Thoughts on the Dirgha Agama's History

Lately, I’ve been attempting to write an introduction to the Dīrgha Āgama. Part of that introduction is a discussion of the history of the collection, such as speculating about what an ur-Dīrgha Āgama may have looked like. It’s a subject that scholars understandably avoid because the picture is complicated and there’s little objective history to confirm or refute different scenarios that we might consider. The result is that, beyond the story of the First Council, the different accounts of how the Saṅgha fractured into multiple traditions, and comparing the contents of the three extant Dīrgha collections, we don’t have much to base our theories on.

Still, I think it’s worthwhile to try to formulate something, and perhaps others can add their two cents. I’m sure there’s scholarship I’m unaware of simply because I don’t have access to a university library.

The Three Extant Collections

Let’s begin by looking at the contents of the three Dīrgha collections, using the smallest of them as the frame of reference and placing the parallels in the other two beside it:

Dharmaguptaka Theravāda Sarvāstivāda
1. Mahāvadāna 14. Mahāpadāna 5. Mahāvadāna
2. Mahāparinirvāṇa 16. Mahāparinibbāna 6. Mahāparinirvāṇa
17. Mahāsudassana (MĀ 68)
3. Govinda 19. Mahāgovinda 14. Govinda
4. Janavṛṣabha 18. Janavasabha 13. Jinayabha
5. Kṣudranidāna 27. Aggañña (MĀ 154)
6. Cakravartīrāja 26. Cakkavatti (MĀ 70)
7. Padāśva 23. Pāyāsi (MĀ 71)
8. Sandhāna 25. Udumbarika (MĀ 104)
9. Saṅgīti 33. Saṅgīti 3. Saṅgīti
10. Daśottara 34. Dasuttara 1. Daśottara
11. Ekottara
12. Triskandha
13. Mahānidāna 15. Mahānidāna (MĀ 97)
14. Śakrapraśna 21. Sakkapañha (MĀ 134)
15. Anomiya 24. Pāṭika 9. Bhārgava
16. Sujāta 31. Siṅgālovāda (MĀ 135)
17. Prasādayati 29. Pāsādika 15. Prāsādika
18. Personal Gladness 28. Sampasādanīya 16. Prasādanīya
19. Mahāsamaya 20. Mahāsamaya 24. Mahāsamāja
20. Ambāṣṭha 3. Ambaṭṭha 35. Ambāṣṭha
21. Brahmacala 1. Brahmajāla 47. Brahmajāla
22. Śroṇatāṇḍya 4. Soṇadaṇḍa 33. Śroṇatāṇḍya
23. Kūṭatāṇḍya 5. Kūṭadanta 34. Kūṭatāṇḍya
24. Dhruva 11. Kevaddha 29. Kaivarti
25. The Naked Wanderer 8. Kassapasīhanāda 46. Kāśyapa
26. Trivedajnana 13. Tevijja 45. Vāsiṣṭha
27. Sramanaphala 2. Sāmaññaphala 44. Rājā
28. Poṭṭhapāda 9. Poṭṭhapāda 36. Pṛṣṭhapāla
29. Lohitya 12. Lohicca 28. Lohitya (2)
30. Lokaprajnatpi
6. Mahāli 32. Mahallin
7. Jāliya 30. Maṇḍīśa (1)
10. Subha 42. Śuka
22. Mahāsatipaṭṭhāna (MĀ 98)
30. Lakkhaṇa (MĀ 59)
32. Āṭānāṭiya 23. Āṭānāṭa
2. Arthavistara
4. Catuṣpariṣat
(MN 60 Apaṇṇaka) 7. Apannaka
8. Sarveka
(MN 105 Sunakkhatta) 10. Śalya
(MN 4 Bhayabherava) 11. Bhayabhairava
(MN 12 Mahāsīhanāda) 12. Romaharṣaṇa
(MN 102 Pañcattaya) 17. Pañcatraya
18. Māyājāla
(MN 95 Caṅkī) 19. Kāmaṭhika
(MN 36 Mahāsaccaka) 20. Kāyabhāvanā
(MN 85 Bodhirājakumāra) 21. Bodha
(MN 100 Saṅgārava) 22. Śaṃkaraka
25. Tridaṇḍin
26. Piṅgalātreya
27. Lohitya (1)
31. Maṇḍīśa (2)
37. Kāraṇavādin
(MN 51 Kandaraka) 38. Pudgala
39. Śruta
? 40. Mahalla
41. Anyatama
(MN 55 Jīvaka) 43. Jīvaka

There are a total of sixty sūtras that appear in the three extant collections. This is largely because a large number of sūtras was added to the Sarvāstivāda version, and because the Theravāda/Dharmaguptaka and Sarvāstivāda versions disagreed on whichsūtras ought to be in the Dīrgha collection and which ought to be in the Madhyama collection. I’ve noted the parallels in the Majjhima Nikāya and Madhyama Āgama to show this disparity that existed between them.

When we compare the Dharmaguptaka Dīrgha to the Theravada Dīgha, they are very similar, differing only in that each added a few sūtras that the other did not. The table below illustrates this:

Dharmaguptaka Theravāda
1. Mahāvadāna 14. Mahāpadāna
2. Mahāparinirvāṇa 16. Mahāparinibbāna
2. Mahāparinirvāṇa 17. Mahāsudassana
3. Govinda 19. Mahāgovinda
4. Janavṛṣabha 18. Janavasabha
5. Kṣudranidāna 27. Aggañña
6. Cakravartīrāja 26. Cakkavatti
7. Padāśva 23. Pāyāsi
8. Sandhāna 25. Udumbarika
9. Saṅgīti 33. Saṅgīti
10. Daśottara 34. Dasuttara
11. Ekottara
12. Triskandha
13. Mahānidāna 15. Mahānidāna
14. Śakrapraśna 21. Sakkapañha
15. Anomiya 24. Pāṭika
16. Sujāta 31. Siṅgālovāda
17. Prasādayati 29. Pāsādika
18. Personal Gladness 28. Sampasādanīya
19. Mahāsamaya 20. Mahāsamaya
20. Ambāṣṭha 3. Ambaṭṭha
21. Brahmacala 1. Brahmajāla
22. Śroṇatāṇḍya 4. Soṇadaṇḍa
23. Kūṭatāṇḍya 5. Kūṭadanta
24. Dhruva 11. Kevaddha
25. The Naked Wanderer 8. Kassapasīhanāda
26. Trivedajnana 13. Tevijja
27. Sramanaphala 2. Sāmaññaphala
28. Poṭṭhapāda 9. Poṭṭhapāda
29. Lohitya 12. Lohicca
30. Lokaprajnatpi
6. Mahāli
7. Jāliya
10. Subha
22. Mahāsatipaṭṭhāna
30. Lakkhaṇa
32. Āṭānāṭiya

DN includes six suttas not found in the Dharmaguptaka DĀ, and that DĀ includes three that are not in DN. Generally speaking, we can conclude that both versions have about the same amount of material added independently of each other, but the two collections share the vast majority of their contents. This is because one of the texts added by the Dharmaguptakas is very large. I think we can safely say that both were originally the same or nearly the same collection, which diverged over time in each tradition.

The Sarvāstivāda version is quite different, however, sharing only 20 sūtras with the Dharmaguptaka/Theravada versions. This makes the whole situation messier, but it also narrows the common ur-Dīrgha down to a smaller set of texts.

The First Chapter of the Original Collection

These are the 20 sūtras shared by all three collections (using the titles found in the Dharmaguptaka version):

Common Sutras of theDīrgha Collections
1. Mahāvadāna
2. Mahāparinirvāṇa
3. Govinda
4. Janavṛṣabha
9. Saṅgīti
10. Daśottara
15. Anomiya
17. Prasādayati
18. Personal Gladness
19. Mahāsamaya
20. Ambāṣṭha
21. Brahmacala
22. Śroṇatāṇḍya
23. Kūṭatāṇḍya
24. Dhruva
25. The Naked Wanderer
26. Trivedajnana
27. Sramanaphala
28. Poṭṭhapāda
29. Lohitya

It’s interesting that all three extant collections have 20 sūtras in common. Traditionally, Buddhist collections were grouped into chapters of 10 sūtras each. Texts were later inserted or removed from these sets of ten, but that was clearly a common practice when the first collections were assembled. If there was originally an ur-Dīrgha that predated the surviving collections, I would expect that it consisted of 10, 20, or 30 sūtras depending on how many chapters it contained.

It so happens that ten of these common sūtras occur in the same section of all three extant collections:

  • The first Chapter of the Collection on Morality (Sīlakkhandhavagga) in DN
  • The third Chapter of the Collection on Morality (Śīlaskandhanipāta) in the Sarvāstivāda DĀ
  • The third section of the Dharmaguptaka DĀ, which was untitled

Each of these chapters have the common theme of stories about encounters with non-Buddhists. Let’s look at the contents of that chapter in each collection. Again, we’ll use the smallest of the collections as the frame of reference and place parallels in the other two next to it.

Dharmaguptaka Theravada Sarvastivada
20. Ambāṣṭha 3. Ambaṭṭha 35. Ambāṣṭha
21. Brahmacala 1. Brahmajāla 47. Brahmajāla
22. Śroṇatāṇḍya 4. Soṇadaṇḍa 33. Śroṇatāṇḍya
23. Kūṭatāṇḍya 5. Kūṭadanta 34. Kūṭatāṇḍya
24. Dhruva 11. Kevaddha 29. Kaivarti
25. The Naked Wanderer 8. Kassapasīhanāda 46. Kāśyapa
26. Trivedajnana 13. Tevijja 45. Vāsiṣṭha
27. Sramanaphala 2. Sāmaññaphala 44. Rājā
28. Poṭṭhapāda 9. Poṭṭhapāda 36. Pṛṣṭhapāla
29. Lohitya 12. Lohicca 28. Lohitya (2)
6. Mahāli 32. Mahallin
7. Jāliya 30. Maṇḍīśa (1)
10. Subha 42. Śuka
25. Tridaṇḍin
26. Piṅgalātreya
27. Lohitya (1)
38. Pudgala
31. Maṇḍīśa (2)
37. Kāraṇavādin
39. Śruta
40. Mahalla
41. Anyatama
43. Jīvaka

As we can see, there was a great expansion of this chapter by the Sarvāstivādins, who more than doubled it’s size compared to the Dharmaguptakas. Only two of these additions have parallels in MN, suggesting that they may have been original compositions by the Sarvāstivādins. Not having access to the texts themselves, it’s hard to judge just how original they were. Sometimes, smaller sūtras whose parallels are found in AN or SN were expanded into larger ones and then placed in the Sarvāstivāda MĀ, so it stands to reason a similar process may have taken place when they expanded their version of DĀ.

We could very easily decide that this was a chapter of the original ur-Dīrgha: Ten sūtras depicting conversations with non-Buddhist ascetics and brahmins. Let’s call this the first chapter.

The Second Chapter of the Original Collection

But what about the other ten common sūtras? They aren’t as easily placed in a chapter with a single theme. They consist of:

  • Two Abhidharma-related sūtras (DĀ 9-10)
  • Two sūtras that describe the Buddha’s history (DĀ 1-2)
  • Three sūtras that describe gods and spirits (DĀ 3-4 and 19),
  • Three sūtras about faith and disbelief in the Buddha (DĀ 15 and 17-18).

Suffice it to say, they appear to be a miscellaneous set of texts that happens to sum up to ten. Could it be that this was a second chapter put together on various topics in the ur-Dīrgha? I think so.

The Splitting of the Original Second Chapter into Two Chapters

At some point after the initial compilation of the ur-Dīrgha, it would seem that the expansion process began. This made it necessary to add a third chapter to the collection, but it wasn’t done in a straightforward way. Instead of simply appending a third chapter, the sūtras that were originally in the second chapter were divided between two new chapters. The new second chapter collected together the most important sutras about the Buddha and Dharma, and the third chapter became the miscellaneous collection that lacked a common theme. This is more difficult to document exactly because each tradition ended up with a different set of sūtras in their second and third chapter. It would seem that this expansion occurred before the Theravāda/Dharmaguptaka schools arose but after the Sarvāstivāda split away from their parent tradition. Or, the Sarvāstivādins may have simply been the more creative tradition early on as well as in later eras.

Sectarian Reorganizations of the Three Chapters

In each sectarian version of the Dīrgha, these three chapters are arranged in a different way.

1. The Theravāda Arrangement

I believe that the Theravāda version may be the most straightforward representation of the expansion process, if we accept that the Sīlakkhandhavagga was the first chapter of the collection. That is:

  • The Sīlakkhandhavagga is the oldest group of suttas (with a few additions)
  • The Mahāvagga was the next stage of expansion that occurred when Vinaya stories were turned into independent suttas
  • The Pāthikavagga was where later additions were collected into a chapter of miscellania without an intentional theme

2. The Dharmaguptaka Arrangement

The Dharmaguptaka version has these three chapters plus a fourth one in a different order:

  • The second chapter containing the Vinaya sūtras was moved to the front of the collection, perhaps owing to the Dharmaguptaka reverence of the Buddha.
  • The third miscellania chapter was moved to second place, which may have represented the Dharma to them given that it contained the Abhidharma sūtras and sūtras on dependent origination.
  • This left the original first chapter in third place.
  • They also attached a mythological collection to serve as a fourth chapter.

This organization followed the Dharmaguptaka principle of dividing canonical collections into four parts. Their Vinaya has four parts and their Abhidharma consists of four parts. They must of divided their Āgamas into four parts, too.

3. The Sarvāstivāda Arrangement

The Sarvāstivāda version consists of three chapters, but they didn’t maintain the same organization of the two new chapters that the Theravādins and Dharmaguptakas did. This could be evidence that the creation of a third chapter happened during the sectarian period but before the Theravādins and Dharmaguptakas parted ways and became separate traditions. I think that parting of ways most likely happened when Aśokan missionaries traveled to South India and Northwest India carrying very similar or identical versions of the Dīrgha Āgama. Then, over centuries of time, those two version diverged in the ways already described, which are fairly minor in the greater scheme of things.

For the Sarvāstivādins, though, the two new chapters were organized in a different way, so this may have happened before Aśoka. Or, perhaps, Sarvāstivādins were much more open to reorganizing their collection, which isn’t an unreasonable scenario given how many more sūtras they added to it.

Their first chapter, which they unimaginatively titled the Chapter of Six Sūtras (S. Ṣaṭsūtrakanipāta), consisted on both Abhidharma sūtras and texts drawn from the Vinaya:

The Chapter of Six Sūtras Theravāda Dharmaguptaka
1. Daśottara Ch. 3, DN 34 Ch.2, DA 10
2. Arthavistara
3. Saṅgīti Ch. 3, DN 33 Ch.2, DA 9
4. Catuṣpariṣat
5. Mahāvadāna Ch.2, DN 14 Ch.1, DA 1
6. Mahāparinirvāṇa Ch.2, DN 16 Ch.1, DA 2

It’s not surprising that Sarvāstivādins would place Abhidharma sūtras at the top of their Dīrgha given how important Abhidharma became in their tradition. Then, like the Dharmaguptakas, they placed Vinaya stories about the Buddha right after them, giving them pride of not-quite-first place. The Dharmaguptakas did the opposite, placing the Abhidharma sūtras in their second chapter. Theravādins placed them dead last, which might well represent how late they actually were in Buddhist history.

The Sarvāstivādins called their second chapter the Chapter of Pairs (S. Yuganipāta), and it consisted of nine pairs of related sūtras. To give a couple examples of what they considered to be pairs, the chapter contains the Janavasabha and Mahāgovinda parallels, which both culminate in stories told by Brahmā. Another pair are the Āṭānāṭiya and Mahāsamaya parallels, which are both esoteric sūtras about mythical spirits and gods. This chapter also contains sūtras that are in both of the second and third chapters in the Theravāda/Dharmaguptaka collection:

The Chapter on Pairs Theravāda Dharmaguptaka
7. Apannaka MN 60
8. Sarveka
9. Bhārgava Ch.3, DN 24 Ch.2, DĀ 15
10. Śalya MN 105
11. Bhayabhairava MN 4
12. Romaharṣaṇa MN 12
13. Jinayabha Ch.2, DN 18 Ch.1, DĀ 4
14. Govinda Ch.2, DN 19 Ch.1, DĀ 5
15. Prāsādika Ch.3, DN 29 Ch.2, DĀ 17
16. Prasādanīya Ch.3, DN 28 Ch.2, DĀ 18
17. Pañcatraya MN 102
18. Māyājāla
19. Kāmaṭhika MN 95
20. Kāyabhāvanā MN 36
21. Bodha MN 85
22. Śaṃkaraka MN 100
23. Āṭānāṭa Ch.3, DN 32
24. Mahāsamāja Ch.2, DN 20 Ch.2, DĀ 19

As can be seen, the majority of these sūtras occur in the Majjhima Nikāya, and they may well of also appeared to the Dharmaguptaka MĀ if still existed for us to compare (assuming that it was as similar to MN as their DĀ is to DN).

Thus, it does seem likely that the Sarvāstivādins subjected their Dīrgha to wholesale reorganization, but kept the original chapter largely unchanged aside from doubling its size. This in itself argues for that group of sūtras are being the core of the original Dīrgha, as this old core would have been tampered with less than later additions.


You may need to read carefully and critically Ven. YinShun’s “第二節 中阿含與長阿含” (MA and DA) in the book The Formation of Early Buddhist Texts (原始佛教聖典之集成), pp. 703-754. CBETA 線上閱讀

1 Like

Read also:
Bucknell, Roderick S. 2014:
The Structure of the Sanskrit Dīrgha-āgama from Gilgit vis-à-vis the Pali Dīgha-nikāya”, in Dhammadinnā (ed.), Research on the Dīrgha-āgama (Dharma Drum Institute of Liberal Arts Research Series, 5), Taipei: Dharma Drum Publishing Corporation, 57–101.


And this one can be found here :nerd_face:

1 Like

I’m curious why the conclusion is that elements of one collection not found in the others are additions, and not possibly the result of texts being “lost” or “rejected”.

Later, you discuss some of these added texts being traceable as expansions of shorter texts found in other canons (which seems like solid evidence), but I’m just curious if there are other reasons for believing this, and if the evidence is different for different suttas.

1 Like

Those are both good questions.

My experience is that it was rare for sutras to be created from scratch. Usually they had some sort of precedent if original sutras were created. Sometimes smaller texts were enlarged with material in other texts, passages were combined in unique ways, [edit: larger texts were divided into smaller ones,] or vague parallels were created that only shared a general theme or subject. And that makes sense for a religious tradition that was attempting to preserve the teachings of its founder. They weren’t intentionally creating new teachings very often. So, all things being equal, my default assumption is that missing parallels were lost at some point.

It was also not uncommon for sutras to be moved from one sutra collection to another by different traditions (if we assume they all started from one version). The additions to the Dīrgha Āgama were probably relocated from other parts of a given tradition’s Tripitaka. We can see that happening in the Theravāda and Sarvāstivāda versions because we have nearly all of their Sutra Pitakas to compare. For the Dharmaguptakas, it’s more of a mystery.

There are a couple reasons why I assume the three versions of the Dīrgha Āgama all grew over time.

We can look at general patterns when comparing the surviving canons. We have the Madhyama and Samyukta Āgama of the Sarvāstivādins, and they are both somewhat larger than their Theravāda counterparts. The Madhyama has 70 more sutras, and the Samyukta has over 13,000 sutras if we count all of the variants. A partial commentary to the Ekottarika Āgama in Chinese describes the Sarvāstivādin Ekottarika as being much larger compared to the one that was translated to Chinese, which contains less than 500 sutras. It sounds like it resembled the Theravāda Anguttara Nikāya, but for all we know, the Sarvāstivāda Ekottarika was even larger. It’s lost today, so we can only guess.

The conclusion I reach then is that the Sarvāstivāda Dīrgha Āgama being about 33% larger than the Theravāda Dīgha Nikāya fits the pattern of their Āgamas being larger. The Theravādins would have lost a large part of their Sutta Piṭaka if the Sarvāstivādins weren’t adding material to their own. They are the outlier, so I think it’s more likely they were doing something different.

When we compare DN to the Dharmaguptaka Dīrgha Āgama, it’s not as striking a difference. We have only their Dīrgha Āgama, Abhidharma, and Vinaya to form conclusions about the Dharmaguptaka’s Tripitaka. What exists of it is very similar to the Theravāda canon, but the two did diverge by adding different things independently. It was much less growth compared to the Sarvāstivādins, but there was some. And that’s what we see when we compare their two versions of the Dīrgha. Some incremental changes that make them different, but they have the majority of the sutras in common.

A second reason is one that I mentioned in the essay. Chapters in EBT collections usually consisted of ten sutras each. The majority of chapters are that size, so it seems likely that later insertions created odd numbered chapters, making many of them a bit more than 10. When I see that the Sarvāstivādin Dīrgha Āgama has three chapters of 6, 18, and 23 sutras, it seems unlikely that it would be the original version of the collection. The chapters should be closer to 10 each, which is what we see in the other two versions. That supports the conclusion that the Sarvāstivādins have rearranged and expanded their Dīrgha Āgama, while the other two are closer to the original version.

The part of my theory that I think is the weakest is my feeling that the Theravada DN is arranged in the order that the collection grew. That is, the first chapter is the oldest part and the last chapter is the most recent. I say that partly because the other two differ significantly, but it depends on whether Sīlakkhandhavagga is really the original core of the collection. I’m guessing that it is because it has a full ten sutras in common across all three collections and has the same title in two of them. It seems too much of a coincidence to me.

Another thing to remember is that we don’t know what the time frame was between the oldest and the newest parts of the collection. It could be a couple decades or a couple centuries. It’s difficult to know for sure since so little is known about the initial centuries of Buddhist history.


According to Ven. YinShun, the Āgamas of the Sarvāstivādins are “adding new texts but not deleting the old ones”. This is why they are somewhat larger than their Pali counterparts. Because of these compilation elements, the Madhyama and Samyukta Āgama of the Sarvāstivādins, and the tradition of the Sarvāstivādins become the most essential textual sources for the comparative study and understanding about the gradual formation of the four Nikāyas/Āgamas. See below text:


巴利的四部,是經過銅鍱部嚴密編纂的 。經文的數目太多,四部間不可能沒有重複;但多數是編入《中部》、《長部》與《增支部》的,在《相應部》與《中部》中,不再保留,所以重複的較少。

這是漢巴聖典(說一切有部與銅鍱部等)再編定時,彼此方法的根本不同。如沒有漢譯的,說一切有部的《雜阿含經》、《中阿含經》,沒有說一切有部的傳說,對於四部阿含的次第形成,是不可能明了的。由於立新而不廢舊,所以儘管有增附的新成分,而在四部阿含成立的研究上,不失為第一流的資料! (pp. 789-790)