Shb., Mans. priyena, putrena, G. parākramena, Shb. and Mans. parakramena; G. prāṇasatasahasrāni, Shb. pra[nasatasa]has[r]ani, Mans. praṇasa[tas]ahasrani; G. Gamdhārānam, Shb. Gamdharanam, Mans. Ga[m]dharanam; G. pasumanusānaṁ, Shb. paśumanuśanaṁ, Mans. pasum[a]nusana; Shb. mahamatranam (per contra note Sanskrit priyena, putreṇa, parākrameṇa, prāṇaśatasahasrāņi, Gāndhārāṇām, paśumanuṣyāṇām, mahāmātrāṇām -all with lingual). And in the Shb. dialect we have n for n a couple of times in suffixes, namely, chamanaye (Skt. *kṣamanaye) and aviprahino (Skt. aviprahīņas).1 Mans. has aviprahi[ne] (with 'Magaforms must have existed, as is shown by rupena (Skt. rūpeṇņa), sakkena (Skt. śakreṇa), ariyeṇa (Skt. āryeṇa), mattena (Skt. mātreṇa), dhammena (Skt. dharmeṇa), kammaṇā (Skt. karmaṇā), sahassāṇi (Skt. sahasrāṇi). Note especially analogical n in the suffixes of oropana- (Skt. avaropaṇa-), rosana- (Skt. roṣaṇa-), savana- (Skt. śravaṇa-), kubbāṇa- (Skt. kurvāṇa-), nibbāṇa- (Skt. nirvāṇa-).] (Note deva-puttena, D. XXI, 1, 11 ed. PTS.) According to Wackernagel, the dental n of kṣepnu- is not analogical but phonetic. But I ask how then does he explain the lingual ṇ of ksepṇā-? For if kṣepnu- were phonetic, then we should expect *kṣepnā- and not kṣepņā-. And it is not possible to explain the lingual by analogy; but nothing is easier than to explain the dental n of kṣepnu- by analogy. And it should be observed that Vedic trpnoti, Pāli pāpuņoti, Girnar prāpunati, Shāhbāzgarhi prapunati show that the conversion of n to n by ordinary rules is not prevented in the combination -pn-.

1 Shb. aviprahino is not to be judged wholly in the same way as chamanaye; they both have analogical ʼn in the suffix, but aviprahino derives its ʼn from the uncompounded *hino. Cf. Skt. paripăna- beside phonetic suprapāṇa-, paryuhyamāna- beside prohyamāṇa- (both from the 'root' vah), parāhna- beside pūrvāhṇa-, aparāhṇa- (Wackernagel's alternative explanation of this last does not convince me). On the other hand chamanaye is due to 'external grammatical analogy.' But the underlying principle, namely, the spread of n at the expense of n, is the same in both cases. On the n-cases of u-stems to which original g-stems have been transformed in the dialect of Shb. and Mans. (likewise Pāli), see Michelson, AJP. xxx1, 61. Shb. Tambapaṁni, Tambapaṁniya, and Shb., Mans. Pitenika- have Magadhan'n for native : see Michelson, IF. XXIV, 55; and it is far more probable that Shb. pranatika also is to be judged the same way from the evidence of Mansehra paṇatika, rather than to be explained by analogy; see Michelson, AJP. 1.c. (Mans. Tambapamniya is not certain as the text in ZDMG. XLIV, has a lingual : hence the form with dental n in EI. II, may be a mere misprint). Shb. ka[lanam] with dental n disappears in Bühler's ed. in EI. II, and is replaced by ka[lanam]. The form ka[lanam] in the German ed. is only a misprint, as is shown by the devanagari transcript; the same error occurs a couple of times in B's comments in the German ed. I follow Bühler in considering Shb. -garana to be a mere blunder for *-garaha. But it

dhan' -e for -o) corresponding to Shb. aviprahino; but has a lacuna in the passage corresponding to chamanaye. The Girnar text differs slightly in the wording of the corresponding passage, and has a lacuna where we would otherwise find a correspondent to aviprahino. We may therefore infer that the dialect of Mans. agreed with that of Shb. in this particular, as it is agreed that the speech of both is essentially the same; there is also a considerable amount of probability that the dialect of the Girnar redaction agreed with them, as it agrees with them in replacing n by n in case-endings; moreover, these dialects show a considerable number of special points of contact; see AJP. xxx, 291, JAOS. xxx, 87-89.1 The bearing the above has in solving our problem

might be one for *-garahana, in which case we should assume an analogical extension of dental n.

1 Another argument to show the plausibility of assuming analogical ʼn for phonetic in suffixes in the dialect of Girnar in the absence of any direct evidence, is the fact that this is found in Pāli (though not invariably). And there are some very striking agreements between the two as contrasted with the other dialects of the Fourteen-Edicts of Asoka.

For example, dh is retained in idha; hm becomes mh; -smi becomes -mhi; kata- corresponds to Skt. krta-; -āya as the dat. sing. of a-stems and as the 'oblique' cases of a-stems; -ayam as the loc. sing. of a-stems; r-endings in verbs; the words tärisa-, etārisa-, yārisa-; etc. And there are some special agreements between the dialects of the Girnar, Shāhbāzgarhi, Mansehra redactions of Asoka's Fourteen-Edicts and Pali as opposed to the Magadhan' dialects of the Asokan inscriptions. For example, the sounds r, ñ, n, ll (written / on inscriptions) as the equivalent of ly in Skt. kalyāṇa- (G. kalāṇa-, Shb., Mans. kalaṇa-, Pāli kallāṇa-, ‘Māgadhan' kayāna-); the pronoun ahaṁ (‘Māgadhan' hakam); etc. On the other hand, there are some special points of contact between the 'Magadhan' dialect of the Asokan inscriptions and Pali. Especially noteworthy is the agreement in the treatment of in conjoint consonants, and the word hevaṁ (G., Shb., Mans. evam): on this last see Michelson, IF. XXIII, 128. But Pali has also one special agreement with the dialect of the Shāhbāzgarhi and Mansehra redactions of the Fourteen-Edicts and the dialect of the Siddapura-Edicts of Asoka as compared with other dialects of the Asokan inscriptions, namely, the transfer of original-stems to u-stems. Two things are certain, namely, that Pali is a literary language only and does not represent any one spoken vernacular, and that whatever dialect forms the basis of the written language, that dialect does not coincide exactly with any one of the dialects of the inscriptions of Asoka: compare Windisch in his essay on Pāli in the Transactions of the International Congress of Orientalists held at Algiers, and Michelson, AJP. xxx, 287, 297.

is this: punya- goes back to *prnya-; *prnya- is derived from *pra- by means of the suffix ya-.1 Now *prna- might have a young analogical doublet *prna- with dental . From *prna- *prnya- would be derived exactly as *prnya- from *prna. From *prnya-, Girnar pumñam comes without difficulty; for ny becomes ññ (written mm) in this dialect; see above.

1 On the suffix nya-, assumed by Meillet and Brugmann, see Bartholomae, 1.c. Even if nya- (which phonetically would become nya- in the present instance) were correct, that would only necessitate shifting our ground in small degree: for nya- could be replaced by nya- as easily as ṇa- by na-.

III. On Certain Euphonic Embellishments in the Verse of

Propertius

BY PROFESSOR B. O. FOSTER

STANFORD UNIVERSITY

THE metrical art of a classical poet may be analyzed with considerably more precision than the art he has bestowed upon the selection and combination of pleasing sounds. The reason for this is apparent: words must be chosen and arranged primarily with a view to meaning and to metre, and only when these fundamental considerations do not stand in the way may rimes, assonances, and the like be introduced into the verse. We cannot, therefore, be sure, in a given case, that the poet has produced just that effect which he would have liked to produce had his medium allowed him. Ut quimus, quando ut volumus non licet was often, no doubt, the reason for this or that sequence of sounds. Hence extreme caution in the admission of evidence is a matter of prime importance for the student of this phase of the poetic art. One must constantly guard against mistaking for intentional what may be accidental or inevitable, and one must be willing to forego the satisfaction which comes from the discovery of definite rules, and content one's self with making out, more or less dimly, certain of the more persistent traits of the author's composition.

The subject I have set myself in the following paper is a very modest one. I have not attempted a critical analysis of the euphonic art of Propertius, but have merely sought to assemble, with such classification as promised to be convenient (though far enough from exact), adequate illustrations of certain tendencies in his work which have impressed me as significant. Considered together, in the light which they shed upon one another, these tendencies will, I believe, be felt to constitute a not unimportant element in a style which has hitherto been studied mainly from other points of view.