The Pavāraā Sutta and “liberation in both ways” as against “liberation by wisdom”

The Pavāraṇā Sutta (Pravāraṇā Sūtra in Sanskrit) in the Pali Buddhist canon is a text which records what happened on a day when the Buddha and his five hundred disciples held a ceremony called pavāraṇā ( pravāraṇā, Skt), meaning “invitation”. The ceremony takes place at the end of the annual rains retreat (vassāvāsa), when each monk in order of seniority invites ( pavāreti) the other monks in his fraternity to point out any misconduct on his part.1 The story in this text can be outlined as follows: in a certain Pravāraṇ ā ceremony, the Buddha asks the five hundred monks in the assembly if they have found any misconduct on his part. In response to this invitation, Sāriputta praises the Buddha on behalf of the five hundred monks. Then, in reply to Sāriputta’s invitation, the Buddha praises Sāriputta and the five hundred monks. Van. gīsa thereupon extols the Buddha and the five hundred monks.
The only complete version of the Pavāraṇā Sutta in an Indic language is the seventh sutta (sūtra) in the eighth saṃyutta of the Saṃyutta Nikāya (SN 8.7) in Pali. It has five parallels extant in Chinese translation from various Indic originals: 1. MĀ 121: Sūtra 121 of the Madhyama Āgama (Zhong ahanjing T 26, I 610a–c); 2. SĀ 1212: Sūtra 1212 of the Saṃyukta Āgama (Za ahanjing T 99, II 330a– c); 3. SĀ2 228: Sūtra 228 of the Other Translation of the Saṃyukta Āgama (Bieyi Za ahanjing T 100, II 457a–c); 4. T 63: Jiexia jing 解夏經 (T I 861b–862b); 5. EĀ 32.5: Sūtra 5 in chapter 32 of the Ekottarika Āgama (Zengyi ahanjing T 125, II 676b–677b). Shou xinsui jing 受新歲經 (T 61) is almost identical to EĀ 32.5 with only a few variant readings. Although this text is attributed to Dharmaraksa ( ̣ 竺法護 Zhu Fahu) in the later catalogues, its vocabulary is characteristic of Zhu Fonian (竺佛念) rather than Dharmaraksa. ̣ 2 Zhu Fonian was the translator of EĀ, despite the incorrect attributions to Gautama Saṃghadeva (Qutan Sengqietipo 瞿曇僧伽提婆).3 Moreover, this sūtra is referred to in a summary verse (uddāna) as sui 歲 in EĀ (T 125, II 681c7). Therefore, it is beyond doubt that the content of the Shou xinsui jing (T 61), identical to that of EĀ 32.5, was originally a part of the Ekottarika Āgama. Thus, there are six complete versions of the same text available to us. The Xinsui jing 新歲經 (T 62) is a Mahāyānist variant4 of the Pavāraṇā Sutta and its content deviates extensively from the above six versions of the text. Consequently, it falls outside the scope of this study. There are three Sanskrit fragments of the Pravāraṇā Sūtra found in Hoernle (1916: 38–9), Sander and Waldschmidt (1985: 187, Kat.-Nr. 1193)5 and Bechert and Wille (1989: 207, Kat.-Nr. 1598). Zieme (1988) discusses a bilingual fragment of this text written in Old Turkish and Chinese. These materials are irrelevant to the issues dealt with in this paper. Of the six complete versions of the Pavāraṇā Sutta, EĀ 32.5 differs significantly from the other five in the following ways. First, it begins with an episode involving Ānanda that is not found in the other five versions. Second, the other five versions claim that all the five hundred monks in the ceremony (or all but one of them) are arhats (arahant in Pali). Third, all of these five versions mention categories of arhats based on different kinds of liberation, whereas no such mention is present in EĀ 32.5. In this paper I will show that these divergences are correlated with one another and will find out which version could be closest to an ancestral text from which these versions are derived. In this connection I will seek to elucidate diverse notions of liberation as developed in the textual history, and will discuss the possibility that the sectarian stance on the issue of liberation has a bearing on the inclusion of various types of arhats in the five versions of our sūtra and their exclusion from one version, EĀ 32.5, whose sectarian affiliation will be a focus of this article. As to the sectarian affiliations of the six complete versions, SN 8.7 belongs to the Theravāda school still flourishing today. Prevailing opinion holds that the Madhyama Āgama preserved in Chinese translation stems from the Sarvāstivāda tradition.6 The Saṃyukta Āgama in Chinese is also widely ascribed to the Sarvāstivāda7 or perhaps more precisely the Mūlasarvāstivāda school.8 The Other Translation of the Saṃyukta Āgama (T 100) was, according to recent research, transmitted by the Mūlasarvāstivāda.9 While not all scholars accept that the identification of these Āgamas with certain schools is fully established,10 it will be taken as a working hypothesis for the present research. Jiexia jing (T 63) is close to the SĀ, SĀ2 and MĀ versions. Most parts of these four versions resemble one another, with only minor variations, so I will use the blanket term “(Mūla-)Sarvāstivāda” to refer to their school affiliation. The school affiliation of the Ekottarika Āgama is controversial. This collection is ascribed to the Mahāsāṃghikas by Bareau (1955: 55 and 57), Ui (1965: 137–8), Akanuma (1981: 37–9), Bronkhorst (1985: 312–4) and Pāsādika (2010: 88–90), but to the Dharmaguptakas by Matsumoto (1914: 349) and Warder (2000: 6). Hiraoka (2007, 2008) sees some passages of the EĀ as showing an affinity with certain schools, particularly the (Mūla-)Sarvāstivāda. It is clear that the majority of scholars are in favour of attributing EĀ to the Mahāsāṃghikas. Furthermore, as far as the materials I have so far dealt with are concerned (findings not yet published), it can be tentatively concluded that a large portion of the EĀ, if not the majority of it, is of Mahāsāṃghika provenance. It is generally held, in various traditions, that the original Buddhist Order first split into two sects, the Mahāsāṃghikas and the Sthaviras.11 Both the present Theravāda and the extinct (Mūla-)Sarvāstivāda schools are of Sthavira origin.12 Five versions of the Pavāraṇā Sutta can be identified as belonging to schools of the Sthavira lineage: one Theravāda and four most probably (Mūla-) Sarvāstivāda, while the sectarian affiliation of EĀ 32.5 is yet to be decided. Below is my attempt to clarify this issue.

In the Theravāda version and the four, presumably,13 (Mūla-)Sarvāstivāda versions of our sūtra, all (or all but one) of the five hundred monks in this Pravāraṇ ā ceremony are said to be arhats or fully liberated beings free from all taints (漏, āsrava),14 but no such claim is found in EĀ 32.5, where the Buddha says: “Among this assembly here, the most inferior on the lowest seat have/has15 attained the fruit of stream-entry,16 and will definitely ascend to the state of non-regression”. 17 That the stream-enterers here are characterized as going on to attain the state of non-regression implies that non-regression is something additional and that stream-enterers per se are liable to regression. To what, then, does “the state of non-regression” refer? The answer can be found in the Bu zhiyi lun 部執異論 (T 2033), a Sarvāstivāda text by Vasumitra on the process of schisms and the doctrines of the various schools. In its presentation of the tenets held by the Mahāsāṃghika school and its three offshoots,18 one tenet is stated thus: “A stream-enterer is subject to19 regression; arhatship is not subject to regression”. 20 This tenet conforms well to the above passage in EĀ 32.5, which suggests that the stream-enterers are liable to regress, and the “state of non-regression” refers most probably to arhatship. This is a significant indication that EĀ 32.5 may belong to the Mahāsāṃghikas.

A comparison of the same rules in the Vinayas of the Mahāsāṃghika, Theravāda, Sarvāstivāda and Mūlasarvāstivāda schools provides us with further relevant evidence on the affiliation of EĀ 32.5. In the Vinayas of these four schools, the fourth “offence of defeat” ( pārājika) is about a monk claiming to have attained extraordinary states if his claim is false. The Mahāsāṃghika version, while illustrating the rule that “falsely claiming to have attained extraordinary states is an offence of defeat”, 21 gives several examples, including the following: If a monk says: “O upāsikā [female lay disciple], the monks dwelling at such-and-such a place for the summer [rains retreat] have all attained arhatship”, and even says: “I have attained this state”, then he commits an “offence of defeat”. ... If a monk tells a male lay disciple or a female lay disciple: “None of the monks performing the Pravāraṇ ā ceremony at such-and-such a place is an ordinary person. They are all arhats, having all attained marvellous states”, and even says: “I have attained this state”, then he commits an “offence of defeat”. 22 This passage is not, however, found in the section on the fourth “offence of defeat” in any of the other three Vinayas derived from the Sthaviras.23 The seventh “offence of expiation” ( pācattika) in the Mahāsāṃghika Vinaya is about a monk claiming, in the presence of an unordained person, to have attained extraordinary states if the claim is true. The following passage appears in the section on this offence: If a monk tells a woman: “None of the monks dwelling at such-and-such a place for the summer [rains retreat] is an ordinary person”, then he commits a transgression of the Vinaya [which can be redressed by] repentance. ... If the female lay disciple asks: “Has your reverence also attained this state?” and he replies: “Yes”, then this is an “offence of expiation” if he speaks the truth. If a monk tells a female lay disciple: “The monks performing the Pravāraṇ ā ceremony at such-and-such a place are not ordinary people. They are all arhats”, then he commits a transgression of the Vinaya [which can be redressed by] repentance. If the female lay disciple says: “Is your reverence also performing the Pravāraṇ ā ceremony there?” and he replies: “Yes”, then he transgresses the Vinaya. If she further asks: “Has your reverence also attained arhatship?” and he replies: “Yes”, then this is an “offence of expiation” if he speaks the truth.2
This passage too is absent from the corresponding section of the other three Vinayas, namely: the seventh “offence of expiation” (pātayantika) in the Sarvāstivāda Vinaya, and the eighth “offence of expiation” ( pācittiya in Pali, pāyantika in the Mūlasarvāstivāda) in the Theravāda and Mūlasarvāstivāda Vinayas. 25 Although the above two Mahāsāṃghika passages relating to the rains retreat are not found in any of the other three Vinayas of the Sthavira lineage, all of the four Vinayas give two stories about the inappropriate behaviour of the monks during the rains retreat as the reasons why the Buddha laid down the two rules against claiming to have attained extraordinary states in regard to true and false claims respectively.26 From this it follows that the Buddha would have made it clear what should not be done during the rains retreat. Whereas the three Sthavira versions are all silent on this matter, the Mahāsāṃghika Vinaya clearly states, in the two passages cited above, that the Buddha prohibits monks from claiming to have attained extraordinary states, in regard to true and false claims respectively, during the rains retreat, including the Pravāraṇ ā ceremony. It is very likely that the Sthaviras omitted these two passages from their Vinaya lest these passages bring about embarrassing associations with their modified Pavāraṇā Sutta, now preserved in the five versions, which claim that the five hundred monks performing the Pravāraṇ ā ceremony were all arhats (or all with one exception). These two passages in the Mahāsāṃghika Vinaya are probably fairly close to the original. As Nattier and Prebish (1977: 267) indicate, the Mahāsāṃghika Vinaya is “noted by many scholars to be the most ancient of all the Vinayas”. 27 On the other hand, we cannot rule out the possibility that the two passages were interpolated into the Mahāsāṃghika Vinaya in order to criticize the Sthavira versions of the Pavāraṇā Sutta. In either case, it is most likely that EĀ 32.5 is affiliated to the Mahāsāṃghikas, who, in accordance with their Vinaya, disapproved of the claim that the five hundred monks in that Pravāraṇ ā ceremony were all arhats.

EĀ 32.5 is the only version that has the following episode (abridged translation of the text with my notes): The Buddha told Ānanda: “You should now strike the bell immediately in an open space because today is the fifteenth day of the seventh month, the Day of Invitation ( pravāraṇā)”. 28 Then Ānanda carried on a fairly long conversation with the Buddha in verse (probably seven verses in total, see the text in note 29). Having heard these words, the venerable Ānanda, delighted and overjoyed, ascended to a preaching hall, carrying the bell in his hands, and said: “I am now beating this Tathāgata’s drum of faith. All the disciples of the Tathāgata should assemble”. Then he uttered verses thus: “Subduing Māra’s power and grievances, removing the fetters without remainder, in an open space I am striking the bell, on hearing which monks should assemble. Those who wish to listen to the Dharma and cross over the sea of birth and death, on hearing this wonderful sound, should all gather here”. Then the venerable Ānanda, having struck the bell, approached the Blessed One and said: “It is the right time now. May the Blessed One impart instruction”. The Blessed One told Ānanda: “You can take a seat in order [of seniority]. The Tathāgata will know the time himself”. 29 Although EĀ 32.5 contains this detailed account of Ānanda’s conversation with the Buddha and his summoning of the assembly, no such account is found in any of the other five versions of the sūtra. What are we to make of this fact? At first glance, one might assume two possible scenarios: 1. This account was fabricated by the redactors of EĀ 32.5 and was added to the original sūtra, while the other five versions preserve the original. 2. This account was present in the original, but was totally removed from those five versions. There may, however, be no clear-cut answer to this question

Let us first examine the possibility of the first scenario, namely that this account was a later interpolation into the original. There are two apparent inconsistencies in the storyline that seem to betray the lateness of the account, or at least part of the account. First, the Buddha told Ānanda to strike the bell (揵 椎, ghaṇtạ̄ ), and he did strike the bell, but when he made the announcement, he said: “I am now beating this Tathāgata’s drum of faith (如來信鼓)”. This appears to be an inconsistency regarding the instrument used by Ānanda to make the announcement, but it is possible that “the Tathāgata’s drum of faith” is synonymous with the bell.30 Second, there is inconsistency regarding the location where Ānanda made the announcement. In a prose passage we read: “the venerable Ānanda, delighted and overjoyed, ascended to a preaching hall, carrying the bell in his hands”, whereas in one of the verses Ānanda says: “in an open space I am striking the bell”, and also at the beginning of this account the Buddha tells Ānanda: “You should now strike the bell immediately in an open space”. There is an obvious contradiction here: a preaching hall cannot be an open space. From the above discussion and the fact that this section about Ānanda is very lengthy and makes up about one-third of this sūtra (32 lines out of 87 in the Taishō edition), one may infer that this section underwent a process of modification or interpolation. Let us now turn to the second possible scenario: that this section about Ānanda was present in the original sūtra but was removed from all of those five versions. In this connection, Ānanda’s religious status needs to be clarified. Although he was an extremely important figure, as a long-time attendant of the Buddha, Ānanda did not attain arhatship until the commencement of the first council following the Buddha’s demise according to the Vinayas of the Theravāda (Vin II 285–6), Mūlasarvāstivāda (T 1451, XXIV 406a) and Mahāsāṃghika (T 1425, XXII 491b), while the Sarvāstivāda Vinaya records that Ānanda was still a “learner” (學人, śaiksạ ) rather than an arhat when he attended the first council (T 1435, XXIII 447b). As mentioned above, in the five Sthavira versions of the Pavāraṇā Sutta, all of the five hundred monks performing the Pravāraṇ ā ceremony (or all but one of them) are said to be arhats. In contrast, EĀ 32.5 does not say this or even mention the term “arhat”. Since Ānanda was considered by the Buddhist traditions to be a monk who was not yet an arhat during the Buddha’s lifetime, the redactors of the five Sthavira versions that depict the Pravāraṇā ceremony as an assembly of arhats may have found it inconvenient to preserve the episode involving Ānanda in this sūtra. The Theravāda version, SN 8.7, makes no mention at all of Ānanda, as if he were not present at the ceremony. The (Mūla-)Sarvāstivāda versions, MĀ 121, SĀ 1212, SĀ2 228 and T 63, evidently felt obliged to acknowledge the presence of Ānanda, so by using euphemistic expressions they unanimously mention that all of the five hundred monks are arhats except one, and only one of these versions, namely SĀ 1212, bluntly indicates that this exception is Ānanda.
From this discussion of the five versions of the Pavāraṇā Sutta, it is evident that there is a strong tendency in the schools of the Sthavira lineage to play down the role of Ānanda in that Pravāraṇ ā ceremony. They do this in various ways: mentioning Ānanda as a somewhat awkward exception to the assembly composed entirely of arhats; mentioning him in a euphemistic way without revealing his name; or even making no mention of him at all throughout the text. It can be inferred that Ānanda played a certain role in some ancestral version of the Pavāraṇā Sutta, but those five versions eliminated the section relating to Ānanda in order to iron out the contradiction between the role of Ānanda, a nonarhat, and the purported assembly of five hundred arhats. Why they wanted to portray the assembly as composed (almost) entirely of arhats is another issue, which will be dealt with later in this paper. As mentioned above, it can be inferred that Ānanda played a certain role in an ancestral version of the Pavāraṇā Sutta. This could be the episode of Ānanda summoning the assembly as related in EĀ 32.5. But as discussed above, the detailed account concerning Ānanda as we have it today must have been produced by modification of the ancestral version or interpolation into it. Why would the redactors of EĀ 32.5 have taken the trouble to do this? The key to this question probably lies in identifying the sectarian affiliation of this text. Referring to a passage in the Mahāsāṃghika Vinaya in Chinese translation, de Jong (1974: 65) says: The Bhiksuṇ ̣ī-vinaya (p. 314; Taishō XXII, p. 546c27) mentions festivals for the birth of the Buddha ( jātimahā), for his Awakening (bodhimahā), for his Setting in motion of the Wheel of the Law (dharmacakramahā), for Ānanda (Ānandamahā), and for Rāhula (Rāhulamahā). This probably indicates that Ānanda and Rāhula were held in high consideration by the Mahāsāṃghikas. Here is another indication that EĀ 32.5 is affiliated to the Mahāsāṃghikas. In this sūtra, the extravagant elaboration of Ānanda’s conversation with the Buddha and his summoning of the assembly could have been motivated by the Mahāsāṃghikas’ particular reverence for Ānanda. In a similar vein, the long set of introductory verses in the Prefatory Chapter of the Ekottarika Āgama features Ānanda and praises him highly (T 125, II 549b–550c). EĀ 32.5 stands out as the only extant Mahāsāṃghika version of the Pavāraṇā Sutta that preserves the Ānanda episode and even modifies and elaborates on it. The five representatives of the Sthavira lineage remove this episode from their versions of the sūtra to avoid the inconsistency between a non-arhat and the purported assembly of five hundred arhats. Why do the five versions of the Pavāraṇā Sutta, belonging to the Theravāda and (Mūla-)Sarvāstivāda, depict the assembly as a gathering of arhats? The reason could be that the Sthaviras were inclined to interpolate into their sūtras different categories of arhats, especially those “liberated in both ways” and those “liberated by wisdom”, in order to justify such a doctrine concerning different types of liberation. I will demonstrate that this was common practice in those schools of the Sthavira lineage and that the diverse kinds of liberation represent later developments in the textual history. EĀ 32.5 of the Mahāsāṃghikas apparently preserves the ancestral version, with no mention of categories of arhats endowed with different kinds of liberation. Different types of liberation The five Sthavira versions of the Pavāraṇā Sutta enumerate categories of arhats thus: 1. SN 8.7: Of these five hundred monks, Sāriputta, sixty monks have the three knowledges, sixty monks have the six supernormal knowledges, sixty monks are liberated in both ways, while the rest are liberated by wisdom.32 2. MĀ 121: Of these five hundred monks, ninety monks have attained the three knowledges, ninety monks have attained liberation in both ways, while the remaining monks have attained liberation by wisdom.33 3. SĀ 1212: Of these five hundred monks, ninety monks have attained the three knowledges, ninety monks have attained liberation in both ways, while the rest [have attained] liberation by wisdom.34 4. SĀ2 228: Among this assembly of monks, ninety monks possess the three knowledges, one hundred and eighty monks have attained liberation in both ways, while the rest are all liberated by wisdom.35 5. T 63: Of these five hundred monks, ninety monks have attained the three knowledges, ninety monks have attained liberation in both ways, while the remaining monks have attained liberation by wisdom.36 These versions speak of different types of arhat, including those who have the three knowledges (tevijjā), those who have the six “supernormal knowledges” (chalabhiññ ̣ ā, only in SN 8.7), those who are liberated in both ways (ubhato-bhāga-vimutta), and those who are liberated by wisdom ( paññā-vimutta), as if these were categories of arhat mutually exclusive of one another. But, as noted by Choong (2007: 41) and Bodhi (2007: 56), the main distinction is drawn between arhats “liberated in both ways” and arhats “liberated by wisdom”. This is because the three knowledges constitute the last (or highest) three of the six supernormal knowledges; they can certainly be acquired by those “liberated in both ways” and even by those “liberated by wisdom” if this term is defined in a stricter sense, as will be discussed below. Gombrich (1996: 112) points out that when the word paññā-vimutti (liberation by wisdom) appears in the texts, it is usually paired with ceto-vimutti (liberation of mind). He says (1996: 112): “I do not think that these words originated as technical terms ... There is only one release: it is a mental
event, triggered by insight”. In other words, ceto-vimutti paññā-vimutti as widely found in the texts should be taken as a single phrase expressing just one thing, namely mental (ceto) liberation (vimutti) triggered by wisdom ( paññā). Brahmali (2011) similarly states: “Wisdom-liberated is a universal term that describes all arahants ... Wisdom-liberated simply means that one has reached the highest liberation by fully penetrating reality with wisdom”. Why did the term “liberation by wisdom” come to be distinguished from “other types of liberation” and become a technical term, rather than a universal term? Gombrich (1996: 116) notes: It is a standard feature of the style of the suttas that words appear in pairs which are synonymous or nearly so. This feature is so pervasive that one need hardly argue for it ... On the other hand, it is equally typical of disciples and exegetes to pick over the words of the master to try to extract every grain of meaning from them. Below I will examine the diverse meanings attached to “liberation by wisdom” in various Buddhist texts, and seek to elucidate how the concept of “liberation by wisdom” developed and gave rise to the notion of arhats “liberated in both ways” (ubhato-bhāga-vimutta) as against those “liberated by wisdom” ( paññā-vimutta). 1. “Liberation by wisdom” as equal to “liberation of mind” in status but different in emphasis A sutta in the An. guttara Nikāya states: “Through the fading away of passion there is liberation of mind (ceto-vimutti), and through the fading away of ignorance there is liberation by wisdom ( paññā-vimutti)” (AN I 61).37 Removing “passion” is indispensable for liberation, and equally indispensable is eradicating “ignorance”. This is based on the following textual evidence. As found at many places in the Nikāyas, when the Buddha describes what happened at the moment he achieved liberation, he says: The mind was liberated from the taint of lust. The mind was liberated from the taint of becoming. The mind was liberated from the taint of ignorance. When it was liberated, there came the knowledge: “It is liberated”. 38 When the Buddha describes how his disciple achieves liberation, i.e. arhatship, he says the same words in the present tense instead of the past tense, as recorded many times in the Nikāyas. 39 Therefore, liberation as the attainment of Buddhahood or arhatship entails “liberating the mind” from lust (kāma), becoming (bhava, in the sense of rebirth)40 and ignorance (avijjā). Here “lust” (kāma) is synonymous with “passion” (rāga) in the passage at AN I 61; thus, in this passage “the fading away of passion” and “the fading away of ignorance” should be reckoned as equally indispensable for liberation. Therefore, “liberation of mind” and “liberation by wisdom” are equal in status, but the two emphasize different aspects of liberation: release from passion and release from ignorance. Liberation with both aspects is the “single” ultimate goal a practitioner can achieve. 2. “Liberation by wisdom” as superior to “liberation of mind” In MN 64 (I 437), Ānanda asks the Buddha why some monks attain liberation of mind (ceto-vimuttino) and others attain liberation by wisdom ( paññā-vimuttino). In the MĀ counterpart (T 26, I 780b), however, the question is why only some monks reach the highest level of attainment quickly.41 This corroborates the following remark by Gombrich (1996: 113): “This text thus strongly suggests that there are ... two qualitatively different experiences of release ... this text is the product of a scholastic debate”. When this Pali text is compared with its Chinese counterpart, this scholastic debate can explain the presence of the distinction between “liberation of mind” and “liberation by wisdom” in the Theravāda version of the text and the absence of such a distinction from the Sarvāstivāda version. Gombrich (1996: 118) says that the term paññā-vimutti came later to be regarded as hierarchically superior to ceto-vimutti and “this seems incompatible with AN I, 61”. Here “AN I, 61” refers to the passage discussed under “1”, above. As De Silva (1978: 120–1) points out, whereas “liberation by wisdom” ( paññā-vimutti) alone refers to final liberation, “liberation of mind” (cetovimutti) alone is hardly ever used in that sense. De Silva (p. 121) says: “Cetovimutti is repeatedly said to be derived from samatha ‘calm, tranquillity’, while paññāvimutti is said to be the result of vipassanā ‘introspection’”. This observation explains why “liberation of mind” is often described as temporary and inferior to “liberation by wisdom”. Since vipassanā is usually regarded by the tradition as the sine qua non of liberation while samatha is seen as subordinate to it and not essential for liberation,42 “liberation of mind”, often ascribed to samatha, is liable to be regarded as inferior to “liberation by wisdom”, often ascribed to vipassanā. De Silva (1978: 122) gives an example of various meditative states being referred to as “liberation of mind”, which implies temporary liberation. This is found in the Mahāvedalla Sutta (MN I 296–7), where eight types of meditation are each referred to as “liberation of mind”, namely adukkhamasukhā cetovimutti (i.e. the fourth jhāna), four appamāṇā cetovimutti, ākiñcaññā cetovimutti (i.e. ākiñcaññāyatana, one of the four formless attainments), suññatā cetovimutti, and animittā cetovimutti. This sutta corresponds to MĀ 211 in Chinese translation (T 26, I 790b–792b). While this Chinese counterpart also mentions four of these eight meditative states, i.e. the fourth jhāna, ākiñcaññāyatana, suññatā and animitta, it never refers to them as “liberation of mind”. This fact casts doubt on the interpretation of “liberation of mind” as temporary and inferior to “liberation by wisdom”. Below I will further demonstrate the fluidity in the definition or interpretation of paññā-vimutta (liberated by wisdom), a term corresponding etymologically and semantically to paññā-vimutti (liberation by wisdom) that is pervasive in the texts. 3. “Liberated by wisdom” as liberated by wisdom alone without formless attainments or without any samatha meditative attainments The Buddhist texts describe the four formless attainments (arūpa-samāpatti) as attainments in samatha or concentrative meditation higher than the four jhānas (dhyāna in Sanskrit). In line with the trend to devalue samatha and hence “liberation of mind”, those who conceived the category of arhats “liberated by wisdom” ( paññā-vimutta) as stated in the Susīma Sutta (SN 12.70: II 119–28) intended to provide a looser criterion for assessing arhats as far as samatha meditative attainments43 are concerned. The divergent accounts in the different versions of this text and the various interpretations of the text lead to disagreement on how loose the criterion can be, i.e. what is the minimal attainment in samatha possessed by an arhat. In the Pali version a number of monks claim to be “liberated by wisdom” without having developed the formless attainments and the first five of the six supernormal knowledges. The commentary glosses their claim as: “We are jhāna-less, dry-insight practitioners, liberated by wisdom alone”. 44 Bodhi (2007: 55, 62) holds that even the compilers of this sutta wish to insinuate that the arhats “liberated by wisdom” lack the distinguished states of concentration, including the four jhānas, but they dare not say this directly. This opinion can be backed up by Gombrich’s (1996: 126) view that the lack of the supernormal knowledges (abhiññā) may imply that the meditative attainments (i.e. the jhānas)45 which bestow those knowledges have not been reached. In contrast, Cousins (1996: 57) states that the references to arhats “liberated by wisdom” in the earlier texts (including the Susīma Sutta) seem mostly to say that they had not developed the formless attainments or the first five abhiññās, 46 but the later tradition even accepts that there were such arhats who had not developed all or even any of the four jhānas. In other words, according to the earlier texts arhats “liberated by wisdom” lack the formless attainments but still possess the jhānas. Likewise, Brahmali (2011) argues: “... in the Nikāyas paññāvimutta refers to an arahant who is liberated after attaining, at the very least, first jhāna”. Let us now turn to the Mūlasarvāstivāda version of the Susīma Sutta, SĀ 347 (T 99, II 96b–98a), extant in Chinese translation. In this version, a monk who claims to be “liberated by wisdom” concedes that he does not attain any of the four jhānas or the peaceful emancipations that are formless, transcending forms. In contrast with the ambiguity in the Theravāda version, the Mūlasarvāstivāda version explicitly admits the existence of arhats who have dispensed altogether with any attainments in concentrative meditation. Commenting on the Pali, SĀ and SĀ2 versions of the Pavāraṇā Sutta, Choong (2007: 41) says: “The wisdom-liberated, listed as the last category of Arhant, denotes the most basic and common level of attaining liberation shown in the three versions”. Thus, the five versions of the Pavāraṇā Sutta other than EĀ 32.5 all rank “liberation by wisdom” as the variety of arhatship that is the easiest to achieve and hence the commonest, just as the Susīma Sutta suggests. Therefore, those “liberated by wisdom” make up the largest number of arhats among the various types of arhats according to all five versions of the sūtra. 4. “Liberated by wisdom” as inferior to “liberated in both ways” The term “liberated in both ways” (ubhato-bhāga-vimutta) is likely to have been adopted in reaction to the development of the idea of arhats “liberated by wisdom” ( paññā-vimutta) discussed above, and in the texts this term almost always appears alongside “liberated by wisdom”. There is disagreement in the definition of these two types of arhat, which may well have resulted from the disagreement in the definition of “liberated by wisdom” mentioned above. The Kītạ̄ giri Sutta (MN I 477) lists seven spiritual types, of which the two highest are “liberated in both ways” and “liberated by wisdom”. The distinction between these two types of arhat is this: a person “liberated in both ways” touches in person47 those peaceful emancipations (vimokha) that are formless and transcending forms, and remains in them; whereas a person who is “liberated by wisdom” does not have such experience of those formless emancipations. Apparently, “those formless emancipations” refers to the four formless attainments which transcend the jhānas in the sphere of form. By implication, those who are “liberated by wisdom” may have experience of the jhānas. Unlike the Pali version, the Chinese version of the Kītạ̄ giri Sutta, MĀ 195 (T 26, I 751b), states that a monk who is “liberated in both ways” touches in person the eight emancipations and remains in them, whereas a monk who is “liberated by wisdom” does not have such experience of the eight emancipations. The same definition of the two types of arhat is also found in the Chinese SĀ 936 (T 99, II 240a) and even in a Pali Abhidhamma work, the Puggalapaññatti (p. 14). The last five of the eight emancipations are the four formless attainments and the “cessation of perception and feeling” (saññā-vedayita-nirodha) according to many sūtras, 48 but the identity of the first three is unclear in the sūtras. In the Dhammasan. gaṇ i, an Abhidhamma work, the first three emancipations are connected with the four jhānas. 49 Bodhi (2007: 69, n. 43) explains them thus: “The first three emancipations are equivalent to the four jhānas, but they deal with the state of jhāna in terms of its objects rather than in terms of its subjective experience”. Therefore, according to the exposition of “liberated by wisdom” and “liberated in both ways” as found in the Pali Kītạ̄ giri Sutta, an arhat (arahant) “liberated by wisdom” may have experience of the jhānas, while an arhat “liberated in both ways” differs in having higher attainments in concentrative meditation, viz. the formless emancipations. In contrast, according to the exposition of these two types of arhat in the foregoing two Chinese sūtras and the Pali Abhidhamma, an arhat “liberated by wisdom” lacks all attainments in concentrative meditation, including even the jhānas. This contrast forms an interesting parallel to that presented above: the Pali version of the Susīma Sutta allows arhats “liberated by wisdom” to be interpreted as possessing the jhānas, whereas its commentary and the Chinese version of this text (SĀ 347) starkly describe arhats “liberated by wisdom” as altogether lacking concentrative attainments, including the jhānas. This can hardly be a coincidence. It suggests the possibility that the term “liberated in both ways” was adopted in order to provide a higher criterion for full-scale liberation in reaction to the tendency to loosen the criteria for liberation in the development of the idea of “liberated by wisdom”. A further argument for the lateness of the concept of “liberated in both ways” is advanced below. Different paths to liberation In the Nikāyas the Buddha is repeatedly reported to state how he attained the four jhānas step-by-step, and then, on the basis of the fourth jhāna, developed the three knowledges in succession. At the moment he acquired the third knowledge, his mind was liberated from the taints of lust, becoming and ignorance. When his mind was liberated, there came the knowledge: “It is liberated”. This formula is found in many Pali suttas, such as AN 8.11 (IV 176–9), MN 4 (I 22–3), MN 36 (I 247–9), MN 85 (II 93), MN 100 (II 212), etc.50 The Buddha also exhorts his disciples to follow the same course of practice leading to liberation in many suttas such as MN 27 (I 181–4), MN 39 (I 276–80), DN 2 (I 73–84), DN 5 (I 147), etc. Unlike the above, the Tapussa Sutta (AN 9.41: IV 440–48) describes the Buddha’s path to liberation thus: he achieved the four jhānas step-by-step, and then the four formless attainments in succession, and finally he attained the “cessation of perception and feeling”. When the Buddha had attained and emerged from these nine meditative attainments in forward and reverse order, he knew and saw thus: “Unshakeable is my liberation of the mind”. 51 Here we have two different versions of the path to liberation. As far as attainments in concentration are concerned, a significant difference is that the first version requires only attainment of the four jhānas whereas the second requires all of the recognized nine attainments in concentration (i.e. the four jhānas, the four formless attainments, and the “cessation of perception and feeling”). It is inconceivable that the Buddha should have related his own experience of liberation in two such different ways considering that, according to the tradition, he supposedly achieved liberation “once and for all” rather than doing it twice. Thus a question arises: which of the two versions of the path to liberation is likely to be the earlier or closer to the original? Bodhi (2007: 57) observes: A number of standard texts define the concentration included in several groups among the thirty-seven “aids to enlightenment” as the four jhānas. In particular, we find ... the right concentration factor (sammā samādhi) of the noble eightfold path ... defined as the four jhānas. The “noble eightfold path” is a standard description of the path to liberation found throughout the canon. It is noteworthy that, as Bodhi points out, one of the eight path-factors, right concentration, is defined as the four jhānas. In comparison with the idea that the “cessation of perception and feeling” is an indispensable prerequisite for liberation, Wynne (2002: 39) suggests: The four jhānas are less problematic: it is asserted over and over again that they alone are the meditative states necessary for the attainment of liberation. It is therefore more likely that they represent the earlier doctrinal understanding. Therefore, the first version of the path to liberation (the four jhanas followed by the three knowledges) is most probably the earlier account of the Buddha’s teaching on the path. The second version resembles the path taken by those who are “liberated in both ways”, who progress through the four formless attainments and the “cessation of perception and feeling”. Wynne (2002: 37) proposes: “the list of spiritual types found in the Kītạ̄ giri Sutta, or at least the highest types in its hierarchy, was formulated by those early Buddhists who advocated formless meditation as a path to liberation”. He goes on to suggest that with the term “liberated in both ways” placed above “liberated by wisdom”, they wanted to say that their version of the path was superior to those paths that avoided formless meditation. This reinforces my argument that the term “liberated in both ways” was conceived at a later stage, perhaps in reaction to the development of “liberation by wisdom” in such a way that this expression was taken to mean “liberated by wisdom alone” without experiencing the formless attainments or even any meditative attainments in concentration. According to the above discussion, there are so many different and contradictory ways of interpreting “liberation by wisdom / liberated by wisdom” that they were certainly not taught by one man, the Buddha, as the tradition maintains, but rather resulted from a scholastic debate at a later point in time. It is reasonable to assume that the four meanings of “liberation by wisdom / liberated by wisdom” elucidated above represent the stages in the redefinition of this term in roughly chronological order. Originally “liberation of mind” and “liberation by wisdom” referred to the same thing. Later on the Buddha’s disciples or Buddhist exegetes tried to extract further meanings from these terms. First, “liberation of mind” and “liberation by wisdom”, while being treated as equal in status, were interpreted as emphasizing different aspects of liberation: release from passion and release from ignorance. Then “liberation of mind” was devalued as temporary, while “liberation by wisdom” alone was regarded as final, on the grounds that the former was attributed to samatha (deemed subordinate) while the latter was attributed to vipassanā (deemed essential). Perhaps at the same time another trend of interpreting “liberation by wisdom” developed in such a way as to lower the threshold for arhatship in terms of meditation so that as many Buddhist practitioners as possible could qualify as arhats “liberated by wisdom”. Such interpretation of the arhat may have led to polemics resulting in the use of the term “liberated in both ways” to designate the “fully-fledged” arhat with experience of all the nine attainments in concentrative meditation. Therefore, those who advocated “liberated in both ways” treated “liberated by wisdom” as inferior. The term “liberation by wisdom”, originally general and neutral in sense, gradually became technical and controversial. The development described above resulted in such ramifications that we now find diverse types of arhats in the Buddhist texts, including the five versions of the Pavāraṇā Sutta. “Liberated in both ways” versus “liberated by wisdom” as interpolation Below is my attempt to show that the category of arhats “liberated in both ways” as opposed to that of arhats “liberated by wisdom” represents later interpolation into the texts, including the text mentioning the seven types of noble disciple and the five versions of the Pavāraṇā Sutta. In addition to the Kītạ̄ giri Sutta, Wynne (2002: 35, n. 16) provides us with eight references to “liberated in both ways” as one term of a sevenfold list of noble disciples in the Nikāyas. I will now examine the texts containing these references and their parallel texts.52 (1) DN 28: III 105 This sutta is equivalent to an individual sūtra in Chinese translation (T 18) and to Sūtra 18 of the Dīrgha Āgama extant in Chinese, which is widely attributed to the Dharmaguptakas.53 While the sevenfold list also appears in T 18 (I 255c), the other parallel sūtra, DĀ 18 (T 1, I 76b–79a), makes no mention of the sevenfold list or of “liberated in both ways”. This discrepancy suggests possible interpolation of the sevenfold list into DN 28 and T 18. (2) DN 33: III 25354 This is equivalent to DĀ 9 and T 12. The term “liberated in both ways” as one of the seven types of disciple appears in DN 33 and T 12 (I 232c) in the section on sets of seven things, whereas this term – indeed the entire sevenfold list – is not found in the corresponding section of DĀ 9 (T 1, I 52a–b). (3) MN 65: I 439 Both the MN and MĀ (T 26, I 747a–b) versions of this text contrast Bhaddali’s disobedience with the obedience of the seven types of noble disciple with regard to the Buddha’s directives. The two versions illustrate the willingness of these disciples to follow the Buddha’s instructions by claiming that they would even lie down in the mud if asked to do so.55 This episode is absent, however, from the EĀ version of the text (EĀ 49.7 at T 125, II 800b–801c), which does not even mention the sevenfold list. Therefore, the authenticity of this episode together with the seven types of disciple is dubious. (4) AN 2.5.7: I 73–4 This passage has a parallel in a Sanskrit fragment in Tripathi (1995: 190, 26.51–4), which does not include words meaning “liberated in both ways” or “liberated by wisdom”. (5) AN 7.14: IV 10 No parallels in Chinese, Tibetan or Sanskrit. (6) AN 7.53: IV 77–8 No parallels in Chinese, Tibetan or Sanskrit. (7) AN 8.22: IV 215 This text describes the eight marvellous and wonderful qualities possessed by Ugga the householder. The sixth of these, as stated by Ugga himself, is as follows: It is not marvellous that when I entertain the San. gha, gods come and tell me: “Householder, such-and-such a monk is ‘liberated in both ways’, such a one is ‘liberated by wisdom’, such a one is a witness-in-person, such a one is an attained-to-view, such a one is liberated by faith, such a one is a Dharma-follower, such a one is a faith-follower ...”. But, venerable sir, I am not aware that while waiting upon the San. gha, I give rise to the thought: “I will give him little” or “I will give him much”. Rather, venerable sir, I give with a thought of equality.56 The seven types of disciple are listed here. This and the preceding sutta, AN 8.21, are in fact two versions of the same text but for a few divergences.57 In these two suttas the first and the eighth qualities possessed by Ugga are roughly identical and the remaining qualities are exactly the same except for the fifth and the sixth. The fifth quality in AN 8.21 is as follows: When, venerable sir, I attend upon a monk, I attend upon him respectfully and not with disrespect

The sixth quality in AN 8.21 is as follows: If, venerable sir, a venerable one preaches the Dharma to me, I listen respectfully and not with disrespect. If he does not preach the Dharma to me, I preach the Dharma to him.59 The fifth and sixth qualities in AN 8.21 are combined to form the fifth quality in AN 8.22: When, venerable sir, I attend upon a monk, I attend upon him respectfully and not with disrespect. If, venerable sir, a venerable one preaches the Dharma to me, I listen respectfully and not with disrespect. If he does not preach the Dharma to me, I preach the Dharma to him.60 This “fifth quality” was apparently produced by conflating the above two different qualities in order to make room for the sixth quality highlighting the seven types of disciple. In the Chinese parallel version, MĀ 38, the sixth quality is stated as follows: When I am giving to the San. gha of monks, gods dwelling in heaven tell me: “Householder, such a one is an arhat, such a one is approaching the stage of arhat, such a one is a non-returner, such a one is approaching the stage of non-returner, such a one is a once-returner, such a one is approaching the stage of once-returner, such a one is a stream-enterer, such a one is approaching the stage of stream-enterer ...”. I do not recall having a thought of discrimination while giving to the San. gha of monks.61 This version of the sixth quality is similar to that in the AN 8.22 version, but instead of giving the sevenfold list of disciples, this version enumerates the four pairs of persons, the eight types of individuals (cattāri purisayugāni atṭha purisapuggal ̣ ā).62 Here we have three versions of the sixth quality possessed by Ugga. Among the three versions, only one has the sevenfold list of disciples, including the monk “liberated in both ways”. This occurrence seems suspicious. (8) AN 10.16: V 23 No parallels in Chinese, Tibetan or Sanskrit. In addition, the Tapussa Sutta (discussed above) likewise has no parallels in Chinese, Tibetan or Sanskrit. In view of these facts, the sevenfold list of noble disciples or arhats “liberated in both ways” as against arhats “liberated by wisdom” could have been arbitrarily inserted in the suttas or have replaced another set of noble disciples. This may also be the case with the five versions of the Pavāraṇā Sutta that contain various categories of arhats, including those “liberated in both ways” and those “liberated by wisdom”. We have seen several instances in which the categories of arhats “liberated in both ways”, etc. are interpolated into Pali texts but not into the texts of other schools. A converse case, however, can also be found. For example, the Mūlasarvāstivāda SĀ 936 (T 99, II 240a) in Chinese translation enumerates the seven types of noble disciple while its Pali parallel, SN 55.24 (V 375–7), enumerates the four fruits. In this case the Pali version is the more credible since this text is concerned with the Buddha’s explanation for the Sakyans’ question about why a lay disciple who drank alcohol could have been declared by the Buddha to be a stream-enterer, which is one of the four fruits. The Chinese version of this discourse, by replacing the four fruits with the sevenfold list, fails to address the issue of that lay disciple’s becoming a stream-enterer. Therefore, the interpolation of the list of noble disciples including those “liberated in both ways” occurs in both Theravāda and Mūlasarvāstivāda texts. The term “liberated in both ways” as against “liberated by wisdom”, or in the context of the sevenfold list or of liberation entailing the formless attainments and even the “cessation of perception and feeling”, is found at many places in the Theravāda Nikāyas as mentioned above, at one place in the presumably Dharmaguptaka Dīrgha Āgama (T 1, I 62b), at four places in the Sarvāstivāda Madhyama Āgama (T 26, I 582a–b, 610b, 616a, 747a–b), at two places in the Mūlasarvāstivāda Saṃyukta Āgama (T 99, II 240a, 330b) and at two places in the Other Translation of the Saṃyukta Āgama (T 100, II 434b– c, 457c) also attributed to the Mūlasarvāstivāda. Therefore, the four primary Nikāyas of the Theravādins and the three Āgamas presumably affiliated to the Dharmaguptakas, Sarvāstivādins and Mūlasarvāstivādins all contain the term “liberated in both ways”. These schools were all descended from the Sthaviras, opposed to the Mahāsāṃghikas at the first schism, according to the sources of various schools.63 Curiously, the term “liberated in both ways” is not found in the Ekottarika Āgama, 64 which is a bulky corpus of 472 sūtras whose school affiliation has not yet been determined. It is likely that the school to which the EĀ belongs did not uphold the idea of “liberated in both ways”. While the school affiliation of the EĀ is still controversial, many scholars associate it with the Mahāsāṃghikas and, as mentioned above, I also suggest that at least a large part of the EĀ is of Mahāsāṃghika provenance. In this study EĀ 32.5 is identified as affiliated to the Mahāsāṃghikas. Does the whole EĀ belong to this school? If so, could the Mahāsāṃghikas have disagreed with the Sthaviras in classifying arhats into those “liberated in both ways” and those “liberated by wisdom”? These questions need further investigation.


Of the six complete versions of the Pavāraṇā Sutta, SN 8.7, MĀ 121, SĀ 1212, SĀ2 228 and T 63 are probably of the Sthavira lineage. The sectarian affiliation of the remaining version, EĀ 32.5, is one of the issues dealt with in this paper. I have argued that EĀ 32.5 belongs to the Mahāsāṃghikas on the following grounds: 1. It contains an idea of stream-entry which conforms to the Mahāsāṃghika doctrine. 2. The way it describes the assembly of monks is in accordance with two passages in the Mahāsāṃghika Vinaya, which do not allow for the depiction of an assembly in a Pravāraṇā ceremony as composed entirely of arhats as found in the other five versions. In contrast, the three relevant Vinayas of the Sthavira lineage lack counterparts for these two passages. 3. An extravagant elaboration of an episode featuring Ānanda is included in EĀ 32.5 but is absent from the other five versions. This episode, which bears signs of reworking, could have been motivated by the Mahāsāṃghikas’ particular respect for Ānanda as implied in their Vinaya. The Ānanda episode is missing from the five versions presumably because the Sthaviras wanted to iron out the inconsistency between the role of Ānanda, a non-arhat, and the purported assembly composed entirely of arhats. The reason why the five versions portray the assembly as a gathering of arhats could be that the Sthaviras wanted to interpolate into this sūtra different categories of arhats, especially those “liberated in both ways” and those “liberated by wisdom”, as they did in many other sūtras. This way of modifying the texts is revealed by a comparative study of several sets of parallel sūtras. Some sūtras have “liberated in both ways”, etc., while their parallels do not. Among such instances is the presence of arhats “liberated in both ways” and those “liberated by wisdom” in the five versions of the Pavāraṇā Sutta in contrast to the absence of such arhats from EĀ 32.5. The concept of “liberation by wisdom” or “liberated by wisdom” underwent a process of redefinition in the textual history. After this concept was redefined in such a way that the criterion for liberation was lowered in terms of attainments in concentrative meditation, there arose polemical exchanges which brought about the use of the term “liberated in both ways” to designate “fully-fledged” arhats with experience of all the nine attainments in concentrative meditation. This development resulted in such ramifications that we now find diverse types of arhats in the texts of the Sthavira lineage, including the five versions of our sūtra. It is reasonable to assume a common ancestral text from which the six versions of the Pavāraṇā Sutta are derived. Of the six, EĀ 32.5 appears to be closer to the ancestral text than any of the other five versions, as far as the passages discussed here are concerned.