^{page 454 note 1} The author follows closely Ziemer (Vergl. Synt. d. indoger. Comp.) here as in almost every other place.

^{page 454 note 2} Of course in any inquiry into linguistic origins we must deal with concrete terms like ‘great’)(‘small,’ and not with moral abstractions like ‘good’)(‘bad.’

^{page 455 note 1} Note also Brugmann, Gr. ii. § 135: In mancher Beziehung geht -- mit dem Participialsuffix --den gleichen Weg.

^{page 455 note 2} I agree with Schmidt, J. in K.Z. xxvi. 337Google Scholarsq. that this is the Aryan form of the suffix spite of Brugmann's argument to the contrary in his Grundriss ii. § 135, Anm. 1. I write the stem however as -yens-s-, the s being really a nominative ending reaffixed in the Aryan period just as later in Latin participles, e.g. feren-s. If this explanation is reasonable it ought to be capable of a statistical proof. The upgrowth of the comparative from the participial sense must have mainly taken place in sentences like the R.V. passage quoted above (ix. 66, 17^c), that is to say with the comparative adjective in the nominative case. We have then the nominative variations masc.; to this last -y ĕ neut. was added: -ēs-: -ĕs-==yē-: -yĕs-. In Sk. -yĥ-s was treated as -yĥs- to form the acc. sg. and nom. plur. masc.: had grades in o, i.e. , and yes was further graded to -is. The perf. participle sin -ven-s had the same treatment precisely, with the additional stem -vṇ-t, evolved out of a case form . A statistic of the comparatives in -yas- and -īyas in R. V. excluding sányas ‘old,’ návyas ‖ návīyas ‘young’ which are not true comparatives reveals fifty-five nominatives masc. in -yā-s, sixty neuters in -yas, two vocatives (masc.) in -yas, eight accusatives sg. and nominatives plur. in over against but fifty-nine case forms in -yas-. This statistic seems to me to amply vindicate Schmidt's contention for an analogical origin of the stem form -yes-. I leave to Schmidt and Meringer the defence of the proposition that gives , noting however that it is accepted by Streitburg against Brugmann I.F. iii. p. 150.

^{page 455 note 3} These instances conform precisely to the point of the last note—that the compv. sense grew out of the participial mainly in the nom. sg. masc. and neuter.

^{page 456 note 1} This it could not metrically do.

^{page 456 note 2} This sense is to be also seen in ἄλλως: I 699 , ‘he is headstrong, even utterly.’ Hdt. i. 60 , ‘and quite comely.’ In Aesch. Eum. 726 we may look upon π⋯ντως as a sort of translation of ἄλλως.

We have here a fine instance of the fact that Greek and Latin sometimes mixed their ă, and č and did not have that perfect phonetic orthography generally ascribed to them; cf. my remarks in Modern Language Notes ix. col. 262 sq. In ἄλλως as explained above I see a doublet to ὅλος. In Latin we have the forms salvos, sollus, sōlus, sŏleo (Breal, Mém. de la Soc. de Ling. 5, 437 f.). The Greek correspondents are (cf. Thess. )(Dor. cf. κώρα) ὅλος. The confusion of ἅλλο- ‘all’ with ἅλλο- ‘other’ perhaps accounts for the smooth breathing in Ionic .

A most interesting correspondence, never before pointed out so far as I know, is that between and sollemnia, games (performed for the state as a whole?). In regard of the suffixes -μπια and -mnia I remark that in careless articulation the group mn gives m < p > n, e.g. ‘dampnation,’ passim in the English ballad poetry; we sometimes have the spelling sollempnis in Latin MSS. In I would see with a loss of -ν-. The permanence of the group -μν- is not proved by words like where the felt participial suffix -μενος (cf. ) may have had effect. ‘god-mount’ is a name derived I believe from . Liv. i. 5 tells of a sollemne that was established on the Palatine mount,—so in a general name for mountain we can see a regression from ‘mountain-festivals’—and these occurred doubtless at a fixed period (cf. Lat. sŏleo). Attention is called to the variants as in ‘whole.’

In respect of the doublet, salvus ‖ sollus, Thurneysen (K.Z. 28, p. 261) explains ă for ŏ as due to the following v. De Saussure had previously brought together a group of words in which -oF- was represented by Lat. -av-. Who can say which is older? If we assume av to be older, then the explanation is simpler phonetically; the low-back-wide vowel ă is rounded by the following -w- to a something very close to ŏ, mid-back-wide-round. In the same way -ev- in Latin was rounded to -ov- in novus: ν⋯Ϝος.

^{page 457 note 1} Perhaps we are to construe this as .

^{page 458 note 1} The origin of this -περ with relatives may perhaps be found if we consider its practical equivalence with -τε, as in . Then -περ is to be ascribed to the relative stem with labialization, and derived from a primitive Greek . The source of the -ρ is perhaps to be sought in the enclitic περ, e.g. (B. 246) ‘though being very shrill,’ cf. Lat. per-iucundus; we can translate by ‘ever so shrill,’ and ὅσπερ by ‘who so ever’ in English.

^{page 458 note 2} We must reckon with the possibility that after ‘tu’ ‘alio ingenio’ is implicit, thus ‘they have one nature and you have another.’

^{page 458 note 3} Apparent exceptions are λ 417, K 556 (in both cases ἠ⋯ περ) and Λ 395 where a predicate verb is to be inferred from the context.

^{page 458 note 4} There are several points in which the usage of Homeric corresponds with the exclamatory-interrogative uses of ā. Cf. L. and Sh. s.v. āt 1. B. 3 a—b, with L. and Sc. s.v. . Eng. ‘why’ like āt is exclamatory, interrogative and apodotic in conditions. If the comparison of ātque and ἤπερ holds good then we should expect an ^*āt ^*‘than.’ The form at-que may have crowded out the simple form because of quam ‘than,’ just as -que ‘and’ became in general the important part of the compound, āt ‘but’ is perhaps a derived meaning from ‘than’ as in English we have the pair, ‘nothing else but’ ‖ ‘nothing else than.’

^{page note 5} Except that precedes in point of time , we might claim the same phenomenon there.

I note that etymological relation obtains possibly between μᾰλλον ‘rather’ and Lat. mālo ‘prefer,’ which I find it hard to believe a double affection of *mage volo, magvolo ‖ mavolo, mālo. If mālo is not such a compound it then remains to explain mavis, mavult on the analogy of nevis ‖ nonvis, nevult ‖ nonvult etc.