ob-sid-io-; con-tag-iōni-, -iōn-, beside con-tag-io- (touching), § 89. √tag (tangere touch), etc. Cf. suffix -tiōn-, -tiōni- under -ti(§ 98).

2. -ya- as a secondary suffix.

-ya- is frequently used as a secondary suffix in every IndoEuropean language, and hence must have existed as early as the date of the original-language.

Sanskrit. E.g. div-ya- (adj. heavenly), st. div- (heaven); çún-ya- (canine), st. çun- (dog); pitr-ya- (fatherly), st. pitár(father); rahas-yd- (secret), st. ráhas- (ntr. secrecy). Stems in -a lose their final a before -ya-, e.g. dhán-ya- (rich), st. dhána(ntr. riches), etc. By means of step-formation of the underlying stem are formed stems like rtav-yd- (adj. seasonable) from rtù- (season); mádhur-ya- (ntr. sweetness), st. madhurá- (sweet); kāur-ya- (ntr. theft) st. kōrá- (masc. thief), etc.

This sf. is added to a nomen actionis in -tu- (v. post.); this -tu- is thereupon mostly raised to -tav-, but more rarely remains unraised. Thus by the raising of -tu- arises the compound suffix -tav-ya- so commonly used; with unraised -tu- arises the form -tvya-, which occurs in the more archaic period; from this latter there arises, by evaporation of the v from the group toy, the sf. -tya- used in certain cases in the ordinary lang.; by loss of the y arises the form peculiar to the Vedic lang. -tva-. These sff. -tav-ya-, -tv-ya-, -t-ya-, tv-a-, originally identical, have the same function as simple -ya-, namely that of a participium necessitatis.

Before -tav-ya- (or -táv-ya-) most root-vowels are raised, thus all final vowels, and also medial i and u, e.g. dă-tavya-, √da (give); ē-tavya-, vi (go); stō-tavya-, √stu (praise); khēttavyafor *khēd-tavya-, √khid (split); yōk-tavya- for *yog-tavya-, √yug (iungere); but pak-tavya-, √pak (cook); kar-tavya-, √kar (make), etc., without step-formation of root-vowel, because in Sk. a was felt to be a raised vowel, in contradistinction to its weakenings (§ 6). There is found also, in certain cases, the auxiliary vowel

§ 89. i, i, between the end of the root and the suffix, e.g. bōdh-i-tavya-, √budh (know); grah-i-tavya-, √grah (take). Examples of the shortened form of the sff. are found in Vedic kŕ-tvya-, later kŕ-tya-, √kar (make); therefore we must probably assume -tyato have arisen from -tvya-, also in forms in -tya-, which have no parallel Vedic -tvya-, as e.g. in í-tya-, √i (go); stú-tya-, √stu (praise); bhr-tya-, bhar (bear), and the like. It is however strange to find such root-vowels unraised or weakened. Vēdic kár-tva- (faciendus, as ntr. subst. kár-tva-m opus) is accordingly to be explained as formed from *kar-tv-ya-, stem kár-tu- (v. post.) by means of -ya-; here y has been lost, whilst vis retained. The function of this form distinctly points to this explanation.

Since -ya- is identical with -iya-,—the vowel being lengthened before y, as happens frequently (§ 15, 2, a),-and since -iyamay moreover stand for -ya- (§ 15, 2, b), we may confidently assume that the participia necessitatis in -aniya- correspond in their sf. to a fundamental form -an-ya- from *-ana-ya- (the loss of final a in -ana- before -ya- is regular, cf. § 14, 1, b), and therefore are formed by means of the same sf. -ya-, which we have already found used in this function, cf. bhrātr-iya- (brother's son, origly. 'fraternal') and pítr-ya- (paternal); parvat-īya- (hill-) from parvata (hill), and ráth-ya- (waggon-) from rátha- (waggon); etc. The sf. -iya- is thus -ya-. This sf. -iya-=-ya- is added to a nomen actionis in -ana- (as in -tav-ya-, -tv-ya-, it is to a similar one in -tu-; on nouns in -ana- v. post.), which, as aforesaid, regularly loses its final -a before -ya-. This sf. -anīya- also has the root-vowel almost always raised a step, e.g. stav-anīya-, √stu (praise); gay-aniya-, √ģi (conquer); dă-niya-=*dā-anîya-, √da (give); but kar-aniya-, √kar (make); pak-anīya-, √pak (cook), etc. Verb-stems in -aya- lose this whole sf. before -aniya-, e.g. kōr-aniya-, st. kōr-áya- (steal).

In the Vedic we find also sf. -én-ya- thus employed; herein we can hardly fail to recognize a variation of f.f. -an-ya-; ën

has arisen from an probably through the influence of the fol- § 89. lowing y, e.g. uç-énya-, √vaç (wish); yet we find also vár-ēnya-, √var (choose) with accented root.

Greek. Here too -ya- as a secondary sf. is very common, e.g. πάτρ-10- (paternal), πaτρ-iá (f. race), from stem wаτéρ(father); owτnp-to- (saving), σwτηρ-ía (rescue), stem σwτńp(saviour); Téλelo- for *Teλeo-yo- (§ 65, 2, c; complete), stem Téλes- (ntr. sg. Téλos end); aλýleɩa, i.e. *åλŋ0eσ-ya (truth), stem ἀληθέση (adj. true); θαυμάσιο- (wonderful) for *θαυματ-ιο (§ 68, 1, c), stem Oaûpar- (ntr., n. sg. Oaûpa wonder); Sika-10(just), stem díxa- (fem., n. sg. Síêîn justice); πýxv-1- (adj. elllong) from πxʊ- (masc., n. sg. πîxv-s cubit), etc. Here also, before sf. -10- origl. -ya-, the stem-termination o=origl. a is lost, e.g. oúpáv-to- (heavenly), stem ovpavó- (masc., heaven); πотáμ-ιο- (river-), stem πоτаμó- (masc. river), etc. Here belong also the stems of participles and other consonantal stems used in fem. alone, like φέρουσα, i.e. *φεροντ-ψα; λελοιπυΐα, i.e. *λελοιπυσ-ψα, *-υτ-ψα; δότειρα, i.e. *δοτερ-ψα; μέλαινα= *μελαν-ya, etc.

2

The form exactly corresponding to the Sk. part. in -táv-ya- is in Gk. -Téo-, i.e. *-Te-Fyo-, e.g. So-Téo-, Sk. dā-tav-ya-, √do (give); ἰ-τέο-, νι (go); πλεκτέο-, πλεκ (twist); θρεπ-τέο-, ντρεφ (rear) ; πεισ-τέο-, νπιθ, pres. πείθ-ω, aor. ἔ-πιθ-ον (persuade); etc. Perhaps we ought to place here the forms in -ota for -τια, e.g. εὐεργεσία from -τία, cf. εὐεργέτης ; ἀκαθαρσία, cf. ȧκálaρто-s, etc.; this -Ta may in other cases have also been developed from origl. sf. -ti- (§ 98).

1

Latin. -ya- as secondary suffix, e.g. in patr-io- (adj. patrius), fem. subst. patr-ia, from pater-, praetor-io- from praetor-, victor-ia from victor-, scient-ia from scient- (sciens), rēg-io- from reg- (rex); audac-ia from audac- (audax); princip-iofrom princip- (princeps), nefar-io- from nefas, sacerdot-io- from sacerdot-, etc. Here also the origl. final a of stem is lost before -ya-, e.g. domin-io- (dominium) from domino (dominus), somn-io

§ 89. (somnium) from somno- (somnus), colleg-io- (collegium) from collega, etc.

Note 1.-Adjs. in -tīuo-, as nā-tīuo-, ac-tiuo-, have been compared to Sk. -tavya-, though they do not coincide in function. From -tavya- is said to have arisen -taivya-, by insertion of i (from y following) into the preceding syllable, thence -taiva-, -teivo-, -tiuo-. However there is no very strong reason for identifying Lat. -tiuo- with Sk. -tavya-, a comparison which is moreover phonetically very doubtful. It is possible that they are new formations in Latin, from stem nato-, acto-, etc.; cf. noc-iuo- beside nocuo-, and mortuo- for the form *mor-to-, which we should have expected.

Note 2.-In Latin the participium necessitatis ends in -endo-, earlier -undo-, e.g. dic-endo-, dic-undo-, etc. In the former part of this sf., en, un, hence earlier on, we may probably recognize the an of the Sk. an-iya-: the -do- is prob. the sf. -do- so common in Latin, e.g. cali-do-, uali-do-, timi-do-, etc., beside calēre, ualēre, timère, etc.; in uiri-di- beside uirēre, -di- appears. This -do- probably is derived from the same verb-root da (Lat. dare, which has, however, apparently become confounded with origl. ✔dha 'facere'), which we see in cré-do, con-do, etc. Cf. formations like ira-c-un-do-, rubi-c-un-do-, uere-c-un-do-, fa-c-un-do-, in which the relation of necessity is not found,—a relation which is not really essential to forms in -un-do-, -en-do-, -n-do- (Corssen, Krit. Beitr. 120 sqq., and Krit. Nachtr. p. 133 sqq.); -bu-n-doin fur-i-bu-n-do, treme-bu-n-do-, must be treated as a part. necess. of origl. bhu, Lat. fu. Accordingly we recognize in -do-, the latter part of this sf., a new formation in Latin. We can scarcely assume that the f.f. -an-ya- (v. supr.) has in an exceptional way in the first place become *-an-dya- by insertion of d before y, as not unfrequently happens in the languages, e.g. middle-Lat. madius for earlier maius, Gk. (vyóv, i.e. *Syvyov, Sk. and f.f. yugám; from which form *-an-dya- the loss of y would then have to be assumed in the same way as perhaps in minus for *minius, -bus sf. of dat. abl. pl. for *-bius (v. post. declension), so that e.g. coqu-en-do- (coquendus) would stand for *coquen-dyo-, *coquen-yo-, f.f. kakan-ya-Sk. pakaniya- (G. Curt. Gk. Etym. 590 sqq.). However, as regards Lat., interchange of d and is not capable of proof; on the contrary, this language shows a tendency towards accumulation of suffixes and suffixative composition with verb-roots, as e.g. -cro- (laua-cru-m, sepul-cru-m), √kar (make); -bo- (acer-bu-s, mor-bu-s, super-bu-s) for *bhvo

У

from bhu (be); here probably we must reckon -bili- (sta-bili-s, § 89. fle-bili-s, comprehensi-bili-s, flexi-bili-s) and -bulo- (sta-bulu-m, fa-bula), further formations of this -bo- (these forms are otherwise explained by others); -bro-, -bra- (candela-bru-m, light-bearer; in this example probably no one will deny the derivation of -brofrom bhar (bear); uerte-bra, late-bra), from bhar (bear); -gno- (mali-gnu-s), vgan (beget); -ig- (rem-ig-, n. sg. remex), √ag (drive, do).

VI. Stems with suffix -va-.

Stems with sf. -va- are found in every Indo-Europ. language; in Lat. and Sclavonic, amongst others, it is a favourite suffix. Stems in -van- are akin to these, and are seen especially in Sk. The sf. -vant- we treat hereafter separately.

Indo-European original language. Certainly demonstrable is ak-va- (masc. horse), √ak (run; cf. āk-u- quick).

Sanskrit. áç-va- (masc. horse), vaç (cf. ãç-ú- quick); é-va(masc. going), vi (go); pád-va- (masc. way, waggon), √pad (go); pak-vá- (adj. cooked), √pak (cook); úrdh-vá- (directed upwards, raised), i.e. *ardhva- (§ 7, 2), √ardh (grow), etc.

-van- is akin, e.g. pád-van- (masc. way), cf. pád-va-; mád-van(intoxicating), √mad (become intoxicated); ŕk-van- (praising), beside ŕk-vant and rk-vá-, √ark (praise), etc.

Greek. On account of the loss of v in Gk. the sff. in question are hard to recognize. Clearly we may place here e.g.. ππо- (horse) for *ix-Fo-=Lat. equo-, origl. ak-va-, etc. ; πoλλó(many, collateral form to woλú-) from *πoλ-Fo-, f.f. par-va-, Vpar (fill); in some other cases the root cannot be further traced, as e.g. Xai-Fó-Lat. lae-uo- (left); oλo-, Iôn. ovλo(whole), f.f. *ox-Fo-Lat. sollo- from *sol-vo-, Sk. sár-va-, etc.

The sf. -van- is seen in al-Fóv- (aióv lifetime, time), a lengthening from a presupposed ai-van- (cf. Lat. ae-uo-, Goth. ai-va-, Sk. é-va-), vi (go); perhaps also πéπ-ov- (n. sg. masc. Téπov ripe) stands for *TETT-Fov-, cf. Sk. pak-vá-, √πеπ, origl. kak (cook).

Latin. The sf. origl. -va- is common; besides eq-uo- (horse)

$ 90.