STOP Early Journal Content on JSTOR, Free to Anyone in the World This article is one of nearly 500,000 scholarly works digitized and made freely available to everyone in the world by JSTOR. Known as the Early Journal Content, this set of works include research articles, news, letters, and other writings published in more than 200 of the oldest leading academic journals. The works date from the mid-seventeenth to the early twentieth centuries. We encourage people to read and share the Early Journal Content openly and to tell others that this resource exists. People may post this content online or redistribute in any way for non-commercial purposes. Read more about Early Journal Content at http://about.jstor.org/participate-jstor/individuals/early- journal-content . JSTOR is a digital library of academic journals, books, and primary source objects. JSTOR helps people discover, use, and build upon a wide range of content through a powerful research and teaching platform, and preserves this content for future generations. JSTOR is part of ITHAKA, a not-for-profit organization that also includes Ithaka S+R and Portico. For more information about JSTOR, please contact support@jstor.org. STUDIES IN GREEK NOUN-FORMATION Based in part upon material collected by the late A. W. Stratton 1 DENTAL TERMINATIONS III Words in -Ttjs, -ttjtos By Walter Petersen The Greek suffix -rqs, -rr\ro% (-ras, -r tiros), used chiefly in the formation of adjectival abstracts, goes back to an Indo-European -tat-, for it occurs in practically the same meaning in the Aryan languages (Skt. and Av. -tat-) and the Italic 2 (Lat. -tas, gen. -tatis). The analogy of most suffixes whose history can be traced leads to the conclusion that this arose by extending the old abstract suffix -td- by means of -t-, or -ti- in case of the parallel -tati-, as is shown by Auf recht KZ. 1. 159 ff., and accepted by Brugmann, Gr. 2. I. 2 450 f., and most other scholars, although Hirt, Handb. der gr. Laut- und Formenl. 2 292 f., sees in it the remnant of an old IE word related to *tewa 'power,' assuming that -tat- and the synonymous -tut- go back to the same original -Hwat-. A number of Greek words in -ttjt- have equivalent words in the same suffix in Sanskrit, Avestan, or Latin, either with complete correspondence of stem as well as ending, or with minor variations. In quoting these it is self-evident that there is no guaranty of IE origin, for each language may have formed a word independently, particularly Greek and Latin, in which the productivity of the suffix was large, and all the more so when such a word first appears in a late writer or in one who, like Aristotle, had a tendency to indulge in new formations of this type. Brugmann, op. cit. 451, mentions the following four words as occurring in more than one language: okbrrfs ' wholeness ' = Skt. sarvdtdt- Av. ha u rvatat- 'wholeness, perfection' : Skt. sdrva-s, Gr. oXos < 6\fo-s ; anaioTris 'lefthandedness, awkwardness ' = Lat. scaevi- tds : Gr. )8poTrJTa are both themselves old words and correspond to old words of other languages. Three other words common to Greek and Latin are certainly not Indo-European. Cicero himself confesses (Tim. 7. 20) with an apology that he coined medietas, which corresponds exactly to Gr. fMecroT7]s 'middle position' : neaos Lat. medius. Lat. austeritas is borrowed from Gr. avaT-qpbrrjs 'harshness' : awTrjpbs, and Christi- dnitds is either borrowed from Gr. XpKTTiaporrjs 'Christianity,' or formed from Christidnus which is itself borrowed from the Greek. Probably, on the other hand, Gr. Xeior^s 'smoothness' and Lat. levitds have a common origin. This implies that the primitive of the latter was a prehistoric *levos = Gr. Aelos instead of the extant levis, but transfers to the t'-declension were common enough at all times. Two pairs of Greek and Latin words differ only as to grade of the root vowel: aKpbrr)s 'extremity' : &npos summits and Lat. dcritds 'sharpness, force' : deer 1 'sharp'; bpObr-qs 'straightness, correct- ness' : bpdbs and Lat. arduitds 'steepness' : arduus (cf. Meyer- Luebke, ALL. 8. 321) . Common origin of course presupposes original complete identity of form, but the divergence may be due to later analogical changes which affected primitive and derivative in the same way. A number of Greek and Latin words differ only by the addition of an i in the latter, e.g., 9766x77$ 'sweetness' : 77611$ and sudvitds : sudvis. Here too original identity may have been disturbed by subsequent transfer to the ^-declension in Latin, which transfer would naturally affect primitive and derivative alike unless they were no longer associated with each other. Similar are: Papirrjs and gravitds : j3apvs 'heavy,' fipaxvrris and brevitds : 0pa%vs 'short,' yXwc6r77S and dulcitds : yXvKvs 'sweet.' When there are more extensive differences between the suffixal parts of Greek words and those of related languages, we can merely 1 Occurs as o-stem as well as i-stem. Cf. Charisius 63 and 93 P. 46 Walter Petersen speak of the possibility of identical ultimate origin disturbed in the same way as in case of the t-stems. So air\6rr)s is related to Lat. simplicitds, 5e£toT??s to Skt. ddksatdti- and Lat. dexteritas, Zhacfrpbr-qs to Lat. levitds, i\ev9epi6rr}s to Lat. llbertds, 1 irkrjpoTrjs to Lat. plenitas, clKvk6tt]s to Lat. salsitds, Bripibrris to Lat. feritds. 2 From the small number of complete and convincing correspond- ences between Greek and other languages we may conclude that the productivity of the suffix was very slight in IE times, and this is further borne out by its rareness in Sanskrit and early Greek. As opposed to hundreds of these words in later Greek and in Latin, the Homeric poems have only nine: 6.(v)5poTrjs, 5t]iottis, naKorris, veorris, <£t\oT»7s, ppa8vTr)s, TaxvTrjs, iorrjs tottis. In the Sanskrit, which alternated between the suffix forms -tat- and -tati- (the former only dat. instr. loc. sg.), there are only fourteen words ending in one or the other or both, of which two belong to the later language and all the rest are Vedic only. 3 Generally speaking, this slight develop- ment of an abstract suffix 4 is just what might be expected, for the luxuriant development of abstracts is a matter of literary prose, of science, philosophy, theology, literary criticism, etc. 1. -tt]t- in derivatives from adjectives. — Since the suffix -tat- was usually added to an adjective primitive, and since by far the largest group of adjectives from IE times was the o-stems with feminines in -a- or -%-, which would appear with -o- in derivation, it follows that in most words -tat- was preceded by o, so that -otdt- was from the beginning the most common combination. Of the fourteen Sanskrit words in -tat(i)- all except two have the corresponding -a-tdt(i)-, e.gi upardtdt- : upara- 'lower,' sarvdtat{i)- : sdrva- 'complete,' ddk- satdti- : ddksa- 'clever.' Of the nine Homeric examples all except two end in -orris, e.g., KanoTrjs : naicos 'bad,' veoTr/s : veos 'young.' 1 Lat. libertas < *llberitas would correspond to Gr. *eXev0epoT?;s, and thus differs from £Xeu0epi6T7)S only by the absence of the suffixal i, as does Skt. ddksatdti- from Se&drijs. 2 Since Sijpiottjs is derived from the substantive Bripiov, it is obvious that in this instance a common origin is impossible. 3 Cf. Whitney, Skt. Gram. 3 476 I. 4 However, Homer has 33 abstracts in -avvr], which evidently had a more poetic flavor than -tt\t. See pp. 74 f . Studies in Greek Noun-Formation 47 With the latter cf. Lat. novitds : novos 'new.' Similar Latin examples are bonitds : bonus 'good/ dignitas : dignus 'worthy.' From the Avestan might be mentioned: uparatdt-, identical with the Sanskrit word just mentioned, and amdrHatat- : ani9sa- 'immortal.' Of the 658 words in Gr. -ttjs which I have found, all except twenty- eight end in -ottjs, and of these 492, or about three-fourths of all words in -tt)t-, are regular derivatives in -ori)s from adjectives of the o-declension, or at least such adjectives are actually found, and, with rare exceptions, must have been the real primitives. No doubt in case of a few words some speakers associated them with the substantive use of the primitive rather than the adjectival (e.g., 4>'iKos 'friend' instead of 'friendly'), and in other instances such substantives might have been the real primitives, but these would easily be counterbalanced by others which seem to belong to an irregular type only because the real adjective primitive in o does not happen to be extant. Only a few additional older examples of regular derivatives from o-adjectives can be mentioned. In Homer is found 0iX6tt?s : i\os, Sr]ioTr)s : Sri'ios, in Pindar d/3por»;j : a/3poj, in Aeschylus e.g., \eiorris : Xeios, in Herodotus de^ioTijs : §e£ios, KaOapcorrjs : naJBapios, \anTcpbrr)s : "ka/xirpos , oiKeiorTjs : oliceZos, inKpoTrjs : mKpos, ircaroTrjs : TTICTTOS, (TKCLIOTTIS I (TKCLIOS, dTtvbrt]% : s 'swift.' In Aeschylus first appears Tpaxvrrjs 'roughness' : rpaxvs, in Herodotus y\vKVT7]s 'sweetness' : y\vKvs and waxvTr}s ' thickness ' : iraxvs- Similarly, from the Classical period, d/i/3X6r»;s, papvriis, ppaxvTrjs, dacvTijs, Spi.p.vTr]s , evdvrqs, 48 Walter Petersen €vpi>T7]s, drfKvTTfs, dpaLS 'well-fattened, large,' of which the latter was probably substantival in origin (cf. Brugmann-Thumb, Gr. Gr. 474). Consequently there is no combination -trr/s corresponding to the Latin -4tas in qualitas 'quality' : qualis, nobilitas : nobilis, etc. Similarly the Aryan languages lack -itat- except in the Avestan kdvitat- from the substantive tow'-. However, Greek does have one derivative from an exocentric adjective whose last component is a substantive in i, sc. axapioT-qs 'stupidity' (found in Polybius) : axapts, 2 with -iottjs instead of *-itt\s. This word is an example of what has happened in derivation with -T7]T- to every primitive except o- and v-stems. While the expected -vrrjs held its own for the latter, all other words yielded to the immense preponderance of words in -orris, and were somehow assimilated to them, either by the abstraction of -ottjs 3 as a unified suffix and its addition to the stem of the primitive, or by modifying the primitive stem so that -orris was the result. It is the same thing as e.g. tu/iar-6-ets 'bloody' : alp,a aip-ar-os, (rja-6eis 'shady' : aiCLa. The simplest case is that of the vr-stems, for which the unendur- able combination -vr-Trjs and consequent obscuring phonetic changes were avoided by the o-extension. The earliest example is cited from Chrysippus by Plutarch : xapttJ'rorTjs 'gracefulness of manner' : \apieis, -evros. Also two from the Roman and Byzantine periods: dvTorris 'the beingness' : &v, ovtos, iravTorrjs 'universality' : 7rfis, iravros. Cf. Lat. voluntas< *voluntitds : voluns, -untis. On the other hand, the Avestan added -tat- directly to the ni-stem, e.g., in uxsyqstdt- 'condition of increase' : uxsyant- 'increasing.' 1 The primitive irpi after \evK6rris : Xeuicos, itoct6ti)s : savr- after 6X6x175:8X05. Studies in Gbeek Noun-Formation 49 The only other derivative from an adjective in an explosive is a7)\t,K6Tr)s 'childhood, nonage' from the K-stem a0r}Ai£ 'young.' For its formation, exactly like xupwrbrris , cf. Lat. capdcitds : capax, -acts. It is as late as the sixth century a.d. Adjectival n-stems of the Greek are all secondary and have no counterpart in other languages 1 except the comparatives, which are also found in the Germanic. Since this language group does not know the suffix -tat-, Greek words in -ttjt- from z'-stems are without analogy in other languages. In the Classical period are found only three derivatives from non-comparatives, all from Aristotle: appe- vbrrjs 'manhood' : appi\v 'male,' p.tko.vbri]% 'blackness' : p.e\as, -avos 'black,' hbri\% 'unity' : eh, evbs 2 'one.' After the last was patterned its Byzantine compound oifievbTi)$ 'nothingness' : oWels. These words agree in adding -otjjs to the unchanged stem like xapievrorris , unless p.e\a.vbTi)s goes back to a parallel o-adjective related to Skt. malind-s 'dirty,' which would then also be found in compounds like [MeKavb-irrepos. Cf. Brugmann, IF. 9. 366 ff. After the Classical period development took place in two direc- tions. On the one hand the type appevbrrjs was extended to compara- tives: ekaTTOvbrrjs 'a being less' : eXarrcov, neitovbrrjs 'greater magnitude' : ndfav, ttKuovoths 'length of syllable' : wXe'ucv. On the other hand there is found a new type in which the suffix -ov- disappears before -orris, for the first time in repajibr-qs (Theophrastus) 'softness,' derived from the non-comparative repap^v, -ovos. Then also similar derivatives from comparatives: apeioTrjs 'excellence' : apeiaiv 'better,' peibrrjs 'lesser part of a number' : p.eiuv 'less/ TrXetoTTjs 'plurality' : ifkeiuv. Possibly also |8eXrt6r7;s : PeXriuv (Schol. Pind. 0. 1. 5), although the reading is doubtful. All these were no doubt patterned after compounds 3 in which o appeared for op of the first member, as p.ub-po)v , a.Kp.b-derov (aKp.oiv), 'AttoAAo- dwpos ('AttoWwp), which were an IE inheritance. Cf. Brugmann, Gr. 2. I. 2 84 f. 1 Adjectival exocentric compounds with c-substantives as final member could of course occur anywhere. 2 I classify here because there is no trace of the original m-stem in the whole paradigm. 3 Perhaps the immediate patterns were rather the numerous words in -oe\bT7)S : aekr)s, ivapybrtis : ivapyrjs, iinp,epbTrjs : eiU/U€- pijs, OeoTrpeTrbrrjs : deoirpeirijs, wk-qpbrris : ifhqpijt, virepirKrjpbrris : vwepirXrjpris, vyibrrjs : vyiris. In the Thesaurus is also found Suqvenb- tt)s : dirjveKrjs. No doubt a plurality of factors contributed to the prev- alence of this type. If, e.g., *ir\rjpebrr]s would have existed in Attic, its contraction to *ifKt)povTr)s would have isolated it and caused assimilation to the regular type, so that ■KKripbr^s would have resulted after all. Moreover, the darkening of the structure of the oblique cases of the-ek\aa, ivapytia, tin&ptia, bylfia. Studies in Greek Noun-Formation 51 real or apparent irregularities. Thus dirXorTjj 'singleness,' first occurring in Xenophon, seems to be derived from air\6os, airXoDs, but is really a regular derivative from the parallel &7rX6s, -17, -bv, found, e.g., in the Cretan dirXet, and identical with Lat. simplus. 1 Similarly the Byzantine SiwXdrqs is to be referred to SiirXos instead of SurXoos. As to woKKorris (Damasc), it is of course based on the stem iroXXo- (e.g. in gen. sg., iroXXoO) rather than on iroXu-, as p.eyd\6- rrjs on iMeyakov, etc., instead of p.4yas. The late Byzantine yp-qyo- p6rr)s 'watchfulness,' apparently a verbal derivative from yprjyopia, is derived from an adjective yp-qyopos, found, e.g., in the Christian inscription CIG. 8686 rd ypyyopov &s. The Homeric 16x77$ 'will, desire ' is to be derived from a prehistoric adjective *is identical with Skt. i§a-s 'seeking, desiring,' and thus belongs to the most common type. Cf. Pott, Et. Forsch. 1. 269, Curtius, Gr. Et. 6 402, Angermann, Curt. Stud. 3. 122, Boisacq, s.v. Fick 1.* 543 would assume *Fi6tt]s as the original form and thus relate to lejuat, but there is no evidence of a f in any dialect. Her- werden's assumption of a Fiorari (Lex. suppl.) in Alcman is a double error: the fragment is from Alcaeus, and the MS reading is wrqri. Cf. Sommer, Gr. Lautst. 13. As to /Siirrjj 'livelihood' (h. Horn. 8. 10), which is usually con- sidered a derivative from the substantive jSlos 'life,' I believe with Aufrecht, KZ. 1. 160, that it is to be referred to the prehistoric adjectival use of flios as found in Goth, quius ' alive ' and Eng. quick, and, with long 1, in Skt. jiva-s Lat. vlvus, etc., so that this also would belong to the most frequent class of words in -rr\T-. If it is derived from the substantive, the suffix, as Aufrecht says, would be otiose. While this would not in itself be a valid objection, it will be seen below that as such a derivative it would not have a pattern to follow at such an early date. Possibly, however, it might be a singular instance of congeneric assimilation to another word in -tijs derived from an adjective, for in the passage cited nanbTris 'wretchedness, poverty' is opposed to the plentiful livelihood suggested by /3i6tj?j itself. In this case, then, the author of the hymn formed the word himself, and it was due to remodelling an existing filos, (810x17, or fHoros of virtually identical meaning. • Cf. Walde, Lat. Etym. Wiirterb., s.v. 52 Walter Petersen Herodian ir. px>v. Xe§. 40. 5 mentions as a doubtful word havorijs, quoting a fragment of Sophocles: iv S, iravaeaQ' a/jLepicov nSxOcov kclI davorfJTos. He then states it might be sound if it should prove to be a nominal derivative. Modern opinions vary widely. Wackernagel, Gott. Nachr. 1909. 58, thinks SaiorrfTos should be read because of the circumflex on the penult, but the fact that the two words were synony- mous and that the accent of the one influenced the other is no argu- ment for assuming that one was spurious. Liddell and Scott would refer to the Macedonian 8avos 'death,' so that it would mean 'mortality, misery.' However, we certainly cannot accuse Sophocles of pedantically forming a word of this kind from a foreign word not actually in use. The solution seems to me to lie in assuming that the root-vowel was long a, 1 and that the primitive was the adjective Savos 'burnt, dry,' thought of, however, in the active sense 'burning, consuming,' a transfer of meaning frequent enough in the tragedians. Cf. e.g. viroirros 2 'suspecting' in Eur. Hec. 1135, not 'suspected.' Thus the primitive of Savorys would be synonymous with that of Srjiorris, and the derivatives would naturally be synonymous and associated with each other, the former no doubt being patterned after the latter. It is this which accounts for the common accent and the common idea of the burning, consuming struggle rather than the spuriousness of the rarer and later form. As far as can be made out from the corrupt text, Herodian designates as spurious dpffdrrjs, used by Critias 3 for bpp.i\. This mean- ing suggests a relation to the verb bpvvfu, which cannot be that of direct derivation, for a derivation from the future opau or aorist &prj\i!- 'young,' eyKaipidrys : eynaipos 'seasonable,' wxP'ottjs : uxpbs 'pale,' ytiTvioTys 'neighbor- hood' : ydrwv 'neighboring.' The first is to be explained by the association of the synonym erepoxpoia 'difference of color,' unless the real primitive was *irepbxpovos , accidentally not quotable. The second and third were similarly influenced by their synonyms iJXt/cia and a.(j>T]kt.ida., as was eyKcupiorris by eyKaipia and o^xpi-brqs by uxp'ia. For ytirvioT-qs the influence of yeirvia is particularly plain because of the weak grade of the primitive stem-suffix (yeirv- instead of ydrov-), otherwise unknown for words in -ttjt-. These six words in -iottjs suggest that kvektvOipvos , which does not occur before Chrysostomus, may not be the real primitive of the Aristotelian aveXevdepioTrjs , but that the latter is formed from the common aveXevdepos under the influence of its synonym aveXevdepia and the simple eXevdeptos and eXevdepibrr/s. It is also possible that words in -iottjs were at least partially patterned after, e.g., IXevdepidrris, which, though derived from eXevdepios, might have been referred to eXeWepos. More complex associations seem to have caused the following irregularities. Instead of the earlier fiaKporrjs ' length' : naicp6s ' long,' p.riKbT'qs is found in Galen. In this case the superlative firjKiaros as well as the synonym to firJKos were the disturbing elements. In the same way \r)86rr)s 'forgetfulness,' if derived from the adjective Xr/dalos 'forgetful,' lost its at under the influence of its synonym \rj$r) or the verb Xtyflw. However, both of these words may rather be due to substitution of suffixes, for which cf . page 60. When -orris was preceded by a r, the existence of two syllables with the same consonant might lead to simplification by haplology. This would be particularly prone to happen in the oblique cases, in which three successive syllables began with r followed by a vowel, and most of all in the genitive singular, in which the combination to was repeated in the first and third place, as it would, e.g., in aicpa- TOTrjTos. This has been widely assumed to have been the cause of 1 Cf. d-y-yeXiirjjs from the substantive S-y-yeAos, p. 60. 54 Walter Petersen the irregular iroTr)s and ttivvtEs, e.g., by Ebel, KZ. 1. 303, and Brug- mann, Gr. 2. 1. 2 452, and it is really the only probable explanation offered, for the other suggestion of Brugmann (loc. cit.) that they were primary formations (cf. ttottjs and the verb-forms ir«roju(H and iirodrjv, invvTas and ttlvvo-kco) meets with the difficulty that there were no patterns to follow. Derivation with -tat- from verbs was unknown everywhere in the IE languages except in a few late Greek and Latin examples, while iroTrjs was principally Homeric, and at all events so old that the suffix at the time of its formation had displayed small productivity, so that no model for congeneric assimilation could possibly be found. As far, then, as irorris is concerned, it was derived from the verbal adjective ttotos 'drinkable,' and was origi- nally *iroTo-Tir)T- 'drinkability,' and was secondarily applied to that which was drinkable, the drink itself. This derivation receives further support from the fact that of the ten Homeric occurrences nine 1 are found in the genitive singular, where, as we have seen, haplology would take place most easily, and one in the accusative (cr 407), while the nominative singular, where conditions were least favorable, does not occur anywhere. Evidently, then, haplology was so frequent in that case that the short form became the regular form, and was followed more hesitatingly by the other oblique cases, while the nominative remained *iror6rr;$, but became isolated from the other cases and was lost altogether, so that the -irorris of the dictionaries is a form which never existed except as a sporadic irregularity. The genitive, on the other hand, maintained itself because imbedded in stereotyped phrases like edr)rvos t)8e TroTrjros (5 788). In favor of a similar origin of mwras 'wisdom' is the fact that its only occurrence (Anth. P. 7. 490) is also a genitive {■Ki.vvTa.Tos for *7uwr6raTos) . We may, however, allow the possibility that its association with the related synonym irivvTr) may have caused the loss of the to, in which case its occurrence in the genitive would be accidental. Haplology I also conceive to be the cause for the genitive anadap- TrjTosior *aKadapTOTr]Tos 'uncleanliness':&Kd0apros, which, although not found in the MSS, was at one time adopted by Tischendorf in Apoc. 17. 4. While this is probably not the correct reading, and i These are: A 780, T 306, S 788, e 201, t 87, k 58, 379, 384, p 603. Studies in Greek Noun-Formation 55 the fact that it is the only word in which -rt\r- is not preceded by o or v shows it to be merely an occasional formation, it may nevertheless be a genuine form which slipped in from some source unknown to us, and its occurrence in the genitive is a point in its favor. Even in those words which have only one r in the nominative singular the oblique cases would show the succession of syllables which might lead to haplology, and in one instance that seems to have led to a form actually preserved in writing. I am referring to the Cretan accusative reoraOeorara, found in the inscription SGDI. 5011. 9 (cf. Bechtel, BB. 25. 162, Buck, Or. Dial. 70). That this accusative actually comes from vebras is shown by the occurrence of the latter in the next inscription (5012. 6) in the same sense 'body of young men.' In the same line as veora is found also veoras as genitive singular, but to assume that this too is caused by haplology violates the rule that the quality of the second vowel displaces that of the first in all such changes when their qualities differ. Cf . Brugmann- Thumb 160. Possibly some abstract in -to. comparable in form to @iott] was associated closely enough to cause the adoption of its genitive form in -rets. It is evident that any adjectival abstract in -Trjs could suffer composition, as could any other substantive, and it therefore requires no comment when we find, e.g., in Plotinus avro-erepoT^s without intervening adjective *avTO€Ttpos. Similarly aW-ayiorris, 1 avd- upaibT-qs, avT-iaorris, 1 avro-\oyix>T7f$, avTO-p-aKapLOT-qs, avTO-£r)poTr)s, avTO-6K6T7]s, avTO-o^oidT^s , avTO-iroaorris , avTO-irpaoTijs , avro- \brtfs , -iroXvOeo-adeor^s , iroKv- Kvpiorris, ^vn-jxeabrris. All of these words are from late philosophers and theologians, beginning with the second century a.d. Adding the number of words of the different types so far discussed, which include all those which, directly or indirectly, go back to adjective primitives, the result is that 586, or 89 per cent of all words in -ttjs, are derived from adjectives. This coincides rather closely 1 Alongside of aiBayidrris is found avroayidrris with existing primitive airodyios. It is just as probable that aiffayidrris was not made directly from its components, but remodeled from the latter by subsequent application of the principle of elision. Simi- larly abruroTris alongside of abroiaorrft from avToiaos. 56 Walter Petersen with the statistics of Paucker, KZ. 23. 157, for the Latin, who found that over 93 per cent of the Latin words in -tds, -tdtis came from indisputable adjectives. 2. -ti]t- in derivatives from substantives.— Most of the remaining Greek words and almost all of the remaining Latin words in -tat- have substantive primitives, e.g. ave^/ibr-qs 'relationship of cousin' : avepibs 'cousin,' BeoTTjs 'divinity' : 6eos, Lat. civitds 'citizenship' : civis 'citizen,' virginitas 'maidenhood' : virgo. The Sanskrit simi- larly has dstatati- : dsta-m 'home,' cdmtdti- : gam- 'good fortune, blessing,' the Avestan, e.g., daevotat- : daeva- 'false god, demon.' One word, sc. a(v)8poTris 'manhood, strength' (cf. Av. hu-n9r 9 tdt- and see p. 59), Greek seems to have inherited from the parent language. That this, however, was only an occasional formation, and that -tdt- could not have been productive in forming derivatives from sub- stantives, is shown by the absence of the combination -a-tdt- in all languages. The extremely large class of substantives in a would certainly have left such a combination somewhere, just as the o-, u-, and i-stems gave -o-tat-, u-tdt-, -i-tat-. Evidently derivation of words from substantives became more common only after the combi- nations with the common adjectival suffixes had become fixed, and when the need of such derivatives from substantives in a was felt, they were made to end in -or-qs, e.g., Tpaire^oTrjs : rpairt^a, v\6rr)s : ii\r), for a-adjectives were only the feminines of masculine neuter o, and would not have been known as basis of derivation. In Greek this inherited (av)hpoTi)s became isolated early and used only in stereotyped combinations. Even its phonetic peculiarities (p. 59) would prevent its serving as a model for others, and so it remained alone for centuries, unless ^iottjs (p. 51) was derived from the substantive /Slos, or, what is more probable, became associated with it after the loss of its real adjective primitive. However, even so, since its suffix conveyed no meaning, it could not serve as a model for a living category. Consequently this word also did not lead to imitation, and it was not before the end of the fifth century b.c. that the next word in -rr\T- derived from a substantive appeared. This is toef/iorris 'cousinship' : avexpibs, first found in an Attic inscription of the year 408-409 1 (IG. 1. 61. 15). Next come two 1 The word is found in a quoted Draconian law, so that it is really much older than the date of the inscription if the law has been quoted verbatim. Studies in Greek Noun-Formation 57 words in Plato : Kvadbrrjs 'abstract nature of a cup' : kvclOos, and Tpoxeforijs 'abstract quality of a table' : rpairef a. These are clearly the result of conscious philosophical abstraction, as are also the Aristotelian irodoTTjs 'the quality of having feet' : tovs, and vrepbr-qs 'abstract quality of wrepa.' Also in Aristotle, but with a more natural appearance, are found Qr\pibry]s 'bestiality' : Bt\piov 'beast' and Xi7t6t??s 'fatness' : rb Xiiros 'fat.' All other instances are post- Classical. Two or three first appear in the Septuagint, and about 45 in the Roman and Byzantine periods. All of this shows that this type of derivative was a gradual growth, and that even after a few patterns existed other words followed at first only hesitatingly and to some extent self-consciously. In view of the absence of a hard-and-fast dividing line between substantive and adjective and the frequency with which adjectives become substantives, it is not surprising that derivatives in -rt\r- should spread from the former to the latter. Thus kcikottis could be thought of in connection with 6 kolkos ' coward ' as ' the nature of a coward' as well as with «a/c6y 'cowardly,' and $1X6x77$ 'love, friend- ship' might as well be 'the quality of a friend' as 'friendliness.' Or fiea-OTTjs medietas might as well be referred to the neuter rb pioov 'the middle' as to the adjective (iiChbrr)s 'friendship.' Contributing factors were the influence of congeneric words, as was shown for ave\l/t,brrjs, and the fact that derivatives from exocentric adjective compounds could serve as patterns for forming similar derivatives from the substantive final members. This is evidently the cause when the adjective primitive is a compound with privative a- and therefore the opposite of the final substantive. After &.-narbrr]s 'corporeality' : (ros), bevbpbrr)s (bivbpov), 5taj8oXoTi7s (5ta/3oXos), 1 eTapxbrrjs (eirapxos), iraipbrrjs (ercupos), €07?j36r»7$ (tyrifios) , fworTjs (£wov), 1 debrijs (Bebs), lirTorris 2 (tiriros), KadoKi.Kbrr)s ((cafloXucos), 1 KopvaiOTT]s {Kopvcuos) , kvkXottjs (kvk\os), pay KTrpbrrjs (p.ayit.\brris in the meaning 'love,' which, though derived from the adjective <££Xos, might occasionally have been associated with the substantivized feminine 061X77 'friend' in the same way as with the masculine (p. 57). The only derivative from i-stems is wjyibrrjs 'natura serpentis' : S<£is, cited from Athanasius. For its formation cf. axapt-brris from an exocentric adjective compound (p. 48). 1 In these words adjectives which look like their primitives also exist, but the meaning of the derivative in every case points to the substantive: e.g., 8ia/3oXoTijs is not ' slanderousness,' but 'devilishness.' 2 To be carefully distinguished from iypdrris -ov ' country-man ' and Zmrinjs -ov 'horseman.' 3 Cf . Brugmann, op. tit. 81 f. Studies in Greek Noun-Formation 59 Two late words in -orris are derived from substantives in diph- thongs. In the Byzantine apx^porris 'high-priesthood' : apx<-epevs association with the adjective iepbs 'holy' has led to the loss of the ev. In case of ypaorris 'old-womanishness' : ypavs, found only in the CGL., the treatment of the primitive stem evidently follows that in compounds like the Aristophanic ypao-aopr/s 'scaring old women.' Stems in explosives, like those from adjectives, add -orris to the unchanged stem. From a root- noun in 5 is found iro56rris (Aristotle) : irovs irodos (p. 57), from neuter r-stems : 6pos, while in the second -orris is regularly added to the weak stem. Two words in -tt)t- come from substantival tr-stems: Xittottjs 'fatness' (Aristotle): to Xtxos 'fat,' and a\.v'oTt\% 'faultiness' (CGL.): to vivos. Aside from the similar derivatives from adjectives (p. 50) there were two causes: the similarity of the nominative with that of the masculine o-declension (cf. to p.4pos and 6 p,bpos), and the patterns offered by the alternation of o- and €6ov 9 f., 210 £f. Studies in Greek Noun-Formation 61 the same idea, and since many adverbs are derived from adjectives, the consequent close association of such a pair of words might lead to the spreading of -ti\t- from derivatives of adjectival to those of adverbial primitives. Outside of the Greek this happened only in the Avestan, e.g., avae-tdt- as though 'woeness' : avoi 'woe!,' yavae- tat- 'perpetuity' : yavoi 'forever.' Between these and the five Greek examples there can be no connection, for the latter are all at least as late as the Roman period and form a rather closely associated group of words derived from adverbs of place. Two words in -orris come from adverbs whose appearance sug- gested case-forms of o-adjectives: e/crorTjs 'a being euros' i.e. 'absence' and ivBdr-qs 'the interior' : evdov 'inside.' The later eiravorris 'the upper part' : iiravoi 'above' was assimilated to the prevailing type 1 of words in -tt)t-, whereas *iwavuTris would have stood alone. The two remaining examples end in -v-ttjs because the primitive adverb ended in v (iieral-vTris 'interval' : fiera^v) or looked like the nominative of an u-adjective: iyyvrris 'nearness' : €7711$. 4. -tt)t- in apparent verbal derivatives. — As opposed to many another suffix -tat- retained its secondary character extremely well, and derivatives from verbs are very rare in Greek and Latin and are unknown in the Aryan languages. Latin has only two certain but evidently artificial examples: 2 differitds (Lucr. 4. 636): differo and indulgitas (Caelius Antipater) : indulgeo. Greek also has only a very few (mostly doubtful) examples which evidently have no con- nection with the Latin. The spread of -ttjt- to derivatives from verbs implies the existence of patterns which, although formed from nominal primitives, could be associated with a verb secondarily, because they developed the notion of activity which is an intimate bond of association between abstract nouns and verbs of the same root. Thus CkoTr\s may have been sometimes associated with 0i\&o 'I love' instead of <£t\os when it designated, e.g., 'friendly service, entertainment,' as in o 55 os Kev i\6T7]Ta irapaaxXI- Ot 8r)'ioTr)s 'hostility' : 817 ios became 1 This was helped by the regular association of adverbs in -co with the -o-adjectives from which they were derived. 2 Cf. Stolz, Hist. Gram. 555. 62 Walter Petersen 'battle-strife, slaughter' and could be referred to Sr/iow 'I slay,' all the more because this meaning was the usual one. On the other hand, Idr-qs 'will, desire,' although its meaning implies activity, could not be felt as a verbal derivative because no verb to which it could be referred was in existence. We can be most certain of secondary connection with a verb for tottjs ' drinking, drink ' (p. 54) < *toto-tt}t- : irorbs 'drunk,' originally 'the quality of being drunk, ' and then applied to objects characterized by that quality. After the loss of the second syllable connection with the adjective primitive was no longer possible, and it could only be grouped with verb forms like whrop,ai and lirbdvv. There are only three 1 probable instances, all of them late, of words in -tt\t- which imply association with verbs in formation. From Galen and Jo. Damasc. is cited daKvor^s mordacitas, for which there is in existence no possible primitive except the verb 5aww 'bite,' although we cannot ignore the possibility of a primitive adjective, *5a/»6s 'biting' being accidentally not quotable. Cf. Angermann, loc. tit. Nilus 88 A has aLv6rr]s 'brightness,' which was either formed directly from aiva> ' show, ' or else was due to contamination of the regular synonym avbTr)s from the adjective 4>avos with the verb 4>aivca. The very late p.eya\vi>6Tr)s 'magnitudo' must without question be referred to the verb p,eya\{ivu 'exalt.' Other instances are only apparent. For opcoTTjs, which suggests ftpvvfu, see page 52. As to havOponvbr^s 'incarnation' (Cyrillus), which seems to be derived from the verb ivavSpuwio), it was more probably a true adjectival abstract and felt as 'the condition of being among men,' presupposing either an adjective Hvavdpwiros 2 'among men,' or else the assumption of the existence of such a word. It may also have arisen through the influence of the simplex avdpuTrorvs on ipavdpoyirncris 'incarnation.' 5. The accent of words in -tvt-.- — All except nine Greek words in -tt)t- accent the preceding o or v, e.g., KaKorrjs, i,\6t7js, &kvt7)s, y\vKirr)s, Tpawe^br-qs, eKT6rijj. This corresponds to the accentua- 1 If the reading Sepip6rr)s- ■/) rav SkvSpwv au£7)rr)s>Se^U>Tris, *bQwTi)Tos >8eS-i6rriTos), it would seem that the original place of the accent must have been either the final syllable of the primitive stem or the root. However, an examination of the Greek exceptions throws doubt on this simple conclusion, for five of the nine Homeric words accent the -T7JT-, showing that there was then no tendency whatever to adhere to the later rule. These are: a(v)8porr)s , ^padvrf)s, raxvrf)s, irorrjs, bi}'iorr)s. The later exceptions 1 are: rpaxvrr)$ (Aeschylus+), also accented rpaxvrris, davorrjs (Sophocles), and invvras (Anth. Pal.). There are also traces of an accentuation Kov4>orr)s (regular Kov6rr}s) in Attic, from which Wackernagel, loc. cit. 59, draws the conclusion that its opposite fiapvrrjs, the only possible pattern for its accent, should really be papvrf]s. The explanation offered by Wackernagel, loc. cit. 50 ff ., is an alleged IE law according to which an i, u, or r preceding a suffix could not take the accent, which would then be shifted to the following syllable. Thus, cf. Skt. purutdma- and purii,-, fbhumdnt- and fohu-. Homeric fipaSvrris, raxvrf)s, and a(v)Sporr)s (: *nf-) would therefore show the inherited accent just as would veor-qs, KaKorrjs, and i\6rr)s for derivatives from o-stems. The Attic, it is claimed, retained traces of this law in rpaxvrris and *@apvrr)s, whereas examples contradicting the same were either accented falsely by the scribes or due to analogy of other words. Without entering on a discussion of all the minute and subtle arguments adduced, we may well admit this much, that there was a general IE tendency not to accent i, u, or r, since these vowels were the reduced grade of fuller vowels and were due to loss of accent. Cf . Debrunner, Griech. Wortbild. 77. It was, however, merely a tendency, and continually offset by analogy, congeneric assimilation, etc., so 1 Also xp«7/3tiTas is found, e.g., in Liddell and Scott, but this is pure conjecture. Since it occurs only in an inscription of the third or second century B.C. its accentuation is unknown, but probably regular. 64 Walter Petersen that we need not expect a difference of this kind ever maintained consistently either in IE or Greek times. In the formation of every word its patterns are a much more powerful influence than vague phonetic tendencies, and it is by association with other words that, not only the exceptions, but the regular words must largely be explained. Wackernagel himself explained Kovorr)s as patterned after *fiapvrr)s, but we may just as well explain rpaxvrr)s as patterned after the similar sounding raxvrr)s, and davorr)s is not made suspicious by its following the accent of drj'ioTrjs, but is doing what every new word had to do. Nor is the influence of associated word on accent confined to words in -r-qr-. Since it displaced an earlier -ret-, the two suffixes must at one time have been found alongside of each other in some words, and this would lead to the later words following the accent of the earlier as found, e.g., in wivvrr), fiiori), avrf), yeverf). Thus Tnvvras may have received its accent from ivivvrr), and in prehistoric times, e.g., *Stj?ot^ or *raxvrr) may have caused Srj'iorris or raxvrr)s. Even other associated words with accented suffix might have been contributing factors, e.g., edwdr), eSwSvs, Ppoirvs for irorr)s ; avrf), 4>ovrj, Kv5oi.fj.os for hr)iori)s ; ahcr) for a{v)hpori)s ; crirovSr) for raxvrris. 6. Some semantic aspects of -ri\r-. — As far as the mere occurrence of meanings is concerned, one may expect the same types for -ri\r- as for other suffixes forming adjectival abstracts. They all usually designate an attribute or state or condition suggested by the primi- tive adjective, but sometimes, on the one hand, come to designate an action, which brings them in contact with verbal abstracts, sometimes, on the other hand, become concrete and designate either objects or individual persons or groups (collectives) characterized by the quality of the primitive adjective. 1 If -rrjr- displays any impor- tant peculiarity of use, it is merely this, that in contrast to most suf- fixes of the kind these developments are of the greatest rarity. More than any other formative -rr/r- has the definite function of forming abstracts from adjectives and of designating an attribute or state with only a rare variation due to the development of individual words and influence of equivalent suffixes. 1 Cf. Brugmann, op. cit., 641 1. Studies in Greek Noun-Formation 65 The most characteristic function of an adjectival abstract and the one farthest removed from verbal ones is the designation of a permanent characteristic or attribute. This function is shown, e.g., in the following groups of words from the Classical period: words designating color or brightness, as epvdpoTrjs and TvppoTrjs 'redness,' xKupbrrjs 'greenness,' y\avn6Ti)s 'blueness, greyness,' irokibr-qs 'grey- ness,' \evKOT7]s 'whiteness,' ixeKavoTrjs 'blackness,' Aa/tTrpor^s 'bright- ness,' (j>awTt]s 'darkness'; words designating taste qualities, as y'hvicvTrjs 'sweetness,' o^vrrjs and v6tt)s 'sourness,' akfivpoT-qs and akvKorrjs 'saltness,' tlkpottjs 'bitterness' ; words designating qualities given by the cutaneous senses, as fiapvTrjs 'heaviness,' kov6- Tt)s 'lightness,' airaXorris, fiaXaicoTrjs , and p.a\6an6Tr)s 'softness,' (TKKrjpoTrjs and crrepeor^j 'hardness,' dp/3Ai>T?7s 'bluntness,' rpaxvTt]s 'sharpness,' ffreppor-qs 'solidity,' vypbr-qs 'liquidity,' fleppoTTjs 'warmth,' xf/vxpbrris 'coldness'; words designating speed of move- ment, as Ppadvrrjs 'slowness,' raxurijs and ukvttjs 'swiftness'; words designating spatial attributes, as fiaKporris 'length,' fipaxvT-qs 'shortness,' evpvrrjs 'width,' arevoT-qs 'narrowness,' waxiir-qs 'thick- ness,' XewTorrjs 'thinness,' pi/cporj/s 'smallness,' evdbrrjs 'straightness,' fiKaiaorris and KafjnrvXorrjs 'crookedness,' aTpoyyvhoTqs 'roundness,' KvpTOTTjs 'convexity,' /cotAori/s 'concavity'; words designating mental and moral qualities of man, as Sejiorrjs and Seivdrrjs 'cleverness,' (XKaioT-qs 'awkwardness,' av\6rr)s 'want of skill, meanness,' kcikott/s 'badness, cowardice,' xPW~brr}s 'goodness,' /cocrpiorj/j 'orderliness,' adeoTrjs 'ungodliness,' av8peu>Trjs 'manliness, daring,' 8padpboir6Tr)s 'joint nature of man,' a-wnaTOTrjs 'corpo- reality,' vkSrrjs 'materiality,' ypvxoT-qs 'spirituality.' Others look more natural, even though many of these were no doubt coined consciously. Aristotle has Qr)pd>Tt)s 'bestiality' and XmtAtijs 'fatness.' Later are ayyekbrris 'nature of the angels,' avbpbrrjs 1 'manliness,' foafidkoT-qs 'devilishness,' fworr/s 'animal nature,' debris 'divine nature,' kvk\6- rrjs 2 'circularity,' voottjs 'intellectuality,' o4>i6tt]s 'snakelike nature,' SejSacrror^s 'dignity of SejSaoros,' Xptcror^s 'nature of Christ, Christhood.' For similar words formed from nouns of relationship the suffix probably became the bearer of this part of the meaning, as in ave\{/t,6- Tt)s 'cousinship,' dSeX^or^j 'relation of brothers and sisters,' vibrqs 'sonhood.' Between a permanent attribute and a temporary or at least inessential state or condition no hard-and-fast line can be drawn. Very many adjectives and adjectival abstracts denote now one, now the other, varying according to the object referred to or other condi- tions. Thus 'white' and 'whiteness' designate an attribute of snow, but a state or condition of the human hair, ' coldness ' an attribute of ice, but a condition of the human body, Gr. Kanbryis an attribute when used of men's character (B 368), but a condition when used of distress or misery (e 414) ; TaweivoT^s ' lowness ' is an attribute when used of land (Diod. 1. 31), but a condition when applied to the miser- able plight of the Athenians at Syracuse in Thuc. 7. 75. The following words (from the Classical period unless stated otherwise) designate a state or condition exclusively or at least ordinarily; vebrrjs 'youth,' vqwibTrjs 'childhood,' irpeafivTas (CIG. 2448. 4. 28, 6. 29) 'age'; evebrrjs 'dumbness,' koxjwttjs 'deafness,' tv4>\6tt)s 'blindness,' xwX6t??s (first in Plutarch) 'lameness'; o^Xiottjj 'wretchedness,' p-aKapibr-qs 'happiness,' p,apy&rj)s 'raging passion, madness,' dXXorpiir^s 'estrangement'; aKparor^s 'unmixed state,' 1 Probably not a direct descendant of Homeric i.(v)Sporfis with short initial syllable. See p. 60. 8 Association with the adjective k6k\ios 'circular' may have played a part in the formation of this word. Studies in Greek Noun-Formation 67 eroifjJ>Ttjs 'preparedness/ Kevbrys 'emptiness,' ^porijs 'dryness,' doKepbrr/s 'muddiness,' airavibTrfs 'scarcity,' (raivpor-qs 'putridity.' Also two of the derivatives from adverbs (chap, iii) designate a condition: iyyvrrjs 'nearness' and Iktott\% 'absence.' Others again are derived from substantives. When the primitive itself is abstract the meaning of the derivative differs little if at all. Thus \oiphrt\% 'pestilent condition' is close to \oipbs 'pestilence,' owbrqs 'faulti- ness' to rd vivos 'fault, harm,' Xtjjuottjs 'soreness of eyes' is equivalent to \yftri, (TxqixaTOTris to irxwa. Here, of course, it will be difficult to distinguish from congeneric attraction and substitution of suffixes. When the primitive is concrete the derivative distinctly desig- nates a condition or state whose essential characteristics are those of the primitive substantive, as irvpor-qs 'fiery condition, heat' : irvp 'fire,' Popfiopdrtis 'muddiness' : pbpfiopos. Oftener the primitive designates a person: iraidibrtis 'childhood' : ircudiov, €»?£6t?7s 'pu- berty' : &£t7|8os, peavioTijs 'youth' : veavias, ypabrr/s 'old-womanhood': ypavs, Tarpbrris 'paternity' : rar^p, ercupbrris 'companionship' : ercupos, airekevdepbrris ' state of a freedman ' : airekebdepos. A special group designates the official position of some dignitary: apxiepbrris 'high-priesthood,' iirapxbTijs 'the government of the tirapxos or prefect,' inrapxbTrjs 'office of the virapxos,' KadoXtKbrris 'office of the Ka0o\iKos or controller,' payia-Tpbrris 'office of the pay itTTpos.' None of these words antedate the Christian Era. One of the principal semantic points of contact of adjectival abstracts is with action nouns or verbal abstracts, inasmuch as the notion of an attribute or permanent quality can easily pass over into that of an intermittent quality or of an activity which is the result of such a quality. 1 This development has not been undergone by those abstracts which are derived from adjectives expressing a tendency to certain kinds of action, at least not as long as the tendency is stressed, for such a tendency is a true attribute. Thus Iraporrjs ' tendency to go forward ' (cf . et/u) , bpy Chords ' tendency to anger, irascibility' (cf. bpyifa), mOavbrris 'persuasiveness' (cf. 1 Cf. Paul, Principien der Sprachgeschicht#, 335. 68 Walter Petersen irelflu), ixiprikoTris ' imitativeness ' (cf. juijueojuai), even though indi- rectly associated with the verbs mentioned, are true adjectival abstracts of ira/xos, bpyiXos, TiBavbs, p.Lp,rfKbs. The action itself came to be designated, probably regularly, by the following words already mentioned : Homeric Iotijs and St/iot^s (pp. 61 f.), davorf/s (Sophocles, p. 52), and dpaoTtjs (p. 52). As a word which was an action noun occasionally was mentioned Homeric i\oTris (p. 61), to which might be added ttot^s (A 780?). Other examples of an occasional idea of activity from the Classical period are: PhuSttis 'violence' used of bringing a man before the courts in Antiph. 130. 16 rj tovtuv piaibrris Kai Trapavop.ia. kov4>6tt]s ' lightness ' is 'lightening, relief (poxQav) Eur. frg. 119. \ap.irpbTi)s 'brightness, splendor' is 'splendid conduct' in Dem. 565. 22 ris ovv rj \ap.wpbT7]s , rj rives ai \rjrovpyiai Kai ra ' ava\6>p.aTa tovtov ; oIk€iot»/s 'friendliness' is rendered 'living together' Isocr. 216 C etkero ri)v otK«6r»7Ta 'EXe^s. irapaopbTi)s 'awkwardness' (only Plato Tim. 87 E) rather denotes awkward movements: Sia rr\v irapa^opbrriTa irT&fiaTa irapexov. Later examples of words in -tijt- with the notion of activity are: aixpoTi]s 'ugliness' in the sense of 'fellatio,' Schol. Ar. Ran. 1308. aXeor»7$ 'an assemblage,' Galen Lex. Hippocr. §e£ior?7s in the sense 'greeting,' Paus. 7. 7. 5 TpoTiveiv .... eiri SeijiorT/ri. SoXioTrjs in the sense 'deceitful action,' LXX Ps. 37. 13 SoXior^ras okqv rr/v rjp.epav enekir-qcrav. £Xeeiv6T7]s in the meaning 'pity,' Schol. Eur. Or. 960. Kparaiorris in the meaning ' power,' LXX Ps. 45. 4 erapaxOrfaav ra opt] kv rfj KparaibrriTi. avrov. Similarly Kvpiorrfs 'dominion' in Col. 1. 16 eire KuptorT/res 1 elre apxai ent Qovcriai. \rjpbrris as ' foolish talk,' Schol. Ar. Nub. 783. Tpofiibrris ' a former life or way of living,' e.g., Clem. Al. 460 e£ idvwv Kai ti)s irpopibrriTOS eKdvqs eiri ri)v ■k'lcttiv bppqaas. Because these words approached the category of action nouns occasionally they were not necessarily felt as verbal derivatives, and as often as not a probably associated verb was not even in existence. On the other hand, such secondary association with verbs might be possible even in spite of the existence of real nominal primitives in 1 Here the plural seems to be an indication of the idea of activity, i.e. it represents different manifestations of a quality, which brings it close to the verbal idea. Studies in Greek Noun-Formation 69 cases like \rip6rrjs (with Xijpe'w), oiKovpbrqs (with oiKovpko), and irpofiibr-qs (with xpo/3i6w). In so far as a few real derivatives in -tijt- from verbs seem to have been patterned after such instances, they have been mentioned in chapter iv. Although it is a natural and common development for adjectival abstracts to become collectives, inasmuch as the word designating an attribute may be used to designate everything characterized by such an attribute, collective use of Gr. -ttjt- is very rare and never became productive in any sense of the word. This is in distinct contrast to Lat. -tat-, which shows such common examples as civitds 'state' : civis 'citizen,' nobilitas 'nobility,' familiaritds 'friends.' The only Classical example of collective use of Gr. -ttjt- is veor-qs 'youth,' which, though usually abstract, is repeatedly used of young men, first in Pindar I. 8. 150. In Crete this use actually became stereo- typed, so that 6 vebras designated a body of young men in an official position, a counterpart to the yepovaia of Sparta. Compare the inscriptions SGDI 5011 f. (first half of third century B.C.). Later examples of collectives designating persons are a8e\0TT]s 'brother- hood' as used in the New Testament (I Pet. 2. 17; 5. 9) and b.v6pwirb- rri$ 'humanity,' which means 'genus humanum' as used by Philo (qu. det. 21). A collective also is fivpioTtjs, equivalent to jwpids in the phrase iv HvpidrrjTi 'countless times' (LXX Sap. 12. 22). Possibly also KapdioTTjs 'praecordia' (Gloss.): Kapdia 'heart' would be somehow a derailment of the collective use, but it looks as though -ott)s was a suffix of appurtenance. Since no citation is given, so that we are ignorant of its exact application, since it is utterly without analogy, and since we do not even find a possibly associated word which might have served as pattern, it is useless to attempt an explanation, and probably it was a false form. Equally rare is the use of abstracts in -ti)t- to designate indi- vidual persons, except when due to conscious metonymy. The only instance before late Roman times seems to be 4>Ck6ti}$ 'love' as pre- supposed by the diminutive 4>ikoTTapiov l avXijra ' darling flute-player' 1 Aside from the hypocoristic doubling of the r, this word is irregular because the primitive <£iX6n;5 appears to have been treated like a -ri?- stem, which is intelligible because nominative vocative singular did not show the second t in ^iXotjjt-, and in the use here presupposed the other cases occurred only rarely. 70 Walter Petersen in Aristophanes (Eccl. 891). Cf. also Nicet. Poet. 1. 278 vvktI ■nrei,5o£6r?7s 'Gloriousness' (Nil.), rnxepdr^s 'Kindness' (Euseb., Athan.), 8avp.ai\eia, both forms being guaranteed by the meter in some passages. Cf. Liddell and Scott s.v. Studies in Greek Noun-Formation 73 from the Classical period, while other similar abstracts are even of primary formation, e.g., p.avia : p,aivofiai 'I am mad,' wevla : wivop.ai 'I am poor.' Similarly, collectives are represented by numerous old and common examples like parpia 'brotherhood, clan,' iraipia 'club,' Koirpla 'dunghill' (: ubirpos 'dung'), oida 'house' (: olnos ' room ') . Of the many other unambiguously concrete words might be mentioned ecrria 'hearth,' Kovia 'dust,' KoiKla 'belly,' ovaia 'wealth, property.' Aside from this difference of the greater extent of its sphere of usage, it seems that words in -ia were more popular and free from the frigid learned tone which characterized most of those in -tt]t-. See pp. 74 f. Of the occasionally competing suffixes I shall mention only -eo--, 1 which, although found in different uses in the great majority of words, forms a group of synonyms to words in -VT-qs, They were mostly of early poetic origin and were being displaced in prose by words in -ti\t-. Cf. the following pairs with writers in which first found: evpos (Homer) and evpvrrjs (Hippocrates), dpacros (Homer) and 6paTt)s (Hippocrates, Thucydides), ir&xos (Homer) and iraxi>Tr]s (Herodotus), /Sapos (Aeschylus) and PapiiTijs (Thucydides), pados (Aeschylus) and fiadbT-qs (Lucian), ifKaros (Simonides ?) and irka.TVTr)s (Hippocrates). On the other hand, both Ta%os and raxvrris are Homeric, while /3p&8os, found first in Xenophon, is even opposed to ppaSvrris in Homer, and dacros (Alciphron) is later than Saa-iirrjs (Aristotle). There are also three similar already Homeric words which are derived from irregular adjectives and are displaced by later forms in -ott?s instead of -vrr]s. Cf. nrJKos and ^kottjs (Galen) and naKpoTrjs (Aristotle), k&Wos and koKottis (Chrysippus), piyedos and jue-yaXoT??* (Chrysippus). 2 The only suffix of any importance which competes with -tt)t- and forms only or almost exclusively abstract words, is -avvt\, e.g. in 8t,Kaiopwv and au>^>poa\ivT\ : aixjipwv, even though *oXit6^>pwc does not occur. Studies in Gbeek Noun-Formation 75 not probable that many of these were actually formed by them, for they designate characteristics like goodness, bravery, swiftness, and slowness, which would be noticed by everyone, a fact which points toward popular origin. Most of them were therefore words of general currency, many of which are met for the first time in poets simply because at the time of the epic and lyric poets and Aeschylus there was as yet no prose literature in which they could appear. On the other hand, sixty-six of eighty-eight Classical words in -avvq, or 75 per cent, are found for the first time in poetry, and of the twenty-two for which prose is the earliest source, all except five 1 were derived from v-stems, for which, as we have seen, it was the regular abstract-forming suffix. It was, then, just as rare for prose writers to venture to coin or use new words in -i\oTr]s (3611. 4, Cos). 2 The other inscriptions also have a rather small number of examples, but not a single new formation, except TrpecrpvTas just mentioned. Of the other examples one group is found in metrical inscriptions, which betrays Homeric influence, as do also these words themselves: veorris (IG. 12. 9. 1195, Oreus; 9. 2. 648. 6, Larissa Pel.), QiUtvs (IG. 12. 5. 591. 5, Ceos; 7. 1885. 3, Boeotia), tclxvtvs (IG. 12. 7. 117. 11, Arcesine), ibr^s (CIG. 3557, Mysia), pi6rr)s (CIG. 6206, 6290, Rome). The remaining examples seem to belong mostly to the conventional language of official documents or of the church: &,yvbrr)s (CIG. 1133. 15, Argos), b.v6poyirbrr)<; (CIG. 8964), ApxawrTjs (Ditt 2 . 376. 42, Edict of Nero to Corinth), 5«wt??s (IG. 12. 2. 46. 6, Mytilene), eavfj.aaidrris (CIG. 3467. 10), debris (CIG. 8964), labras (IG. 5. 1. 1432. 41, 43, Messene), Kaflapeioras (IG. 5. 1. 1432. 10, 15, 43, Messene), KadaptArris (probably IG. 12. 8. 16. 10, Hephaestia), kootu6t?7S (Ditt 2 . 519. 31, 38, 58, Athens; IG. 12. 7. 240. 18, Minoa; etc.), ~ka.p.irp6TT)s (e.g. IG. 5. 1. 464. 9, Laconia), ixeaorris* (Ditt 3 . 1125. 10, Attica), vebras (IG. 5. 1. 1427. 4, Messene), baibr-qs (e.g. Insc. Priene 61. 12; IG. 9. 2. 1104. 32, Magnesia), irocrbrris (IG. 14. 956 A 11, probably also 7. 2227. 3, Boeotia), irpabrris (IG. 12. 7. 240. 19, Minoa; possibly ibid. 401. 1, Aegiale), Trpavrys (CIG. 2788, Caria), aenvbrris (e.g. IG. 12. 1. 84. 6, Rhodos), xPWrbrris (IG. 12. 7. 395. 28, Aegiale; ibid. 5. 2. 269. 11, Mantinea), d^rrjs (Insc. Priene 17. 8, 29; Ditt 2 . 226. 108, 350. 15). 1 Also in the form o£k«A7tjs e.g., Insc. Priene 55. 27. It is by far the most fre- quent in inscriptions of all words in -tijt-. 2 Restored by conjecture, but very doubtful because other occurrences of this word in inscriptions are metrical. 3 Here the use of the word in -tijt- is due to the philosophical style of the inscrip- tion. Studies in Greek Noun-Formation 77 WORD-LIST ypaorrfi CGL., Gloss, [p. 59. irpaorrjS Lys.+[p. 76. avro- Eustath. [p. 55. o-Tpa^oTijs Eustath. avrj^orrfi Pach., Elias Cret. ifjyflfSorrrs Theophil. [p. 58. pai/Sorj/s Eustath. o-TtX/SoTTjs Plut.+ o-Ka/i/SoTr/s Athanas. Ko\ofioTi)s Plut . + oXiyoTijs Plato + irapakoyoTrjS Nicet. Chon. d&oAoyoTj/s Hippoc. apyorrf; Epiph. /AapyoTifi Soph. + XaipapyoTij'S Philo ivapyorrji Poll. [p. 50. yopyorrjs Hermog.+ evSoTrjs Dionys. Areop. 121 B, Max. Conf. Schol. [pp. 61, 71. TroSorrjs Aristot. [p. 59. draSamjs Did. A. 404 B. flcdTTjs Plut.+[pp. 58, 76. d- Plato+ iroXvdtoa- (?) Orig. [p. 55. in*- Did. A. 321 B. tXa9r)vaioTrji Galen. yevvaioTrjs Eur., Thuc.+[p. 70. dpaiorijs HippOC. + i&paioTr)? Cornut., Clem. A1.+ oKtpaiorijs Polyb.+ XadpaioTrji Procop. 8iKpaioT»/s Hippoc. 78 Walter Petersen wpawTrys Xen.+ aid- Greg. Cypr. [p. 55. fuxraiorrji LXX+ d- Diog. L. [p. 55. KpaTaiorrjs LXX+ [p. 68. i- Eustath. aporepo- Theod. Stud. ■y£\oioT7js Athen., Cyrill. aXXoiorrfs Hippoc, Plato 6/toidrijs Plato + &v- Plato + airo- Dionys. Areop. 820 A, 913 D, Hermias [p. 55. d<£- Amphiloch. lrowTrfi Plato + oiroiorrr; Nicomach. Geras. eTtpoidrrjs Plato + erepoxpoioTrjs Pyrrho ap. Diog. L. iriorrjs Hippoc.+ -rjmoTijs Hecat. Abd. ap. Joseph. + vrpnoTrfi Plato + Ka9apioTr)<: Herod.+ [p. 76. frnKapioTiji Plato+ [p. 70. fivpto- Jo. Chrys. [p. 55. airo- Jo. Chrys. [p. 55. a^apioTTfi Polyb. [p. 48. Aypiorris Xen . + eXevOepioTrp Plato + [pp. 46, 53. av- Aristot., Schol. Aristoph. [p. 53. cirificpiorijs false reading for hn- litporrf; Iambi, (see Thesaurus). OijpioTiji Aristot., Metop. Pyth. ap. Stob. [pp. 46, 57, 66. Spaa-TrjpiOTrj's Eustath.+ 6\e8pi6rr]i Phot. (Kum.). KtuptoTrf; Jo. Act. in Ideler Phys. et Med. Gr. 2. 185. iy- Schol. Eur. [p. 53. emjaopioTijs Iambi. incrpi6rrit\6rrp N.T.+[p. 50. /8c/8ijAotijs Amphiloch. dSrjAoTijs Prot. ap. Diog. L.+ iir-aWrjXoTrjS Apollon. Dysc. irap-a.XXr)X6rr)S Apollon. DySC. KaT-aAA.ijAoT7js Apollon. Dyse., Drac. d- Apollon. Dysc. /iu/i«jAotijs Suid. [p. 68. i?s Hom.+[pp. 57, 69f., 76. 8eo- Menand. Rhet. i^iAottjs Hippoc. + kvkAotjjs Pseudochrys. [p. 58. i^eAAoT^s Aristot. + /wXAo'tijs (?) restored by Toup in Hesych. s.v. vcoxeAaa 71-oAAoTrjs Damasc. [p. 51. KvXXorrjs Eustath. oXdrrjs Aristot.+ [p. 44. avro- Procl. [p. 55. 3ia/3oAoT)js Athanas. [p. 58. tfoAonjs Theoph. Ceram. 981 B. airXoTry; Xen.+[pp. 46, 51. SurXoTT)'; Nicet. Chon. [p. 51. vAorrjs Iambi. Myst. 265. 6 [pp. 57, 58. d- Plotin. rpavAoTTjs Aristot. + <^avAdrrjs Eur. + (TTpoyyvXoTrp Plato + dyKvAoVrjs Timario ovXoTfji Aristot. + vwovXoTrf; Byz., e.g., Cyrill. A. KapvirvXoTrii Hipp0C.+ tik^Aottjs Plato + opvWo- Theodos. Gram. [p. 55. IMxXorrp E. M., Schol. Lycophr. ^cdXottjs Plut.+ Tepo^orrjs Theophr. [p. 49. h-apMTrfs Plato+ [PP- 67 f . Studies in Greek Noun-Formation 81 ^pc/itoTrjs Euclid. + A.T7/KOTT7S Schol. Aristoph. [pp. 58, 60. aa-qixoTty: CGL., Gloss. «jrio77ju.dTT7S CGL., Gloss. oKiy-atfioTrji Aristot. oju-ai/AOT^s Gloss. Theophil. Institt. dv-ai/u.0T»7s Aristot. iv-atfiorry; Hippoc. cAAoytfiOTrjs Gloss. aXKi/ioTtji M. Attal. 48 (Kum.) 8oi«^.dTrjs Justin. Mart., Jo. Chrys. vo/u/j.6Tr]i Iambi. yovip-orr/s Dionys. Areop. 981 A+ /iovifioTjjs Procl., Dionys. Areop., Gregor. <^>povtfioT7js Galen. + Xoijuotjjs LXX+[pp. 58, 67. CTOi/xorry; Dem.+ iropifiorrp Eustath., Georg. Cedr. a)pt/^,oTijs Sehol. Horn. •yi'topi/j.oTTjs Stob.+ crtp-OTrjs Xen . + aiSea-ifioTrp Gloss., Pap. Grenfell 2. 90. 6 (Herwerden) [pp. 70, 75. Xpr)i/j.6rrjs Eustath. Oep/wTi)': Hippoc. + avro- Basil. 8i&via6ti)s Plato + cTt)/xoTj;s Strabo+ avxfJ.6Tii)<; a rejected reading Clem. A. 179 [p. 58. wixoTrp Eur.+[p. 76. dfuonoTris Orig. 2. 1588 C + (TTtyavoTTjs Dion. P.+ Savorjjs Soph. [pp. 52, 63 f. TnOavorrjs Plato + [pp. 67 f. a- Aeschin.+ Xpio-TiavoTry; Zonar. (Kum.) [pp. 45, 58. tKcivonjs Plato + av- Epiph. 1. 176 A, Amphiloch. fieXavoTijs Aristot.+[pp. 48f. /jjavoT7)i Plato + iiravoTr)? Epiph. 2. 104 B [pp. 61, 71. rpavorrp Philo, Plut.+ 4>av6rrp Aristid., Eccl. [p. 62. ayvoTr/s N.T.+[p. 76. (TTtyVOTTfi HippOC. (TTvyvoTrp Alexis+[p. 71. xbthvoTrp Adamant. TreXtSvoTjjs Aretae. [p. 71. evoTrjs Aristot. + [p. 49. oW- Isid. 249 C+[pp. 49, 71. xcvorrjs Plato + d-Ax/xevoTjjs Xen. ei-Xifievorry; Menand. Rhet. 7roX.v-\tfiev6rr)i Menand. Rhet. (TTe(i)vdTrjs Herod. + ya\.r)VOTi)'S Sext. Emp.+[p. 70. amproryfi Epiph. 1. 957 A, Hesych. [p. 50. atVoTjjs Herodian. KaivoTr)'; Thuc. + KeAatvoTrjs Gloss. (fmivorrii Nil. 88 A [p. 62. Swords Antipho, Thuc.+ [p. 76. eXenvoT^s Schol. Eur., Eccl. [p. 68. ■jro6a.voTrfi Maxim. Conf. 2. 408 C [p. 70. TairuvoTrft Herod. +[pp. 66, 71. a-KOTuvorrpi Plato, Manass. XiBivoTip Fr. Here. (Lidd. and Sc.) ovhafuvorrf; Eustath. Koivorrji Andoc, Xen.+[p. 72. o-tvoTrjs CGL., Gloss, [pp. 60, 67. d- Eunap. [p. 50. 8a.KvoTr)s Galen., Jo. Damasc. [p. 62. piKvoTr)? Greg. Nyss., Hesych. TrvKvoTifi Hippoc, Aristoph. + aTfpajuvoTTjs Theophr. •Ttp.voTt]'; Eur.+[p. 76. d- Epiph. 2. 32 C avro- M. Acom. 1.299. 22 (Kum.) [p. 55. 82 Walter Petersen Kpufivorrp (?) an inferior reading Mace. 2. 10. 34 [p. 58. yv/ivorrji LXX, N.T.+ ipvixvorrfi Xen.+ dppevorijs Aristot.+[p. 49. /*ct£ovorqs Iambi, [p. 49. irXtiovorrp Nicomach. Geras. [p. 49. lutvoTip Epiph.+ ihrrrovoTifi Iambi, [p. 49. arikirvorrp Plut., Galen. -+- TCpWVOTTfS LXX+ Xaworifs Xen., Plato + axtvSvvdnjs Galen. fixyaXworrp Epist. Leon. Isaur. [p. 62. aXtpoTyj<; Byz. Studies in Greek Noun-Formation 83 0oA.eporijs Hippoc. riiuparrfi Hippoc. + [p. 70. av- Gloss. «rip*poT?;s Iambi, [p. 50. €p6ri)s Aristot. -f- Plut. + [pp. 46, 50. iircp- Dionys. Areop. 909 C [p. 50. avwopoTrfs Menand. Rhet. £ i/Pottjs Thuc. + ovro- Tit. B. 1192 B [p. 55. trrjpanfi Theod. Prodr. aixrTr)pori)<; Xen., Plato + [p. 45. opcrrr)'S Plato [p. 68. yp-qyoparrfi Th. Lask. _Cod. Par. 3048 (Kum.) [p. 51. " (Tairparrji Plato + XeirpoTijs Jo. Chrys. Aap-irpoTjjs Herod. + [pp. 68, 70, 76. tic- Msc Bononiensis [p. 70. acnrparrfi Anastas. Sin., Gloss. Zach. pap. Xvirparrfi Strabo a-Teppartfi Aristot. + [p. 70. Trvpporrjs Aristot. + irarpoTijs Greg. Nyss., Dionys. Areop. +[p. 60. e/i.-p.erpoTjp Aristaen. vporr]S Philo, Luc.+ ixvpomjs var. lect. for 6\vp6rrfi Polyb. 1. 57. 6, Philo 6^upori;s Polyb. + Ivxvp&rrjs Dionys. H., Philo 6lv- Gloss. i\aoT»;s Ephraem. Syr., Ant.Mon. ypi6rr)s Herodian Epim. 16 [p. 58. o-TtpupoTrp Schol. Horn. S.Se\6rrj i ; Aristot. Kwportfi Hippoc. + KovX° T, ? s J°- Damasc. [pp. 57 f. d- Niceph. Blemm. 769, 772. *>- Niceph. Blemm. 769, 772 jttxot»7« Herm. Vis. 3. 12 yaiupar-q'i Aristot. Ko/ti/foTjjs Isocr. + £