extended phrase, atharvane samhitalakshanagranthe caturâdhyayikavyakaranam sampurnam. The endings of the sections are also marked in entirely uniform manner, prathamasya prathamaḥ pâdaḥ, and so on. There is nowhere any trace of an

enumeration of the rules.

The form of the name as before accepted by me is therefore established beyond the reach of question; and the responsibility for its inconsistency with grammatical rule is to be put upon the shoulders of its authors or transmitters, whose errors, if such there be, we are not called upon to amend in a matter of this kind.

The opening invocation, om namaḥ cribrahmavedâya, is in close accordance with that of the other manuscript (see p. 575, or 245 of the separate impression).

I begin now a detailed collation, with statement of the results following from it. With this I incorporate, here and there, a few corrections, in great part derived from a notice and criticism of the former work given by Weber in the Literarisches CentralBlatt, No. 29 (1863), and reprinted by him in the second volume of his Indische Streifen (pp. 230-233). For the sake of brevity, I denote the new manuscript everywhere by B., and refer to the other as A.

After the invocation, already quoted, follows next om atha 'ngirasah. This, then, is evidently the first rule of the treatise as at present constituted, and my conjecture (p. 575 [245]), that it was "a copyist's error for atharvángirasah," and so a part of the invocation, is proved erroneous. Being, however, unexplained in the commentary (like the metrical and other appendages to the third and fourth chapters, to be noticed later), its right to stand as an original part of the treatise may be questioned. Its form, too, is suspicious: utha 'ngirasah would be a proper heading for the Atharvan text itself, rather than for such a treatise as this. Nor do we expect a Prâtiçâkhya to begin with defining the text to which it relates; a mere evam iha, as in the next rule but one, is all the definition that is called for.

1.2. B. divides into two rules, by a line of interpunction inserted after ca. A. does the same in the final repetition of the rule. This is not an evidence of much consequence, as there are several cases later (namely i.94, ii. 65, iv.96,109) where B. blunderingly inserts a mark of division in the midst of a rule. The exposition of the whole as one connected rule by the comment in A. (though not unattended with difficulties) is of much more weight. The point would hardly be worth dwelling upon, save that the admission of atha 'ngirasaḥ as a rule seems to demand that, in order to make out the number of forty-one rules in the section, we throw rules 14-16 into one (see below), and then find one extra rule somewhere; and there would appear to be no other place to

find it than here. I will not undertake to decide the point absolutely.

...."

i.4. B. has anṛkáraḥ.. as has A. also in every place but one, where it reads annrkárah. There can be no question that the amendment to anļkáraḥ is a necessary one.

i.8. B..... trtiyântân iti çau: a mere copyist's error.

1.14-16. These are given in B. as a single rule, and should doubtless count as such. It was not quite correct to state that A. treated 12 and 13 together, as one rule; a renewed examination shows me that there has been, rather, a complete falling out of the comment to 12, so that 12 is stated, and then 13 explained and repeated. As to how the required number of rules shall be found in the chapter, see above, under rule 2.

i.19. B..... adharakanthyah; A. has the same in the rule and its final repetition, but thah both times in the comment.

i.25. B..... adharaushthyam; A. has thyam in rule, paraphrase, and repetition, but not in the interpretation.

i.30. B..... antasthánám; and so in every other like case hereafter (with an exception or two which will be noted in passing); A. has the same almost uniformly; I adopted the reading antahstha etc. because the rules of the Prâtiçakhya itself do not authorize the omission of visarjaníya before a sibilant followed by a surd

mute.

1.33. This rule is probably to be understood as eke 'spṛshtam (as suggested by F. M. Müller, Sanskrit Grammar, $57, note f). The commentary reads (a little amended) eke 'spṛshtam svaranam karanam bruvate, apare vivṛtam. What, however, should be the difference between asprshṭam and vivṛtam except as a verbal question, of choice between synonyms, it is not easy to see. Possibly a difference in the implied degree of openness is to be recognized.

1.39. B. salakára lṛvarṇam; A. everywhere sabhakaram, followed by mrvarnam, trv-, and mṛv-; that the true reading is that given in the edition is not, of course, to be questioned.

i.42. B. has, apparently, -shṭanaḥ amended to shṭhánaḥ.

i.43. B..... hínaḥ çvásanádaḥ: doubtless an error of the scribe. 1.63. B. shatparasor....: an obvious blunder.

1.66. B., in fact, has.... angurim, which I pointed out as being undoubtedly the preferable reading.

1.68. B. parosh-: a scribe's error.

1.73. B. dirgha pra-: also doubtless an error of transcription. It is to be noted that the Taittirîya-Prâtiçâkhya (at x.24) makes the same prescription as the others do respecting the exemption of a “pragraha" vowel from phonetic alteration.

i.74. B. ikúr-: a mere slip of the pen.

i.77. B.....co 'dáttaḥ; and A. has the same reading. My reading udattáḥ was, I think, a simple oversight; since the singular, used in a collective way, is obviously not less admissible than the plural.

1.82. B..... iti paraḥ, like A. I should, however, still defend the reading of the edition as a called-for emendation.

1.85. B. has pansumans- by first hand, but alters, erroneously, to páñçu-.

i.94-95. B. divides, erroneously, soshmani: púrvasyá 'núshmá "ntaryena vṛttih.

In the edition, 'nushma is an error of the press for 'núshmá. i.96. B..... kháimakház ity ....: an oversight of the scribe. i.98. B..... avavetáni.. ............: a blunder merely.

i.99. B.....'nuttamá sp-: also nothing more than a blunder. I am inclined to think that the expression yathasamkhyum in this rule distinctly implies the existence of four yamas only, and prescribes the insertion of them after the various mutes "in accordance with the number" of the latter: that is to say, of a 66 after a yama first" mute, a second after a "second," and so

first

on.

i.104. B. púrvaḥ svar-, which seems to me a false reading. I have no new light to cast upon the construction and meaning of this very obscure rule.

i.105. B., like A., runs all these extracts into sandhi together. It writes sṛshtahs and práciḥs (putting the figure after the visurjaniya); A. had dropped out the sign of protraction in these two words (as also that in praçiss, where B. reads with the edition). It also has avaçe 'ti, without the sign, in accordance with what I had pointed out to be the requirement of rule 97. It adds a double stroke of interpunction after the of asisd, before the figure. Finally, along with A., it omits the iti after tasd in the last citation; this I either admitted from the Atharvan text by an inadvertence, or inserted by way of an (unnecessary) emendation.

B. adds, as final rule of the chapter, ita uttaram adhikam, one of the phrases reported in the edition as constituting a part of the comment. There was nothing whatever in its appearance in A. to suggest the idea of its being a rule.

The full rehearsal of the cases of protraction in rule 105 is not in accordance with the general style of the treatise, which is wont to dispose of such a matter by a gana, like khanvakhází 'tyádi plutam; and this circumstance, taken in connection with the peculiar mode of treatment of the two last rules by the commentator, is calculated to suggest the suspicion that whatever follows rule 104 is a later addition to the text. The Taittirîya-Prátiçâkhya, at least, deems it no part of its duty to give a statement of the

protracted vowels of its text, and in the Vajasaneyi-Prátiçákhya (ii.50-53) such a statement is made only incidentally.

ii.9. B. naṇanemaḥ....: mere blunder.

ii.16. B. shakárántán nánáp-: an equivalent reading.

ii. 18. B. lopo daň stháḥ stambhoh....: the most blundering version of a rule that is to be found in the whole manuscript.

ii.22. B. nakár: slip of the copyist's pen.

ii.26. B. nakarasya caṭatavargeshu simply; and it has been pointed out already by Weber (in the review referred to above) that this is the true reading of A. also, I having allowed to blunder into the rule three words that in fact belonged to the comment. Of course, the inference which I drew in the note to the preceding rule from the repetition of visarjaniyaḥ in this one falls away as unfounded.

ii.47. B.....rodviv-: error of copyist.

ii.58-59. B. runs these two rules together, and adds ca at the end. The ca may be a genuine difference of reading, but the rules must at any rate be separated, as no na is implied in the latter.

ii.61-62. B. again accidentally omits the pause between these two rules.

ii.65. B. puts a viráma under the t of karat, and adds a pause, thus dividing the rule into two; but it is an obvious blunder.

ii.73. B. has pitush pi-, which is the better reading; but it also gives erroneously pitari.

ii.75. B. ayush pra-: again a preferable reading.

ii.77. B. paridhish pa-: once more, as in the two preceding cases, the true reading, since the Prâtiçâkhya ought by all means to follow its own prescriptions.

ii.82. B..... saḍrúpasya: an equivalent reading.

ii.86. B. stṛtasyasvapishu: copyist's error.

ii.89. B., pr. manu, had nárṇadenope; but it is amended to an accordance with the edition.

ii.92. B., pr. manu, had.... akáravyadhaye...., but is amended to akasya (for akárasya?) vyavaye.

ii.94. B. supports the peculiar expression -mebhyo.

ii.102. B. has in margin patha mṛçi.

iii.3. B. vyavadhav....: mere copyist's blunder.

iii.20. B. ralope púrvasya: a real difference of reading; acceptable, but not necessary.

iii.27. B. -padhá svare: error of the scribe.

iii. 28. The lacuna pointed out as existing here in A. turns out to have involved two rules, which B. gives thus:

chakáraç ca;

pippalyadishu púrvát.

In the former of these, there is doubtless continued implication of svarát, and the rule teaches that ch is doubled, or changed to cch, after any vowel. To this no other restriction applies than that taught in rule 30; and that one, only in the combination chy. The corresponding rules in the other Pratiçâkhyas are as follows. The Rik Prât. (vi.1, r. 3, ccclxxx.) teaches that ch is doubled as well when it is not the first consonant of a group as when it is; while a couple of later rules (vi.3, r. 12-13, ccclxxxviii.-ix.) rehearse a number of exceptional cases in which the duplication does not occur. The Vaj. Prât. (iv.24) prescribes the invariable insertion of e between ch and a preceding vowel, and excepts (iv.25) but three cases. The Tâitt.-Prât. (xiv.8) prescribes the duplication only in certain specified cases.

The second rule declares that 'in pippali etc. there is duplication after the first vowel of the word.' The cases here contemplated, judging from the word taken as example, are of a wholly different character from the others to which the rules of duplication relate, being such as have a double letter as part of their original and proper orthography. I do not, therefore, think it at all worth while to look through the Atharvan text in order to pick out the other words which may have been included in the gana. In every occurrence of this particular word, it may be noticed in passing, a part of the Atharvan manuscripts read pishpali.

The next following rule is na visarjaniyah, and with it begins the rehearsal of exceptions.

The hope which I before expressed (p. 584 [254]), that the restoration of these lost rules would lessen somewhat the list of duplications, is not realized: not a single one of the consonant groups treated in the third additional note requires to have its form as there given modified; we have only to extend the simple letter ch, in a host of cases, to the group cch.

iii.33. B..... prakṛtyáḥ: a blunder only.

The Tâitt.-Prât. also has a corresponding rule (x.24).

iii.39. B.....'ntasthah; and the omission of the and the treatment of the word as a singular were an oversight on my part, since my copy of A. reads in the rule -sthaḥ, and in the comment antastha bhavanti.

iii.53. B..... púrvapadáder.... A. has the same reading in the rule, and purvaḥ was my emendation, suggested partly by the apparent necessities of the case, partly by the paraphrase of the commentator, which runs purvo bhavati: padader akárasya. I do not see how the desired meaning can be obtained without reading purvah; although, even then, the expression is very difficult and peculiar.