**3.** *D. ak¯ arn ¯ . ava* **15: Materials, Peculiarities of the Language and Meter, and Editorial Policy**

## **3.1. Materials Employed**

I have consulted copies (digital and paper) of 22 Sanskrit manuscripts of the *D. ak¯ ar¯ n. ava* from Kathmandu, Tokyo, Kyoto, and New York. I consider that, currently, they are all of the available Sanskrit manuscripts of the *D. ak¯ ar¯ n. ava*. <sup>48</sup> (I have excluded fragmentary documents that appear to have been copied in the recent centuries.) All of the Sanskrit manuscripts were copied in Nepal. Of them, I have selected and used the following four Sanskrit manuscripts to edit the *D. ak¯ arn ¯ . ava* 15:


<sup>48</sup> For the other 18 of the 22 Sanskrit manuscripts that I consulted, see footnote 57 in this monograph. I have microfilm-copies, photocopies, or digital versions of those Sanskrit manuscripts. I collected many of them in Nepal and Japan in and after 2008, when I was given the first fund to study Chapter 15 of the *D. ak¯ ar¯ n. ava*, and some of them from Dr. Tanemura, who visited Nepal to collect Sanskrit manuscripts of various Buddhist texts. In 2016, I obtained digital versions of many of them from Dr. Serbaeva. These digital versions were extremely helpful.

<sup>49</sup> NGMPP A138/6 (paper, dated "*sa[m]vat* 894, *jyes.t.ha ´sukla* 10," in Newar script) and Matsunami 144 (paper, dated "*samvat* 917 *miti karttika k ¯ r.s.n. apañcamidine*," in Newar script) are similar to manuscript B (Matsunami 145); they may be copies of manuscript B. They are newer than manuscript B, and particularly, the former manuscript contains many scribal errors. Therefore, I have not used them in this monograph, although they are older than manuscripts C (NGMPP A142/2) and D (Goshima and Noguchi 41). In (Sugiki 2018a, 2018b), I also used NGMPP A138/6.

<sup>50</sup> Manuscript D is perhaps a direct copy of NGMPP B113/6 (paper, 147 folios, Newar script, dated NS 983, *paus.a*, *sukla ´* 15), which was produced three years before manuscript D. However, some of the folios of B113/6 are out of focus and illegible. Therefore, I have used manuscript D and not NGMPP B113/6. There appear to be no major text differences between them.

I used manuscript A as the base. Manuscript A is most likely the oldest among the available Sanskrit manuscripts of the *D. ak¯ ar¯ n. ava*. Manuscript B is the second oldest, although it was copied in the 17th century and is not too old. Manuscript C contains variant readings that are more in accordance with the orthographical and morphological rule of the standard Sanskrit; however, it also contains more scribal errors. The same thing is said of manuscript D to a certain degree, and, as I mention below, the *Man. d. alarcanavidhi ¯* , whose authorship is ascribed to Ratnasena, seems to have been produced from the transmissional line to which manuscript D belongs. These constitute the criteria of the selection of the four manuscripts. Maeda used none of manuscripts A, B, or C, nor discussed the relationship of manuscript D to the *Man. d. alarcanavidhi ¯* .

The relationship between the four manuscripts must also be considered from the following perspective: It seems that including manuscripts B, C, and D, all the Sanskrit manuscripts of the *D. ak¯ ar¯ n. ava* that I could consult, are direct or indirect copies of manuscript A. In other words, there is a high possibility that all the available Sanskrit manuscripts belong to the transmissional lines that are derived from manuscript A.

Some leaves in manuscript A are partially damaged. Some of the damaged parts are not transcribed in manuscripts B, C, or D. This means the following: the damage to those leaves in manuscript A had occurred before manuscripts B, C, and D were produced; manuscripts B, C, and D were produced based on manuscript A (or some manuscript produced based on manuscript A); therefore, those damaged parts in manuscript A were not transcribed in manuscripts B, C, or D.51 In manuscript B, blank spaces are made for many of those parts. Manuscript B is very close to manuscript A; it appears to have been copied directly from manuscript A. Manuscripts C and D have words for some of those parts that are not transcribed in manuscript B. However, in many of the passages where they are present, those words do not naturally fit the context or do not make sense. Probably, they were interpolated to supplement the lost parts of the text by the scribes of manuscripts C and D, or scribes of other manuscripts on which manuscripts C and D were based. As mentioned previously, some cases can also be found where manuscripts C and D have variant readings that are more faithful to the rule of Classic Sanskrit than manuscript A. However, this does not necessarily mean that older texts are preserved in manuscripts C and D. Those variant readings in manuscript C and D seem to be emendations by the scribes of the two manuscripts or scribes of other manuscripts on which they relied, although it

<sup>51</sup> Some of the damaged parts in manuscript A are transcribed in manuscripts B, C, and/or D: This means that the damage to those parts occurred after manuscripts B, C, and D (or some manuscript on which they were based) had been produced.

could mean that some of the variant readings in manuscripts C and D are, by chance, identical to the readings in unknown older texts, which the scribes did not consult. The same things that I mentioned of manuscripts B, C, and D can be said of the other Sanskrit manuscripts that I consulted.<sup>52</sup> I do not mean that the text preserved in manuscript A is the original one of the *D. ak¯ ar¯ n. ava*. I suggest a high possibility that manuscript A is a version of the *D. ak¯ ar¯ n. ava* from which many Sanskrit manuscripts were produced in Nepal. Additionally, as I suggested in Chapter 2, manuscript A may be relatively near to the original text(s) of the extant version of the *D. ak¯ ar¯ n. ava*.

<sup>52</sup> For example, see the *D. ak¯ ar¯ n. ava*, 32.13cb (according to manuscript A [my unpublished edition]): *t.eti vajraprabhav¯ akhya ¯ m. vi*++++++*ngati ˙ h. pura¯h.* (: "++" means an illegible letter because of damage to the leaf). Of the letters *vi*++++++, a portion of the left part of the letter that follows *vi* is visible, but it is hard to determine what that letter is. The letter *nga ˙* also looks slightly like *n´˙sa*. However, that letter is certainly *nga ˙* in manuscript A. The Tibetan translation of this line is as follows: /*t.i ni rdo rje'i nus par grags/ /yul rnams kyi ni grong du bgrod//*. This line is a part of the discourse to explain the etymology of the name of the holy site Dev¯ıkot.a. The line in question explains the meaning of the letter *t.a*, the last word of Dev¯ıkot.a. Therefore, the *pada ¯* , which is partially illegible because of damage to the leaf (viz., *vi*++++++*ngati ˙ h. pura¯h.* ), must contain the letter *t.*. The Tibetan translation is *bgrod* ("travel"). This suggests that some form of the verb *at.* or *pat.* ("go," "travel," "roam," or equivalent) was present in the original *pada ¯* . Alternatively, the word *-gatih.* ("going" of *vi*++++++*ngati ˙ h.* ), which is equivalent to the Tibetan *bgrod*, is used to imply *at.* or *pat.* ("go"). Based on manuscript A and with reference to the Tibetan translation, I propose to restore the text as follows: *t.eti vajraprabhav¯ akhya ¯ m. vis.aya¯n. a¯n gati ˙ h. \*pura¯h.* (for *pur¯ıh.* , *pura¯n. am¯* , *pures.u*, or either of their singular forms), "[The letter] *t.a* is proclaimed to be the adamantine power. [With this power he] goes to the multitude of sensory objects (also meaning towns in local places)." This makes sense.

For that *pada ¯* , *vi*++++++*ngati ˙ h. pura¯h.* , the two accidents that occurred in manuscript A mentioned above, namely, the damage to the leaf (viz., *vi*++++++) and the letter *nga ˙* slightly resembling *n´˙sa*, appear to have affected the readings in the other Sanskrit manuscripts. The readings in manuscripts B, C, and D are as follows: *vim*(a blank space for about one letter)*sati ´ h. pura¯h.* B; *vim. sati ca saha ´ h. pura¯h.* C; and (a blank space for about three letters)*vim. sati ´ h. vara¯h.* D. In manuscripts C and D, the part *vi*++++++*ngati ˙ h.* was changed to *vim. sati ´ h.* ("twenty"), which does not naturally fit the context. It is evidently an (incorrect) emendation as *vi* and *ngati ˙ h.* (resembling *n´˙satih.*) were forcibly combined to create *vim. sati ´ h.* , despite the existence of some letters between them. In manuscript C, the words *ca sahah.* ("and powerful," or equivalent) were also added, perhaps to accommodate the meter. The word *pura¯h.* was changed to *vara¯h.* ("supreme") in manuscript D. The readings in the other 20 Sanskrit manuscripts that I consulted are as follows: (1) Those similar to the readings in manuscripts A and B: *vi*(a blank space for about three letters)*n´˙satih. pura¯h.* (NGMPP C42/9=NGMPP C94/2, 141v1–2); *vi´satih. pura¯h.* (NGMPP E419/15, 109r5); *vim. sati ´ h.* (a blank space for three letters) *pura¯h.* (NGMPP E650/16, 100r1); *vim. sati ´ h. pura¯h.* (NGMPP D40/6, 53v4; NGMPP D15/4, 75r2; and NGMPP E1729/4, 76r2); and *vim. sati ´ h.* (a blank space for about three letters) *pura¯* (NGMPP A1275/17, 105r6); (2) Those similar to the reading in manuscript D: (a blank space for about three letters)*vim. sati ´ h. vara¯h.* (NGMPP B113/6, 89v5); *vim. sati ´ h. vara¯* (NGMPP E3350/1, 114r6); (a blank space for three letters)*vim. sati ´ h. para¯h.* (NGMPP B113/3, 61v10); and *vi´satih. para¯h.* (NGMPP E1476/3=IASWR MBB-I-66, 82v6); (3) Those that contain *sahah.* , which are, in this respect, similar to the reading in C: (a blank space for about three letters)*vim. sati ´ h. para¯h.* (ac) and (a blank space for about three letters)*vim. sati ´ m. ca sahah. para¯h.* (pc) (Matsunami 144, 144r1); *vim. sati ´ ca sahah. para¯h.* (NGMPP A141/4, 97v1); *vim. sati ´ m. ca sahah. para¯h.* (NGMPP D15/6=NGMPP E1841/2, 105r1); *vi´sati ce sahah. para¯h.* (NGMPP E1555/9, 102v6); and *yim. satti ca saha ´ h.* (NGMPP G238/16, 119v4). This suggests a possibility that the interpolation of *sahah.* originally occurred in Matsunami 144, which is a Sanskrit manuscript older than manuscript C (see footnote 49 in this monograph); (4) The other: (a blank space for about two letters)*vim. sati ´ h.* (NGMPP E422/11, 105r2). The leaf that contains this *pada ¯* is lost in NGMPP A138/6.

Notably, although its language is highly non-standard, which I will clarify in Chapter 3.2, there are not too many corruptions that do not make sense in manuscript A, and the peculiarities of the language in manuscript A are also preserved in the parallel passages found in the texts that were developed on the basis of the *D. ak¯ ar¯ n. ava*. In my previous papers (Sugiki 2018a, 2018b), I analyzed manuscripts A and B and manuscripts C and D to belong to different lines of transmission. Now, I have reached a different conclusion.

For the reason discussed above, I focus on the version of the *D. ak¯ ar¯ n. ava* preserved in manuscript A, the oldest manuscript, which was produced and transmitted in Nepal, and presents a critical edition and translation of that version. Variant readings found in manuscripts B, C, and D and the other related materials that I have reported in the critical apparatus also serve as information of how the text preserved in manuscript A has been transmitted and transformed.

I have also edited a text of the Tibetan translation of the *D. ak¯ ar¯ n. ava* 15 as supporting material. The Tibetan text is useful for those studying the transmission of the *D. ak¯ ar¯ n. ava* in Tibet. I have used the two Tibetan translations (Tib) below. Of them, D is the base text:

D: Sde dge edition, Tohoku university catalogue no. 372. Chapter 15: 169r4–179v5. P: Peking edition, Otani University catalogue no. 19. Chapter 15: 35v4–46v1.

Although only the Tibetan translations are available, I consulted Padmavajra's *Bohita¯*, a commentary on the *D. ak¯ ar¯ n. ava* (Tib, D 1419, Chapter 15: 130v5–156r2), and Jayasena's *Ratnapadmaraganidhi ¯* (D 1516, my edition presented in Chapter 9 in this monograph). A fragment of a Sanskrit manuscript of the former work is extant (NGMPP A48/9, palm leaf, date unknown), but it does not include the part that explains Chapter 15.53 The latter work teaches deities' individual mantras, in which every deity's individual Sanskrit name is transliterated.

The *D. ak¯ ar¯ n. ava* 15 contains various deities' names and ritual terms; it includes the names of many deities, phrases, and passages that resemble those found in other scriptures and commentaries.54 Among those whose Sanskrit manuscripts

<sup>53</sup> NGMPP A48/9 contains three leaves, all of which are partially damaged. The first leaf (whose folio number appears as 133rv) includes a commentary on the last verses of Chapter 22 and the opening verses of Chapter 23 of the *D. ak¯ ar¯ n. ava*; the second leaf (whose folio number appears as 173rv?), a commentary on some verses in Section 1 of Chapter 50; and the third leaf (whose folio number is illegible due to damage to leaf), a commentary on some verses of Chapter 27. From NGMPP A48/9, we can collect some *prat¯ıka*s or quotations in Sanskrit from the *D. ak¯ ar¯ n. ava*. With this handful of *prat¯ıka*s, it is difficult to decide whether there is any significant difference between the quoted text in NGMPP A48/9 and the text preserved in manuscript A. However, some peculiarities found in the latter are also present in the former.

<sup>54</sup> As discussed in Chapter 2, the *D. ak¯ ar¯ n. ava* was composed in the latest stage of the history of Indian Buddhism in the area (east India or Nepal) where both Buddhism and Saivism flourished. It is natural ´

are extant, the texts that contain major parallels are Vajrapa¯n. i's *Laghutantrat.¯ıka¯* (Skt ed., (Cicuzza 2001)), the *Kalacakra ¯* (Skt ed., (Dwivedi 1994)), Pun. d. ar¯ıka's *Vimalaprabha¯* (Skt ed., (Dwivedi 1994)),55 Umapatideva's ¯ *Vajravar¯ ah¯ ¯ısadhana ¯* (Skt ed., (English 2002)),56 and some others that are mentioned in Chapter 4 and the Primary Sources Section in this monograph. Chapters 10 and 29 of the *D. ak¯ ar¯ n. ava* also include major parallels. The texts that are most resembled are Chapter 20 of the *Sr´ ¯ıvajravar¯ ah¯ ¯ıkalpamahatantrar ¯ aja ¯* (abbreviated to *Var¯ ah¯ ¯ıkalpa*: Skt ms., Matsunami 346, 74r1–92r7, paper, Newar script, dated NS 937, *phalgu ¯ n. a*, *sukla ´* 10)57 and Ratnasena's *Sr´ ¯ımahasa ¯ m. varasaparikaraman. d. alarcanavidhi ¯* (abbreviated to Ratnasena's *Man. d. alarcanavidhi ¯* : Skt ms., NGMPP B24/52, 1v1–34r6 [the whole manuscript except for its colophon, 34r6–r7], palm leaf [19rv missing], Newar script, undated). The entire text of Chapter 20 of the *Var¯ ah¯ ¯ıkalpa* is almost identical to the discourse on the Heruka man. d. ala in the *D. ak¯ ar¯ n. ava* 15. Although partially emended and reorganized to make it a ritual manual for actual performance, the entire text of Ratnasena's *Man. d. alarcanavidhi ¯* is also similar to the *D. ak¯ ar¯ n. ava* 15.58 They were composed based

that the *D. ak¯ ar¯ n. ava* has many deities and ideas in common with other Buddhist and Saiva texts. In ´ this monograph, I have not pointed out minor parallels; I have noted only texts that include major parallels.

<sup>55</sup> The compilation of the *Laghutantrat.¯ıka¯* is datable to around the end of the 10th or the beginning of the 11th century. The *Kal¯ acakra ¯* was completed in the early 11th century. The *Vimalaprabha¯* was also composed around the early 11th century: it may be contemporaneous with or is slightly later than the *Kalacakra ¯* . For the chronology of these texts, see (Newman 1998, p. 343; Wallace 2001, pp. 3–4; Cicuzza 2001, p. 13; Isaacson and Sferra 2014, p. 97, footnote 18; Sferra 2015, p. 341, 343; Isaacson and Sferra 2015b, p. 477).

<sup>56</sup> (English 2002, pp. 12–13) said that Umapatideva may have been active between the 11th and 12th ¯ centuries.

<sup>57</sup> In Matsunami 346, this chapter is scribed as "the 18th chapter" (*-pat.ala as.t.ada´ ¯ sah.* ; 95v5). However, this is actually the 20th chapter.

<sup>58</sup> The collation between the *D. ak¯ ar¯ n. ava* 15, Ratnasena's *Man. d. alarcanavidhi ¯* ("R"), and the *Var¯ ah¯ ¯ıkalpa* 20 ("V") is presented below. The sign "—" means that no parallel lines can be found because they were not originally included or because the text is so much changed or reorganized that it is hard to deem it as a parallel line: *D. ak¯ ar¯ n. ava* 15.1–2 (manuscript A 13r10) = R — = V 74r1–r2 ♦ *D. ak¯ ar¯ n. ava* 15.3–5 (A 13r10–r11) = R 1v1–v3 = V 74r2–r4 ♦ *D. ak¯ ar¯ n. ava* 15.6–12b (A 13r11–v1) = R 1v3–2r1 = V 74r4–v1 ♦ *D. ak¯ ar¯ n. ava* 15.12c–15 (A 13v1–v3) = R (19rv missing)–20r2 = V 74v1–v3 ♦ *D. ak¯ ar¯ n. ava* 15.16–23 (A 13v3–v5) = R 20r2–r7 = V 74v3–v7 ♦ *D. ak¯ ar¯ n. ava* 15.24 (A 13v5) = R 20v1–v2 = V 4v7–75r1 ♦ *D. ak¯ ar¯ n. ava* 15.25–28 (A 13v5–v7) = R 20v2–v5 = V 75r1–r4 ♦ *D. ak¯ ar¯ n. ava* 15.29–45b (A 13v7–v12) = R 20v5–21v3 = V 75r4–v7 ♦ *D. ak¯ ar¯ n. ava* 15.45c–51b (A 13v12–14r1) = R 9r6 and 21v3–v7 = V 75v7–76r2 ♦ *D. ak¯ ar¯ n. ava* 15.51c–60 (A 14r1–r3) = R 9r6–v6 and 21v7–22r6 = V 76r2–r6 ♦ *D. ak¯ ar¯ n. ava* 15.61–68b (A 14r3–r6) = R 9v6–10v5 and 22r6–v4 = V 76r6–v4 ♦ *D. ak¯ ar¯ n. ava* 15.68c–72 (A 14r6–r7) = R 23r4 = V 76v4–v7 ♦ *D. ak¯ ar¯ n. ava* 15.73–78b (A 14r7–r9) = R 10v5–11v4 and 23r4–23v1 = V 77r2–r5 ♦ *D. ak¯ ar¯ n. ava* 15.78c–80b (A 14r9) and 89 (A 14r12) = R 23v7 = V 77r5–r7 ♦ *D. ak¯ ar¯ n. ava* 15.80c–84 (A 14r10–r11) = R — = V 77r7–v2 ♦ *D. ak¯ ar¯ n. ava* 15.85 (A 14r11) = R 23v7 = V 77v4 ♦ *D. ak¯ ar¯ n. ava* 15.86–91 (A 14r11–v1) = R 11v4–12v2 and 23v7–24r5 = V 77v5–78r1 ♦ *D. ak¯ ar¯ n. ava* 15.92–94b (A 14v1–v2) = R 24v2–v3 = V 78r1–r3 ♦ *D. ak¯ ar¯ n. ava* 15.94c–98b (A 14v2–v3) = R 12v2–v6 and 24v3–v7 = V 78r3–r5 ♦ *D. ak¯ ar¯ n. ava* 15.98c–102b (A 14v3–v4)=R— = V 78r5–v1 ♦ *D. ak¯ ar¯ n. ava* 15.102c–112 (A 14v4–v7) = R 24v7–25r3 = V 78v1–79r2 ♦ *D. ak¯ ar¯ n. ava* 15.113–117 (A 14v7–v9) = R 12v6–13v1 and 25r3–r6 = V 79r2–r5 ♦ *D. ak¯ ar¯ n. ava* 15.118–123 (A

on the *D. ak¯ ar¯ n. ava* 15. Although its palm leaf manuscript exists as mentioned above, Ratnasena's *Man. d. alarcanavidhi ¯* does not appear so old a work; it was produced from the transmissional line from Sanskrit manuscript A to manuscript D of the *D. ak¯ ar¯ n. ava*, as it contains a passage that only the Sanskrit manuscripts belonging to this transmissional line have.59 The *Var¯ ah¯ ¯ıkalpa* and Ratnasena's *Man. d. alarcanavidhi ¯* were not translated into Tibetan. They were most likely composed in Nepal, where the *D. ak¯ ar¯ n. ava* flourished to a certain degree. I consider Jayasena's *Ratnapadmaraganidhi ¯* (12th century) to be older than Ratnasena's *Man. d. alarcanavidhi ¯* (and it appears that Jayasena had a better understanding of the *D. ak¯ ar¯ n. ava* 15 than Ratnasena did). Although both are manuals for visualizing the Heruka man. d. ala in the *D. ak¯ ar¯ n. ava* 15, no close relationship can be found between Jayasena's *Ratnapadmaraganidhi ¯* and Ratnasena's *Man. d. alarcanavidhi ¯* . Jayasena's *Ratnapadmaraganidhi ¯* is probably older than the *Var¯ ah¯ ¯ıkalpa*.
