Divamnem explained as dvamnem from a du = Sk. 15+ du, middle, = ‘to suffer torture’, (‘Wer einen gerechten erdulden lisst’) was long followed by many. Now a root *dw = ‘to remove’ is suggested, altering the entire cast of the verse: ‘from him the long endurance of misery and dark- ness shall be spared (removed (?))’. The Pahl. suggests div in the sense of ‘deceive’ (cp. dab). Aparem khshayd = ‘the uttermost of ruin’ (?) is inter- esting. The Pahl. trlr. suggests dpara in the sense ‘later’, so Spiegel and Jackson. Khshayd. Y. 71, 17 speaks strongly for a gen. with daregem dy. Its position, it is true, also well indicates a nom. according to every Vedic analogy. For that reason I formerly rendered alternatively as a nom. masc., ‘one otherwise ruling’; cp. Mazdé khshaydchd, Y. 28, 7 (see also Haug). Later I followed, with Darmesteter, the significant shivan = ‘lamentation’, and am still unwilling to twist the sentence out of a natural shape, and, Supposing the existence of a khshayah, or a khshaya, I would compare kshaya from the second kshi. Vaché cannot be gen., but, if we emend to vdchéd, ‘woe of speech’ becomes possible. Avaé(é)tds is the noun in the nom., avae(é)tdts, cp. ameretatds, also yavaé(é)taité(é) (from yavae(é)tas[°tdts]), etc. It would then be either in apposition with vachd having adjective force, or, reading vdéchd, we should render ‘lowness’, that is, ‘vileness of speech’. The syntactical reference of this line b to ashavanem - - - héi in the sense ‘removed from’ seems to me erroneous, see above. (It is foreign not only to the spirit of poetical diction, but of all ancient original matter, to have these marked particulars strikingly alluded to at length, and then separated from the person who escapes them by a feeble separating word, while the wretch to whom they belong, as a punishment, is mentioned in the next

line. The entire verse refers to the idolatrous Daé(é)va-worshipper who would deceive the saint in the interest of the Lie-demon; and it is in anti- thesis to the preceding verse, which expresses the truth of Ahura, and his prophetic adherent.) Haug’s eradicet for naéshat is strikingly out of the way. Naé(é)shad(f) is a 3°4 sg. aor. of ni (n7) = ‘to lead’. The Pahl. trir. read dyad(t). His shivan = ‘lamentation’ may be ‘literally, and not merely figuratively, a translation of khshayd (cp. Darmesteter’s études). From ribishné we see that he took avaé(é)tds as a compound of ava with a form of ¢ = ‘to go’ (see also Haug). The Pahl. trlr. renders: ‘he says (that is, his word és) ‘a vile proceeding’. In tem = ‘darkness’ in c we seem to have an amusing error. But, as tem is correctly used for temanhdé in the line immediately preceding, the only sensible conclusion is that the trlr. arbitrarily repeated the word from the foregoing line, allowing himself to choose that procedure simply because the word éém in the Gathic reminded him of the temanhé just preceding. And this is of importance to explain other similar cases. Here ignorance is excluded as an explanation, and yet the outward shape of a word seems to have determined a free insertion, or a rendering! Ner., according to all the MSS. except that used by Haug, follows; but Haug reports tam - - - bhuvanam (as a masc., possibly having the 224 dhuvana in view). It is possible that this tam may not only render the Gatha, but a now lost Pahl. reading zak, or denman, in which case our amusing error would turn out to be a corruption from the original, no tem = ‘darkness’ having ever been written by the earlier, and more original trirs.

21. Taking Hawrvaté Ameretdtasché as gen., we should have the most formidable heap of genitives in literature, no less than six in a string, followed closely by a seventh. Moreover Ahura’s ‘store of Weal and Death- lessness’ is a somewhat strained and artificial expression, but his Asha, which included all his legal and ceremonial provisions for the welfare of His faithful, naturally has a ‘store’. The words are acc. pl. used for acc. duals, as perhaps ahdbis is used for the dual. The absence of the dual number tells also nearly (if not quite) as much against the words when regarded as gen. If gen. (?) render: ‘May M. A. establish from the fulness of Weal and Immortality (long life unbroken by death), and from the fulness of His holy Order the protecting headship of his own Sovereignty’. The gen. bérdis naturally suggests the rendering ‘out of the fulness of His holiness’, and doubtless the form was chosen on that account; but the gen. with @ properly means apud; the two shades of meaning are easily harmon- ized; cp. d varendo vichithahyd. For 6dpaithydd(¢) with the gen., cp. v. 15 vird(ajd(f)cha with adrujantd, mainyeus with vahistad(t) Y. 33, 6. As to

- 469 sarod, its obvious meaning, as the original of the Pahlavi-pers. sar, has long been familiar. The derived meaning custos = Wachter (Haug), or ‘refuge’ (Ved. ¢drman) was adopted by Haug at a time when he had not yet inves- tigated the authority of the Pahl. commentary, or the Pahl. language. The meaning ‘headship’, ‘rule’ is actually needed to explain the Parsi-persian

word. As to the form, I should regard an accus. pl. fem. as the most improbable. It might be an accus. sg. neut.; but saré occupies the place of a nom. sg. in a Gathic, or Vedic, sentence, and it may be nom. sg. masc. (cp. its exact expression as such in sdremné). There is no reason why it should not be even nom. sg. neut., in apposition, but not agreement, with Ahuré, theme saranh (sarah); cp. Yt. 10, 40 Karetachid(t) aé(é)shim hufré- yukhia yoi nighrdiré(e) sarahu mas(k)ydkanam ashemndjané bavainti. (In Y. 49, 3. I regard saré as hardly possible as a gen. on the same ground as that on which a gen. Mazddo has been so generally rejected in Y. 28, 1. It would be a heaping-up of genitives. In t@ varhéus saré izyai mananhd, vamheus obviously applies to mananhé. Saré is acc. sg. neut. = sard from saranh (sarah)). There are three stems sar (loc. sair?, if read with most MSS.), sarah (saranh) in Yt.10,40, here, and in Y. 49, 3, and possibly sara may be the stem here. It is hardly necessary to cite instances of the different genders from the same roots, but cp. Sk. dyu, m. and dyus, n., jana, m. and jdnas, n.; cp. the genders changed with change of accent, apds, m. and dpas, n., tydjas, n. and tydgd, m., rdkshas n., and rakshds m. (injurer), or even without change of accent, apds, m. and apds n. As to a stem sara beside sarah, cp. Ind. hédas n. s. neut., and hédas n. s. m., stem héda; cp. gépas N.S. masc., and gépas nom. sg. neut. See Lanman, Noun-inflection in the Veda, p. 549 for further examples. The form is either an acc. sg. neut. (?), or a nom. sg. masc. from sara, and the syntax gives such strong evidence that this case alone might establish the existence of a stem in the a declension (or else it is a nom. sg. neut. in apposition (?)). For Mazda (or mazdd) in apposition with a neut., cp. Yt. 1, 8 dasemd yad(t) ahmi spéno, with an abstract, cp. Yt. 1, 7 khstvd yad(t) ahmi khratuS- - chisti§. In apposition with a pl. neutr. (?), cp. Yt. 1, 7 Téirya (ahmi) asha vahista (ashem vahistem (?)). This is however the later Zend of interpolation. As to vazdvaré, which, comparing ayaré, saddaré, etc. one would naturally call a neutr. pl., but which, as indeclinable, may be a sg., it is hard to resist the impression that vaz = vah is the root extended by d + the suffix van = varé; cp. karsvan, karshvaré, and the meaning is ‘strength from abundance’,) J and prosperity’; see the Pahl. as rendered by Ner. Haug takes haurvaté amereta- tascha as accus., but ddpaithydd(t) as = e (suis) bonis = seiner Giiter Wohlstand. The Pahl, might, as usual, be rendered much closer to the Gatha, Yet lame as

the rendering is, it gave the clue to the meaning of every word. Bérézs = béindak, which latter is properly gen. by position; and avd should strictly be taken as referring to it, see @: ‘to the full one, that one who is Righte- ousness’. Vapaithyad(t) = nafshman = ‘own’, and patih, which is hardly to be referred to khshathrahyad = zak 7 valman 7 khiddt; saré = sar- darth, so, uniformly. Urvathd = désté, giving us our earliest information. Vazdvarih may be merely transcribed, but there is always danger in talking about the ‘mere transliteration’ of the words of a mother language into those of the daughter dialect. Mother and daughter are really one. If vazdvaré was Zend, vazdvarih was Pahl. ipso facto. Ner. knew what vazd- varth meant; his pivaratvam at least is no transliteration. .Avd, if read in c, might even be regarded as carrying its force so far as déstd, ‘to his own soul’ is not the needed sense; valman would be better read to escape this last.

22. Astis, if = dsti$, might properly be referred to std = sthd as = ‘stand by’, ‘helper’; cp. Vedic pratishthts, but I think it is hardly desirable to depart from the MSS. simply to show that we are aware that @ may become 7, as in Ind. sthiti etc., etc. The simple and concrete ex- planation of the Pahl. trlr. comparing dsthi =.‘bone’ is more probable, because not so advanced as the idea of ‘helper’, ‘stand-by’ (to the Deity). Cp. also Y. 33, 2 where the MSS. also read astém, and Y. 46, 11; Y. 49, 11, where the astayé are in Hell, a locality where assistance to the Deity could scarcely be rendered. Péddkihd may possibly be a pl. in °thd, and so a literal rendering for chithrd; see the Parsi-pers. which does not see the adverb, also Ner.’s prakatatvam ‘manifest things to (DJ. avd) the wise (are these)’. See the certain freedom, as opposed to error, in what follows. The concrete khudd in this particular phrase is not bad for khshathré; see the certainly free vddinyén = hapti. Vilokayitéd in Ner. is also not bad for hapt?.

XXXII,

See for summary and additional notes S. B. E. XXXI pp. 54—67, changes in opinion decided upon since its composition are however not uni- formly noted.

1, Ahydché (so to be read) refers to the subject in paouru (or pouru). aé(é)ndo v.6, in naé(é)chtd(t) vidvdo in v. 7, to dussasti$ in v. 9, to the subject in yasché vadaré vdizhdad(t) ashduné(é) in v. 10. This person re- presented his evil daé(é)tu, verezéna, and airyaman; see Y. 33, 4(?), Y. 46, 1, Y. 46, 5. Ydsad(t) = ‘prayed’ (not ‘wished’); cp. yds@, Y. 28, 1. Spiegel’s important mahmt = smdsi_ ig, notwithstanding its interesting character, hardly tenable. Haug, following the Pahl. indirectly, is correct. Mahmi

mandi are both loc. in meaning (Justi), and there is no reason whatever (except a desire to differ for the sake of differing) why manéc should not be compared with the Indian °manw, or even with the fem. mand. The latter by a familiar transition might well account for a masc. mana. Manahydcha (Y. 71, 3) proves the certainty of a stem mana in Zend, as manayd points to the fem. and manaé(e)ibyé (?) to the masc. Others insist upon a dat. infin., but, if such be the form, the word is still used in the sense of the loc.

Mahmi = ‘through me’ is improbable. The prosaic attempt made by able scholars to string lines a and 6 together is, I think, a mistake, as is the neglect to supply the ever-recurring form of ah = ‘to be’, where indicated as in line 6, with Daé(é)vd (hent?), and with mahmi mandi, the latter words being obviously intended to agree. Antithesis, a chief agent in poetical diction, is present in ahydchd and mahmit: ‘His baé(é)tu prayed (as the holy prophet did in Y. 28, 1); his are the Daé(é)vas, but in my venerating thought 7s the friendship, or friend, of Mazda’. That a and b should be addressed to the Devils (the Demons, informing them (even in defiance) that the community desire to know the disposition of Ahura Mazda is, I think, unlikely. Haug’s ad ortus = urvdzem 4 is not desirable; his mahmi mandi = ‘in my opinion’ is better. Pahl. Zak is gen. by position; see ahydchd, and Ner.’s asya; ydn is gloss; see Ner.; read ‘- - - in his proprietorship, or kinship (that is, as lord, or kinsman) - - -’.