The sf. -ta- often occurs, as we have already remarked, as the § 96. first element of compound suffixes; thus in -ta-ta-, forming superl. in Gk. (§ 106); -ta-ti- in the Sanskrit (Vēd.) secondary sf. -tā-ti- (fem.), in which we have probably to recognize a further formation of the above-mentioned sf. -ta-, fem. -tā, which is used in a similar function to form abstracts, e.g. sarvá-tāti- (totality), st. sárva- (all); devá-tāti- (godhead), stem dēvá(masc. god); vasú-tāti- (wealth), stem vásu- (possession), etc. The rarer sf. form -ta-t-, e.g. dēvá-tāt-=dēvá-tāti-; satyá-tāt(truthfulness), stem satya- (true), etc., is clearly a shortening of -tā-ti-. With regard to sf. -ti- we shall see that even by itself it is shortened to -t- in Sk., Zend, Gk., and Lat.

In Greek this -tat- is much used in a similar function in form -TηT-, e.g. veó-tηt- (n. sg. veóτns fem. youth), stem véo(new, young); piλXó-tnt- (love), píño- (dear); ßpadú-tnt- (slowness), βραδύ- (slow); ἑνότητ- (oneness), stem ἑν-; παντό-τητε (universality), stem πаvт-, etc. Thus here also, as in not a few other cases (e.g. gen. dat. dual.; before sf. -Fevт- § 90), the consonantal stems follow the analogy of the a-stems. In Latin the sf. becomes -tā-ti-, -tā-t-, e.g. dūri-tāti- (hardness), dūro(hard; on i for o, v. § 40); anxie-täti- from anxio- (on ie for ii, v. § 38); ciui-tāti-, ciui-; uetus-tāti-, uetus, etc.

Note.-Acc. to Benfey (Or. und Occ. ii. 521 sqq.) Lat. salutis from *saluot-, itself a shortening of *saluo-tāt-, like Zend haurvat- from haurva-tāt-.

A secondary sf. -ta-na- occurs in Sk. e.g. hyas-tana- (yesterday's), hyas (yesterday), to which Lat. -tino- in such forms as cras-tino-, sērō-tino-, closely corresponds; cf. Iran. -ta-na-, Zend -çanh-, Lith. -tina-.

Concerning sf. -ta-ma-, forming superl. like -ta- and -ta-ta-, v. § 108; on -ta-ra- used in comp. degree, v. § 105. It may be that the sff. -tar-, -tra-, which will be handled in the next section, are likewise contracted forms of -ta- and -ra- combined, for archaic abbreviations of the elements of suffixes are undeniable in some

cases.

§ 97. Stems with sff. -tar-, -tra- ; -tar- forms a nomen agentis and fut. part. act.; -tra- forms nouns which mostly signify an instrument.

That the nouns in origl. -tar- in Indo-Eur. were even at that period employed as pres. and fut. participles, we cannot prove with certainty, because those functions are found only in the Asiatic and S. European divisions of the speech-stem. There was no doubt originally only one formation, whose earliest form has held its ground in those nouns of this kind which are used as words of kinship, i.e. -tar- for all genders, n. sg. masc. and fem. -tar-s, neut. -tar-. This sf. is added immediately to the root, which mostly is raised one step; in case of derived verbs it is added to the verb-stem, e.g. ma-tar- (the 'female producer,' mother), √ma (produce, bring forth); pa-tar- (father), √pa (protect, rule); bhrā-tar- (brother), bhar, bhra (bear, preserve); dā-tar- or perhaps da-tar (giver), √da (give); probably su-tar(woman), su (produce, bear), whence sva-sutar- (woman related, i.e. sister); gan-tar- (begetter), √gan (beget), etc.

The correspondence between the languages tends to prove that already in early times there existed a kindred form in *-tara- (for -tara- used to form comparative, v. post.), whence came -tra-; whilst -tar- represents persons, this -tara-, -tra- was used of things, and hence does not form nomina agentis, but usually indicates the instrument. Formations such as dak-tra(tooth), √dak (bite); gā-tra- (limb), √ga (go); krau-tra- (ear), √kru (hear), etc., can scarcely have been wanting in the origl.language.

The origin of the suffix -tar-, -tra-, is obscure. We conjectured (§ 96), that it is composed of two suffixes -ta- and -ra-, as -mana- from -ma- and -na-; we might make an equation thus, -tar- : *-ta-ra- (tra) :: -man- : ma-na- (mna). As -mana-, -man-, is raised to -māna-, -mān-, so also -tara-, -tar-, is raised in the languages to -tāra- (Lat. -tūro-), -tār- (Lat. -tōr-).

Sanskrit. Suffix -tar-, n. sg. -tā for -tar-s (§ 15, d), acc. § 97. sg. -tar-am, in words expressing kinship, e.g. pi-tár- for *pa-tar(father), ma-tár- (mother), bhrā-tar- (brother), etc. The formations of this kind which were felt as nomina agentis are distinguished by a higher step-formation of -tar- to -tār- (n. sg. masc. -ta- for -tar-s, but acc. sg. -tär-am); this is also shared by svásar- (sister), for *sva-star-, *sva-su-tar- (literally kinswoman). The suffix -tar- belonging to nomina agentis is added to the end of the root. With the exception of medial a, the root vowels are raised one step before this suffix, e.g. dā-tár(dator), √da (give). The fem. affixes -ya, e.g. n. sg. dā-trí, i.e. *da-tryā (§ 15, c), from *dā-tar-yā; instead of the primary form of the suffix, which has held its ground in words expressing relationship (mā-tar- fem., but may be masc. as well) also in feminines, a further formation has here worked its way in; cf. forms such as pi-tr-ya- (fatherly), from pi-tar- (father); kar-tár-, √kar (make); pak-tár-, √pak (cook); bōddhár- for *bōdh-tar(§ 59, 2), √/budh (know), etc. Before this suffix, as e.g. in past part. pass. and elsewhere, many roots show an auxiliary vowel i, more rarely i, e.g. ģan-i-tár-, √ ģan (beget); grah-ì-tár-, √ grah (seize), etc.

These stems in -tar- serve for a periphrastic future, in such a way that the masc. is used for all genders; in pers. 1 and 2 the nominative form of the singular has become stationary (thus passing over likewise into the dual and pl.), and is welded together with the pres. of the verb as (be), while the 3rd pers. does not require the verbal form, e.g.

Sing. 1. dātásmi from dātá asmi (I am [about] to give).

2. datási from dātá asi.

3. dātá (rarely dátásti from dātá asti).

Plur. 1. dātásmas from data (we might have expected

dātáras) smas.

2. dátástha from dātá stha.

3. dätáras.

§ 97.

In the earliest Sanskrit (Ved.) these forms appear also accentuated on the root, e.g. dá-tar-, etc., wherein we ought perhaps to recognize an earlier system of accentuation, since the rule is for accent and step-formation to go together.

Suffix -tra-, almost always neut., n sg. -tra-m, rarely fem., n. sg. -trā, e.g. çró-tra- (ear), √çru (hear); gá-tra- (limb), √ga (go); vás-tra- (garment), √ras (clothe); vak-trá- (mouth), √vak (speak); dáš-tra- masc., and dáš-ṭrā fem., acc. to sound-laws for *dãç-tra-, -trā (tooth), √daç, dãç (bite), etc. Also with auxil. vowel i, e.g. khan-i-tra- (shovel), khan (dig), etc. Moreover the root sometimes appears furnished with the stemtermination a, as in pres., e.g. páta-tra- (wing), √pat (fly), pres. stem páta- (3 sg. páta-ti); kŕnta-ta- (plough), kart (split), pres.-stem krntra- (3 sg. krntá-ti), etc.

Greek. The suffix original -tar- does not serve to express the future-relation; it appears as -Tep- in words of kinship, as -τηρ-, -τоρ-, when forming nomina agentis, in the latter of which formations the feminine is distinguished by the affix -yahere also (cf. § 97). 1. Words of kinship, e.g. wa-тéρ- (father, acc. πατέρ-α), μη-τέρ- (mother, acc. μητέρ-α); 2. nomina agentis, e.g. So-τýp- (giver, acc. So-Tŵp-a), also Sw-Tńρ, √do (give); the fem. is formed from unraised suffix -tar-, dó-тeipa, i.e. *So-Tep-ya, f.f. da-tar-yā; σw-Týp (saviour), stem σw, fem. ow-Teipa, etc. Forms like yeve-Tip, vyev (beget), must probably be held to have stems in original a underlying them (cf. Sanskrit). Beside these also is -Top-=origl. -tar-, ¿ñ-тop(speaker, acc. ῥή-το-α), ν ῥε= ἐρ (speak); Είστορ- (ΐστωρ, oTwρ knower, witness), √Fid (know); Sw-Top- (Swτwρ Hom. Od.= Swτńp), etc.; þρá-тop- (n. pl. øpá-Top-es), origl. bhrā-tar(brother), has become removed from words of kinship in form as well as in meaning (member of a opáтpa). In Topo- we see almost certainly a further formation from -Top-, after the analogy of the a-stem, thus in diák-тopo- (guide, Hom.); ả-λáσ-TOPObeside ȧ-λáo-Top- (malignant, avenger), √λað (forget); a solitary

-TUρ-, -TUρo-, is found in μáp-Tтuρ- (gen. μáρтupos witness), § 97. μáρ-TUρo-, √ originally smar (remember).

The feminines in -Tpia are distinguished from those in -Teipa only by the loss of the a of the suffix original -tar-; from -tar-ya came -trya, i.e. -tria, e.g. πоin-тρia, verbal-stem πoin(make). The secondary formation by suffix -ya- is generally very common here, as e.g. from stem Ta-Téρ- is formed a stem πά-τρ-ιο- (paternal), fem. πα-τρ-ιά (origin, race); σωτήριο(saving), stem ow-Typ-; these secondary formations intruded into the fem. and supplanted the original stem in -tar- with few exceptions.

The feminines in -тpíd- (n. sg. -тpís) are either late-formations peculiar to the Greek, formed by means of a later suffix -¿d-, or (cf. G. Curtius Gk. Etym.3 p. 583 sqq.) -тpıd- is merely a phonetic variation from *Tpy-, so that here a suffix -tri- would have to be presupposed, e.g. avλn-тpid- (flute-player, fem.), verbal-stem avλŋ-; πα-тρíd- (fatherland) from stem πa-тéρ-,

etc.

The suffix origl. -tra- appears as -Tрo-, -Opo- (neut.), -тρа, -Opa (fem.); the aspiration is probably caused by the r, e.g. víπ-тρо- (neut. washing-water) for *vß-тро-, √иß, original nig (retained in via, wash,=*vy-yw, § 63, 1); äро-Tро- (neut. plough), from verbal-stem ȧpo- (plough, in åpó-w, ȧpó-σw, ȧpó-σai), √åp. In ia-Tpó- (masc. healer), verbal-stem ia- (iáoμai heal); Sai-Tpó- (masc. carver), cf. Sal-opaι (divide), nomina agentis are formed in -Tро- (cf. -Tоρо- above). Further Bá-0po- (ntr. base, step), √ẞa, original ga (go); þý-τpa (fem. agreement), ν ρε; μάκτρα (kneading-trough), νμακ (knead, μάσσω *μακψω, generally softened into μαγ); φρά-τρα, Ιon. φρήτρη (clan), √ppa=ḍep, origl. bhra, bhar, cf. Opâ-Top-=origl. bhrā-tar-; кoμń-Opa (sleeping-place), verbal-stem koμа- (коiμάw put to rest), etc. The suffixes -7λ0-, -0λo-, fem. -7λn-, Oλn-, e.g. Xú-To- (neut. liquor, fluid), vxv (pour); Oúo-oxo- (neut. implement for Bacchus-worship), vou, the σ appears in other

=