Introduction

The focus of the present chapter is on understanding the Ekottarika-āgama collection. By way of providing a basis for an assessment of the Ekottarika-āgama, I begin with a case study of a doubling of the same tale found in different parts of this collection, namely the tale of the former king Mahādeva, already taken up in the two previous chapters. I first translate another version of this tale found in the introduction to the Ekottarika-āgama.[n.1] [8] The text translated below is EĀ 1 at T II 551b26 to 552a22 and T II 553c5 to 553c23; the first of these two parts has already been translated in Huyên-Vi 1985: 40–42. Then I compare narrative and terminological aspects of this version with another version of the same tale that occurs among the Elevens of the Ekottarika-āgama, which I translated in the previous two chapters. Based on the indications that result from this comparison, I then survey relevant information related to the translation of the Ekottarika-āgama.

The shorter Mahādeva tale found in the introduction to the Ekottarika-āgama occurs in the context of a narrative, according to which the monk Uttara, to whom Ānanda entrusts the preservation of the Ekottarika-āgama, had in a former life been a descendant of King Mahādeva. The narrator of the tale is Ānanda, who relates the story of Mahādeva to Mahākassapa.

---

Translation

In the distant past, in this auspicious aeon, the Tathāgata Kakusandha, an arahant, fully and rightly awakened, endowed with knowledge and conduct, a well-gone one, a knower of the world, a supreme person, a leader on the path of Dharma, a teacher of gods and humans, called a Buddha, an assembly of fortunes, had emerged in this world.[n.2] [9] The former Buddha Kakusandha is not mentioned in any of the parallel versions.

At that time there was a king called Mahādeva, who governed by relying on the Dharma, never being unfairly partial. He had an extremely long lifespan and was handsome beyond comparison, rare to find in the world. For eighty-four thousand years he enjoyed himself as a prince, for eighty-four thousand years he governed as a crown prince by relying on the Dharma [of princes], and for eighty-four thousand years he governed the continent by relying on the Dharma of kings.[n.3] [10] A description of the successive periods of eighty-four thousand years spent by Mahādeva as a youth, etc., can also be found in MN 83 at MN II 76,18, MĀ 67 at T I 513a27, and EĀ 50.4 at T II 807a4.

Kassapa,[n.4] [11] For Ānanda, who according to tradition was junior to Kassapa, to address the latter by his proper name stands in an interesting contrast to another episode, noted by von Hinüber 1991: 124 and found at Vin I 92,37. On being asked to assist Mahākassapa in an ordination, Ānanda indicates that he does not dare to pronounce Mahākassapa's name (which he would need to do during the motion). On the way tradition depicts these two disciples and their interrelation cf. also, e.g., Przyluski 1926: 296 and 376f, Migot 1952: 539f, Frauwallner 1956: 161, Bareau 1971a: 140, Tilakaratne 2005, von Hinüber 2008: 25f, and Anālayo 2016b: 171–174. you should know that at that time the Blessed One was dwelling in the Mango Grove.[n.5] [12] MN 83 at MN II 74,15 (Ee as well as Ce) introduces the location as Makhādeva's Mango Grove (Be and Se: Maghadeva's Mango Grove). MĀ 67 at T I 511c24 and D 1 kha 53a1 or Q 1030 ge 48b6 similarly refer to Mahādeva's Mango Grove, 大天㮈林 or lha chen po’i ā mra’i tshal. EĀ 50.4 at T II 806c22 just speaks of Mahādeva's Grove, 大天園. After his meal, he was doing walking meditation in the courtyard, as was his usual custom. I ⟨was⟩ his attendant.[n.6] [13] My translation follows an emendation suggested in the 佛光 edition that replaces 及 with 乃. At that time the Blessed One in turn smiled, and from his mouth five-coloured rays emerged.[n.7] [14] A smile by the Buddha, but without any reference to five-coloured rays, is also reported in MN 83 at MN II 74,16: sitaṃ patvākāsi (Be, Ce, and Se: pātvākāsi), MĀ 67 at T I 511c25: 欣然而笑, EĀ 50.4 at T II 806c24: 便笑, and D 1 kha 53a1 or Q 1030 ge 48b7: ’dzum pa mdzad do. Only T 152 at T III 48b27 has the same motif of the Buddha's smile being accompanied by the manifestation of five-coloured rays: 欣然而笑, 口光五色. Having seen it, I knelt down in front of the Blessed One and said: "Buddhas do not smile in vain. I wish to hear the whole story. A Tathāgata, an arahant, fully and rightly awakened, never smiles in vain."

Then, Kassapa, the Buddha told me: "In the distant past, in this auspicious aeon, a Tathāgata by the name of Kakusandha, an arahant, fully and rightly awakened, emerged in this world. In this place, he taught the Dharma to his disciples in full.[n.8] [15] Adopting a variant without 復於. Again, in this auspicious aeon the Tathāgata Konāgamana, an arahant, fully and rightly awakened, also emerged in the world. At that time in this place here, that Buddha also taught the Dharma to his disciples in full. Again, in this auspicious aeon the Tathāgata Kassapa, an arahant, fully and rightly awakened, emerged in the world. In this place here, the Tathāgata Kassapa also taught the Dharma to his disciples in full."[n.9] [16] A reference to the three former Buddhas is not found in the parallel versions.

At that time, Kassapa, I knelt down in front of the Buddha. I said to the Buddha: "I wish that the Buddha Sakyamuni also teaches the Dharma fully to his disciples in this place here.[n.10] [17] Adopting a variant without 後. Then this place will have served in continuation and without interruption for four Tathāgatas as their diamond seat."[n.11] [18] The motif of sitting on the same place as used by former Buddhas (and by King Mahādeva) is not found in the parallel versions.

At that time, Kassapa, Sakyamuni Buddha sat there and told me: "Ānanda, I have been seated here in the past. In this auspicious aeon a king appeared in the world whose name was Mahādeva … up to … for eighty-four thousand years he edified [his subjects] by relying on the Dharma of kings and educated them by way of virtue.

"After many years, he in turn told his barber:[n.12] [19] Adopting the variant 北 instead of 比. 'If you see white hair on my head, then tell me right away.' Then that person, having heard the king's order, after some years saw a white hair that had appeared on the king's head.[n.13] [20] Whereas MN 83 at MN II 75,6 reports that many years passed by before the barber discovered a white hair, MĀ 67 at T I 513c1, D 1 kha 53b3 or Q 1030 ge 49b1 and D 4094 ju 76b6 or Q 5595 tu 86b6 just indicate that this happened at a later time, 於後時 / dus gzhan zhig. In EĀ 50.4 at T II 808a26, on receiving the instruction, the barber examines the hair for a long time and thereupon discovers a white hair. He in turn knelt down in front of the great king and said: 'Great king, you should know that on your head a white hair has appeared.'

"Then the king told that man: 'Take golden tweezers, pull out the white hair, and place it in my hand.' At that time, having received the king's order, that man in turn took golden tweezers and pulled out the white hair. Then, having seen the white hair, the great king at that time gave expression to a poem:

"'Just now on my head, A hair [indicating] decay and disappearance has appeared,
The divine messenger has come,
It is the proper time to go forth.[n.14] [21] Adopting the variant 應 instead of 當; on variations regarding the reference to a divine messenger cf. below notes 54 and 55.
"Now I have already tasted human happiness, I shall now strive for merits by which one rises to the heavens, shave off my hair and beard, put on the three Dharma robes, and out of firm faith go forth to train in the path and leave behind these many troubles.'[n.15] [22] A reference to an aspiration by Mahādeva to leave behind the many troubles is not reported in the parallel versions.

"At that time King Mahādeva said to his first[born], the crown prince by the name of Dīghāyu: 'Dear, do you know now that on my head a white hair has appeared? My intention is to shave off my hair and beard, put on the three Dharma robes, and out of firm faith go forth to train in the path and leave behind these many troubles.

"You succeed to my throne, govern by relying on the Dharma. Do not neglect this, going against my instruction and acting like an ordinary person. The reason is if there should be such a person, who disobeys my instruction, then he is acting like an ordinary person.[n.16] [23] According to MN 83 at MN II 75,28 and D 4094 ju 77a6 or Q 5595 tu 87b1, by discontinuing the practice instituted by Mahādeva the crown prince would become the last man, antimapurisa or skyes bu tha chad, or according to D 1 kha 54a3 or Q 1030 ge 50a1 just a common man, skyes bu tha shal. MĀ 67 at T I 513c21 indicates that by acting like this he would let people fall into extremes, 人民墮在極邊. According to EĀ 50.4 at T II 808b11, by not following the example set by his father, the crown prince would be like bordercountry people and be reborn in a region without Dharma, 便為邊地人也 … 便生無法處. An ordinary person for long dwells in the three unfortunate destinies and amidst the eight difficult circumstances.'[n.17] [24] Such a reference is not found in the parallel versions.

"At that time King Mahādeva, having handed over the throne to the crown prince and granted treasures to the barber,[n.18] [25] MN 83 at MN II 75,16 reports that the king granted a village to the barber; according to EĀ 50.4 at T II 808b13 he bestowed some farmland on the barber. in that place in turn shaved off his hair and beard, put on the three Dharma robes and out of firm faith went forth to train in the path and leave behind these many troubles. He well cultivated the holy life for eighty-four thousand years, practising the four[fold] even-mindedness (brahmavihāra) of mettā, compassion, [sympathetic] joy, and equanimity. At the end of his life he passed away and was reborn in the Brahmā heaven.

"Then King Dīghāyu kept in mind the instructions of his father, the king, and never gave them up even for a moment. He governed by relying on the Dharma, not being unfairly partial. Soon, after less than ten days, he in turn also became a wheel-turning king endowed with the seven treasures, that is, the seven treasures of the wheel-treasure, the elephant-treasure, the horse-treasure, the jewel-treasure, the precious womantreasure, the steward-treasure, and the general-treasure.[n.19] [26] Whereas MN 83 does not mention the seven treasures at all, the other versions associate these already with Mahādeva; cf. MĀ 67 at T I 512a2, EĀ 50.4 at T II 807a2, D 1 kha 53a6 or Q 1030 ge 49a5, D 4094 ju 76b2 or Q 5595 tu 86b1, and T 152 at T III 48c7. These are reckoned the seven treasures. He also had a thousand sons who were brave and wise,[n.20] [27] A description of the thousand sons of Mahādeva's son Dīghāyu is not found in the parallel versions. The motif of having a thousand vigorous sons occurs in D 1 kha 53b1 or Q 1030 ge 49a6 and in D 4094 ju 76b3 or Q 5595 tu 86b3, where it is associated with Mahādeva instead. able to get rid of many troubles and to dominate the four directions.

"Then King Dīghāyu, [following] the example of the previous king (as above) … composed the poem:[n.21] [28] My translation follows the 宋, 元, and 明 editions, which continue here with the remainder of the discourse. The present section is only found at the end of this fascicle in the Taishō edition, T II 553c5 to 553c23.

"'Just now on my head,
A hair [indicating] decay and disappearance has appeared,
The divine messenger has come,
It is the proper time to go forth.
"Now I have already tasted human happiness. I shall now strive for merits by which one rises to the heavens, shaving off my hair and beard, putting on the three Dharma robes, and out of firm faith go forth to train in the path and leave behind these many troubles.'

"At that time King Dīghāyu said to his first[born], the crown prince by the name of Sudassana:[n.22] [29] My rendering of the third king's name 善觀 as Sudassana is based on another occurrence of the same name in EĀ 38.7 at T II 723a21: 善觀辟支佛, counterpart to the Paccekabuddha Sudassana in its parallel MN 116 at MN III 69,11; cf. also Akanuma 1930/1994: 642. Whereas most of the other versions do not provide the name of the third king, EĀ 50.4 at T II 808b29 gives his name as 冠髻 (with a variant reading as 冠結). 'Dear, do you know now that on my head a white hair has appeared? My intention is to shave off my hair and beard, put on the three Dharma robes and out of firm faith go forth to train in the path and leave behind these many troubles.

"You succeed to my throne, govern by relying on the Dharma. Do not neglect this, going against my instruction and acting like an ordinary person. The reason is that if there should be such a person, who disobeys my instruction, then he is acting like an ordinary person. An ordinary person for long dwells in the three unfortunate destinies and amidst the eight difficult circumstances.'

"Then King Dīghāyu well cultivated the holy life for eightyfour thousand years, practising the four[fold] even-mindedness of mettā, compassion, [sympathetic] joy, and equanimity. At the end of his life he passed away and was reborn in the Brahmā heaven.

"Then King Sudassana kept in mind the instructions by his father, the king, and never gave them up even for a moment. He governed by relying on the Dharma, not being unfairly partial."

[Ānanda said]: "Kassapa, do you know who at that time was Mahādeva? Could he have been someone else? Do not see it like this. The king at that time is now Sakyamuni. He who at that time was the King Dīghāyu is now me, Ānanda.[n.23] [30] EĀ 50.4 at T II 810a27 instead identifies Ānanda as the last of the series of kings who kept up the way of ruling instituted by Mahādeva. According to Jā 9 at Jā I 139,28, Ānanda had been Makhādeva's barber. He who at that time was Sudassana is now the monk Uttara. At that time he constantly accepted the Dharma of kings, never gave it up or forgot it, did not discontinue it. Then King Sudassana, keeping the command of his father, the king, governed by relying on the Dharma and did not discontinue the king's teaching. The reason is that the instructions given by one's father, the king, are a hard thing to disobey."

---

Study

My comparative study of the above tale falls into three parts. In the present first part I examine the reference to three former Buddhas in relation to the often proposed Mahāsāṅghika affiliation of the Ekottarika-āgama. Then I examine narrative differences between the above tale and the other Ekottarika-āgama version, before surveying differences in translation terminology between these two, and finally returning to the issue of the collection's school affiliation at the end of this chapter.

Three Former Buddhas in the Mahādeva Tale

Of the different versions of the Mahādeva tale, the narrative translated above from the introduction to the Ekottarika-āgama is the only one to mention three former Buddhas. The motif of the Buddha's smile that signals the impending delivery of a canonical jātaka, combined with the indication that in the same place three former Buddhas had also been seated, does recur in a different context in the Mahāvastu.[n.24] [32] The Mahāvastu reports the Buddha informing Ānanda that in this spot three former Tathāgatas, arhats, Fully Awakened Ones, had been seated: the Blessed One Krakucchanda, the Blessed One Kanakamuni, and the Blessed One Kāśyapa, Senart 1882b: 318,11: etasmin ānanda pṛthivīpradeśe trayāṇāṃ tathāgatānām arhatāṃ samyaksaṃbuddhānāṃ niṣadyā abhūṣi bhagavato krakucchandasya bhagavato ca kanakamunisya bhagavato ca kāśyapasya. Similar to the description given in the above translated Ekottarika-āgama tale, according to the Mahāvastu this information then motivates Ānanda to invite the Buddha to sit in this place as well, so that it will have been made use of by four Tathāgatas. However, in the Mahāvastu tale the Buddha is not a former king, but rather a young brahmin unwilling to visit the former Buddha Kāśyapa.

The Mahāvastu version of this tale also stands alone among its parallels in bringing in three former Buddhas.[n.25] [33] The parallel versions that also have the episode of the smile mention only the former Buddha Kassapa, MN 81 at MN II 45,14 and MĀ 63 at T I 499a16. The episode of the smile is absent from two other parallels which, however, do refer to the former Buddha Kāśyapa in their narration. These are the Saṅghabheda-vastu of the Mūlasarvāstivāda Vinaya, Gnoli 1978: 22,25, and an Avadāna collection preserved in Chinese translation, T 197.10 at T IV 172c20. Given that the Mahāvastu is a Vinaya text of the Lokottaravāda-Mahāsāṅghika tradition,[n.26] [34] On the Vinaya nature of the Mahāvastu cf. Tournier 2012. this similarity in the motif of the three former Buddhas introducing a past life of the Buddha – as a young brahmin in the Mahāvastu and as a king in the Mahādeva tale – may be one of the reasons why Bareau finds indications that point to a Mahāsāṅghika affiliation in the introduction to the Ekottarika-āgama.[n.27] [35] Bareau 1955: 57 notes that "les indications contenues dans la préface de la traduction chinoise de l'Ekottarâgama montrent que, selon toutes probabilités, la recension de celui-ci ainsi traduite appartenait à une secte mahâsânghika." Bareau does not provide further indications regarding the particular passages he has in mind.

Further perusal of the introductory section brings to light other such indications.[n.28] [36] I already drew attention to these passages in Anālayo 2009k. The introduction to the Ekottarika-āgama reports that at the outset of the first communal recitation, saṅgīti, Ānanda was hesitant to take on the role of reciting the discourses, suggesting that Mahākassapa should rather be given this role.[n.29] [37] EĀ 1 at T II 549b29; T 1507 at T XXV 31c18 explains that Ānanda respected him not only for his seniority, but also because Mahākassapa had been his father for five hundred past lives. The Mahāsāṅghika Vinaya preserved in Chinese translation reports a similar hesitation by Ānanda, who suggests that another monk should take up the role of the reciter.[n.30] [38] T 1425 at T XXII 491b24. Such a hesitation is not mentioned in the accounts of the first saṅgīti in the Vinayas of other schools.[n.31] [39] The Dharmaguptaka Vinaya, T 1428 at T XXII 968b15, the (Haimavata?) *Vinayamātṛkā, T 1463 at T XXIV 818a15, the Mahīśāsaka Vinaya, T 1421 at T XXII 191a18, the Mūlasarvāstivāda Vinaya, T 1451 at T XXIV 406b29, the Sarvāstivāda Vinaya, T 1435 at T XXIII 448b13, and the Theravāda Vinaya, Vin II 287,12.

The introduction to the Ekottarika-āgama also reports that, at the conclusion of Ānanda's exposition of the Ekottarika-āgama, there was an earthquake, a rain of celestial flowers, and the gods in the sky expressed their approval.[n.32] [40] EĀ 1 at T II 550c7: 時地大動, 雨天華香至于膝, 諸天在空歎善哉. Similar miraculous manifestations taking place at the conclusion of the first saṅgīti are reported in the Mahāvastu, whereas Vinayas of other traditions do not mention such occurrences.[n.33] [41] The Mahāvastu, Senart 1882b: 71,12, reports the other monks asking Mahākāśyapa why there is an earthquake, the sound of celestial drums, and a rain of celestial flowers, kin tu, bhoḥ, dhutadharā samakaṃpi, medinī sasaritā sasamudrā, devadundubhiravāś ca manojñā, divyamālyavikiraṇaṃ ca bhavanti?

Yet, the same introductory section also presents problems with the Mahāsāṅghika identification. The introduction to the Ekottarika-āgama reports that at the first saṅgīti Ānanda recited the four Āgamas in the sequence Ekottarika, Madhyama, Dīrgha, Saṃyukta.[n.34] [42] EĀ 1 at T II 549c28: 契經今當分四段, 先名增一, 二名中, 三名曰長多瓔 珞, 雜經在後為四分 (adopting the variant 先 instead of 次); according to T 1507 at T XXV 32a23, the division into four Āgamas and the placing of the Ekottarika in first place was originally devised by Ānanda. According to the Mahāsāṅghika Vinaya preserved in Chinese translation, however, he rather recited them in the sequence Dīrgha, Madhyama, Saṃyukta, Ekottarika.[n.35] [43] T 1425 at T XXII 491c16: 尊者阿難誦如是等一切法藏, 文句長者集為長 阿含, 文句中者集為中阿含, 文句雜者集為雜阿含, 所謂根雜力雜覺雜道雜, 如是比等名為雜, 一增, 二增, 三增, 乃至百增, 隨其數類相從, 集為增一阿含. Again, according to the introduction to the Ekottarika-āgama the Kṣudraka collection contains Mahāyāna scriptures. [n.36] [44] EĀ 1 at T II 550c10: 方等大乘義玄邃, 及諸契經為雜藏. The Mahāsāṅghika Vinaya instead notes that the Kṣudraka collection contains tales related to Pratyekabuddhas and arhats.[n.37] [45] T 1425 at T XXII 491c20: 雜藏者, 所謂辟支佛, 阿羅漢, 自說本行因緣.

Although these two indications do not sit too well with the Mahāsāṅghika hypothesis, they could be the result of an overall trend in the introduction to the Ekottarika-āgama towards enhancing itself vis-à-vis other Āgamas and towards giving predominance to Mahāyāna teachings. In such a context, a rearrangement of the sequence of the Āgamas that places the Ekottarika in first position would be a natural occurrence, just as an interpretation of the Kṣudraka collection as being the canonical repository of Mahāyāna teachings.

The precise nature of the Kṣudraka collection is in fact not a straightforward matter.[n.38] [46] For a survey of different versions of this collection cf. Lamotte 1956. In the Theravāda tradition one also finds different perspectives on this matter. The reciters of the Dīgha-nikāya differ from those who recite the Majjhima-nikāya on whether at the first saṅgīti the Khuddaka-nikāya was allocated to the basket of discourses or to the basket of Abhidharma.[n.39] [47] Sv I 15,22. Modern-day Theravāda traditions also disagree on what works should be included in this collection and thereby considered canonical, with the Burmese incorporating works such as the Nettippakaraṇa, the Milindapañha, and the Peṭakopadesa, whereas in other Theravāda countries these are not considered to be part of the Khuddaka-nikāya.[n.40] [48] Cf., e.g., Abeynayake 1984: 33–46, Collins 1990: 108 note 11, and von Hinüber 1996/1997: 42f.

Although the two indications contrary to the Mahāsāṅghika hypothesis are not decisive, it needs to be noted that problems with the Mahāsāṅghika hypothesis also manifest elsewhere in the Ekottarika-āgama. One example is the regular reference in this collection to twelve aṅgas,[n.41] [49] A reference to twelve aṅgas can be found in EĀ 29.5 at T II 657a2, EĀ 39.1 at T II 728c6, EĀ 49.1 at T II 794b14, and EĀ 50.8 at T II 813a25; for a survey of these listings cf. Nattier 2004b: 193f. whereas the Mahāsāṅghika Vinaya speaks of only nine.[n.42] [50] T 1425 at T XXII 227b25; a discrepancy already noted by Hirakawa 1963: 63f. This discrepancy could be due to the Vinaya preserving an earlier record of the listing of aṅgas than the Ekottarika-āgama, given that the listing of twelve appears to have developed out of an earlier listing of nine.[n.43] [51] Lamotte 1956: 263 note 2, Kalupahana 1965: 616, von Hinüber 1994: 122, and Nattier 2004b: 168. Therefore this is also not conclusive.

The same also applies to a discrepancy in the count of monastic rules, where the Ekottarika-āgama speaks of two hundred and fifty rules,[n.44] [52] EĀ 48.2 at T II 787b10: 有二百五十戒. a number the actual count in the Mahāsāṅghika Vinaya preserved in Chinese translation does not seem to reach.[n.45] [53] The Mahāsāṅghika Vinaya, T 1426 at T XXII 555b15, gives the following listing: 已說四波羅夷法, 已說十三僧伽婆尸沙法, 已說二不定法, 已說三 十尼薩耆波夜提法, 已說九十二波夜提法, 已說四波羅提提舍尼法, 已說眾 學法, where the śaikṣa rules are not counted. According to Pachow 1955: 11, the śaikṣa rules are sixty-six (cf. T 1425 at T XXII 399b7), which together with the seven adhikaraṇa-śamatha (where it is anyway doubtful if these should be considered as "rules" properly speaking) would result in an overall count of two hundred and eighteen rules. That this is also not conclusive can be seen by turning to the Theravāda tradition, where an even more significant discrepancy in this respect occurs. Discourses in the Aṅguttara-nikāya speak of just over a hundred and fifty rules,[n.46] [54] AN 3.83 at AN I 230,17: sādhikaṃ … diyaḍḍhasikkhāpadasataṃ, an expression found again in AN 3.85 at A I 231,18, AN 3.86 at A I 232,33, and AN 3.87 at A I 234,11. whereas the actual count of rules in the Theravāda Vinaya results in two hundred and twenty-seven.[n.47] [55] The Theravāda pātimokkha for bhikkhus comprises four pārājika, thirteen saṅghādisesa, two aniyata, thirty nissagiya pācittiya, ninety-two pācittiya, four pāṭidesanīya, seventy-five sekhiya, and seven adhikaraṇasamatha. According to an explanation proposed by the Pāli commentary, not all of the rules recorded in the Vinaya had been promulgated by the time of the coming into being of these Aṅguttara-nikāya discourses.[n.48] [56] Mp II 346,30: tasmiṃ samaye paññattasikkhāpadān' eva sandhāy' etaṃ vuttaṃ; cf. also, e.g., Dutt 1924/1996: 75f, Law 1933: 21, Bhagvat 1939: 64, Pachow 1955: 8f, Misra 1972: 33, Dhirasekera 1982/2007: 145, and von Hinüber 1995: 15.

In this way, these various points for or against the Mahāsāṅghika hypothesis do not seem to allow the drawing of a definite conclusion. Moreover, assessing the significance of Mahāsāṅghika elements in the introductory section, such as the appearance of three Buddhas in the Mahādeva tale found in the introduction to the Ekottarika-āgama, would also depend on ascertaining when this introduction became the preface to the Ekottarika-āgama collection and when it reached its present form.

Narrative Differences between the Two Mahādeva Tales

The appearance of three former Buddhas in the Mahādeva tale in the introduction to the Ekottarika-āgama is also remarkable in so far as the version of the Mahādeva tale found among the Elevens does not have any reference to past Buddhas. Such a difference would be more easily understandable if this motif had been found in the longer version only, as one could then imagine that it had been abbreviated in the shorter extract. Given that the version in the introduction is the shorter of the two, it is remarkable that the motif of three Buddhas does not recur in the longer version of the same tale in the same discourse collection.

Elsewhere the Ekottarika-āgama shows considerable interest in past Buddhas, to the extent that it has two records of the past Buddha Dīpaṃkara predicting the future Buddhahood of the one who was to become the Buddha Sakyamuni.[n.49] [57] EĀ 20.3 at T II 599b14 and EĀ 43.2 at T II 758b26. Dīpaṃkara is not mentioned in the early discourses in the four Pāli Nikāyas or in other Chinese Āgamas. In the Theravāda tradition, his predicting of Sakyamuni's Buddhahood occurs only in the late Buddhavaṃsa, stanza 2.60 at Bv 13,1; cf. also below p. 429f note 50. This makes it safe to conclude that there would have been no reason for the reciters of the Ekottarika-āgama to exclude a reference to former Buddhas from the version of the Mahādeva tale now found among the Elevens, had such a reference been originally found in it.

The absence of any reference to former Buddhas is not the only narrative difference between the two Mahādeva tales. The version found among the Elevens also does not report any miraculous manifestations accompanying the Buddha's smile. According to the description given in the introductory section to the Ekottarika-āgama, "the Blessed One smiled and from his mouth five-coloured rays emerged". Elsewhere the Ekottarika-āgama recurrently describes various miraculous events, so that there would have been little reason for the reciters of the longer version of the Mahādeva tale to omit a description of the fivecoloured rays, had this been part of the story they had received from their predecessors.

Another difference is that the version in the introduction does not present Mahādeva as a wheel-turning king,[n.50] [58] The qualification of Mahādeva as "governing the continent" in EĀ 1 at T II 551c4: 治化天下 does not seem to imply world dominion and thus would not be an implicit reference to his status as a wheel-turning king. EĀ 50.4 at T II 806c29 describes his dominion as encompassing all four continents (which according to ancient Indian cosmology make up the whole world): 四天下, an expression found also in the description of Mahādeva's dominion as a wheelturning king in T 152 at T III 48c19: 四天下. MĀ 67 at T I 513c13 uses the same expression 四天下 when describing the dominion handed over by Mahādeva to his son. MĀ 67 at T I 512a1 also employs the expression 天下 when depicting the range of his army, which I take to imply that, without the wheeltreasure that opens up the path through the ocean, his army would have been able to control only Jambudvīpa as one of the four continents. In sum, I take it that the reference in EĀ 1 at T II 551c4 to 天下 would refer to Jambudvīpa, not to the whole world of four continents that are governed by a wheel-turning king. but only qualifies his son in this way. Comparing the different versions of the Mahādeva tale gives the impression that the motif of the wheelturning king was subsequently added to the narration.[n.51] [59] Cf. also above p. 128. The Pāli version does not have the motif of the wheel-turning king at all and thus appears to testify to an early stage of textual development, before this motif had made an impact on the discourse. The Bhaiṣajyavastu version and the discourse quotation in Śamathadeva's Abhidharmakośopāyikāṭīkā identify Mahādeva as a wheelturning king and briefly list his seven treasures.[n.52] [60] D 1 kha 53a7 or Q 1030 ge 49a5 and D 4094 ju 76b2 or Q 5595 tu 86b1. The Madhyama-āgama discourse and the discourse found among the Elevens of the Ekottarika-āgama not only identify Mahādeva as a wheelturning king, but also provide a detailed description of each of his seven treasures.[n.53] [61] MĀ 67 at T I 512a3 and EĀ 50.4 at T II 807a2.

When viewed against what appears to be a gradual development in the parallels, the Mahādeva tale in the introduction to the Ekottarika-āgama seems to reflect an intermediary stage in the application of the motif of the wheel-turning king to the Mahādeva tale, as this conception already manifests in relation to Mahādeva's son, whose seven treasures are listed without a detailed description, but has not yet been applied to Mahādeva himself.

The stanzas spoken by Mahādeva, once white hair has been discovered, vary in each of the extant versions. A noteworthy detail here is that the other versions agree in speaking of the white hair as a "divine messenger",[n.54] [63] MN 83 at MN II 75,18 and Jā 9 at Jā I 138,24 use the expression devadūta, MĀ 67 at T I 513c8 and EĀ 1 at T II 552a1 the corresponding expression 天使, and D 1 kha 53b6 or Q 1030 49b4 and D 4094 ju 77a2 or Q 5595 tu 87a3 the equivalent lha yi pho nya. with the sole exception of the tale found among the Elevens of the Ekottarika-āgama, which instead speaks of the "body's messenger".[n.55] [64] EĀ 50.4 at T II 808b1: 身使. The reference in the tale among the Elevens appears to be the result of a confusion between deva and deha, a confusion that evidently did not occur in the tale found in the introduction to the Ekottarika-āgama.

All versions agree that the sight of the white hair stirred Mahādeva to go forth, with the version in the introduction to the Ekottarika-āgama being the only one to add to his inspiration to go forth that he wanted to "leave behind these many troubles".[n.56] [65] EĀ 1 at T II 552a4: 離於眾苦. On deciding to go forth, according to the introduction Mahādeva bestows treasures on the barber.[n.57] [66] EĀ 1 at T II 552a11: 財寶. According to the version among the Elevens, he gives him some farmland.[n.58] [67] EĀ 50.4 at T II 808b13: 田業.

The two Ekottarika-āgama versions also disagree on who of their protagonists should be identified as being a past life of Ānanda. According to the account found in the introduction, Ānanda was the son of Mahādeva. According to the discourse found among the Elevens, he was the last in a series of eighty-four thousand generations of descendants of Mahādeva who kept up the custom instituted by Mahādeva.[n.59] [68] EĀ 1 at T II 553c20 and EĀ 50.4 at T II 810a28 (where he is then followed by a king who did not keep up the custom instituted by Mahādeva, identified in EĀ 50.4 as a past life of Devadatta); according to T 1507 at T XXV 34b28, Ānanda was the wheel-turning king Dīghāyu (and thus the son of Mahādeva): 阿難白引往昔為轉輪聖王, 名曰長壽. In sum, the version in the introduction differs from its Ekottarika-āgama counterpart as follows:

• mention of three former Buddhas,
• description of miraculous manifestations accompanying the Buddha's smile,
• Mahādeva is not introduced as a wheel-turning king,
• white hair is a "divine messenger", instead of the "body's messenger",
• Mahādeva goes forth and "leaves behind these many troubles",
• Mahādeva bestows on the barber treasures, instead of farmland,
• Ānanda was the son of Mahādeva.

The differences between two versions of the Mahādeva tale in the Ekottarika-āgama are of such magnitude that it seems safe to conclude that they belong to different transmission lineages. In other words, these two are not a shorter and a longer version of the same story with some variations, but rather two different narrative developments of the same motif.

When evaluated from the perspective of oral transmission, it seems highly unlikely that these two different versions of the Mahādeva tale could have come into being within the same text. Not only their coming into being must have happened independently, it seems also difficult to imagine that they would have been transmitted orally over long periods as parts of the same text. Oral transmission tends to stereotype, wherefore within a single text differences naturally tend to become less during the period of transmission.[n.60] [69] With this suggestion I do not intend to propose that it is impossible for some differences to exist within the same orally transmitted text. An example from the Aṅguttara-nikāya would be AN 6.44 at AN III 347,16 and AN 10.75 at AN V 137,19, where the same introductory narration leads to two different replies given by the Buddha, as a result of which one discourse is found among the Sixes, whereas the other is located among the Tens. Here two diverse records of how the Buddha reacted to a particular situation have been transmitted alongside each other within the same text. The case of the two Mahādeva tales is different, however, in that an execution of the same narrative motif within the same textual collection has resulted in numerous differences of various types.

For the two versions to stand side by side within the same text, exhibiting the number of differences they do, the most natural scenario would be that one of them is a later addition to the Ekottarika-āgama at a time when this collection was no longer transmitted orally.

Different Translation Terminology in the Two Mahādeva Tales

Due to their diverse narrative coverage, the two versions of the Mahādeva tale in the Ekottarika-āgama have only four proper names in common (leaving aside the name of Ānanda in the frame story). These four are the name of the location and the names of the first three kings. In the other versions, the location is invariably given as Ma(k)hādeva's Mango Grove,[n.61] [70] In addition to being found in the parallel versions of the Mahādeva tale, the name of this grove occurs also in the Mahāvastu, Senart 1897: 450,18, which reports that a ṛsi by the name of Śroṇaka went to stay in the mahādevāmravana, thereby also employing the king's name together with an indication that this was a mango grove. combining the name of the king with an indication of the type of grove in which he lived. In the Ekottarika-āgama versions, however, the account in the introduction only speaks of the type of grove, whereas the version among the Elevens speaks just of Mahādeva's Grove and thus does not refer to the kind of trees found in this grove.[n.62] [71] EĀ 1 at T II 551c5: 甘梨園 and EĀ 50.4 at T II 806c22: 大天園. However, these different renderings may just be based on different Indic originals.

In the case of the proper names of the kings, the names of the third king in the two versions definitely go back to different originals.[n.63] [72] EĀ 1 at T II 553c9: 善觀 and EĀ 50.4 at T II 808b29: 冠髻 (or 冠結). The names of the first and second king, however, can safely be considered as being based on the same Indic term. Whereas the version in the introduction employs a transcription of Mahādeva as 摩訶提婆, the version among the Elevens translates his name as 大天.[n.64] [73] EĀ 1 at T II 551b29 and EĀ 50.4 at T II 807a1. In the case of the second king, both versions opt for a translation of the name, the introduction using 長壽, whereas the version among the Elevens instead employs 長生.[n.65] [74] EĀ 1 at T II 552a5 and EĀ 50.4 at T II 808b4.

The net result of this is that all four proper names that are common to the two actual tales differ. When evaluating this finding, however, it needs to be kept in mind that in the course of translating a whole work it can easily happen that the same proper name is rendered differently. Thus, for example, a listing of famous seers found twice in the Dīrgha-āgama, by the same translator Zhú Fóniàn (竺佛念) who is also credited with translating the Ekottarika-āgama, shows several variations.[n.66] [75] DĀ 10 at T I 87a16 and DĀ 26 at T I 105b24; cf. also Meisig 1990. Listings of the same seers in the Madhyama-āgama also show variations within the same collection.[n.67] [76] MĀ 152 at T I 667c23 (repeated at 668a7) and MĀ 158 at T I 680c6, which show the following variations for some of the names of these famous ṛsis: 毗 奢蜜哆羅 / 毗奢蜜哆邏, 夜婆陀揵尼 / 夜陀揵尼, 應疑羅婆 / 應疑羅娑, and 婆惒 / 婆和 (in the first case the difference occurs already within MĀ 152, where the reading in the main text of the first instance is found as a variant reading for the second instance, the reverse then being the case for MĀ 158). A particularly striking case occurs in a Saṃyukta-āgama discourse, where the name of the monk Koṇḍañña is rendered in two different ways in the same discourse, with the shift from one transcription to the other occurring rather suddenly, in the middle of the text.[n.68] [77] SĀ 379 at T II 104a11 reports that the Buddha, who has just set in motion the wheel of Dharma, for a second time asks 憍陳如 (=Koṇḍañña) if he has understood the Dharma, whereupon 拘隣 (=Koṇḍañña) replies that he has indeed understood, 復告尊者憍陳如: 知法未? 拘隣白佛: 已知. Both transcriptions are listed in Akanuma 1930/1994: 43 as alternatives for the same name of Aññāta Koṇḍañña. Up to this point, SĀ 379 has been using the transcription 憍陳如, employed also in parallel versions to this discourse found in the Dharmaguptaka Vinaya, T 1428 at T XXII 788b24, the Mahīśāsaka Vinaya, T 1421 at T XXII 104c18, the Mūlasarvāstivāda Vinaya, T 1450 at T XXIV 128a9 as well as T 1451 at T XXIV 292b29 and at T XXIV 406c5, and the Sarvāstivāda Vinaya, T 1435 at T XXIII 448c14. From this point onwards, SĀ 379 keeps on using 拘隣, a transcription found also in a version of this discourse in the Ekottarika-āgama, EĀ 24.5 at T II 619b6. Unlike SĀ 379, the other versions are consistent in their translation terminology. Su 2010: 28 note 46 suggests that this sudden shift of terminology in SĀ 379 could be the result of an incompletely carried out revision, during which earlier instances of 拘隣 were replaced with 憍陳如. Such instances indicate that the lack of consistency in the rendering of proper names in the two versions of the Mahādeva tale found in the Ekottarika-āgama is in itself far from being conclusive and the two versions could still stem from the same translator.

Besides proper names, however, the two versions also show other differences in translation terminology. Mahādeva's rule as a king is depicted in the introduction with the phrase 以法治化, whereas the discourse among the Elevens employs the expression 治以正法. Here the difference between the reference to dharma (法) and to saddharma (正法) may simply be due to different expressions in the originals.[n.69] [78] EĀ 1 at T II 551c1 and EĀ 50.4 at T II 807a1.

When taking up the length of his lifespan, the introduction uses the expression 壽命, whereas the version among the Elevens just speaks of 壽.[n.70] [79] EĀ 1 at T II 551c1 and EĀ 50.4 at T II 807a1. The usage in EĀ 50.4 would explain the choice of the rendering 長生 instead of 長壽 for the second king. Again, when describing the periods of Mahādeva's life as a young prince and as a crown prince, the introduction uses the expressions 童子身 and 太子身, whereas the version among the Elevens speaks of 童子時 and 太子時.[n.71] [80] EĀ 1 at T II 551c2 and EĀ 50.4 at T II 807a5.

Although the above variations in the expressions used to render what are specific aspects of the story are not of much significance, variations also occur in relation to what would be pericope descriptions. When Ānanda kneels down to ask the Buddha why he smiled, the introduction uses the expression 前長跪, whereas the version among the Elevens instead employs 右膝著 地.[n.72] [81] EĀ 1 at T II 551c7 and EĀ 50.4 at T II 806c26.

When it comes to Mahādeva's going forth, the introduction narrates that he "shaved off hair and beard", 剃除鬚髮, and "put on three Dharma robes", 著三法衣, in order to "train in the path", 學 道. According to the discourse among the Elevens, he "removed hair and beard", 下鬚髮, and "put on Dharma robes", 著法服, in order to "enter on the path", 入道.[n.73] [82] EĀ 1 at T II 552a3 and EĀ 50.4 at T II 808b14. In the case of the expression 下鬚髮, besides nine occurrences in EĀ 50.4, the same EĀ 50.4 at T II 808b6 and 810a3 also has the alternative phrase 剃鬚髮.

Although one would expect a pericope to be rendered with some degree of consistency, each of these variations is in itself still not conclusive. Given the time gap that must have occurred between the translation of the introduction to the Ekottarika-āgama and the rendering of a discourse found in its last part of Elevens, it would not be surprising if the translator was not consistent in his renderings. In fact Zhú Fóniàn does not appear to have been a translator with particularly consistent translation terminology.

Moreover, Dào'ān (道安) reports that the translation of the Ekottarika-āgama was undertaken during a time of warfare and therefore under conditions that would certainly not have been conducive to a consistency check of the translation terminology.[n.74] [84] T II 549a18. In fact the Ekottarika-āgama collection often shows a considerable degree of inconsistency of translation terminology.

What gives further weight to these variations, however, is the circumstance that some of the expressions used in the discourse found among the Elevens do not make an appearance elsewhere in the Ekottarika-āgama.[n.75] [85] In what follows, my indications are based on a digital search of the CBETA edition which, with its manifold advantages, also comes with the limitations that inevitably go with digital searching. This is the case for the complete phase 著法服, as only 法服 on its own occurs in other Ekottarika-āgama discourses. In contrast, the expression found in the version in the introduction, 著三法衣, occurs frequently elsewhere in the Ekottarika-āgama.[n.76] [86] EĀ 9.1 at T II 562a27, EĀ 9.2 at T II 562b20, EĀ 16.4 at T II 579b26, EĀ 24.2 at T II 616c8, EĀ 24.4 at T II 618a23, EĀ 28.4 at T I 652a5, EĀ 29.9 at T II 658c7, EĀ 32.4 at T II 676b22, EĀ 35.7 at T II 700b23, EĀ 38.6 at T II 720b22, EĀ 38.7 at T II 723b15, EĀ 38.11 at T II 726a15, EĀ 41.5 at T II 739b29, EĀ 42.3 at T II 752c1, EĀ 42.4 at T II 753b9, EĀ 43.7 at T II 763c21, EĀ 45.5 at T II 773a12, EĀ 46.10 at T II 780b28, EĀ 47.9 at T II 784c12, EĀ 49.9 at T II 804c11, EĀ 50.8 at T II 812c29, and EĀ 51.3 at T II 816a9 (here and below, I only take into account separate discourses, disregarding recurrence of a particular expression in the same discourse). In such contexts, the phrase 學道, found in the introduction version, occurs also regularly, whereas 入道 makes its appearance only rarely in the Ekottarika-āgama, and in any case not as part of the pericope description of someone's going forth.[n.77] [87] Except for EĀ 50.4, I have only been able to locate this expression in two discourses. One of these is EĀ 8.3 at T II 561a19+24, where the phrase 入道 is used in a reference to the Tathāgata. The other is EĀ 12.1 at T II 568a2+9 and 569b9, where the expression 一入道 qualifies the practice of satipaṭṭhāna, thereby being a counterpart to the expression ekāyano maggo in the parallel MN 10 at MN I 55,31 (= DN 22 at DN II 290,8); on this expression cf. the discussion in Gethin 1992: 59–66, Kuan 2001: 164, Anālayo 2003: 27–29, Sujato 2005: 177–186, Harrison 2007: 208, Nattier 2007: 196–199, Wen 2011, and Anālayo 2013g. Again, the expression 剃除鬚髮, found in the introduction, is standard in descriptions of going forth (if beard and hair are mentioned at all), whereas the phrase 下鬚髮 does not seem to recur elsewhere in the Ekottarika-āgama, the discourse among the Elevens being the only instance where this expression occurs.

In other words, the expressions 著三法衣, 入道, and 下鬚髮 to describe Mahādeva's going forth in the discourse found among the Elevens appear to be specific to this particular text, differing from the terminology that is used in such contexts elsewhere in the Ekottarika-āgama. This gives the impression that such variations are not just different choices employed by the same translator who is working on a long text in difficult conditions. It would not be easy to devise a reasonable explanation why the same translator should change his translation terminology just for this one discourse.

The above instances are not the only ones of this type. The discourse among the Elevens mentions that after the meal the Buddha got up, 食後起.[n.78] [88] EĀ 50.4 at T II 806c23. The version of the Mahādeva tale in the introduction and two other discourses in the Ekottarika-āgama that report what the Buddha did after his meal do not mention that he got up and thus just use the phrase 食後.[n.79] [89] EĀ 1 at T II 551c5, EĀ 17.1 at T II 581c13, and EĀ 41.10 at T II 743a6.

The discourse among the Elevens provides the information that the Buddha was staying in [Mahādeva's] grove with the expression 園中止.[n.80] [90] EĀ 50.4 at T II 806c22. One instance that bears some similarity to the expression used in EĀ 50.4 can be found in EĀ 51.7 at T II 818b21, where Anāthapiṇḍika informs his son of the fact that the Buddha is "staying in my grove", 止吾園中. As in this case the 止 precedes the reference to the grove, the formulation does not match the expression found in EĀ 50.4. This phrase does not appear to recur elsewhere in the Ekottarika-āgama, which instead tends to report the Buddha's sojourn in a grove simply with 園中, as is the case for the Mahādeva tale found in the introduction,[n.81] [91] EĀ 1 at T II 551c5. Other occurrences of this type are too numerous to be listed separately. or even just with 園.

The Mahādeva tale located among the Elevens begins by indicating that the Buddha was in the company of a great community of monks, 與大比丘僧, numbering one thousand two hundred and fifty.[n.82] [92] EĀ 50.4 at T II 806c22. Whereas the version found in the introduction does not report the number of monks that were staying together with the Buddha, other discourses in the Ekottarika-āgama that present the Buddha in the company of the same number of monks instead use the expression 與大比丘眾,[n.83] [93] EĀ 30.3 at T II 660a2 and EĀ 45.7 at T II 773c21. an expression also regularly employed when the number of monks is instead five hundred, or even more than one thousand two hundred and fifty.[n.84] [94] Although references to five hundred monks are too numerous to be listed separately, examples of the phrase 與大比丘眾 used in relation to the much higher number of monks that were held to have accompanied previous Buddhas can be found in EĀ 20.3 at T II 597b18, EĀ 23.1 at T II 610b1, EĀ 30.3 at T II 665a13, EĀ 43.2 at T II 758a8, and EĀ 52.2 at T II 824a23. That is, the usual rendering of such references to the community of monks accompanying the Buddha in the Ekottarika-āgama employs 眾 instead of 僧.

The version among the Elevens describes the Buddha addressing Ānanda with the phrase 佛語阿難.[n.85] [95] EĀ 50.4 at T II 806c28, which is the first of twenty occurrences of this phrase in this discourse, although in one instance in EĀ 50.4 at T II 809a22 the phrase 佛告阿難 can be found, with the (for EĀ 50.4) probably more original variant reading 佛語阿難. This formulation does not appear to occur elsewhere in the Ekottarika-āgama,[n.86] [96] There are, however, two instances where the expression 佛語 occurs when the Buddha addresses someone else; cf. EĀ 31.2 at T II 667b20 and EĀ 33.2 at T II 686a14. which instead records instances where Ānanda is being addressed by his teacher with the expression 佛告阿難,[n.87] [97] EĀ 32.5 at T II 676c2, EĀ 40.5 at T II 739b18, EĀ 40.10 at T II 743c27, EĀ 42.3 at T II 750c22, EĀ 44.10 at T II 768c8, EĀ 45.2 at T II 770c22, EĀ 45.3 at T II 772a17, EĀ 45.5 at T II 773a19, EĀ 46.8 at T I 780a13, EĀ 48.2 at T II 786b8, EĀ 48.3 at T II 787c11, EĀ 48.4 at T II 791b8, EĀ 49.8 at T II 802a1, EĀ 49.9 at T II 804a12, and EĀ 51.8 at T II 820b13. or alternatively 世尊告 阿難. In other words, the verb employed in the Ekottarika-āgama in such contexts is 告 instead of 語. The version in the introduction uses the corresponding phrase 佛告我, where the proper name 阿難 is not mentioned since here Ānanda himself narrates what happened.[n.88] [98] EĀ 1 at T II 551c9.

A similar pattern holds in cases when Ānanda asks the Buddha a question. The Mahādeva tale among the Elevens uses the form 阿難問佛,[n.89] [99] EĀ 50.4 at T II 807a6, which together with the abbreviation 問佛 and the alternative expression 阿難復問佛 occurs ten times in the discourse. not found elsewhere in the Ekottarika-āgama, which instead uses 阿難白佛言 when Ānanda is asking the Buddha a question.[n.90] [100] EĀ 23.5 at T II 613b21, EĀ 23.6 at T II 613c24, EĀ 26.9 at T II 642a24, EĀ 36.5 at T II 703b27, EĀ 40.5 at T II 739b22, EĀ 42.3 at T II 751c11, EĀ 45.5 at T II 773a18, EĀ 48.2 at T II 787b22, and EĀ 51.8 at T II 820b15. When Ānanda asks several questions, the phrase for subsequent instances then tends to become 阿難復白佛言.

It seems to me that these examples of phrases that describe standard situations are fairly conclusive evidence of different translators at work. With all due consideration given to variations in terminology during a prolonged translation carried out under difficult circumstances, the above pattern clearly points to a peculiarity of the idioms employed in the Mahādeva tale among the Elevens that differs markedly from the rest of the Ekottarika-āgama collection.

This impression finds further confirmation in other variations between the two versions of the Mahādeva tale that involve key terminology which could reasonably well be expected to be used consistently by the same translator. This would be the case at least when it comes to translating a single work, which does not leave much time for the translator to change his translation terminology in the way this might happen with works by the same translator rendered at different stages of his working career.

One example is the householder-treasure, one of the seven treasures that are the property of a wheel-turning king. Whereas the introduction speaks of 典藏寶, the discourse among the Elevens uses the phrase 主藏寶.[n.91] [101] EĀ 1 at T II 552a18 and EĀ 50.4 at T II 807a4. The expression used in the introduction recurs in another listing of the seven treasures in the Ekottarika-āgama,[n.92] [102] EĀ 36.5 at T II 707c11. but the phrase employed in the discourse found among the Elevens does not seem to recur elsewhere in the Ekottarika-āgama.

In this case, however, it could still be that the same translator has come up with different ways of rendering the householdertreasure, as other discourses in the Ekottarika-āgama employ still another rendering as 居士寶.[n.93] [103] EĀ 17.7 at T II 583b28, EĀ 23.1 at T II 609c15, EĀ 24.4 at T II 617b29, EĀ 39.7 at T II 731b17, and EĀ 39.8 at T II 731c21. Yet another rendering can be found in EĀ 48.3 at T II 788a12: 守藏之寶 (which also differs from the other instances by having this treasure as its last). Such an explanation would, however, not be convincing when it comes to another discrepancy that involves the rendering of the term arahant. This discrepancy occurs in the context of a standard set of epithets that describes the Buddha as being an arahant who is fully awakened. The introduction uses the expression 至真, followed by 等正覺. In contrast, the version among the Elevens employs 無所著, followed by the same 等正覺.[n.94] [104] EĀ 1 at T II 551b27 and EĀ 50.4 at T II 806b27; on 至真 and 無所著 cf. the discussion in Nattier 2003b: 214 and 217ff. The qualification of a Buddha as 至真, 等正 覺 appears to be the standard translation used elsewhere in the Ekottarika-āgama.[n.95] [105] Instances of this usage are too numerous to be listed fully, hence I content myself with just giving the first few references in the collection: EĀ 10.10 at T II 566a17, EĀ 12.4 at T II 569c7, EĀ 13.5 at T II 574a27, EĀ 17.2 at T II 582c28, etc. An exception to this pattern is EĀ 26.9 at T II 639b9, which uses the expression 無所著, 等正覺 to qualify the Buddha, although the same discourse also has the standard rendering 至真 at T II 639c22. This discourse has several elements testifying to late influence, particularly evident in an explicit reference to the Hīnayāna, EĀ 26.9 at T II 640a5; for a more detailed study cf. below p. 473ff. Thus the occurrence of the expression 無所著, 等正覺 may well be an indication that EĀ 26.9 contains material that is not original to the Ekottarika-āgama collection.

A similar type of difference also manifests in relation to the term brahmavihāra, which the introduction renders as 四等心, whereas the discourse among the Elevens employs 四梵行.[n.96] [106] EĀ 1 at T II 552a14 and EĀ 50.4 at T II 808b15. The introduction's version's expression 四等心 recurs in several other discourses in the Ekottarika-āgama,[n.97] [107] EĀ 24.6 at T II 624b29, EĀ 26.9 at T II 639b5, EĀ 27.8 at T II 646b6, EĀ 29.10 at T II 658c19, EĀ 31.2 at T II 667c21, EĀ 38.9 at T II 724b24, and EĀ 48.3 at T II 789b12. but the rendering 四梵行 does not seem to occur anywhere else apart from the Mahādeva discourse found among the Elevens.

On considering all of the above noted differences, it seems to me unavoidable to conclude that the two versions of the Mahādeva tale in the Ekottarika-āgama were not translated by the same person. The magnitude of differences surveyed so far needs to be considered against the background that the two versions of the Mahādeva tale have relatively little text in common. The first part of the rather short introductory account, which refers to three former Buddhas and reports Ānanda's request for the Buddha to sit in the same place, is without a counterpart in the discourse found among the Elevens. Instead of a reference to former Buddhas, the discourse among the Elevens has a long description of the seven treasures of a wheel-turning king. Moreover, this version continues after the third generation of kings with its narration leading up to the final of eighty-four thousand generations of kings, exploring his life and conduct in much detail, none of which is mentioned at all in the introduction to the Ekottarika-āgama.

In this way, what the two versions have in common is the tale of Mahādeva who on seeing the first white hair renounces his throne and goes into seclusion to practise the brahmavihāras. The introduction then repeats this with few variations for the next two kings, after which it concludes. That is, the actual amount of text that is common to the two versions is fairly small. To find such a number of variations in this relatively short portion of text is surely significant. With all due consideration to the possibility of scribal errors and inconsistency of translation terminology by the same translator, the differences noted above must be the result of different translators at work.

The evidence surveyed so far shows that the two versions of the Mahādeva tale in the Ekottarika-āgama appear to be based on different original narrations, which were then translated by different translators. The version found among the Elevens shows recurrent disagreements with translation terminology employed elsewhere in the Ekottarika-āgama, indicating that this discourse was not part of the original translation of the Ekottarika-āgama into Chinese.[n.98] [108] This in turn provides a significant indication regarding the nature of the 分別功德論, T 1507, which has a brief reference to the Mahādeva tale at T XXV 32c8. According to this reference, King Mahādeva and eighty-four thousand generations of kings after him practised the brahmavihāras, moreover, Mahādeva was the only mahāpuruṣa among them (which I take to refer to his status as the Buddha in a former life). The eighty-four thousand generations are not mentioned in EĀ 1, but only in EĀ 50.4 at T II 809a22, making it probable that T 1507 refers to this discourse. In other words, by the time of the coming into being of this reference in T 1507, EĀ 50.4 must have already been part of the Ekottarika-āgama. Regarding T 1507, Mori 1970: 456 explains that, after the completed translation of the Ekottarika-āgama, "the commentary was done later consulting the former's translation already made for their mutual concord." Nattier 2012: 15 points out that T 1507 "originally thought to be a translation of a commentary on an Ekottarikāgama …. appears, at this point, to be a commentary on T125 itself"; cf. also below p. 463 note 71 and, for a detailed study of T 1507, Palumbo 2013.

This suggestion finds corroboration on considering the placement of the Mahādeva tale among the Elevens in the fiftieth chapter of the Ekottarika-āgama. Each of the first three discourses among the Elevens in the fiftieth chapter of the Ekottarika-āgama begins with the Buddha taking up an elevenfold topic and then continue with his detailed explanation of this topic.[n.99] [110] EĀ 50.1 at T II 806b13, EĀ 50.2 at T II 806b26, and EĀ 50.3 at T II 806c10 each begin by mentioning a set of eleven, followed by the enquiry about what these eleven are, 云何(名)為十一, which then leads on to a detailed exposition. The fact that in this chapter only these three discourses match the category of Elevens has already been noted by Yìnshùn 1962/1983: 759, who points out that in the previous chapter only EĀ 49.1, EĀ 49.2, EĀ 49.3, EĀ 49.4, EĀ 49.6, EĀ 49.7, and EĀ 49.10 fit the same category. This conforms to what one would expect under the header of "Elevens", as the point of such a subdivision in the Ekottarika-āgama, found similarly in the Aṅguttara-nikāya, is precisely to collect discourses which in some way involve the number eleven.[n.100] [109] Whereas the extant Ekottarika-āgama and the Aṅguttara-nikāya go up to Elevens, the original conception of the numerical collection appears to have gone only from Ones to Tens. As pointed out by von Hinüber 1996/1997: 40, in the case of the Aṅguttara-nikāya "originally, it seems, AN had only ten Nipātas. This can be deduced from the fact that at the end of the Dasakanipāta not only groups of 10 items occur as the title implies, but also groups of 20 (AN V 304), 30 (AN V 305), and 40 (AN V 306) items, which is typical for the last chapter of a text." In fact the Abhidharmakośavyākhyā, Wogihara 1932: 188,25, speaks of an Ekottarika-āgama that goes up to Tens only. A count from Ones to Tens is also mentioned in the description of the Ekottarika-āgama in the account of the first council or communal recitation in the Mūlasarvāstivāda Vinaya, T 1451 at T XXIV 407c1: 若經說一句事二句事 乃至十句事者, 此即名為增一阿笈摩. The corresponding passage in the Dharmaguptaka Vinaya appears to reflect a development from an original collection of Tens to a collection of Elevens, as it explains that the Ekottarika-āgama goes from Ones to Tens and from Tens to Elevens, T 1428 at T XXII 968b20: 從一事至十事, 從十事至十一事, 為增一. An account of the compilation of the Tripiṭaka, 撰集三藏及雜藏傳, T 2026 at T XLIX 4a19, confirms that the Ekottarika collection goes from Ones to Tens, to which discourses related to eleven were added to form the Elevens. Przyluski 1926: 105 notes 1 to 4 points out that several specifications given in T 2026 at T XLIX 3a27 about the numerical sections of the Ekottarika-āgama correspond to T 125, but differ from the situation in the Aṅguttara-nikāya. This makes its indication of the gradual development of the Ekottarika collection from a collection of Tens to a collection that also covers Elevens directly relevant to the case of T 125. In fact, Dào'ān's (道安) preface to T 125 at T II 549a6 explains that the term ekottarika means increasing by one up to Tens, 數終十, 令加其一, 故曰增一也 (on the corresponding passage in T 2145 at T LV 64b2 cf. Palumbo 2013: 40). It seems to me that the use of the phrase 故曰 makes it clear that this is a word explanation of the term 增一, not a description of the actual condition of T 125, which is taken up at a later point in Dào'ān's preface. Therefore I think it is more probable that the expression 令 加其一 refers to the character of this collection as increasing by one, not as conveying the idea that one more nipāta has been added to the original ten nipātas. The translation of this part of the preface by Legittimo 2014: 73 as "one was added to these. This is why it is called 'increased by one'" is not convincing; the name of the collection would hardly have been derived from the fact that a later addition was made to it. Palumbo 2013: 40 note 70 holds that the notion "that 'numbers end with ten' (shu zhong shi 數終十 in the preface) is a purely Chinese notion." Yet the division of texts into tens is well attested throughout the early Buddhist textual collection and has according to Renou 1957: 2 precedents in Vedic literature. In sum, it seems to me reasonable to assume that this explanation, given at the outset of the preface, reflects awareness of the original conception of the numerical collections. The same type of explanation is also found in T 1507 at T XXV 32a26: 以一為本, 次至十, 一二三隨事增上故曰增一 (although a variant reading changes ten to eleven), which, as mentioned in the previous note, would be referring to the original conception of the Ekottarika-āgama, not to the actual condition of T 125, which was of course known to the author(s) of T 1507, cf., e.g., T XXV 34b6: 誦增一, 正得十一事.

The fourth discourse among the Elevens in this chapter of the Ekottarika-āgama is the Mahādeva tale, which does not contain even a single occurrence of the number eleven. The same is the case for subsequent discourses in this chapter.[n.101] Particularly noteworthy is that EĀ 50.7 at T II 812b16 (just as EĀ 52.5 at T II 825c23 and EĀ 52.6 at T II 826a4) takes up a topic involving fives, which is a clearly out of place in a chapter on Elevens. In view of its detailed exposition of the seven treasures of a wheel-turning king, the Mahādeva tale could have naturally found a placing among the Sevens of the Ekottarika-āgama.[n.102] [111] This is, in fact, the placing of another detailed exposition of the seven treasures of a wheel-turning king in EĀ 39.8 at T II 731c24. Among the Elevens, however, the Mahādeva tale is clearly a misfit.[n.103] [112] Although not each and every discourse in the numerical collections clearly exhibits this pattern (cf. Allon 2001: 14f), the fact that such a misfit can be a sign of a problem in transmission can be illustrated with the example of the Karajakāya-sutta. The Karajakāya-sutta occurs among the Tens of the Aṅguttara-nikāya, AN 10.208 at AN V 299,11, but does not bear a relation to the number ten. Comparative study of this discourse in the light of its parallels makes it highly probable that at some point in its transmission the Karajakāya-sutta lost an exposition on the ten courses of action, which would have been the original reason for its inclusion among the Tens; cf. Anālayo 2009c, Martini 2012, and Dhammadinnā 2014. In the case of T 125, even discourses whose text can be found with very few differences in wording outside of the Ekottarika-āgama as individual translations attributed to translators active before the time of the translation of the Ekottarika-āgama have a placing in the collection that conforms to the basic numerical principle: EĀ 30.3 at T II 660a1 (corresponding to T 128b at T II 837c12) is found among the Fours and at some point in its long exposition indeed broaches a topic related to four, followed by a question as to what these four are, T II 665a16: 云何為四, and in reply to this question comes a short listing of the four. EĀ 48.3 at T II 787c2 (corresponding to T 453 at T XIV 421a6) is placed among the Tens of the Ekottarika-āgama and has indeed a reference to ten perceptions, T II 789b5: 此十想者. On these two discourses cf. also Lévi and Chavannes 1916: 191 and 263 as well as the study of T 453 by Legittimo 2010b.

Another peculiarity in translation terminology further confirms the fact of a later addition. The Mahādeva discourse found among the Elevens begins with the phrase "at one time the Blessed One", rendered 一時婆伽婆, followed by an indication that "at that time the Blessed One", 爾時世尊.[n.104] [113] EĀ 50.4 at T II 806c21. On the standard opening phrase of a discourse cf. Anālayo 2014b: 41–45. The rendering of bhagavant as 婆 伽婆 does not recur anywhere else in the Ekottarika-āgama,[n.105] [114] The expression 婆伽婆 occurs also in T 1507 at T XXV 35b16, explicitly introduced as an equivalent to 世尊; cf. also the comment above note 98 regarding a reference to the Mahādeva tale in T 1507. nor is it found in the Dīrgha-āgama (T 1) or in the Udānavarga (T 212), attributed to the same translator Zhú Fóniàn.

Instead, the peculiar opening of the discourse, with 一時婆伽 婆 followed by subsequent occurrences of bhagavant rendered instead as 世尊, recurs in a group of Madhyama-āgama discourses now found individually in the Taishō edition, but which appear to have been translated by the same translator.[n.106] [115] Before turning to the relevant instances, I need to record my indebtedness to Jan Nattier for having drawn my attention to this pattern and for having already suggested, in an email dated 13 April 2010, that the occurrence of the expression 一時婆伽婆 could be a sign that the present discourse was absorbed into T 125 from another translation. Discourses that open with the 一時婆伽 婆 phrase, followed by the alternative rendering of bhagavant as 世尊 throughout the rest of the discourse, are T 47 at T I 837a9, T 49 at T I 839a8, T 50 at T I 842b6, T 51 at T I 843c16, T 53 at T I 846c8, T 56 at T I 851a26, T 58 at T I 853c23, T 60 at T I 856a7, T 64 at T I 862b8, T 65 at T I 863b13 (adopting the variant reading 伽 instead of 加), T 66 at T I 864b5, T 70 at T I 875a14, T 73 at T I 879a11, T 75 at T I 882a23, T 77 at T I 886a28, T 79 at T I 888b18, T 82 at T I 901b26, T 83 at T I 902b7, T 90 at T I 913c7, T 91 at T I 915a7, and T 94 at T I 917b16; whereas T 55 at T I 849b26 has the reading 一時婆迦 婆, which is then also followed by 世尊 in the rest of the discourse. This formula appears to be a common characteristic of the twenty-four discourses that according to the research by Hung et al. 2009/2010, based on indications by Mizuno, appear to be from the same translator, except for T 92 and T 93 (although the formulation 一時婆迦婆, found in T 55, recurs in T 89 at T I 913a20, which presumably does not belong to this group). T 92 and T 93 begin by indicating that the Buddha had passed away and thus naturally do not have a counterpart to the above phrase. The 一時婆伽婆 phrase also occurs in several Saṃyukta-āgama and Ekottarika-āgama discourses preserved as individual translations. However, in the case of one such instance of 一時 婆伽婆 in T 133 at T II 855c6, the arahant epithet of the Buddha is rendered in the alternative way, 至真, 等正覺; cf. T II 855c11. Thus T 133 does not concord with the way the arahant epithet of the Buddha is rendered in the individual Madhyama-āgama discourses listed in the note below and in EĀ 50.4. Notably, six discourses out of this group list the epithets of the Buddha, and each of them precedes the qualification 等正覺 with the term 無 所著,[n.107] [116] T 60 at T I 858a2, T 66 at T I 864c7, T 73 at T I 879c4, T 75 at T I 882b18, T 92 at T I 916b10, and T 93 at T I 917a21. the expression also used in the Mahādeva tale among the Elevens. In spite of these similarities, however, as the research by Hung (2013) shows, in other respects the translation terminology in the Mahādeva tale now found among the Elevens of the Ekottarika-āgama differs from this group of Madhyama-āgama discourses, so that for the time being the provenance of the Mahādeva tale now found among the Elevens remains unclear.[n.108] [117] The research by Hung 2013 also shows that the twenty-four Madhyama-āgama discourses mentioned above in note 106 differ substantially from the translation terminology employed in the rest of T 125, making it improbable that these stem from the same translator; cf. also above p. 105f note 35.

What is indubitably clear, however, is that the phrase 一時婆 伽婆 found in the introduction to the second Mahādeva tale further confirms that another translator must have been at work. The introduction and conclusions of a discourse are stereotyped to such a degree that it can safely be expected that the same way of rendering will be used consistently by someone engaged in the translation of a single work.

Besides the peculiar phrasing found at the beginning of the second Mahādeva tale, its conclusion also shows the signs of another translator being at work. The second Mahādeva tale concludes with the expression 佛說是已, followed by reporting Ānanda's delight.[n.109] [118] EĀ 50.4 at T II 810b18. I am indebted to Jen-Jou Hung for this finding. Since the first Mahādeva tale, EĀ 1, does not have a formal conclusion, I did not notice this feature of EĀ 50.4 when comparing the two versions. Thanks to the discussion in Hung 2013: 127f, this additional feature of EĀ 50.4 has come to light, corroborating that a different translator must have been at work. This phrase does not recur elsewhere in the Ekottarika-āgama, which instead regularly uses the phrase 聞 佛所說 to introduce the delight of those who have heard what the Buddha had taught. With this additional evidence, I think it is now time to come to a definite conclusion. The second Mahādeva tale must stem from the hand of another translator.

By way of winding up my comparative study of the two versions of the Mahādeva tale, it seems clear that these two discourses are based on different original narrations and have been rendered into Chinese by different translators. This in turn provides strong evidence for concluding that an already existing translation of the Mahādeva tale was not part of the original text and was incorporated among the Elevens of the Ekottarika-āgama.[n.110] [119] Palumbo 2013: 113f notes that Mizuno already identified EĀ 50.4 as an interpolation (my ignorance of Japanese has prevented me from consulting Mizuno myself). This contrasts with the assessment by Legittimo 2010a: 153f that the Ekottarika-āgama has been passed on faithfully without any change, "sans interpolations ou censures, et sans être soumis aux manipulations des idéologies changeantes des siècles suivants, le recueil a conservé fidèlement les données anciennes telles qu'elles furent transmises jusqu'au jour de sa traduction"; cf. also Legittimo 2010b: 256: "the Chinese Ekottarika-āgama collection remained unchanged since its translation at the end of the fourth century."

The motivation for such integration of the Mahādeva tale's depiction of an ideal Buddhist king from elsewhere into the Ekottarika-āgama could be related to political considerations, given that the history of Buddhism in China is one of constant struggle for the emperor's recognition. The depiction of the ideal Buddhist ruler in the Mahādeva tale would have come in handy for propaganda purposes in such a setting, providing an example for how, from a Buddhist perspective, an emperor should conduct himself.[n.111] [120] In a study of the writings of Xuánzàng (玄奘) from the perspective of their intended impact on the emperor, Deeg 2009: 51 speaks of "the speculum-motive: the intention to hold a mirror in front of his ruler to show him how an ideal Buddhist kingdom looks like." Pāsādika 2010: 95 comments on a jātaka tale found in EĀ 52.9 at T II 829b11 (translated in Pāsādika 2007) that this narration appears to present a "subtle criticism, warning and simultaneously encouragement towards renewal discreetly addressed to those in power in early medieval China".

In a study of models of Buddhist kingship in early medieval China, Palumbo (2012: 316) highlights in particular the penchant of the translator Zhú Fóniàn in this respect, noting that "Zhú Fóniàn's vision of a world ruled by Buddhist holy men, whom a devout monarch would reverence and the masses obey, was absolutely unprecedented in China."

In my study of the Mahādeva tale found among the Elevens in the last two chapters,[n.112] Cf. above p. 113–164. I suggested that the original point of the depiction of the wheel-turning king in this tale was precisely not the providing of a model to be emulated. Instead, in this discourse the wheel-turning king seems to function as a humorous persiflage of ancient Indian kingship. This persiflage in turn forms part of a soteriological project that replaces the acme of worldly kingship with the superior principle of renunciation, and this in turn leads up to the Buddha's disclosure of the supreme path to liberation. Nevertheless, tradition soon took such tales as providing a model of the dharmic king, a model exerting considerable influence on notions of kingship in various Buddhist countries and traditions.

An employment of the Mahādeva tale in the ancient Chinese setting would be well in line with this tendency.[n.113] [122] DuBois 2004: 539 speaks of "the participation of Chinese political actors in worship of Maitreya and reinterpretation of the cakravartin". The wish to provide a depiction of an ideal Buddhist king could also explain a peculiar feature of the Mahādeva tale found among the Elevens. Unlike its parallel versions, this Ekottarika-āgama tale has narrative pieces that appear to stem from a different discourse that also describes a wheel-turning king, found separately in the Dharmaguptaka, Sarvāstivāda, and Theravāda traditions, and known in Pāli as the Cakkavatti(sīhanāda)-sutta.[n.114] [123] Be, Ce, and Se have the title Cakkavatti-sutta, whereas in Ee the title is Cakkavatti-sīhanāda-sutta.

One of these pieces found in the Ekottarika-āgama version of the Mahādeva tale as well as in the Cakkavatti(sīhanāda)-sutta and its parallels describes in detail the negative repercussions on the whole country that ensued when a king, on ascending the throne, did not follow the tradition of dharmic kingship established by his predecessors. People became short-lived, diseased, and poor, poverty led to theft, which in turn led to a general increase in crime and a deterioration of living conditions.[n.115] [124] This narration is found in EĀ 50.4 at T II 810a14, which has counterparts in DN 26 at DN III 65,15, DĀ 6 at T I 40b23, and MĀ 70 at T I 522a28 (a similar parallelism can be found between EĀ 50.4 at T II 808c13 and DN 26 at DN III 60,9, DĀ 6 at T I 39b24, and MĀ 70 at T I 520c14); cf. also above p. 159f.

Such pieces would have made the Mahādeva tale a presentation of the wheel-turning king worth including alongside other discourses that take up this motif in the Ekottarika-āgama, so as to provide a model of Buddhist kingship to the ancient Chinese audience.

The Ekottarika-āgama Translation

The original translation of the Ekottarika-āgama appears to have been based on a text recited orally in 384 by Dharmanandin/Dharmananda,[n.116] [125] Dào'ān indicates that Dharmanandin/Dharmananda was a reciter of two Āgamas, T II 549a11: 誦二阿含, presumably the Madhyama-āgama and the Ekottarika-āgama then translated by Zhú Fóniàn. which was then translated by Zhú Fóniàn.[n.117] [126] The catalogue 出三藏記集, T 2145 at T LV 6a13, mentions a still earlier Ekottarika translation, which appears to be the collection now found as T 150A. Whereas early Tripiṭaka catalogues such as T 2145 (出三藏記集) just report this translation,[n.118] [127] T 2145 at T LV 71b29. about two hundred years after the original translation the catalogue T 2034 (歷代三寶紀), known for not being particularly reliable,[n.119] [128] Cf., e.g., Nattier 2008: 14f. reports that Gautama Saṅghadeva, the translator of the extant Madhyama-āgama (T 26), subsequently did a retranslation of the Ekottarika-āgama.[n.120] [129] T 2034 at T XLIX 70c12. Such a retranslation is not mentioned in Saṅghadeva's biography in T 2059 (高僧傳), which only records his Madhyama-āgama translation.[n.121] [130] T 2059 at T L 329a23.

Closer study of the translation terminology used in the extant Madhyama-āgama (T 26) and in the Ekottarika-āgama (T 125) makes it clear that this is so different as to make it safe to conclude that these two collections have not been translated by the same translator.[n.122] Hung and Anālayo 2016, Radich 2016, and Radich and Anālayo 2016. An alternative explanation would be that Saṅghadeva only revised the already translated Ekottarika-āgama (T 125). Yet, even that is not entirely straightforward, since by that time Dharmanandin had left China. Without access to the orally recited original, it is not easy to see how Saṅghadeva could have carried out a revision of the text. Lín (2009) suggests that it may have been rather the original translator Zhú Fóniàn himself who did a revision. For the original translator it would be natural to do a revision of his own work which, as mentioned earlier, had originally been carried out under difficult conditions.

According to the preface to the Ekottarika-āgama by Dào'ān (道安), the translation undertaken in 384 had four hundred and seventy-two discourses in forty-one fascicles.[n.123] [134] T II 549a15+26: 四十一卷 … 四百七十二經. Dào'ān's indications given at this point in his introduction were based on his own supervising of the copying of the original translation and were apparently made just before his passing away.[n.124] [135] At T II 549a17 Dào'ān mentions his own role in checking the translation. Matsumura 1989: 361 note 10 comments that "the fact that Dào-ān died in 384 proves that this introduction was written in the very year when [the] Ekottarika was translated. Therefore the information provided in this introduction is highly reliable." This makes his description a first-hand eyewitness report and therefore different from information provided in catalogues that were compiled centuries later.[n.125] [136] Even the 出三藏記集, T 2145 at T LV 10b21, in spite of being presumably based on Dào'ān's no longer extant catalogue, gives a different fascicle count for the Ekottarika-āgama which is less than what Dào'ān mentions in his introduction, speaking of only thirty-three fascicles, 增一阿鋡經三十三卷.

The extant Ekottarika-āgama has indeed four hundred and seventy-two discourses (not counting the introduction), [n.126] [137] T 125 has fifty-two chapters, of which the first corresponds to the introduction. Thirty-three of the remaining chapters have the standard count of ten discourses. The others are as follows: three discourses (chapters 7 and 30), four discourses (chapter 6), five discourses (chapters 5, 36, and 41), six discourses (chapter 48), seven discourses (chapters 13, 28, and 45), nine discourses (chapter 52), eleven discourses (chapters 17, 19, 31, and 44), twelve discourses (chapters 32 and 38), and thirteen discourses (chapter 20). The count of discourses, without taking into account the introduction, is thus indeed four hundred and seventy two. According to Legittimo 2014: 80 note 44, however, "in the present Taishō edition, the Ekottarikāgama actually contains 476 sūtras. One sūtra was mistakenly attached to the previous one. Thus, in the present state, the Ekottarikāgama seems to contain 475 sūtras. Three to four sūtras therefore remain unaccounted for, which is an increase of less than 1 per cent compared with the numbers given in the preface"; cf. also Legittimo 2014: 76: "the extant collection seems to contain three or four more sūtras than at the time of its translation." This is not correct and is the result of a counting error. but these come in fifty-one fascicles. If Dào'ān's indications are correct, which seems fairly probable, they point to a consistent number of discourses in combination with a substantial difference in actual volume, an increase of over 24 per cent. Although some degree of variation in fascicles can result from a bare rewriting of the same text, such difference could also be a sign of a change of the original translation. The second of these two possibilities finds support in a discrepancy, further discussed by Su (2013), in relation to the summary verses, uddānas. According to Dào'ān the first twenty-six fascicles of the Ekottarika-āgama carried uddānas.[n.127] [139] T II 549a16: 上部二十六卷全無遺忘, which is followed by an indication that the uddānas for the second half were lost, 下部十五卷失其錄偈也. Dào'ān then continues by reporting that with his helpers he worked to restore what had been lost, which might imply that they restored some of the lost uddānas, of whose important function Dào'ān was apparently well aware; cf. T II 549a16. In fact, as discussed by Su 2013, the second half of the Ekottarika-āgama does have some uddānas, although considerably less than the first part. This does not fully correspond to the present situation, where what roughly corresponds to the first half of the Ekottarika-āgama does have uddānas for the most part, but some are missing. Moreover, as pointed out by Su, references in other works to discourses presumably once found in the Ekottarika-āgama translated into Chinese can no longer be located in what we have now as entry no. 125 in the Taishō edition.

Taken together, these discrepancies and what emerges from a comparison of the two versions of the Mahādeva tale in the Ekottarika-āgama suggest that some sort of revision of the translated collection took place. Regarding a possible revision of the Ekottarika-āgama, what previous scholarship has found out about other works by the translator Zhú Fóniàn is significant:

• Legittimo (2007) points out similarities between T 384 (菩薩 從兜術天降神母胎說廣普經) attributed to Zhú Fóniàn and the Vimalakīrtinirdeśa, of which a version had already been translated previously (T 474).[n.128] [141] Legittimo 2007: 1082 notes that one such similarity, found between T 384 and the Vimalakīrtinirdeśa-sūtra, also manifests in the Ekottarika-āgama, EĀ 45.4 at T II 772b19. The passage in question provides a listing of five nutriments alternative to the standard listing of four nutriments found elsewhere in the early discourses. This alternative fivefold presentation reckons the eight liberations as a type of nutriment, 八解脫食.
• Legittimo (2008) notes parallelism between the same T 384 and the Saddharmapuṇḍarīka-sūtra, of which, too, a Chinese translation was already in existence (T 263).
• Pu (2008: 43f) notes parallelism between T 309 (最勝問菩 薩十住除垢斷結經) attributed to Zhú Fóniàn and the previously translated T 630 (成具光明定意經).
• Nattier (2010) then takes up T 309 for a closer examination and points out several cases of parallelism with T 221 (放光 般若經), with T 630 (成具光明定意經), and with T 403 (阿 差末菩薩經), all translated prior to T 309.
• Palumbo (2013: 127) shows that a Sarvāstivāda prātimokṣa preserved in Chinese translation from a Dūnhuáng (敦煌) manuscript has served as the basis for a part of Ekottarika-āgama discourse 48.2.

Nattier (2010: 251) concludes that the case of T 309 shows that, instead of translating an Indic original, Zhú Fóniàn was "drawing substantial material from existing Chinese scriptures while at the same time reframing and rearranging it within a genuinely new composition".

Based on an examination of the biographical section on Zhú Fóniàn in the catalogue T 2145 (出三藏記集), Nattier (2010: 253) then distinguishes between two periods in Zhú Fóniàn's career. The first belongs to the final part of the fourth century, when he was working in close association with foreign monks. The second period lies mainly in the fifth century, when he worked predominantly on his own. According to her conclusions, works produced by Zhú Fóniàn on his own during this second period stand a chance of being the result of a more creative approach on his side, instead of constituting renderings of Indic originals.[n.129] [142] Future research on works attributed to Zhú Fóniàn during this second period of his activities might also benefit from the observations in Silk 2006: 49. If Zhú Fóniàn should indeed have done a revision of the Ekottarika-āgama, then this revision would fall into the second period of his activities identified by Nattier.[n.130] As discussed above, p. 102ff, I do not find the alternative scenario developed by Palumbo 2013 convincing.

Regarding the fact that the number of discourses now found in the Ekottarika-āgama matches the indications given by Dào'ān (道安), from the viewpoint of a reviser of the translation it would have made sense to keep an eye on maintaining a constant count of discourses during a process of textual revision that incorporates additional textual material, in order to avoid casting doubt on the authenticity of the material contained in the collection. Putting the repute of the collection unnecessarily in doubt by not ensuring some degree of correspondence with the known overall count of discourses would have resulted in the exact opposite of what a revision would have tried to accomplish, namely making sure the collection is considered sufficiently important and genuine even by those with stringent attitudes to canonicity, so that it is perceived as worth the effort of passing it on to future generations.

Maintaining the same discourse count could have been accomplished if the Mahādeva tale had replaced a short discourse found among the Elevens, similar to those three that stand at the beginning of the fiftieth chapter in which the Mahādeva tale is found as the fourth discourse.[n.131] [143] Regarding the possibility that some discourses were replaced by new material, it is perhaps significant that the survey in Akanuma 1929/1990: 120– 156 shows that the Ekottarika-āgama has a considerably lower percentage of parallels in the corresponding Pāli Nikāya than the other Āgamas preserved in Chinese translation; cf. also Lü 1963: 244, who observes that "of its 472 sūtras, only 135, that is less than one third of its contents, agree with the Aṅguttara Nikāya." This stands in contrast to Ekottarika-āgama Sanskrit fragments, where the survey in Tripāṭhī 1995: 31 shows considerable parallelism with the Aṅguttara-nikāya. The same is also the case for the partial Ekottarika-āgama, T 150A, where Harrison 1997: 276 in his detailed study concludes that of the forty-four discourses found in this collection "36 have close parallels in the Pāli Aṅguttara-nikāya, while five have parallels in the complete Chinese translation of the Ekottarikāgama, the Zengyi ahan jing (T 125)." The fact that T 125 differs so much from other Āgamas and Nikāyas could be the result of a process of revision during which some of its original discourses were lost and need not be the result of school affiliation only. As a result of changes of this type, the Ekottarika-āgama would indeed have become more voluminous than it had been before, without affecting the count of discourses.

Changes of this type would also naturally lead to a loss of some uddānas, namely whenever a discourse that is immediately followed by an uddāna is being replaced by another discourse without an uddāna. This would also explain the discrepancy between Ekottarika-āgama quotes in T 2121 (經律異相) and the actual discourses now found in the extant Chinese translation of the Ekottarika-āgama, discussed by Su (2013), in that T 2121 might be quoting from the original Ekottarika-āgama translation, before its revision.

Alternatively, however, it could also be that at times two discourses were merged into one. This would then make space for the addition of the Mahādeva tale or other new material while keeping the discourse count constant. As discussed in previous chapters, there is in fact considerable evidence for the occurrence of such a merger elsewhere in the Ekottarika-āgama.[n.132] Cf. above p. 51–111.

Besides providing information on the size and discourse count of the Ekottarika-āgama, by the very fact of its existence the preface by Dào'ān also provides a clue as to why someone might wish to add material to an existing collection that has already been accepted as canonical. As pointed out by Lancaster (1999: 537f), the emphasis at that time in China on translating canonical material of Indian origins had the following result:

[It] helped to create the situation where contemporary Buddhist works of China were denied an avenue for distribution. Unless a Buddhist document was a translation from the Indic … [it] would not be included in the collection that was copied and spread from place to place. With this restriction on inclusion, works written in China were neglected. Even the great sage Daoan 道安 found it necessary to append his writings as prefaces to the jing [i.e., the canonical scripture, such as the Ekottarika-āgama]. In that way, his work would be copied when the jing was reproduced; otherwise, there was no way to publish. In this situation, it is not surprising to find a large number of pseudographs, Chinese works claiming to be translations from Sanskrit.[n.133] [146] On the related problem of explanations given during translation that then become part of the translated text cf. Funayama 2006.

The danger of already accomplished work getting lost would have become particularly clear by the beginning of the fifth century, when besides Zhú Fóniàn's translation of the Madhyama-āgama, another Madhyama-āgama translation carried out in 398 by Saṅghadeva had come into circulation, as a result of which Zhú Fóniàn's Madhyama-āgama translation seems to have lost favour and eventually disappeared. [n.134] [147] Judging from T 2147 at T LV 178b20 and T 2148 at T LV 216a15, the Madhyama-āgama translation by Zhú Fóniàn (referred to as a translation by Dharmanandin/Dharmananda) was lost by the time of the early seventh century. This in turn also meant that Dào'ān's preface to the Madhyama-āgama, to which he alludes at T II 549a5, was lost. Thus even appending one's writings to a canonical text did not always ensure that these were transmitted to future generations. The biographical account of Saṅghadeva in T 2059 (高僧傳) makes a point of indicating that the translation of the two Āgamas transmitted by Dharmanandin/Dharmananda (and translated by Zhú Fóniàn) had not been executed well.[n.135] [148] T 2059 at T L 328c28.

If by the beginning of the fifth century the bad repute of Zhú Fóniàn's two Āgama translations and in particular the loss of popularity of his Madhyama-āgama translation were in the air, it would have been quite understandable if he should have decided to undertake a revision of his Ekottarika-āgama translation in order to ensure that at least this work would be considered worth passing on by future generations, instead of being also replaced by some other translation.[n.136] On the apparently felt need to ensure that the Ekottarika-āgama translation is considered worth transmission cf. also below p. 458f.

Regarding a somewhat creative attitude that might have informed such a revision, the introduction to the Ekottarika-āgama furnishes another significant indication. The text provides an instruction that, in case the name of a particular location where a discourse took place has been forgotten, one should simply supply the name of any of the well-known cities where the Buddha usually dwelled.[n.137] [149] EĀ 1 at T II 550b13. A similar instruction is also found in T 1507 (分別功德論).[n.138] [150] T 1507 at T XXV 33b19; on this work cf. above note 98 and Palumbo 2013.

The Ekottarika-āgama and T 1507 might be the first instances of such an instruction coming to the attention of the Chinese readership, as similar indications made in the Mahāsāṅghika and Mūlasarvāstivāda Vinayas were only translated at a later time into Chinese.[n.139] [151] T 1425 at T XXII 497a6 and T 1451 at T XXIV 328c15 and 575b29; cf. also Schopen 1997/2004: 395–407.

Independent of whether this indication was part of the introduction to the Ekottarika-āgama from the outset or was added to it based on an extract from some other source, the original point of this instruction needs to be understood in the light of what appears to be a general lack of interest in ancient India for historical details.[n.140] [152] Coward 1986: 305 explains that "the early Buddhists shared … the Indian indifference to historical details. Historical events surrounding a text are judged to be unimportant." According to Pollock 1989: 610, behind the ancient Indian attitude towards historical details stands "a model of 'truth' that accorded history no epistemological value or social significance". Gokhale 1994/2001a: 1 adds that according to tradition "the Buddha does not indulge in 'low' talk (tiracchānakathā …) … [such as] tales of kings and their high ministers, armies and wars … now precisely these are the stuff out of which the chronicles of history are made." In other words, in an ancient Indian setting the suggestion to supplement any location freely would have been understood to be simply an expression of the relative unimportance of locations and other such narrative details.

When considered from an ancient Chinese viewpoint this instruction acquires quite a different significance, given the concern of the Chinese with historical record keeping. In an ancient Chinese setting the implication would be that, according to the text itself, even such crucially important information as the location where something took place can be freely supplemented. In the case of a translator who already had a tendency towards a more creative rendition, it seems not too farfetched to assume that he would feel some addenda to improve on the same text would be quite in line with "the Buddha's intention".

In the case of the Ekottarika-āgama, such an interpretation of this instruction would find further encouragement in the nature of the collection itself, which is an anthology of various texts that often bear little relation to each other. This is the case with the Ekottarika-āgama and the Aṅguttara-nikāya to a greater extent than with other Āgamas and Nikāyas, where often a stronger sense of thematic cohesion makes itself felt.[n.141] [153] Legittimo 2012: 350 observes that "due to its numerical scheme … the Ekottarika-āgama shows a particular propensity … to incorporate new material".

In this way the historical setting and the nature of the collection might explain why, as clearly shown by a close study of the two versions of the Mahādeva tale, a substantial text was added in China as a discourse to the Ekottarika-āgama collection.

This in turn provides some background for assessing the topic of the school affiliation of the collection, a topic which I explored above in relation to the introductory section of the Ekottarika-āgama in particular.[n.142] Cf. above p. 173ff. The assessment by Bareau, discussed above, concords with that of several other scholars, who have pointed to evidence in support of the hypothesis that the Ekottarika-āgama collection extant in Chinese stems from a Mahāsāṅghika line of transmission.[n.143] [31] Cf. Mayeda 1985: 102f and for recent contributions Pāsādika 2010, Kuan 2012, 2013a, 2013b, and 2013c.

Hiraoka (2013), however, points out narrative affinities between texts found in the Ekottarika-āgama and in the Sarvāstivāda tradition(s). Yet, it needs to be kept in mind that such type of affinities do not necessarily have a decisive bearing on the question of the school affiliation of a discourse collection, since narrative material easily travels across different traditions, texts, and reciter lineages.[n.144] An illustrative example would be the observation by Radich 2011: 27 in relation to the Ajātaśatru narrative that the two narratives that "match most closely to that of the MSV [Mūlasarvāstivāda Vinaya] (even more than the Sarvāstivāda works just discussed) are from traditions quite distant from it: Jaina texts, and Sv [Sumaṅgalavilāsinī]". In the present case in fact, as already stated by Harrison (2002: 19), the Ekottarika-āgama in fact "can hardly be Sarvāstivādin".

Given the finding that an entire discourse became part of the collection only in China, it is quite possible that other pieces of various origins also made their way into the collection only at that time. This makes it difficult, if not impossible, to use individual instances in the way they are now found in the Chinese text as definite evidence for or against the school identity of the reciters who transmitted this collection before it reached China.

Nevertheless, from the viewpoint of comparative studies of the early discourses a common pattern can be observed that also sheds some light on the Ekottarika-āgama.[n.145] My assessment in what follows is based on Anālayo 2011c and a range of comparative studies published in article form. As a general tendency, discourses from the four Pāli Nikāyas are often close to discourses in the Dīrgha-āgama (T 1) as well as to discourse quotations in the Dharmaguptaka Vinaya (T 1428) and the *Śāriputrābhidharma (T 1548), whenever these are extant. Another group, distinct from the former, whose members have much in common, comprises the two Saṃyukta-āgama collections (T 99 and T 100) as well as discourse quotations in the Mūlasarvāstivāda Vinaya and the Abhidharmakośopāyikaṭīkā (D 4094 and Q 5595). The Madhyama-āgama (T 26) and discourse quotations in the Sarvāstivāda Vinaya (T 1435), whenever extant, are also members of this group, although at times exhibiting their own specific idiosyncrasies.

In the case of a number of discourses in the Ekottarika-āgama (T 125) that give the impression of being genuinely early material, these at times involve substantially different presentations compared to all of the discourse material mentioned so far. This in turn makes it fairly probable that the reciters responsible for transmitting the Indic original of the Ekottarika-āgama belonged to a tradition which at a comparatively early time in the history of Buddhism followed its own trajectory. This would fit the case of the Mahāsāṅghika tradition(s) and would also explain the Mahāsāṅghika elements identified by various scholars in the extant collection.

Yet, due to the problems discussed in the previous pages, in particular the occurrence of discourse merger as well as the finding that material has been added in China, it seems to me impossible to force the Ekottarika-āgama collection into the mould of one school affiliation or another. Instead, I contend that the Ekottarika-āgama collection, as now extant in Chinese translation, is as a whole best considered as not being amenable to an identification of its school affiliation.

In the case of Āgama material, school affiliation is after all only something that during transmission gradually affects texts whose beginnings predate the coming into existence of the different Buddhist schools. In view of this it is quite natural if a such school affiliation, acquired during transmission, could also be lost during a later stage of transmission.[n.146] An example for a comparable process, albeit affecting only parts of a single discourse in some editions, is the intriguing finding by Karashima 2014 of what he refers to as a "Sarvāstivāda encroachment" on DĀ 10 that must have happened during the written transmission of the Dharmaguptaka Dīrgha-āgama in China. This is precisely what seems to have happened when the Ekottarika-āgama reached China.

In this way the Ekottarika-āgama collection extant in Chinese offers an illustrative example of the multitude of influences that can shape a discourse collection. At the same time it serves as a reminder that the attribution of school affiliation is merely one of the categories that can be employed to handle a particular text, and in the present case clearly not the most useful one.[n.147] Boucher 2008a: 190 notes that "the problem with always seeing nikāya as at the heart of … differences is that it occludes other possible explanations … we might do well to consider explanations … beyond our modern proclivity for school affiliation." Hartmann 2013: 48 concludes a study of the Maitreyasamitināṭaka and the Maitrisimit by noting that, if his observations are correct, then the question of school affiliation pursues something of little, if any, relevance to authors and users of the text in question, "dann geht die Frage nach der Schulzugehörigkeit bei vielen Werken … einfach ins Leere, dann zielt sie nämlich auf Kategorisierungen, die für Verfasser und Benutzer offenbar keine Wirklichkeit oder zumindest keine praktische Bedeutung besessen haben."