LIBRARY OF THE University of California. (J1FT OK Class I60e 1^*5 THE LATIN THIRD DECLENSION A STUDY IN METAPLASM AND SYNCRETISM A DISSERTATION PRESENTED TO THE FACULTY OF ARTS, LITERATURE, AND SCIENCE OF THE UNIVERSITY OF CHICAGO IN CANDIDACY FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UN TY BY HENRY FARRAR LINSCOTT CHICAGO Zbe 1Hntx>ersitt> of Cbtcago iPreas 1896 CONTENTS. Introduction - .... -5 Metaplasm ... - - - 7-23 1. In General - - ... 7 2. Theory of Double Formation - - - 10 3. The Elements of the Third Declension - - 13 4. Application to the Metaplasm - - - - 19 5. Origin and Development of the Process - - 20 Syncretism 24-34 1. In General .... - - - 24 2. In Italic - - 3° The Case Forms in Detail - 35 INTRODUCTION. The Latin third declension has been selected as the basis for a discussion of metaplasm for the reason that it is a typical instance of such declensional amalgamation. In addition to the presentation of facts bearing directly upon this subject, an effort has been made to discuss the relations of the third declension to the corresponding systems of other languages. The treatment of syncretism is of a more general nature, including the phenomena of all declensions and of the Italic, Latin, and Oscan-Umbrian periods. In the third section, it has been my endeavor to discuss in detail the origins of the case endings peculiar to each lan- guage and to define the metaplastic and syncretic effects, wher- ever manifested. The following authorities have been consulted with greater or less frequency: Balg, G. H.: A Comparative Glossary of the Gothic Language. New York, 1887-89. Breal, Michel : Les Tables Eugubines. Paris, 1875. Breal and Bailey: Dictionaire Etymologique Latin. 3 e Edition, Paris, 1891. Bronisch : Die Oskischen L- und E-Vocale. Leipzig, 1892. Brugmann, K.: Griechische Gratnmatik in Iwan Midler's Handbuch der klassischen Alterthumswissenschaft. Bd. II, 2. Aufl., Miinchen, 1890. Brugmann, K.: Grundriss der vergleichenden Gratnmatik der indogerma- nischen Sprachen. 2 vols., Leipzig, 1886-93. Buck, C. D.: Der Vocalismus der oskischen Sprache. Leipzig, 1892. Biicheler : Umbrica. Bonn, 1883. Corssen : Ueber Aussprache, Vokalismus und Betonung der lateinischen Sprache. 2 vols., 2. Aufl., Leipzig, 1868-70. Delbruck : Vol. 3, Brugmann's Grundriss. Leipzig, 1894. : Syntaktische Forschungen. 5 vols., Halle, 1871-88. Feist: Grundriss der germanischen Etymologic Strassburg, 1888. Fick, A.: Vergleichendes Worterbuch der indogermanischen Sprachen. 3. Aufl., Gdttingen, 1874-76. Georges: Lexicon der lateinischen Wortformen. Leipzig, 1 890. 5 Grassmann : Worterbuch zum Rig-Veda. Leipzig, 1875. Henry, Victor: Precis de grammaire comparee du grec et du latin. Paris, 1888. Jackson : An Avesta Grammar. Part I, Boston, 1892. Kluge : Etymologisches Worterbuch der deutschen Sprache. 5. Aufl., Strass- burg, 1889. Kluge : Vorgeschichte der altgertnanischen Dialekte in Paul's Grundriss der germanischen Philologie. Vol. I, Strassburg, 1889. Meyer, G.: Griechische Grammatik. 2. Aufl., Leipzig, 1886. Meyer, Leo: Vergleichefide Grammatik der griechischen und lateinischen Sprachen. 2. Aufl., Berlin, 1875-77. Mommsen, Theodor: Die unteritalischen Dialekte. Leipzig, 1850. Neue : Formenlehre d.er lateinischen Sprache. 2 vols., 2. Aufl., Berlin, I875-77- 3- Aufl., 1892- Pauli : Altilalische Studien. Vols. I-V, 1883-87. von Planta, Robert: Grammatik der oskisch-utnbrischen Dialekte. Bd. I, Strassburg, 1892. Schweizer-Sidler : Grammatik der lateinischen Sprache. Halle, 1888. Stolz, Fr.: Lateinische Grammatik. 2. Aufl., Miinchen, 1 890. Wharton: Etyma Latina. London, 1890. Whitney, W. D.: Sanskrit Grammar. 2 ed., Boston, 1888. Zvetaieff : Inscriptiones Italia Inferioris Dialecticce. Moscow, 1886. B. B. — Beitrdge zur Kunde der indogermanischen Sprachen, herausgegeben von Ad. Bezzenberger. Vols. I ff. I. F. — Indogermanische Forschungen, Zeitschrift fiir indogermanischen Sprach- und Alterthumskunde. Vols. I ff. K. Z. — Zeitschrift fiir vergleichende Sprachforschung, begriindet von A. Kuhn. Vols. I ff. P. B. B. — Beitrdge zur Geschichte der deutschen Sprache und Literatur, herausgegeben von H. Paul und W. Braune. Vols. I ff. Rh. M. — Rkenisches Museum fiir Philologie. Vols. I ff. M. U. — Morphologische Unlersuchungen auf de?n Gebiete der indogermani- schen Sprachen, von K. Brugmann und H. Osthoff. Vols. I-V. Other works have been consulted, as indicated by the references in the notes. It is a pleasure to acknowledge the assistance of Professor Carl D. Buck, under whose supervision this work has been undertaken and completed. METAPLASM. Metaplasm may be defined as that process by which are trans- ferred or established upon noun or adjective stems any case end- ings original to a declension other than that to which the stems affected belong. In short, the effect of the process is to estab- lish in a given system of declension case forms extraneous in origin, but necessarily identical in function with the forms which to a greater or less extent are displaced. The original endings may be either partially or totally eliminated, may exist as the predominating type still possessed of greater force than the encroaching element, or may not appear, and if appearing may exist in few and sporadic instances. In its origin and development, metaplasm depends upon analogy. 2 The original condition, in any specific instance, must be a series of case endings distinct in form and peculiar each to its particular system of declension. Between these two parallel series of forms, a point of contact must first have been estab- lished. This may result from phonetic change, causing identity of form, 3 from a similarity in meaning or function/ or by reason of circumstantial elements of formation inherited from the parent speech. 5 1 For the subject of metaplasm in general, cf. Brugmann, Gr. II, 722 ; B. Torp, Die Flexion des Pali; Wetter, Zur Geschichte der Nominal-Declination im Russ.; Courtenay, K. S. B., VI, 19 ff.; K. Bojunga, Die Entwicklung der neuhochdeutschen Substantiv-Flexion. 2 Wheeler, Analogy, 9 ff.; Paul, Principien, 95 ; Bartholonut, K. Z., 29, 524 ff.; Brugmann, Gr. II, p. 722. 3 £. g., Goth, /-stems masc. with gen. and dat. sg. after <?-stems, balgs, gen. balgis, dat. balga like dags, dagis, daga. A E. g., Skt. naptr beside napdt after words of relationship. Wheeler, Analogy, 9 ff.; Brugmann, Gr. II, p. 722. -E.g., Skt. usa in beside usa saw to usa s niter s//ia s ; Lith. kirtnis, gen. kirmio, beside Lat. vermis, wherein inherited likeness of form between i- and w-stems has caused a metaplasm. 7 8 THE LATIN THIRD DECLENSION When a point of contact has been established by reason of similarity of form, analogy may occasion a like similarity in the remaining case endings previously distinct in form. Likewise, if the contact be not in form but in meaning or function, the analogy proceeds on the lines observable in the case of analogy from similarity of meaning and diversity of form. 1 The effects of the metaplastic force are of two distinct types, (i) one series of forms may be entirely eliminated, or the type newly created may exist beside the original as a distinct word, e. g., Skt. ndpir beside napdt, or (2) the result may be an amalga- mation, in which traces of each of the two series of forms may be observed. The former type may be characterized as Metaplasm by Transfer; the latter as Metaplasm by Amalgamation. The rep- resentative instances of each type may be enumerated as follows: I. METAPLASM BY TRANSFER. In the Aryan languages, many consonant stems have been transferred to the ^-declension by reason of identity of form at the ace. and inst. sg. In Pali and Prakrit the process has been so extended as nearly to eliminate the consonant declensions. 2 In Sanskrit ndptr has arisen beside napdt by the influence of other words of relationship ; usas has been re-formed on the anal- ogy of nouns in -as as sthas ; from the contact of -vant- and -van- stems have resulted many pairs of forms inflected after either system. 3 Other instances are, in Greek, the extension of -t- and -S- inflexions, e. g. , TSptS- to ifyns, Ihpiv and AeW, -ovtos for -ovos; the transfer of masculine -^j-stems to the a-declension, e. g., SoKpaT^s, 'XoKpa.Trjv ; in Latin, the transfer of «-stems to the ^-de- clension; in Germanic, the passage of the consonant stems into the vocalic declensions, e. g., Goth, fotus, handus, tunpus ; O. H. G.zand,fuoz;* in Balto-Slavic the transfer of consonant stems to 1 Balgs '. dags = balgis '. dagis; sthas '. usas = stham '. usam. 2 Brugmann, Gr. II, p. 723; Lassen, Inst. Ling. Prac, p. 314. Kuhn, Pali Gram., pp. 67 ff. 3 Bartholomae, K. Z., 29, 540 f.; Brugmann, Gr. II, pp. 724 ff. 4 Kluge, Paul's Grundriss I, p. 389. Braune, O. H. G. Gr. 2 § 216. Anm. 1. THE LATIN THIRD DECLENSION 9 the /-declension and the change of /- to /0-stems, e. g., kirmis, gen. kirmio; Skt. krmis. 1 II. METAPLASM BY AMALGAMATION. This type appears in Aryan in the composite declension of -*>«-stems containing forms peculiar to the /-declension, e. g., ace. balinam, beside balibhis, balisu; Av. ka'nitwm beside ka'nibyo and in the extension of //-forms to the vocalic and r-stems ; in Greek in the amalgamation of the -id- and -/- : -//-stems ; in Germanic in the /-declension of adjectives, an amalgam of /- and /0-stem forms, 2 in the confusion of a- and /-stems in English 3 and in such declensional types in Gothic, as frijondam to nora. pi. frijonds : nahtam to nahts* : baurgim to baurgs and dulpais to dulps; in Balto-Slavic in such types as dantis, gen. pi. dantu : szirdis, szirdu; O. tt.jelent, gen. pi. jelenz and in the appearance of / of the /-stems before suffixes in m among the consonant stems, e. g., Lith. akmenimis, O. B. kamencmc ; in Italic in the so called Latin third declension, to which consideration will now be given. The third declension of Latin seems to be the most thorough- going and complete instance of declensional amalgamation observable in the Indo-European languages, to so great an extent have the original types, the /- and consonant declensions, lost their individuality in the composite system which has resulted. It appears that this instance of metaplasm is the result of certain circumstances of stem formation, inherited from the parent speech, or exactly of a system of double formation, characteristic of the Indo-European period. It is, therefore, pertinent to consider this subject by reason of its probable bearing upon the main question in hand, and as well for any light which may be thrown upon the third declension as a whole, its constituent elements, and the relations of those groups to related words of other languages. 1 Brugmann, Gr. II, p. 732. 2 Streitberg, P. B. B., 14, 210 ff.; K. Z., 31, 51 ff. ''Kluge, Paul's Gr. I, p. 899 ; Sievers, Ags. Gr. § 252. 4 Kluge, Paul's Gr. I, 387. 10 THE LATIN THIRD DECLENSION PARALLEL FORMATION IN INDO-EUROPEAN. It appears that the extension of the primary root to substan- tival or adjectival formations was not in every instance accom- plished by the addition of a single, specific, suffixal element. On the contrary such extension seems to have been directed along various axes of formation by the addition of different stem elements to the same root. Hence there may have existed side by side parallel forms, constructed upon the same primary root with practical identity of meaning but different systems of declen- sion. Of the parallelisms of formation which may be inferred, two, in particular, are clearly defined. These may be character- ized as follows : I. Upon a given root there may have been formed, (i) a root-noun, or consonant stem, (2) an <?-stem, and (3) a form in 1, or, if the same principle be applied to suffixes, consonant, 0- and /-stems may exist side by side upon the same root with substan- tial identity of meaning. II. In like manner a given root may appear (1) as a root- noun or consonant stem, (2) as an 0-stem, or (3) as a formation in -to-. These types, assumed at this point, are substantiated by the following material: 1. Consonant O- and /-Stems. Skt. dvar, dur " door " : Skt. dvara-s, Lat. forum, Grk. 6vpa, Goth, daur : ~Lat. fores (pi.), Lith. durys (pi.) " door." Skt. pdth " path " : Grk. ttcito-s : Lat. pons, O. B. pate " way," O. Pr. pintis, Skt. pathl-s. Grk. xv v '• Skt. hahsa-s "goose" : Lith. zasis "goose." Skt. an-dks- : Skt. catur-aksd-s " four-eyed " : Skt. dksi-s"eye." Grk. (Doric) <5s : Lat. auris, Lith. ausls "ear," O. Pr. ausins, Av. usi (du) "ears." Skt. nds " nose " : Skt. nasa, O. B. nose, " nose," Lat. ndsu-s : Lat. ndres (pi.), Lith. nbsis "nose." Grk. ow£, : Lat. unguis, O. I. niga : Lith. tiagas "nail," Skt. nakhd-s. THE LA TIN THIRD DECLENSION 1 1 Skt. ndbh : O. Pr. nabis, Skt. nabhi-s " navel " : O. H. G. naba, Lith. naba. O. B. grade " city," O. N. gardr : Goth, gard-s " house," Lith. zardis, 1 Lat. urbs. Grk. 6rjp : ha.t. feru-s, /era : O. B. zverc, "wild animal." Goth, yfo/fo " fish," O. I. ease : Lat. pisei-s. Skt. a/ba-,5- "axle," Grk. a/xa^a, O. H. G. a7/jyz : Lat. axis, Lith. a^zw, O. B. osd "axle." Lat. os, Av. asto (gen. sg.), "bone" : Skt. dsthis, O. B. /&?j7<f "bone." Skt. cva "dog," Av. spa, Grk. »aW, Lith. .rz# : Skt. cuni-s, Av. sum's, Lith. szun'is. Lat. <rar<?, Umbr. /£«/-// : Lat. earnis. Grk. Ka/jLTrrj, Lat. eampus, Lith. kdmpas " corner " : Lith. ketur- katnpis "four-cornered." Lat. elavus, O. I. *■/**' : Lat. eldvis, Grk. k\^(/t)is. Skt. pratistha " standing-place " : pratisthi-s. Lat. haedus : Goth, gaits. O. B. rfa/a : Lith. da/is. Skt. abhi-hrut " injury " : abhi-hruti-s. Goth, heiwa-frauja "house-master," O. 'Q.po-sivc, Skt. eeva-s, civa-s : Lat. «z>zV, Osc. ^z>.r. O. N. hlaun (n) "buttock," hlauna-sverd : Skt. crbni-s " hip," Av. sraonis, Lith. szlaunis, Lat. cluni-s. Grk. irripva, O. H. G. fersna "heel " : Skt. parsni-s " heel." Goth, pruts-fill" leprosy " : Lat. pellis. Skt. vrsan "manly " : vrsni-s, Av. varsnis " ram." Av. raohsna and raohisnis "bright." Skt. dena-s and deni-s " stone." Skt. usnd-s and usni-s "warm." Grk. a.Kpo-%, Skt. acra-s : Lat. #<w, deris, Grk. oxpi-s, Osk. akrid, Skt. -acri-. Lat. saeer, -a, um, Umbr. sakra : Lat. sacer, -e, Umbr. sakre. Goth, fagrs "fair" : Umbr. /a<:;w (nom. pi.) "pacati." Av. //jra- and /rjr/j " pointed." Lat. seciiris and O. B. sekyra " ax. " 'A zV>-stem transferred from z-declension. 12 THE LATIN THIRD DECLENSION Lat. buret and bun's. Skt. vamrd-s and vamri " ant." Skt. jlvd-s and jlvi-s " alive." Skt. eubhrd-s and cubhri-s "white." O. B. bylz and <Sjp/<f. Grk. aiXrj and avAis. Grk. KavA.o>, Lith. kaulas "bone " : Lat. eaulis. Skt. nctkt " night," Grk. vv$, Goth, naht-s : Skt. >idkta-m : Skt. ndkti-s, Lat. ;/f.v, Lith. naktis, O. B. «<?i/<r. Skt. dafdt-" tenth " : Grk. Se/orrds, Goth, taihunda, Lith, deszirn- tas : Skt. daeati-s, Lith. defzimtis "ten," O. B. desett. Lat. hortus, Osc. hurz, Grk. x°P ro « : Lat. cohors. Suffix -/a/- beside -tdti-. 1 2. Consonant (9- and 70-Stems. Skt. pad "loot," Grk. ttws, Lat./^, Unibr. /m : Skt. pdda-m "step," Grk. txLIov, Umbr. perum, Lith. peda : Skt. pddya-s, Av. pa'dya-, Grk. 7re£os, Lat. aca-pedius, Lith. lingwapedys "light- footed." Skt. 7vY "village," Av. vis, O. P. c'// : Skt. vefd-s "house," Av. vo3sa, Grk. o"kos, Lat. tv^tfi - : Skt. veciam, Grk. olklov. Skt. z^/6 "voice," Av. ^a/z'j, Lat. vox : Skt. vdkd-s "word," Av. Z'a^ra- : Skt. vakya-m "speech," Lat. convicium. Skt. /vl/' "king," Lat. rex, O. I. rl, Goth. n?z£-.r : Skt. rajyd- " kingly," Lat. regius, Goth, reiki. Skt. napat " grandson," Av. napa~3, Lat. «<^<w : Skt. naptya-s, Grk. a-vei/'tos. Skt. /C'-w-w '-earth," Av. zJj, Grk. x^ v : Lat. humus, Grk. ya.it.ai: O. B. zetnija, Lith. s?w<? "earth." I. E. <&*« "house," Skt. dan, Av. <£?©£", Grk. 8eo--7rdr7/s : Skt. ddma-s, Grk. 8d/u,os, Lat. domus, O. B. </<??;/? : Skt. ddmya-. Skt. rafc "light," Lat. ///.v : Skt. roed-s, Grk. Actios, Lith. lau- kas : O. B. /&;#, //^cr, Lith. laukis. Skt. yuj "yoke," Grk. a-£v£, con-jux : Skt. yuga-m, Grk. £uyo-v, Lat. iugum, Goth, juk : Skt.^^ja-jr. •Brugmann, Gr. II, pp. 290 ff. THE LATIN THIRD DECLENSION 13 Skt. crad-dha "trust," Lat. cor, Grk. <rjp, Lith. szirdu (gen. pi.) "heart," Grk. KapSia. Skt. spdc "spy," Grk. <jku$, Lat. auspcx : Lat. vestispica : Lat. auspiciutn. O. P. /C'artf "army," O. B. kara, Lith. liaras "war" : Goth. harjis "host," Lett, karsch, O. I. cuire. Skt. ndva-s "new," Grk. veds, Lat. novus, O. B. novz : Goth. ««///>"new," Lith. naujus, Skt. navya-s. Skt. dnta-s "end" : Goth, andeis, Skt. antya-s. Lat. orbus, Grk. dp^o-^oTr/s : Goth, a/-^/ "heritage," O. I. 0/-^. Lat. /?7vz, Lith. a^a "berry," O. B.y'tfgv/ : Lith. tfgy.y, Grk. d£os. Lat. porculus : Lat. porcilia, O. H. G.farheli. Skt. svdpna-s "sleep," Lat. somnus, Lith. sapnas, Grk. u7rvos : Skt. svdpnya-m, Lat. somnium, Grk. Iv-iotviov. Grk. otvov, Lat. vinum, Arm. £*'»/ "wine," Lith. apwynys "hops." Goth, kaurn "corn," Lat. grd/ium : Lith. zirnis "pea." Skt. mdrta-s "mortal," Grk. jSpords, ap.-fipoTo<i : Skt. mdrtya-s, Av. masiya-, Grk. ap-fipoatos. Skt. sahdsra-m "thousand," Grk. SeKa-^tAtoi : Skt. sahdsrya-s Grk. x^A' 01 - THE ELEMENTS OF THE THIRD DECLENSION. This fact of parallel formation may first be applied to the third declension in a general manner before its effect in the met- aplasm is treated. Latin and Oscan-Umbrian possess many i- stems which may 1) correspond to <?-stems in related languages or 2) exist beside f-stems in the Italic group, e. g., avis, Osc. cevs, Skt. ceva-s ; piscis, Goth, fisk-s ; colhs, Lith. kdhias ; pellis, Goth, prutsfill; caulis, Grk. KauAos, Lith. kaulas ; humilis, Grk. X#a/u,aAds ; sitnilis, Grk. d/AaAds ; levis, Grk. Aeios ; lint's, O. B. lenz ; tristis, Skt. trstds; agilis, Skt. ajirds : imberbis and imberbus ; iner- mus and inermis; exanimus and exanimis ; biiugus and biiugis; sacer, -a, -urn and sacer, -c. These relations may be explained as the result of double formation. In the latter case both forms have been preserved: 14 THE LATIN THIRD DECLENSION in the former but one, the /-stem type, and, in consequence, a dis- tinction is at once apparent when comparison is made with lan- guages which have preserved only the f-stem forms. A further fact increases this probability. It might be expected that, if the cause be parallel formation, in certain instances, at least, <?-stems ought to exist in Latin beside /-stems in other groups. Such, indeed, is a fact, e. g., Lat. haedus, Goth, gaits ; Lat. ferus, O. B. zvert "wild animal." The contention that parallelism is the cause for this relation of Latin /-stems toe-stems in other groups is supported by further considerations. O- and /-stems often exist side by side in Latin and, as regards relative function, are assignable to certain groups, to be observed in other languages, as follows: — i. Forms in o and /may have the relation of adjective and fem- inine abstract, c. g., Skt. hitd-, Grk. 0£tos, Lat. con-ditus, Lith. pre- ditas, beside Skt. a-hiti-s, Av. deftis, Grk. 0e<ns, O. B. detd. For Latin this relation appears in quietus, quies; datus, dos; grdtus, grates (pi-) J status, statim. 2. The parallelism may exist in substantives, e. g., Grk. ajxa^a, O. H. G. a/isa, Skt. aksd-s beside Lith. aszis, O. B. ost, Lat. axis. This type appears in Latin in bura, buris ; forum, fores ; cldvus, cldvis ; palumbus, palumbis ;' sequester, -tris, sequester, -tri, seques- trum, sequestre. 3. Adjectives in may stand beside adjectives in i, e. g., Grk. 7repKvos, Skt. pfffti-s; und-s, Goth, wans; Grk. ewts : Skt. dena-s and df>/i-s; cubhrds and cubri-s. Av. raoh'sna and raohsnis. Latin possesses the same relation, e. g., sacer, -ris (Plaut.), Osk. sakrid U. sakre ; sacer, -ft, O. sakru, U. sakra : graci/is, graci/us 2 : steri- lis, sterilus^ : dapsilis, dapsilus* : decor em, decorus : indecoris, inde- corus : hilaris, hilarus : celero, ceteris. 4. From Indo-European times compound adjectives formed upon 0-stems have often assumed, 1) the /-form or 2) the /-form 'Cato R. R. 90: Mart. 13, 67, 1. 2 GraciLr, Ter. Eun. 314 ; gracila, Lucil. Non. 8, 48. iSterilam, Paul. Fest. 314 : sterila, Lucr. 2, 845. '■Dapsilis, Plaut. Pseud. 396, Aul. 167. f UNI THE LATIN THIRD DECLENSION 15 beside a form in o, 1 e. g., Skt. abhaya-hasti-s to hdsta-s : dhumd- gandhi-s to gandhd-s : prdty-ardhi-s to ardhd-s : saho-bhdri-s, jar- bhdrl-s to bhdra-s : su-niti-s, dgra-niti-s to nita-s : trir-dfri-s, cdtt/r- ap-i-s to dfra-s : A v. mazda-yasnls to yasna : zarapuslrls to Zara- puslra- : Grk. a«oms to koitt;, Lith. ketur-kampis to kampas ; nakt'e- kovis to kbvas ; pig-kalbis to £«/^a : 0. h. pre -pros it adv. to prosle ; ils-plznt to plznz? This relation is seen in Italic in U. perakre to akru-tu ; sevakne to acnu ; Lat. biiugis, iniugis to iugum ; semisomnis, somnus; inermis, inermus, artna; imberbus, imberbis, barba; exanimis, exanimus, animus; perennis, perennus, annus; effrinis, effrenus; prcecoquis, pracoquus. It is next in order to consider the w-stems and the category of parallel formations to which they belong, and to discern a possible bearing upon the question of metaplasm and the for- mation of the third declension. Streitberg has established, for the suffix io, two ablaut grades, -to-, strong, and -I , weak. This weak grade has been preserved in Italic in Lat. alts, alim; Cornells, Cornelim ; fill (voc.sg.); Ingeni (gen.sg.); Umbr. tertlm "tertium "; tehtedim, " tectorium " ; sansl, sad (ace.) to voc. sancle : Osc. mediclm, memslm. These forms prove that this weak grade of -10- was inherited and existed until after the separation of the dialects. It is, further, probable that this type was, at an earlier period, a more extensive category, than appears from the material of his- torical times, for the greater relative frequency of the forms in Oscan-Umbrian indicates that they were more numerous in Italic and probably in early Latin. Again the forms seem to have been a dying force in Latin of the historical period and this may indicate that the few forms are remnants of a more extensive cat- egory previously existing. If this be true, it is in point to inquire 1) what tendency has occasioned their elimination and 2) to what system of declension have they been transferred. The conjecture that they have been merged in the /-declension is supported by the following considerations: 1 Cf. Mahlow, AEO., p. 121. 2 Leskien, Handbuch, p. 94. 1 6 THE LATIN THIRD DECLENSION i. The Argument from Inherent Probability. It is evident that the two systems of declension come in con- tact at the following points: — -/- -io- Nom. sg. -is : -is, e. g., Lith. Izbdis; Goth, bruks, un-nuts. Ace. sg. -i-m : -i-m, e. g., Lith. zbdi. Gen. pi. -iom : -torn, e. g., Lith. zbdzu ; O. B. krajt ; Goth. harje. Nom. ace. pi. n. -i-a : -i-a, e. g., Goth, kunja : O. B. polje. This condition, in itself, constitutes a strong antecedent prob- ability that the two declensions were in close contact and would become merged in one system or the other. 2. Analogies of Other Languages. a) Germanic — The so called /-declension of adjectives in Gothic is a conglomerate of *- and /0-stem forms, resulting from a merging of such types, as bruks, un-nuts with /-stems such as ga-mains, Lat. communis; /trains, Skt. (rents. 1 Among substan- tives may be noted andins ace. pi. to andeis. b) Balto-Slavic— Many /-stems have passed into the io- declension. This is a process directly opposite to that assumed for Latin and yet the analogy has weight, showing as it does the contact of the /- and -/^-sterns. Instances are krytis, gen. -Us and czio ; szlitis, gen. -Us and fem. szliti : antis, gen. -Us and czio, Skt. dtis, Lat. anas 1 : O. B. gospodt, gen. gospodja, dat. gospodjo : ognt Lat. ignis, inflected as a -/^-stem, e. g., ognja, ognjo." 3. Evidence within the Italic Group. a) The correspondence of Lat. and Osc.-Umbrian /-stems to w-stems in other groups, e. g., pix, gen. pi. picium, Grk. maaa, Lith. pikis : postis, Skt. pastya-m, O. H. G. fasti : Osc. aiteis, gen. pi. aittium, Grk. aLWa : faux, gen. pi. faucium, Grk. </>u<ra but Skt. bhiika-s : verres, Lith. verszis, Lett, wersis : vehes, Lith. vcTis, O. Pr. wessis : lux, O. B. hue, luca, Lith. laukis : calx, gen. pi. calcium, Lith. kalkis. 'Brugmann, Gr. II, p. 269. 2 Leskien, Handbuch, § 38. THE LATIN THIRD DECLENSION 17 b) The existence of io- and /-forms in the same word, e. g., prasepe and ace. pi. prcesepis to prcesepio, prcesepiis ; dldris, dldr- ius ; auxilidris, auxilidrius ; articuldris, articuldrius ; balnedris, balnedrius ; caligdris, caligdrius ; cubiculdris, cubiculdrius ; inter- caldris, intercaldrius ; militdris, militarius ; palindris, palindrius ; vulgaris, vulgdrius ; alvedre, alvedrium. c) Oscan in single forms gives evidence of such a transfer to the /'-declension. Upfils, 1 Latin Ofilius, is in the weak grade -*"-. The genitive is Upfaleis, 2 in the form of the /-declension. But the oblique cases of the -/-stems have full -io-torms, e. g., Goth, harjis (gen.), Lith. zbdzjo. Hence we should expect a form parallel to meddikiai. Again, aiteis to Grk. alo-cra < ait-ia and luisarifs, beside Latin lusoriis, point to the same conclusion, that there has been a transfer to the /'-declension. It thus appears that the correspondence of /'-stems in Latin to 0-stems in other groups may exist for two reasons, 1) because of the parallelism of 0- and /'-stems in the parent speech, and 2) by reason of the parallelism of 0- and w-stems and the trans- fer of the weak form in -j- to the /-declension, as in Lat., e. g., faux, Grk. <£v'cra, Skt. bhuka-s or O. aiteis, aittIum, Grk. ala-aa, I£-aiTos. The third declension presents two further anomalous stem relations, /-stems in Latin may correspond to //-stems in other languages in the well-known relations sudvis, Grk. ^Su's, Skt. svddus; tenuis, Skt. tanus ; brevis, Grk. fipaxys, etc., and -<?.y-stems may appear in Latin as /'-forms, e. g., sedes, Skt. sddas, Grk. ISos, nubes, Skt. nabhas, Grk. ve^>os. As regards the former condition, it is evident that an expla- nation may be sought in the parallelism of 0-, /'- and //-stems, e. g., Skt. babhrus "brown," Lat. fiber, Av. bawris "beaver"; Skt. tdntu-s and tdnti-s "thread"; Lith. asztrus, O. B. ostrz "sharp," Skt. dstrd "good," Lith. asztrds ; Skt. dcru-s, dcra-s, dcri-s; Skt. dhdru-s "sucking," Grk. Br\\v%, Grk. 6r/\r) ; Skt. patdru-s, patdra-s "flying"; gdtu-s, gdti-s "motion." This condition may account for such a relation, as mollis, Skt. mrdhu-s, but not for the cases •Zvetaieff 121, 122. 2 Zvetaieff 113. 1 8 THE LATIN THIRD DECLENSION in which the element u is present. It has been suggested that these forms in -vi- are the feminine formation in -id. ' There is, however, a possibility that another element is present. Beside the stems in -u- there may have existed forms in -uo- fern, -ud or -ui- fem. -ul-, e. g., Skt. ddru, dru "wood," Av. dd"ru ; Grk. Spv pa, Lith. derva ; Skt. ddrvis, ddrv'i "spoon" : Skt. tanus, Grk. Taia's; Grk. rava/rds, Skt. tanva-s "slender," Lith. tenvds ; Skt. ghrsu-s and ghrsvi-s " lively " ; Skt. yahii-s, yahvd-s, yahvi "young" : Skt. raghu-s "swift," Grk. eAaxv's ; Lith. lengwas : Skt. paracu-s "axe," Grk. 71-cAckvs ; Grk. iriXtKnov : Skt. puru-s, Grk. ttoAu? ; ttoXAos : Skt. fifu-s " young " and su-cicvi-s " well growing." The three types appear in Latin, as follows — i) -«-stems, densus, Grk. 8o.<tv$ ; cdrus, Skt. cdrus : 2) -uo-, arduus, Skt. urdhvd-s, Grk. 6p#ds ; 3) -///-stems, sudvis, tenuis, etc. In the case of the last type, it is only intended to point out that the forms in -ui- would naturally fall in that category and to suggest that they may have helped to form the group together with the feminines in -ul-. Occasionally /-stem forms seem to exist beside ^-sterns, e. g., sides, Skt. sddas, Grk. cSos : nubes, Skt. ndbhas, Grk. v£<£os : moles beside molestus : pubis, tdbis, plibis, sordis, squdlis : pulvis beside pulver, pulveris; cinis, ciner ; vomis, vomer : cucumis, cucumer. Brugmann has explained one, at least, of the /-stems, paralleled by -^-forms, as the result of double formation. 3 It seems, also, that the parallelism noted in Skt. sddas and sddi-s is widely extended, e. g., Skt. pathas, pathi-s ; jdnas, jdni-s, jam ; vdnas, upamati-vdni-s ; rdbhas, su-rabhi-s ; mdhas, mdhi-s; dhruvds, dhnivi-s ; vdcas, vaci-s. Grk. /AtVos, /xtJvi-s ; Av. rava~3, ravis, Grk. p^p'-Sios. Similar relations appear in Latin, e.g., nubis, nubs; sordis, sors ; tdbis, tdbis ; pubis, pubis ; moles, mo/is ; sidis, sidis ; plibis, plibs. Hence the entire category may perhaps be explained by this parallelism. 'Brugmann, I. F., IV, 218; J. Schmidt, K. Z., 25, 139 and K. and S. B., IV, 266; Johanssen, K. Z., 30, 403; Danielsson, Gram., Anm. I, 25. 2 Brugmann, K. Z., 24, 44; J. Schmidt, Pluralbild., p. 148, otherwise J. Schmidt, K. Z., 27, 328; cf. also Thumeysen, K. Z., 30, 489. THE LATIN THIRD DECLENSION 19 APPLICATION TO THE METAPLASM IN THE THIRD DECLENSION. Parallelism in formation has served to explain the apparent anomalous relation of Lat. /-stems to 0-stems in other groups. As has been intimated, however, the same facts, in a different application, bear upon the question of metaplasm in the third declension. It has been shown that consonant and /-stems may exist side by side upon the same root with substantial identity of meaning. Brugmann 1 has suggested that the metaplasm has been occasioned by this condition and by the existence of such doublets as ttoct-, twcti- : clvitdt-, civitati-. To this theory, how- ever, the following supplementary matter may, perhaps, be added. If the conclusions of the previous pages be sound, the condi- tions tending to promote the metaplasm may be increased by the fact that /- and consonant stems may exist side by side in Latin by reason of the fact that certain /0-stems have become merged in the /-declension. If, then, we assume such doublets from two distinct sources, the rise and development of meta- plasm may be traced as follows : It is probable that two forms of the same word belonging respectively to the /- and consonant declensions and existing side by side with identitv of meaning and function, would be often con- fused, that the case forms peculiar to each would be used indis- criminately and without regard for their point of reference. Thereby a point of contact between the two declensions would necessarily arise. The probabilitv of such a confusion and point of contact is supported by the analogies of other languages, e. g.; 1) Arvan- — O. P. vipibis* to vip, Skt. vif, Av. vis. 2) Germanic — Gothic consonant stems show the influence of the /-declension, 3 e. g., baurgim to baurgs ; alhim to alhs ; spaurdim to spaurds ; waihtins to waihts ; dulpais to du/ps. 3) Balto-Slavic — a) The transfer of consonant stems to the 'Grundriss II, pp. 292, 727. 'So Spiegel, A. P., Keilinschr., 2 p. 177 : Bartholomx reads vipaiifsand holds for influence of <?-stems ; cf. Brugmann, Gr. II., p. 723, and Kossowitz, Inscr. Pal. Pers., under vit'ihis. 3 Braune, Got. Gram. 3 § 116. 20 THE LATIN THIRD DECLENSION /-declension, e.g., Lith. dant'is, Skt. ddnt- ; szirdis, Lat. cor, Skt. hrd; debesis, Grk. ve'^os, Skt. ndbhas, O. B. zviV, Skt. vie : /$) metaplasm in the Balto-Slavic period before suffixes in m and in the locative plural, e. g., Lith. akmenyse, older -ise, O. B. kamentchz ; Lith. moterimis, O. B. materismd, matertchz. Assuming, therefore, as the first condition such a confusion in the case of specific words, it is probable that the approximate contact between the two declensions would become crystallized in some particular case form and that, at that point, one of the two possible terminations would eventually prevail to the exclusion of the other, first in specific words belonging originally to either declension and, second, by an analogical extension in other or in all the forms in each declension. Such a condition may be termed the formal contact in distinction to that approximate type which was noted in the element of confusion. This formal con- tact would occur at the point where the two declensions have the strongest affinity. That point seems to have been the dative- ablative plural. For the earliest Italic we may suppose such rela- tions as dotibus beside *nepot-bus ; mentibus, *ferent-bus; sacri-bus, *matr-bus ; funi-bus,*homin-bus ; faaci-bus, *due-bus, and possibly within the same word norti-bus, *noct-bus ; clvitati-bus, *clvitdt- bus. The affinity of the two declensions at this point is apparent. But, further, in other languages this metaplasm of the i- and consonant stems appears most frequently or exclusively at this point, e. g., in Germanic and Balto-Slavic before suffixes in -m- (cf. p. 9). The only point of infection must necessarily be the first point also, and this condition adds further probability for that assumed in Italic. Hence the formal contact in the dat.-abl. plural, occurring first, perhaps, in such words as nox and civitds, must have been, later, extended to all the words of each declen- sion by analogy. Then the similarity at this point may well have produced similarity at other points and thus have extended the metaplasm. THE DEVELOPMENT OF THE PROCESS. The development of the metaplasm and its extension to the various cases may be traced as follows : The first stages of the THE LATIN THIRD DECLENSION 21 process must be assigned to the Italic period, by reason of the agreement of Latin and Oscan-Umbrian at certain points, as follows : 1) The dative-ablative plural. Latin and Oscan have the /-stem form i-bos 1 without exception, 2 e. g., O. teremniss, aisu- sis, ligis. Umbrian has -us without exception, e. g., fratrus, ncrus, kapidus, homonus, karnus. If this Umbrian type be accepted as secondary and a later formation after the tf-declen- sion, 3 the forms do not disprove the hypothesis given above. 2) The ablative singular. Among the /'-stems the oldest forms are in -Id in Latin and Oscan-Umbrian, e. g., O. Lat. marld, omnei : Lat. turrl, classi, etc., : O. slaagid, sakrid, akrid : U. poni, peracrei, ocri-per. This form was transferred to the consonant declension in the Italic period, e. g., O. Lat. airid, covcntiomd, bovid : Fal. opid; O. prcesentid, prupukid, serevkid ; U. pedi, persi. The fact that -I is largely replaced by -e in Latin of the classical period, that Oscan has forms in -3d by the later influ- ence of the 0-stems, and that at a later time -e is the prevailing type in Umbrian, cannot obscure the evident condition, noted above, that the two declensions become merged in the form -id at the point of the ablative singular. 3) Dative singular. It is possible that a metaplastic devel- opment occurred at this point. A positive statement cannot be made, since in the development of Latin and Umbrian the indi- vidual case endings have become identical in form. Hence no differentiation can be made. The probability is, therefore, mentioned at this point and a more extended discussion reserved for the treatment of the specific case forms (p. 43). 4) Genitive singular. At this point a somewhat different condition must be noted. Latin and Oscan-Umbrian do not agree in form. The former has the consonant type -es > -is and ' Or for Oscan -ifis, Buck, Voc, p. 49. 2 Latin has senatorbus, C. I. L. I, 196, which may well be doubted on the ground of senatoribus (2) and mulieribus on the same inscription. Cf. Stoltz Lat. Gram., 2 p. 344. Allen, Early Latin, p. 29. 3 Brugmann, Gr. II, p. 711. 2 2 THE LATIN THIRD DECLENSION •os > -us, while the latter preserves the /-stem form eis, O. -eis, U. -es. An element of similarity is, however, noticeable. The metaplasm is complete in each group. Latin, in no instance, preserves the /-stem form : Oscan-Umbrian has that form con- stantly and without exception. The conditions are different from those in the nominative plural, in which the forms are well dis- tinguished in Oscan-Umbrian. It is probable, then, that in the Italic period there was not a complete metaplasm as in the abl. sg., but that the forms -eis and -es were used indiscriminately. Latin has preserved -es and Oscan-Umbrian -eis. It is evident that the metaplasm advanced no further in the Italic period. Its force, however, was retained long after the period of separation. In Latin it lived to cause identity of form to a greater or less extent in the ace. and abl. sg. and in the plural cases other than the dat.-abl., and remained as an active factor in the historical period, making constantly for a complete amalgamation of the two systems. On the other hand no such activity is discernible in Oscan-Umbrian. The sole result was the survival, in the gen. sg., of the /-stem form -eis, a partial step only, since confusion must have been established in the Italic. To be noted, however, is the fact that in Latin the consonant forms have been preserved in the singular in marked contrast with the tendency of the earlier period. The Italic languages are characterized by a marked frequency of /-stem forms among the adjectives. Original /-stem types are more consistently preserved in words belonging to that declen- sion, and are more generally transferred to adjectives of differ- ent origin, e. g., for the /-stems abl. sg. in -I in distinction to the -e of substantives, acri, celebrl, equestn, tristl, etc.; ace. pi. in -Is, omuls, mortalis; gen. pi. in -turn and neuter pi. in -ia. The same characteristics are observable among adjectives, originally conso- nantal in inflexion, e. g., ingenti, inerti, perpetl, evident! 1 : ace. pi. amantis, prudentls, hebetls : neuter pi. always -ia except in Vetera; gen. pi. 'turn, e. g., amantium, inertium. It is evident, therefore, that the /-stem forms possess excep- tional vitality among the adjectives, both in persistence in their ' Cf. Neue, Formen.3 II, pp. 51 ff.: divite is the prevailing type for dives. THE LATIN THIRD DECLENSION 23 original position and in extensive transfer to the consonant declension. Likewise /-stem forms often exist among the adjec- tives beside corresponding consonant terminations, when the same words are used as participles or substantives. The basis for this distinction seems to lie in the difference in function. The facts noted above may not reasonably be separated from the further fact that in the entire Italic group there is a tendency toward /-formations among the adjectives [cf. p. 13 above), a tend- ency supported by the analogies of other languages. It may, then, be accepted that there come into Italic from the parent speech a considerable number of compound adjectives in /, exist- ing beside a substantival simplex in 0. Further, in the case of adjectives in 0, beside forms in / in Indo-European, the /-form has become the prevailing type in Latin (p. 14). In the instance of parallelism between 0- and w-stems, the latter type was, in general, adjectival in function, as appears in the relation, Skt./<f/, Gr. ttws, Lat. pes : Skt. pdda-m, Gr. iriSov, U. perum, beside Skt. pddya-, Gr. Tre'^os, Lat. acupedius, Lith. lingva- pedys "swift-footed." It has been assumed that the weak grade in -j- of this suffix has been merged in the /-declension. If this be true, the relation adjective in / to substantive in becomes a more numerous, and, in fact, a very considerable category in the Italic languages. The group must have constituted a very extensive body of /-stems existing among the adjectives, and possibly the largest proportion of words possessing that function belonged in this category. It may, then, be assumed that the /-stem formation came to be viewed as the distinctive type of the adjectival function, and that the category, for that reason, possessed sufficient linguistic weight to make the original /-stems less susceptible to meta- plastic influence, and to cause an extension of its forms and an infection of consonant stems more extensive than that noted for substantives. Such a heavy and compact mass of forms may well have influenced the whole category in this way, and, like- wise, may have occasioned the transfer of some adjectives from the 0- or consonant declensions by reason of the tendency of identity of function to eliminate diversity of form. SYNCRETISM The term Syncretism is applicable to instances in which a given case form in any declension possesses a series of functions which, in the final analysis, must be referred, not to a single syntactical force, but to two or more such primary forces. A single case form thus assumes the functions previously pos- sessed by two or more cases, and the grammatical apparatus is simplified and its diversity of form reduced by the total or approximately total exclusion of the forms thus yielding to syncretism. This phenomenon appears in the Indo-European languages in two types : i) A single case may have performed double syntactical duty from the earliest times, or 2) the con- dition may be the result of a genuine contamination and merg- ing of functions. The former must be viewed as an inheritance from the parent speech ; the latter as a result of some linguistic force tending to reduce complexity in the declensional systems. There exist the following instances of syncretism as an inherited and original condition : 1) Dative-ablative plural ; 2) genitive-ablative singular, 1 except 'The ablative had a separate form only in the pronouns and in the sg. of the 0-stems. Otherwise it was merged with the gen. in the singular and the dat. in the plural. The former condition seems natural {cf. p. 29 and Delbriick, Grundriss III, p. 191), but why is it identified with the dative in the plural? Lanman (Noun Inflection, p. 583) holds for linguistic economy, the dative being taken as the nearest form in infrequency of usage. Gaedecke (Ace. im Veda, p. 144, Anm.) assigns actual syntactical contact as the cause, and Delbriick (p. 190) adds further cases of the same nature. To these may be added the Lith. usage with verbs of stealing in a sense allied to that of separation, e. g., pavoge tdm seniui tq. azkq. " he stole the goat from the old man " ; also various constructions in Homer of the nature of a dative of advantage, e. g., Aavotvi \01ybv &/xvvov, A 456, "ward off destruction from the Danai," d/iuv^uei'cu &pe<r<nv, E 486, " to protect the wives," beside the same force with genitive, e. g., oXXd Zeus Krjpas dfivvev waidds iov, N 1 09. 24 THE LATIN THIRD DECLENSION 25 in pronouns and 0-stems save in Avestan and Italic ; 3) nomina- tive and accusative neuter, both singular and plural j 1 4) nomina- tive and vocative of dual and plural ; 5) in the oldest type of the dual, the various functions are borne by three forms, the nom.-voc.-acc, the inst.-dat.-abl., and the gen.-loc. A condition analogous to the above may be noted in cases of accidental identity of form by reason of the union of the case ending and stem vowel, as in the dative and locative of the d- declension, e. g., a -\- -ai = -di : a -f- i = di." The other type of syncretism, that process of simplification of case systems which has occurred within the separate history of the individual languages, has been considered by Delbriick as the result of three possible influences : r) contact occasioned by confusion of prepositions, 2) contact in actual syntactical usage, and 3) identity of form. As regards the first element, it is evident that prepositions may be used with different cases and with substantially identical meaning. This is evident, even in Sanskrit, in which the dis- tinctions are generally well preserved. 3 For other languages more striking instances may be added, e. g., Gothic — in filuwaurdein seinai, Matt. 6 : 7, "because of their much talking," beside in pizozei waihtais, Eph. 3:1, "for this cause " : ana staina, Matt. 7 : 24, " upon a rock," but ana baurgs, Tit. 1 : 5, "in the cities" : afar twans dagans, Matt. 26 : 2, "after two days," but afar pamtna hlaiba, John 13:27, "after the sop " : iah atstandands ufar ija, Luke 4 : 39, " and standing above her," but warp riqis ufar allai airpai, Matt. 27 : 45, " there was darkness over all the earth" : Lithuanian — cf. uz with both ace. and gen. in sense " behind." 4 Old Bulgarian — po paiz " upon the way," po morjo "on the sea," beside /0 vise grady "in all cities" : vz tz dene "on that day," vz sija nostz "on this night," beside be ob nostz vz molitve boziji " and he tarried all night in prayer to God " : 'For the explanation of this condition cf. Delbriick, Grun. Ill, p. 189. 2 Brugmann, Gr. II, p. 738. 3 Whitney, Skt. Gram. 2 293a, 1127. 4 Schleicher, Handbuch, p. 290. 26 THE LATIN THIRD DECLENSION za tvoja. beseda. "because of your speech," beside za stracha jodijiska " from fear of the Jews." It is a question, however, whether the preposition was so vital a factor in case relations at the time at which the syncretisms arose, as at a later time. If the prepositions be viewed as crys- tallized case forms of demonstrative stems, it seems probable that at first they were accessory only and that the case endings alone served to express the various case relations. Later the preposition must have become more intimately associated with the idea thus conveyed, until it was felt to be essential and equal to the case ending in grammatical value. Still further the preposition has in many cases assumed predominating force over the ending. Hence, granting that the preposition possessed force equal to that of the case, a diversity of function may well have assisted the syncretism. If, on the contrary, the prepositions were merely accessory forms, the difference of function in such words may be a result rather than a cause of the syncretism. Such an instance may possibly be found in the use of Latin ab with both abl. and inst. function. The original force seems to have been that of the abl., e. g., Skt. dpa, Grk. a-rro. Hence the double function, in senses "by" and "from," may have been suggested by the syncretism of the abl. and inst. The same word in Gothic, ab, is used with instrumental value, e. g., ab saurgom afhapnand, Luke 8:14, "choked with cares." On the other hand, prepositions may acquire a variety of functions, other than by a syncretism of cases, e. g., Av. a i wi, used with both loc. and dat. in the sense "upon," but without syncretism of the cases; antarj "between, among" used with inst. and loc; faro " before, beside" with abl., gen. and loc. : pasca "after, behind" with ace, inst., abl., gen.' Again prepositions may develop a variety of meanings with- out syncretism of the cases, e. g., Av. pcfti "with" or "at" with ace, inst., abl. and loc, beside Skt. prdti "into" with ace, Av. avi "to" with ace, "in" with loc. and "from" with abl. without trace of syncretism in case endings. As regards the third factor, identity of form, it is evident that 1 Jackson, Avesta Grammar, Pt. I, p. 204. THE LATIN THIRD DECLENSION 27 such a condition, arising by reason of phonetic change, would readily occasion syncretism. In Goth, the dat. case of the 0- stems having dat., inst., abl. and loc. function may be referred to any one of the following forms, *dagat loc; *dagdi dat.; *dago inst., or *dqgod abl.' It is also possible that partial identity of form may promote syncretism. The conditions would be two given cases, possessing a partial identity or similarity of form. Analogy might, then, cause elimination of the difference of function bv that force by which resemblance of form suggests likeness in function. 1 ' This element may possibly be noted in the case of the abl. -inst. in Italic, e. g., b, bd ; a, ad ; u, ud ; /, Id. A further factor is that of syntactical contact. It seems that each case must have possessed a primary force which may be viewed as its distinguishing characteristic. These primary forces must, however, be regarded as the centers of syntactical areas, from which radiate numerous secondary functions. Hence, though the primary forces be wholly different in character, these secondary syntactical extensions may not only approach but also intersect or coincide. Such a contact becomes evident from the comparison of the cases, notably the ablative and genitive and the dative and locative. 1. Ablative and Instrumental. The original forces of these cases must have been respectively separatio)i or source and accompaniment or association. In the secondary forces, however, the two cases come in contact as follows : a) Instrumental — A derived force appears in the usage, which expresses instrument ox means, 1 e.g., Skt. bhadr&m kdrnebhih (rnuyama " may we hear with our ears what is auspicious " ; (astrena nidhanam* "death bv the sword" ; Greek x €L P^ T€ f MLV Karep€$ev, A. 361, "caressed him with her hand " ; Ai#ois ZfiaWov, X. A. 5, 4, 23, "they pelted them with stones": Old Bulgarian jisplztiise se strachonu, Luke 5 : 26, "they were filled with fear" ; 1 Cf. Kluge, in Paul's Gr., p. 386; Brugmann, Gr. II, pp. 599, 617, 628. 2 Wheeler, Analogy, p. 29. ''Whitney, Sanskrit Gram. 2 § 280. 28 THE LATIN THIRD DECLENSION toja. bo meroja, Luke 6 : 38, "by that measure" ; trzstt It vetromc dvizemy, Luke 27 : 24, "a reed shaken by the wind" : Lithuanian lazdd miiszti "to strike with the staff"; keliu vaziuti "to go by the road." This force readily passes into an expression of occasion or reason, e. g., Skt. krpaya 1 "through pity"; tena satyena "by that truth" : Grk. ayvoia iia/jLaprdvovai, X. C. 3, 1, 38, "they err by reason of ignorance" : Old Bulgarian azz ze scde gladomz gybla " I perish from hunger"; Lithuanian badii gaiszti "to perish from hunger" ; drugiii sirgti "to be sick of a fever." b) Ablative — The ablative may come to mean procedure, as from a cause or occasion, e. g., Skt. v&jrasya cusmdd daddra 3 " from the fury of the thunderbolt he burst asunder" : Grk. rio 8' avr €7ri/i€/A<£eai, B. 225, "on what account do you again find fault" : Old Bulgarian strazdasteji otz duchz nicistz, Luke 6:18, "afflicted by unclean spirits" : Lithuanian jis nusigando to zbdzo "he was terrified at the word "; asz bbdzus to valgio, " I loathe food." There is evidently a contact in the force, occasion, or cause. The instrumental may also be used with verbs denoting separation, 3 and the two cases are used with the verb "drink" to denote the cup or vessel. 4 2. Dative and Locative. The primary force of the locative was situation or location. Extended usages, however, are found in 1) the forces " in case of," "respecting," "with reference to," e. g., Skt. tdm it sakhitvd Ima/ie 5 "him we beg for friendship" : Old Bulgarian lie jaste ne szb/az- nitz se vine "who shall not be offended at me"; ubudite sja pravde, 1 Cor. 15 : 34, "awake to (with respect to) righteousness" : 2) with verbs of "arriving," "placing," "showing," "bestowing," or as a goal or object of motion, action, or feeling, e. g., sd id deve u gacchati "that, truly, goes to the gods"; sam crutya piirvam asmdsu "having before promised us" : O. B. nalezestjo 1 Whitney, Sanskrit Gram. 2 § 280. 2 Schleicher, Lith. Gr. p. 267. 3 Whitney, Skt. Gram. 2 § 283, and Delbruck, Gr. Ill, § 1 10. ■•Uelbriick, Gr. Ill, §§87 and 113. 5 Whitney, Skt. Gram. 2 § 304 b. THE LATIN THIRD DECLENSION 29 t'emd narodu, Luke 5:1, "as the people pressed upon him"; kosmd sja podolzce rizy jego, Matt. 9 : 20, "she touched the hem of his garment" : 3) with nouns and adjectives having similar forces. 1 With these usages may be compared those of the dative in 1) the force "with reference to," e. g., Skt. isumk rtivdna dsandya "making an arrow for hurling" : O. B. vznemljate sebe, Luke 21: 34, "take heed with respect to yourselves" : 2) with verbs of "giving," "assigning," "moving," etc., e. g., Skt. yd ?/d ddddte sdkhye "who gives not to a friend"; O. B. pomanase pricestt- nikonu, Luke 5 : 7, "they beckoned unto their partners" : 3) verbs of mental status signifying "give attention," "have regard for," "please," etc., e. g., Skt. yddyad vocate vipribhyah "whatever is pleasing to Brahmans," kirn asmdbyham hrrnse "why art thou angry with us"? O. B. revtnule doromc bo/isimt, 1 Cor. 12:31, "covet the best gifts." There are assumed for Indo-European eight cases : nomina- tive, vocative, accusative, instrumental, dative, ablative, genitive, and locative. The instances of syncretism tending to eliminate this diversity of form are as follows. In Greek such a process is evident at two points, 1) the dative, locative, and instrumental 2 in the so-called dative form, and 2) the genitive and ablative 2 in the genitive form. In Latin the instrumental, ablative, and locative have become merged in a single form, in most cases that of the ablative {cf. p. 30). The four oblique cases, instrumental, ablative, locative, and dative, have become syncretized in Ger- manic, the nominative and accusative functions are, in certain cases, borne by a single form, 3 and the nominative is often used as a vocative. Irish is on a line with Germanic, since the so-called dative form has the functions of the instrumental, dative, loca- tive, and ablative. In Balto-Slavic both Lithuanian and Old Bulgarian have syncretism of the genitive and ablative. The form for the singular of the 0-stems is that of the ablative, else- where the genitive. 1 Whitney, Skt. Gram. 2 § 304 b. 2 Brugmann, Gr. Gr. 2 pp. 207, 205. 3 Brugmann, Gr. II, pp. 526, 547. 3° THE LATIN THIRD DECLENSION SYNCRETISiM IN ITALIC. Latin shows the following instances of syncretism : i) the ablative, locative, and instrumental in the so-called ablative form and in both singular and plural of all declensions ; 2) the dative and locative of the a-stems ; 3) the dative and loca- tive of the /-stems; 1 4) the genitive and locative of the <?-stems; 2 5) the nominative and accusative plural of /- and consonant- stems and possibly, also, the same cases in the ^-declension. 3 Oscan and Umbrian have the following certain cases of syn- cretism : 1) the ablative and instrumental, e.g., O. ablative, Amvi- anud, BtrviANtJD "a Boviano," eisucen ziculud "ab eo die''; inst. ammid egmas tovticas " rei publicae causa," tristaamentud " testimonio," altrud ligud " cum altero lege," cum preivatud " cum privato," dolud malud " dolo malo," mvjltasikad " multa- ticia," serevkid " auspicio," rehtud amnItd "recto circuito," muinikad tanginud "communi sententia"; U. abl. akru-tu "ex agro," ehe esu poplu "ex hoc populo," anglu-tu " ab angulo" : inst. adputrati "arbitratu," nomne "nomine," paca "causa," esu bue peracre "hoc bove opimo," persklu " supplicatione," pure " igne," pone " posca," vea " via." 2) Dative and locative of the «-stems, 4 e. g., O. dat. anters- tatai " interstitae," deivai gexetai " deae genetrici," entrai " interae," PernaI " Pernae," Anagtiai " Augitiae " : loc. Bansae "Bantiai," eisai via! mefiai "in ea via media," aasai purasiai "in ara igniaria," medikkia! " magistratu." U. dat. ase " arae," cerfie " cerfiae," mefe " mensae," tote " civitate " : loc. scalsie "in patera," totem-e, tote " in civitate," sate " in sancta," Akedunie " in Aquilonia." 3) Dative and locative of /-stems (//. p. 44). In the plural, as in Latin, the dat., abl., or inst. form pos- 1 Assumed on the basis of the forms which are locative. Cf. Brugmann, Gr. II, pp. 604, 605. 2 Brugmann, Gr. II, p. 586. 3 Brugmann, Gr. II, p. 666. * Here as in Latin the syncretism may be viewed as an accidental identity of form (p. 25). THE LA 'J IN THIRD DECLENSION 31 sesses four functions, dat., abl., loc, and inst., e. g., O. dat. dek- manniuIs " decumanis," huntruis " inferis," supruis' " superis," lIgatuis " legatis," anafriss "imbribus," MAATfjis kerrjuis " manibus cerealibus" : inst. aisusis "sacrificus" : loc. op eizois " apud eos," anter teremnIss " inter terminos," exaiscen ligis, " hisce in legibus." U. dat. fratrus " fratribus," homonus " hominibus," aketus " agentibus," nerus sihitir " principibus cunctis" : inst. veskles vufites " vasculis consecratis," veskles snates "vasculis umectis," adepes arves "adipibus arviis," kapidus " capidibus," dupursus " bipedibus," asesetes karnus " non sectis carnibus " : loc. pre-veres " ante portam," tuvere kapidus " in duobus capidibus," vasus " vasis," fesner-e " in fanis," fondlire " in fontulis " ; abl. vapersus "ab sellis." In the early history of the Italic group there appear to have arisen two syncretisms: 1) that of the abl. and inst., and 2) that of the dat. and loc. The former is evident in all declen- sions and both numbers. The causes may have been syntactical contact (p. 27) and in certain cases approximate identity of form. The latter may be noted in the entire plural and in the singular of the a- and /-stems. Herein also functional contact may have been the cause, and in the case of the a-stems, identity of form. In the plural, therefore, four functions were brought within the scope of a single form. For the consonant, /-, and «-stems this form was the dat. -abl., for the 0- and «-stems the instrumental. For the consonant declension slightly different conditions must be noted. The abl. force in this declension was borne by the gen. originally. There are, however, no traces of this con- dition in Italic. But there can be no objection to the supposi- tion that the metaplasm and transfer of -Id to that declension was as early as the syncretism of the inst. -abl. Hence as 3d, ad, iid assumed this double function, it is probable that id did the same for both /"- and consonant stems, and further, as the inst. forms 0, a, u were displaced by the ablatives, it is probable that -id also prevailed to the exclusion of the inst. of the con- 1 Bugge, Alt. -It. Stud., p. 40. 32 THE LATIN THIRD DECLENSION sonant stem, whatever that may have been. Accordingly the conditions in the various declensions would be as follows, sin- gular — <?-stems, a loc. in -ei and an abl.-inst. -3d; a-stems, a loc.-dat. -ai and an abl.-inst. -ad; /-stems, a loc.-dat. -ei and an 'abl-inst. -id; consonant stems, an abl.-inst. id and a loc. i; plural — the dat., loc, abl., and inst. functions in a single form. In the Latin of the classical period further conditions are noticeable. The locative function has become merged in the ablative-instrumental with relatively few exceptions in the instance of the o- and a-stems. The causes for this syncretism appear to have been the following. After the rise of the meta- plasm in the third declension there existed an abl.-inst. in -la beside a locative in i. At an early period, before / changed to e, the similarity of form might have suggested an identity of func- tion, somewhat before the parallel syncretism in the other declen- sions. Further, in the plural of all the declensions, the locative function was borne by the same form as the abl.-inst. In the latter instance the existence of so large a body of forms having all three forces might well occasion, by analogy, a similar condi- tion at other points in the various declensional systems. It remains to consider whether this condition is characteristic of Oscan-Umbrian as of Latin. First, it is noticeable that the locatives of the o- and a-stems are better preserved than in Latin. In the latter language the instances are relatively few in num- ber, e. g., huml jacere, belli domlque, Romce. In Osc.-Umbrian, however, the type is well preserved, e. g., O. MtriNiKEi terei "in commune territorio," thesaurei " in thesauro," Frentrei " Frentri," aasai purasiai " in ara igniaria," eisai viai mefiai " in ea via media" : U. desire onse " in dextro humero," Fisie-m, Fisie "in Fisio," lolem-e "in civitate," rupinie "in rubinia," Akedunie " in Aquilonia." It seems, therefore, that this syn- cretism has not progressed so far as in Latin. A further ques- tion involves the consideration whether it has affected Oscan- Umbrian in any degree, and, if so, to what extent. For the /-stems the question is complicated by the fact that in Umbrian the abl.-inst. form in -e cannot be distinguished from THE LATIN THIRD DECLENSION 33 the dative or from a possible locative in -e derived from I.-E. -e. T In the case of the abl. in -I it is alone possible to distinguish the form, and of these cases, relatively few in number, none appear to possess a certain locative force. However, such a function is noticeable in the instance of the same form, transferred to the consonant declension, e. g., testru-ku pedi "ad dextrum pedem," nestru-ku pedi " ad sinistrum pedem." It would seem, therefore, that the same force must have belonged to this form in its original position, the /-declension. In Oscan there are three instances of the abl. of the /-stem, e. g., slaagid, akrid, sakrid. Of these the first alone seems to bear on the question in hand. The word certainly denotes a point in space and can possess neither abl. nor inst. force, as defining the preceding word sakaraklum. Biicheler 2 has identified the usage with the Latin "e regione" in the sense "opposite to," "facing." The natural translation of the form, however, seems to be "in loco" and the construction thus becomes locative. If the necessity for avoiding such a force in an abl. form be removed, it seems possible to view the word as locative. Such an objection may best be removed by a consider- ation of the abl. forms of the o-, a-, and ^-declensions. In Oscan no instances of a locative usage without a preposition occur, but with certain prepositions it seems to appear, e. g., up ei'sud saka- raklud i36"apudid sacrum," ap tovtad 231 " apud populum." The evident meaning is locative, 3 as in the plural, e. g., ap eizois 231. But further Umbrian gives evidence of an abl. form used with loc. force, e. g., testru "at the right," termnuco stahituto "ad terminum stanto," vukukum "in luco Jovis," andervomu ser- situ "inter rogos sedito." It would seem, therefore, that the syncretism of the locative with the abl. -inst. in the singular had, in a measure, made its 1 Brugmann, Gr. II, pp. 604, 612. 2 Mommsen, p. 231. 3 The objection to such a view is that the locative force may inhere in the preposition. Delbruck (Gr. Ill, p. 677) identifies the word with Skt. dpi, Av. a* pi, Grk. kiri. Both Greek and Sanskrit use the form with the locative and that seems to have been one of its inherent usages. Hence ap would seem to require a locative force in the word depending upon it. 34 THE LATIN THIRD DECLENSION appearance in Oscan-Umbrian, but to no such extent as in Latin. In general, also, it seems that it was a much later process than that by which the ablative and instrumental functions were amal- gamated. The syncretism in the instance of the nom. and ace. plural of the third declension was relatively late and purely a Latin phe- nomenon. It is supported by analogies of other languages, e. g., Grk. t/jcTs, Tpis; 1 O. B. nom. sg. of ^-sterns in z ; 2 Av. ddtdro, ace. Skt. ddirn; O. B. nosti raky ; In O. N. and West Germ, the nom. and ace. plural have frequently become confused. 3 This cannot, however, be viewed as genuine syncretism. It does not seem that there can have been any intimate syntactical contact, for the functions of the two cases are well defined. A possible explana- tion for Latin may appear in the fact that in the consonant declensions the forms had become identical by metaplasm, e. g., -es from /-stems and ace. -es < -ens. This identity might affect the closely related /-stems. 1 Brugmann, Gr. II, p. 664 ; Brugmann, Gr. Gr. 2 p. 136. 2 Brugmann, Gr. II, p. 532. 3 Kluge, in Paul's Gr. I, p. 386. THE CASE FORMS IN DETAIL. Nominative. — In this case, the endings of the two declensions are in general well preserved. In one category, however, confu- sion may be noted, viz., the original /-stems, which have a nom. sg. differing in no respect from that of the consonant declen- sion, e. g., mens, Skt. matis "thought," Goth, ga-munds "remem- brance," O. E. ge-mynd, Lith. isz-mintis "understanding," O. B. pa-?netc : mors, Skt. mrtis "death," Lith. mirtis, O. B. sz-mrzt6 : ars, Skt. rtis "ratio," O. B. rett : fors, Skt. bhrtis "support," Goth, ga-baurps "birth," O. E. ge-byrd : nox, Skt. ndktis, Lith. naktis O. B. nosU -.pons, O. B. pate "way," O. Pr. pintis, U. putites (nom. pi.) gens, Skt. jatis, Av. fra-zefntis : dds, Skt. datis, ditis, Grk. Sdo-is, 8oio-t5, Lith. dutis "gift" : gldns, O. B. zeladd "acorn" : quies, Av. sti^tis, O. P. siyati : anas, Lith. dntis " duck," Skt. atis. Further the /'-stem character of these words is proven by other facts, i) the consistent form of the gen. pi. in -turn 1 and 2) the presence of early forms of these words with a nom. sg. -is, e. g., opis, Plaut., Bacch., 893, stirpis, Cic, Leg., 1, 8, 26; municipis, C. I. L., 2, 1964; Col. 5, p. 67 ; sortis, Plaut., Cas., 358; mentis, Ennius in Priscian, p. 764, Varro, L. L., 5, § 59 ; lentis, Priscian, p. 764. Various explanations may be sought for these phenomena. It is possible that they are the result of double formation, that there were originally two types and that the consonant stem form has prevailed to the exclusion of the parallel type peculiar to the /-declension. This view may certainly be applied to specific cases, for of the types *nokt- and *nokti-, civitdt- and clvitdti- the former have prevailed in Latin nox and dvitds. In the majority of cases, however, the /-stem was the only form existing from Indo- European times, and hence this condition cannot be accepted as the specific cause in all instances. 1 Neue, Formenlehre 3 I, pp. 276, 277. 35 36 THE LATIN THIRD DECLENSION It is further possible that the reduction of these nominative forms is the result of the loss of / by syncope.' This hypothesis seems hardly tenable for the following reasons. Syncope may be noted for Latin in the instance of a syllable following the princi- pal accent, in case there follow one or more additional syllables. 2 It does not appear that syncope occurs in Latin in a final syllable, save in the instance of the suffixes -ro-, -to- and -ri-, -It-. In these cases, certainly, special conditions may be noted. Hence it will not escape notice that for mors, gens, dos, etc., this process must be supposed under conditions different from those noted above. Further the supposed syncope seems to be of a decidedly sporadic character. It appears that syncope in final syllables, when manifested, affects consistently all vowels save u, e. g., Lat. ager <*agros, sacer <*sacris : Osc. hurz "hortus," cevs "clvis," humuns "homines"; Umbr. pihaz " piatus, "/<?>« <*faunis, fust, ferest, etc. ; 3 Germ, dags, gasts, nahts gen. <*na/ites.* It is, there- fore, in point to inquire why, if such a syncope developed in Latin, it did not affect the parallel formations in o, e. g., mortits beside mors; quietus, quies ; catus, cos ; hortus, co-hors. Likewise if syncope be claimed for -/-, it cannot have been thorough in effect, e. g., vliis, crdtis, grates, vafis beside cos, dos, lis ; sementis, gens, pons ; fortis, ars, mors ; vrbis, corbis, scrobis, turpis, urbs. It is also possible to assume that the apparent reduction of *artis to ars, *gentis to gens, etc., is a further step in that meta- plasm by which the i- and consonant systems have become amalgamated. There are certainly no distinctive reasons why a nom. sg. might not be affected by such a process as readily as other cases. There is nothing more anomalous in the hypothet- ical fact that ^mortis yields to mors than in the evident fact that *morteis has been replaced by % mortes, mortis. But the plau- sibility of this hypothesis is increased by the fact that /-stem forms have developed in the nom. sg. of certain consonant stems, 'Corssen, II, pp. 597, 598, 599. 2 Brugmann, Gr. 1, p. 475; Stolz, Lat. Gr. 2 pp. 321 ff. 3 Buck, Voc, 194 ff. ; Flauta, I, 225 ff. ; Brugmann, Gr. I, pp. 475-6. 4 Brugmann, Gr. I, p. 477 ; K'luge, in Paul's Gr. I, p. 359. THE LATIN THIRD DECLENSION 37 e. g., juvenis, Skt. yuvdn; eat/is, Skt. cvan, Grk. kvwv, Lith. szii; metisis, Skt. «/</!>', (irk. ^7/1/. The original character of these words seems to be proven by the consistent appearance of -urn in the gen. pi. Again, the instances of a nom. sg. in -is in the words in question all occur after the metaplasm has become fully devel- oped and established in certain of the case forms. In a word, the dat.-abl. plural had -ibus and the gen. sg. -es at a period so early as to leave no trace of the original types. On the other hand, a nom. sg. in -is is proven in certain words in which it later appears as -s. A reasonable inference may, then, be that the reduction of *artis to ars, ^mentis to metis, etc., occurred at a period later than that at which the -eis of the gen. sg. was replaced by -es. If, then, this change from -is to -s took place at a time after the rise of the metaplasm, it seems that it was not a cause of that process, that it was rather a resultant of the metaplasm, an effect and not a cause. Hence it would appear that the linguistic force underlying this change was analogy and not syncope. A second group of forms has in the nom. sg. a consonant form, while in the same words in other languages -os appears, e. g., faux, Skt. bhukas "hole "; pax, Skt. pd(as " noose" ; puis, Grk. ttoXtos ; lens, Lith. glindas "louse"; trux, Goth, pwairh-s "angry"; vlvdx, Lith. gyvokas "living"; latex, Skt. latakas ; strix, Goth, strifes "stroke"; falx, Grk. <£oAkos, <£aA/<7??,- lanx, O. B. lakz "crooked," Lith. lanka "valley"; lux, Grk. Acv'kos, Lith. laukas, Skt. rbcas ; senex, Skt. sanakds; vertex, Skt. vdrtakas. To these forms may be added formations in -/ex, referred to -feco-, x and those in -ax, -ix, -ox, e. g., auddx, feltx, ferox. It is first evident that the forms in question cannot be the result of the syncope of final -o-, for there is no evidence of such a loss of that vowel in final syllables. Moreover, the words bear evidence of an /'-stem character, in that they have a gen. pi. in -iitm, e. g., faucium, pultium, etc. 2 It therefore seems that these words must have been /-stems originally or have assumed their 1 Brugmann, Gr. II, p. 239. 2 Neue, I, 3 pp. 272 ff. ; cf. Charisyas, I, 1 17. 38 THE LATIN THIRD DECLENSION known form through the intermediate stage of the /-stems. The former hypothesis is to a certain extent reasonable. An /-stem form may have existed beside the parallel form in the c-declen- sion, as in avis, Skt. cevas. Then the former, preserved in Latin, might have lost its nom. sg. as did mors, gens, dos, etc., e.g., I. E., *polto-s = Grk. ttoXtos ; I. E., * polti-s = Lat. * po/tis = puis, as * mortis = mors. Such an explanation, however, does not seem to apply to a considerable number of forms. In the case of the suffixes -go-, -ago-, -igo-, -ago-, -igo-, it is difficult to assume forms in -/-. These suffixes in question appear in the different languages as follows : Aryan go-, e. g., Skt. dtka-s, Av. attka : Skt. markd-s, Av. mahrka- ; Skt. cuska-s, Av. huska-, Skt. antaka-s, sindhuka-s, etc. 1 Greek, Latin and Celtic go-, e. g., 'nnriKos, do-TiKos, dvSpiKo's; amnicus, civic us, bellicus, mordicus ; O. I. sui/e-c/i, ses-c, bres-c : 2 -g- 3 (consonant stems), e. g., fxupa£, Skt. maryakd-s-; vea£, O. B. novakz ; O. I. nathir, /air, fa/; A -gio-, e. g., 7repio-cros, TrepiTTos, oYo-0-01, Skt. dvika-s, navvo-aia, Lat. pannucia, /xeracraai, tiricraaL, peipaKiov, avSpaKiov, 7raAAd/aov; patricius, sutbricius, fiovicius, artificium, opificium ; O. I. fem. in -ic/ie, Lat. icia, e. g., tairismiche. Balto-Slavic go-, Lith. pi/ka-s, judo-kas, gyvbkas, kirktikas; si/pnb-kas, didb-kas, mazb-kas. O. B. znakz, prekz, mecdkz, to/ikz, kolikz. -gio-, Lith. kume/ike, ratikike, picszkc ; O. B. ovica, detica, karab/icc, kamenccc, jadzcd, pis^cz ; weak grade -gi-, e. g., Lith. jaunikis, czusikis, kiszkis, p/iiszkis, kul'ikis, provikis ; Lett, p/usch- kis, kaschkis, namikis. It is clear that the types -g-, -go-, and -gio- came into Italic. It is possible, therefore, that the weak form -gi- also existed at an earlier period. In fact, this hypothesis is highly probable, in view of the presence of the full form -gio-. It is further supported r) by specific forms in the Italic languages, e. g., U. sumrsim-c 1 The consonant type appears only in viprif (k) (Brugmann, Gr. II, 385); the -ijo- stem in apiciya, Grk. iiriixaa (J. Schmidt, K. Z., 28, 122). 2 Brugmann, Gr. II, pp. 238 ff. 3 The material in Latin is obscured by metaplasm, but there can be no doubt of the presence of the consonant type, e. g., comix to U. curnacu (ace. sg.). "Brugmann, Gr. II, p. 385 ; Wh. Stokes, B. B. XI, 84, 155. S/TY THE LATIN THIRD DECLENSION 39 to Lat. ad murcim; 1 pracoquus, praicoquis, prtzcox; merces beside merx in Char. p. 27 ; 2) by the relations audacia, auddx; efficdcia, efficax; mendacium, mcndax ; pelldcia, pelldx ; artificimn, artifex? etc. ; 3) by such relations as pix, Grk. iriaa-a, Lith. pikis ; faux, Grk. <f>v<ra, Skt. bhuka-s ; calx, Lith. kdlkis ; frnx and friigis (Varro), Goth, bruk-s; lux, 0. B. luca, hut. It may, therefore, be not unreasonable to suggest that many of these nouns in x were originally weak -to- formations, that they have passed into the /-declension in Latin and that their characteristic ending -is has been lost, as in the case of mors, gens, etc. Such an hypothesis, at least, explains the apparent /-stem character of the nouns evident at the point of the gen. plural. For the adjectives, it is necessary to assume that the category contains many original consonant stems, that it is an amalgam of the two types -q- and -qi-. All distinctions have been leveled, however, and the consonant stems have become virtually /-stems in common with many other adjectives. Genitive. — At this point the metaplasm must have affected the forms in the Italic period, though the infection cannot have progressed beyond the stage of confusion (p. 21). Since the separation, however, different treatments have prevailed. Latin gives no evidence of the original /-stem form -et's, 3 Skt. avis, Goth, anstais, Lith. nakt'es. In all instances for both /- and con- sonant stems the form is the original consonant type -es, -os. These endings appear unchanged in Apolenes, Salutes, partus, Kastorus, Venerus, Cererus, patrus, Gesarus, hominus, nominus, prcevdricd- tiduus. The form of the classical period is -is < -es, -os having yielded to the parallel ablaut form. It is evident that the elimi- nation of -eis from the /-stems took place at a very early period and that the metaplasm and confusion established in the Italic period produced this result soon after the separation. For 1 Murcim, Varro, L. L., 151 ; other texts read murcia (strong form). Cf. Biicheler, "Umbrica," vocab. 2 It is noticeable that the noun forms are either fern, or neut. in line with the facts of the development of substantive force in adjectives. Cf. Brugmann, Gr. II, p. 444. 3 Exc. parenteis, an error. 40 THE LATIN THIRD DECLENSION among the genitive forms in -us we note an /-stem, partus. The loss of -eis must have taken place before the decadence of the type -os, since an /-stem would probably not take such an ending after it had lost its linguistic force and was preserved in a small number of forms only. In Osc.-Umbrian, on the other hand, the /-stem form has prevailed, e. g., O. /-stem, aiteis ; con. stems, Appeluneis, Iuveis, LUVFREIS, KUMPARAK.INEIS, MAATREIS, MEDIKEIS, tangitieis, Cdmeis \ U. /-stem, punes, ocrer, sorsalir, con. stems, matrer, nomner, farer. In the prevailing type in Umbrian, ei is reduced to e, 1 sorsalir is a variant, but the / is viewed as from I. E. ei. 2 Accusative. — The prevailing type in Latin is -em, the ending of the consonant stems. This form appears without exception in its original position, the consonant declension. For the /-stems a less general statement can be made. In the majority of cases the metaplasm has been complete in its effect, and the ending original to the consonant declension has been transferred to the /-stems. But the original ending has not been completely eliminated, as in the instance of the gen. sg. Certain words always or nearly always have an accusative in -im, while in others it appears beside -em. In manuscripts of the same author both forms are used, e. g., ndvim, Hor., Carm., I, 32, 8 ; Cic, Att., 7, 22, 1: ndvem, Hor., Serm., 1, 1,6; Cic, de Orat., 1, 38 ; messim, Plaut., Most., 161 ; messem, Plaut., Rud., 637 ; Trim, 32. 3 The ending -im is best preserved in certain words, 4 e. g., turrim, burim, securim, pitppim, sitim, restim, vim, 5 pulvim, tussim. How- ever, the form -em is not unknown in these cases, e. g.,turrem, C. I. L., I, 1259 ; Suet., Calig., 46 ; puppem, Juv., 6, 102 ; securem, Livy, 8, 7, 20 : 9, 16, 17 ; restem, Plaut., Rud., 1036; Juv., 10, 58. These facts seem to indicate that at this point, the ace. sg., the /-stems yielded to metaplasm later than in the gen. sg. If 1 Planta, I, p. 147. 2 Planta, I, p. 148. 3 Neue, Formen. 3 II, pp. 196 ff. 4 Schweizer-Sidler, Gram., p. 152. s Vim probably for *vlm; cf. Brugmann, Gr. II., p. 550. THE LATIN THIRD DECLENSION 41 it be correct to assume that the complete stage, exemplified in the latter instance, is preceded by a stage of confusion, then the ace. sg. may be characterized by that term. Further, it seems that the progress of the metaplasm in this can be observed within the literary period. Certain words, such as turn's, 1 have em much more frequently in the later writers. Hence we may assume that the metaplasm, at this point, is at an incomplete stage, but yet near to and constantly approaching completion in the elimination of the ending -im. For the Oscan-Umbrian period we may infer, 1) that the /-stem ending remained unaffected, e. g., O. slagim, tiurri, SAKRIM, DEKKVIARIM : U. SPANTIM, AHTIM-EM, SEVAKN] | 2) that the consonant-stem ending -em was replaced by the <?-stem ending after the metaplasm between the 0- and consonant stems had developed, e. g., O. leginum, tanginom, medicatinom : U. arsfer- turo, curnaco, capirso, uhturu, salu. The sole variant from these types appears in the Umbrian in -e.m -e among the /-stems, e. g., sakre, uvem, staflare. It is hardly possible to assume that this ending has come from the consonant declension, for in the Osc. -Umbrian period that ending -em was eliminated from its original position, yielding to -am from the ^-sterns. It is not reasonable to assume the reappear- ance of these forms at a later time in Umbrian alone and in the /-declension. The facts of Umbrian orthography show that / was open relatively to / and that, further, / was sometimes indi- cated by e. On the other hand, no such conditions govern the case of original e. There could be no reason for a differ- ence of orthography and change of e to i* In the final position an original -im might well appear as -em, even in the majority of instances. Such are the facts, for Umb. has -em -e and -im -i in the proportion fourteen to five, e. g., peraknem, perakne (3), perakre, purtifele, randem-e, sakre, sevakne, staflarem, staflare, Tadinate, Tarsinatem (2) : ayliiu-'E.u, ferim-e, smursim-e, spantim, spanti. Planta 3 holds that -em stands for -im, while 1 Georges, Lex. der Lat. Wortform., "turris." 2 Buck, Voc, p. 45 : Planta, I, p. 100. 3 Planta, I, p. 100. 42 THE LATIN THIRD DECLENSION -im is to be explained on the ground that the forms are -w-stems with ace. -tin. The grounds for such a reference are. however, not readily apparent. Ablative. — It has already been shown that metaplasm occurred at this point in the Italic period (p. 30) and that -Id was transferred to the consonant stems. In Oscan this form alone appears 1 (p. 33). For Latin and Umbrian no such general statement can be made. In Latin of the classical period a form in -e is in the ascendancy for both -i and consonant stems. Umbrian, likewise, contains numerous forms in -e, e. g., pure, vapede, karne, tikamne, nomne, menzne, curnase, pase, due. This form in -e may be viewed as a possible locative or instrumental suffix, from i loc. or -e? a 3 inst. From the point of view of Umbrian, however, the last possibility is excluded. 4 Hence, for the Italic group as a whole, the decision must lie between the loc. and inst. in -e. Preference may be given to the former alternative for the follow- ing reasons. The suffix e for the inst. has not been generally accepted. To suppose -a would involve the separation of Latin and Umbrian. It is probable that -Id came into the conso- nant declension as a syncretized abl.-inst. form, and that the inst. would then cease to exist as a separate form, as in the vocalic stems. From this point the forms may be traced as follows. The consonant declension in the Italic period must have possessed an abl.-inst. -Id and a locative -/. Later there arose a syncretism between these three cases, and hence the forms -Id and -/ (Latin -e) came to be used indiscriminately for all three cases. The type -i has been lost in Oscan ; in Latin and Umbrian it has remained in the form -<?. Within the Latin period a further development has taken place. The type -<? has been transferred to the /-stems, and this is in accord with the facts noted for the gen. and ace, that the consonant stem forms predominate in the singular. We may 1 The consonant stems have also been affected by metaplasm, e. g., ligud, tanginud (cf. Brugmann, Gr. II, 594). 2 Brugmann, Gr. II, p. 594. 3 J. Schmidt, K. Z., 27, 292 f. Pluralbild., 41 f. *Buck, Voc, p. 39. Planta, I, p. 563. THE LATIN THIRD DECLENSION 43 also note the same incomplete stage and the same element of confusion as in the ace. sing. In the consonant declension the type -id remains beside that in -e,' e. g., capiti, Catull., 68, 124; scelerl, Sen., Phaedra, 693; coventionhi, C. I. L., I, 196 -.sanction;, C. I. I,., I, 198; virtutei, C. I. L., 1, 34 : airld, C. I. L., I, 61 : corporl, C. I. L., I, 206. Among the /-stems i exists beside -e, e. g., sorti, Plaut., Cas., 428 \parti, Plaut., Men., 479; and very frequent in igni, Tiavi, puppi, turrl, etc. The same confusion is noticeable here as in the ace. sg. in the case of individual authors, e. g., nepti, Tac, Ann., 3, 24 : nepte, Tac, Hist., 5, 9 : dassi, Verg., ALn., 8, 1 1, classe, JEn., 1, 379 : civi, Cic, Farm, I, 9, 15 ; cive, Verr., 2, J 3> 32- Umbrian, also, has forms in -e among the /-stems, and it may be assumed that this type has been transferred from the conso- nant declension. However, the influence of the consonant stems in the metaplasm seems to have been a Latin phenomenon. Elsewhere in Umbrian the /-stem forms have prevailed, and hence the presumption is against the hypothesis of such a trans- fer. But, further, it is possible that e has arisen within the /-declension, and that the distinction is non-essential and ortho- graphical. Planta 2 has shown that i may appear as e in Umbrian and so it is reasonable to accept the explanation, which obviates the necessity of assuming a transfer. Umbrian has certain forms in -e among the /-stems, which possess a locative force, e. g., ocre-m, ocre. No definite determi- nation of these forms can be made. They may be 1) locatives in -e, 2) identical with the dative in -e with a loc. force still present, or 3) the ablative with loc. function, as in the ^-declen- sion, testru, termunco, or the Lat. ablative-locative. 3 Dative. — The endings of the dat. sg. in the different dialects are the following: Latin has i in both /- and consonant declen- sions, e. g., urbi, regi, leg?, virtutei, Apolenei, fraudei. Oscan has 1 It is evident that -e lias prevailed to the exclusion of -i to a much greater extent in the consonant declension than among the /-stems, since the type -J is relatively infrequent among the former. 2 Planta, I, pp. 107, 108. 5 15rugmann, Gr. II, p. 612. 44 THE LATIN THIRD DECLENSION -ei, e. g., Herentatei, 1 \inrz\ovvu, Diuvei, Aiovfci, Iuvei, FUTREI, KVAISTUREI, LEGINEI, MEDIKEi, PaTEREI, SVERRUNE1, vezkei. The form for Umbrian is -e, e. g., Tadinate, Tarsinatc. ocre, adferture, karne, nomne, Iovie, pase, kapide. There are also two forms in J, e. g~, Iuvip "Iovi patri " and Marti. It is evident that the O. -ei must be referred to the locative of the /-stems -ei and this reference involves a metaplasm, since the form appears in the consonant declension. The Umbrian types e and rarely i are not so clear. The forms may have arisen either from -ei of the /-stems or -ai of the consonant declension. Hence it is possible to avoid the assumption of metaplasm and refer each to an origin within its own declension. It might, how- ever, be expected that Umbrian would agree with Oscan and this probability is further supported. It has been noted that Umbrian has two forms in -J. This condition is more nearly in accord with the treatment of -ei than -ai in Umbrian, 2 since the latter does not appear as -I. In Latin, also, the forms may be separated and referred, respectively, to -ei in the / and ai in the consonant declension. 3 The forms of both languages, however, are in accord with the assumption that the metaplasm occurred in the Italic period. The two declensions had become merged in the dat.-abl. plur. and abl. sg. and this condition might pro- mote identity in the dat. sg. If the metaplasm occurred in this period, the forms of all three languages can be referred to the same origin, the /-stem ending. Latin also possesses a small number of dative forms in -e, c. g., Iunone, C. I. L., I, 172, metre, id., 177, salute, 179, Marie, (3) 62, patre, 182, Diove, 188, victore, 638, llctore, 1060, Pilemone, 1 104, Hercule, 11 70, love, 1402, and in formulas, e. g., iure dicundo. This type may be from a locative of the /-declension in -e.* It is further possible that the -e is an orthographical variant for the normal type -J. 5 This is supported by the anomalous parallelism 1 Possibly the only z'-stem, from -tat or -tati-. 2 Planta, I, pp. 143, 147, 148, 368. 3 Brugmann, Gr. II, p. 603. 4 Brugmann, Gr. II, p. 604. sSolmsen, I. F. IV, 137. THE LATIN THIRD DECLENSION 45 in Diovei victor e, C. I. L., I, 635, and by a similar representation of -1 in the nom. plural, e. g., ploirume, illvire and abl. plural cava fumes, mentovines, nuges. Nom. and Ace. Plural. — The forms of these cases for the Italic period were respectively : /-stems, nom. plural -es < -eies, c. g., O. aIdilis, tris " tres"; U. sacres, sacrcr, puntes, pacrer : ace. plural -ins, e. g., U. avif, aveif, trif 1 : consonant-stems ; nom. plural -es, e. g., O. humuns, meddiss, usurs, 2 ace. plural -ens < -tis, e. g., U. nerf, capif, vapef, vef, buf. In Oscan and Unibrian the forms are unaffected. Latin, however, has conditions differ- ing from those of the Italic period. The original forms would give for the /-stems, nom. -is, ace. -is : for the consonant declension, nom. -es, ace. -es. As a matter of fact, however, the /-stems have both -es and -Is in the nom. and ace. In the conso- nant declension there are no certain instances of -es, -es is the prevailing form, but -ts occurs, though less often than among the /-stems. 3 From these facts may be inferred, 1) that in the nom. plural the /-stem form was transferred to the consonant stems at a very early period : 2) that the ace. plural in each declension was pre- served, becoming -Is and -es, respectively ; 3) that a syncretism arose among the /-stems, and in consequence -es and -is came to be used as a pair of forms, valid for either function ; 4) that this confusion among the stems has affected the consonant declen- sion and caused the appearance of forms in -is in both nom. and ace. function. The confusion of -es and -is may be explained on the ground that, in the closely related consonant declension, the forms were already identical. This identitv of form may have suggested an identity of function also among the /-stems. Genitive plural. — The types which prevailed during the Italic period were respectively -ium and -inn, e. g., O. aittium ; i.iiMi- tum, fratrCm, nerum ; U. peracrio, peracnio ; fratrom, kratru, buo. 1 Planta, I, p. 505. -' I'lanta, I, p. 229 ; Buck, Voc, p. ig6. ' Brugmann, Gr. II, pp. 665, 666. 46 THE LATIN THIRD DECLENSION In Latin one may observe a stricter adherence to the original forms than in any of the case endings previously considered. The /-stems have -turn with great uniformity in cases where an ace. in -im and an abl. in I rarely occur. Such are the types represented by auris, neuters in -e, -a/ and -ar, imber, dddes, and the forms with reduced nominative, urbs, ars, gens, etc. On the other hand, most consonant stems have -um, e. g., dux, pes, pater, etc., and notably the original consonant forms, invents, cants. It is, however, evident that infection and confusion are present even in the gen. plural. This may appear in two ways, i) con- fusion of a general nature, and 2) metaplasm manifested in par- ticular categories. The former type is represented by such instances as apum beside apium, nubum, nubium, ccedum, ccedium, etc. 1 These words show the characteristic of the opposite declen- sion, though less often than the original form and in many cases rarely. The facts are in line with those noted for the ace. and abl. sg. and, indeed, such a metaplastic effect might well be expected, since the declensions are so generally affected at other points. Of the particular categories to which reference has been made the most important is that of the stems in -nt-. These are undoubtedly consonant stems. These words have -turn beside -um even in substantival usage, e. g., parentum, parentium, dientum, rfientium, nefanium, nefantium. It is possible that this condition has arisen as follows. The forms were originally participles and hence closely allied with the adjectival function. It is further evident that /-forms have assumed predominating force among the adjectives and have displaced consonant forms more often than among the substantives. Hence we may assume that -turn was first established in the participial and adjectival usages. From this point of contact -turn might well spread to the sub- stantives. Among the /-stems there are certain words which have -um with unvarying consistency, e. g., hospes, O. B. gospodt ; anas, Skt. Ms, Lith. anils, and many nouns which may have been /-stems at one point in their history (p. 39), e. g., artifex, senex, municeps, 1 Neue, Formen. 3 I, pp. 259 ff . THE LATIN Til IK I) DECLENSION 47 prineeps, etc. It is noticeable that these words are words of more than two syllables which have lost their /-form in the nom. sg. The dissyllabic forms have a different history, since mors, pons, etc., retain -turn. It is remotely possible that the heavier forms were more completely subjected to metaplastic influence than was the dissyllabic type; that the nom. -is was lost earlier and that, in general, the metaplasm was more effective in all the case forms. The fact that no traces of -is remain, as in sortis, mentis, etc., may point to this hypothesis. e UNI UNIVERSITY OF CALIFORNIA LIBRARY, BERKELEY THIS BOOK IS DUE ON THE LAST DATE STAMPED BELOW Books not returned on time are subject to a fine of 50c per volume after the third day overdue, increasing to $1.00 per volume after the sixth day. Books not in demand may be renewed if application is made before expiration of loan period. we:*.t-£i 15m-4,'24 ^D 00096 /