"The Doctrine of Magic Female Spirits’ A critical edition of selected chapters of the Siddhayoge$varimata(tantra) with annotated translation and analysis DPhil. thesis, submitted on 7 June 1999, by Sudit Terese ‘Merton College“The Doctrine of Magic Female Spirits’ : A critical edition of select chapters of the Siddhayogesvarimata(tantra) with annotated translation and analysis ‘Submitted in Trinity term 1999 for the Degree of Doctor of Philosophy by Judit Torzsok, Merton College, Oxford “This thesis contains the first edition and anuoated translation of twenty-three (of thirty two) chapters of the shor redaction of the Sidahayogesvarima, te oldest scriptural Source ofthe theological school known a the Trika of Kashmir and one ofthe earliest, Sotees ofthe Tantric cull of mantragoddesses and female spirits (Yoginis), “The general introduction contains a discussion of the dating of the text, its two redactions and its place in the canon ofthe Vidyapitha, which teaches the pantheon and cult of predominantly female deities. ‘The general nttoduction i followed by a stuly of the pantheon of the ‘Siddhayogesvarimata and the rituals associatad with its goddesses. It concludes thatthe cult of the three mantra-goddesses includes elemnens ofthe archaic worship of the so- called Eight Mothers (Astamaikh), fusing it with the cult of Sarasvat the orthodox. goddess of leaing. “There follows detailed analysis ofthe language ofthe Sidahayogesvarimata, ‘contrasting its phonology, morphology and syntax with classical Sanskrit, Parallels in epic and Buddhist Hybrid Sansksit and the possible linguistic influence of Prakrt and ‘Apabhraméa are pointed out throughout. ‘Since only to manuscripts of te text are available, editorial policies and the role of parallels and emendations are discussed in detail before the critical edition Tn the critical edition thre registers of apparatus report lacunae in the manuscripts, quotations and testimonie used for establishing the text and variant readings. Paralle} passages from other edited and unedited Tantras have been used extensively in Teconstructng the text from the poorly transmitted manuscripts. Editorial procedures are presented in endnotes ‘Nine appendices are supplied: three contain editions of parallel passages from other unedited Tantas, (wo give transcriptions of some unedited portions ofthe Siddhayogesvarimata, one provides a comparison of two versions of the ‘Svacchandataniaas linguistic evidence for tantric forms, and two identify and analyse citations attributed to the Siddhayogesvarimatain the Prayasctasamuccaya of Hydayasiva (previously unedited) and in the Tanidloka of AbhinavaguptaAcknowledgements This thesis would not have been completed without the support and help of, a number of individuals. Fist of all, | would like to thank my supervisor, Prof. Alexis Sanderson, ‘who helped me throughout the formation of this thesis with his vast knowledge, and patently eorested many eos made. Alltheremsining oes are mine.) hank Dr jarunaga Isaacson for correcting and proof-reading a number of chapters and for his encouragement throughout, and Prof. Csaba Totdssy at ELTE University, Budapest, ‘who taught me Sanskrit and who has always supported me in all sorts of ways. For their help and support in various ways 1 thank: Dr. Fivind Kabrs, Dr. James Benson, Elizabeth English, Somdev Vasudeva, Dr. Dominic Goodall, Jurgen Hanneder and Csaba Dezs6. Tam grateful to the following institutions for allowing me to have copies of manuscripts: the National Archives in Kathmandu, the Asiatic Society of Bengal in Calcutta, the Indological Department of the French Institute of Pondicherry, Cambridge University Library and the Bodleian Library. Oxford, For their financial support, I should like to thank George Soros scholarship made it possible for me to come to Oxford inthe first pac College, whose generous Senior Scholarship enables mnost of this thesis, ‘and Emmanuel College, Cambridge. where | completed this work as a Research, Fellow. Finally, I thank all my friends and family For their encouragement. If theses were to be dedicated, | would dedicate it to my mother.Table of contents Introduction [General introduction..... ‘= The importance of the’ text = The title = The two recensions ... Zhe date of the SYM and its piace among the tantras = Summary of the edited chapters .. WU, Goddesses and their supernatural powers TH. Aiga language... ="Sandhi and phonology Morphology. > Morphosyntax ~ Irregular Compounding’. = Syntax The lexic = stylistic features = The metre IV. Editorial policies YV. The manuscripts of the Vi. Conventions in the apparatus VI, Orthographic conventions ‘Vili. Signs and abbreviations The text Chapter 1 2 Chapter: Chapter 3 Shaper 4 Chapter Chapter 6 Chapter 7 Chapter 8 Chapter 10 Chapter 1 Chapter 12 Chapter 13 Chapter 14 ‘Chapter 15, Chapter 16 18 Chapter 19 Gapera2 pier Chapter 29 Chapter 30 Chapter31 1... Chapter 32, I Notes on the constitution of the text and Aiga forras Chapter 1Chapter Chapter Chapter Chapter Chapter Chapter Chapter Chapter Chapter Chapter Chapter Chapter Chapter Chapter Chapter Chapter Chapter Chapter Chapter Chapter Chapter Chapter Translation Chapter Chapter Chapter Chapter Chapter Chapter Chapter Chapter Chapter Chapter Chapter Chapter Chapter Chapter Chapter Chapter Chapter Chapter Chapter Chapter Chapter Chapter Chapter Appendices 1 Transcription of chapter 2. Ciatons inthe Priyasciasamuccaya 2 Pullin th Abhi “supplement o Appendix 4Transeipion ofthe inserted folio nN ‘5. Two versions of the Svacchandatantra 1— 2.100 ...6. Yoginilaksana in the Tantrasadbhava, T. Colophon: 8. The SYM — 9. The Brahmaydmalas list ofthe texis of the Vidyapith ‘Works consulted aeIntroductionI. General Introduction 1, The importance of the text “The Sanskrit text edited and translated here forthe first ime, the Siddhayogesvarimata or The doctrine of magic Female Spirits, is one ofthe demonstrably earliest seriptural sources in India in which the non-orthodox (non-Vedic / Tantric) worship of female deities predominates. It belongs to the tradition ofthe Trika (Triad), whose distinguishing marks include the central worship of three mantra- ‘goddesses, Parl, Pardpard and Apari, and which is the tradition followed by Abhinavagupta (fl. c. 975-1025 CE)! one of te most original exezetes inthe history of Indian religious thought. “The text teaches first the principal mantras of the system, which are used in daily Worship ofthe deities as well as at initiation and in rites of magie. It then describes the procertue of initiation in a rather dense way. The description of initiation also includes three ancillary topics: (1) the rules of conduct to be observed by initiates, (2) the way in which the worshipper isto transform his body into a deity or Female Power, and (3) the structure ofthe universe, whose diferent levels are governed by various groups of goddess-lke figures known as Mothers. The main subject ofthe rest ~ about two thirds = of the text is various rites of magic which involves also the invocation and worship of, the appropriate circle of deities / Female Powers and preceded by observances, associated with the protection mantras. The last chapters contain information on the relation of the text 1 other traditions and on its mythical and allegedly oral transmission, {As the above summary shows, the Siddhayogeivarimata principally deals with ritual prescriptions and consequently its mais interest i that it contributes to our understanding of ritual and th history of ritual in Inia, In addition to this, it contains important iconographic details of deities it isthe only text to give the detailed iconography ofthe three principal mantra goddesses ofthe Trika and it also has a remarkable passage on the iconography of the ferocious form of Siva, Bhairav Moreover its chapter 29 describing different categories of women recognised as embodiments of magic Female Powers can be shown to have served as textual source for the Buddhist tantras of Samvara on the same subject. Thus, its also an important On the dating, tee Sanderson (1988:158). The Tika of Abhinavagupss exegesis is of course diferent from the Trika of tye SYM, bu the later is part of the former. On the evolution ofthe Tika see Sanderson (1988.164).piece of evidence in establishing the dependence of Buddhist tantras on their Saiva counterparts, ‘Lastly, the language ofthe Siddhayogesvarimata provides us with an early and ‘unpurified example of Aisa Sanskrit} (Aigo ic. the language of Siva, who proclaims these texts and who is also called simply a, the Lord). This language evolved as the language of a certain group of tantric scriptures, especially ~ although by no means ‘exclusively as the language of seriptures usually dominated by or involving goddess ‘worship, particularly inthe earlier period oftheir formationt. I is based on Sanskrit ‘with a number of deviations or changes which show the influence of the spoken languages ofthe time, influence of Prakrit and! Apublramsa. It seems that this peculiar language was also associated in a number of cases with the esotericism of the text: the more the language of the text differs from the classical Sanskrit of the orthodox, ihe mote esoteric its teaching is. Although the Aisa language is described in the easiest way by pointing out its deviations from standard or classical Sanskrit it is not to be considered simply as ‘bad Sanskrit, but asa language on its own right. 2. The title “The tte ofthe text oocurs in slight variations inthe colophons and margins as Siddbayogesvarimata’ and Siddhesvarimata, while te name Siddhmata (or Sidhal Siddhétanta) and Yogesvarimata also occurs several times in citations in the 2 The textual dependence ofa number of Buddhist tanta: of Sarnvaraon Saiva counterpart has been amply shown by Prof. A. Sanderson ina series of lectures in All Souls College, Oxford, Hilary term 1998, The parallel passages referred to above have alo been identified by him, 3 Ausefl description of deviations from Sanskrit is found inthe edition by Goudriaan-Schoterman (1988:44-109), but their ent, the Kubjikdmacs, is probably much later than the Siddhayogesvarimara, Some other examples, such a that ofthe Svacchanda, must be weated with cae, since the avalabic caitions follow the purified txt adopted bythe commentator ofthe text, Kgemardj ee also the Aika texamples from the Nepalese manuserpts of the Sirdbtriaukdloaa (as opposed tothe purted tersion of the ertial edition) by Prof. Sanderson as given in Goodell (19983301) 4 This statement does not take al the problems ofthe question into account On Aisa usage inthe scriptures ofthe Siddhnt, see Goodall (1998: Ixv-Ix). The Bharavatantas are also manly writen ‘with many Alda forms, while some Skt tanras such asthe Milinfvjayotara ofthe Tika have much Fewer Aisa examples than the Siddhayogesvarimat. On te argument that more Aga examples may show the antiquity ofthe ext see the above cted work by Goodall However, some Sata tanias ‘which ae nt among the very earliest, but are among the very esoleric ones ~ suchas the Jiyadrathaylmal or the Kaulsinanirpaya~ also abound in Aga forms. 3 the eolophons, the text ofthe manuscripts as wel asthe Tanrdloka often cite the mistaken vite “yogidvar® instead of *yopesvar™. The former seems wrong. for the word is used asa synonym for ‘opin. which can be paraphrased by ‘Goddesses of Yous! (yogesvar), but not by ‘Rulers / Goddesses of Fogle (jogisvah. The seibal confusion between -- and is unexceptional, due othe similarity ‘of the two vowels in pronunciation. However, itmust be noted that this ecafasion probably dates ‘uit carl, and already Jayaratha seems to mdersand the tile inthis form. For he reads yogivara in “Tanrdloke 243d and gloses i a follows: . yogindm aiSvarah syd ly athab. Nothing implies thot Abhinavagupa read yoglvarainstead of yogesvara hereTantraloka® The first element ~ siddha— is certainly used in the adjectival sense Its primary meaning is ‘perfect or ‘perfected / accomplished’, which can be applied to the following noun denoting Female Powers or Yoginis (Yogesvar'), ov altematively, te the doctrine (mata) itself. However, in the context of this tantra, siddha seems more likely to refer to supematural powers known as Siddhi-s, and means ‘magic’ of ‘conducive to magic — again possibly referring (othe Female Powers, or/and the doctrine itself. ‘YogeSvari meaning [female] master of Yoga’ denotes the same as Yogini, Without dwelling on the subject of the etymologies and meanings of Yoga, I think the word Yogint or Yogesvari is net necessarily intended to convey that these sometimes witch- like female spirits or goddesses have any particular relation to certain classical yogic practices. Instead, che word probably refers to the similarity of supernatural effects traditional Yogins can obtain (such as being infinitely small large, strong ete) and powers the Yoginis of this system can bestow. This i a tentative interpretation of the ‘word Ihave met neither in the primary nor in the secondary literature on the subject. However, the presumption that yogic practices do rot necessarily play a crucial role here? is also confirmed by the fact that alternative names or categories ofthese Female Powers are Sakint, dakin etc." ‘Thus, among the various possible translations ofthe title, the most probable is “The doctrine of female spirits endowed with supernatural powers. Since our text uses the ‘The Tanudloka uses ths fll ite Siddhayogesvarimata on several ocasions eg in 2.41, 3.220, BI, 15321 ete. and Yogesvarimata once in 28.61. Siddhmata oceus in 7.40, 15.157 among. ‘others, wile its variations (Sidaba /Siddbstatr) in 1243, 8.4 ete. On the ful list of ozcurences, See N. Rastogi’ inteductory first volume othe edition ofthe Tanrifoka, 260 and p266 and ‘Appendix 8 ofthis edition. It seems thatthe simple designation or tile Mala’ never refers tothe Slddhayogesvarimata, although layaatha wrongly takes i as such ad 15.156e4-157ab. [tis possible that Athinavagupta deliberately avoided such an ambiguity o ground for confusion between the doctrine ofthe anse Tika represented by the Siddhayogesvarimaia and the antiritualist Mata, ‘Moreover it cems also unlikely that he wed Mata to denote the Picumata-Brahmoyasmala, although there ae arguments to prove the opposite (see eg. Sanderson 1986: 148). Fr evidence that Mata does ot denote the SYM, see Appendix 8 ‘hiss pethaps also confirmed by the fat that on hee orcasions N wrongly writes Sidahi ‘Siddha inthe colophons. possibly undertanding them to be synonymous, See Apte's defi sida: ‘endowed with supernatural powers or faculties’ a That he adjective Siddha’ probably belongs to te YogeSvarl element rather than to “mata (doctrine) ‘can be supported by a parallel expression often used to denote the practitioner: siddhayogY (sed eg. ‘728s, By extension, the doctrine ofthese Younis s ofcourse also conducive to supernatural effec “Tis statement does not imply that Yoga isnot imperat forthe atainment of supernatural facukies: ‘yogic practices in fact play an important ole in the manipulation ofthe external world trough that of the ternal, However, Yoga ls practiced bythe Sadhaka and not bythe Yogins, therefore I doubt that ‘Yoga and Yogins sould be dieclyastocated. Nevertheless. its possible that they ae indiresty linked as assumed above, both being conducive to supernatural effects, possibly through youic ' Mention mast be made of the fact here shat while the short recension of the SYM has proctcally no teaching on Yoga the long recension cited in the Tanaloka seems to ha= had sections on yoric topics sce ete citation ad Tanroka 2519-213b onthe topic of nidicdrapa and the mention of ‘Ruan in the citation ad Tantaloks 3220-225. This, however, doesnot contradict the interpretation of the word Yogial above.first word, siddha asa verse-filling adjective fairly often the title simply refers to the teaching of female powers / spirits!" ‘The text isin fact both a doctrine expounding the nature and rites of these spirits (the teaching about Yoginis) and a teaching bestowed and transmitted by them (teaching, by these Yoginis), asthe last chapters ofthe text explain.!2 3. The two recensions Comparing our manuscript evidence (both of Nepalese origin) of the Siddhayogesvarimata with citations inthe Tanttoka, it must be concluded that we have. a different, shorter recension than that of the Kashmirian exegetes"3, There is evidence of two kinds for establishing the existence of two recensions. Firstly, there are some citations which can be identified in our Nepalese recension, but which differ slightly from their ns as cited in the Tantrloka, In cases where only a few words differ, one may presume that the quotation in the Tantrlloka is not fully precise or that its language is purified if the Nepalese version has more clements!4, However, the order and number of lines in the citations are also sometimes different from the Nepalese recension.15 This fact shows that other small differences may also derive from original differences between two existing recensions. ‘Secondly, there ace a number of quotations which are missing from our Nepalese recension, sometimes to such an extent that even the topic is left unmentioned. One of several examples is the list of fony Yoginis associated withthe syllables of the Pardpard ‘mantra, given by the commentator, Jayaratha, ad Tantriloka 30.25-26, another is the This then synonymous withthe allerative tile Yogesvarimaa and vhimately with Siddhimate, the lates paraphrasing Siddhayogesvari® with the feminine adjective Siddha. On the further alternative ‘ofthe tile, see subseetion 4 onthe place ofthe SYM inthe Saiva canon. TE Thave not distinguished above between the alternative tls °maa and *tantra for both words mean ‘doctrine. and their use wsvally depends onthe position ofthe title in the verse ~ tantra is used in 044 ‘and “mata in even pas, he later being much more frequent. (See. e.g. mate /“matam in 3.498, 3istb. 3.534, 84d, 22.38, 29.3, 2914, 29.19¢; and “raniram in 31-13, 32.4a) The colophons ‘Show the some alternation, using ®mata nineteen and “tanza four times. Thave used a combination of these titles as the tite ofthe thesis im orde o include the most common form ending with “mata ‘while indicating te group of scriptures to which it bongs, Taira, The same compounded ile was {eed in the edn ofthe Kubjitdmara(anta). {SPs fact was first remarked in Sanderson (1990:31). For a detailed discussion of ll the passages, see Appendix & Inthe present discussion I ite only some representative examples of differences and ‘imitates, 18" This seems tobe the ease with SYM 1.12-13 cited ad Tanloka 11.81.82, 15 See e.g, SYM 626-31 ax reproduced by ayaraha ad Tantdloka15.333-334, cited in the apparatus ad oc. In this eate the Nepalese recension seems slighly longer. See also the description ofthe ‘tansmission ofthe Tantra (Syatikrama), whichis summarised in 3 verses in a shor chapter in our Tecension (chapter 32}, while the citations by Jayaratha ad Tanaka 361-10 give a much more ‘tailed account comprising about 23 verses. However, the two accounts are not unrelated, both tmeniioning Bhairava, Bhaitav, Svacchandabhairava, Garuda, Sukra and Rima. For a comparison see testimonia ad fo. and Appendix 8.detailed description of the universe cited throughout chapter 8 ofthe Tantrifoka, or a fairly iong passage on the desireable characteristics of a human skull as a container of alcohol.16 The differences between the passages in the Nepalese recension and the citations also show that the style or way of narration of the longer recension of the Siddhayogesvarimata was different, giving more detailed descriptions, and it was perhaps less cryptic on ritual prescriptions”. “The existence ofthe two recensions, a shorter Nepalese one and a longer surviving, in citations by Kashmirian authors, raises the question oftheir relative dating, There are arguments forthe earlier dating of both recensions “The fact that the Nepalese one is shorter suggests that it is primary, for inthe large majority of cases of textual transmission in India, texts have become much longer rather than shorter, especially scriptures of tantric and purdnic type, Hewever, there are ‘examples for compressing texts ata later date, stich asthe history of the various recensions of the Kalotaratantra!®. Consequently, the argument concerning length cannot be decisive in either way without additional supporting evidence. ‘Another argument forthe earlier dating ofthe shorter, Nepalese recension is that ‘early Nepalese palm leaf copies of tantras seem to preserve earlier versions in general, and are more faithful to their originals, asin the case ofthe Svacchandatantra!® Although both our manuscripts of the Siddhayogesvarimata are relatively recent ones, cone of them is most probably the transcript ofan early palm-leaf manuscript, which was still available atthe beginning of this century 29 However, this argument can be opposed by the fact thatthe earliest citations are nevertheless those in the works of Kashmirian cexegetes, which would support the primacy of the longer recension. “The available evidence in support ofthe primacy ofthe longer version seems stronger, yet the relative dating of the recensions needs more decisive evidence. However, this relative dating unfortunately may not help in determining the date of the original, which is discussed below.?! 16 Tanurtloka and -Viveka ad 2725-286. 17 See eg the detailed insiructionsatibuted tothe SYM to draw a mandala in Tantrfoka31.155- 163. Noe, however. that ayaratha does not cite the SYM here although Abhinavagupta seem to be rawing on this text tis also posible dat at tis point Abhinavagupla Uses an anonymous manual ‘Gnavailzble forthe commentator (ee the reference toa Kulapaddhati in SYM 8.125) ofthat he bases the description on eal instruction, for Jayaratha sates only that Abhinavagupa teaches the ritual prescriptions of the SYM (tam eva vidhim da), but doesnot use any expression to indicate that we Seal witha citation ora paraphrase ofthe SYM (such as tam eva granthum pata). 1 sce the introduction n NR. Bhat’ edition ofthe Sérdbatricaukalonara(1979:xi) 19 For a comparison ofthe two recension of hs text see Appendix 5. The Nepalese version seems to bbe primary to that used (and peshaps sometimes alo slightly purified) by its Kashmirian commentator, ee. ‘On the manuscripts, see introduction othe edition. 21 suppose that even if additional evidence comes to ight itis very unlikely 0 be i the form of ‘entensive citations from avery early date, which would change the termini discussed below.4, The date of the Siddhayogeévarimata and its place among the tantras “The dating ofthe Siddhayogesvarimatais as difficult as that of other seripeural sources, since it was inthe interest of its authors to conceal any reference tothe date of composition, and to present the text as eternal and thus undatable revelation 22 Consequently, dating is possible only on the bass of citations and on the basis of relative dating according to its relation to other tantas. “The Siddhayogesvarimata must precede the Malinivijayottara, the basis of ‘Abhinavagupta's exegesis, since the Malinivijayottara itself states that it derives from the Siddhayogesvarimata. More precisely, it claims to be a shorter version of the original Siddhayogesvarimata, and also mentions thatthe Siddhayogesvarfmata consists of ninety million verses.23 We do not need to presume that a text ofthis improbable length ever existed, for the length of tantras is often exaggerated, partly because this exaggeration isa way to express the recognition of their importance, and partly because itis meant to convey that we inthe present decadent age, in the kali yuga, have only an inferior version of what was considerably longer and thus superior in aeons past n the case of the Malinivjayottara the assenion thatthe ‘original Tantra known to Bhairava ‘was reduced several times is also a way of expressing the redactional independence of the Miliniviayotara itself ‘The long recension of the SiddhayogeSvarimata cited by Jayaratha also contains an account of how the text became reduced in the transmission, having ‘only’ 78125 verses by the time it reached the world of mortals, Thus, the Malinivijayottaa's statement ‘may simply be understood to convey that it regards the Siddhayogesvarimata as an ‘older and more prestigious tradition and /or text to which it wishes to attach itself 25 72 On te an-hstoricism of Vedic revelation ss a model and the Mimamsl se Pollock 1989. Ant- histoneism wit reference to the Tans is discussed in Sanderson (1997:14), mentioning also the device of prophecy (by Bhairava) as away round the dficuly of refering t evens. 33 foct te Msaayortarasoys that th chon version counting twelve thousand verses) of the Maintvjayatanra consisting of thity milion verses, and that this Maliniijaya was comprised ftom the ninety million verses ofthe Siddhayogesvatmat: sidhay ogeSvaranta Svat” core. Siar? ed) navakotprvitaram at way kahit param bhedatraavisarpitam /mintvijaye ‘ante ktivitayalaksiteyogamargas tvayprokta svistmo mabesvara/ Bhlyastasyopasamrah rok dvadasabhistthé sabarai soi vstvo grate ndlpabuddhibib / Malinivijyotara 1.8 ib 2 The passage is cited ad Tantraloka 36.1-7ab, after which another passage is also cited by Jayaratha, ‘which gives the names of eight lage units with the numberof Slokas they contain. have not been Tbe to inerpet the whole passage with certainty buts clear that the following numberof verses are tributed t six ofthe sections: Pada 0, Mala 100, Uddhira 200, Unara 400, Kalpa 1000, Kalpaskanda 2000; the Byhadutara seems to have 100 aksaras (understand for Slokas?) andthe Samhita 14,062 (). The alleged total is then probably 17 912 verses, which is sul about eighteen ‘mes more than the numberof verses in the short recension. 1B There are several ther passages where the Milli vjayouara refers tothe Siddhayogesvarimat. (See siddhayogesvarimate/*matam 18.474, 17.3, 18384, 19484, 23.374; sidahayogesvartanre 2218;Since the Malintvijayottara was probably known to the Saiddhdntika Sadyojyotis?, ‘who probably predates Somananda (fl. c. 900-950 CE)” considerably, it may be presumed thatthe Maliniviayottara was composed before 800.28 The only ferminus ante quem we can establish is this date, for it would be too hypothetic to attempt to determine how much time is needed forthe composition and the acceptance of a scripture. However, it may not be too far-fetched to presume that the Siddhayogesvasimata ~ or its long recension ~ was probably composed around the seventh century CE. As the terminus post quem would be even more hypothetic, no attempt has been made here at its determination, [As tothe relation of the Siddhayogesvarimata to other tantras, it considers itself to belong to the Bhairava branch as one of the colophons shows? and the text itself ‘mentions on several occasions*!. This also ismplies an opposition to the less esoteric branch of the Saiva scriptures, to the Saiva Siddhanta,%? Moreover, the yorintmate 22.26) tis difficult to decide fl these references are genne or not, expecially because hey may involve passages missing from the shc= recension. However, suxpect that a numberof them serve only to teassen the above mentioned attachment of the Milinivjayonara othe SYM. (teferences inthe context of Dhdranis seem particularly suspect, forthe SYM does not seem to contain uch about yogic practices ofthis kind). JB See Goodall (1998Inx0), based on evidence adduced by Prof. Sanderson and by Diwakar Acharya ‘Mokyakanicd 16 © Mainiijayotaa 1176: Moksakriks We < Malinlvijayotara 1.18; Moksakirika 25 eMilingijayotara 254; Moksakiviks 2e-73b < Milinivijayara 1-18-19. Prof, Sanderson has Siso identified a tation of Malinivijayouara 260 without attnbution in Sadyojyotis's commentary ad ‘Seayambhuvasdtrasameraha 3.11 . 73 line 8 He has also propsed that Sadyojyotis's use ofthe terms pralaykala and vjinckal derives from the Mlinivjayotara (cited in Goodall 1998: 184, note ‘TD Inspite ofthe above evidence, Prof. Sanderson has poinied ou that all hese citations and parallels may also have come from the longer redaction of the Svyambhovagama, whose shor eduction unde the tile Svayambhuvasaoasamgraha shows several other parallels with Teka scriptures see cg paalels cited from Svdyambhuvasitrsamgraha chapter 21 a SYM chapter 10) Se Sanderson (1988:158) 28°This date canbe pushed back with the dating of Sadyojyots. thas been remarked recently by rot Sanderson and Dr Isaacson that Sayojyots doesnot attack Dharmakiri onthe question ofthe Existence of Atman in the NareSvargpariks therefore he must have lived before Dharmakiti became well-known and famous so a¢ not to be nestgit in such discussion. This means that Sadyoyyots's date may be pushed back as far as around 700 of possibly even ear and thus the Malinvjayouara ‘can perhaps be dated before that date. ‘Bye date may be established onthe bass shat all anrieiterture must postdate the PSSupatas including Kaupdinya, vio is hypotetucally placed inthe fourth century. However, since Kaundinya's dating has not been established wih erty, this terminus post quem is very speculative, More ‘ecinive evidence has been pt forward in Sanderson (1997:34), arguing thatthe adoption of cements lof Gteck astrology in ealy Sava texts (signs ofthe zodiac, weekday et.) shows that they cannot be ‘ated earlier than the fourth century AD. 50"This isthe interpetation ofthe odd colophon to chaper 21 I propose. which has ‘Bhairavavirasambitd as the tile (sripture of the Bhairav heroes”. {suspect tha may preserve the ‘emnant of an oder for ofthe clophons, which probably refered tothe fct thal the SYM was a Bhairava tana. See the colophons ofthe Tanasadbhava, which state thatthe text belongs 10 the ‘Bhairava branch, and to the VidySpiha within that branch: ii Bhairavasotas! mahdtanirevidySpthe ‘sapakoripramage srianasadbhdve pal ST See 1-19 (implicil), 32.6 and 32.13, the late two passages attaching the txt to the ‘Syacchandatantt, which then most probably predates the SYM. 32 That this opposition is more important inthe SYM than that between Bhairava Tanias and the ‘Vidyapita is clea fom 19. On a definition of the Siddhinta, see Sanderson (1988:136) and (1992: 282.257), The later applies tothe developed form ofthe Siddhanta rater than Lo the seriptura eve‘Siddhayogesvarimata claims to belong to the Vidydpitha (32.5),t0 the Collection of Female Mantras’, which is the branch where the worship of mantra-goddesses comes to the foreground, Within this branch, it knows of the Union tantras (Ymala tanras, listed in 29.17d-18 and said to belong to the Right Current by the Jayadrathaymala cited ad Joc.), whose principal cutis that of Bhairava and his consort, as the ‘Frightening Lord of the Skull’ and the Furious Goddess of the Skull”. Other tantra titles include those of the Left Current of the ‘Seat of Female Mantras (an exact parallel to the Jayadrathaydmal' ist) and a number of others which probably belonged to the Middle Current. From the testimony of the Jayadrathay:imalait seems that the SYM was part of the Right Current ofthe Vidyapitha, The SYM, however, does not divide the ‘VidyApitha into different currents. Below Ihave reproduced a table and some words of explanation from Sanderson (1988:137), which shows the place ofthe TrikaS and thus that of the Siddhayogesvarimata - within the Saiva scriptures. ‘Whatever is above and tothe ieft sees whatever is below and to the right as lower revelation ..] AS we ascend through these levels, from the Mantrapitha to the Yimala-tantras and thence tothe Trika and the Kali cult, we fin thatthe feminine rises stage by tage from subordination to complete autonomy, Tantras of Kali ‘Trikatantras Sakti tantras ‘Yamala-tantras Vidyapitha Mantrapitha Bhairava-tantras Saiva Siddhdnta MANTRAMARGA ‘While the SYM mentions only the $vacchandatantra from what is below the ‘Vidyapitha™, it gives a list of titles of the Vidyapitha, which runs parallel tothe list of For the occasional overlapping of doctrine and practice betv.een the Siddhanta andthe Bhairava branch athe scriptural level, see examples in Sanderson 1985, 3° The surviving tantas ofthe erly (non-kaula) Trika~the SYM, the Mlinivjayouara and the “Tanrasadbtva~ donot use the term Trika to refer to themselves, aris this erm mentioned in early lists of the Vidyopitha. Sanderson (199031) cites the Kularatnoddyota (Fo. 79) which defines the ‘Tika as having he three mantra goddesses (Pad, Pararhand Apa) and the two alphabet deities (Malini and Sabdariibhairaval Mauka) may be concluded ht the term was coined after the basic texts andthe system ofthe Tika became popular and in particular afte it produced a number of ‘giptres which needed tobe subsumed Under a single Ue 34 The reference tothe 64 divisions in 29.19 seems to imply thatthe SYM knows about the 64 ‘canonical Bharava Tantra although the reference icelf is ambiguous), but it does ot mention any ‘ther than the Svacchanda.the JayadrathaydmalS, These two and the Brahmayaimalds list are summarised in the table below, forall three are demonstrably early tantras ofthe VidyBpitha?6 and their lists may help in establishing a relative chronology. The classification into Currents (srotas) has been adopted from the Jayadrathayamala, The Brahmayamala mentions this, categorisation, but does not apply it to the title within the Vidyaptha.? Tiles common 1 at least two texts have been highlighted in bold type, Asterisks signal thatthe title also oecuts in thelist of the Sekapthiyasumhia, which divides them ino Left and Righ®8, The place (ofthe Siraocheds /Sirobyais ambiguous inthe Jsyadrathaydala, which identifies itself with i and ‘lassfies it under the lft aswell a belonging to oth Lett and Right. The table reflects this 3 inconsistency. JRY SYM BY Sarvavira irakhya (Mantramatin)"® Trigola | SYM* Siddhasara‘! Siddhas? Paficimrta/ Sricakra Paiicdmrta* (Prapatica* yoginija) Visvadya Visvadya (Aahorest) ‘Yoginijala*(-samvara) Yoginijila* Yorintjtiar Kalakhya (Aghoresvar) Khecara (Keigaghoresvart) 35 The Jayadrathayamalés passage is cite asa close parallel in he apparatus 2d SYM 29.15.19. 36 As Prof Sanderson has shown, all hee are quoted in the Taniztota. For the SYM, evidence has bon shown in the apparatus fore Jayadradhay mals (cted usually as Tandardabhatiaka) see es ‘Tanuloka and -vivekaad29100ab and Jyadrathaymala Ah saika(A 151.16) 0. 149189 21603 forthe Brakna mad idetiicaton se eg. Goud-an and Gupa (198142) For the ‘Brahmaydmala Sanderson 19972 noe 4 shows the terminus a> quem tobe much eae, since this text is mentioned inthe original Skandapurdna preserved ina manuscript dated $10 AD. See also ‘Advacnsen-Bakke- Isaacson (19984) footnote 9. 537 For the passage see Appendix 9. Tits inthe Brahmaydiala below have been aligned with he possible parallels inthe other wo texts. they ae unlikely to dente the same tex, they are in brackets. The order of enumeration tus has not been kept, but ithas been respected forthe other (Wo texts, Note thatthe Brahmaymala classifies eight BhairavaTantas headed by the Svacchanda under the Vidya. This may show the close relation of early Vidyaplhs texts Sachs the SYM withthe ‘Svacchands; but ii also likely 1 be due w te conflation of thelist of tangas coming from the Southern (Daksio) face in otber classifications (ee eg. Hanneder 1996:19), and the ist ofthe Right ‘Curent Daksina-sous) in the Vdyipia. The former comprises the Bhaava Tanta, while telat has the VAmala Tatras according othe JayadrathaySmala. This confsed sate ofthe list is also seen won the texts ofthe Let Curent ofthe Vip are mentioned ater the Siddha Tatras, thus, ‘pasted om thelist of her Vidya ext 5 ong passage rom the lost Srkanhya is cited in Takgakavarts [Nityadsamgrababhidhdnepaddha, and has been denied by ro. Sanderson Within tis passage the section on the sixty-four Bhaiava tanras is also cited in the -Viveka ad Taotrdloka 1.18. The fll pssage bas been transcribed in Hanneder 1996: Appendix 2. Some ties occur several times, and {heir order an alleged relation each ther i often very diferent fom what maybe infeed fom the ‘ther tee sources therefore Thave ony indicated the oceumenceof aie 539'As Pro. Sanderson has shown, the fist satka ofthe Jayadrathaydmala, which i alterativly called Siraccheds preserves something of a probably earlier Vamatanr. See eg. Sanderson (1990:214). The lernative names inthe Jayadrathay dala ft some text ~ such as Tria forthe SYM ~can be ‘enufed because the two Tas in which they oocur un parle 0°This may refer tothe Malinfvjayotara. and its worth noting that i “le follows the SYM. called Sidi) immedi in he Brahnay mala i The SYM itself sates in 32Se that it has an aerative name Sira “2Ths tile also wed forthe SYM by Abhinavagupia io Tunaloka 3.155.Sidhana® (Lakintkalpa) Savara (Man) Tilaka* (Mahamart) on Hydaya Yoginihrdayets idyabheda Vidyapitha ravidy gana’ (2Banuripas) (Girohrtay (Aghorastra) Sicaccheda* Siraccheda* Mahdsammohana* — Mahisammohana* Sanmoha* Nayottara® Nayottara® Nayottara* Mahiraudra Mabaraudra Bhavass aukra®) 2udrayamala’7* Rudrayamala* = Rudrayimala* rahmaymala* Brahmayimala* Brahmay&mal Visouyamala® Visnuydmala* —Vignuydmala* Skandayamala’ Skandayamala* — Skandayamala* Aumaydmala Umayamala Yamayamala (Giraceheda) ‘Viyuyimala Kuverayimala Indrayamala Svacchandabhairava* Krodhabhairava™ ‘Unmattabhairava®™ Ugrabhairava Kapélfbhairava*™ Shamkirabhairava Sekharafbhairava} Vijayabhairava™ Only three of the above titles have survived, none of which has been edited so far: the Siraccheda (in the Jayadrathaymals), the Brahmayamala and the SYM. While the first is probably a later version of the original, the later two have been identified with certainty. However, the Brahmayamala scems to be the only text to have survived in the ‘same redaction as that available for the Kashmirian exegetes. The presence of these ‘Tantras in all ists pre-dating the Kashmirian exegetes shows that they are likely to be the earliest scriptures of the Vidyapitha, which implies that the SYM is probably the ‘© The titles Sidhana and Savara/Savara /Subara may be understood as one single name ‘Sytarasiona, ‘hiss, of couse, not identical with the well-known Yoginthdaya ofthe Trpurasundart cult. (See Sanderson 1988:158) ‘This text may well be diferent from the Vidyabeda/ ‘pha the other two lists. 46 Prof. Sanderson emends this wordt ath, Thus te text reads: tad eaivanayottaram /Saukram (gira thd prokiam vamasroed vnirgatam. ‘7 Thistle seems to have been become extremely popular in two ways. Soue tater Kavla Thika txts, such a the Vijaanabhairava and the Parra atach themselves to this ana; and while no demonsrably ancient ext has survived unde thistle there are dozens of mare o es ecent remakes! ‘oft found in libraries throughout India,earliest seripture ofthe Tika‘ Ifthe position of the titles on thelist is significant such as, it seems, itis in the case ofthe canon of the Siddhanta!?— then the SYM — or its long recension ~ may be the earliest surviving scripture ofthe Vidyapitha. 5. Summary of the edited chapters ‘The following brief summary is intended to give some orientation to the reader. ‘The text is writen in the usual form of a dialogue between Bhairava, who expounds the doctrine, and the Goddess, who asks questions and at times summarises what has been taught (Chapter | stars with the question ofthe Goddess, who asks Bhairava why the mantras do not function as they should. Bhairava replies that they lack tneir inner vigour, which he has hidden from the unworthy, but will reveal in the subsequent chapters. Chapter 2 first describes the gura and lists the signs he is supposed to show when ‘possessed’ by the Power of Rudra. This is followed by the exposition on the three kinds of Powers / Female Spirits who rule the worid. They are equivalents of the three ‘main mantra-goddesses, who are briefly mentioned at the end of the chapter, and their function and character echo the theory of three Gunas in the Samkhya: Sattva (white/ benevolent), Rajas (red/passionate) and Tamas (dark/malevolent). ‘Chapter 3 gives the Malini-code, ie. the parts of the body of the Alphabet Goddess identified with letters ofthe Sanskrit alphabet. From vers: 2%, the chapter teaches the Pardpari, Apard and Pari mantras with the Malini code. The last verses promise various supernatural effects for those who know the mantras and recite them. Chapter 4 describes the so-called ancillary mantras, whose principal function isthe protection ofthe practitioner. Chapier 5 teaches the mantras ofthe Gtardians ofthe Directions. Chapter 6 is entitled the Samayamandala’, the mandala used at the Samaya initiation, by which people are accepted in the particular Saiva community. The description of the ‘mandala includes the detailed visualization of the three principal mantra goddesses (verses 19-27), which is preceded by the concise teaching of purificatory rituals. The Visualisation of the deities is followed by intemal and extemal worship the later includes offerings of human flesh and wine, From verse 36, the disciple is called in and his body is purified. He is lead to the mangala blindfolded, and casts a lower on it. His 48 On the relation ofthe Siddhayogesvarimatao other texts of the Trik, see Sanderson (1995:231F) apd Sanderson (1988:14019. 48 See Goodall (1998:lAxi the relative order the lists imply ofthe few sted Sidhanas that do survive seems not incompatible withthe other itera evidence gathered here for relative dating. This ile, however cannot be applied here with auch certainty, for we have very few texts surviving pri from de early Tiss of the Vidyspihainitiation name is determined by the place on which the lower falls. The chapter ends ‘with a number of rules to be observed, which are given tothe initiates after the ritual (verses 45-54), The closing lines (55-57) probably refer tothe topic of the following. chapter. Chapter 7 lists the main events of initiation proper (diksa), which ensures success in ‘magic and / or final release. The purification of the levels of the soul, which are equated with the levels ofthe universe, is performed by three purifications (verses 7-9). The practitioner is joined to the highest (transcendent level, if he wants to attain final release; he is joined to one level below if he intends to enjoy worly success (verses 15- 16), From verse 19, the guru is prescribed to teach the transformation of one's body into Bhairava (made up of female powers) according to this system, and this gives the opportunity to describe this procedure in detail, From verse 33, sarious supernatural effects are mentioned and attributed to the mantra goddesses, ‘Chapter 8 is also on initiation. It touches « number of topics very briefly - sometimes repeating what has been sai in the previous chapter ~ and gives additional details on the performance of ths rite. It seems that this chapter and chapter 7 are complementary, and the two are meant to be put together in order to have all the details. Verses 9-13 ‘mention the initiation mandala (different from the basic Samaya one described in chapter ©, and from verse 28 we are given a miscellaneous list of Mothers, which provides some evidence to determine the number and names of the levels ofthe universe (Tattvas) according to this doctrine. The closing part of the chapter ison the number of fire-ituals, but the last verses have only survived in fragments Chapter 9is very fragmentary, therefore [have relegated its transcription to Appendix 1. twas clearly written on fire-fferings, and the last verses, which have survived somewhat better, list a number of substances to be offered. Chapter 10 teaches observances asseciated with the ancillary mantras, to be performed before stating a ritual to obtain supernatural powers. The observances are reminiscent ‘of Pisupata and Lakula ones. Chapter 1 is about the visualisation ofthe white mantra goddess, Para, bestowing immortality and securing welfare. Chapter 12 prescribes another visualisation of the same goddess, who this time promotes eloquence and learning. From verse 13 an alternative is given to this practice, which involves a vegetarian observance. Chapter 13 gives the details ofthe invocation of Yoginis at night in a cremation ground, who bestow supematural powers. The associated mantra is that of Pardpara; and some of the blood- “Then one should place the tree slab godess Apard,destryer of the pains ofthe humble swith a similar appearane'on the left prong ofthe trident Inthis description, whe we say let {from the point view of the mandala} we mean ight from the point of view ofthe performer of ve rill and "ight means infact Set Tin the Same vay). Tis female manu [1 Apa] looks like Pariparh excep tht] she is tavny-black One should place the beautiful dest ofall eters the Alphabet Gaddess, Malin] on the middle prong, She illuminates this world and i beauaful ike miliads of shining suns. Then one Should place at the op of her head crest (a eus] with eight petals and » pericarp. On the pericarp Of the lotus. one should place the beauful one-yllable godess, Pars, She 1 translucent] Ike the best crystal and she pours cut nectar al ver. She bestows well-being and sucess In this visualization, while Paris above the other goddesses as a higher form of the Alphabet Goddess reluted to the brahmanical goddess Sarasvati7, Parpart and ‘Apariare not aranged in a further hierarchy, but are treated as equally ferocious Kapilika deities. The equality between Pardpard and Aparais also suggested by their Position in the TriSdlabjamandala in the Siddhayogesvarimata and the Malinivjayottara, ‘where the lotses corresponding to these two goddesses are at the same level, lightly below the lotus of Pard.* However, there is threefold hierarchy, suggested by the 3 CF footnote 4 above This paragraph of the wansation is almost flly identical with Sanderson (1399:51) 1 As Pro. Sanderson has show, se Senderson (1993-51) ‘8 The two mandala almost identical, see Tanrloka and -viveka 3.155 1, The detailed description ofthe mandala ofthe SiddhayogeSvarimata canbe ound only inthe Taurloka, notin our msattribution ofthe colours white, red, and tawny-black. In both the SiddhayogeSvarimata and the lost Trisirobhairavé?, the hierarchy suggested by the colours ~ corresponding to the pan-Indian colours of satva, ras and tamas is further supported by the hierarchy ofthese deities inthe placement of their mantras. For in the vidydtrayanyasa, one is to place Para, Pardpari and Apari on the head (or te top-knot), heart and feet respectively.!? Moreover, this ‘ordering’ ofthe goddesses is reflected inthe scant remarks in the Malinivijayottara on visualization, where Paris mild, Paripard is alte ferocious but not frightening and Apard is tervtying.!" 2, Abstract ectypes of the goddesses The triadic arrangement can be observed in the classification ofthe abstract ectypes of the goddesses. These female spirits (Sakti) called auspicious or mild(aghorih), terrible (ghorih), and more terrible than the terible, or surpassing the terrible (Ghoraghoratarah or ghorataryahy!? are infact three kinds of forces that govern this world, Among the scriptural sources, it is the SiddhayogeSvasimata (2.23-31) that gives the most detailed account oftheir function. [The power of Rudra] is calle ‘yogesvai" and her form is threefold my dear. Now I wil ell you their division as they exis in this world. The powers / Sakis who bestow grace by clearing up the darkness of ignorance which resides inthe body ofthe bound sou are called “the Auspicious Ones’ they bestow Sivs(hood]. The Rudeas, governed by those Auspcious Powers and focusing thei minds on Sadiiva, release the soul fom its bondag’ [other powers / fatis] who ‘obstruct the way to liberation are known as ‘the Terible Ones'. The Rudras who are possessed by them perform creation and dissolution, they play inthe Body lke ehilren with clay bulls. Those [powers} who cause a downward flow (of the souls into lower levels ofthe universe) and gratify the soul, who make it indulge in objects of enjoyment and in the condition of being bound, obstruct the way to liberation and dhey are called "the Ones Surpassing the Terrible’ The Rudras ‘whose minds are empowered by them and who are governed by them, throw down and down, ‘those who ate given tothe pleasures of the bound soul Since the possession bythe tree (kinds ‘of] powers is always beneficent /Sivalness](Saikara), they are called the ‘Ones More Terrible ‘Than the Terrible’, the “Terrible Ones’ and the “Auspicious Ones’.!3 Thus, established in the Power of Rudra they ae to govern tis world 9 For he colors of te dies inthe Tirbhaiavaanra cf. Tanwaloka 31.115-18 ad Sanderson 1990:2:55, OC. Malnnayonare 837 10 Paspardinsvaripoarlzavamly mabibalm/icchirpadharkm dh kimcidugri na bhisapim “apatt vanaSrige tu Dianinkeapigan/ichaopadbaran devin praacrvintinm par dpyivann devin candrakoyayuaprabhi,(Mlinvjayoare 8. 24-14) "As Prof Sanderson has pointed out the two designations reflect two interpretation of the hed unit cite Aran 0) pornstar hoses Tesecod fe llowed by Kauoginya ad Ptapuasia 323 13s De Isaacson bas pened out ome, there a pun onthe name of Biv Sasara [Renin hee, who is identified with the tee Kinds of Powers. Athough the tree Pada of he Aghoramantca {ue not names of Siva, bt of Raa (8 the mana blow nana te udrarpeby a, ey are identi with Sakasi ths pastage,‘The origin of this triad goes back to a Vedic mantra, which was employed as the ‘mantra of the Aghora face in Saiva and Bhairavatantas.' Without suggesting any is of some relevance in the context ofthe Trika that Sdyana in his commentary atribetes sitvika, rjasa and timasa nature tothe three elements of the Vedic mantralS, for the same association is made in Tika scriptures and exegetical works as we shall see. ‘Although one could easily associate satta,rajas and tamas with these three groups of Sakts, the SYM does not explicitly make this connection. However, the Malinivjayotara, the Jayadrathayaimals and the Tantrloka add further triadic sets in association with these powers, as summarized in the following table:16 Aghor&h: auspicious / mild, Parih (Malinivijayot:ara), Iecha / Siva (ayadrathaySmala}, Srsti / nonduality (abheda) / knower (pramatr) [Tantrdloka 3.72.74, 3.103-104) Ghorab: terrible, Pardparah [Malinivijayottara}, Jaana / Sakti [Jayadrathayamala), Sthiti/ duality-cum-nonduality (bhedabheda) means of knowledge (pramana) (Tantraloka] Ghoraghoratarth / Ghorataryab: more terrible than the terrible / surpassing the terrible, Aparah [Malinivijayottara], Kriyd / Nara (~ Anu) [Jayadrathayémala), Samhira / duality (oheda) / object of knowledge (prameya) (Tantratoka} 3. Female Spirits ‘The SiddhayogeSvarimata is chiefly concemed with how to attain supernatural powers, usually with the help of female spirits called yoginis, dakinis or $8kinis. This topic occupies almost two thirds ofthe text, the remaining one third mainly describing ‘mantras and initiation. The way in which a Sdhaka isto perform these, is given in chapters 10 and 13. Firs, he is to perform an observance called vidydvrata or vidyaigavrata, associated with the ‘ancillary mantras. This includes besmearing the body with ashes, wandering in uninhabited places, laughing with boisterous laughter (attahasa), and reciting the appropriate mantras loudly. After this, he isto perform a ‘worship of the deities (yéga) with the erghapdtra. Tis is followed by the recitation of 1 Aghorebhyo ‘tha ghorebhyo ghoraghoratarebhya/sarvatah Sarva sarvebhyo namas te astu ‘udraripebhyah (Taitriya Bribmapa XS. and Maitryaniyasumhitd 29.10) 13 aghoranamako daksinavaktartpo devas tasyavigraha aghorthsitvikarven Sindh /anye tu ghord venogrih/apare tu timasatvenaghocdd api ghoratarah, Siyana ad lo. {eee Sanderson (1990.55.58)‘mantras (apa), accompanied by the prescribed fire offerings (agnikarya). Then he is to perform worship again and finally, invoke the appropriate goddess and / or Yoginis whose power he wants to employ. ‘The Siddhayogesvarimata gives details of numerous groups of female spirits (voginicakra-s) to be worshipped and it also describes various ways in which these spirits can be summoned. This latter must usually happen at night in the dark fortnight (krsnapakse), in a cremation ground, The Sidhaka must be naked, smeared with ashes, and he should lure female spirits by offerings of impure substances such as meat or the ‘perfect nectar’ (sidahdmyta — the mingled sexual fluids). He is to offer his own blood from his left arm (vdmaiiga) to tame these spirits In patala 26, we find a brief classification ofthese ‘acting yogis". Apart from distinctions such as goddess yoginis(devati-) and human yoginis (manusya-) there is a threefold classification within the yoginis associated with acts or deeds (Karmayoginis cor dakints): auspicious, frightful and lowly dakinis (Siva dakini, eudradakint and dakini ksudra), Moreover, there is another passage in patala 28 (verses 29-30), which seems to have a similar classification in mind. However, here the threefold classification applies not only to dakinis as above, but to all yoginis. They are associated with three different parts ofthe day, three colours and three types of supernatural powers. ‘These pre-eminent [yogini-s} areas white as the moon a the ‘ist part ofthe day. At midday, they are red like indragopakas [red ants} A the last watch of the day they are black and they bestow success in black magic. They [the white ones) give rise to pacification and welfare, {the ‘ed ones] to hypnosis and atiation, and (he black ones] to the faculty of arousing enmity and of killing The only simitar passage inthe Trika scriptures can be found inthe Tantrasadbhava (chapter 16). Although the Tantrasadbhava's Yogini taxonomy is based on the archetypes of the seven mothers (saptamairkab) or on an inflection of these mothers (casita, cumbia etc.) in one passage itre-lassifies these seven under three groupe: the first group is one of heavenly Yoginis of sittvia nature, the second group has Yaksa- _yoginis of rajasa nature and the third consists of Monster-(raksasa-)yoginis of témasa nature.!7 However, the Tantrasadbhava differs in that it does not attempt to classify supernatural powers on the basis of the three classes of Yoginis. "7 teigvabhavtchs> sma by eth sattvarjastimasih/brahminy (Jams tnthsindryts devia _svika ma /kaumdrtvagoavigor yakgasattvs varSnane/varahicandikraudrikulait ‘aksasiteakich>4. Classification of Supematural Powers On the basis of various passages on supernatural powers (siddhis) scattered in chapters 11-31 of the Siddhayogesvarimata, itis possible to reconstruct the three groups of siddhis which seem to be attributed tothe three classes of creatures.!8 1. sittvika_siddhi-s : well-being being well-fed (pusti / épydyana), expiation / pacification (s4nti)!9, (saving things} in case some disaster occurs (upasarge ‘samutpanne)®, conquering death (mrtyunyiaya), eloquence / poetic talent (Kavitva)?!, the ability tobe infinitely smal, big ete. (anienidigunt, Final release (moka), 2. rajasa siddhi-s : subjugating people to one's will (vasyay, attracting people (esp. women, akarsana)?3, going to the underworld (patilecaratvam), flying (khecaratvam), disappearing (antardhénany,pil-Sdai (he pil putin the mouth ssid vo make one invisible, gulikésiddhi), and a Siddhi with the magic wand and a bowl (siddhakasthakamandalu)™. 3. tdmasa siddhi-s, or the twelve kinds of black magic (abhicdras ) listed in chapter 24 ‘murder (marana), expelling someone (uccdtana) annihilation (jambhana), paralysing. (stambhana), benumbing (mohana), ‘nailing down’ (kT/lana), taking away someone's speech (vdcdpahara), making someone dumb (mdj-atva), deaf (badhirya), blind (andhana), impotent (Sandhikarana), changing one’s form (ripasya parivartanarn).25 ‘Although various other tantras also have a threefold classification of Siddhis (e.g, the Svacchandubhairavatantra, 2.145), the details are quite different. The basis of, classification differs, t00; or they tend to classify Siddhis according to their difficulty and the kind of mantra recitation one should apply. Thus the Svacchanda has mental 18 The ist have econsruced follows the loge ofthe Siddhayogesvarimst long the ines ofthe shove mentioned threefold patter. Nevertheless, mention mast be made of he fact that nthe ‘lassifcaion of Siddhis. although the pater is present is ot pefecy systematic or consistent: e.g atractng ¢hypnotiing people canbe classified under any ofthe thee eategories, depending on fne's aims, However, hik tha inspite ofthese inconsistencies the classification is present ‘hough not always expicily — inthe Siddbayogesvarimata 19"The first two ae associated withthe White Yoginis in SYM 28.29 cited shove and with the white ark eg. n 6.27 (where she i called dpyayanakar) and in 185. 207s is mentioned in 736d with reference to the white Pash 21 Conquering death and cloquence ae obtained through the Sichana of Pat 22 Tae ast two ~ cls ope Shand ia reese ne aegred he Highes Sidi in 9c 33 These two ae associated with the red Yous in 2829 24 The lat five ae lsied in 29.9-100 as Middle Sidi 25 29.10cd-11 alo includes varius kinds of subjugation of people to one's wil, which shows thatthe elasificaion isnot very consistent on this point.recitation for expiation or pacification (<ént, muttering for well-being (pus) and loud recitation for all the other Siddhis, which are categorized as black magic (abicfra). The closest parallel to the Siddhayogesvarimata is perhaps a passage in a later compilation of, various Siddhis, in the Indrajavidyasameraha. Although the classification of Siddhis follows the three types of recitation and not the thee Gunas, the text prescribes mental recitation for pacification, expiation and final release, muttering for subjugation and attraction of people and loud japa forthe ‘lowly’ Siddhis.26 5. Where the threefold classification fils “The evidence shown so far suggests a systematic arrangement of goddesses and Siddhis, where Par and her ectypes promote the ‘white siddhis', Pardpard and her ectypes the ‘red’ ones and Apari with her ectypes the “black ones. However, if we look atthe observances, the mantras and visualizations (dhyénas) associated with various superhuman effects, the whole tiple system collapses. In practice, itis almost exclusively the mantras and dhydnas of Para and Pardpard that recur regulary, and defying the logic ofthe postulated triadic system, iis Parpard and not Apard who is associated with all the dark Siddhis. Below are two examples to demonstrate the association of Pardpard with the dark Siddhis. The fist shows that — in spite of our expectations — Partipar is visualized for killing (mrapa) The second one teaches the visualization of an important group of ‘Yoginis, whose names correspond tothe dark Siddhis of annihilation jambhana), ‘benumbing (mohan), making someone blind (andhana) etc. The text prescribes that they are to be visualized with the same form as Paripara (not Apard) ‘One should recite Pardparh and look at her, who has ed eyes. Aer 108 recitations, one wil be able to kill immediately. (7.34) Destroyer, Supefie, Left (2), he Who Makes One Bind, and She Who Cuts (2; She Who Nails People Dowa, She Who Tames andthe Killer — [these yopnls] age always 1 be worshipped and svalized with he form of Paripard described above. (Paula 25)27 This disappearance of Apard — or, rather, her merging into Pardpard — inthe context of magic could explain the Pifgalémata's unusual attibution of colours, That, text, while keeping Pard white, attributes the colour black to Pardpard, while Apar is said to be yellow 28 26 cantik pausike mokse manasa japam Searet/vatyalrsav upamsu syd vacikam ksudrakarmani (petits p21) jambhant mohant vdrndsndhiot fkarrt evs ca/ kilt damaalcaiva tains pramathanitica/ _pysparoktaripena pia dhyeyas ca sarvadé 8 See Sanderson (4950: 52)Moreover, the Sédhanas of Para, aiming at conquering death, gaining wel-being, and acquiring poetic talent, seem to follow a very different pattem from that ofthe attainment of other Siddhis. For in these Stidhanas (Siddhayogesvarimata pata 11 and. 12), Parl is visualized asa solitary deity, without a retinue of Yoginls, as inthe Kavitvasidhana below. Listen, 0 Goddess, othe highest secret, which isto he protected wit cae, by which poetic talent wil come about listen tot attentively. After worshipping the dees propery, wih thet ‘own forms, making eff, and after making fit offerings as prescribed one should do the visualization, (One should visualize Pard with her ox form, iting on Jtus inthe at, with the book ofall knowledge inher left hand, o Beaunifl One, and hulding a beautiful, heavenly rosary of erystal inher right nnd. One iso vsualire a guland on her neck, a garland of heavenly beauty, made ‘up with betds which are round ike the bus ofthe kalba tee and which shine forth like fre. “This garland reaches down to her feet and s spotless as crystal all over. One should then sale her as pouring out the divine nectar of minority, inthe middle of a Kadamba grove One should see her pouring forth the nectar ofall knowledge in great Moods and one should see this nectar enter in one's mouth, and that ono el hs the same fom. After this, the best of Stahakas should visualize that this nectar comes out of hs mouth asa flow of Sistas. fone has done this visualization, he will be able «u produce fascinating ornate poetry within a month He wil be a eacher of al doctrines and ater six months, he will be able o produce Satras himself. He will know all the sciences a the fri ofthe Myrobalan inthe hand (clearly, as if they were self-evident truth) Whatever has something to do with words and whatever isto be known ia this world, willbe his, bth s ois formulaon and content. (12.2-12) (On the other hand, forthe attainment of other, more violent, Siddhis the Sddhaka ‘must follow a very differen: procedure: he should summon Yoginis in a cremation ‘ground in the way described above. In this context, itis the 2arpard mantra that he should recite. ‘The following passage also shows that in this later case, the actual bestowers of Siddhis are the hordes of Yoginis, and there is no elaborate visualization, unlike in the case of the Parisidhan: (On the 14th day ofthe dark half of the lunar month, after fasting fo tue nights, one should go tothe eremation ground without companions. Naked with the top-knot vatid, facing Nor with eect body, the mantin should recite Pardpara fearlessly. The high-ainded hero concentrating and with tranquil mind, should perform this ull Yoginls gather together on all Sides. Seeing their various (rightening forms, he should not be afraid, ne should mentally recite the female manta (Pardpar}.(13.11-14) ‘The contrast between practices associated with Pari on one hand, and withthe other goddesses and Yoginis on the other, suggests that we may have two entirely ferent sources for these practices. Paras an ectype of the brahmanical goddess ‘Sarasvati isa solitary, mild and ‘orthodox’ deity; in patala 12, the opsional observance (vrata) associated with her involves a strictly vegetarian dit with fruits, roots,vegetables, barley-meal and milk products2® Altogether we may say that this deity is associated with brahminical leaning, welfare and orthodox purity. On the other hand, the cult of Yoginis is nota cult of one or two specific goddesses (Paripard and Apari are just two representatives of them), but one of innumerable female spirits who indulge in impure substances and are associated with witchcraft, Although some of them may take up mild forms in the first watch ofthe day, their main occupation is. ‘getting intoxicated with blood and amyta in their frightening forms. Moreover, if these Yoginis are classed or mentioned by name, they are divided into eight and not three. This eightfold division is present at various levels. The female mantra with which one invokes the Yoginis the Parpard mantra, is divided into eight segments, each representing a goddess: Aghota, Paramaghor8, Ghorardp, Ghoramukhi, Bhima, Bhisana, Vamant and Pibani.™! These eight Yogints of Pardpara fare also placed on the body from top to toe, similarly to the three main mantra goddesses®, The Alphabet Bhairava Sabxtardsibhairava) has a retinue of eight ‘goddesses, each representing a consonant group: Brahmi, Vaisnavl, Mahesvari, Yamya, Visavi, Kamamoti and Aghoresi The typology of worldly Yoginis in chapter 29 —from which a substantial portion is certainly lost ~ i also based on the Finally, while the mapdela drawn forthe Samaya initiation is empowered by the three ‘mantra goddesses placed on the three prongs of a trident ‘n the middle, the Initiation ‘Mandala has an eight-petalled lotus inthe centre, representing a group of eight. goddesses stating with Aghori.35 ‘The Siddhayogesvarimata seems to try to fuse the system of eight Yogi “Mothers and the cult ofthe pure goddess Para jin its cult of the three goddesses by placing two representatives of the Yoginis as Pardpard and Apara into its triad, thus 2 Siadhayogesvarimac 121568163 30 See Panis 22 teaching the mixed nature (misakam pan) of Yogias 31 See the mana given im 3.23-39 321 reer to these eight Younls as pars of Prdpats boy (yee). Sec eg SYM 726-27 3 Ses SYM 161-83 24 The fest four are named in SYM 29.25, 32,39 and 4S, Yim is ot mentone, but can be ‘dented fom te description ofthe last Youn, whose signs ar the Fang (argu) and the Rod (anda), which are the atibutes of Yama, CL. also dad being an epithe of Yama (punishment), and the compound Yamadansi in Apte. The enaiing Yopn-aniies can be “Malintjyouara 3.14 Mabes! Braman ata Kaun Vaya ah / Ain ‘Yop cu td aith// However itis so pesible tha the remaining Ove were called Visa Karpamol and Aghor just as inthe cae ofthe consonant dees of Sadarafbairava mentioned hore. 33 this canbe infeed fom SYM 6.55 and 8.10. This manala may also have ardent wit tbe ee fodder, butit snot lea rom he very erp esrpon in chap 8attenuating the fundamental differences between the solitary Pari and the Yoginis.36 ‘The threefold classification in fact serves as avery good device to merge the two cults by superimposing the classification of Gunas on them. However, even though ths triple classification (bhedatraya) is reinforced at several levels, it cannot conceal the ‘underlying bifurcation, the contrast between Para and the Yoginis ‘The conclusion I would like to draw on the basis ofthe above consists of four elements 1. Tpropose the hypothesis thatthe cult ofthe Triks probably evolved from an cexoreistic base, dominated by the worship of eight goddesses as a variation ofthe cult of the Bight Mothers, in which perhaps the pure goddess, Pard, was also a leader of the eight impure Yoginis". It is also conceivable that Para had a separate cult, for which the ardtrimtika may be an additional testimony. This system was then re-arranged by the superimposition of the threefold classification borrowed from the Sdmkhya and related traditions 38 2. Through the adoption of the iconography and powers ofthe orthodox goddess, Sarasvati, into the figure of Para, an element of the orthodox domain was also colonised and combined with the features ofthe popular goddess of illnesses®. (Cr. Para's double role as Sarasvatl to bestow leaming, and as a godless promoting health in separate Sédhanas.) This adoption of the orthodox goddess in tur served for Kashmirian brahmin exegetes as a basi to colonize the root texts ofthe Trika ‘from above’. 3. In the Siddhayogesvarimata, this colonization of a brahmanical deity and the {identification of the Yoginis is important primarily for the S8dhaka, For it isthe Sidhaka's aim to control all forces of the universe, including the orthodox domain with its Sistric knowledge and purity. At this point, the cult was inclusivistic for the ‘Sadhaka’s sake. On the other hand, when later exegetes colonized the texts of the Trika, the process was reversed: the exegesis was done forthe sake of brahmins seeking liberation (mumuksu), and it was the inclusion of Pard-Sarasvati that made the cult particularly susceptible fora brahmanical colonization which saw Pard-Sarasvati as the 36-The fat that Par has a Yogint tinue does not change the fact that itis er solitary form which dominates in visualizations. Her Yogincaka is formed probably to “interae’ he into the Yogint hordes 37 "This a role reminiscent ofthe god Aiyant in Tamil Nadu as described by Dumont 1953, 3 iis, of course, also possible o presume that a adiion with a threfold clasification adopted the he eight Yopnts and that of Park, ot that these cults merged in some oer way. The pists athe ide arangement nt ony ~and nt neces he it and Frees coset 3 tn his ater role, her function may remind us of he popular goddesses of illnesses such a Stal (also associated withthe white colour and cooling) or Mryammap although no direct relation can be shown between them,centre ofthe eult, and ‘absorbed! the texts through her in order to control impure forces through the pure. In this sense, the story ofthe SiddhayogeSvarimata and the cult of the Trika i a story of competing forces attempting in turn to conquer each other's domain. 4, The Siddhayogesvarimata represents an eatly stage of the cult, where certain inconsistencies still reveal that there is a twofold classification of the Yoginis and Par (bhedadvaya) behind the threefold one (bhedstraya). The Malinivijayottara, however, seems to make a further step towards systematization, and there the threefold classification seems to be perfect. Therefore, I propose that in the Milinvijayotara, the visualization ofthe deities (Pari ~ mild, Paripara ~ slightly ferocious, Apard ~ frightening)! has been adjusted to accord with their places and hierarchy in the placement oftheir mantras (mantranyisa, Pari on the head, Par pari on the heart and ‘Apari on the feet) Inthe Malinfvijayoztara, the goddesses exist almost exclusively as mantra deities and they do not figure in Sidhanas or yogic practices (dhdrand+-s), Just as the Malinivijayottarahhas a more systematic way of presenting the parts of the alphabet ‘goddess Malini (from top to toe’, from nato pha, nédiphdnta), it seems to have attempted to adjust a detail ofthe visualization — making Paripari just a litle ferocious, so that it should not contradict the triadic arrangement. 4° On ‘purty and power, sce Sanderson 1985. 4 Malitvijayonara 872-74 42"This has been demonstrated by Somdev Vasudeva in lecture, in May 1996, at All Sous College, OxfordIll. Aifa language ‘The grammar below is an attempt to describe the way in which the language of the SYM differs from classical Sanskrit. The imegular forms are called Aisa ~ characteristic of the language of Lord Siva (16a) ~ following Ksemardja's usage of this word in his commentary on the Svacchandatantra.! White one may hesitate to agree with Kgemardja that these must be hallmarks of divi. style, they perhaps should not be considered simply ‘erroneous forms that would make a learned isan blench’, or ‘grotesque solecisms' either This is a language with its own rules, whose basis is Sanskrit but hich shows influence from Prakrit anc! Apabhraméa. Aithough Sanskrit is a convenient point of comparison to describe what can be called Aiga it does not mean that Aisa is simply ‘erroneous Sanskrit’ —-just as a pidgin language is not erroneous English or French, even if it may best be described in comparison to English or French. In the description of Aisa contrasted with Sanskrit it must be pointed out that while standards of morphology in classical Sanskrit are quite unambiguous due to their importance in the indigenous tradition of grammar, the rules ofthe syntax and lexicon show much more flexibility. For standard Sanskrit syntax, the point of comparison below is Speijer (1886), and forthe lexicon it isthe three commonly used and most ‘comprehensive bilingual dictionaries (Bothlingk and Roth, Monier-Williams and Apte). Moreover, there are certain forms and ways of usage which may violate rules of standard Sanskrit to some extent, but can be considered widely accepted in epic and purdnic literature. Although these deviations are not exclusively characteristic of Aisa, 1 have tried to include them in this grammar rather than to accept them in the text silently, since there is no comprehensive critical work available onthe iregulaites of purdipic ‘Sanskrit, Ihave occasionally indicated parallels from Meenakshi (1983), but it must be remembered that this work on epic syntax cites examples only from the Sabha ark! Udyogaparvans and from the Ayodhya- and Yuddhakanda ‘The grammar below includes almost all forms Ihave labelled Aisa in the Notes on the constitution ofthe text and Aisa forms. However, there are some forms which occur in passages whose constitution is very uncertain, and a numberof these have not been included, since they may not be characteristic of the text in general. On the other hand, certain imregularties which have been rejected as scribal erors are mentioned, for although I have emended them, itis conceivable that they formed part of | Kyemarajis wage ofthis word was alo adopted by Ara (1988), who wa llowed in Goodall (4998: Goodall also quotes a number of passage rom Array (198101) where Kyemarja uses the term Aika lernating wih Aivara and Arg characteristic of se). 2 Expressions quoted from Goodall (19982) footnote 138‘The principle I have followed to determine iregularities is a practical one. Since the text is very badly transmitted and we have only two late manuscripts, the frequent ‘occurrence of an iregularity was not considered sufficient forthe form to be taken as Aiga. Thus, any irregularity needed to be justified with either the exigencies of the metre ‘or with similar forms or principles in Prakrit, ApabhramSa or Buddhist Hybrid Sanskrit (BHS), Occasionally, Ihave also accepted forms which were not required by the metre, ‘but which were probably coined on the analogy of other irregular forms thatthe metre required elsewhere. However, there remains the problem of the frequently altemating masculines, neuters and unmarked nominatives without any metrical reasons, which can be original Aiga as well asthe results of seribal eror.{n these cases, Ihave decided to correct forms which were likely to have undergone seribal corruption, i. irregularities atthe end of a line or missing / mistaken Anusviras or Visargas. In cases were the editorial correction would have involved too much change in the text without any apparent paleographical justification, I have retained the inegular forms as Aisa, such as -0 endings for am. ‘Two important objections may be raised against the above principles. 1. Firstly, it may be suggested that other texts with Aisa forms should also be taken into account, This has been done to some extent, fr there are frequent references to the collection of Aisa forms in the critical edition of the Kubjikmatatantra by Goudran and Schoterman, to parallels pointed out by Prof. Sanderson? and toa few additional parallels listed by the editor. There are two reasons why the numiber of references to ther texts is relatively small. One is that there are very few critical editions of tantric, texts with Aiga forms, which is perhaps partly due to the fact that this branch of literature was considered worthless for along time both by Western and by some Indian scholars.* The other reason is that every text differs from the other as to what Ai forms it prefers and how often it uses them. To take over principles or forms from other 2am ery grata to Prof Sanderson fr pining out several separ (orm and pales which occur equely in manuscripts offer texts, hus poviding convincing evidence forthe acepance of these Forms. Is hoped thatthe Aa roma he hus comped il be published soon to help future gators of ante tent Epihough the situation improved especially afte he itis, one silo subject On te traional view, Montilla induction his ditonary may be ct‘ seatched everywhere for good MSS. ofthe most popular Tana. Everywhere {was tld tha the RadrrySmala Tantra washed in mos esteem, But fe eateful xamaton ofits comms { Jcided ‘hat it was nether worth editing nor waning. Another problem shat though some Siddhantika texts which conan numerous Aisa forms have ben eid, the dion follow th Southern feeenio(), which represen grammatcaly purified versions of the sae et (sir to the ‘Manstoaratss Souther ecenston). CI ete list of Als differences na Nepaese manuscript ofthe Sirdhtrizatiklotara compares to NR Bhat eiton compiled by Pro. Sanderson and repel a ‘Goodall 1956s)texts uncritically$ would mean to change the language ofthe edited text, which would ultimately also distort our view on Aiga in general. 2. Secondly, it may be objected thatthe edition represents a purified text, with corrections too numerous to be acceptable in all cases It is true that the edition probably sives a slightly more correct text than the original presumably was. However, the acceptance ofall errors a original would result in text which would probably be much ‘more irregular than the original. The underlying problem is ~ as Goudriaan and Schoterman (1988:57) observed concerning the text of the Kubjikamata- thatthe text does not use irregular forms and sequences in a systematic way. ‘Tt happens that a certain phenomenon is backed by the whole manuscript evidence on textplace X, while itis almost absent on text-place Y; or. its strongly backed almost everywhere, but entirely absent on one place without any motivation for such a deviation discernible’ Policies of a critica edition need to be consistent and systematic, but this ‘means that the editor will inevitably need to systematise the original to some extent, Which had no such principles or pretensions, Since fully correct and justifiable reconstruction ofthe text is impossibe, the least distotive method should be applied. ‘Yet, in spite of the standardisation inthe critical edition, the reader ‘should remain aware ofthe floating state of affairs inthe sources’ (Goudriaan and Schoterman 1988:57). I, SANDHI AND PHONOLOGY A number of irregular Sandhi features occur in order to preserve the correct metre. 1 have not included here Sandhis which occur due to the loss of nominative or accusative ccase-endings; these are discussed in the section on morphology .7_ However, all kinds To adopt readings of ter manuscripts of oer Tanta would mostly mean unrital adption tr one should st alo eit the source texts. Tis, af course. paces a vicious cite: pares would be needed rom lt text forth eiton, bt oe would ned eriteal edn of these paral ‘The analyst of Als forms presented ere an empl break thisce, which obviously means {hat hs grammar wil surely conti more eros snd misintepretations than works to ee om the same vbject rs tobe hoped ths sther cial etna of Tana te come ret whateva his been mistaken here this ean be example with the vecumences of Sanskrit “yuan 29.50 bth manuscrips have the la form “yaya which is obwiusly nt requed bythe mee, bt hasbeen adopted becane the same phonetic change occurs in ther words (whe the change required bythe metre, ¢ ‘samupetam > samopetam) In21-410, however, only Nreads ths Ais orm, and Ihave adopted D's eet sty It maybe jected that Dasa pure version here but hen Why does it have the ‘Akan 2.50" In any ese the llerating forms ae ot motivated by the ec to conform to the ‘ie, and they pos represent original ncoitncy ine tel. Sice we have oly (wo Imanssripts, such cass are dificult solve ina atfactry way. (Do ancberserbe before D may Also have corrected the Ait form atone place while forgetting abou the conection the oer tecurence) 4 Gondrian and Schoterman discuss this phenomenon under the title oepular conto of vowel (1988:6-62) inthe section on Sandi. Thi st man includes words wih lost ase-endings, xcept 1), and g which involve double Sands. Two cases sted in) may alo be interpreted as masculine pronouns forthe feminine. Since these occurences cover very diferent suuctures and may mat simply beth ese of regular Sanghi, have inced them i th discussion on morphology an stax xviof lengthenings which have no apparent reason other than to preserve the metre are included in the description of Sandhi as well as the elision of final -t, External and internal Sandhis are not treated separately. ‘A Lengthening and substitution in order to obtain a long syllable metri causa 1. Lengthening of -a- pcvamukham (for parvamukham) 6.16e. simira (for samira) 8:32a, ahitd® (for rahita®) 22.4b. For the frst and third examples, see also Pischel §70 on the lengthening of the final short vowel in formation of a compound in Prakit remarking that in many cases the Jengthening is purely metrical. Cf, alvo Pischel §75. GGoudriaan and Schoterman (p. 57-58) note numerous examples forthe lengthening of the short -a- in certain phonetic environments, most frequently in composition after stems in -i and 1, and conclude that this phenomenon is connected withthe oscillation between feminine stems in -/-i and -yi. However, the lengthening of -i stems to -yAis not common (cf, also 1.B3e below, therefore I have not accepted forms such as Saktyntam or aghorydyastakam although they are fairly common. This is one ofthe points where my editorial policies may erroneously favour correct forms. Ihave also conected ahutyastakahomena to ahutyastakahomena, which may, however, be imerpreted as ahutya + estaka-homena, but this seems unlikely in the contexts where it occurs. CF. 6.56a, 8.26¢ My policy is in accordance with BHSG §3.5ff, where itis noted that the lengthening of 2+, which happens most commonly to the final -a including stem-final in composition, is almost wholly due to metrical reasons. 2. Lengthening of -i- Ahuttbhib (for ahucibhif 8.26, 7.12b. ‘See Pischel §73 on the lengthening of short vowels ia medial and final syllables exclusively on account cf metsical consideration. See also BHSGS 3.124 3. Substitution of -u- with -0- samopetam | .4e, °samopetd 18.21 (for samupeta-) ‘The following has been accepted as Aisa, although the substitution is not required by the metre: *sdyojya® for “sdyujya® 29.50, Ff. Pischel §§77-78 on vocalic elevation in Prakrit deviating from Sanskrit. Also see the numerous o vajrinaitydi -> varinety&di) 46a (N.b. mantra giving) ‘This may be treated as elision of -i (as in Gouriaan and Schoterman 1988:60), which is supported by a somewhat similar phenomenon in Prakrit, ef. Pischel §143. _svasthanaiva (svasthane eva —> svasthdna eva ~> svasthinaiva) 632b ‘mitarénya (matardh any —> métard anyah -> métarényah) 8.294, samsthitanya (samsthitah anyAh —> samsthité anya —> samsthitanyah) ‘hasamtoccair (hasantah uceair -> hasanta uccair ~> hasantoccair 10.74 (N.b. a5 emended) 8 Ina leer dated 21 May 1997,phalguneti (ah + iti-> a+ iti —> eti)21.260 (Nba proper name)? bhairaveva (ah + iva > a + iva—> eva) 21.474 (N.b.: a proper name) ‘The last five examples may be considered examples forthe omission of case endings with :uch underlying principles asthe Apabhraméa zero plural marker for ‘matar-(Tagare §84) and the application of regular Sandhi. (ef. Goudriaan and ‘Schoterman 1988:69), Double Sandhi is not unexceptional in the Ramayana either, as, ‘Satya Vrat Sdsti(1964:197ff) points out Lack of Sandhi / irregular hiatus Below I list only hiatuses within the same pada. As in epic Sanskrit, a hiatus between pidas was probably not considered a fault. See Satya Vrat Sasti (1964:194) on Padancayatiin the Ramayana have not listed all of the examples, but the ones cited show the verious motivations which can result in this Aiga solution. 4) To keep the form of a mantra (and to avoid hypopettism): hrdayaya it 4.4 ) To avoid hypometrism: mahé-oghaib 12.82; manasa ipsitan kiman 16,50a ©) To avoid an unmetrcal pida with the second and third syllables short: ‘mantrite udake sati 16.376 ‘Type b) has been noted as Aiga by Ksemarija ad Svacchanda |1.95¢: K:: maya aharmukham iti sandhyabhavas chindasah ‘See Goudriaan and Schoterman on the occurrences of iregular hiatus within a pda, p63f¥ and BHSG§ 4.51ff remarking that hiatus in separate vids is very common in verse, but may have been Sanskritised in prose, where they occur less frequently. For the same phenomenon in Pali, see Geiger 67. For examples for in-pdda hiatuses in the Ramayana, see Satya Vrat Siti (1964:19240. D The -s + t- Sandhi (and -§ + ¢-) or the ‘cosmetic masculine’ ‘The underlying principle of this Sandhi is that masculine and neuter forms are interchangeable (cf below). The Sandhi applied is that before a word starting with endings withthe dental sibilant are preferred tothe Anusvara endings. Examples are: dhydnanyésam samakhytam sakalas tu vardnane 6.32cd (N.b. nyasais treated as neuter in the majority of cases) sthanaSuddhim dravyasuddhim bhataSuddhis tathaiva ca 6.6ab 1d ndsdputas te ca 16.16.14ed 9 This can also be explained as corectSandhi with the loss of the nominative ending, which was the interpretation preferred in Goudriaan and Schoterman (1988:69). They attrbus di lack of nominalive endings toasty in sandhi matters’ and also note at'a special preference for stem-forms i found in [Proper names inthe context of enumeration’ CF. also the elision of nominative ending before ii ‘emarked by Prof. Sanderson and cited in Goodall (1998:xvi)In 16.13-29 the tendency is that karais treated masculine before tu, and neuter in most other positions: fakdras my 16.194, flaklras tu 16.214, fakdras tu 16.22a, hakiras 16.29¢; cf. also tvaggatas t yakiro vai 16.27c. A similar phenomenon i thatthe -S ending is preferred before a ca in the next word, such as: pprakrtim purusam caiva Sivas caiva 7.7ab (purusais neuter as the non-final member of thelist) adbikram tatha bhogam layas caiva 7.1 ab (adhikiray is for adhikras, since itis also non-final) ‘As the examples ofthe -Sc- Sandhi show. the -st- or Se~ Sandhis occur regularly at the end of lisis of three, where the last word is followed by ca /tatha /tu. Although this Sandhi seems regularly employed in lists of three itis not applied consistently everywhere, e.g. in the case of Kara Jn Gousriuan and Schoterman this phenomenon is treated as a cumulation of various scribal errors under different subsections in the chapter on “pseudo-sandhi* inaccuracy in the writing of word-endings (pp. 52-53), or as anomalous uses of the cases (pp.87-90). However, since they are probably not simply seribal erors, and their use is motivated by a phonetic environment, I have labelled them as special Sandhis. Cr. also Goudriaan and Schoterman on the seribal error of adding -s before a t- or - before , which addition may be motivated by the same phonetic environment, but does not ‘concern the above listed cases, where the endings are not added but changed. 1 propose to explain this phenomenon as an attempt to Sansritise the text. Sanskrtisation here does not mean the rendering of the text more correct according to the rules of Sanskrit grammar, but thatthe authors may have attempted to make the text sound more Sanskrit in accordance with their idea of what Sanskrit should sound lke. ‘This meant that they perhaps tied to increase the number of Sanskritic endings ie create nominal endings which did not sound Prakrcc such as -am or -0 would. This resulted in an inereased number of masculine endings which are made conspicuous by the application of Sandhi such as -as +- and -af + c-. These masculines are employed even in places where their use isnot fully justified ~ since everything else i treated neuter ~ and thus their occurrences in these environments seem rather cosmetic, E The Prakritic -o ending In some cases I have accepted an inegular-o ending (standing for -ah or -ai) before voiceless consonant, since changing it to regular Sandhi would have required to much interference with the text. prsto caiva11.2c, tritayo samayi 12.15bThis form is supported by Prakrit -oendings and it seems to be much more frequent in NN than in D. Notable is N's frequent se of tato for tatah /atas as opposed to D's regular Sandhi in most of these cases. Cf. Pischel §363, and BHSG §4.38 for o before ‘voiceless consonants and pause. F Treatment of hiatuses 1. Common hiatus fillers ‘There is large number of occurrences of ca, wand hi clearly functioning as hiatus- fillers. This phenomenon can also cause problems inthe correct interpretation of some lines, for one needs to decide which occurrences are meaningful and which are fillers. Only one case is mentioned here, where the repetition of ca in the very same pda shows that one ca is obviously superfluous forthe contruction of the sentence: catviriméati cinyas ca 8.3 c, See Goudriuan and Schoterman pp. 64-66. ‘That the use of conjuntions and eva isnot regular has been also remarked by Ksemardja ad Svacchanda 3.33¢ Atmayige kre caiva dehasuddhib prajayate K: co hyarthe.evasabdo jayatesabdad ‘anantaram yojyah. (N.b.: here Ksemarija tres to upgrade the style substantially by reinterpretng the filler caiva) 2. The intrusive -m- ‘This hiatus-filler is listed with numerous examples in Goudriaan and Schoterman (65- 66). CF. also BHSG § 4.59 and for Pali Geiger §73. Here I shall give only one example for each type 8) To avoid a correct hiatus: anyasmin kula -m- utpannd 22.6¢ ») To avoid Sandhi in a compound: ta-m-diné 16.42b (cf. BHSG $4.60) ©) To avoid an incorrect hiatus: virunidisi-m- asthdpya 5.Sa (this i the most common ‘motivation, cf. BHSG §4.59) 4) Inthe majority of cases it is difficult to judge ifthe -mis a hiatus filer or an iregular neuter ending. Cf. e.g, trikarddrim iva 6.18, jabgham ud°16.22b. It is possible that the gender was simply considered changeable in order to avoid a hiatus.!9 Itis also possible to accept -d- as ahiatus-filer, following BHSG §4,64 (cf. Goudriaan and Schoterman p.66!!), but Ihave found only one occurence which may be in tis category: tasya siddhir vardrohe yad idam gopayisyati 22.40d: where one ray also understand the use ofthe neuter yad forthe masculine yah. 10 There isa somewhat similar phenomenon inthe use of possessive pronouns in French. The ‘masculine form replaces the feminine before feminine nous in order fo avoid hiatus, eg. mon ‘troduction fr ma +introduction. Spoken Brish English also has a similar Sandhi with the so-called ingusive R, although there is ofcourse no geader confusion: aw and order pronounced as law -- and ‘zder onthe analogy of words in which the final writen is pronounced in intervoeaic position. TT Both grammars emphasize the rarity of -d- as hats-fille, and Goudraan and Schoteran find the evaluation ofthe form dificult.Il, MORPHOLOGY A Formation of words 1, Formation of the feminine “The fist examples from a) toc) may be de tothe influence of vowel-discolouration in ‘Apabbraméa (see eg. Tagare §18 (3), although ~ as Tagare mentions in §37 —it is not ‘common phenomenon in Apabhramia 8) -aké for ~ka *sidhakii 3.204; igulakil ¥6.2Se: asesaphalutiyakdm 16.45. Cf, Goudriaan and ‘Schoterman p.68. This alternates with correct tori. b) ant for int ‘As Goudriaan and Schoterman (p67) point out, the hesitation between these suffixes goes back to the Vedas (Atharvaveda 4.12.1). CE. Sprapasanim 3.394, °vardhani 6.94, °vinisenim 6.230 ©) -atdfor it In fact, for -ir, since the masculine stands here forthe feminine (cf. 2 below) Playatipi for palayitet 19.166 4) -ifor~ikd (or pethaps loss of the -ka ending metri causa) andmi® for anmiki® 6.17 ©) -istem forint ‘yogibhib for yoginibhib 16.74, 18.28b, 21.1e, yogi for yogin? 22.14. Cf. Goudriaan and Schoterman p.68 observing that yogi seems to serve as a feminine of yogin. Cf. also the confusion between -i/-in /istems discussed in BHSG §10.1ff The underlying principle is probably the confusion of stems which all become -iin Prakrit and ‘Apabhramsa. For Prakrit «in > -iand ~i/7 > -i see Pischel §405 and §384 respectively. In addition to this, Apabhramsa also changes i to i/-a, see Tagare §18 (3). See also the following examples adduced by Prof. Sanderson'2: Netratanra 20.16 yogiyogena, where the sense is yoginiyogena, as Ksemarija recognises ad 0c: yoginindm yogena, also Malinivijayotiara 19.264: yogi yogikule kull 8) Ghoranmukhi for Ghoraamukha According to Panini 4.1.58, proper names ending with °mukha or °nakha must take the feminine in -. The SYM consistently uses an irregular form to denote the fourth tmantra-goddess of Paripara, Ghoramukhi, and the name figures inthe same irregular form in the mantra asa vocative (Ghoramukhi) See .g, 3.23-38 and notes to the 1 Ina eter dated 01/06/1999,translation, The same phenomenon has been noted in the Ramayana by Satya Vrat ‘isl (1964:203) concerning the name Sdrpanakhi/. The confusion is easily explicable as a Prakritism, since stems on -Fand ~tend to converge in Prakris. 2, Lack of feminine formation ‘This usually happens with consonant stems and with forms ofthe nomina agents; and the underlying principle is that due tothe vocalisation of consonant stems and the ‘general assimilation of stems to ~2, these are probably felt tobe feminine forms. Masculine -as tem used for the feminine -as (all examples are for ®manas). This must be due to the confusion with thematic feminines in -d, and the phenomenon should in fact be categorised under thematisation of -as stems: but since itis perceived in standard Sanstait as a masculine form, Ihave included i here: ekagramandl 19.2b, Srpusvaikamand 31.5a ato brstamand devi29.1a Cf. Pischel §410 (Apsard for apsaras) Masculine instrumental of -ant stem for feminine -antz: Bhavati for bhavatyl(m.c.) 29.12a Cr. the same kind of instrumental (mahay= mahatd ) used for the feminine in Prakzit, Pischel §396. 3, Irregularities of the abstract noun formation in -tva ‘The suffix -tva is not used very consistently: (m.c.) sakalatve must be for sakale 7.16a, for the immediate context has both sakale and niskale twice and laye once. On the other hand, the abstract suffix is omitted in 2.244, which has siva®for sivatva®, Cf also bhairavam for bhairavatvam in 22.124 For other examples fr the lack of -tva elsewhere, see Goudrisan anc Schoterman p67 4, Miscellaneous irregular formation of nouns to adjectives Several of these forms may be explicable with the AiSa flexibility of using nouns for adjectives and vice versa, ~ Grade confusion for aisényam 6.304, taipkalam for trikalam 6.49a The former iregulatity may be partly explicable with the disappearance of the diphthong -ai in Prakrit, ef. Pischel §60. The later may be parallelea by the occasional elevation ofthe first vowel in synthetic formations in Prakri, of §77. ~ Raja / ja for rAjas: ejavarna® for rajasavarpa® 16.33c, rajam 21.184 ‘The last two examples may be seen as displaying the general tendency of shortening ‘words with the elision ofthe last syllable, 5. Comparative / superlative confusion. -kaniyas for kanistha (in fact, for kanistha, while kaniyasis thematized to kaniya/ kaniyé), kaniy&iigustha 6.172. This confusion itself may not be Aisa, for Abhinavagupta uses Kanlyasto denote the litle finger in Tantroka 8.95, but this ‘meaning or usage is not recorded in dictionaries. The phenomenoa is paralleled in Buddhist Sanskrit, where -tara is used as superlative suffix, BHSG §22.40.B Nominal declension 1, Lack of endings Nominative and accusative endings are often elided ifthe metre requires Sandhi without them, kanlyaigustha for “am 6.17; nédeyavidhinimisam 6,460 vidyavrata samlrabhet 10; arake cakravegi tu mohany ante tu circayet for accusatives) 19.1 Lab; tasya véksiddhi Jityate 19.184 Cf. BHSG §8.22 and §§8.31-8.35; Pischel $364 (both for Prakrts and Apabhampéay; Tagare §80 (use of zero as a termination of the direct case) Goudriaan and Schoterman p.. (Also ef. confusions beween nominative und accusative mentioned below) The tendency towards having only two cases, a nominative/accusative or direet and ant oblique case similarly to Apabhramsa, is discemile with almost all stems. This phenomenon will be treated in general when discussing irregularities of syntax. 2, Stems on -a and - Teshould be noted here that al other stems tend tobe assimilated to -2/~Astems, usually by extension, For examples ef each stem below. 4) Stems in -2 are usually declined regularly, except for incongruence of gender and number (¢f. syntax and lexicon). However, the omission of the syllable -nd- in the genitive plural does occur: aes (for agesdnim) eva mantrindim |.13a, Cf. Goudriaan and Schoterman p.69. Ksemaraja also notes this applied to a feminine -dstem as Aiga ad Svacchanda 4.3844 .- sparso yadvat pipilikiscm?> K: yadvat pipiliketisamcarantindm pipilikinim ivety arthab, piplikacm> ity atra n&Sabdasya lopa aisvarah. (I is not clear from the edition if Ksemaraja read pipilika or pipilikim for pipiikinam) For this contraction, of. BHSG §8.124, assuming that it is formed on the analogy of ‘consonantal stems. ‘An isolated case of turning an -a stem into an ~in stem in the instrumental metri causa has been noted: visargind for visargena 21.45» by Stems on -d usually have “Ayd instead of “ayd in the singular instrumental: mudrdya viichitapradam for mudraya v° 6.164 , pardyd for parayé (+ sa) 7.16a, (+ tu) 7.238 tryaksardya (+ va) 7.80, pardparaya (+ ca) 7.8c, vajramudrya 21.14c. This may be explicable asa tendency towards having one oblique ending in -4ya Cf. Goudriaan and ‘Schoterman p.70. Tagare §89 BHSG §9.48, 3. Stem on -i (and -u )xxvii 8) Gender confusion: there isa tendency to treat all stems in -ias feminine, although this is not applied as a rule. Cf. BHSG §10.9ff; Tagare §94 (p166) remarking that the declension of masculine -i/-u stems is more influenced by the feminine tems than by that of the masculine -a stem (but masculine and feminine -stems are wel differentiated iin Prakrit, the latter having the same terminations as -d stems; cf. Pischel §§377-388) ') Case confusion: as noted above in case of stems, a tendency towards having one oblique case can be discerned: paiktyd 3.23d (instrumental for the genitive) ) Creating -a stems from -i stems. This is done in two ways: by extension “mdrtayah for “mdrtib 10.10d and “paiktayab for paiktih 16.15d (occurring in BHSG §10.7 the stem rsaya- used for rsi, for which cf. also ssayo-guptam in compound in SYM 32.13a with the BHS maharsaya-siddha- sevitam) of by replacement °akytah for “akrtayah 29.10 ‘The phenomenon may be partly due to the fact that the ~j stem was felt to be feminine (Cf. above) and was thus replaced by the ubiquitous -a stem masculine. . 4) Confusion between stems in ~jand in: the case of yogi ‘Mahiyogim for mahayoginam 30.4c Cf. Goudrisan and Schoterman p.70, BHSG §10.1ff, Pischel §405 ) If we accept lengthened -a-s in compounds such as *Saktydntam for Saktyantam, the ‘phenomenon may be explicable as a confusion between -i and -fstems, the latter often being extended to an -@ stem. However, Lhave not observed any other extension of -j to-distem, the phenomenon being probably restricted to -istems. 4. Stems on -i (and -d) a) The most common phenomenon is the extension of -istems to -d: yogesvaryah 2.21b °vidhdyinyah 2.28a, tarjanya°6.17c, devya(h] (for devifb)) 7.2c, 12.36, 16.406, 16.52c, 19.9¢ , rudrdnya?*10.94, kauberyabhimukhah 13.12b, atigulyas 16.24a, yoginya[h] acc. pl. 21.9c. Cf. Pischel §384ff, BHSG §10.86 and 10.164, Tagare §97ff 'b) This also happens to monosyllabic stems keeping the ~iin the stem: (loc) striyayam for striyém 735a. Cf, BHSG $10.6 Prakrit does not distinguish between words of one ‘or more syllables either, the terminations corresponding to -4 stems. Pischel §384, ©) Confusion between between the nominative and the accusative, especially in the plural, is very common, thus having -ih fo: -yah:°ripinth (perhaps supply / understand dadyit?) *6.174, yoginih for yoginyah 22.5a, 22.20c, “dhévinih 22.374. They are identical in Prakrit, Pischel §374 (and of course in Apabhramsa too cf. Tagare §§88,93) N.B. This means that a word in-Fan have three nominative plurals: regular yoginya, extended yoginya, and the accusative yoginth. Moreover, with te suvstitution ofthe feminine by the masculine the -inah ending is also possible, such as °abhilisnah in22.94, (cf the Apabhramsa borrowing from the masc. and neuter dectensions in the feminine Tagare $88) 5, Stems on -r. ‘Asa preliminary it must be noted that judging from the metre -r was most probably pronounced and conceived as --. Tis is also supported by the orthography ofthe mss, which may however date from a later period. Eg. 6.3b would be unmetrical with a very tuncommon violation of the metre (two faghus) ifthe -r were not substituted by -ri. The same has been remarked by the editors (Adriaensen-Bakker-Isaacson) ofthe original ‘Skandapurina (1998:27-28). Cf also the examples given in Satya Veat Sat (1964: 181) to show that intemal vowel Sandhi is not applied in some cases in the Raméyapa, These examples all involve initia --s which are treated as i -i- in the compound, showing probably again the actual pronunciation of rt that time rather than the omission of internal Sandhi: paramarsip, “gandharvars*, rksasarksa® and paramargind. Extensions to ~d stem: the case of mitr-. Cf. Goudriaan and Schoterman p.73. = mitarah (nom. pl. , probably understood as plural of * métard, an extended stem) 8.29d, 8.300, 8.30b, 8.30d, 8.324, 8.32f, 8.354, 8.36c, 8.364, 8.38d, 16.17b, 22.64, 22.154, 29.44, Noteworthy is that in the meaning ‘divine mother, the stem mAard, which is dectined like the feminine in -a, appears in Prakrit according to Hemacandra 3.46. (Pischel §392); BHSG §13.18 = extension to ~d via extension to 7: mituydh 8.34 (four titnes ~ in one series) Cf. the rare duhitr for dubity- BHSG §13.15 matré? 3.50a (metri causa) cf. BHSG §13.15 6. Stems on mute and -s. 4) Feminine stems in mute. The few examples show the assimilation of these stems to -& stems, Cf, BHSG §15.9; Pischel §413 (including the example vad); Tagare §75 vae- is extended tovdea in 10..1d — but it also has an accusative singular vik in 3.23b. (probably due to the merging of the nominative and .he accusative) dis is alternatively used wit the stem difae-g. in 6.31¢ This -dstem form is fairly ‘current, bu ef. Ksemardja ad Svacchanda 3.17ab judging it Aisa, ardayadims ttah patica diisu vidisisu ca K-di8isy ityAdir aiSah pathab, CC. alteration of gir/grd discussed in Goudriaan and Schoterman 9.74. kis possible that nisim 16.32c, which I understood ris+hiatus-filling -m, is made on the analogy of Prakritdisim (accusative) recorded in Pischel §413. Cf. digi -m- in SYM 55a by Thematization of -s stems. ~ This is very common, usually through the elimination of the -s cf is -as ending, adhatD adhd? N 2.288; tadgatas cf payah smytah (being mesculine) 3.184;Siram for Sirah 4.4c(-m- may be hiatus filler) 6:30d; brahmasirasyaitad *10.8c; Siracchedam 29.17x; cddha? for cidhal (metric) 5.94; *paya®for payo® 5.106, kaniya? for kantyast®(6.17ay;rudratejopabrmhitah 32.9b; * vircetena (for “cetasa) 21.46e;, margasie (tt *Sirasi) 21.2Se;, rajasya 22.254, The feminine °mand asin 31 Sais, probably also the result this thematization ‘This type of thematization teja for tejas) has been noted as Aisa by Ksemardja ad ‘Svacchanda 5.64: tejodghita ity aiSah pathah. Cf. BHSG §16.10ff, Pischel §409 (both BHS and Prakrit keeping the correct lunthematized stem too just as Aisa cf. Pischel §407) = Another way of thematizing -s stems is by extension: **cetasab for °cetds 10.4b, 13.4b; manasena for manasd 16.372. This is rarer in Prakrit too ef. Pischel §409 (where ‘one ofthe two examples is °monaso) BHSG § 16.2.9 (recording buth cetasa and ‘manasa) ~The thematic derivative may also be used, ifc. °minasah for °mandbt 19.20b, whi regular. 7, Stems on -nt ‘Only one iregularty has been noted, the thematizaton ofthe present active participle. **hasantoccair hasantah + uecait) 10.74. Cf. Goudriaan and Schoterman p.75. For this, the following parallels have been adduced by Prof. Sanderson!3, This extension of present active participles, which reflect Middle Indo-Aryan (ee Pischel §560; cf. BHSG §18.4.31) is not uncommon in our corpus, as the following examples show: 1 rahasanta-. Svacchanda, NAK 1-224, f. 105rI-2 (A) (> Tantrasadbhava, NAK 1-363, f. 134v 2 [B) dhyayito yogibhir nityam prasannavadaneksanah pprahasanta -m ivdbhatijyotsnarasmibhi niemalaih 1d rasmibhi A: rasmibhir (corrected against the metre) B ‘The form has been edited out in the text of the Kashmir edition, Svacchanda 10.603 ddhyato vai yogibhir nityam prasannavadeneksarab ‘Prahasan sa ivabhati nirmalajfnarasmibhib ‘nd in that of the Grantha manuscripts (IFI, T. No. 507, p.226 (Al: No. 570. p. 224 [B]; No. 1032, p. 228/929 [C}} sa dhyato yogibhir nityam prasannavadaneksapah ‘Prahasann iva cdbhati jyotsnéraSmisunicmalait rasmisuninmaainApeB : sunnah C: asmjsunie® Asc garjanta-. Svacchanda, NAK 1-224, f. 104v5 (A) (> Tantrasadbhava, NAK |- 363, f.113v6 (B]): garjantair (air corr: ai A’: ai B ganavrndais ca); > ‘Svacehands 10.587 and the Grantha iss: grajadbhir 3. Vypayantah (nom. sg. masc.) in Tantrasadbhava, NAK 1-363, f.3¥ (1.49): 13 Personal communication ina lees.sa eva bindur ity ukto vyapayanto vyavasthitah yap pabubhdvasha samsitesasaray asa ydayanta oid £65 wdaya (nom. mas) bid: yan. JFvalanta- bid. f. 107v: jvalantah (nom. sg. masc.); ibid. prajvalantait" 8, Stems on -n, 4) Thematization of -an to -a. Cf. Goudriaan and Schoterman p76 varmam for varma 4,64 (-m- may be hiatus filler); karmam (ace. sing.) 6.480, karma (ace. sing.) 7.15b, °karmesu 18.264; °atme for ®atmani 14.49, °ndmena 21.83 Cf. BHSG §17.2-21: ef. the appearance of the thematized stem of rjan- in Jaina “Mahi in Pischel §399. +b) Thematization of -an to ~ina Cf. Goudrivan and Schoterman pp.76-77 navatmano for navétma 18.¢ adhvanam for ava (also neuter for mase.) 22.114 Cf. BHSG § 17.31-47, Pischel §402 (noting the same vacillation between stems, recording addha / addhino = adhva), ) Thematization of ~into ina, All examples are for the nominative singular. ‘mantrinah 6.1 1a, raktayajfiopavitinah 19.7, pltayaiNopavitinah 10.9, vratino 13.78, rdpinam 18,19b; phalabhiginah 32.94, siddhikarksinab 21.360 Cf. Goudriaan and Schoterman p.77, Pischel §406 4) Confusion of -in/-i/-7 the case of yogin ‘This has been mentioned w.ien dealing with -i stems. Cf. yogim for yoginam 30.4¢ (cf. Pischel §405), yogibhih for yoginibhih 16.74, 18.28b, 21-le; yogi® for yogini22.1d ¢ Numerals 1. Differences of gender are usually ignored Cf. Goudriaan and Schoterman p.77, trayah for tisrah (but the masculine and feminine are confused in the sentence in any case ef Pischel §438, BHSG §19.8-9) 6.17b; ekaikasmin 22 34c (masc. for the fem. and also loc. forthe gen., cf. pronominal declension below and Pischel §435) 2. The use of tra with the appropriately declined form for *triméat 6.19¢: astdtrimsims tathd varniin Cf. BHSG §19:30ff and Goudriaan and Schoterman p.78. ‘This shows assimilation to -a stem nouns, as does thematised dasa, D Pronominal declension 1, Thematization and nominal endings tam for tat 1.10e; yarn for yat3.2c; anyam for anyat 2.34b; visve for visvasmin 2.234; ppirve for parvasmin 5.3a; imaift 7.7¢ Cf. Goudriaan andSchoterman p.78, BHSG §§21.11 (ace. sing.) 21.20 (loc. sing.) 21.71 (ima; Pischel §424 (not allowing nominal loc. sing.-s) 2. Gender confusion 4) The most common is the use of masculine for feminine. tasya for tasyab 2.350, 3.8b, 3.14a, 3.214, 16.494, 18.72, 30.3b, 29.43b, 29.496; yasya for yasyah 30.2c; tasmic for tasyas 32.7a; ye for yah 8.299; ete for eth 21.172; ebhir 21.176; tasmin for tasyamn 13.tc; caisimn (2) for sam 13.204; sa for sé 18.1 1d; tesaim for tdsim 22.54; esd for fem. 22.10; etesdm 22.38a¢. Cf. Goudriaan and Schoterman p.79; the gen. sing. demonstrative being frequently the same for the masc. and the fem. in Prakrit in Pischel § 423 (and BHSG §21.18 using Prakrt forms); for the confusion of mas. and fem. pronominal declensions, or rather the substitution of the fem, with the masc, in ApubhramSa see Tagare §123 (iv) ff. by The use of masculine for the neuter also occurs ayam sphujam for idam sphutam 16.504: tasya siddhir vardrohe yad idam gopayisyati (or shall we understand a hiatus-filling c+: ya -<- idam?) 22.404 CC. ayam for idam in Ardhamigadhi, Pischel $429. The morphological disappearance of the neuter in Apabhraméa (Tagare §76), which began in Prakrit (Pischel §357) may explain this substitution «) The feminine pronoun instead ofthe collective masculine, tésdm (referring to mantrandm and vidyndm, the later being closer) 2.39, tas (ref. to Bhaginis and Viras) 21.36e. This confusion of masculine and feminine shows, ‘Apabhraméa influence (“agare §76 and §123 (vi)) 3. Number and case confusion a) Number confusion occurs with the dual, which i in fact not much used in nominal Yonigahvara 17 pa tener dated 0170671999tejam-ca-candrarp-ca-vimisritatvat 305 Yonigahvara (rewritten by Jayaratha ad Tantréloka 3.67 as tefahéasAkasphutavimisritatvat) Also: jardmarana-céintekara in Lalitavistara 5.41. Cf. Renou 1961 §76 (pp.84-85) on the same phenomenon but mainly conceming Dvandvas. IV SYNTAX A Lack of congruence of noun phrases 1. OF gender Cf. Satya Vrat Sistri (1964:237-9), Goudriaan and Schoterman p.86 a) Neuter adjective qualifying masculine substantive: smptam (varpah) 3.8e, jhasaptamam (varnah) 3.9, “lagnam (varnab) 3.14, ndbhigam (varpal) 3.14, sénusvaram prakinitam (varpah)3.16b; evam eva kariyo yam vidhih sarvatra coditam /smanyam devadeveSi ratasyam simudiyikam 19.25; etdini sitrasthdndn kathita dvidasimrte 29.204 bb) Masculine adjective qualifying neuter substantive: ratyaksas (Sistrani) 12.173, kathyamano (cekramm) 21.1 ©) Masculine / neuter inconsistency in a series of adjectives: prathamam daksinam proktam vimaé caiviparah smrtsh (qualifying either varnah or ‘bhasanarn) 3.10ab, { variant: casaptamalvm N/D 3.104), thyinanyasain samakhyatam sakalas tu 6:32cd; kavargam kirtito devi 16.20a; bhairavardpinam / navatmino tha kkartavyam 18.19be: cétigani pajayet /svasvadiksamsthitaA cAstram 18.23bc 4) Neuter substantive qulified by masculine forms in a subordinate clause: roksayet sarvadravysni yajfrthe ye prakalpitah 6:9ab £) Feminine substantive qualified by masculine adjectives (devyah) prajabhilisigah 22.94, *caksusab 22.270 2. Of number Cf. Satya Vrat SAstei (1964.239-241), Goudriaan and Schoterman p.87 a) Plural for dual: netréni6.41b, °ndyiki 7.264; daSanapaiiktayah 16.154: ud 16.22b (after nitamba and jarghs}; te ca 16.14c (after ndsiputas, which is singular for dual) by The too frequent use of the collective singular: jatav ekavacanam ‘There are numerous examples for tis, only a few have been listed. Sometimes they may be motivated by the metre, or they may show hesitation characteristic of colloquial language (. the section on colloquialisms below). They also seem to substitute the rarely used dual more often than in standard Sanskrit (where singular Dvandvas are used for expressing a dual forming one concept or group).bhasanam (for dual) 3.94, sitrunena for dual 6.13c, kaniyéigustha (for dual) 6.1 *Adike (for °Adisu) 6.28a, svasthdnaiva (for svasthanesu) 6.32, devya (for gen. plur.) 6.45b, tathosrolakamahisikrospukijam 6.50ab, karmany eva tatas tasmainn 7.123, * bhuktimuktit 8.3b, sarpyojananiyojane (lo, singular dvandva) 825, aksaralaksam (for pl. understood forthe lakhs in the context) 14.3a, japadhydnam (regular singular Dvandva) 14.5b, nisdputas (-s preferred before t-; cf. -st- Sandhi) 16.14, ‘The phenomenon ~ at least concerning the singular for the plural ~ has been remarked by Ksemarija ad Svacchanda 3.138a, where one must understand a group of mantras by the singular although he did not categorise it as Aia arvoddhytena mantrena K: jitav ekavacanam, «¢) Other incongruences of number have heen corrected, ‘They are more frequent in N than in D. CF, 3.454: yada samtosito gurub D/ yal satositd guru N. As this ‘example shows, most cases are surely the results of scribal corruption, such as misreading -o / 4 and omission / addition of Visargas in the feminine, In correcting. these errors, Ihave applied the practical principle mentioned above, ie. Ihave corrected errors which leave the metre undisturbed B Confusion of the nominative and the accusative ‘These examples have been kept together for the underlying principle must be the merging of these two cases into a nominative, while the othr cases seem to merge into single oblique case. ‘Where confusion was likely to be the result of scribal corruption, I have emended or ‘corrected the text. However, itis possible that I have not retained all the examples of this confusion in the original. Cf. Goudriaan and Schoterman pp.88-89. a) Nominative for the accusative siddhic(N only, not retained) for sidahim 5.2c; ayam vidhif (possible o understand in a separate seutence) 7.54; tritayo . japan 12.:5-16; vdmam nitambabhigar tu (treated as neuter) jarighd padas tathaiva ca afigulyas ca nakhas caiva tavargena kramat kuru 16.23cd-24ab; vethayo'pi va. homayet) 19.21b; bhaginyo 29.44; cakramudrithav tasyd darSayet punah 29.43cd; asau for amum 5.4b. by Accusative forthe nominative ‘This occurs usually with -i stems listed above: °rdpioib (perhaps also possible to understand as accusative supplying dadya) *6.174; yoginih for yoginyah 22.50, 22.206; °dharinih 22.374. But ef also: gulikasiddhim evam hi 29.9¢ (pethaps for ‘euphonic reasons?)©) Confusion ofthe nominative or accusative with other cases are rarer. Noteworthy, however isthe use of the instrumental inthe nominative sense, which may be due again to mixed active and passive constructions and not to case-merging. Cf. Goudriaan and Schoterman p.88, understanding thatthe nominative is used forthe instrumental. ajayet .. mantrind 18.23 b-d _Yajec cakre ca vidhivad yoginisiddhim icchata 21,120d It seems that “siddhim icchatdis commonly used as nominative in Aiga to denote the practitioner, probably due to its being a useful stock-phrase at the end of an even pda, Cf, Kaulajfdnanimaya 3.24: etan ta pajayed devi kaulikam siddhim icchatd, The same phenomenon has been noted in Goudriaan anc! Schoterman p.89 with examples of Jchatl standing forthe nominative. C Confusion of the non-accusative oblique cases 1, Locative for the Dative-Genitive tasmin (for tasmai) 3.45b, 3.46a, tasmin for tasyah 13.1¢, lolupaya tu sidhake 22,394 ‘The overlapping function of the dative and genitive in classical Sanskrit is well-known, ‘That the locative is also interchangeable with them with words of promising, giving etc. has been noted by Speijer §145. However, since this interchangeability is perhaps less ‘common, Ihave included it in the lst. The last example above shows that they are more interchangeable in Aisa for in classical Sanskrit the adjective and substantive should agree. Cf. alterating genitives and locatives noted by Goudriaan and Schoterman p.90. 2. Dative-Ablative for the Locative tanmatrebhyas tat cdnyah samsthitah 8.32cd 3. Dative-Ablative for the Genitive narakebhyah patitvena 8.29 4, Instrumental-Dative-Ablative for the (Genitive-)Locative (in duzi) sandabhyi: 29:34 5. Locative for the Genitive ‘This seems quite common. ‘vimadindm patitvam hi ksmabhavat sédhake bhavet 19.22ab ‘yan cnyani karmani anuktdny api sidhake 19.242 uudayam tesu yad devi mase mise tad ueyate 21.3ed etesu hrdayo devi mahibalaparakramah 21.34ed 6. Locative / Accusative ‘Two of the examples below concem the verb Asi, and in all three ca:es the -mis explicable asa hatus-filler, therefore Ido not think thatthe locative / accusative changeis very characteristic. It seems that in Aiga when aSri- was used in the sense of ‘dwelling in’ or ‘stationing oneself a’ then the locative was felt more appropriate to stand wit yasmin dvipa -m Arita 12.2% tya 16.32 Anyasmin kula -m- utpanndh 22.66 7, Instrumental for the Ablative, Ablative for the Instrumental and their interchangeability ‘Their functions overlap also in standard Sanskrit e.g. in expressing cause, but they are not as interchangeable asthe examples below show. 4) Instrumental for the ablative: indriyais ca sumudbhitay 8.326 b) Ablative for the instrumental (or corrupt fr the aceusaive?): yarhakramat for ‘yathakramena /2kramamm 21.466 ©) instrumertals and ablatives alternating in the same sentence (expressing cause): rena, ajena , ksmabhavat, raudiya, aksatai, padmahomat, bilvair homéft]19.21-23; ‘na snehdn na ca lobhena na cérthena na trsnaya/ na garvan na ca Kirtya vA22.39ed- 40a 8, Instrumental / Locative ‘These seem to be completely interchangeable and altemate with each other frequently. 4) Instrumental for the locative: raviget sidhakena tu 12.204; evam paripard devi trirdre varada bhavet; pard vai ‘navardte tu sadrétrenaiva capara 16.51cd-52ab ; pirvena ya.nansirtyam (for ‘Pirvasmin) 19.9a, 19.10b; adyena for ddye i.e. arake) 19.103; yatra yatra tu amsena (for amse) virasyatra rucir bhavet 30.7ab. 'b) Locative forthe instrumental: svamantre upasamhdram kurydc caivétmano hpi 18.29: guptamn parva krtam deve 32.126 ©) Altemating: yavarge vasavi tara karmamoti ind 16 43ab ‘On the convergence of the lacative and instrumental, Prof Sanderson wrote the following! ‘The confusion of locative and instrumental singular in -e and -enais well attested in our texts. See, e.g., Yonigahvars 136: vdmavarte nyaset, 166: arcane ‘ridhanena, 328: pliryate madirdsave, 423: meldpe dvldasth smytah / astau ‘akiena margena catuh Sambhavabhitale, Siddhayogesvarimata quoted at ‘Tanuralokaviveka a 30.26a, where after cight and a half Slokas maintaining locatives in parallel phrases instrumentals are introduced in their place: vakire salt cafe add /romicarkrp oka nut / VahnirdpS rakérena tejoripdrakiraja/ phakére yoniripa tu taksre pararipini Tinta utes bere vedi / pcre oahrmsd elec pikadiyika/ 48 fy leuer dated 21 May 1997.‘At the back of this phenomenon is, I propose, the convergence of these ending in Middle Indo-Aryan. In Apabhramsa the endings -em, e~, -,-/~, ahi, -ehi~ and -ehi serve both cases; see Tagare §§81-82. Traces ofthis are already discemible in the Jaina Maharastl of the Paimacariya; see 3.60: vaddhanti pala yatnatah’ ') ‘.kdmye karmani Sasyante mnasi muktikaiiksinah’/Saive Kecid ibecchanti Bhairave na kadicana 1.19 23-This isan exegetical device wed by Ksemardja (ad Svacchands 4.10Ted: hei kikksivat ‘parvatripiyojyam), but it does nt reall slve the problem.