On vāri and the restraint of Mahāvīra

DN 2 offers an important survey of key teachings of several contemporaries of the Buddha, including Mahāvīra Vardhamāna, the “Jain ascetic of the Ñātika clan” as he is referred to in Pali.

The Pali passage has been a struggle for translators as it is linguistically obscure and doesn’t obviously relate to other known teaching of the Jains as recorded in the Pali canon. The passage occurs at DN 2:29.4 and MN 56:12.2.

sabbavārivārito sabbavāriyutto sabbavāridhuto sabbavāriphuṭo

Let’s see if we can untangle this with the assistance of Jain scriptures.

This is said to be a “fourfold constraint”. There is a better-known fourfold constraint that is mentioned at DN 25:16.3, which corresponds with the known set of that name in Jainism. Thus the teaching here has been suspected of being corrupted, or perhaps a satire on Jain views. Given that vāri means “water” in Pali, and that the Jains were noted for their rules regarding water (see MN 56:11.2, there using udaka), some interpreters, including myself, took this as a reference to Jain discipline regarding water.

However, it turns out that the phrase has a close parallel in a genuine early jain text that records the teaching of Mahāvīra himself. Not only that, but almost the entire section has close parallels in the suttas.

The text is Isibhāsiyāiṁ, the “Sayings of the Seers”. This is a rather extraordinary text which records saying by some 40 or so sages, most of whom do not seem to be Jain. Several are Buddhist—including verses by Sāriputta, which I hope to look at in the future—some are brahmins, while others have no known affiliation.

Text here:


Translation and study here:

The relevant section, chapter 29, has a series of verses attributed to vaddhamāṇa, who rather curiously in the Jain tradition is not identified with their teacher Mahāvīra Vardhamāna, but with another sage. However this seems to be incorrect, not least because the same teaching is ascribed to him in Pali too. Thus we can assume this in, in fact, attributed to Mahāvīra.

Given that the whole text has close Buddhist parallels, it is natural to wonder what the relation is. Perhaps the Buddhists vorrowed from the Jains, or vice versa, or both, or both drew from the same source, or both shared the teachings before splitting in two (assuming the early Buddhist tradition was less differentiated than it became later). All or none of these could be the case, and I don’t have enough familiarity with the texts to draw any solid conclusions.

I’ll present the Jain text here, although since much of it is repetitive, I’ll abbreviate. I give a rough translation, but since I don’t know Ardhamagadhi, take it with a grain of salt. There’s a nice study and translation of the text; the translation is very free but still helpful. Pali parallels are included.

savanti savvato sotā, | kiṃ -a sotoṇṇivāraṇaṃ? /
puṭhe muṇī āikkhe: | kahaṃ soto pihijjati? ||1||

The stream flow everywhere, what is their blocking?
I ask the sage this question: how are the streams dammed?


“Savanti sabbadhi sotā,
(iccāyasmā ajito)
Sotānaṁ kiṁ nivāraṇaṁ;
Sotānaṁ saṁvaraṁ brūhi,
Kena sotā pidhiyyare”.

Note that the verb pihijjati / pidhiyyare is extremely rare, it occurs only here in the Pali canon.

In both the Buddhist and Jain texts, this opening question recieves a good and pertinent response, although they are completely different.

vaddhamāṇeṇa arahatā isiṇā buitam. ||1||
The perfected seer Vardhamāna said:

panca jāgarao suttā, | panca suttassa jāgarā /
pancahiṃ rayam ādiyati, | pancahiṃ ca rayaṃ ṭhae ||2||

Five sleep amid the wakened, five wake amid the sleeping
By five are dust gathered, by five dust prevented.

SN 1.6:3.1:

“Pañca jāgarataṁ suttā,
pañca suttesu jāgarā;
Pañcabhi rajamādeti,
pañcabhi parisujjhatī”ti.

The Pali commentary explains these as the five hindrances and five spiritual faculties, whereas in the Jain text they refer to the five senses, which are “asleep” (dormant) in the enlightened, but “awake” (rampant) in fools.

The Jain text next goes on to discuss each of the five senses, in verses that loosely parallel SN 35.94. The parallel is not as close as the previous two verses, but the general tenor is the same. The Jain text for some reason starts with the ear, and repeats the five senses in the exact same form. The Pali, on the other hand, has the normal sequence, and introduces variations in each verse. I won’t give all of them, just the eye is enough.

rūvaṃ cakkhum uvādāya | maṇuṇṇaṃ vā vi pāvagaṃ /
maṇuṇṇammi -a rajjejjā, | -a padussejjā hi pāvae ||5||
maṇuṇṇammi arajjante | aduṭhe iyarammi ya /
asutte avirodhīṇaṃ | evaṃ soe pihijjati ||6||

When a sight arises in the eye, delightful or bad,
don’t desire the delightful, nor repulse the bad.
Not desiring the delightful, nor repulsing the disliked,
not attached, not rejecting: that is how the stream is dammed.

SN 35.94:4.1:

Disvāna rūpāni manoramāni,
Athopi disvāna amanoramāni;
Manorame rāgapathaṁ vinodaye,
Na cāppiyaṁ meti manaṁ padosaye.

Having seen pleasing sights
and unpleasant ones too,
get rid of all manner of desire for the pleasant,
without hating what you don’t like.

Notice that Ardhamagadhi maṇuṇṇaṃ serves for Pali manorama. Both have the same sense and are derived from man (“mind”). The more exact Pali equivalent is manuññaṁ which is found rarely (eg. MN 66:11.3). It’s interesting that maṇuṇṇaṃ appears in Isibhāsiyāiṁ only here and in the verses attributed to Sāriputta, where it is in fact a close match for the passage at MN 66:11.3.

I’m not sure of the senses of asutte in the last line of the Jain verse. It would seem to mean “not sleeping” as it does earlier in the poem. But the sense requires rather “not attached, not favoring”; in similar passages the Pali has ananurodha (“not favoring”). Perhaps it is a form of asatta, “unattached”.

The next series of verses doesn’t seem to have any close Pali equivalents, though there are some phrases in common, which I note, although this is surely not all. A curious detail, however, is that the prose framework for SN 35.94 speaks of the “tamed” and “untamed”, just as these verses do. Thus the above verses speak of restraining through the senses, and are more-or-less shared. Each has a context that is unshared, but the context—prose in one, verse in another— introduces the concept of the “tamed”.

I provide a very rough translation, leaving them mostly unpunctuated to show their tenuousness.

duddantā indiyā panca | saṃsārāya sarīriṇaṃ /
te c’ eva -iyamiyā sammaṃ | -evvāṇāya bhavanti hi ||13||

when the five senses are ill-tamed, one transmigrates in transmigration
once restrained, they serve as tools for deliverance

duddanteh’ indieh’ appā | duppahaṃ hīrae balā /
duddantehiṃ turaṃgehiṃ | sārahī vā mahā-pahe ||14||

unrestrained senses drive the soul downwards to hell
as wild steeds drag the chariot off the highway

indiehiṃ sudantehiṃ | -a saṃcarati goyaraṃ /
vidheyehiṃ turaṃgehiṃ | sārahi vvāva saṃjue ||15||

When the five senses are well-tamed, one proceeds in that domain,
like trained steeds that stay on the highway

puvvaṃ maṇaṃ jiṇittāṇaṃ | vāre visaya-goyaraṃ /
vidheyaṃ gayam ārūḍho | sūro vā gahit’ āyudho ||16||

first conquer the mind, restrain in the domain,
ride and command the elephant like an armored warrior.

jittā maṇaṃ kasāe yā | jo sammaṃ kurute tavaṃ /
saṃdippate sa suddh’ appā | aggī vā havisāhute ||17||

One who has conquered mind and defilements, rightly practicing austerities,
shines bright as the fire on an altar rich in offerings.

sammattaṇṇirataṃ dhīraṃ | danta-kohaṃ jitindiyaṃ /
devā vi taṃ -amaṃsanti | mokkhe c’ eva parāyaṇaṃ ||18||

The sage rightly conducted, tamed, victor over the senses,
is praised even by the gods, liberation is their destiny.

A couple of Pali parallels:

an4.28:7.5: Devāpi naṁ pasaṁsanti,
snp5.15:3.5: Vimuttaṁ tapparāyaṇaṁ.

savvattha viraye dante | savva-vārīhiṃ vārie /
savva-dukkha-ppahīṇe ya | siddhe bhavati -īraye ||19||

One everywhere desireless, tamed, restrained in all restraints
who has let go of all suffering, wanders achieving success.

Note dhīra in preceding verse, also in the following:

pli-tv-kd1:22.14.5: “Yo dhīro sabbadhi danto,*
The sage everywhere tamed

snp2.2:12.3: Saṅgātigo sabbadukkhappahīno, Na lippati diṭṭhasutesu dhīro”.
The sage has slipped their chains and given up all pain; they don’t cling to the seen and the heard.”

Our main point of interest here is the final verse, where we find savva-vārīhiṃ vārie, which obviously has the same sense as the Pali sabbavārivārito.

The Jain text has only this one phrase, not the four. The “four constraints” are normally ethical precepts in Jainism, starting with not killing, and that sense is indeed found at Isibhāsiyāiṁ 31.39ff. Most likely the fourfold scheme is summarized here in verse.

The Jain context is enough to conclusively rule out any connection between vāri here and “water”. It is, rather, a causitive form, “to make stop” i.e. “restrain”. This sense is introduced in the first verse of the poem and consistently reinforced throughout.

The form vārīhiṃ is, I believe, a locative plural, as opposed to the Pali commentary which glosses with instrumental. The Pali form is ambiguous as it is compounded, but such variations in oblique forms are not unusual.

Returning at last to the Pali phrase, a couple of notes.

  • Dhuta in the sense “shaken off (evil by means of ascetic practices)” is a characteristic Jain term.
  • For sabbavāriphuṭo compare ophuṭo at mn99:15.5. In both cases phuṭ appears in a string of terms from the root var, and is possibly a corrupted form, or at least has the same meaning.

Thus we can translate:

sabbavārivārito sabbavāriyutto sabbavāridhuto sabbavāriphuṭo
(A Jain ascetic) is restrained in all restraints, is bridled in all restraints, has shaken off evil in all restraints, and is curbed in all restraints.


Sadhu — thank you, venerable! :pray: Very happy that this vāri issue has been resolved finally.

Do you have access to resources with more information on the Isibhāsiyaim? I couldn’t find very detailed articles on e.g. where it comes from, the chronology and how scholars date it, etc. I’d like to learn more as there seems to be some consensus on it being an early text.

I also think that there needs to be more research on Buddhist-Jain connections. Most ink has been spilled on Buddhist relationships to Brahmanism, especially the Upanisads, with a mixture of helpful and far-fetched scholarship. And this is certainly very valuable and relevant.

But, here we already see a lot of correspondences. Basic early ideas like the āsavas, the idea of countless eons of rebirth being the relatives of sentient beings, a life-long religious mendicancy with organized rules, non-Brahmanical sramana movements, and so many more essentials in Buddhism are pre-dated and/or coeval with Jainism. It seems that whereas the Brahmanical cultural influence was heavy, important for converts and status, and very influential, it was really among the sramanas that a lot of these actual practical ideas were at home. Who knows? This type of research may begin to shed light on major Buddhist collections and ideas — like the SN Sagāthāvagga, etc. — rather than mostly vague allusions to Brahmanical concepts.


The intro to that linked file is pretty good. But that’s all I have.

For sure yes, if you uncover anything let us know.

One thing that strikes me about this passage is that it’s not just a shared verse, but a whole passage that is mostly found in the two traditions, instantiated in different ways, but still with a similar message. It’s like what I’ve found in the Vedic literature. Sometimes it’s not just words or phrases, but shapes of narrative that provide the link.


I’ve found that the ear is a good place to start coming to terms with the senses.

I wonder if you could provide the exact order in the text?

1 Like

Same as the normal Buddhist order, just the first two swapped. The Brahmanical texts are less standardized than the Buddhist, but I don’t know if that is generally less true of the Jain texts or just here.


pidhiyyare is cognate with the Sanskrit third person passive plural form pidhīyante (pidhīyate being singular) - from verb (a)pi-dhā (meaning ‘to shut’), where pi is an irregular contraction of for the preverb/upasarga ‘api’.

The -re at the end of the pali word (in my understanding) is actually a misreading of -te in the sanskrit original in the pre-pali manuscript (when the word is written in the Kharosthi script ‘te’ looks extremely similar to ‘re’) - from where Pali inherits this word (but it has evidently been thereafter regularized with the ‘re’ pl. ending in pali for most verbs).

te in kharoṣṭhī

re in kharoṣṭhī

The pali variant readings of this word (in the pārāyaṇavagga) are

  • pithiyyare (sī. syā. pī.),
  • pithīyare (sī. aṭṭha.),
  • pidhīyare (?)
    which indicates the word was sometimes written with a long medial ī (which was at other times converted into a iy) - but the reading of the final -re for the -te seems uniform.

The cūḷaniddesa gives a variant of the same word as explanation – kena sotā pidhiyyareti kena sotā pidhīyanti…
So does the atthakathā - kena sotā pidhiyyareti kena dhammena ete sotā pidhiyyanti

Yes. The final vārita in (sabbavārivārito) is a causative past passive participle form of the Skt. root vṛ – and means ‘has restrained’. The vāri (in sabbavāri) is, I think the Pāli-ization of the Sanskrit future passive participle form vārya (“has to be restrained”).

Hence the 4 bahuvrīhi compounds probably mean:

  • sabba-vāri-vārito (sarva-vārya-vāritaḥ) = one who has restrained everything that had to be restrained.
  • sabba-vāri-yutto (sarva-vārya-yuktaḥ) = one who adheres to all that’s supposed to be restrained – cf. Skt yukta
  • sabba-vāri-dhuto (sarva-vārya-dhutaḥ) = one who has shaken-off (evil) by such a practice – cf. Skt dhuta
  • sabba-vāri-phuṭo (sarva-vārya-sphuṭaḥ) = one who has manifested (in himself) all that’s to be restrained – cf. Skt sphuṭa

But Pali didn’t go via Kharosthi.

The ending is rare but regular, found in jāyare, miyyare, nopanīyare, sandhīyare, piyyare, jīyare, etc. It’s reflexive (middle) plural 3rd present.

Thanks, yes.

Right, that makes sense, I think it is otherwise unattested in Pali.

It looks to me like the Ardhamagadhi savva-vārīhiṃ vārie supports this reading, but I don’t know enough about it to say.

Just FYI, here and in one other case, phuṭa is closely associated with var and it seems it’s a misreading (perhaps from var, via vut) or perhaps an unattested sense of phuṭa in the meaning var.

pañcahi nīvaraṇehi brāhmaṇo pokkharasāti opamañño subhagavaniko āvuto nivuto ophuṭo pariyonaddho.

I have seen other words where only passing through Kharoshthi (and ancient misreadings therefrom) can make sense of certain Pali spellings - specially where regular sound shifts do not work. I am not theorizing with just this one example.

The passive -re ending (as in pidhīyare / pidhiyyare) has otherwise no phonological basis.

kena sotā pidhiyyare is manifestly in the passive voice and not reflexive, specially where the agent appears in the instrumental.

The other verbs you mentioned are also found to be passive constructions, as in

  • chahi ṭhānehi jāyare (pli-tv-pvr3)
  • yena jātā na miyyare (Snp3.8)
  • khīṇāsavā mānena na upanīyare (atthakathā on AN6.49)
  • khajjare bhuñjare piyyare ca (ja497)

I’ve never really studied this form in detail, so let’s learn. Let’s add to the list:

  • vuccare “(they) are said”
  • upāsare “(they) attend”
  • haññare “(they) are killed”
  • modare “(they) rejoice”
  • dissare “(they) are seen”
  • nopanīyare “(they) don’t represent”
  • bhāsare “(they) speak”
  • seyare “(they) sleep”
  • nīyare “(they) are led”
  • upapajjare “(they) rearise”
  • dicchare “(they) give”
  • paṭijānare “(they) confirm”
  • pamuccare “(they) are freed”

I’m sure there are more! These are all from verse. We have a mixture of active and passive forms; probably an unusually high proportion of passive. But the passive forms are all marked by -(i)ya added to the root as normal. So we can’t say that the middle form itself is passive.

That’s not a passive construction; the agent and verb agree in plural, the instrumental in singular is a qualifier here (compare samatte nopanīyare with attani upaneti).

khīṇāsavā mānena na upanīyare, na upanenti, na upagacchantī
those with defilements ended do not present, include, approach with conceit

Another example of this: The patronymic (in Sanskrit) Pārāśarya is found in Pāli EBTs in four variant forms - pārāsariya, pārāpariya, pārāsiriya & pārāsiviya. These variants would normally be inexplicable as far as regular phonetic laws and sound shifts are concerned.

Here (below) they are written in Kharoṣṭhī - they read pārāśariya, pārāpariya, pārāśiviya & pārāśiriya respectively - but they look nearly the same (and I am therefore nearly certain that the lexical variants arise from misreading the source text from a pre-Pali Kharoṣṭhī manuscript - before conversion into Pali in Brāhmī script).