shows suffixes -as- and -i- interchanged in some stems, just as § 102. in Latin.

The dative of such nouns in -as-· —no longer felt to be a case, and hence shortened-serves as an infinitive in Latin; e.g. ueher-e, f.f. vaghas-ai, Sk. váhas-e, present-stem uehe-, original vagha-, root original vagh; dicer-e, f.f. daikas-ai, present-stem dice-, f.f. daika-, dic; monĕr-e, f.f. manayas-ai, present- and verbal-stem monē-, f.f. manaya-, root original man, etc. In fieri, fierei (also fiere), both from *feies-ei, f.f. dhayas-ai, i has, as often, been retained beside è (e) =original ai. This form is in nowise distinct from the usual infinitive active (cf. L. Lange, über die bildung des lateinischen Infinitivus Praesentis Passivi. Denkschriften der philos. histor. Classe der Kaiserl. Akad. der Wiss. in Wien, Bd. x., and published separately there); fro has indeed mainly an active form; the root of this word is dha (set, do), and fio a present formation in -ya- of intransitivepassive function; the f.f. of fio is therefore *dha-yā-mi, in Sk. with unoriginal weakening of a to i and middle termination dhiya-te from *dhaya-te, or else the final-sound of the root is lost in Sk. and -iya- stands for -ya- (§ 15, b). In either case the Sk. form is late and not original, and useless for the expla nation of the Latin. From dha-ya-mi arose regularly in Latin *fe-io-mi, feio, fio; f.f. of fieri, fiere, is therefore *dhayas-ē; in fieri fi has become fi, not an original shortening, the older fieri being retained by Naeuius, Plautus, Pacuuius.

This formation is in Latin so closely joined to the pres.-stem that, except where the stem has the stem-addition -a-, it omits the -a- of the original suffix -as-, e.g. es-se (posse=pot-esse), f.f. as-s-ai (not *as-as-ai, which would have produced *ese-re, *ere-re), esse for *ed-se, f.f. ad-s-ai, √ed (eat); fer-re for *fer-se; uel-le for *uel-se (§ 77, 1, b); da-re, root and pres.-stem da; fo-re for *fu-re, √fu, u having become o under the influence of the r; i-re, early *ei-re, f.f. ai-s-ai, pres.-stem i, ei, original ai, vi. The analogy of the present has here throughout exercised

§ 102. its influence, and has called forth these new formations peculiar to Latin (perhaps the forms cited were at an earlier period of the language *eses-e, *edes-e, *feres-e, *ueles-e, *eies-e, which would correspond exactly to Sk. forms like asas-ē, adas-ē, bharas-ē, varas-ē, ayas-ē).

This -se is added also to the perf.-stem in -is-, which is found in Latin only (v. post.), e.g. peperis-se, dedis-se, fécis-se, etc. Forms like dixe, uexe, seem to be syncopated, like dixti for dixisti; possibly however they are older forms from the perf.-stem without is (v. post.), and thus stand for *didic-se, *ueueg-se. Whether the full suffix -es- = -as- (*didic-es-e, *dicsis-es-e) ever existed or not, depends on the antiquity of these formations.

Note.-Impetrasse-re, leuasse-re and the like (used only in case of derived-verbs in a, and peculiar to the earlier language only) are used as fut. inf.; 1 sg. would be *impetrasso, etc., cf. facesso, incipisso, etc. The peculiarity of these forms lies therefore not in the suffix, which is the usual one, but in the verbal

stem.

The infinitive forms of the medio-passive in Latin are hard to explain. The assumption of Bopp offends against the soundlaws (vgl. Gramm. iii. § 855, p. 273 sqq.); the form in -i (dici) is held by Bopp to be a curtailed form of the earlier -i-er (dic-i-er), whose er he thinks is a transposition of re=se (acc. of the reflexive; cf. amo-r=*amo-se), thus explaining dici-er from *dici-se; *dici would correspond to Sk. forms like drç-é (§ 87), unless dicier were a shortening of diceri-er from *diceri-re, *dicesi-se (i.e. *daikasai-svam), just as laudari-er is produced by dissimilation from *laudare-er, *laudare-re, *laudase-se, i.e. from inf. act. with se attached, which forms the middle voice in Latin. Notwithstanding that this view recommends itself in that it makes the inf. med. to be formed from inf. act. precisely as the med. generally was formed from the act. in Latin (*amari-se: amare :: *amo-se: amo), it is difficult to see the reason of the transposition of se, re, to er (from *laudare-se,

*dici-se, there would have arisen according to Lat. sound-laws § 102. perhaps a form *laudare-s, *dice-s or dici-s, like laudaris, 2 sg. med., from *laudas-i-se; or also *laudare-re, *dici-re or *dice-re). Pott (the last time in 'Doppelung, etc.,' Lemgo and Detmold, 1862, p. 266 sqq.) makes the division laudarie-r, earlier *laudasie-se, thus assuming no transposition of -se, -re, to -er; -sie he holds to be an earlier termination of the active -re. But how is ie then to be explained? In legier and the like, Pott assumes loss of the first r in consequence of dissimilation; *agerie-r (this -rier is preserved in fer-rier), thence *agrier and agier by loss of the first r, finally *agie, agi [or perhaps *agir, agi?; cf. sis, Umbr. sir, sẽ, from sies]. Also Leo Meyer (vergl. Gr. der griech. und lat. Spr. ii. 124) explains legier from *legerie-r, *legesie-se; laudarier from *laudasie-se, which -sie, -sye, he holds to be "nothing else than a peculiar early infinitive-termination, which may perhaps be closely connected with Sk. -syai, in Vēd. rauhíshyâi [in our spelling rōhišyāi], —for rauhisyâi—(increase), and á-vyathishyâi,—for á-vyathisyâi (not tolerate)." But we hold with Benfey (v. supr.) these infinitives in -syai to be inf. from fut,-stems rõhisya-, vyathisya-. Leo Meyer, raising this objection himself, adds: "Possibly this infinitival sye or more fully esye is likewise an early dative of an old suffix formation asya, a further formation, by means of suffix ya, of the old suffix as, well known in the Lat. act. infin." Moreover Leo Meyer holds it conceivable that forms like ducier may be not shortened from *ducerier, but derived from stems like dūcio- (f.f. daukya- therefore). We should then have to assume fundamental forms perhaps such as *daukyāi-svam [dat.+acc. of reflexive]. From this very uncertainty on Leo Meyer's part it is obvious that none of his conjectures are upheld by any decisive arguments. We should scarcely venture to maintain datives in i, è (ei), from a-stems in Latin; in the dative the forms in question could only have been *laudario-r, *duciō-r, earlier *laudasiō-se, douciō-se. Also the separ

to us.

§ 102. ation of the med. (pass.) inf. from the act. will not approve itself Lange (in his above-named work) takes forms like legier as shortenings for *legi-fier, i.e. as compounded of the pres.stem legi- (legi-t), with infin. fiere, fieri; medial loss of ƒ occurs e.g. in lupis for *lupois from *lupo-fios (v. post. Case); amaui for *ama-fui, amasti for *ama-fuisti (v. post. § 173, 2). Forms like amā-rier, da-rier, he explains from *ama-siere, *da-siere, i.e. from the present-stem and an infinitive *siere, f.f. *-syas-ai, for *esiere, f.f. asya-s-ai, formed from √es (be), pres.-stem *asya(with passive function), like fieri from √dha, pres.-stem dhaya-. The present-formation in -ya- appears indeed in ves in Latin, but has a future, not a passive relation (v. post. § 165 Lat. V.).' So also according to Lange ferrier is formed (f.f. therefore probably *bhar-syasai or perhaps *bharasyasai), whilst in other cases this *-sier, -rier, is added to pres.-stems only which end in a vowel (amā-rier, monē-rier, molli-rier).

The forms legi, amari, are explained by Lange from *legies (from *legi-fiese), *amasies (from *ama-siese), with frequent loss of final s (§ 79), and contraction of ie to i, like later sīm, sīs, for siem, sies. Thus from common fundamental forms have been developed (1) legier, amarier, with the s- of the presupposed *-fiese, *-siese, changed into r, and (2) legi, amari, where the s has been lost.

Accordingly in the Lat. inf. pass. also we should see nothing but infinitives in -se, because they all would be compounded with either infin. fieri, f.f. dhayas-ai, or *siere, f.f. (a)syas-ai. This view also is suspicious in some respects; above all we cannot conceive an inf. of √es (be) with passive function.

G. Schönberg (Zeitschr. xvii. 153 sqq.) has recently explained forms like amarie-r as dat. of stem *amasi+se, from *amäsiai-se, forms like legie-r, on the other hand, as dat. of stem legi- (representing stem leges- of the act.)+se, from *legiai-se, thus separating the latter from as-stems, probably correctly, like Leo Meyer (v. sup.); he has recourse to an interchange of consonantal

stems with i-stems, and of suffix -as- with suffix -i- (adducing § 102. examples). It must, however, be allowed that the like dat. forms of i-stems are otherwise unauthenticated.

Thus Latin infinitives passive (med.)-clearly a recent formation of the language-must probably be treated as not hitherto explained with certainty.

XXI. Stems with suffix -ka-.

The suffix -ka- (cf. pronominal-stem ka-) is not common primarily, but secondarily is on the contrary a very favourite one (e.g. in the function of forming diminutives, cf. Lud. Schwabe, de deminutiuis graecis et latinis liber. Gissae, 1859, p. 44 sqq.). The numerous other suffixes, whose principal element is k, need not be considered here, excepting -ska- (whose s, it is true, is obscure), because it formed one kind of present-stem as early as the original language (§ 165, VI.), e.g. ga-ska-.

Sanskrit. Very rare as a primary-suffix, e.g. in çus-ka(dry) for *suš-ka (§ 55, 2, n), √ çuš (çúš-yati dries), original sus (cf. Lith. saús-a-s, Sclav. such-ů dry, Zend hus-ka-; dhā-ká(masc. receptacle), √dha (set). As secondary suffix common, e.g. sindhu-ka- (adj. derived from Sindhu), stem sindhú- (nom. propr.); putra-ká- (masc. little son), putrá- (masc. son), etc.

Greek. Primary in On-κn (store-place), v0e (place, lay); very common secondarily (cf. Budenz, das Suffix κós im Griechischen. Göttingen, 1858), e.g. φυσι-κό-, φύσι-, θηλυκό-, θῆλυ-, καρδια-κό-, καρδία, λογικό- (if rightly separated thus) λόγο-, etc. Suffix -ισκο- here forms diminutives, e.g. παιδίσκο-ς, παιδίσκη, stem παιδ-, etc.

Latin. Here also but few primary formations can be pointed to with certainty, as pau-co- (adj. few), cf. Gk. πaû-po(small), Goth. fav-a (adj. few); lo-co-, early stlo-co- (masc. place), from stal, stla, further formation from √sta. Secondarily very common, as in Gk., e.g. ciui-co-, stem ciui-, urbi-co-, stem urbi-, belli-co, stem bello-, etc.

$103.