On Lamps and Islands: The Ambiguity of Dīpa in Indic Languages

Today, I’m editing my translation of DĀ 6, which is parallel to DN 26, the Cakkavatti Sutta. This sutra shares a passage made famous in the Parinibbāna Sutta in which the Buddha says:

“Attadīpā, bhikkhave, viharatha attasaraṇā anaññasaraṇā, dhammadīpā dhammasaraṇā anaññasaraṇā.

“Mendicants, live as your own island, your own refuge, with no other refuge. Let the teaching be your island and your refuge, with no other refuge.

Island, here, is read as a synonym for a protected place of refuge, making this appear to be a typical case of Buddhists repeating themselves in different ways for clarity. However, when we look at Chinese Āgama parallels, this passage takes on a different meaning. This is because dīpa also means lamp or fire in Indic languages like Pali, but not in Sanskrit, which separates the two meanings into dīpa (lamp) and dvīpa (island refuge). This makes a number scenarios possible, such as BHS dīpa (island or lamp) being misread as Skt. dīpa (lamp). However, we shouldn’t immediately assume the Chinese passages are mistaken, especially since they corroborate each other’s reading of dipa as lamp or fire instead of island. We may have an example of a northern Buddhist use of light metaphors, which the Theravada tradition doesn’t share.

Let’s look at the parallels.

DĀ 2 (Parinirvana Sutra) and DĀ 6 (Noble Wheel-Turning King Sutra) both have this reading:


You must be your own fire and burn in the Dharma; don’t burn something else. You must be your own refuge and take refuge in the Dharma; don’t take refuge in something else.

Here, the translator has translated dīpa as 熾燃, which literally means to burn intensely, like a raging fire, suggesting the original may have been a verb “to light” or “set fire.” This would be a Dharmaguptaka reading of the passage, which isn’t terribly clear. Either it’s understood that a lamp or torch is being lit, or the practitioner is lighting themselves like a lamp or torch, but it would seem most likely a metaphor of creating a source of metaphorical light (wisdom).

We also have the good luck that the Sarvâstivāda canon didn’t place the Noble Wheel-Turning King Sutra in their Dīrgha Āgama, so we find a parallel in the Chinese Madhyama Āgama. The parallel passage in this case is at the beginning of MĀ 70:


“Monks, you must light the Dharma lamp yourself and make the Dharma your own refuge. Don’t light another lamp, and don’t take refuge in another Dharma.”

Here, again, we get a better sense of dīpa being read as a lamp, but there’s also a verb (然) meaning “to light” it. This is an interesting metaphor because it occurs in others contexts in Chinese translation. Lighting the Dharma lamp is found in a number of texts, often Avadāna and Mahayana texts, and it seems to mainly refer to making the Dharma known to others or the world, and thereby putting out the darkness of ignorance. Examples include:

  • The Asoka Avadāna and Divyāvadāna material that was inserted into the Saṃyukta Āgama (T99.168b13 & 178c12)
  • The Kāśyapīya Saṃyukta Agama, No. 94 (T100.407a16)
  • Past Stories of the Bodhisattva (T153.69b19)
  • The Flower of Compassion Sutra (T157.202c18, 219b28, 224a7) - particularly notable because these passages seem to reference Cakkavatti Sutta
  • The Chinese Lalitavistara (T187.603b22)
  • The Chinese Buddhacarita (T193.87b11)
  • etc.

Needless to say, this wouldn’t be an unknown metaphor using a source of light to describe how wisdom banishes ignorance from the world when shared with others. It’s interesting to look at this case in the context of other passages in which Theravada sources seem to erase metaphors of light, perhaps because they took on essentialist interpretations in later eras.


Thanks for your post.
Wouldn’t the words that follow dīpa, “attasaraṇā anaññasaraṇā” lend support to reading dīpa as ‘island’? A safe, protected place?


Without any context, it’s a natural way to read the Pali, yes; Buddhist texts often pair synonyms together in this way to make their meaning clear. This is why the Agama readings are significant. Multiple sources read Dharmadīpa as a lamp rather than an island.


for what it’s worth, here’s the commentarial (Sumaṅgalavilāsinī) gloss:

Attadīpā’ti : mahāsamuddagatadīpaṃ viya attānaṃ dīpaṃ patiṭṭhaṃ katvā viharatha.


Well that’s interesting and strange to see such a difference. I was curious so I did a bit of search on DSBC, and found a few examples that may be relevant, including a Sanskrit parallel of that line.

Mahāparinibbānasutta (DN 16)

Evaṁ kho, ānanda, bhikkhu attadīpo viharati attasaraṇo anaññasaraṇo, dhammadīpo dhammasaraṇo anaññasaraṇo.

Mahāyāna Mahāparinirvāṇa Sūtram

evaṃ hi bhikṣur ātmadvīpo bhavaty ātmaśaraṇo dharmadvīpo dharmaśaraṇo 'nanyadvīpo 'nanyaśaraṇaḥ |

This seems to agree with the common interpretation of the Pali.

Śīlapāramitāyāmanarthavarjanaṃ Pañcamaḥ Paricchedaḥ

tatra jambūdvīpaḥ pūritaḥ syādvaiyāpṛtyakarairbodhisattvaiḥ |

jambūdīpapramāṇaiścoddeśasvādhyāyābhiyuktairbodhisattvairekasya pratisaṃlapanābhiyuktasya bodhisattvasyopasthānaparicaryā kartavyā |

It uses both dvīpa and dīpa as part of Jambudvipa?

yathā candrapradīpasūtre kāyasaṃvaramadhye paṭhayate-na hastalolupo bhavati na pādalolupaḥ hastapādasaṃyata iti ||

bodhicittavimalāgnisuprabhaṃ dharmadīpa samujvālayiṣyati ||

Here is dharmadīpa. But in this line, it seems to be connected to fire and light.

Maybe an original double meaning of both seclusion (island), and dhyana / samadhi (light)?


Just talk about some translation biases here, in my own native tongue, the word “lamp” has a lot of positive meanings, we have sayings to teach young children such as “Get close to the lamp, you are bright, get close to the ink, you are dark.” And lamp is also an exclusive word for any light-producing devices💡. Whereas island in our language is rather neutral in meaning, so it’s very easy for the translators to choose one word over the other.

I’m not a native English speaker but my impression is that “lamp” does not have much positive meaning when compared to “island”. Furthermore “lamp”, to my non-native ear, sounds a lot like lamb. :grinning_face_with_smiling_eyes:

But those are just biases of course.

This looks like the Skt. Parinirvana Sutra that still exists (not the Mahayana version that I think only exists in Chinese). But that reminds me of several other Chinese parallels to look at – the alternate Parinirvana Sutra translations.

I’m not finding a parallel in T5, but T6 appears to have a similar passage at 180b1, but it isn’t the same quote. It reads:


Again, the two things are an oil lamp and a refuge, but it’s less the Buddha giving advice to Ananda as just saying that he’s created these things by teaching. And the interpretation is the same as in DN 26, DA 6, etc, describing the four abodes of mindfulness.

I’m not finding a parallel in DA 7, either.


What I’d really like to do is find a parallel somewhere in Xuanzang’s translations to see how his texts read it. Perhaps it was resolved to dvipa when the switch to Skt. happened. It seems as though Sankritization caused confusions sometimes.


That makes sense, since it’s much shorter and has the direct parallel. Interesting that they have it filed under Mahayana sutras. Although now that I’m looking at the categories, they don’t have any category for sutras that are not Mahayana.

It looks like Gunabhadra went the other route with the interpretation, going the island (法洲) route in the Samyukta Agama (SA 36, SA 638, SA 639). Same with T 192, the Buddhacarita translated by Dharmakṣema. From SA 36:


There are certainly more examples of 法燈 across the canon, though. The term also reminds me of Dīpaṃkara Buddha (然燈佛).

I accidentally typed “lamb” earlier. I hesitated to delete it, because it seemed nice. Mendicants, after I’m gone, be like Dharma lambs…


It does, doesn’t it? That occurred to me as well. The Buddha “Burning Lamp.” I can dimly recall an Avadana story about an ascetic making a wick of his head and setting himself on fire like a human lamp or candle, too. In the Mahavastu, maybe? Wow, the things this passage in MA 70 seems to tie together.

I wonder if it’s true, then, that MA and SA come from different canonical lineages, even though they appear on their face to both be Sarvâstivāda. It must be something that arose after the sectarian traditions put their canons together.

1 Like

Years ago I was reading a book about modern Korean Buddhism that mentioned a rare practice still found occasionally in South Korea. The practice was for monks to burn off a finger as either an offering of sorts or a sign of dedication to the bodhisattva ideal (I can’t remember the exact reason they did it). The practice could be traced back to a Buddhist text, although I don’t remember if its the Avadana you’re thinking of or not.


I believe it was also the Buddha Maṅgala in the Buddhavaṃsa that burned himself, although translations/interpretations seem to differ.

1 Like

There’s this awesome story about Venerable Dabba the Mallian who was the lodgings assigner and while doing his job used his finger as a torch to show the late arrivers to their room:

Kd 14:4.4.31
Dabba then entered the fire element, and with his finger glowing, he walked in front of those monks.

Of course everybody rather tried to be a bit late on purpose …


In the Chinese translation of Mūlasarvāstivāda vinayakṣudrakavastu(根本有部毗奈耶雜事) translated by Yijing(義淨), the verse is stated as:

自爲洲渚 、自爲歸處,法爲洲渚、法爲歸處;無別洲渚,無別歸處。

which has the same meaning as DN16 of Pāli canon (Let yourself/dharma be your own island/refuge; without the other.’) The part which contains the verse is quoted entirely in 大寶積經 compiled by Bodhiruci(菩提流支). It seems that Yijing, who studied in Nālandā for a decade, used the scripture in Sanskrit in which a clear difference exists between island(dvīpa) and lamp(dīpa), as it is pointed out above.

I once heard a non-rigorous guess from a scholar-monk which states that the translators of corresponding Chinese sūtras might be well aware of the ambiguity between island and lamp (in some Prakrits like Pāli), as modern readers are, but they would have chosen lamp deliberately for the understanding of readers of the Chinese culture: He said that since most of the translators were experts of Sanskrit (or at least so-called Buddhist hybrid Sanskrit), it is quite likely that they knew that island is right.

Further, he noted that every Chinese translation chose 洲(island in the river) instead of 島(island in the sea) as the translation of dvīpa of the passage: this shows that they understood dvīpa as the island in the huge river like the Ganges, which might become the refuge in the flooding during the rainy season. (Is it so? I am not quite sure although it is clear that the Buddha dwelled in the inland region of the Indian subcontinent for his entire life.) He concluded that such subtropical phenomenon was not familiar to Chinese readers, thus they chose (or, made a pun of) light instead.

@dayunbao It seems that you are referring to 연비(燒臂; burning of the forearm) ritual in the Korean Buddhist tradition. Although the possible origin might be extended to the earlier sources, it is said that the ritual has been directly rooted in sentences of 梵網經 (which is considered as Chinese Apocrypha in general, and although the title means Brahmajāla Sūtra, it is NOT related to the Brahmajāla Sutta (DN1) of the Pāli Canon) and sentences of Saddharma Puṇḍarīka Sūtra.
After the simplification over centuries, in Korean Buddhist tradition nowadays, the term, 연비 refers to making a slight and temporal point burn by an instant touch of burning incense during some ceremonies. However, there still exist few old monks who burnt and excised some fingers entirely. Ironically, Ajahn Brahm met one of them, Ven. Hyeguk, during his visit to Korea in 2016:

Ajahn Brahm and Ven. Hyeguk are about to shake hands (The photo might be disturbing.)
(source: 고승들은 공(空)을 이야기했다 : 종교 : 사회 : 뉴스 : 한겨레)


I only just now realized that in the picture you originally posted Ajahn Brahm is going to shake the Korean monk’s hand. I thought he was ruffling the woman’s hair, which would be a strange thing for him to do, ha-ha.

1 Like

Thank you! I believe an argument was recently put forward that the Chinese Samyukta Agama is Mula-Sarvastivada, while the Madhyama was from a mainline Sarvastivada canon. Here we may have a piece of corroboration for that.

It might also be a case of passages changing over time. If we can find a parallel in Xuanzang’s Abhidharma texts that also reads the passage as island, then it may be Sanskritization that’s the culprit rather than sectarianism or regionalism. Yijing and Xuanzang both collected texts India proper rather than Central Asia, so that may also be a source of the difference. As always, though, it’s a blurry picture because of the lack of early Indic originals.

1 Like

As far as I heard, Mahāvastu of Lokottaravāda, which was written in Buddhist Hybrid Sanskrit, also uses dvīpa in the corresponding passage.

1 Like

For what it’s worth, I happened upon the term used in the SA by looking for parallels in the Abhidharmakosabhasya. The passage doesn’t seem to be present, but I did find 洲 for dvīpa as used for Jambudvipa, etc., and inferred 法洲.

catvāro dvīpāḥ-




uttarakuruś ca;


This is an interesting point that I passed over at first. Yet, 洲 does occur in a couple passages in DA with the same meaning of an island refuge that we would expect in the Cakkavatti sutta passage. In DA 21, the Brahmajala sutra, this series of synonyms occurs three times:

[91c14] 世間有沙門、婆羅門廣博多聞,聰明智慧,常樂閑靜,機辯精微,世所尊重,能以智慧善別諸見。設當問我諸深義者,我不能答,有愧於彼,於彼有畏,當以此答以為歸依、為洲、為舍,為究竟道。

It would seem the passage in DA 6 either was interpreted specifically to mean fire, or it had a different word altogether.

MA doesn’t have a passage that uses 洲 as a refuge; it only occurs as the word for a continent.

SA has quite a few examples like DA 21, including the same passage that appears in the Cakkavatti and Parinibbana suttas. It’s the outlier that makes it difficult to say that it’s definitely a change of interpretation that happened sometime after 400 CE. Perhaps for a time, there were traditions that interpreted it either way, and the lamp interpretation is the source of the Dipankara avadana stories. Then the passage was resolved to “island” later, obscuring its original inspiration.

Difficult to support judgements about whether the Pali reading of dipa as “island” is original or later, given that SA indicates it may go back quite a long time.