, 152-176, 178, 181-183, 185-207, 209-220, 222-234, 239, 241-279, 282-283, 286-305, 307-308, 311-323, 327, 330, 332-333, 335-337, 339-340, 351-352, 355-361, 363-370, 372-387, 389, 391-423. The Siloka is an addhasamavutta syllabic metre, which means it has two dissimilar lines which make up a pādayuga; two pādayugas (or occasionally three) make a verse. There are normally 8 syllables to each half of the pādayuga (pair of lines).Occasionally we find 9 syllables if one is resolved; or, more rarely, 10 if 2 are resolved in the same half of the pādayuga - for the latter see 8a (1st & 6th); 185a (1st & 5th); 302f (1st & 4th).

The Siloka has a pathyā (normal) structure and 7 variations. The normal structure can be described as follows:In what follows = a light syllable; = a heavy one; = anceps, the syllable can be either light or heavy; = one light, or one heavy, or two light syllables.

Odd line:


Even line:

⏓⏓⏓¦⏑−⏑⏓ x 2

In the text as it has been established here there are pathyā lines in 82% of the Siloka pādayugas.For the references see the tables in the Index of Metres. Another way to emphasise how high this percentage is, would be to note that there are only 7 Siloka verses in the whole collection that do not have pathyā lines in them. Nos 23, 69, 87, 183, 196, 274, 303.

7 variations occur in the first half of the pādayuga. They have the following structure (with the percentage of their occurrenceNote that percentages are approximate only.):


⏓⏓⏓¦⏑−⏑⏓ 2%


−⏓−¦⏑⏑⏑⏓ 2%


−⏓−¦−⏑⏑⏓ 3%


−⏓−¦−−−⏓ 7%


⏓⏓⏓¦−⏑−⏓ 1%


⏓⏓⏓¦⏑⏑−⏓ 3%


−⏓−¦−−⏑⏓ (very sporodic)

There are also 8 lines that are irregular.27a; 116c; 218c; 222c; 260a; 266a; 274c; 315a. These are commented on in the text.

Verse no 150 is unusual because, as it stands, it is in Anuṭṭhubha metre, but whether we should count this as a separate metre, or as Siloka with 2 Anuṭṭhubha variations in the prior line is not sure.

It is well established that in the opening of the Siloka metre, 2 shorts are normally avoided in 2nd and 3rd positions. That this is so can simply be proved by reference to the texts themselves, and noting that the number of times this opening appears is very much less than a random choice could possibly entail. Besides this though, we can also see that changes are regularly made to word forms in order to avoid the opening.

When we examine the Dhammapada itself we can see that changes have been made in the following words in order to avoid the unwanted opening: 13b, 14b (vuṭṭhī); 143a (hirīnisedho); 245a (hirīmatā); 269b (munī); 374c (labhatī); 399c (khantībalaṁ); 413c (nandī-). This pattern is repeated in many of the Siloka texts found in the canonical works.

However, on closer inspection we can see that there are a significant number of occasions in the Dhammapada when the pattern ⏓⏑⏑⏓¦ does in fact turn up in the opening, sometimes even when it would be easy to ‘correct’ the metre if the redacters had felt that it was wrong. Note that there are no ‘correcting’ variants at all recorded for the following lines, where this ‘wrong’ opening occurs: 3a; 4a; 68c; 99b; 148c; 265a; 268c; 363a; 375c; 420a; 421a. That is 11 times, and is more than the number of corrections m.c. that are listed above. Because of this I have not felt obliged to correct this opening when it occurs, if there is not good manuscript evidence for it, and have therefore let it stand at 1c; 2c; 121e; 140b; 382b; 415c. In all 7 instances there are metrically more correct readings found in the PTS edition, but they are for the most part based on the slim evidence of just 2 old Thai manuscripts, which can be shown to have a habit of ‘correcting’ the metre, when they felt it to be wrong.


Vetālīya & Opacchandasaka

The most popular of the mattacchandas (measure) metres in the Dhammapada is the Vetālīya. Whereas the Siloka, Tuṭṭhubha & Jagatī metresThe latter two are discussed below. are syllabic metres, and organise their lines by counting the syllables, in the mattacchandas metres it is the total number of measures (mattā) to the line which is the determining factor. In Vetālīya the prior lines normally have 14 measures, and the posterior lines 16. With both the cadence is the same: −⏑−⏑⏓. There are 105 lines in 30 verses in this metre, which is roughly 5.5% of the total number of verses.

There are only two verses in the Opacchandasaka metre, Nos 184 & 371, and some odd lines that turn up in what are otherwise Vetālīya verses.179a; 342d; 344a; 362d. This metre normally has 16 measures in the prior line, and 18 in the posterior. The cadence is similar to Vetālīya, with an extra, heavy, syllable in penultimate position: −⏑−⏑−⏓.

The description of the openings is the same for both metres, so they will be treated together here.

In the prior lines the openings are:

−−⏑⏑ (sometimes syncopated to −⏑−⏑), 33 instances (= 52%).15c; 16c; 17c; 18c; 24a; 44c; 45a; 80c; 145c; 179c; 180ac; 184a; 235c; 236ac; 238ac; 240c; 284a; 285ac; 324c; 341c; 342c; 343c; 344ac; 348c; 362c; 371ac; 388c.

⏑⏑−⏑⏑ (sometimes syncopated to ⏑−⏑⏑⏑) 17 instances (= 26%).15a; 16a; 17a; 18a; 44a; 80a; 95c; 145a; 184c; 237a; 240a; 284c; 324a; 341a; 342a; 343a; 350a.

In 4 places we find this opening −⏑⏑− (179a; 235a; 348a; 388a); once −⏑⏑⏑⏑ (284a); once ⏑⏑⏑⏑− (95a); and once −−− (349c). In this text I count the following prior lines as irregular: 24c; 45c; 237c; 334c; 349a; 350c; 362a.These are commented on in the notes to the text.

In the posterior lines the openings are:

−−−⏑⏑ 24 instances (= 36%).15d; 16d; 17d; 18d; 45c; 80d; 95d; 145d; 180b; 184b; 235d; 236b; 237d; 238b; 284d; 285d; 334b; 341d; 344d; 348b; 350d; 362d; 371b; 388d.

⏑⏑−−⏑⏑ 24 instances (= 36%).16b; 18b; 24b; 44bd; 45bd; 80b; 145b; 179d; 180d; 184d; 238d; 240d; 284b; 285b; 334d; 341b; 342b; 343b; 344b; 348d; 350b; 388b.

In 2 places we find this opening ⏑⏑⏑⏑−⏑⏑, which looks like a variation of the 2nd opening above (235b; 324d); once −⏑⏑−⏑⏑We should perhaps regard this as a syncopated variant of the second opening. (342d); and once ⏑−−⏑− (179b). There are 5 irregular lines: 236d; 240b; 324b; 343d; 362b.These are discussed in the notes to the text.

The first of the openings listed above seems to have a secondary form, with a heavy syllable where we would expect to find a light one, giving the form −⏑−−⏑⏑. This opening occurs with such frequency,15b; 17b; 24d; 95b; 237b; 349b; 349d; 371d. The last two of the lines listed here are syncopated. Possibly 343d should be regarded as an example of this variation also, see the note to that verse. that there can be no doubt that the redacters felt that it was an acceptable variation. Various theories have been suggested to account for this variation,See Warder, PM, pg 121ff. for an assessment of the various ideas put forward. but upto now none of them are really very satisfactory.In The Word of the Doctrine Norman makes various suggestions on how we could regularise these lines to give a normal opening, but it is worthwhile pointing out here, that a number of the changes suggested there are unusual and that not one of the suggestions has any manuscript support for it.

It is perhaps worth noting here that in verse, the last syllable in the line is always counted as heavy, no matter what its real length is. We could perhaps suggest that this phenomena, which is known as pādantagaru, may well find its compliment here, where the first syllable sometimes has to be counted as light m.c., no matter what its true length is. We could perhaps call this phenomena pādādilahu, which would also be complimentary to the phenomena of pādādigaru, which is found in gaṇacchandas verses.


Tuṭṭhubha & Jagatī

Tuṭṭhubha is a syllablic samavutta metre normally having 11 syllables to the line,Sometimes, because of resolution, we find a Tuṭṭhubha line with 12 syllables: 20e; 40b; 108d; 125d; 328c; 346d; 347d. In each case the resolution is of the 1st syllable giving the opening: ⏑⏑−⏑−. and 4 lines to the verse. Approximately 7% of the verses in the Dhammapada are in this metre. The shape of the Tuṭṭhubha line can be defined thus: ⏓−⏓−¦⏓⏑⏒¦−⏑−⏓.

Jagatī is similar but has an extra, short, syllable in penultiumate position, giving a line which is defined thus: ⏓−⏓−¦⏓⏑⏒¦−⏑−⏑⏓. 2% of the lines are in this metre.

There are many forms to the break, but note that the sequence −⏑⏑ is very common, accounting for approximately 73% of the breaks, no matter where the caesura falls:


,−⏑⏑ 19b; 40ab; 46a; 54d; 83a; 94b; 108d; 127cd; 128cd; 141b; 142ad; 144d; 151ad; 208c; 221abd; 280b; 281c; 306c; 325bd; 326a; 328b; 329b; 331c; 338d; 345a; 346ad; 347b; 353a; 354bc; 390cd (41 lines = 29%).

−,⏑⏑ 19c; 20de; 40d; 46b; 54ac; 83bc; 84b; 108a; 127ab; 128ab; 141ad; 144cd; 151c; 177c; 208b; 280d; 281a; 306ad; 309cd; 310c; 329c; 331b; 338ab; 345c; 390b; 326d (36 lines = 25%).

−⏑,⏑ 20c; 46c; 94c; 125ab; 142c; 144c; 151b; 177d; 208a; 280ac; 309a; 310ab; 325ac; 328c; 331ad; 346c; 347c; 353b; 353d; 390a (25 lines = 17%).

,−¦−⏑⏑ 208d; 328d; 338c; 345d (4 lines = 3%).


Other regular forms:

,⏑⏑⏑ 54b; 108c; 281b (3 lines = 2%).

⏑,⏑⏑ 20b; 84a; 94ad; 328a; 329a (6 lines = 4%).

,⏑⏑− 19a; 20a; 125d; 142b; 309b; 310d; 330c; 346b (8 lines = 6%).

,−⏑− 46d; 83d; 84d; 177b; 326bc; 345b; 347d; 354a (9 lines = 6%).

−,⏑− 354d (1 line = 1%).


The extended form of the metre, pausing at the fifth and restarting from the same syllable, occurs in 3 places, showing the following forms:

−,⏑⏑− 306b (1 line = 1%).

−,−⏑− 141c; 177a (2 lines = 1.5%).

Normally in the Tuṭṭhubha break there is a light syllable in 6th position and most scholarsHelmer Smith simply ignores the break in the appendix on the metres in his edition of Saddanīti (pg 1151-1154). Warder (PM, pg 208) says that it ‘should perhaps always be corrected’. have refused to countenance the possibility of a heavy syllable occurring in that position. However, as I have shown elsewhere (see my Parayanavagga), in the early verses there is one pattern to the break which does have a heavy 6th which occurs so regularly that it must be regarded as an acceptable variation. In this edition of the Dhammapada we find the same break occuring in 3 places:

−−,⏑ 144a; 221c; 281d.

There are 3 other occasions where we have a heavy 6th. They are ,−−⏑ 144b; 353c. & ,⏑−⏑ 108b. In each case we should probably correct the way we take the reading. See the notes in the text for a discussion.

There are 4 places where the break has only 2 syllables owing to replacement occurring at the 6th (see above).

−,− 40c; 125c.

,−− 19d, 20f.

It should be noted that 347a is unusual in that there is nowhere we can easily count a break as occuring. The following lines, which are irregular, are commented on in the text: 84c, 329d; 330d.