MA S TER NEGA TIVE NO. 92-80599 MICROFILMED 1992 COLUMBIA UNIVERSITY LIBRARIES/NEW YORK as part of the 'Foundations of Western Civilization Preservation Project" Funded by the NATIONAL ENDOWMENT FOR THE HUMANITIES Reproductions may not be made without permission from Columbia University Library COPYRIGHT STATEMENT The copyright law of the United States - Title 17, United States Code - concerns the making of photocopies or other reproductions of copyrighted material... Columbia University Library reserves the right to refuse to accept a copy order if, in its judgement, fulfillment of the order would mvolve violation of the copyright law. AUTHOR: RADFORD, ROBERT S. TITLE: LATIN MONOSYLLABLES PLACE: [S.L.] DA TE : [1 9031 *.» COLUMBIA UNIVERSITY LIBRARIES PRESERVATION DEPARTMENT BIBLIOCRAPHIC MICRnFORM TARHFT Master Negative ff Original Malerial a$ Filmed - Existing Bibliographic Record Restrictions on Use: BKS/PROO Books FIJL/BIB NyCG92-B320 J. B Record t of O - Record added today IDsMYCG'92-B320^18 C C : 9,6 6 8 E«'L T ;• a m CF-*5:xx L:eng PB 2 1903/ TYP:a STsp DCFg? CBC;? MOD? INTi? GPC:? BIO:? REPs? CFI ,-? DM: 040 100 1 OR 2 POL: R a cJ f o r d , R o b e r t S o iri e r v i 1 1 e RR: Ai rqu I s i !.: i oijs NYCG-PT FRM: MS; : E:Ls AD: 05-07-92 Sl^m'i ATC: IJD: 05-07-92 F I C : ? COMy??? FSI 5? ILC;??-;^'? Us? COL : EriL:; GEN: B5E: 245 14 The Latin MonosyllaW.eB in their Relation to Accent T^nd Quanti ty= I hEmi ^ /•• n rofCirm3.= I bA Study in the Verse of Terence Cs.i . ., « I bs .n , , ^f I cl 9033 . C6a;i-.ia3 p. LDB GRIG QD 05-07-92 TECHNICAL MICROFORM DATA fx^h^^n o/?^-^-^-'^-~7-V- REDUCTION RATIO: ( / K IMAGE PLACEMENT: lA ^IA) IB IIB — DATE FILMED:___6_-_l£Jl INITIALS 1\a OX- FILMED BY: RESEARCH PUBLICATIONS. INC WOODDRIDGe7ct c Association for Information and image IManagement 1100 Wayne Avenue. Suite 1100 Silver Spring, Maryland 20910 301/587-8202 Centimeter 1 2 3 iiU iiiilmiiiiii Inches Mil TTT 5 Ml 6 iJiiiili 7 8 9 10 ilmi|i||i |iiiiliiiiliiiiliiiilii I TTTT 11 12 13 iiiliiiiliiiiliiiiliiiiliiii 1.0 Ui 28. It ^ 1.4 2.5 2.2 2.0 1.8 1.6 I.I 1.25 TTT 14 15 mm iiliiiiliiiil T MflNUFfiCTURED TO RUM STRNDPRDS BY APPLIED IMfiGE, INC. ExtraUed from the Transactions of the American Philological Association^ Vol. xxxiv, 1903. ^KA /cl^xn. VL — The Latin Monosyllables in their Relation to Aceent and Quantity. A Study in the Verse of Terence} By Pkof. ROBERT S. RADFORD, ELMlKvV COLLEC.E. I. Problem and Method. It is the purpose of this paper to investigate the cause of the syllable-shortening which occurs after short monosyllables in early Latin verse, and also in a more genenil way the j)art which monosyllabic words play in the accentuation of the Latin sentence. The data illustrating the quantitative and accentual relations of monosyllabic words will be drawn almost entirely from the six plays of Terence, although simi- lar results have been obtained by the writer from an exami- nation of ten of the plays of Plautus. In a study which involves the Latin sentence-accentuation it is scarcelv neces- sary to state that the writer is greatly indebted to the brilliant labors in this field of Ritschl,^ Klotz, Lindsay,- Skutsch,Wack. ernagel, and many others, not to mention the detailed inves- tigations of Ahlberg, Podiaski, Kiimpf, and O. Brugman. Special indebtedness will be acknowledged in each case, but it will be readily understood that my total indebtedness to these scholars is greater than can be indicated in single references. The two most frequent forms of syllable-shortening in early Latin occur in iambic words and in words preceded by a short monosyllable. In the case of iambic words the shortenino- is 1 A paper treating the same sul)ject in relation to the verse of Plautus was read by the writer before the Johns Hopkins Philological Association, Ai)ril 24, 1903. An abstract of this paper has already been published in the Johns Hopkins Uni- versity Circulars, Vol. XXIII, and the paper will appear in full in an early number of the American Journal of Philology. 2 I think it proper, however, to state that at the time when I reached the main conclusions of this article 1 was not acquainted with that part of Ritschl's Prole- gomena which treats the same subject, nor with Wackernagel's article in Imlogcrm. Porsch.^ Vol. I. Y«il. xJixiv.J Lmtin M&nosjiiabies, now very generally attributed to the word-accent upon the initial syllable; a slight modification of this view would be to hold that tbc shortening is due to the analogy of pyrrhic wo^rdi. s«cli, as rtw and mgii, m which the quantitative rela- tions do not retard the development of a clear stress accent. A second and still larger class of shortened quantities arises when a short monosyllable, or dissyllable which has become a iiionosyllahle by elision, precedes another word, as sed abstu- Msii, sed niyr/i/i. What is the cause of this shortening ? Why is it that the short monosyllable has the power of short- ening a following long syllable? Upon this question students of early Latin verse are far, I think, from having reached an agreement It happens, necessarily, from the very structure of iambic verse, that a long syllable shortened in an iambic sequence is in every case either preceded or followed by the verse-accent ; upon this coincidence rests the older view, still held by Seyffert, Leo, and Brix-Niemeyer, of an artificial^ shortening produced by the verse-accent. Accentual scholars are themselves greatly divided in opinion at this point, and it does not seem necessary to mention all the theories which hme licen proposed, that of Klotz being especial!)- difficult nf acceptance, at least in the extreme form in which it is stated^ Gnmdzii^f, p. 6^ ff., p. 45 ff. Ahlberg ( Dc corn ft. iamb. PimtL fmaesiimes, Lund, 1901, p. 52 If.), and Lindsay {Tif Cafiipi, London, 19CX), p. 35*), have, it is true, clearly pttt forward the view that some form of word-grouping has taken place, and that the group-accent has in some way come * Tlic use of tills term must not be imderstood as implying that I cjuestion in anapaestic verse or eveii occasionally in the beginning of a verse or colon of iaiiiMc or trocliaic verse the occnrrence of an ** acltticiiil " or metrical shortening pn»(luced by the very concrete, the very real verse-accent. If the cases of sed- iiittm, sed-llhim were only a few in number, I should be ready to accept the metrical exi»lanation, to which 1 have no the«>relical objections. But since these cast s nre very numerous, and since we know from many ancient testimonies that it is the special characteristic ©f Latin iaiibic verse to reproduce the carlence of collocjuial speech, and thus tO bfjlg tlie verse-accent (as a sul»ordinate factor) into cooperation with the word-accent, the uietrkal eiplanation appears to me untenable for iaml)ic verse as t whole* - Ill his earlier writings (e^, L&i, Ijimg.tf. 202), Professor Lindsay hesitated betweea tlii» liew «ii4 the explanalion given by Kloti, Crundz^ p. 68 ff. 62 Robert S. Radford. [1903 into play, but neither of these scholars has undertaken to explain how the primary or secondary accent has come to rest upon the atonic monosyllable in a group like sed ilU\ sed dbstullstL So far as regards the word-grouping, Havet also long ago pointed out that sed-abstiilisti forms to a certain extent * one body ' (see the quotation of his views in Plessis, ed. AdelpJioe^ p. 5), and in his Mctriquc {e.g. §§ 353, 363) he regularly links the monosyllables by a hyphen with the follow- ing word to indicate their proclitic character, but the particu- lar explanation which he gives of the shortening in this group is so intimately connected with his theory of * initial intensity ' that it does not call for further notice here ; on the theory of * initial intensity * I shall make some observations in a later section of this article (see p. 95, below). I wish to bring forward for the solution of this problem a method which as a whole has been hitherto untried. I pro- pose in brief that for the moiiient we shall wholly dismiss the question of syllable shortening, shall observe the short mono- syllables when they precede not long but short initial sylla- bles, and determine independently the place of both the primary and the secondary grammatical accent in the groups thus formed. This means that we shall first determine the place of the primary accent in the groups sed Jiic (dissyllabic group), sed agit (trisyllabic group), sed Jiomines (quadrisyl- lable group), and the place of the secondary accent in the similar groups sed amove, sed amavernnty sed Jiilaritudo. It is not possible, to be sure, to determine directly the place of the accent in all these groups, nor is it necessary. A single group, the trisyllabic, offers exceptional opportunities for the determination of its accent, and affords the one point where an accent-law may be established, capable of extension in principle to all the groups. A trisyllabic group of this type (w. w^) may assume two forms; it may either form an anapaest, sedagiuit, or a tribrach, sed agit* Now from the first of these forms, the anapaestic, we can derive no aid; for, owing to the structure of iambic verse, the group can never occur without receiving the verse- accent either upon the initial syllable, sM agnnt (much the W ijl' • Ji' JUtl IT m I' Laiin Motwsyllables, 63 J iiiore frequent form), or upon the ultimate sed agthit (less frequent form). Hence in this case we can draw no conclu- S011S as 10 the natural accent. But liitli tie second form of the trisyllabic group, viz., the tribrach form, sed agit^ the case is wholly different ; for so far as concerns the structure of iambic verse, the group is perfectly icxible and may receive the ictus equally well and with nearly equal frequency either upon the initial syllable, Sid agii, or upon the second syllable, sed dgit. So far, how- ever, as concerns the word-grouping, the group must not admit the ictus upon the second syllable, if in consequence of peculiarly close grouping it has acquired the grammatical accent of a single tribrach word, i.e. the grammatical accent, sM agiif belonging to an improvised compound (as we may say)^ just as the accent ineo belongs to a permanent com- pound. For it is a well-known rule of Latin iambic verse that a tribrach word receives the metrical accent in general only upon the initial syllable and thus almost always in agree- ment with the grammatical accent, i.e. regularly geuere, very imtljgenire. Before we proceed to apply the test just indicated to the verse of Terence, it will be necessary to comment briefly Upon the general character of the Latin monosyllables. Accoitiing to an of^repeated rule of the grammarians the monosyllables are usually without the accent.^ Thus Priscian (Kcil, G.L. Ill, p. 479, 20=5 Scholl, De acccntu linguae lat.^ p. 194) in speaking of the accent of iam (in iam dudum) says : Gravcm., ul omnia fere monosyllaba praepositiva ; see also I'M/., B. 478, 21 (accent of n/); p. 24, 21, etc. This rule does «ot apply of course to „,o„osynabic nouns and verbs, as many other testimonies of the grammarians show (cf. Scholl, i2rii«¥., p. loSf.Xbiit only to those words which, owing to their meaning, are naturally unaccented in many languages, vii., the monosyllabic prepositions, conjunctions, pronouns, and adverbs (see, #.^., Wallin, Yale Psjclwl. Lab. SittdieSy IX p lit; Meycr^Luhke, G rammed, mmmn, Spr., I, p. 503; ^ I* llilkr reaches the same conclusion on metrical grounds, Res Metrica\ 64 Robert S. Radford. [1903 G. Paris, Role de V accent lat., p. 19). Hence there seems no good reason for doubting the substantial truth of the gram- marians' rule, which means no more than the similar rule respecting the accent of prepositions, i.e. the prepositions — hoih praepositiones compos itae and adpositac — are unaccented in a considerable majority of the cases in which they occur, as confero, conferre^ im navem, ex bello. Besides, as will be shown later, the grammarians often distinctly recognize that the monosyllables may acquire an accent when they form part of a compositum or word-group. Thus it is evident from the grammarians* statements that the monosyllabic conjunc- tions, pronouns, and adverbs bear precisely the same relation to the accent as the monosyllabic prepositions. This con- clusion is expressly confirmed by Audax (Keil, VII, p. 360, I £f.*), to whom we owe the clearest account of the accent of these particles that is to be found in Latin literature subse- quent to Quintilian : non omnes partes orationis aequales sunt . . . nam et pronomen subiacet noniini, et verbo servit adverbium . coniunctio quoque et praepositio ad clientelam maiorum partium pertinent . hae ergo partes, quae adpendices sunt, sic maioribus copulantur, iit tanqitain in iinam partem orationis'^ coalescant, proprium vero fastigium perdant, non omnes dumtaxat, sed pleraeque. It is necessary also to review briefly the treatment in iambic verse both of primary and of secondary word-accents belong- ing to the syllable-group, ^ ^ w : A. Primary Accentuation. —To determine with precision the place of the primary grammatical accent in trisyllabic groups of the form vy, w w is possible only through the fact that tribrach words such as gencrc do not as a rule admit the metrical accent upon the second syllable. Yet the statement sometimes made that Latin tribrach words never under any conditions admit the accent genire in iambic verse is not * This is substantially the same as the anonymous quotation in Scholl, De aec.y p. 175 f., entitled « Interr. et resp." which, however, is not placed, as might be expected, in the chapter on the accent of conjunctions. 2 • Una pars orationis ' is the special term which the grammarians employ of the compositUy which, like htiittsce modi, istius uiotH, etc., are known to be such through the test of the accentuation; cf. Scholl, De ace, p. 124 ff. Vol. jcxxiv.] Latin Mt^nosj^ilabies, 65 66 Robert S. Radford. [»903 altogether accurate, so far at least as regards the first foot of a ¥crse or colon. ; and since we can scarcely expect that tribrach groups shall be treated more rigorously in this respect than tribrach words, it is important to state the usage of the iambic poets in respect to the first foot somewhat more fully than is done by Klotz, Gmuds., p. 274 ff., or by Kitsch 1, Froieg., p., ccxxv If, The evidence is conclusive that the Roman poets have sometimes admitted in tribrach words and in words ending in a tribrach the accentuation £-efi/re in the first foot of a verse or colon, although far more rarely and with much greater hesitation than they have admitted the accentuation /feffore in the same place. The usage of the later poets may be found in L. Miiller, AVj i/r/r., p. 168: Seneca and Prudcntius have each one such accentuation of the tri- brach in the first foot {/i}gimm, g-m/re), not to mention the more frequent cases in less careful writers such as Ausonius, Avienus, Terentianus, etc. From the Christian poets also some examples have been collected by Hiimer, Latdmisih' Ckrisiiickefi Rkjtkmefi, p. 27. Examples from the metrical inscriptions are given by llodgman, Hartfard Studies, IX, p. 139, i^. CLE, 67, 3 iidqite ; 92, 3 Siephdne ; 211, 3 iiimla. Ahlberg {De fr^ceieusmaiieis aMtiqume fmsis iat,, Lund, 1900, I, p. 32) accepts the accentuations sequiminei Merc. 782, muiiifis (first foot of second colon) J/<;a/. 169, Mincrua, Baech. 893. Hence, even in cases where a correction would involve little textual change, it appears unnecessary to follow Langen {Phihfogus^ XXXI, p. 109), and recent editors in emending MiL 1 1 20' ililiif (Gotz and LeO' : itmn\ Andt\ 478 Mcine{\]n\^i.: kk mparatiim\ Caecil. emm. frgm. 232 egone (retained by Ribbeck^), or to adopt with Scholl a change of order in Cas. 564: komhiim amatorem (cf. Humphreys, Trans. Am. Phil. Assoc. ¥H (li76)i p. 132 t)}' Examples in other feet than the irstt however, are so rare in the republican and early imperial poetry that they must be regarded with extreme sus- picion and are commonly removed from the text by transpo- * Aid tlio lli-iiftf, Mimi. 507; i^ntf Cure. 119 (in cretic verse; Gotz, igon)\ I ri^rinM, Ace. tr«g,frgfn. 381. I i sitions, yet it is important to recognize that a few such cases occur in our Mss. in texts otherwise free from suspicion, i.e. Most. 1 100 scr/rc, A feu. 8yy validus, Andr. 596 corriga\\ Sen. Med. 26J fetniuM ; see also Klotz, Gnindz., p. 274. Quite 2 similar to the treatment of tribrach words is the treatment in iambic verse of the compound phrases (which arc often written as one word) ending in a tribrach or an anapaest, such as adeOy iniln, veluti, interibi, in t ere a, intereos, proptered, prop- tereos, etc. According to the ancient orthography these phrases may equally well be written separately, i.e. in ibi, ad eo, inter ibi, inter ed, etc. (see, eg., CIL. I, ind. p. 609), but so far as concerns the accentuation, they apparently never admit of separation into their component parts (decomposi- tion), but always receive the metrical accent upon the ante- penult, i.e. in ibi, ad eo, inter ibi, intt^r eos. B. Secondary Accentuation. — I have so far discussed only cases of the primary accent in tribrach words, and students of Latin verse-accentuation have generally been content to confine their treatment to these cases. Of almost equal im- portance, however, are the cases of tribrach sequences which involve the secondary accent, that is, in which the syllables immediately preceding the primary accent form a tribrach series, as cdldniitdtem, miseridrum, fdmilidrcm, Jiildritiido, etc. I have not been able to find in current discussions of Latin verse-accentuation any treatment of cases of this sort except a brief reference to somewhat similar cases in an article by Lindsay, PhiloL, LI (1892), p. 373 (footnote). In point of fact the secondary accent in cdlamitdtein is observed almost as strictly by the Roman dramatists as the primary accent in g^nere. Of the nearly 200 cases of this kind occur- ring in Plautus, the secondary accent appears to be disre- garded in only two or three, e.g. in the iambic verse-close of AliL 562 nidlitiose tamcn and in the bacchiac verse-opening 4 5 6 of Cist. 3 dp^rfiistis. [With respect to the latter case it is possible, but not especially likely, that the Latin bacchius admits the accentuation ^u k^ 1. instead of the usual -^ do _:_ ; cf. Christ, Metr.f p. 419 f.] Terence has apparently violated Vol. xxxiv.J Latin Monosyllables. 67 68 Robert S. Radford. [1903 tic secondary tribracli accent twice, once in the middle of the ifcrse, once in the verse-close: Andr. 941 aim tud r^lfgwne Mum ; Heaut. 906 dpirfsere SstMm, In both these' cases ©tier scansions than those which I have adopted are possible, le. rilii^ione, op^mire, but not especially probable (cf. Hauler! Eini, P/iorm,, p. 54»). The elision of the final syllable in both cases prevents a double coniicl and apparently renders tie single conflict somewhat less harsh, but is far from pro- ducing a recession of the accent, i.e. religiofie\ as has some- times been supposed. It must be added that editors of Terence have iiot always sufficiently regarded this secondary accent in tieir conjectures ; thus the reading >r///V^/Xr;;/) has been adopted by all recent editors except Umpfenbach in Andr. 232 instead of the Ms. faailiatem, but is wholly inadmissible in view of the unusual conflict, and the same is true of Fleckeisen's conjecture cttpiditat{e\ Heaut. 208. Finally, this accentuation occurs once in the first foot of Saturnian verse, if the quantitative view of the Saturnian be correct, i.e. Naev. 109 (Havet) Sieilifm/s paciscil We may sum up tic conclusions which we have reached as lollows: In words and compound phrases forming a tri- brach such as genere, inibi {in ibi\ etc., the grammatical accent is in rare instances disregarded in the first foot of a U'Crsc or colon ; tiere are also a very few cases, chiefly in the irst foot or the verse-close, of the disregard of a secondary tribrach accent in words like chldmmtem. It is evident that in any tribraci word-groups which wc may discover to exist tic same licenses wlil be .admitted. It only remains to note briefly tiat tie treatment of syllable-groups forming a dactyl (_ %/ v) is considerably less strict ; that in dactylic words such as pectore a primary grammatical accent is ratier freely dis- regarded ia fie flfst loot and occasionally disregarded in the other feet (cf. Klotz, Grunds,, p. 274 ff.); further, that a secondary dactylic accent, such as that of cdmmddttdtes, was rather freely disregarded by tie Roman iambic poets prior to the time of Hitciriis » in... tieIr strenuous endeavors to form ^ Phaedrus has avoided these accentuations almost entirely; compare, how- ever, in the wise-close App. 21, 12: deiinuisti pedh. i « « ^ ^ 4 legitimate iambic and trochaic verse-closes — thus in the verse-close, with harsh double-conflict : Andr. 569 at si cor- rigitur, quot commoditattfs vide ; PJionn. 843 ; cf. 284 ; cf. 676; Hec. 122; Ad. 880; Pacuv. trag. frgm. 164 R. ; id. inc. LIV R., etc. — also occasionally in middle of verse, with eli- sion of final syllable, i.e. with single conflict: Andr. 844; Hec. 797; Yjoh^t. com. frgm. 113 R. ; Afran. com. frgm. 7, etc. ; finally, we may note that dactylic compound phrases such as attamen^ quomodo, ncscio, and obviam admit to a certain extent — most often in the first foot — of being treated as two words through decomposition, i.e. at tdmcn, quo modo^ ne scio, ob viam (cf, Klotz, Grunds., p. 276; Ritschl, Prolcg., p. ccxxxvii ; Skutsch, ForscJi,, p. 158). The reason for the difference of treatment in the verse-accentuation of tribrach and dactylic words need not be discussed here (compare, however, Klotz, Gnmdz.^ p. 278 f.); I shall only remark in conclusion that, owing to the decomposition (recomposition) which is admitted even in the most frequently occurring dac- tylic compositay it is not possible for us to employ as rigorous tests for the determination of the regular accent of dactylic groups (_, w vy) as can be employed in the case of tribrach groups (v^, w w). II. Comparison with Greek Verse. It is evident that if, in pursuing this investigation, we shall find the Latin tribrach group accented either wholly or in very large part upon the initial syllable, i.e. scd agit, the question will still present itself, whether, after all, there is not some reason in the special kind of iambic or trochaic verse, some reason in the placing of the caesurae for the non- occurrence of the accentuation sed-dgit. With this question in view, I have examined about 1 500 Greek iambic and tro- chaic verses in Philemon and Aristophanes, and I find that in Greek, where the influence of the stress-accent does not exist, the two accentuations occur with equal frequency in trochaic verse and in the longer iambic verses, in 500 verses of this sort the metrical accent falling 9 times upon the sec- Vol. xxxiv.] Latin Monosyllables, 69 ©ed syllable and an equal number of times upon the initial syllable, wWk In tlic iambic trimeter the accent upon the .second syllabic bcars' to tie Inkial accent the ratio approxi- inatcly of 2 : 3.^ Thus the trimeter alone is found to be some- what more favorable in its structure to the initial accent for tic reason that of its two chief caesurae the simujuinaria is more frequent than the semisiptetmrm. The two ni.ost coni- .«©© forms of a Greek trimeter containing" one of these trisyllabic groups may be seen from the following verses: Wkhjrgm. 90, I (Kock) : pj^i^u rk Kitak ttwowi | ttoAAoi', Uattora., M,/rgm. 90, 6 ToaovTo yly wi \ to kukov ^Xikov ntp yu. The results obtained through an examination of Greek verse may be stated in detail as follows: In more than 600 iani'bic trimeters of Philemon the ratio of the medial to the initial accent is 12:19; the medial accent occurs m frgtuni, 3«» 5; 44. 4; <5o, 2; 79, II ; 88, 9; 89, i ; 90, i ; 90, 11 ; 98, 1; 131; 207; 240. In 450 trimeters of Aristophanes's Eqtmits (extending through v. 1025) the ratio is 5:8; the niedtal accent occurs in Eq. 124; 140; 202; 482; 938. In 150 iambic tetrameters of Aristophanes the ratio is 4:4; the medial accent occurs Eq. iiZ\ 433; 859; 899. In 350 tro- chaic tctraraetcrs and dimeters of Aristophanes {Nub., Eq,, Mm,, Av.) tie ratio is 5:5* the medial accent occurs Eq. 280; Jf/.28o; 388; 396; 790. In the 1 500 verses examined the medial accent occurs 26 times, the initial accent 36 tinies. In niiaking tic count cases M tic secondary accent, such as Piil/r^n. 100, 2, rk eXaXiyo-ev, were estimated according to tie Latin accent-law. A distinction was also made between real and apparent trisyllabic groups, e.^, in Phil. /r^m. 7, i jkp iwl ri^^ '^1^% tic prepositioa M is, of course, proclitic, and the irst part of tie resulting group, yap cttI-to pPj/ia, is not trisyllabic but quadrisyllabic. We may conclude, I think, from these statistics that in respect to the placing of the metrical accem lipott^ tic fest; « tic^. second syllable of such * Hii rmtio will hold good approximately for all the trimeters of the Middle and New Comedy according to the references given hy Perschinka, D^ med. ei nm, mm, irim, iamk, Dimri, Phikl Vindok III, 330 ff. I ) i » 'I f 4 I 70 Robert S. Radford, [1903 phrases as \ia ^ia, irpo^ e/ie, tcl kuko, the Greek is absolutely indifferent, as it is upon the whole indifferent in the accentua- tion of the tribrach word.* III. Summary of Tekentian Usage. The same test yields very different results when applied to the Latin poets. The accentuation of all the tribrach groups occurring in Terence is given in detail in a later section of this paper (p. 78 ff.), but since I wish, before proceeding farther, to discuss the origin and the effects of this accentua- tion, I shall here briefly summarize the results obtained in the fuller discussion. There are in Terence 176 certain or highly probable cases of tribrach (trisyllabic) groups of the form . sed homines, although the initial accent, sid homines, appears also to have been in current use. Il ii Cipif '■ 'tlllijiigli the tendency of the monosyllables to coalesce lis pronunciation with the following word that trisyl- laMc groups have been formed io Latin and have received re-accentuation in accordance with the three-syllable or the carier initial accent law. As has already been indicated, tic pause that divides a word of one or two morae from the following word may be shown by numerous metrical tests to be extremely brief in a quantitative language like the Latin ; |«t,.ffen with tic iiaiinisilng pause, a recession of the accent catiiiot cas% arise without a fixed or usual order of words. Hence, i4 as appears to be the case, a general recession of the accent has taken place in these groups, this must be due to the analogy of the many phrases which have acquired a Med outer. We may suppose that the analogy of the numerous verbal compounds with monosyllabic prepositions, such as m^&, imeo.pereo, snbeo, has first exerted its influence and led to the recession of the accent in all the trisyllabic locutions of ixcd form or frequent occurrence. The latter, the §xed locutions, are indeed very numerous; thus with enim alone we have ei enim, at em'm, sed enim, ncquc enim, quidetmn, quod enim, quia emiwt, ego emim, etc., all of which are virtiial €0$t^siim^ .and m very frequent use. Again, in consequence of the rtllcs of word-position by which the per- sonal and demonstrative pronouns are attached in the sen- tence directly t*i pepositive conjunctioiii, to interrogative words and affirraatif e parlfeN, a Wiltltude of fixed locutions arise containing ego and the various forms of is ; i.e. et ego (Seyffert, S/nd Plant., p. 12), nam ego (ibid., p. 20), /ol ego (ICellerhoff, Stndem. Stud.. H, p 60 ff.), quid ego (Kampf, Pronom. Personal., p. 31 1%, fmd ^0 {ibid., p. jj), at eg&, sed ego, dum ego, mM €g9, ele. ; M ea, et ea, sed enm, at emn adeitm, in eum, ab eo, quid €0, etc. ; with homo also we have the frequent phrases hie horn^, is homo, quis homo, qui homo, etc. In short, of the 180 cases of tlie tribrach group found i in Terence, nearly two-thirds occur apparently in fixed locu- tions of this character. Finally, the analogy of the fixed locutions is followed by the merely fugitive and infrequent combinations, %& tie analogy of quid agit and idagit is fol- lowed by sed agit, and the accentual group in its developed form does not apparently require the closest possible connec- tion in sense, provided always that the single words involved belong to the same clause. Thus we have not only the fre- quent phrases quid ais and /// ais, but also Phonn. 380 quem amicum tuom ais fuisse, and not only the frequent phrase ubi erit, but also Hcc. 474 mco crit ingenio. Lindsay {Lat, Lang., p. 167 ff., and Journal of Philology, XX, 150 ff.) and Skutsch {Forsch., p. 157 ff.), in treating the sentence-accentuation in Plautus and Terence, have already discussed at length the proclisis of the Latin prepositions, but they have overlooked the similar proclisis ^ of the mono- syllabic conjunctions and adverbs. Yet the proclitic charac- ter of the Latin monosyllables, as a class, has always been recognized in a general way by Latin metricians, and has often been invoked in the explanation of special rules of Latin prosody; see L, Miiller, Res Mctr?, pp. 164-170, 460-467 ', Ritschl, Proleg., pp. ccxiii, ccxxxiii, etc.; Podi- aski, Quomodo Tcrentius in tetramctris iambicis et trochaicis, etc., pp. 7, 10, etc.; O. Brugman, Queviadmodum in ianibico senario,Q,\.c., p. 18; Kohler, De verb. ace. in troch. sept. Plant., p. 29; cf. Weil et Benloew, U Accentuation lat., p. 56, and Corssen, Ausspr., IP, 862 ff. The Roman grammarians also recognize no distinction in character between the prepositions and other monosyllabic words, as the testimony of Audax, quoted above (p. 64), clearly shows. Finally, the Latin iambic poets of the classical period, when admitting a short monosyllable to form part of the resolved arsis,^ treat the ^ I do not forget in making this statement that many German scholars employ •enclisis' as a general term for word-grouping (Tonanschluss), and avoid entirely the use of the term * proclisis,' which, as is well known, is not of ancient origin, but a coinage of G. Hermann's. On my own account I have not hesitated to employ 'proclisis,' after observing the general use of the term among Romance scholars. 2 By * arsis ' is meant the strong, or accented part of the foot. Vol. juixiv.] Lmiim MGnosyllables, 'Z ^repoeitbns aad coEJunctioDs precisely alike ; i.e. they use im-Hmore and H-^more, with equal frequency, as quasi-quad- risyllabic words; for examples, see B. Schmidt, De Senccac ^^£J' ruiimikts pmsGdiads ei meirkis, p. 46 f . In saflof tfcat the Latifi, monosyllables are regularly pro- clitic, I do not mean to deny that there are many single phrases in which monosyllabic words have acquired an en- clitic Mse through the observance of some fixed order. Such f Irases are mescm quis {mfsehquisX nisi si {nisisi\ simul ac {simulm\ etiam nunc (etiammtm), and the like, a fuller enu- meration of which may be found in Corsscn. Ausstr., IP, 835- 861. Especially frequent in these phrases is the quantitative type \j V, o, which, by an extension of usage, seems some- times to be pronoynced as a single word in the caesurae of the chief Latin verses, on the basis of the form alone. Impor- tant as these enclitic phrases are, they constitute a very small part of the total use of m.onosyIlabic words. The parts of the siibstantiire verb are also regularly enclitic; compare the frequent writing mnatasi, amminmsi m our Mss. Hence, it is necessary to distinguish carefully between the groups qltid if as alone and quid (fusses/. The first is trisyllabic (qiad ^/»jf),; tie second is qtiadrisyllaWc,, and has a variable accent i^/iid iffiiis-fsi or qmid opus est), just as we have found the accent to vary in quid agiiur and sed kmnines. The case is similar with ei ega m ai ifa alone, and the fuller combinations According to the view here adopted, trisyllabic groups, such as sed em'm, sed ej^e, sed kmmo, etc., have originally arisen through the proclisis of the monosyllable, \mt with the 'rtiwlt till a.. 'dissf liable witli short: penult has practicallv become cisciitic whenever a monosyllable precedes. This explanation may appear at first to be somewhat at variance with the results reached by Wsiekeriiagel in. lis extremely valuable and comprehensive article, •*Peber e!n Gesetz der Indogermanischcn Wortstellung," Indogcrman. Forsch., I 'i^QiX P' 333 ff. la the part of this article devoted to the Latin word-order (p. 406 ff.^), Wackernagel has shown, with * Sec, also, the suinmary in Stok. Hist&r. Cramm, d. lat Spr., I, p. 105. 74 Robert S, Radford. [1903 great completeness, that the Latin personal and demonstra- tive pronouns manifest a strong tendency to occupy the second place in the sentence, — a place which is not only commonly occupied, in Latin, by such unemphatic words as enim, quidcm, etc., but which is associated, in most of the Indo-European languages, with accentual weakness, or encli- sis. It follows that the personal and demonstrative pronouns are enclitic whenever they follow the prepositive conjunctions, as well as in some other cases. While it did not fall within the scope of his article to treat the verse-accent of the early Latin poets, Wackernagel has clearly demonstrated, in sub- stance, that at some period of the Latin language the initial accent existed in many of the groups which we are now dis- cussing, viz. sed ego, si d cum, quid co, etc., as well as sed tu, Sihimihi, and the like. I gladly recognize the great value of Wackernagel's independent proof of the fact, but in respect to the process, although he has found it convenient to employ the term 'enclisis' throughout, I cannot see that his results are necessarily opposed to the account which I have just given. In the first place, In a very large number of the trisyllabic groups whicli we are now discussing, the weakly accented word does not, as a rule, occupy the second position in the sentence, i.e. dd cum, dd cnnu, tibi ego, sdt crat, turn agam, hie homo, etc., so that no theory of enclisis will apply to these cases. Secondly, the personal and demonstrative pronouns, being usually employed without emphasis, are more or less weakly accented in all parts of the sentence; hence they naturally gravitate to the second, or enclitic posi- tion, which is, however, rather to be considered a proof than a cause of their weakness. In view of these considerations it seems probable that the initial accent has arisen in the man- ner already described, although it is not to be denied that in some cases word-grouping through enclisis may also have op- erated, and the two processes may have gone on side by side. In any case the results are clear enough. The mono- syllables which are, as a rule, atonic, often acquire the accent of the word-groups into which they enter, as in sid- homo, sdd-homines, and with the accent they acquire the Vol. xxxiv.] Latin Monosyllables, 75 power of shortening a following unaccented syllable, as in sid-haic^ sid-drgintL Groups containing the pronouns ille, iste\ ipm^ and iHier weakly ai^ceiileii. woini% such as unde^ omnis^ esse, geetiMf kerck, ergo, etc., often adopt a similar accentuation; tins, on tie analogy of s^d ego, sid ea, sid enm, we have sed iius, sid ilia, sid istum;^ on the analogy of sM age, sid emim, we have sid irg&, etc. Through false analogy the popular pronunciation foretimes gave to the monosyllable an accent incorrectly and, consequently, dis- regarded a genuine grammatical accent, as we may see from accentaatiiins like sid MxWtfir Mrtum, which are occasion- ally admitted by Plantus. Cases of this last kind are very rare In Terence; Spengel cites only iho dbsecro {Andr, 781), mM Miewst (Emm, 233); cf, sid fnterim {£un. 607; I/eaut. "1 3 882(1)); tiiam inveniri (Andr. 939). s Tic following table will serve to show at a glance the manner in which the syllable-shortening arises: Words. Word-groups. SHORTENrNG BY Analogy. Dissyllabic irus dgit sid hie tUi amicus iri sid haic tlbi Urgenti Trisyllabic ginere sid ea qudd enim qui erit quid habuisset quid opus sid Ilia qudd iius ita ^sse titam inveniri lit ix-me Quadrisyllabic gineribus quid opus-est sid etiam qui homines sid dbsecro * sid interim quid interest • AH sticfi cases as sedMseer&, in wWch a trisyllabic word of this kind is short- ened after a manosyUabl% iw« dft la *' false analogy." 1 The examples here used are taken from Spengel, Einleit. Andria, p. xxviii ff. * Compare Ahlberg, Procel.^ p. 49 : Haec vocabula cum vocabulo praecedenti proxime coniuncta quasi enclitica fiunt : et ipsus^ ut ipsus, sed eccum. This is a statement of the result, not of tlie process : a special form of accentuation, so 76 Robert S. Radford. [1903 4 The question may well be asked at this point, What is the accentuation of trisyllabic groups formed by long monosylla- bles, i.e, groups of the form _. v. ^ ? Such groups are not in all respects similar to tribrach groups, for the reason that a somewhat longer pause (' latcns tcmpus; Quintil. IX, 4, 98) falls after a long monosyllable than after a short; hence the long monosyllable possesses a greater independence; compare also the accent ghcribus with cxitibcas. I propose to discuss the accent of these groups in full elsewhere; it will be suf- ficient to point out here that the initial accent is the usual, though not the sole, form of pronunciation for the republican period. Thus Plautus and Terence have only hoc aoets of the republic. In Ahlberg's collection we find forty-two examples of the type vides db ea (inclusive of nine examples which involve syllable-shortening, as in rogat ut iIiMm% hut of tie type sed ab ia we find only a single case, le. CisL S94 eg0 addnitm^ which maybe excused by the license of the first foot, cf. p. 65, above.^ Hence proceleusmatici of the types sed ab /a, id ui irus, et h&mo ubi, fieqne ego dmo are evi- dently avoided by the iambic poets, although in these assumed types the thesis is formed in a thoroughly legitimate manner (see Ahlberg, /./., I, loff.), and although the dissyllabic arsis also is one of the most usual forms, see ibid.y p. 23. Hence, since neither the assumed formation of the thesis is avoided taken separately nor that of the arsis taken separately, it is clear that the avoidance of these types is due to the effect produced by the two formations when occurring together, and this effect is none other than the false accent sed ab //i, meqm eg^ dm^. The twelve examples of the type vidrs db ca cited by Ahlberg from Terence are as follows — from the iambic trimeter (/./., p. i35^-)* ^^^^^'- 737 ego quid agas; Heatit 872 ego d6mi ero; Fkorm, 98 ea sita erat(see Ahlberg, p. 156); Emm, 509 video ab ea; Pkorm, 48 alio, ubi crit; — froni the iambic sept and oct. (p. 14O' L): Phonn. 491 capiti. |1 Idem ego; Eun, 309 modo quod ames; Ad. 192 si ego tfbi illara; — from the trochaic sept. (p. 149 f): Plionn. 346 vide q«ld agas ; HemL 966 tibi qui erat ; Eun. 224 vide quid agas ; — from the tf ochalc ocfc (p. Ifi): Eun, 618 rogat ot is such a proccl. as Fleck, reads with the Calliopian Mss. in Heaut. 931: ti' iS iritt where Umpf. and Pi, read with A /«/, et id. 78 Robert S. Radford, [1903 Mil. 994; Rnd 731 ; True. 879; Aul. 734; Epid. 641 ; Men, 162; Most. 305; 833; Pers. 832; Capt. 461; Cure. 160; 170; Trin. 715; Amph. 748; Pacuv. trag. frgm. 99; Pseud, ii'^i. The Latin accentuation of sed agit as a single word is also made probable by some of the general rules of Latin prosody. Two of these may be mentioned here : (i) the absence of a full or genuine word-end within the trisyllabic group is shown by its admission in all the uneven feet (see p. 92, below) to form the iambic anapaest, i.e. sed agunt, since the thesis of this shortened, this exceptionally swift anapaest does not in general admit division by a word-end ; see Ritschl, Proleg., p. ccxxxvii; Klotz, Grunda., p. 307.1 (2) The general law of Latin prosody against the placing of monosyllables before the pauses is evidently based upon their proclitic character; for it is a rule of the Graeco-Roman poetry that a full word-end must fall at the end of a metrical period (Rossbach-West- phal, Metr. IP, p. 106), and such a word-end scarcely falls in Latin within a complex like si-bona. V. Terentian Usage in Detail. It seems desirable to explain clearly the system of meas- urements upon which the following statistics for the accentua- tion of tribrach groups in Terence are based. In cases which involve a primary accent, such as id agis, the measurement of the tribrach requires no explanation, but in cases involving a secondary accent, it must be borne in mind that the secondary group should be measured from the place of the primary grammatical accent. Thus in E?in. 931 : modo ddiVeskeiitn^ Ins, the rule is not violated because the first part of the phrase is quadrisyllabic when estimated from the place of the pri- mary accent in adulescentnlHS ; on the other hand hSmo Mu- lesceus {Pkorm. 1041) and p61 ego amator (cf. Eun. 936 cum 1 The metrical law of Lachmann and Ritschl which forbids the divided thesis of the anapaest, etc.. is subjected to thoroughgoing criticism by Maurenbrecher, Hiatus tnui Verschleifung im alt. Lai., Leipzig, 1899, P- 26ff; Maurenbrecher's careful analysis (/./., p. 31) proves only that the law is not one of absolute validity, which we have little right to expect it to be in the case of poets so little bound by fixed conventions as the Roman dramatists. Vol. xxxiv.] Latin Monosyllables, 79 amaiSre)w[Q the only admissible accents in genuine trisyllabic groiips. In tie cases, also, where the last member of a tri- syllabic phrase coalesces through elision with some following word, as m §Mid ig{&) /miZ/r (Plant. 3Iost. 531), there is clearly no violation of the rule ; for this phrase may be considered m^ imdt lip" nut of fiiii# aad eg^ alone {qmU-ego), but of ijnid and igo-kodk; d.Andr, 684 uM Mi erii, though we find only Mi Mi alone. Similarly the complex character of the follow- ing prepositional phrases with apud is clear : Andr. 254 niihi ipud-fdrum; Meamt 377 seio, ipud-patrem. Finally not every chance ocewrfcace of the forms w, w m and m. w» iz, constitutes a trisyllabic sentence-group, but the words in- volved must at least belong to the same clause and be coniiccted m the sentence-enunciation. Where the words belong to diilereal clauses, set off by punctuation, the rule does not apply. Hence I omit from these statistics examples both (1) of the form w; 6 v^ and (2) of the form ^ ; xj XI., viz., ( I ) Eun. 630 fit. ubi ; Hcant. 1 54 fit, libi ; 628* ^ ego>^rus;, F/mrm, 1029 sic dabo : age; Hec. 610 pol. Ijabi; 637 fit lea; Jnf. 646 Quid?|K*go; cf. 943 haecquidem. II %e; 946* quid? II ego (Dziatzko); 982 da modo. H age.— (2) Amir, 713 siqufd. | age ; NrmtL 974 ego, id obesse ; Enn. 252 nigo; ait?; cf. 3^1 sine. | at eniro istaec. To prac- tically the same head should be referred two apparent excep- tions in which the monosyllable is closely connected with the preceding word through elision, and at the same time is separated ffom the following dissyllable by the principal caesura, r,^, : ^jiii, 512: UM v^ni, caus(am), ut | Ibi mandrem, r^pperit. Mid. 394 Tiriimpliat | h6c piovl || Ut; 1015 sed ea; — Andr. 420 tibi erit; 684 ubi erit; Phorm. 889 datum erit; Hec. 474 meo erit; — qufd ita J#i^/. 371; Enn. 366; 725; 861; 897; 959; 1008; MeaMt 6fO; F/ierm. 568; — lieaut. 874 ncque ita ; 941 sed ita; Ad, 161 in ita; 483 nisi ita; — Fhorm. 47 ubi era; 634 lit erus; cf. 471 et quidem ere, which may, however, be taken as etqufdeni: ere. Umpf. and Dz, read tibi ire Andr. 508, but the text is corrupt and the order much confused in the Mss. ; Fairclough now reads tibi ere,. Fleckeisen tibi reniintio, ire. An exception might seem to be offered also by I Ice. 'jc)g : Edepol Ti^ meam ems esse 6perara dtJputdt parvi preti, but we need not hesitate to accept here the order which is found in D (and also in F): ii erits, i.e. isse ems. [Note also qu6 erus Pompon, com. frgm. 45 R. According to a collection ^of' f xiiii^ples whieli, in part » based upon the Le- maire Index and is, therefore, only approjcimate.ly complete, Plautus has the type sed ems {err, era) 31 times. The four passages in which either sed ems m sed enis (with length by position) is possible,,. 9m tO' be read, according to the latter scansion, siilce ccrtftio cases of eris' are not rare in Plautus. The accent nisi ims in a single apparently correct text {Poen. 839) may be explained as due to the license of the first foot 1 f of a colon (fifth foot of the septenarius).] — quid opus, Andr. 490; Phorm. •j^Zl-^P/wrm. 440 siquid opus, i.e. si quid opus; 654 sed mfhi opus (probable scansion); 681 tibi opus; 716 Ita opus; Ad. 996 quod opus. The ambiguous passage Eun. 223 SI sit opus, vel totum is to be read opfis, as is also Ad. 617 id anus mihi ; for certain cases of ophs in Terence, see Meant. 80; Ad. 254, etc. [Note also Afran. coin. frgm. 145 R. fit opus; Titin. com. frgm. 4 R. id opus; cf. Pompon. com. frgm. 66 R. age anus. The usage of Plautus is similar in respect both to opus and to anns ; a single exception, in opus Fidnl 75, is due to the license of the first foot.] For apparent exceptions occurring in (jnid opus-cst and similar phrases with est, see p. ^j, below. Hec. 538 ut ego opinor; for the numerous examples of this phrase in Plautus, where it is always similarly accented, see Kiimpf, Pronom. Personal, p. 4; cf. Andr. 179 neque ut opinor; /://;/. m quom ibi adessent ; 242 quae habitudost ; 522 quid habuisset; 588 fa alienas ; 606 pol ego is essem ; 764 volo ego adesse; 926 quod ei amorem ; Meant. 191 ad eam in urbem ; 592 tibi opis ; 6%7 kt id omitto ; 836 pro ali- mentis ; Phorm. 94 mihi onus ; cf. 175 ego in eum incidi ; 332 quia enim in illis; 412 ^go adipiscar; 509 quod homo inhu- manissumus; 531 s^d utut ; 545 Geta alienus; 553 siqufd opis; 1041 homo adulesccns ; 1046 quod is iubebit; zi. Ad. 232 tum agam ubi ^ illinc. ( Total 73.) (/S) Cases in which the quantities of the final syllables are not perhaps altogether certain in single cases, but are suf- ficiently certain collectively. It is generally agreed by Ter- entian critics that many final syllables which were * half-long ' or prevailingly long in Plautus are to be considered as defi- nitely short in Terence ; thus verbal forms in -at and -ct which were originally iambic, such as erat, amat, amct, habct, are pyrrhics in Terence (see Bomer, De cornptione vocab. iamb. Terentiana, p. 12); homo retains a long final only in arsis (Bomer, /./., p. 18; Fabia, ed. Ad., p. 55); ego, ibi, ubi have a long final only in arsis, and even then in very rare cases 1 A dissyllable when elided is commonly treated as a monosyllable in Latin. Vol. xxxiv.] Latin Monosyllables, 83 84 Robert S. Radford, [1903 (Klotz, Grunds,.^!^. 5 if-; Bomer, /./., p. 34, p. 63; Hauler, Ank jr. Pk&nm.^ ¥. 176). On the other hand verbal forms of the first and secund persons, sncli as ago, em^ ems, habes, abis, are probably lo be considered as having more often a long final in Terence; hence cases like Andr. 614 id ago, 714 domi ero,* will not be included in this collection. On the principles littt Stated'"' 'ire '\mm tie following examples of groups in vol v- ing the so<:aHed 'half-longs': (i) erat, a,„at, amct, habct ; Eun. 736 sat erat (or sdterat, like poterat ; see Leo, Plant. Forsck,^ p. 2(^\ L. M idler, Res. MetK, p. 466 f., and compare the word-divisioQ in the Piaiitus Mss.) ; Meant. Gig sed erat ; 966^ qif e«it; Pkmm:,. 97 sita erat; 768 .sat erat (saterat); Ad. 494 mfhierat; Eun. 986 quid? amat?; Ad. 341 quom amet; Andr, 954 quia habet; Meant. 835 decem habet; P/mrm, ID41 si habet; Ad. 382 sibi habet; — (2) Aomo: mi ioiiiO' Amdr. f2i; Pkorm. lOOj; Ad. t%6\ — Andr. j-jZ p6l homo ; Eun. 960 quis homo ; — (3) tf/i/(f ), ibi, nbi: cf. Phorm. 198 modo a pud portum ; Andr. 343 scd ubi ; 928 is ubi ; Enn. 719 pirem ubi; Meant. 983 ct ibi; cf. Phorm. 827 sed ubi MH» ; cl Ad. sa? iii4 wM ; 5,70 scfo uM. On the other hand ill two ambiguous passages we should scan ///;/ and ibi: Eun. 414 is iiW molestus; Ad. 584 quid ibi facit; the latter is rightlj s© scanned by Spengel, for the original quantity is preferable on other grounds in the pure seventh foot of the septenar i us. { Total 24. ) (4) Tie following phrases occur with ego in Terence : an ego^ di'^git^ mfm4d.^9 (i,c, #r fuM ego%. it ego, (et qnidem ego\ M ip?! Udem ego, ndm ego, nique ego, nisi ego, qndsi ego, quid ego, qnSd ego, qudm ego, quim ego, quom ego, sM ego, tibi ego, imm ego, vir ego, ubi ego, ut ego: — Andr. 252 ; 508 (id ego, rightly scaanei-" as trociaic' iwrsc hj Fleck., and Faircl.) ; 5 19 ; 563; 612; S|0; 886; 944; Enn. 142; 265; 293; 496; 822; 930; 958; 1081; 1086; Meami. 191 ; 252 ; 529 (quid ego ni); 563; 631 ; 663; 686; 956; 993; lO^ilP/iorm. 491 ; 519; 587; 685 ; 844 ; 1000 ; 1031 ; 1052 ; Hec. 98 ; ( 195 : el qufdem ego, or etqufdepi ego); 408; 524; 564; 850 (rightly scanned as tro- ^ This is tlie nccentuatioii of Spengel and Fleck., and is approved by C. F. W. Muller, Ptami. Prm^ pp. I5J» lit ; olief editors accent wrongly domi ero. chaic by Umpf. and Dz.); Ad. 128 ; 256 ; 378 ; 568 ; 749* ; 784 ; S77 {ecqnld ego); 916 ; (946, Umpf.) ; 972 ( Total 50). In the example ecqnid ego cited from Ad. ^-jj we have what appears at first the inadmissible accentuation of a trochaic word upon the ultima, cf. Podiaski, /./., p. 62 ; but this accentuation is only apparent; for all the compounds formed of two mono- syllables admit in early Latin of being resolved into their original parts (Ritschl, Prohg., p. ccxxii f . ; Klotz, Grundz., p. 324). Hence we may write here if we wish ec quid ego, just as we often find this division in the Plautus Mss. {e.g. in B ec quid Cas. 242, et quid Aviph. 577, etc.), and as Leo writes in Sen. Oed. 263 quid quid ego fugi ; see also the lat- ter's remarks. Plant. Forsch., p. 236. The extent to which the expressions containing ego have acquired a fixed order and become phraseological may be seen from the fact that the full form qnid ego without elision occurs in Terence 13 times, quod ego 7 times, at ego and scd ego 4 times each, // ego 3 times, etc. That qnidcgo and quidille possess most of the characteristics of actual compounds may be seen further from the fact that they take precedence over the compound qnidni or qninni^\ for in connection with ego and ilL qnidni {qninni) suffers tmesis and the forms qnid ego ni {Meant. 529), quid illam ni{Ad. 662) result; for additional examples, see Plessis on Ad. 662 and Brix-Niemeyer on J/il 11 20. Hence so far as^ the actual usage of the language is concerned, Priscian (Keil, in, 24, 23 f.)appears to be mistaken when he says that the conjunctions enter into composition with no words which are declined except with the indefinite pronouns, i.e. in siqnis, neqnis, nnmqnis. We may be sure that if the Latin gram- marians had had occasion to develop fully this topic and to discuss in an independent manner the compounds capable of being formed with the help of conjunctions, they would have shown differences of opinion at this point, just as they differ widely in the lists which they give of the compound conjunc- tions and of the prepositions which may serve to form com- pound verbs (on the latter contrast, for example, Priscian, Keil, 1 For the latter form, see Cledonius, Keil, V, 66, i6 ; Corp. Gloss. Lai. IV, 158, 19, etc. Vol. xjciiv.] Latin Monosyllables. 8S m Robert S. Radford. [1903 III, 56> 9f. with Donatus, IV, 366, 10 f., who excluding only ^i^d aid pms leaves ample xtmm for such compounds as c&nirqfaeem^ pmptervolmre} fropteresse^ etc.). We can cer- tainly recognize »o difference in closeness of connection between si-qmis and si^ego, between ec-quis and et-ille, and if the ■qttcstion be decided on the basis of the accent to which Priscian so constantly appeals ia. the determination of compo- siia and on the basis of such usages as quid ego m\ we should be compelled to recognize the phrases which are made up of €©|lj«nctions '^md of the personal or demonstrative pronouns as ^omposita in the sen.se of improper compounds.^ On the other hand, if we adopt the traditional orthography, i.e. the usual word-division, as our standard, we shall recognize more Ittstiication for Priscian*s statement; for sic/nis, eeqids, quis- qmiSf etc., are written very frequently together in Mss, and inscrii)tions, while we find, e.g, in the Plautus Mss., the orthography /i?/i;^£i, quiaego, sedego^ siego, etc., less frequently. That sM ego is the only accent known to Terence is also strikingly shown by the small number of cases in which the scansion appears doubtful. For Terentian critics are agreed that the scansion ego is quite rare in Terence, and conse- qpently we are not at liberty to assume this scansion freely. la piiiat of fact, wMk s/d egO' h certain in 50 passages, we shall need to assume sed ego in only 3 passages, viz. HeauL 309 ita timuL || at egwpiiiini conjunctions et enim ^ (etenim) and sed enim {stdmim)hme the initial accent : Keil, III, 91, II 1 (aaScholl, Be me., p. 192) Composita vero dtgue /ieuim, sidenim. Haec enim ex accentu composita esse nos- €iiittir. Sitnilariy smiemm is iwined as a compositum also in a gloss on Max. Vict,, Kdl, VI, 203, \2\si enim, i.e. probably sicnim, is named as a emnpositum by Donatus, Keil, IV, 365, 2; ibid,, p. 3% 5, and if Cledonius, Keil, V, 24, 25 ; com- pute, aiSD^,, lie tfealinciit dE qmdistic (in early Latin often quiMsiic) as a eompositnm by Priscian, Keil, III, 8$, 33 J Corp. Gloss, LmL V,6«, 55 ' qiddistic sub uno accentu est profecto iicl omnino/ There can be little doubt that the numerous iltlef ■cnwpound.s of emm^ iave a similar accent, ix. at enim, neqme mim, quid enim, quod enim, M^ emim, etc. Among tie glossarists Festus treats ncccnnt{—non ennt) and necerim {ssmee et$m) as single words (p. 162, 11 Miill. ; Paul iw. Fest,., p. 162, 21 MulLX int tie most complete evi- dence may be drawn from Gstz and Gnnderniann's Corpus Ghssarioram Lat, The conjunctions here glossed as com- pound words are sede[c]eum, II, 181, 19 (where the gloss einhe apparently indicates tiat tic force of the simple conjunction is. lost as completely .as in, tie Italian ebbene {^eiime% i very good '); sedenim^ IV, 565, 52 ; ntcnim, IV, 470» 45 * Q^^i^- enim, IV, 461, i ; quidita, II, lij, 14; IV, 158, 33» etc.; cf. qmidistie:, V, §22^^ $$; quid igitmr, IV, 421, 16. In addition we ind glossf s ii'pon tie pronouns and verbs atego and astego, 1 £i enim is nearly always written in oiif texts of Latin authors as one word; it is scarcely necessary to say that no such uniformity exists in the Mss., which no more write et enim than they write ob viam and in vicem invariably as one word. On account, however, of being much the most frequent of all the compounds of enim, etenim is usually written as one word, and this orthography is much more frequent than sedenintt quidenim, etc., which also occQr. 98 Robert S. Radford, [1903 II, 284, 34 (where, however, the Greek stands first); et ego, III, 342, 22 (where also the Greek stands first); ide{^g^^o { = iyo) auT6[v']\ II, 80, 40 ; ninnqnid ego {i.e. num quidego as a rule), = egone, IV, 369, 35; idagis, idagit, the latter = ii^epyel, II, y6, 12, 13 ; qiiianiat, II, 167, 5. The grammarians also name qnidita among the compound adverbs {quidita, Audax, Keil, VII, 348, 14 ; quid ita, Dositheus, VII, 410, 24), and an abbreviation for the frequent phrase ego ejtiin is found in the Commentarii Not. Tironian. It is instructive to com- pare also the traditional orthography in our texts of etenim, poterOy potcram, satago, retr{p)ago, rctr{o)eo, velitti (cf. velut, mmnt\ adeo, ideo, postidea^ ant{c)idea, ub{i)ubi, nccuter, niuter (according to some, cf. Lindsay, Lat. Lang., p. 39), necopinantem, etiani, quoniani, etc., although it should be remembered that the separate writing et enim, sat ago, vel uti, nbi iibi, nee opifiantem is also frequent in the Mss., and even the division ad eo, id eo is sometimes found. B. Word-division in Inscriptions and Mss. — Finally, the word-division in the inscriptions and Mss. has an important bearing upon the question of the accent ; for, from the time of Quintilian (I, 5, 25) the Roman grammarians often expressly connect the question of punctuation and word-division with the determination of the accent, often in the formula * ratio distinguendi (dividendi) regulas accentuum corrumpit ' ; see the numerous passages of the grammarians on this subject, which are collected by Scholl, /./., p. 127 f. They also fre- quently recommend an unusual word-division, i.e, haiiscio instead of //«//(<:/) .fr/^(Pseudo-Phocas, Keil, V, 441, i), on the ground that it corresponds more fully to the accentuation. In the case of the Latin prepositions, both monosyllabic and dissyllabic, the omission of punctuation in writing is well known, but the similar usage by which the monosyllabic con- junctions, adverbs, and pronouns are joined in writing with the following word has received less attention, although it is by no means rare. Marius Victorinus, Keil, VI, 23, 7 ff., pre- scribes the omission of punctuation in nechoc and neeillud ]\xsX as in ingalliam and initaliam : sed ne ea quidem, quae cum praepositione dicuntur, circumpungetis, ut circumduci et cir- Vill. xxxiv.] Latin Monosyllables. 99 cwnveniri, et nonnulli et paulopost, nee haec, ut ingalliam, initaliam, ncclioc, nccilliidi quae iofinite dicuntur. In accord- ance witli this orthograpliy the mooosyllables are at times written together with the following word in inscriptions of the best period, as may be seen from the examples collected by Corssen, who cites Ausspr. 1 1 2, p. ^^ etcoronis, itauti, ,|llii^ll«: