I I RES METRICA BY W. R. HARDIE OXFORD UNIVERSITY PRESS LONDON EDINBURGH GLASGOW NEW YORK TORONTO MELBOURNE CAPE TOWN BOMBAY HUMPHREY MILFORD PUBLISHER TO THE UNIVERSITY 1887 PREFACE /^^,^ I AM reluctant to trouble the reader with what is more or less personal, but it seems necessary to explain how this book came to exist. In the spring of 1912 I was asked to review Schroder's Horazens Versmasse, fiir Anfdnger erkldrt, and I did not find that it solved the problem ; it was not a thing which, if translated, would be helpful to 'beginners' in this country. In the summer it occurred to me to try the experiment of writing an account of Horace's lyric metres myself, and I wrote it, merely as an experiment and without the slightest inten- tion of publishing it. Shortly after this I learned from Professor Lindsay that he was interested in getting something done towards making metrical knowledge more accessible to students ; what he had in mind was finding somebody to translate Bickel's section on Metre in Gercke and Norden's Einleittmg in die AltertJmms- ivissenschaft. But Bickel's treatment, he agreed, was not satisfactory for our purpose ; like Schroder's it was too technical and too condensed for the ordinary reader. I showed him my experimental account of Horace's metres, and chiefly through his instigation and en- couragement I wrote in the following summer accounts of some of the more familiar metres such as the Hexameter and Iambic Trimeter. These I now thought of as chapters in a possible, though still hardly probable. 417894 vi PREFACE I book. The experiment which I made with most re- luctance was that of writing a chapter on Greek lyric verse but this also was done, in September 1913. The conditions of teaching in this country are not like those of Germany, and few teaching posts allow much tmie for writing or research ; so it was only at rather long intervals that I could take this metrical business in hand. ^^ I say this not in the least as regretting or resentmg the i conditions, and not at all as an apology for the deficiencies \ of the book. I am not sure that with more time to give to it I should have made it any better ; it might even have been, for practical purposes and for the ordinary reader, worse. Such as it is, it owes its existence largely to Professor Lindsay's encouragement. For its defects he is not responsible ; he removed some of them, and he has contributed valuable advice and suggestions. He is in no way responsible for what is said about Saturnian verse in the chapter on ' Metre at Rome '. It was at one time a part of the design that he should contribute a chapter on the metre of the Roman Drama. But it now seems best to refer the reader to his account of Plant me metre in the Introduction to the larger edition of the Captivi (Methuen and Co., 1900). The method of the book will be thought unscientific, but it seemed best to avoid any attempt at deductive, exhaustive, and systematic exposition. To begin with what is highly abstract and general-as with definitions of Rhythm and Metre— is to run a risk of being involved at once in what is disputable and speculative, or what can scarcely be clear and significant to the reader until he has already become familiar with numerous particular facts. To avoid what is controversial, to remain on what PREFACE vii is fairly firm ground, has been one of the main objects kept in view. Controversial matter has been segregated as far as possible by relegation to an Excursus. It is hoped that some general notions of rhythm and metre will gradually become clearer to the reader as he pro- ceeds, and a Glossary of metrical terms has been appended in order to enable him to ascertain whether an idea which he gathers from the text of the book was meant to underlie it or not. It has been a further aim to treat metre in a rather more historical way than has sometimes been done, and to keep it in closer contact with literary study and literary interests. But a strictly historical method would have been as difficult to carry out as a deductive and theoretical method. The Chronological Table is intended to supply the defects of the text in this matter of sequence in time or consecutiveness. A strictly deductive or a strictly historical method would have made each chapter more dependent on what receded it. It seemed desirable to make each chapter thing which would be fairly intelligible if read separately. jThis results in a certain amount of repetition, which would e a fault in a work of literary art. In a text-book tility may be allowed to prevail over artistic considera- ions. The main business of the writer of a text-book s to devise ways of putting things which will make them :lear, forms of statement which are not misleading or :omplicated or abstruse : the grouping of facts is also Dart of his business, the arranging of them so that they will support one another and their connexion will appear. [t is not his business to devise new theories. But neither :s it his business to accept theories, whatever their present ^iii PREFACE vogue, without criticism. Even the writer of a text-book must ake up some position, and he may perhaps usefully n a Pi^face try to state briefly what he has been do.ng when ^e has' Jot been occupied solely with what I have defined as his main business. 1" regard to the history of the Hexameter and especially of its caesura, the writer of this book disagrees w'h som'e current methods and is inclined to dispute, or . disable • as evidence, a good deal of recent stat.st.cs In regard to Iambic and Trochaic Verse evdence ,s offered-statistics, I think, too simple to be d.sputable- ?hat the peculiarly Roman line (with spondees >n wha for a Greek were the wrong places) did not d.e out gradually, but gained ground, from Ennius to Accms ^nd then was ejected by a reaction or new movement^ An outline of the history of metre at Rome has been ftU^pted (pt. n, chap, ii) which is intended to^^ow briefly what the Romans were trymg to do at d^eient periods and what were the difficulties they had to contend with. The tragedies of Seneca have been kept m v.ew as throwing light both on Horace's metrical ideas and on Roman metrical practice of the imperial age. As to Greek lyric verse, the writer is aware that he will be regarded by Mr. White and others as Kpo-- 6>. Kal iKK.aiXnvos, when he expresses doubts abou £ quadrisyllabic structure of Aeolic verse and about Choriambo-Ionic structure of the ' E-Pj"- J*'- however, is quite deliberate, and, though it does not re on complete knowledge of the evidence, it does no r on complete ignorance of it either. In an Excursus (afte,| chap, i of pt. 11) will be found an attempt to show tha I PREFACE ix three passages which have been quoted as important do not prove what they are supposed to prove (i.e. that the Choriambo- Ionic arrangement is as old as the fifth cen- tury B.C.). To revert to dactylic scansion and to suggest that ' anacrusis ', though unsupported by tradition, is a more useful idea than ' hypercatalexis ' is a proceeding ikely to evoke the objection, ' So you think that modern metricians may know more about Greek verse than the Greeks did'; and to reply 'That is exactly what I do think' will perhaps elicit still more severe condemnation, if the holders of the now prevalent views condescend to notice it at all. I have suggested a comparison with Ptolemaic and Copernican astronomy, but this of course may be disallowed. ' The Greeks ', it will be said, ' did ot construct the stars ; they did construct their own erse, and must have known what it was.' To this JI should reply * Do we accept their grammar, or their ritical treatment of texts?' They spoke their language and wrote the texts, but is it not the case that the arts of analysis and interpretation lagged far behind creation, land were developed surprisingly late? Do we believe uintilian when he says of the 'Historical Infinitive': \ sttipere gaudio Graecus'. simul enim a.ud{tuY coepit' ; or vhen he takes Virgil to have written * qui non risere arentes ' in the sense of ' who have not smiled upon or to a parent ' — a plural resumed by a singular /mnc in the ut next line ? Is it a happy description of the shorter gen. pi. in Latin {deum) to say that it is the ace. sing, used est "or the gen. pi.? Do we think it, I do not say true, but espossible, that Homer wrote /xera 8' ov g-(J)l jraTrjp Kie i^^'^a/cpua Xetpcdv (the father in question being by this time 1^^'Jead — killed in an earlier book)? And what would Bentley X PREFACE have said of a critic who gravely discussed rccv kv tol9 ocpdaXfxoh napOei/cDu as a phrase for the pupil of the eye {Koprj) — this not in a dithyramb, but in a plain prose treatise — as the writer De Sublimit ate does ? Examples like these could easily be multiplied. Why should the Greek treatment of metre be supposed to be infallible ? Besides, I am not at all sure that we know what the Greek treatment really was in the sixth or fifth centuries. Still less do we know what metre was before Sappho, or in Indo-European times. Further, one might point out that writers on English metre have again and again assumed that poets wrote something that is not best described as they themselves would have described it ; e.g. that Milton did not really write a line of five iambi in Paradise Lost. But I do not lay great stress on this last argument ; I should lay more stress on it, if I had more belief in the theories about English verse which these writers have propounded. There is also the case of French verse. How much agreement is there about that, although metricians would have the advantage of being able to interrogate contemporary poets who are writing it? Greek verse was much more complicated and varied : what is the probability that Hephaestion. would be able to give a true account of lyric forms which had for long ceased to be composed ? * For these reasons it does not seem to me at all illegiti- mate to inquire what is the best way of describing verse-forms rather than what is the way supported by tradition. S CONTENTS PART I THE SIMPLER AND MORE FAMILIAR FORMS OF VERSE CHAPTER PAGE I. THE HEROIC HEXAMETER: (i) Origin and General Structure . . i (ii) Subordinate Divisions .... 14 Excursus : {a) Hexameters in the Lyric and Tragic poets • . 26 {J}) Statistics relating to Virgil's verse . 27 IL THE COLLOCATION OF WORDS IN EPIC VERSE (ELISION, HIATUS, ETC.) ..* 30 III. THE ELEGIAC COUPLET . ^. . 49 IV. ANAPAESTS 56 V. IAMBIC VERSE : (i) The Iambic Trimeter and ' Senarius ' 68 Excursus : (a) ' Choriambic substitution ' . . 86 [U) The Roman disregard for ' Porson's Canon ' . . . . . . 87 xii CONTENTS CHAPTER PAGE (c) The nature and history of the old spondaic ' Senarius ' at Rome . 88 (d) The length of the first syllable in an impure or heavy iambus . . 91 (ii) The Iambic Tetrameter Catalectic . 92 VI. THE SCAZON OR CHOLIAMBUS . . 98 VII. THE TROCHAIC TETRAMETER AND 'SEPTENARIUS' 103 VIII. HENDECASYLLABICS OR PHALAECEI 114 PART II I. GREEK LYRIC VERSE : (i) Archilochus, Alcman, and Sappho . 118 Excursus (a) 'Aeolic' Verse .... 136 (ii) Simple or Homogeneous Metres {fiovo- etSfj) ....... 143 (iii) Pindar and the Dramatic Poets . . 165 Excursus (b) On the quadrisyllabic scansion of ' Dactylo-epitrite ' verse, on ' Iambic Syn- copation', and on 'Ictus' . . . 177 II. THE HISTORY OF METRE AT ROME 196 III. LYRIC METRES OF HORACE . . 223 APPENDIX (A) Glossary of some Metrical Terms . . 261 (B) Chronological Table 270 PART I CHAPTER I THE HEROIC HEXAMETER I. Origin and General Structure It is a singular fact about Greek literature that it begins with poems of great power and splendour, which exercised a vast influence in later times and in their greatest parts were hardly, if ever, surpassed by the work of later poets. But of course it is only for us that Greek literature begins in this way. There is nothing extant that we can certainly or con- vincingly describe as pre-Homeric. That much did exist is certain. It has long been recognized that the Homeric poems are not really very early or primitive, but the crown and consummation of a long period of poetic effort and experiment. This is no less clear in their form and versification, with which we are here concerned, than in their diction and style and contents. Hexameters are often spoken of as sung, and we hear of a musical accompaniment, but it is probable that what was really sung was a more primitive form of ballad in shorter lines, such as conjecture or imagination can sometimes recon- struct out of Homeric lines. Thus rjpdfieOa /liya kvSo^, e7ri > V OLO CTT apL(TT€pa pcofLfjcrat PS)V 1887 P tee e. r *■ ,c, r ' THE HEROIC has the a.r of being an ancient tripudiun, ov threefold dance of an armed warrior. The longer hexameter '"t was the work mainly of lonians. At all events it belongs to a d.alect whose uncontracted forms lent themselves to dactyhc verse (M'^«"/'«'- XaA^.«, lpl», &?yvpiow\ and, further, forms of words have been adapted to dactylic verse by successive — '->^ (^'l kUvaroliK6.i.a-:o,, with the first ^y"-Me lengthen d> Much of the diction has been cast into shapes which readily n^ake up a hexameter. There are a great many complete hnes that are recurrent or conventional {kvkKikoI anxo', e.g. avrap ,wh n6-'W 1 ^v...^.^^ 1 X XP^.o. ,pSr», is a convenient term oiH.^'r,, ™\'^^;^X which is made audible, in verse, by a normal short syllable. It may be made audible in music by a short note, J^ ; or made percepfble by . a movement of the body, as in dancing. HEXAMETER 3 an indefinite string of groups of four short sounds. But the perpendicular lines which we have drawn to separata them into groups have no real existence. The groups are made perceptible as groups by an ictus on the first of each (in dactylic verse — anapaests belong to the same yivo/v^w Ov>v_;w &C. Next, the grouping can be made still more obvious by putting a long for the first two shorts. This gives us an indefinite string of ' dactyls ', and it must be cut into ' lengths ' to give us verse. ' Lengths cut off' is Aristotle's phrase {royLT] pvOjjLov, TfirjToi). A hexameter is a 'length' of six fjiirpa, feet or units, a tetrameter a ' length ' of four. But if these lengths are to be perceived as separate groups, they must have some sort of structure, the beginning or end must be made perceptible. In the hexameter the end is made perceptible by decreeing that the last foot shall ahvays be of two syllables. But the effect of this closing cadence would be obscured unless the preceding foot were the normal one. Therefore a hexa- meter ends in a dactyl and spondee, or at all events in a dactyl and a disyllabic foot. There is another possibility or alter- native (it is found in Homer and was fancied in a later age by Alexandrian poets and their imitators at Rome) — the last two feet may consist of long syllables^ or the line may end in three long syllables and a doubtful syllable {iKrj^oXov 'AttoXXcouo^j. In this case the foi^rtk foot will usually be a dactyl. This theoretic construction, the construction of a hexameter i?i vacuo, leaves us with two problems to consider, a smaller and a larger one. The smaller question is about the nature of the last foot or last two syllables of a line. Such effects as Virgil's Turres et tecta Latinor'(-um) ardua cernebant iuvenes B 2 4 THE HEROIC suggest that our statement ought to have been ' the end of the group of six is marked by making the last foot a spondee'. If so, in an ending like atiigii arva, the short syllable will be a syllaba anceps in the sense that it can count as a long. It is at the end of a line or group, not thrust up, as it were, against a following syllable whose beginning ends it. A short pause or silence completes the time, like a short rest in music, the pause for which the Greek symbol was A ; so that the last foot is — v^ A . That seems the most reasonable way of contemplating the matter. The objection to it is that it does not accord with ancient theory, and that there are a few lines — but very few — which it does not explain. The last foot was thought of by the ancients not as a dactyl which combined its two short syllables into one long, but as a dactyl which dropped its last syllable, — w (w). On this view the last syllable is anceps in the sense that if actually long it may count as short. Against this ancient theory a reader of Virgil may raise the objection : ' but there are lines which end with a real and unmistakable spondee, for the last syllable is not the last syllable of a word ; there is " elision " : Turres et tecta Latinor(um) ardua cernebant iuvenes.' 'And, again', he may proceed, 'are not lines which end in almost equally unmistakable spondees (like primus ab oris) far more numerous ? ' But elision at the end of a hexameter is not certainly older than Callimachus.^ And it is chiefly in ^ "H/xtcru /xot ^vx^js CTt to weov, riynav 8' ovk 6i8' €iT^ epos eiV 'Aidtjs Tjpnacre, nKrjv dcpaves. It has been suggested that Ennius introduced this effect in Latin by writing ' altisonum cael(um) ' at the end of a line. Ennius is credited with inv iting monosyllabic forms cael and gau for caelum and gaudium, on the analogy of Homer's Kpt and dw. Elision would not explain gau so readily as it does caei. But it may be doubted whether Ennius ever s ■ . HEXAMETER 5 Latin that spondees are in an overwhelming majority. In Homer one line in five (or if we nowhere write a u e0eX>cu- (TTLKov, one line in four) ends in a short vowel not closed by a consonant (as in vfja, 'ArpuSao, edrjKe). In Ovid, Me^. vii, the proportion is i in 12; in Virgil, ^^/. r in 14-3, Georg. I in II, Aen. i. 1-500, i in 17-2. The Romans had a liking for making a last syllable long when it could legitimately be done, as is seen in Horace's Sapphics and Alcaics. So, perhaps, the ancient theory of a trochaic ending is to some extent justified by the history of the verse. The lines which the other theory does not explain are chiefly these two in Virgil : inseritur vero et fetu nucis arbutus horrid(a) et steriles platani, {Georg. ii. 69.) and et spumas miscent argenti vivaque sulpur(a) Idaeasque pices. i^Georg. iii. 449.) But the text of neither passage is above suspicion. Much earlier we find in an epigram of Simonides : i) /xey' 'AOrjvaioLa-L (p6(ii>9 yeue$', tjulk 'ApicTTO- yeiTcoy "I-mTapyov Krelve kol 'Ap/ioSto?. But it is not safe to infer that either the hyperbaton or the trochee belonged to recognized metrical practice. When a proper name had to be put into a metrical inscription the rigidity of metrical laws had to yield a little. There are inscriptions extant in which metre has to submit to much harsher treatment than this. The second question is much more Important. In cutting really wrote such lines at all. They may have been fictitious examples of solecisms or eccentricities. We know (from Pompeius, 289. 10 k.) that Lucilius not merely said that there were 100 kinds of solecism, each with its own appropriate name, but actually described them : e. g. Tmesis : 'as if we were to write', 'as if Ennius wrote "Massili—portabant iuvenes ad litora— tanas." ' 6 THE HEROIC off a ' length ' of six dactyls we cut off what was too long, what cannot really be delivered in one breath, without a pause or break, as a single whole. Aristoxenus laid it down that for feet of this y€U09 {T^Tpdcrrj/ioi or TeTpd\povoL Tro^ey, of the yiuos 'lXov is four feet. This may or may not be an absolute and final law of rhythm. But it seems sound, and at all events the hexameter as we know it is not one continuous whole, but a whole made up of two parts. Mere theory will not enable us to divide it. Mere theory might suggest 2>-'c3i ^"dTL, tis irore rds iroXv^pva-ov. {Oed. Tyr. 151.) This is part of a lyric system ^ it is not an epic line. We must go back from theory to history and recall rjpdfjLeda fiiya kvSos \ k7r€vis kcrecpdvojTO {II. xi. 36), liovv ^viv evpvfxeTojnov (x. 292), as in Ennius' s ponebdt ante salutem, an original long quantity of the termination perhaps survives. I HEXAMETER ii is often a * pause ', but it is not a pause of measurable or fixed duration.^ It is not a metrical 'pause'; by that is meant an interval of silence which can be measured in terms of the duration of short syllables or y^pbvoi npc^Toi : e. g. when a group of three dactyls is ' catalectic ' there is a pause at the end which is St^povos or 8ia-7]fL09 (arboribusque com|ae ^). It is a metrical division with which some sort of rhetorical pause or pause in the sense usually coincides. If it is doubtful where the metrical caesura occurs, the doubt must be set at rest by the sense or the rhetoric. Thus in the line : pascite ut ante boves, pueri ; submittite tauros, if the punctuation is right which puts only a comma before pueri and a semicolon after it, the caesura is after ' pueri ', it is ipia T€V)(^€ Kvu€(rorL there is no bucolic diaeresis in any sense that it is worth while to contemplate. To reckon as a bucolic diaeresis any ending of a word at the end of the fourth foot will give misleading results. Gleditsch {Metrik^ p. 119), following Hartel, says that 60 per cent, of Homer's lines have bucolic division, or 15,200 verses in all. This figure must be arrived at by counting lines like eXcop^a T^vyjc kvv^(t> ^7, 88, 98, 99, 100. If I am right about these, and if these 100 lines are a fair sample of Virgil's verse, there should be about 2,000 hephthemimeral lines in the Aeneid ; and to these I would, of course, add about 1,000 more, those of the type 'non comptae mansere comae '. Thus about a third of the lines of the Aeneid HEXAMETER 29 would be hephthemimeral, and about two-thirds penthemimeral. 'Trochaic' would be Mr. Butcher's 'abnormal trochaic' — 119, But there are more than that, though not, I think, many more : lines to be added are those like Norden's litora deseruere, latet sub classibus aequor (which would slightly reduce the total of 1,037 or 1,020 or whatever it is). For these reasons I regard both Norden's statistics and ^ Mr. Butcher's as unconvincing and unsatisfactory. If out of 9,878 lines in the Aeneid as many as 8,349 have penthemimeral caesura, Virgil's verse could no longer be extolled for its variety of cadence ; it would, on the contrary, be exceedingly monotonous. And if there are 1,156 lines with trochaic caesura, it would be perceptibly less unlike Homer's verse than it is. The counting, it would seem, should be done over again. But would the result be worth the labour ? Possibly ; but it might result only in putting more accurately what can be roughly estimated by surveying a few hundred lines. In the first 100 lines of book i I do not get a result seriously different from that elicited from book vi : 22 hephthe- mimeral lines (excluding ' non comptae mansere comae' — one of these is 'posthabita coluisse Samo. ||hic illius arma': could any caesura that is not the chief caesura justify the hiatus ?). Holding this view, I ' disable ' or disqualify Mr. Butcher's con- clusions about the authorship of Opuscula\ that the Culex was written by Virgil is a conclusion that I am inclined to disagree with in any case. What is said above in general terms about Virgil is confirmed by a survey of 1,000 lines of the Ae7ieid (viii.-ix. 269) : lines with hephthemimeral division, 25-8 percent.; lines in which it is difficult to decide between penthemimeral and hephthemimeral, but in which penthemimeral is certainly not predominant, 5-8 per cent. ; these two together, 31-6 per cent. ; trochaic division, i-8 per cent. ; lines in which it is difficult to decide between trochaic and heph- themimeral, 4*4 per cent. : total lines which are not penthemimeral, 37*8 per cent. CHAPTER II THE COLLOCATION OF WORDS IN EPIC VERSE, ELISION, HIATUS, ETC The structure of a particular line has been represented above (p. 13) by a diagram in which the ^povoL or syllables look like rectangular blocks of stone carefully fitted together. In order to realize the nature of ancient verse, as compared with the verse of English , or German or other modern languages, it may be worth while to pursue this illustration a little further. The words in Greek and Latin verse are rather like rectangular pieces of marble built into a wall. They are exactly measured ; ancient versification contemplates long and short syllables which in length are in the ratio 2:1. The long is twice the short, though in actual pronunciation it cannot have been precisely so. The second syllable in ' volucres ' or woXv- Tpo7ro9, when lengthened as it might be in formal and finished verse, can hardly have been really equal to the first syllable of (TKrjTTTpov. But for metre the syllables counted as equal. Further, words in ancient verse seem to come more into contact than words in the more or less accentual verse of a language like English. They fill up more completely the spaces provided for them by metre, and the end of a word is not unaffected by the beginning of the word that follows it. We may roughly represent this in a diagram as follows, drawing perpendicular lines to mark off the (dactylic) time-spaces provided by metre : r 1 TtOV 1 1 a/ / 10 6ev ,6e a, e{j y a rep CI] At o?, etn : n E Kat 7]f. LV COLLOCATION OF WORDS IN EPIC VERSE 31 If we now take as an example of English verse, in its loosest and most accentual form, Longfellow's line This is the forest primeval, the murmuring pines and the hemlocks, we must represent this in a very different way — how, exactly, it is difficult to say. There must be quantitative spaces ; all verse has quantity, it is not a series of momentary explosions of sound. But in English a strong stress accent may be said to usurp the place of quantity and to be confused with it. The diagram may be something like this : Th IS 15 the Ihc n Dines A and Ihe Here the dotted lines in the first two feet are meant to indicate the effect of the English stress accent. In the first foot the words ' is ' and ' the ' are about equally inconspicuous to the ordinary English ear. But in ancient verse the con- sonants sf^, whether within the same word or in separate words, would mean a long syllable {Kparepos 6e6s). English verse differs in many features from the verse of Greece and Rome, and some of these it may be instructive to consider in dealing with iambic verse. Forest may be said to be an iambus, and some English writers have tried to use it so. But the English accent really forbids its use as a disyllabic foot of which the second syllable is the more important or con- spicuous. The Latin accent of the classical period did not prevent the metrical use of ' dabunt ' as an iambus, and it had no power to make ' p^pulus ' a dactyl. What we are concerned with at present is the comparative looseness with which English words are set down side by side. What is normal in Greek and Latin is rare in English. Elision : the definite article the 32 COLLOCATION OF WORDS sometimes seems to suffer elision in verse, but hiatus is the rule. Length by ' position ' : and pulpit, drum ecclesiastic, was beat with fist instead of a stick — the effect is intentionally comic, and the words ' a stick ' are normally qualified to form the last foot of a purely iambic line, a stick. In the opiis quadratum^ or ex saxo quad7'ato^ of ancient verse, we have to consider three things: (i) 'length by 'position'; (2) elision (with its opposite aphaeresis, which is much less common ; and with crasis, which is also comparatively infre- quent) ; (3) the conditions under which hiatus is allowed. I. The phrase 'length by position ', as it is now used, means that a syllable is long owing to the ' position ' of the vowel before two consonants.' Originally ' by position ' Qkcru was contrasted with (pvaet in the sense of ' convention ', deliberate or artificial agreement, avi'OrJKr} ; dea-ei fiaKpa meant ' long by poetic convention '. It is more convenient to think of ' position ' in the other sense. When a short vowel is followed by one consonant within a word the consonant goes with the following syllable {fid-Xd-Kos), and no lengthening can occur. ^ The consonant at the end of a word goes with the vowel before it, and a possibility of lengthening begins. But with only one consonant lengthening is infrequent. With two, if they are a mute and a liquid, it is frequent, but still optional. With two consonants that are not a mute followed by a liquid lengthening is the rule, and exceptions very few. So we can construct an ascending scale of lengthened syllables, beginning with what are licences and ending with what is all but compulsory. (a) Lengthenings before one consonant within a word, which have no justification except metrical convenience : OLKoifiaTO'Sf aQdvaros, (XTroTriarja-L, Ze^vpir), avue^^e^, wdpe^dj, IN EPIC VERSE 33 liiXavi, ifxaOov (written fxetXapi, e/xfxadoi^). Later poets seldom ventured upon such changes except in the case of proper names (Italia, Sidonius). These syllables are really 0€, * taxique pinusque, Othrysque Pindusque '). Many apparent lengthenings before one consonant in Homer were accounted for when Bentley detected the original presence of the Digamma. In other cases the lengthened vowel was one that had once been long. But there are many cases to which neither of these explanations is applicable. Very frequently the lengthening is before a more or less marked caesin-a^ and the ictus of the first foot had a similar effect (eTrei 8r], rd irepl KaXd). Of Roman poets Ennius and Virgil sometimes did what Homer had done, but other poets were very sparing in their use of such things or, like Lucretius and CatuUuS; avoided them altogether. When Horace wrote 'ne quis humasse velTt Aiacem ' he was deliberately reproducing an old quantity — it is a legal archaism in a king's interdict. Virgil has simt before the penthemimeral caesura and a full stop [AefL X. 433) : tela manusque sinit. hinc Pallas instat et urget. In a few passages he has a lengthening for which no such 1887 D 34 COLLOCATION OF WORDS justification is easily discoverable and which must be set down as Homeric or pseudo-Homeric licences : Aen. iv. 64 pectoribus uihians, ix. 610 terga fatigamus hasta. Roman metricians (e.g. Velius Longus, 52. 9 k., in the first century of our era) had a theory that h might contribute to quantity by position. They were following Greek writers who had before them lines like 7] okiyov 01 Trai8a koiKora yetvaro TvBevs. But to give h this force was no part of the metrical practice of classical poets: for them, ' auctus hymenaeo' or 'languentis hyacinthi' did not differ from, or was no moie normal than, 'gravidus auctumno '. But the discussions of the grammarians seem to have given rise to a belief that h ought to contribute to quantity by position. It is the practice of the early Christian poets (so that, e.g. a hexameter may begin with vir Jminilis)^ while the * paganus pervicacissimus ' Claudian avoids it. {c) Lengthening before a mute and a liquid : similis voliicri, mox vera volucris. A mute and a liquid are easily pronounced together and both can be put in the following syllable, ' volu-cri'. Here a limitation at once comes into view. If one of the letters is in one word and the other in another, this cannot be done, et i'ipa7n alone is possible, not e-t?-ipa?n. So, too, in compounds: 'obrepo' and iKpel are necessary, ' obrepo ' and eKpel are inadmissible. But, while shortening is excluded in these cases, lengthening freely takes place in Greek even when both consonants are in a following word : dpLTrpeTred Tpooea-aiv. 6(f)pd TrpoaB' dWoav. (j)LXd (j)poi^icou. fiera irpodTOLorLV. There are great differences in this matter between different poets^ different dialects, and different periods of poetry. In Homer and Hesiod lengthening is the rule, whether the two IN EPIC VERSE 35 consonants are in the same word as the vowel or in the next one : Kat re irpo o tov kv6r]aev : rd t kacrofi^va irpo r kovTa: Shortening is infrequent, and usually has some special, though it may be slight, justification {pv8\ yap ovSe ApvavTO^ vlb9 . . . , AcppoSiTT], proper names : T€L\€(n7rXfjra, irporpa- iriadai, words otherwise excluded from dactylic verse. Add //. xi. 69 roc Se Spdy/jLara : xv. 142 tSpvcre Op6ua>). The reader of Homer would be surprised to meet with erpefLei/ or with Tei)(€d Tpdocoy. Shortening occurs in Homer only when the liquid is A or p, not with yLt or u.^ The opposite extreme is found in Attic comedy. Shortening is invariable, except that /3X and yX sometimes lengthen the syllable, and y/z, 8/j., yu, Sv regularly do.^ Tragedy admitted lengthening, Trdrpos, viKpos. This ' correptio Attica ' belongs to the history of iambic verse. In Roman comedy also and in spoken Latin shortening was the rule. Lengthening belonged to serious, finished, and formal verse; it came in with Ennius's introduction of the hexameter, but compared with Homer the Latin hexameter admits it very sparingly. Tejiebrae, latebrae, pharetra are found, and a line may end with patres (Stat. Silv. V. 3. 215), or libro {Georg. ii. 77). Here the ictus is on the syllable lengthened. But this is not invariably the case ('iuvat integros accedere fontes', Lucr.). In a compound like 7iavifragus the / is always short, and when the two consonants are in a separate word lengthening is rare^ and is usually a direct and deliberate reminiscence of Homer {spiculaque cli- peique, Aen. vn. 186). Endmgs like jnercenttir Atridae, Etruria crevit, vioe7itd Troiae are common. In other metres also, ^ ndTpus, according to Ebeling, occurs about 120 times (and the dative TTdTpi about 60 times), -narpos appeal's in //. vi. 479 (and nowhere else) ; Kai TTore tis einrfaf ' vaTpos y ode ttoWov ajiHVOJV ', but here there is evidence for the reading diroi. 2 Regularly in Latin also : Mr. Housman contends (and he is probably right) that cycnus must be the spelling, not cygnits, if the first S3'llable is to be short. D 2 36 COLLOCATION OF WORDS lengthening is very rare when the mute and liquid are in the next word. Catullus has it in his experiments in pure iambi (the Phaselus : ' Propontida trucemve Ponticum sinum, per impotentia freta', where, according to Christ, he is using weak longs to depict the rapid movement of the vessel: so in his lampoon on Caesar, ^ultima Britannia', xxix. 4). Latin seems to have had no natural tendency to such effects ; they were lessons learned from the metrical practice of certain Greek poets. They were not native to Athens either ; rather, Attic tragedy learned them from the Ionian epic and the Ionian iambics of Archilochus. (/) Shortening before consonants other than a mute and a liquid occurs very seldom in either language. Pindar has ecrXoy^ and there are a few examples of shortening with iiv (Aes. Agam. 990, Eur. Bacch. 71 v/ii^rjcrco; see Christ, Metrik, p. 14). Apart from these, shortening occurs only with ctac, (ttt, (TT and (TTp, ^, in Latin also with initial s^u. In one or two instances shortening can be avoided by giving to / the sound of y i^Ia-TLaLav, II. ii. 537, re aKLrj, Hes. Erga 589), but in a greater number of cases nothing of the kind is possible and it becomes unnecessary to assume it at all. Shortening occurs chiefly with proper names {ZcckvuOo^, ZeXeia, ^KdfxavBpos:) or words which otherwise could not find a place in a hexameter {(TKeTTapvov, Od. v. 237).^ 1 Shortening with consonants like sc^ str is rare, and therefore pre- sumably difficult or repellent. It may be asked, * How was it possible at all?' There is the extreme view that even in English, in a word like ' restriction ', the first syllable is not short, but is only inconspicuous beside the strong stress accent pn the second S3'llable. The answer is perhaps this : a rule that the syllable cannot be short would be true only if pronunciation were always so rapid as just to avoid 'gabbling' or * mumbling ' consonants ; but the delivery of ancient verse (and probably the pronunciation of Greek and Latin generally) was not so rapid as that, and hence there was no insuperable difficulty about making a syllable with sp or str relatively short. Metre is not concerned with IN EPIC VERSE 37 Roman poets are less strict. In the early drama — even in anapaests, the metre most akin to dactylic verse — there are numerous shortenings which have no place in the hexameter, or in the stricter prosody of the Augustan age. The Law of Breves Breviantes {infra, p. 90) sanctions not only molestSrum, but also voluptdtefii ; as it sanctioned, too, the shortening of a naturally long vowel in dd?m md7isit. But even in hexameters there are things for which there is no analogy in Greek. There is, to begin with, the large exception that down to the time of Lucretius a final s is frequently attenuated and has no metrical effect : tum laterali(s) dolor, certissimu' nuntiu' mortis, (Ennius.) 'frequently', but not regularly, for it often causes length by position : Corneliu' suaviloquenti ore Cethegus Marcu' Tuditano conlega. (Ennius.) Lucretius has this neglect of i" freely, Catullus has it once (' tu dabi' supplicium ', in what may have been his earliest poem, though put last in our editions), and Cicero, in 46 B.C., calls it subrusticum. If it persisted later, it was in comedy only. 2 was not always in Italy distinctly a double consonant ; it was more like English z ox s (cf. Sagtmtum for ZdKvi^do9 : zenatuOy Faliscan gen. of senatiis^ &c.). Hence it was perhaps rather easier for Virgil to put Zacynthus at the end of a line than it had been for Homer {Aen. iii. 270). Seneca in iambics has 'tranquilla Zephyri ' {Agam. 433). Shortening before sc, sp, &c., is found chiefly in Lucretius ('cedere squamigeris: mollia strata ',&c., and even'pendentibu(s) structas ') and in the Satires of Horace, where it occurs nine times ('praemia scribae, saepe stilum vertas', &c. In i. 3. 44 ultimate questions of phonetics, but only with the grouping of syllables which can be classified more or less reasonably as long and short. ss COLLOCATION OF WORDS ('non fastldire. strabonem') it is rendered easier by a full stop). Propertius has it five times (or six times if we include iii. i. 27, Scamandro).^ The formal heroic Epos rejected it almost entirely. Virgil, who has Zacynthiis only once, has this also only once, and with the help of a full stop, Aen. xi. 309 'ponite. spes sibi quisque'. Shortening before sni in the case of the word smaragdus is found in various Roman poets, including Tibullus^ Propertius, Ovid, Lucan, and Statius. Silius has Zacynthiis several times. If it was difficult, though not absolutely interdicted, to shorten the syllable, to lengthen a short vowel in a preceding word was also difficult or repellent. What the Roman poets do is to avoid rather carefully such a collocation of words as would necessi- tate it. Ennius has stabilita scanma sohwiqtie ; Catullus, in hexameters^ has one such lengthening : nulla fugae ratio, nulla spes; (Ixiv. 186.) Tibullus has pro segete spicas^' and one other instance. L. Miiller {^De Re Metrica^ p. 390) collects six examples in all in epic verse. Besides these there are a few in Seneca and Martial, and Juvenal's occulta spolia (viii. 107). The extreme scarcity of the effect seems to make it clear that the avoidance of it was deliberate. 2. Ancient verse, with its syllables carefully placed in con- tiguity, ' ut tesserulae omnes ', naturally abhorred ' hiatus '. If a word ends in a vowel it must be definitely closed or dismissed by running up against the consonantal beginning of the next word. If the next word began with a vowel the two vowels were left facing each other with some sort of unsatis- factory 'rift' or 'gap' between them. It was not only in ^ Catullus had wiitten unda Scamandri. With the probable exception of Propertius, later poets * showed more respect for metre than for the gods ', as L. Miiller remarks, by calling the river Xanthiis. ^ Unless we should read, with some editors, an abl. in -t, segeii. IN EPIC VERSE 39 verse that this effect was disliked. When prose was elaborated and its rhythm studied, hiatus there, too, was avoided. The Isocratean prose of the fourth century rejects an arrangement of words like inel ijSr] ovSei^.^ The chief device by which hiatus was avoided was elision. When two vowels came together the preceding one was thrust out or forced out of the metrical scheme (e/idere, kKSXtp^iv). It disappeared, it had no effect. But, it must be remembered, it is only for metre that it disappears. That it disappeared entirely in pronunciation is incredible. Perhaps it did so, or all but did so, in a few cases, when it was an extremely slight sound. The final e of que^ ve, ne may have been almost •inaudible ; it had a tendency to efface itself in ordinary pro- nunciation apart from verse (cf. vin^ viden ; nee beside neqice ; ac beside atque ; and it has been rendered probable that the final e of some words like inde^ unde, netnpe was similarly attenuated or dropped in dramatic or colloquial verse). But the elided vowel is often by no means a slight one. In Latin it is often long. In quare age it is the short a of age that forms the second syllable of the dactyl. How was the e pronounced ? It is difficult to say. Ancient verse must be thought of as delivered more slowly than our verse or than an English reader tends to deliver it. The metrical scheme consists of spaces or com- partments which have some room in them, room it would seem for a rather sudden change. The sound of e is begun in the second compartment of the dactyl, but instead of going on till a consonant stops it or till it runs up against a consonant, it is supplanted, after occupying some slight and not exactly measurable part of the space — vocalis ante vocalem correpta — ^ Though Latin treated long final vowels very differently from Greek, the Eomans also disliked the accumulation of such effects : Rhet. ad Herenn. iv, § i8 ' compositio est verborum constructio aequabiliter perpolita. ea conservabitur si fugiemus crebras vocalium concursiones, quae vastam atque hiantem orationem reddunt, ut haec est: '"baccae aereae amoenissimae impendebant.'' ' 40 COLLOCATION OF WORDS by the short a that follows. In ordinary speech, apart from verse, something similar may have happened ; magna opere becoming magno opere and then magndpere} It is chiefly in Latin that this extreme kind of elision is found. There is a great difference in this matter between the two languages. Short syllables are freely elided in both, and especially those short syllables that are of very frequent occurrence in the terminations of nouns or verbs (a, e, o, a and e in Latin), but Greek treats long syllables differently. In dactylic verse it shortens the long syllable, if the syllable that follows is short, irXdyy^Qr] eirel . . .; and is careful to avoid placing words so that two long vowels come together {7r\dy)(Bri ijdrj would be strange in Homeric verse, though quar{e) omnes presents no difficulty in Latin). In other forms of verse, and conspicuously and strictly in iambic and trochaic verse, Greek arranges words so that a ^ It has been tliought that in the case of a diphthong one of" the two component vowels suffered ehsion. Certainly, with eppeev vScup coming at the end of a line, a beginning such as eppet vdojp might be explained as eppf' vSaip — a case perhaps of the atticization of the Homeric text. But a pronunciation which would be represented by di'dpa jxo' evuQne is not a thing that it would be well to inculcate or practise. If we hold that an elided vowel did not vanish entirely, it seems clearly best to pronounce oi rapidly. In the case of a single vowel such as rj or o;, the theory has to assume that it is made up of two xpo^^i^ o^" ^torae (»w> w or v^v^) and that one of them goes out. The explanation becomes a rather unreal one ; it is diflicult to feel or to imagine that this is what actually happens. There is perhaps a feeling in some minds that only e and o can be short, and 7; and cv cannot be. But this is an illusion. If it were true, Boeotian Xiyovpav (^ = ?\iyvpdv would be impossible, and KdrovAos could not stand for Caiulus, as it does. Certain Attic spellings attested by inscriptions (ttow, -noili) do, how- ever, point to the disappearance of an t, in a case of similar shortening within a word. But it is doubtful whether these should be admitted in literary texts where Trotw is an iambus. There is no difficulty about ttoiSj as an iambus, for olos and 'Mip'aios can be scanned with their first and third syllables short. IxN EPIC VERSE 41 long vowel is not followed by another vowel, whether long or short. Roman poets sometimes wrote verses with few or no elisions in them, aiming at a special effect of smoothness or rapidity, but no such strictness as this was inherent in the language or in the general principles of Latin verse. What has been said of Greek verse is not universally true. Reservations or exceptions have to be made. (a) The terminations 01 and ai sometimes suffer elision after the Latin fashion : most frequently in comedy, but such elision is not unknown to epic verse {^ovXo/jl' eya>, IL i. 117) or to dramatic and lyric verse {aetpofi ovS' dTraxro/jLac ktX., Soph. Track. 216). Cf. ocrrjixepat {oo-at r]fj,epaL). The Lesbian poets have elisions of the Latin type (dXXa fie kcoXv{€i) ai8ct)9), but not so frequently as Latin. (b) The converse of elision may take place, Aphaeresis. The second or following vowel may be extruded or suppressed {^\i7] 'yo) voTjixa — an example has been found in Homer also, //. i. 277 ixriTe (TV, UrjXeiSTj, 'OeX' epL^ifieuat). This happens to e or a with a long vowel preceding. (c) The two vowels may be fused into one new vowel sound. This fusion is Crasis : Kayw, ovfio^, &c. It is found in the text of Homer (//. viii. 360 dXXcc naTrjp ovfxos (6 ejios)), though not in places where the metre makes a long syllable necessary. Its existence in the drama is undoubted : ovcpSpet = €(p6p€L, rd/jLo. = to, e/xd, &c. Sometimes the vowel of one of the words prevails, but is changed from short to long by the absorption of the other : rdpa = tol dpa, dSaveio-aTO = d kSaveiaaTO. There are other cases where something of the nature of crasis happens, though it is not represented in writing, e. g. rj XdOer' rj ovk kvo-qcrev, ey cb ovt kfiavTov. The word avvaXoKprj is used for this by Bickel. It is difficult to say whether in such cases something of the nature of crasis takes place, or something more like elision 42 COLLOCATION OF WORDS of the Latin type. ' Elision ' is a convenient metrical term for all cases in which the preceding vowel has no effect, and only the following one takes a place in the metrical scheme. Besides vowel-endings Latin also elided a final m and its preceding vowel. The attenuation of the vi in ordinary pro- nunciation seems to be attested by the great frequency with which it is left unwritten in inscriptions (e.g. Taurasia(m) Cisauna(m) in the earliest of the Scipio epitaphs, but the omission is not only early, it is common in inscriptions of all periods). It is indicated also by forms like animadverto. Probably the sound was gradually attenuated ; it grew fainter as time went on. Ennius could leave it unelided in a heroic hexameter (' milia militum octo '), and this seems to have been rather more frequent in Saturnian verse. Virgil did not follow Ennius's example. Horace has the effect once, in satire ('cocto num adest honor idem?' ii. 2. 28). The text of TibuUus presents one example (i. 5. -^^i ' ^^ tantum venerata virum, hunc sedula curet'); there are two in Propertius and one in Manilius. (Attempts have of course been made to remove these by conjecture, but the conjectures are not very con- vincing ; so it is perhaps better to follow L. Milller in retaining them.) It remains to consider the conditions under which Hiatus is admitted. 3. The occurrence of hiatus is a large subject, and a complete account of it would involve a vast number of details and small facts. What is important for the ordinary student is to learn in outline how the practice of poets varied at different periods, to understand some of the general principles which regulate it, and to appreciate the poetic or technical effect which particular writers had in mind when they admitted hiatus. In the Homeric text as it stood 200 years ago hiatus was very frequent, but about that time Bentley made or had made the discovery of the /^, — whereby * ipsam sanavit artem metricam ' — with the result that the number of cases was IN EPIC VERSE 43 immensely reduced. ]\Iost of the harsher examples of it disap. peared (ev feiScos — AnoXXcoui fdvaKTL, &c.). Still, in spite of this, hiatus in Homer is fairly frequent, more frequent than in later Greek epic {^/xii'dev, | €z' Trore roi . . . , Ticreiav Aavaoi \ k^a SaKpva, Szc). Sometimes it is rendered easier by a pause in the sense. Sometimes it occurs between a long and a short syllable, an easier thing than hiatus between syllables of the same quantity, easier between D 3 than between O O.' Lyric poetry and Attic drama were far more strict. In the drama, when hiatus does occur, there is usually some sort of interjection or exclamation or repetition (co 0VT09 . . ., co eta . . . LT€ ^OLK^aL 'lt^ PoLK^ai. Christ, Meirik, p. 40). d, €, and 0, we have seen, are readily elided, especially in common terminations. i and v are less accommodating. darv could not suffer elision at all. l, when it is a verbal suffix, is frequently elided (ei/z'(0* "^^^VH-'i^^ TLOr](T\i)) : in the indeclinable en and dXXodi it is elided almost as readily as e in dXXore. The i of the dative is much less frequently elided {)(pv(T€irj KepKiS' ixpatvev, Od. v. 62). Elision of on is rare and doubtful, ri is not elided, re is eHded with great frequency. Disyllabic prepositions are readily elided, but irpo refuses to become irp . Besides these and similar cases, there is, in Greek dactylic verse, the hiatus which accompanies the shortening of a long vowel or diphthong. Whether this can be regarded as an obscure case of elision has already been discussed. For the ordinary reader it is not that, nor does he quite feel it to be a case of real hiatus ; it is rather a hiatus that is paid for or atoned for, and so cancelled, by loss of quantity. When Catullus wrote the line (Ixvi. 11) : qua rex tempestate novo | auctus hymenaeo he was deliberately introducing two Greek effects into his ^ L. Miiller remarks that the line 'quid struit aut qua spe [ inimica in gente moratur' would be vitiosus with ' hostili ' for 'inimica'. 44 ' COLLOCATION OF WORDS verse ; and many things that are exceptional in Roman versification are of this type, experiments intended to show learning and taste, though the taste is sometimes a little eccentric. The early drama shows no great abhorrence of hiatus, though many instances of supposed hiatus have been removed by emendation in one way or another, and there are clear cases where Plautus selects an unusual word or arrangement of words to avoid hiatus. Sometimes (though not always or even more frequently than not) it occurs when there is a change of speaker,^ and it is well established, though not very frequent, at the juncture between two parts of a long line (after the first four feet of a trochaic tetrameter or iambic ' Septenarius '). But Ennius in his epic, both in regard to elision and hiatus, had adopted very strict principles of versification. In this matter he does not follow Homer. He has fewer elisions than Virgil, and hiatus is not certainly found in the extant lines except twice in the case of a final 7n ('milia militum octo : dum quidem unus homo'; and perhaps 'inimicitiam agitantes'), and, with shortening, in Scipio invicte. Latin verse, as it is known to us in literature from Ennius to Juvenal, and in many writers later than Juvenal, is charac- terized by great strictness in the observance of metrical rules. In the use which he made of elision, especially in the Aeneid^ to add dignity and volume to his verse ('omnia praecep(i) atqu(e) animo mec(um) ante peregi'), Virgil stands rather apart from other poets. He is great enough, like Homer, to be a law to himself. Post-Augustan poets, though they owed much to him in many ways, followed rather the norm of Ovid in the mechanism of their verse. Hiatus also is used with some freedom by Virgil, and he has effects in which later poets did not venture to follow him. Perhaps they thought that some of ihem were due to his having left the Aeneid unrevised * This is not wholly unknown in Greek : an example has been found in Menander. IN EPIC VERSE 45 at his death. They knew that incomplete Hnes were due to that cause, and impressive though the effect of them sometimes is (as in ' numina magna deum ', Aen. ii. 623) they never tried to repeat it. Types of hiatus admitted by Virgil can be very definitely classified : i. After a syllable on which the icfus of the verse does not fall. (a) Two instances of hiatus after a short syllable, where there is a marked pause in the sense : addam cerea pruna. honos erit huic quoque porno. {£c/. ii. 53.) et vera incessu patuit dea. ille ubi matrem. (Aen. i. 405.) These are very exceptional. (b) Hiatus of the Greek type, with shortening of a long vowel. Sometimes this brings into dactylic verse a word otherwise inadmissible ^ : insulae lonio in magno. {Aen. iii. 211.) Sometimes the effect is to bring the same syllable into a Hne with two different quantities, a thing sometimes affected by poets from Homer's ^Ape^, 'Ape^ downwards : vale vale, inquit, loUa. (£d. iii. 79.) Hyla Hyla, omne- sonaret. (lb. vi. 44.) ^ The shortening of va/e here has nothing to do with the ' Law * For elision of insulae was not allowed ; elision of a word of Cretic form (— v^ — ) was carefully avoided, Ennius had written ' Scipi6 invicte'; Catullus, ' uno in lectul6' (Ivii. 7), with less excuse for his hendecasyllabic metre, readily admitted * lectiilo' (Accius in his Annales had the harsh hexameter-ending inaxime Athenae). 2 The effect aimed at was the presentation of a sound dying away or an echo dying in the distance. This interpretation of the poet's intention does not invalidate the metrical explanation ; the metrical graecism is the means by which the effect is achieved. 46 COLLOCATION OF WORDS of Breves Breviantes ' (which could shorten cave and similar words befo7'e a co?2so?ianf. Cf. 'vade vale cave ne titubes ', in Horace). The second vale is, to start with, exactly like the first ; it suffers shortening by what is substantially a Greek principle of versification.^ Such shortening, however, does not seem to have been entirely unknown to the Latin language. It appears to belong to Latin most clearly in the case of monosyllables followed by a short syllable. Plautus has ita me di a7?ient {ita me dt anient in trochaic verse, as he would have sed ita 7?ie di a?ne?it in iambics, ita me di ame'nt in anapaests {Fersa 492), as Catullus has non ita 7ne di a77ient in elegiacs, xcvii. i). This shortening probably answers to actual pronunciation of Latin. If it were merely a piece of Greek metrical practice we should expect it to occur in dactylic or anapaestic verse only. It is found in verse that is more or less informal or conversational : Lucilius, quid servas quo ea7ii ? [Sat. xxx. 21) ; Catullus, te in circo, te in omnilms {libellis : tabe7'7iis^ L. Miiller) ; Horace, Sat. i. 9. 38 si me a7nas. Virgil has it in the Eclogues (viii. 108) an qia a7}iant (also in ii. 65), and even in epic verse, Aen, vi. 507 te, aifiice, 7iequivi. A monosyllable was elided only when the following syllable was long^; before a short syllable it was shortened. (c) In a line of the Georgics (i. 437) Glauco et Panopeae et Inoo Melicertae Virgil has the effect just described (in Pa7topeae\ along with hiatus after Glauco, where is not shortened. He is repro- ducing a line of Euphorion : rXavKcp Kot Nrjpfji kol elvakico MeXiKiprrj. 1 In Horace's line ' vade, vale, cavS ne titubes ' there is no metrical question about cave. The syllable was short, and no metrical principle had to operate upon it to make it so. 2 The only exception in Virgil appears to be Aen. vi. 629 ' sed iam age '. Qui is elided in Lucilius 505 (1029), if the text can be trusted : sicuti te, qui ea quae speciem vitae esse putamus. IN EPIC VERSE 47 TXavKCi) in Greek verse might be followed by a vowel without much difficulty. Virgil's Glauco is almost as abnormal as Horace's et Esgtiiltfme allies ; not quite so abnormal, for Horace's line is iambic, and iambic verse is less tolerant of such effects. ii. After a syllable on which the Ictus of the verse falls, L. Miiller {De Re Metrlca, p. 375) defines four cases in which hiatus occurs, but perhaps the four cases can be reduced to three. (a) After one of the two dominant caesurae, the semiquinaria or semiseptenaria : munera sunt lauri | et suave rubens hyacinthus.^ {Ed. iii. 63.) banc sine me spem ferre tui, | audentior ibo. {^Aen. ix. 291.) In the second of these lines there is a marked pause in the sense, and it is that pause rather than any merely metrical division that is the real cause of the hiatus. If so, this class (a) would be merged in class (d), Infj-a. In the line ter sunt conati | imponere Pelio Ossam {Georg. i. 281.) the hiatus seems to contribute to the sense of effort and the adaptation of sound to sense ; i. e. there is again a cause other than the mere occurrence of the penthemimeral caesura. (b) With a proper name or a Greek name : Nereidum matri et Neptuno | Aegaeo, {^Aen, iii. 74.) a line which also illustrates (a) by its hiatus after mairl. ^ The close of this line would be felt to be a Graecism, and probablj' the hiatus also — a Graecism or, we might say, an annexation for Latin of an effect of Greek verse. The whole history of Roman poetry, on its formal side and from one point of view, may be thought of as a series of such 'annexations' or conquests of literarj' provinces; more or less successful annexation, with such' modifications or restrictions as make the thing effective in Latin verse. 48 COLLOCATION OF WORDS IN EPIC VERSE (c) With two short syllables coming before and after, or at all events before : evolat infelix et femineo | ululatu. stant et iuniperi et castaneae | hirsutae. (d) With a rhetorical pause or pause in the sense : si pereo, | hominum manibus periisse iuvabit. (Which also illustrates (c) ; there may be more than one justification for a metrical effect.) L. Muller adds, ' It cannot be doubted that Virgil had the authority of Ennius for these exceptions to metrical rule'. But there are doubts, (c) at least, if not (b), suggests much rather the example of the ' cantores Euphorionis ' than of Ennius : and ' castaneae hirsutae ' is in the Eclogues, where Virgil was more influenced by an Alexandrian than in his later works. (c) and (d), he further states^ are avoided by other classical poets, explaining Ovid's ' purpureae Aurorae' as justified by a proper name. The writer of a complete Hisioj-ia Metronwi] would be bound to tabulate the practice of each of the more important classical poets. What is attempted here is only to define and illustrate certain types. For the appreciation by the ordinary reader of metrical effects in a poet like Statius or Valerius, the main thing is to have some clear idea of the effects employed by Virgil. They are not admitted with much frequency even by him, and the general conclusion is obvious that hiatus was kept within very strict and well-defined limits by the classical poets of Rome. i CHAPTER III THE ELEGIAC COUPLET * The word eAeyo? is scarcely distinguishable from Bprjvo^ ; it meant a lament for the dead, Horace's qjterimonia {Ars Poet. 75) : kXeyelov was a metrical term for the form of verse, the couplet consisting of hexameter and pentameter. The Romans, however, used the word ' elegi ' in the latter sense, or at all events in a wider sense than the Greeks of the classical period. The elegiac couplet appears in literature early in the seventh century b. c, associated with the names of Callinus and Archilochus. It is a variation upon the heroic hexameter, in the direction of lyric poetry ; the couplet is a short stanza or strophe, and elegiac poets can be classed under ' Poetae Lyrici ' in the wide sense of that phrase. Though accom- panied in early times by the flute, it cannot be reckoned as ' melic '. It is clear that at a quite early date it became a measure that was merely recited, not necessarily sung. It arises in a period of unrest in Greece. The more placid Epos, with its prolonged tale of a heroic past, has now beside it forms of verse which express what is of nearer and more everyday interest. In Callinus the verse is used for a war song (tradition puts Callinus rather before Archilochus), and it has been thought that this w^as its original purpose, a cry of alarm when an enemy is approaching and a summons to arms. But this is uncertain, as is also the derivation of the word eXeyos ', it may be foreign, possibly Carian : if it is Greek perhaps the most plausible derivation is from the 1887 E 5° THE ELEGIAC COUPLET exclamation €: e Aeye e Aeye €. The verse is used to commemorate the dead by Simonides, and from being the verse of the epitaph (e7rfypa/z//ara eTTLToicpta or eTriKTJSeia) it became the verse of the epigram in general. But it had already taken on various tones : an erotic, though pensive, tone in Mimnermus, and in Mimnermus's contemporary, Solon, a note of vigorous and patriotic exhortation. In all its forms it is, of course, far more personal than the Epos : the per- sonality of the Homeric singer remains in the background — Hesiod is less reticent, as is also the author of the Hymn to Apollo — whereas the main purpose of elegiac verse is to give expression to the feelings or thoughts of the poet. The ' Pentameter ' is constructed on a different principle from the hexameter. It falls into two equal parts, and the division between them is marked by the end of a word. It is 2i + 2 J feet. In the second half only dactyls are admitted ; it is a fairly common principle of structure that the primary rhythm should become apparent at the close. There is catalexis twice ; and it was this that made it appropriate for the expression of grief or other emotions. It did not roll on confidently to its close like the hexameter, but twice sank or waned or 'died away in pain'.^ It is called ' Pentameter ' as early as the fifth century b. c. It was the TrevB-qixLjxeph twice ; the actual words or sounds amounted to five dactylic units or fieTpa. The scansion of it as five consecutive feet (— ^=^^ \ — ^=^^ \ |v^v-^— |v_/w— )is heard of, but this is the view only of some late metricians. It is two groups of dactyls, each KaTaX-qKTLKov eh orvXXa^rjy : _^=^_^^_;^ I _^v^_ V-/W But this description of it is not altogether free from diflficulty. ^ "09(v TTd'TajxtTpov TO) ijpojiKa) avffJTTTOi', ovx 6/xodpafxovvTa ttj tov irporepov SwcifAei, dA\' olov awcKirveovTa Kat cvvavoafievvvfjievov rati tov T€\(vTriaai'TOi Ti;\;aij' ol 5' varepov irpui airai'Tas adia^pupMi' ovtqj Ai5vjj.os kv TO) Tiepl TToiTjTuju (Orion). THE ELEGIAC COUPLET 51 If there was a pause as long as a long syllable after the first part (K€i^b9 8lxpouo9 or Sta-qfios)^ why is not the last syllable of that part a syllaba a7icepsV Why not, say, But that is not a legitimate line. It becomes legitimate if we write uicro-erai for OL)(€Tai, so that the syllable is long ' by position '. The verse then is continuous. If so, why the end of a word ? why not — a line that actually occurs in Euripides — The answer to that question seems to be that the first syllable of BaK\e1e would have to be prolonged in a way that belongs to song, and is quite unnatural in mere recitation. It would have to be rerpaa-Ty/xoy or lj . The exceptions quoted — from verse that is not ' melic ' — are apparent rather than real : rjfx^h 8* €19 ' E\Xrj(r\7rovrov aTreTrXeo/ieu ('EXXr]^ ttovtov). The two groups or phrases or KcoXa are connected in a somewhat peculiar fashion. Neither hiatus nor elision is allowed (elision is at all events extremely rare — practically forbidden). The last syllable of the first half must be either long by nature and followed by a consonant, or long by position. The hexameter in an elegiac couplet does not differ in any important way from the hexameter of the epic. Two points may be noticed : (a) The ' bucolic ' division of a hexameter was a favourite one when it was the penultimate line of an epigram (see p. 18). It was a division which had the effect of linking the two lines together. A group of connected words began before the end of the first line. ^ Compare e.g. the Archilochian juxtaposition in Horace (' Asynar- tete') : reducet in sedem vice. | nunc et Achaemenio. E 2 52 THE ELEGIAC COUPLET (b) A (TTTovSeid^cou or line ending in | is avoided in elegiac verse. A line like cara deum suboles, magnum lovis incrementum is most, naturally followed by a marked pause. It is too eminently fitted to be the last of a series to be happily placed as the first of a couplet. Both of these facts illustrate the close connexion between the two lines. But In spite of this connexion the lines are very rarely ' hypermetric ' or metrically continuous. There is Simonides' ingenious introduction of the name ApLo-royeiTcou into elegiac verse {supra^ p. 5). After this nothing, till we come to a couplet of Callimachus : 7]fjLL(rV /jLOL "^V^fj^ €TL TO TTVeOV, rjjlLCTV 8' OVK olS' eiT "Ep09 eiV' AlStj^ rjpTraa-e, wXrji^ d(paves. Callimachus had much influence on Roman poets, and this couplet may have had its effect. But it did not cause Roman poets to make elegiac couplets hypermetric. It may have suggested the effect in hexameters. The two parts of the pentameter are set down, as it were, in juxtaposition, with a diaeresis between them, not linked together like the two parts of a hexameter. Their internal structure must next be considered. There the principle of ' caesura ' comes into play. Coincidence between the end of a foot and the end of a word is subject to restrictions, which are as usual more rigid in Latin than in Greek. Meleager [Afith. Pal. v. 165. 2) has agreement of word and foot in the second half of the line {av/jLTrXave, irorvLa vv^), and Theognis has a penta- meter in which word and foot actually agree throughout (456 ; quoted by Christ, Metrik, p. 208) : oi/ro)? tiiaiTep vvv ovSe^b^ d^L09 ei. The Greek line, with its monosyllabic endings, though unusual, is not repellent ; nor are monosyllabic endings in English repellent : In the Pentameter aye | falling in melody back. THE ELEGIAC COUPLET S3 A Latin line, constructed on the plan of the Greek one, e. g. : illic saltern nunc advena plurimus est is scarcely endurable. The explanation may be that in Latin, when a word coincided with a spondee or dactyl or trochee, the accent inevitably coincided with the ic/iis, and, being a stress accent, over-emphasized it. In Greek, accent could -coincide with ic^us without attracting special attention ; in English, accent takes the place of quantity, and the metre is nowhere without it. Apart from the rarity of a monosyllabic ending the Greek pentameter is subject to little restriction regarding the incidence of word-endings. In Latin, on the other hand, there is a gradual elimination of the unfit or survival of the fittest. Elegiac verse was introduced by Ennius, and written by him with some smoothness and finish, as far as the few extant examples enable us to judge. A few pentameters of Lucilius survive (Sa^, xxii) and show no advance in technique. Catullu-s's elegiacs are by no means on a par with his epyllion in neatness and euphony. Some of them are not 'duriusculi ' (as Pliny thought some of his hendecasyllabics), but 'duris- simi ', e. g. (Ixxiii. 6) : quam modo qui me unum atque unicum amicum habuit (me elided : elision in the middle ; two elisions in the second half ; ending in a word of three syllables), ei misero eripuisti omnia nostra bona (two elisions running, one in the middle of the line : three neuters in a in an awkward string). Cornelius Gallus came next, and we may conjecture that he attained a greater degree of smoothness, though Quintilian calls him 'durior' {sc. *Tibullo et Propertio '). In Tibullus — Propertius — Ovid the further progress lies revealed. Tibullus and Propertius still 54 THE ELEGIAC COUPLET have endings in a word of three syllables (though not frequently) : haesura in nostro tela gerit latere. (Tib. i. lo. 14.) Similar to this is an ending in a word of five syllables (' harun- dinibus ') ; a quadrisyllabic ending (' Pierides ') seems to have been somewhat more euphonious, and maintained itself longer. In Ovid a word of two syllables is the rule : exceptions are extremely few. Ovid is careful also to make the last syllable of the line actually long (unlike TibuUus's * latere '), not relying on the syllaba anceps or the following pause. These are im- provements in themselves, though there is a certain monotony in his unvarying correctness. Accent and ictus are separated by the disyllabic ending; but so they are also with 'latere' and 'Pierides'.^ Christ suggests that the 'feminine' or trochaic division of the dactyl allowed the line to flow more easily to its close, as compared with the ' masculine ' caesura in ' gerit | latere '. The arrangement of words is often highly symmetrical and regular, especially in Ovid, but also in the elder poets. One of the commonest effects is that of an adjective at the end of the first half agreeing with a substantive at the end of the second : et dare captivas ad fera vincla manus, or the converse : aspicio patriae tecta relicta meae. {Ex Foiito i. 2. 48 and 50.) But an ending in an adjective is rare, apart from the possessive - adjectives nieus^ tuus, siius. The composer of elegiac verse should be careful to make his line end with a substantive or a verb. The rule applies to hexameter also. It is a principle of symmetry and emphasis that a line should not end with an ^ Unless the accentuation Pierides still survived or could be felt, Zielinski thinks that Ciceronian rhythm affords evidence for it {Clause!- geseiz, p. 233). THE ELEGIAC COUPLET 55 unimportant or otiose word, and this is even more clearly true of the ending of a couplet. In what precedes elegiac verse has been considered as it is written in the best periods of Greece and Rome by consider- able poets. The pentameter was not always preceded by one hexameter. Inscriptions show the practice of substituting for a couplet a group of three lines : TOi^^e TToQ" "EXX-qi^e^ Nlktj^ Kpdr^L epyco 'Aprjo^ Uepaa^ i^eXdaavre^ eXevOepa 'EXXdSt kolvov ISpvcravTO Alo^ /Scofibu 'EXevdepiov. This appears in the Anthology (vi. 50) with a pentameter inter- polated after the first line, ^vToXfico yjrv^TJ^ Xijfxari 7rei66fj.epoL (see Wilamowitz, Sappho und Simonides, p. 197). In later times there were eccentricities both in theory and practice. There were theorists who held that the last syllable of the first half was a syllaba anceps : hoc mihi tam grande munus habere datur. And a late Greek epigrammatist, Philippos, wrote pentameters, Kara aTL\ov, i. e. without any intervening hexameters {Anthol. xiii. i). If there was any abuse of the metre in earlier writers it lay in using it for purposes other than those for which it was originally intended. Callimachus used it for long narratives in his AtTia (though shortly before his time it had seemed to Aristotle incredible that anybody should write a narrative poem in a metre other than the heroic hexameter), and he was followed by Ovid in the Fasti, and by Propertius in some shorter pieces. Callimachus also set the example of using elegiac verse for a lampoon or personal attack in his Ibis, a theme which for long had been claimed by iambic verse. Callimachus's poem may have been short enough to count as an epigram, but Ovid's Ibis is not. CHAPTER IV ANAPAESTS An anapaest is the converse of a dactyl, di'ria-rpo SaKTuXco, ^w-, not -ww. But the rhythm is the same; it is icro9, par (2:2, ww | v_/w).^ It is doubtful, however, whether an anapaest is exactly and in all respects the converse of a dactyl. In a dactyl there is a strong ictus on the first part of the foot, and the first two times always take the form of a long syllable.^ In an anapaest the second part of the foot is normally, but not always, a long syllable, and the ictus may have been less marked. If we represent a dactyl by -^ww, an anapaest may have been Ow-^ (o«^ | Ow). The first part of the foot is often a long syllable; with this and resolution in the second part an anapaest becomes — O w. The slighter ictus perhaps helps to account for a strange fact in Latin verse. In other metres Latin verse shows a great reluctance to put the ictus on the second syllable of a dactylic word : cardifie, vertite seem to have been repellent to the Roman ear.'* But the anapaest that in syllables looks like a dactyl — dvairaKTro^ SaKTvXo€iSi]9 — is freely used : incliite | parva | prodtte | patria. saeptum altisono | cardine templum. 1 Quint, ix. 4. 48 : ' rhythmo indifferens est, dactylicusne ille priores habeat breves an sequentes ; tempus enim solum metitur, ut a sublatione ad positionem idem spatii sit,' 2 There are some rare and rather disputable instances of the resolu- tion of the long syllable of a dactyl in lyric verse. And it is possible that Ennius admitted v^ v^ — or v^ v^ w w for a dactyl even in hexameters. 3 Zielinski; however, thinks that Ciceronian rhythm points to such an effect evyeLS — dyaQol 8' dpiSaKpves dvSpes)- Such proverbs occur in hexameters, in Homer and Hesiod, some- times with only one short syllable at the beginning {iraQoav Se re UT]7rio9 eyuco, beside pe^deu 8e re vrjnLOS eyvco). When an anapaestic dimeter, or line of four anapaests, had been developed to accompany the march of troops or of a chorus,^ the catalectic form of it agreed in appearance with the verse- proverb, but must have had a different rhythm, for, whereas the dactylic half-verse was three feet with a kind of anacrusis (w, — , or wv^), the anapaestic line was one of more than three feet and must have had the time of four. Associated with the efi^aTrjpLa of Tyrtaeus at Sparta, and later used for the entrance and exit of the chorus in the Attic theatre, anapaestic verse was obviously a march-rhythm." The 1 There is evidence for it also as an exhortation or accompaniment for rowers : Serv. on Aen. iii. 128 ' Cretam proavosque petamus : celeuma dicunt, et bene metro celeumatis usus est, id est anapaestico trimetro hypercatalecto'. If this nautical use is ancient, we can see why Cratinus wrote the verse Kara utIxov for a chorus of seamen in his 'OZvaaris (ir. I44y . P^'^jj j^yj,^ ^^jjy ^'^g oi-yav, KOI -navra Ko'^ov tolxo- iT(.voii' rj^xiv 'WaKTj TTarph kariv, nXiopxv 5' iiyL 'Ohvaaii 9tia'. 2 Like the iambic trimeter anapaests could also halt or limp. A 58 ANAPAESTS short syllables accompany the raising of the foot (dpai^), and with the long syllable it is set down again (Oiai^). Two m anapaests accompany two steps, together they make a Pdais ' (the Roman passus, five feet as a measure of length). Two anapaests make a fierpov, four are called a 'Dimeter'; they are scanned Kara SiTroStav like iambi or trochees, in spite of their greater length. This being the nature of the anapaest, it seems clear that it is and was distinctly felt to be an ' ascending ' rhythm ; we must not follow J. H. H. Schmidt in applying to it the methods of modern music and scanning anapaests as dactyls with anacrusis (v^w: — '^Vw' I — \j \j — WW j — ^y The name dvaTraLcrros implies this, and it is confirmed by the regularity with which a word ends with the end of the second foot: ^ ^ ^ ^ i SiOpovov AioBeu I Kai StcrK-qTrrpov Tifxfj^ 6\vpov I ^eiJyo? 'ATpeiSdu. The same question has been raised about iambic verse, which Schmidt treated as trochaic with anacrusis. About both the testimony of antiquity is that the rising movement gave an effect of greater vigour and energy — the downward dactylic and trochaic movements one of greater fluency and facility — ^ ' cholanapaestic ' metre is found, on stone, in lines addressed by a gouty patient to Asclepius : rdSf aol Aiocpavros (Trfvxofiar aaiauv fxf, fiditap adevapuTaTf. laadfKvos woddypav Kan-qv. (Kaibel's Supplementimi Epigraniniaium Graecorum, in Rhein. Mus. 34. 210.) Wilamowitz pointed out the appropriateness of the movement of the verse, suggestive of a hobbling gait. When Diophantus recovers, he returns thanks in dactylic hexameters. The verse consists of three anapaests and an iambus, a metre which requires further investigation. It is found (with accent taking the place of quantity) in early Byzantine hymns. But it is also as early as Sappho l^y\vKvm/epov d^dxavov opnfTov), ANAPAESTS 59 and in the case of iambi Quintilian expressly says that this effect was felt throughout the line, not merely at the beginning of it. It must have been even more clearly so in anapaests. In a hexameter or iambic trimeter the rhythm may be said to be reversed at the caesura ; in an anapaestic dimeter the initial effect was definitely repeated or renewed in the middle of the line. Sometimes it is further emphasized by the con- currence of a rhetorical effect, repetition or anaphora {vvv kwaKovcrov, vvv kirdpri^ov. Aes. C/w. 725). The chief forms of anapaestic verse are the following : (a) The early ifi^arrjpLa of Tyrtaeus, in which the catalectic form of line is repeated : ayer' o) ^irdpTas evduSpov KovpoL waTepcov iroXLardv, Xaia fikv LTVP TTpo^aXiade, 86pv Se^Ljepa 8' ^vtoXjjlcds. That is ^^- ] ^^- I i=^^_ I - or ^^- I ^^- | ^^- | »-» or w^ _ I i^^ _ j v.^ _ I _ ^ or i=^^ _^^_^^u _. The third foot may be a spondee (which it rarely is in the drama), and the long syllable of the anapaest is not resolved. (b) The acatalectic dimeter of tragedy and comedy (with the catalectic line at the close of a group of lines) used mainly for the H(To8os of the chorus. This is the most important and most familiar form of anapaestic verse ; it will be considered more fully. (c) In the Greek drama dimeters of a freer and more lyrical cast, in Oprjuot and other emotional passages, sometimes using frequent spondees to express distress or perplexity {/on 859 f. CO yj/v\d, TTcos (Ttydaci) ;) and sometimes admitting complete resolution in passages of greater excitement or surprise, either once in the line or twice or even in all three feet save the last : T19 opea 1 PaOvKOfia | Td8' e7reav\T0 PpoTOiv ',^ * A resolved line of this type served the metricians as an example of 6o ANAPAESTS Similarly resolved anapaests are found on a larger scale in the VTTO p)(rj /la of Pratinas, i-4(\Vilamowitz, Sappho und Simonides, p. 132 f.). (d) A longer line, tetrameter or ApL(rTopa^, a group of syllables that make an ' Adonius ', which closed a sapphic stanza. To the Romans the composition of i\napaestic verse pre- sented special difficulties, and their practice is consequently different from that of the Greeks. In early Roman drama the proceleusmaticus (v^ w vj:- w) is freely admitted, not only in comedy, but also in tragedy ; and the diaeresis in the middle of the dimeter is less strictly observed. Greek anapaests not only had diaeresis in the middle of the line, but quite frequently ended each anapaest with the end of a word ; in Latin such an arrangement meant divergence throughout of ictus and accent : incliite parva prodite patria.' ^ Compare lines of Anaxandrides (quoted by Lindsay, Capiivi, p. 77) : (pVCTKUIV, ^OJfXOV, T€VT\uJV, OpiUJV, 9vvvi5€S oiTTai, (pvKida £({>9ai. 64 ANAPAESTS Pacuvius is credited with a deliberate attempt to write anapaests on a new principle, exemplified in the lines : agite, ite, evol vita, rapite, coma tractate per as|pera saxa et humum. Here there is more agreement between accent and ictus, facili- tated by the abolition of the diaeresis. Marius Victorinus^ who mentions the experiment, of course gives no hint that agree- ment with accent was arrived at (' Pacuvius novare propositum volens noluit intra binos pedes finire sensum '), nor do we know whether Pacuvius persisted in his experiment or tried it on a large scale. It could not succeed or establish itself. The result was too unlike the Greek norm ; and to compose in this vein must have been extremely difficult. To write verse that observes two different principles is a tour de force which may be difficult in the extreme, e.g. to write hexameters in English that observe both accent and quantity. In Latin anapaests complete success was impossible. Pacuvius has to end his lines with 'coma, humiim'. To achieve coincidence of accent and ictus the line must either end with a word of one syllable, which no doubt it might occasionally do, as in Accius's delubra tenes, mysteria quae, or with one of a very limited number of quite exceptional words such as 'nostras', 'tanton', 'abit' (where the abnormal accent is due to loss of a syllable). It is notable that Terence never uses anapaests in his few cantica, although they are the favourite metre in the cantica of Plautus. With a catalectic ending the conditions are reversed. Here the agreement of ictus and accent was a thing not difficult to (The exact coincidence of accent and ictus here is of course quite acci- dental. Accent diverged from tcius just as readily : Accent has no place in Greek metre till Babrius, and in Babrius it is regarded onl}' in one syllable of the line.) I ANAPAESTS 65 attain, but difficult to avoid. So it is not surprising to find Plautus writing catalectic dimeters continuously {Kara cttlxov or iugiter) : defessus siim pultando. hoc postremiimst. vae vdbis, &c. {S^ic/ius ^i^f.) Plautus also uses the catalectic line as a concluding line or dausicla after a group of acatalectic dimeters, as do the tragic poets : Accius, Philocieta : unde igni' cluet mortalibu' clam divisus : eiim dictii' Prometheus clepsisse dol6 poena sque lovi fato ^xpendisse supremo.^ This ending is found also in the Satiirae of Varro {Koa-fio- Topvi'Tj) : dum nos ventus flamine sudo suavem ad patriam perducat. But it is not consistently and invariably used by the early poets, and it was entirely abandoned by Seneca. Plautus, besides writing catalectic dimeters, frequently makes his longer lines (tetrameters) catalectic. One or two lines are catalectic in a passage otherwise acatalectic : Persa 783-4 : qui illiim Persam atque omnes Persas atque etiam omnes personas male di 6mnes perdant, ita miser6 Toxilus haec mihi concivit. It is difficult to see why Plautus did not make all his long lines of this type. Aristophanes had done so, without any inducement in the shape of coincident accent and ictus. And the same question arises about iambic verse. The catalectic form (* septenarius ') makes coincidence in the penultimate ' In the second line ' dictus Pr6metheus ' shows that the incidence of the ictus did not enforce length by position. It was when the ictus fell on a vowel originally long hy nature that the long quantity was revived or protected (' mane mane' begins a trochaic line). 66 ANAPAESTS syllable extremely eas^ of attainment and all but invariable. Yet Plautus often writes the complete ' octonarius ', without having much precedent for it in Greek comedy as far as we know (though the 'I^vevrat of Sophocles has revealed a passage of considerable length in satyric drama). In the Roman drama anapaests are not associated with march- ing or with the ela-oSos or 'i^oSos of a chorus. They may happen to be used when a personage is entering, as in Trin. 840 f., but that does not make them ' march-anapaests ' (Lindsay, Captivi^ p. 80). There was no regular chorus, and the 'orchestra' was occupied by the seats of senators. The fishermen in the Rudens enter with iambic septenarii. Anapaestic verse could be used as a lyric anywhere in the play. Many of the more complex forms of Greek lyric were unworkable in Latin, and anapaests were one of the simpler metres that took their place. Seneca's practice would have seemed very strange to the Greeks of earlier times. His treatment of anapaests is in keeping with what he does in the case of other metres. He sometimes writes a long passage in Sapphics, without an Adonius anywhere ; sometimes a long passage in Asclepia- deans without a shorter closing Hne (Glyconeus). So in anapaests his last line is often a dimeter like the rest. He does, however, show some tendency to use a monometer as a closing line, often in the form — ^ v^ : quid plura canam ? vincit saevas cura novercas. {Phaedra 357.) Or the second dipody of the last line takes that shape : nee sit terris ultima Thule. {Med. 379.) In the Agamemnofi (310 f.) he has dimeter and monometer alternating, a form unknown to Greek tragedy. In linking his lines together Seneca is less careful than the Greek dramatists. Hiatus and syllaba aticeps occur sometimes, without being justified by the close of a period or by a marked ANAPAESTS 67 pause in the sense. But the instances are few, and some of them have been removed by emendation. They are notably more frequent in the Octavia than in the other plays, and this difference in metrical practice is one of the fairly numerous and considerable reasons for assigning that play to a different author. The author has also a tendency to make a monometer end in a short syllable (' pulsata freta', 316, 'fletibus ora*, 330), and he has an unusual lengthening before a mute and a liquid in ' plura referre prohibet praesens ' (890). F 2 CHAPTER V IAMBIC VERSE I. The Iambic Trimeter and Senarius * Trimeter ' and * Senarius ' are not merely different words for the same thing or words of precisely the same meaning. The former designates a handling of the metre that was distinctively Greek, though practised also by later Roman poets : ' senarius ' is the more appropriate word for the heavier and more amorphous type of line which prevailed at Rome before the Augustan age. In both languages there were several distinct veins of iambic writing and types of iambic verse, each with principles and restrictions of its own. Quintilian {Insi. Or. ix. 4. 139 f.), speaking of the variation of rhythm in oratory to answer to various emotions, illustrates this by iambic verses : sometimes ■ we aim at the ' tumor tragoediae ', which depends mainly on iambi and spondees ; the iambus with its rising movement, he would say, gives an effect of imperious energy, and the spondee of course lends solemnity to the utterance : en Impero Argis, sceptra mihi liquit Pelops (words of Atreus, perhaps from a play of Accius) ; the verse of comedy, he proceeds, gains 'celeritas' at the expense of ' gravitas ', it admits ' trochees ' (by which he means what we are accustomed to call ' tribrachs ') and pyrrhics : quid igitur faciam ? non earn ne nunc quidem : J IAMBIC TRIMETER AND SENARIUS 69 further, ' aspera et maledica ', when expressed in verse, take the shape of iambi : quis hoc potest videre, quis potest pati, nisi impudicus et vorax et aleo ? Quintilian is here quoting, from Catullus (xxix), lines which admit no spondee and consist of pure iambi throughout. This was a refinement on the abusive verse of Archilochus. The effect was one of measured and dexterous ' celeritas ', like the flashing of a rapier in a skilful hand or the flight of a well- aimed missile.^ Iambic verse has no early history. It comes into view suddenly in the middle of the seventh century b. c, written in a finished and exact form by Archilochus of Paros, to whom notable advances and innovations are attributed in several different kinds of metre. It existed before, no doubt, probably in a ruder and more popular form, with the licences which are later seen in comedy. It belongs clearly to the lonians, and it had some place in the popular festivals of Demeter and lakchos, which flourished among the lonians and their kindred of Attica — festivals connected with the rise of the Attic drama. The study of iambic verse begins most conveniendy with the most regular form of it that had a wide vogue— Catullus's absolutely pure iambi we may postpone, as an exceptional and rather artificial or fanciful thing — the trimeter of Archilochus and the tragedians. The practice of Archilochus differs so slightly from that of tragedy that he need not be separately treated. It was his verse that tragedy adopted, XeKTiKcoTarou rcou ixirpGdv — 'aiternis aptum sermonibus ' (Hor. A. F. 81) — as ^ Quintilian's phrase about Archilochus, ' breves vibrantesque senten- tiae' [yi. i. 60), compares his thouglits or meanings to the hurthng of a spear through the air, or perhaps to a thunderbolt (compare the saying of Cicero, which QuintiHan quotes [ix. 4. 55] : ' Demosthenis non tanto- pere vibrarent ilia fulmina,' &c.). I 70 IAMBIC TRIMETER nearer to real life or actual debate than the elaborate and stately hexameter. There are various indications that the natural texture or tendency both of Greek and Latin was iambo-trochaic (alternation of short and long) rather than dactylic or anapaestic, though both languages lent themselves readily enough to the latter forms of verse also.^ i The iambus is a foot of three times (^ w >u), its rhythm duplex or ScTrXda-ios (the ratio of the parts 2 : i, or in actual ^ order 1 + 2). The second and third times go together, taking normally the shape of a long syllable {u —) ; when they do not do so the ictus is on the second syllable of the foot (w | o ^). It is a rising foot ; the slighter or weaker part precedes the , stronger. But though the ancients recognized and often ^ discuss the single iambus, Greek composition in iambics dealt not with single iambi, but with pairs or dipodies (whence the name trimeter ^ for a line containing six feet) : w — ^ — . Of J the four syllables in such a group there was one that might be heavier without detriment to the general effect — the first. To make the third long would efface the structure of the group : \^ is amorphous. The iambic line consisted of three I ^ Cicero {Orator 189) : ' versus saepe in oratione per imprudentiam. dicimus . . . senarios vero et Hipponacteos effugere vix possumus ; magnam enim partem ex iambis nostra constat oratio ' (where Cicero would have ended with the fault he is describing if he had written, ' partem ex iambis constat nostra oratio '. He shuns that, and gives us instead, as in the previous clause, the favourite prose ending of a double Cretic < -/w y^—) : ib. § 191, he speaks of the view that iambic rhythm belongs specially to oratory, 'quod sit orationi simillimus' ; it therefore belongs to the drama too : ' qua de causa fieri ut is potissimum propter similitudinem veritatis (= reality, real life) adhibeatur in fabulis, quod ille dactylicus numerushexametrorum magniloquentiae sit accommodatior.' ^ Terentianus : sed ter feritur ; hinc trimetrus dicitur, scandendo binos quod pedes coniungimus. ('feritur', is beaten ; each of three groups of syllables has its dominant ictus,) » AND-SENARIUS 71 such dipodies, and it followed that the first, fifth, and ninth syllables might be long : For feet of three times, the longest possible kooXou, according to Aristoxenus, was one of six feet or eighteen times. The trimeter then may be one kcoXou ; but in fact and in practice the trimeter insisted on being divided, in very mu^h the same way as the hexameter. It resented equal division ; like a hexameter it refused to be 2 + 2 + 2 or 3 + 3 when it was embodied in words : a line like f dfir)^dv(£)V I d\yr)86v(ov j kTTi(TTpo(f)aL is not admitted in tragedy, and Aeschylus's line (if he wrote it in this shape) Sp^Krjp 7repdcrauT€9 \ fioyis noXXco wovco may be a case of sound deliberately made to answer to sense : The line too labours and the words move slow. When elision occurred the division was tolerated : S) Trdcra KdS/iov yaV, dnoXXyfiat SoXo), and of course the end of a monosyllabic word might occur in the middle of the line if the word was proclitic and not enclitic : ov fjLOL TO, Tvyeco tov iroXvy^pva-ov fieXec is a perfectly euphonious line. Exceptionally, also, a break in the middle of the line might emphasize a contrast : drifjLLa^ jJlIv ov, irpop-ijOLa^ Se aov. {El. 1036.) The trimeter then, like the hexameter, prefers to be 2\ + 3^ or 3-|+ 2 J. Often, of course, as in a hexameter like Peliaco quondam prognatae vertice pinus or caerula verrentes abiegnis aequora palmis, 72 IAMBIC TRIMETER the division is very slight, and exists rather for metre than for the sense or for rhetoric. It was possible for a comic poet to make a trimeter consist of one long compound word {Wasps 220) : dp^aLOiJLe\L(n8(ovo(ppvvL\rjpaTa, though we may perhaps suppose a very slight division after (tlScopo. The analogy between these divisions and those of the hexameter is so close that it seems necessary to call them caesurae. If they are caesurae^ the fact is of some importance for recent metrical methods and discussions. The syllables of an iambic trimeter admit of being construed as trochees with afiacmsis : (In a trochaic dipody, which is the converse of an iambic one^ it is \}c\^ fourth syllable that is anceps or permitted to be long.) This method of scansion, which is in agreement with modern musical notation, was adopted by J. H. H. Schmidt and others in the nineteenth century, and — for lyrics at all events — the principle has had considerable currency in this country, since it was set forth by Jebb in the introductions to his editions of plays of Sophocles. It has advantages. It may be contended that it gives a better and clearer account of certain forms of verse than does iambic scansion. As regards the trimeter of the drama— verse that is spoken or recited rather than sung — this matter of the caesura is perhaps the most tangible objection that can be brought against it. With trochaic scan- sion it is difficult to see that there is any cumbrousness or heaviness or rupture into two parts in the line of Aeschylus quoted above : Spr\\Kr\v 7r€pda-ai^\T€S [JLoyLS 7ro\\Xa> tto^O). The normal divisions of the Trimeter would have to be called diaereses and not caesurae, on the trochaic theory. Further, AND SENARIUS 73 J) the difference in tj6o9, in feeling or effect, between trochaic and iambic verse is discussed so often and so conspicuously by the ancients that it is difficult to believe that the difference consisted only in the presence or absence of one syllable at the beginning.^ We have seen above that the structure of anapaests is far from suggesting obviously that they should be scanned as dactyls with anacrusis ; and it may be added that the anacrusis is no more plausible in the case of dochmii (v^ v_y — ), (f)pevS>v 8va-(f)p6v(ov afxaprrjiiaTa is not readily thought of as (Ppe\v(Jov 8v(T(pp6v(£)V d\fj.apTrjfj.ara. The essential structure of the tragic trimeter has now been explained. If we speak of it in terms of feet, we may say that only the first, third, and fifth may be heavy or spondaic. There must be a caesura in the third foot or in the fourth. Subordi- nate, but still quite important principles of structure remain. A clear understanding of the nature of the verse was one of the contributions made by English scholars to classical research. It was made in the beginning of last century. One of the rules is still known by the name of Porson (Lex Porsoni"). (i) Resolution. Unlike the long syllable of the dactyl in a hexameter, resolution of which is found only in two or three dubious lines of Ennius {siipra^ p. 56), the long syllable of an iambus could be replaced by two short syllables. This was done very sparingly by Archilochus, not more than once in a line : KJ \^ KXaLCt) I Toc Saai^odv ov to, MayvrjT(i)v KaKoi. ^ There is a passage in Quintilian (ix. 4. 136) which seems to show that when a series of syllables occurred which could be analysed as iambi {y — w — w — ....), the 'rising' elTect was not felt only at the outset ('iambi . .. omnibus pedib us insurgunt et a brevibus in longas nituntur et crescunt '). The believer in trochaic scansion should read the iambic tetrameters in the 'Ixv^vrai of Sophocles (291 f.) with an open mind, or a mind as unprejudiced as he can make it. He will not feel them to be merely trochaic tetrameters with a syllable prefixed. 74 IAMBIC TRIMETER Tragedy at first was similarly strict, but gradually assumed greater freedom. When Euripides brought it down from heroic heights to a level nearer that of everyday life, its vocabulary and metre underwent a corresponding change. In metre resolution became more frequent. Euripides has . resolution as often as four times in a line : irevla B\ was that the two short syllables which stood for the long should be in the same word (w | w -^ , not w^ | v^), as in Tcc ©a(Ti(a)i^), or in Aeschylus : ^KvBrjv ey oI/jl\ou, dpa\TOv els kprjfiLav. ^ A line in the Frogs appears to be an exception : Koi KoiSdpiov teal XrjKvdiov fcal GvXaKiov (1. 1203). The line would be normal if the first syllable of OvXclkiov could be supposed to be short. But all the earlier evidence makes it long, and the only instance of its shortness given by L. and S. is in a late epigram in the Anthology {Atith. Pal. viii. 166), vv^hich has no metrical authority. Further, the context demands a word exactly equi- valent in its syllables to Kwhapiov and \rjKv6iou. The deviation from rule is similar to that which is sometimes necessitated by a proper name. AND SENARIUS 75 In comedy the rule is not observed, though it is not broken with great frequency : €fi€ fiev (TV 7ro\\ov9 Tov I Trarep' e\Xavu€i9 8p6fjLOV9. (Arist. Clouds 29.) (ii) An anapaest was admitted by tragedy only in the first foot {TTorafioov re nriyai. . .), and that under the condition that T0L9 ovpavov is so exceptional — in tragedy — that doubts arise about the text. Comedy did not observe the rule^ and it was entirely disregarded by Roman poets, even when they wrote iambic verse with the greatest strictness and finish, in the Augustan age and later. The Porsonian canon has hitherto been a mere 'empirical fact '. The practice of the Greek tragic poets commends itself to the ear as euphonious, but no definite cause for it was suggested by Porson himself, or for long after him. Recently a promising attempt to assign a cause has been made by K. Witte {Hermes, 1914, p. 229 f.). His theory may be summarized as follows : — The iambic trimeter, like the heroic hexameter, fell naturally into two parts, and the 76 IAMBIC TRIMETER ■« favourite and most frequent division of it was the penthe- mimeral : a b Further, the effect of this division was heightened, with extreme frequency, by making the syllable before it long. If a precisely similar effect followed at b, four syllables later, the division at a would no longer be dominant. Hence if a line begins it may be completed by the words KevyeveLOLV kKirpeirus, where the -av is short, but a quadrisyllable word, or group of words, with the last syllable long may not follow the caesura. Similarly (though here the argument becomes somewhat less convincing) with a hephthemimeral division : "EW-qves ov fievouv, dXXa (riXfiaariu, the hephthemimeral division (which mt/s^ be preceded by a short syllable) may not be followed by a similar division which would obscure or eclipse it, i. e. not by a similar division with the added emphasis of a long syllable before it. That is a brief statement of Witte's principle, and it appears • to be sound, at all events as regards the dominant penthe- mimeral caesura, and the rule once evolved may have been extended to lines with the less frequent hephthemimeral. In a line like v^ — — I — w non ut superni villa candens Tusculi there are two conspicuous groups of syllables arranged in a way that rather distorts or destroys a trimeter, a metrical effect belonging to lyrics and to a wholly different type of verse. v^ — — UvQcovL T av^rjs ovpov vjxvoov. (Find. Pyth. iv.) Witte points out that, in the rare violations in Greek tragedy of Porson's Law, the last five syllables of the line sometimes y^ AND SENARIUS 77 go very closely together in sense, or the words are bound together by elision ^ : d /JLOL TrpooreXdcov criya arj^Laiv eLT ex^^- (Soph. Phil. 22.) Now elision at this point — before the final cretic — is distinctly preferred by Seneca. The effect of * villa candens Tusculi ' did not repel Horace, but it did repel him. In the great majority of the lines in Seneca that end in a word of three syllables, there is elision before that word. If we next come back to Horace we find that in Epode xvii, a poem of a dramatic cast rather than Archilochian iambi, the only violation of Porson's rule is ' homicidam Hectorem '. Tragedy, as it had an exalted, Ionic and poetic, diction of its own, was also stately and regular in metre. The comic poet can make an effective use of the difference. Thus in the Peace Hermes, being a god, naturally speaks the language of tragedy. When Trygaeus knocks at the gate of Olympus, he begins : TTodeu ISpoTov fi€ TTpocre^aX* — (^poTov, for 6vr]Tov^ is tragic, and we may suppose that the sentence was to end with words like i^aicpur]^ (pan?). But he is startled by the apparition of the gigantic beetle, and he goes on : wi^a^ ^HpaKXeL^, TOVTL TL k(TTL TO KaKOV ) TP. LTTTTOKdudapO^. ' Guva^ 'HpaKXeis ' and the form tovtl belong to everyday colloquial Attic. The former is also metrically untragic. The ^ Witte proposes to deal with the first line of the Ion by writing it : ArXas 6 xakKeoiai vuroia ovpavov, with elision before the cretic. But this seems doubtful in Attic tragedy: it would be more plausible in Archilochus, in verse written in actual Ionic. 78 IAMBIC TRIMETER resolution in tl €(ttl to Ka\K6v ; is not irregular or not glaringly irregular, for the article to goes closely with KaKOV. The trimeter of x\ttic comedy enjoyed a large measure of freedom. Some of its features have already been mentioned. But it remains a very different thing from the * senarius ' of the early Roman drama. Most notably, it never admits a spondee in the second or fourth foot. It might be supposed that the admission of an anapaest would carry with it the admission of a spondee, but this is not the case. Further, the anapaest is not admitted indiscrimi- nately and in any shape. It is admitted most readily — with great readiness — when all three syllables are in one word {Kard^a, Kard^a ktX.). A long word has a longer pause after it than a short word, and if a long word ended with one of the short syllables the foot was broken or strained asunder becoming less suitable to take a place among iambi. The effect is therefore infrequent, and it has been observed that in Greek it occurs within a dipody, not between two dipodies : eTrtVlAfOTToy 7j\\Ka>. 8€vp\d TrdXjp \\ /SaSicrrioy. The Latin senarius is stricter in this respect, and it has been suggested that it was so because it ignored dipodic structure ; it had no quadrisyllable group within which a wrongly divided anapaest could find shelter. Hence the line [Asinariaj Prol. ii) Demophilu' scripsit, Maccu' vortit barbare may be regarded as proof that, though Plautus's full and formal name, at all events in his later life, was T. Maccius Plautus, he was also known as Maccus. ' Macciu' | vortit ' divides an anapaest wrongly. The comic trimeter could divide itself in ways that were AND SENARIUS 79 ft excluded from tragedy, even emphasizing a strange division by change of speaker {Birds 175) : IT. pxk-^ov KOiToo. EII. Kal Stj /SXeTTO). n. pxine vvv duoo. The ' proceleusmaticus ' ('^ w w w), arrived at by resolution of an anapaest, is very rare in Greek comedy, much rarer than in Latin, but there are undoubted instances of it. A good example is a line of the comic poet Plato : ovTos, TL9 ei; \ Aeye rayy' \ tl (Tiyas ] ovk epei^. The trimeter of the satyric drama calls for very brief notice. It may be said to be intermediate between the tragic and the comic trimeter, but not in the sense that it is half-way between them. It is nearer the tragic form of verse — notably so in the recently recovered *I\vevTaL of Sophocles — and differs from it chiefly in admitting the anapaest more freely, in places other than the first. Anapaests are fairly frequent in the Cyclops. A fragment of the ' npo/irjOev^ UvpKaev^ ' of Aeschylus : Xii/d Se Trtv€S Koi TreTrXeKTavrjfjievai, {Cho. I049O and in comedy : d^v' aKovo), (Ar. Fax 663.) and two instances found by Wilamowitz in Simonides, one with the choriambus in the middle of the line : a dr] ^OTci \ ^oio/xev ov\8ev elBoTes (1. 4 in the passage beginning a ncu, reXos p-h Zevs : Sappho u?id Simonides, p. 271 and p. 153). These passages form a rather slender basis for the view that choriambic substitution was an important and characteristic feature of the metrical art of the lonians. Milton may have had the passages of Aeschylus in view when he freely admitted a choriambic effect in English : Purples the East — Better to reign in Hell — AND SENARIUS 87 It is fairly common in English verse : Dangerous secrets, for he tempts our powers. (Shelley.) Ready to spring ; waiting a chance ; for this. (Tennyson.) Is it that in a language in which accent comes into play it matters less which of two syllables has the ichis or stress, so that a trochee can take the place of an iambus ? Such a possibility would perhaps account for some of the perplexing varieties of the Saturnian measure in Italy, when, for example, ' Hercoli ' takes the place of * Metelli ' ('donum danunt Herculi— dabunt malum Metelli '). {b) The Roman disregard for ^Porson^s canon\ — No certain explanation has been given of the divergence between Greek and Roman verse in this respect. We may perhaps suppose that the facts are something like these : — The observation of the rule by the Greek tragic poets was instinctive, unconscious or half- conscious, the result of an innate sense for euphony and propriety of rhythm. It was not formulated as a rule of composition, and did not come to the Romans as a precept. Their own early verse differed widely from Greek verse in many ways, so that a subtlety like this was not thought of, and when they began to write iambic verse on more strictly Greek lines, they were still influenced by the heavy and spondaic character of their early verse, and they did not study Greek metrical practice with the minute scrutiny bestowed upon it in a later age by Elmsley and Porson. Further, there was perhaps in Latin a special reason for liking an ending in a word of three syllables ; accent and ictus agreed, there was notable coincidence in the fifth ictus of the line, and also in earlier places (' villa candens Tusculi '). The Porsonian rule would have fettered the composer, so the conditions regarding a trisyllabic word undergo some modification ; they are relaxed to facilitate such endings. When Livius Andronicus set himself to produce Greek plays at Rome in the metres of the Greeks, he might conceivably have laid upon himself the rule that no line may begin or end with a word of two syllables. But it would not occur to him to do that. It would have been a highly artificial restriction, and in words like probant, domos the quantity was sufficiently conspicuous to make them workable as iambi. They had to be admitted, but lines SS IAMBIC TRIMETER ending in words of three or more syllables were preferred. Of the extant senarii of Ennius about 34 per cent, end in words of two syllables, of those of Accius about 38 per cent.'^ Of the first 100 lines of the Oed. Tyr. of Sophocles 62 end in words of two syllables ; of the first 100 lines of the Philoctetes^ 58. If we now turn to Seneca a remarkable result presents itself: of the first 100 lines of the Hercules 84 end in disyllables ; of the first 100 lines of the Thyestes, ^J. The dislike of a disyllabic ending has been replaced by a preference for it ! But, of course, no Porsonic rule is enacted. There is, on the contrary, a preference for its violation. The Romans had always liked spondaic effects in tragedy, and the liking now takes what a Greek would call a comparatively legitimate shape. It is an all but rigid rule that the fifth foot must be a spondee, and this precluded the observation of the Porsonian canon. {c) The 7iature and history of the old spondaic senarius at Rome. — It is in general extremely difficult to compose verse that is regulated by two different principles, such as quantity and accent, at the same time. But when Andronicus and Naevius saw their way to beginning and ending a line with a word of two syllables — when they found that the effect was not repellent to the ear — they had surmounted the chief difficulties.^ In the middle of the line it was not very difficult to keep quantity and accent in agreement. Given the regular incidence of the Latin accent and the normal caesura of an iambic line, agreement resulted almost mechanically : ibis Liburnis | inter alta navium. The Latin accent was not a strong enough stress to enable a short syllable to take the place of a long, but when two or three longs ^ I count those lines in Ribbeck's text which are fairly certain, neither formed by conjecture, nor possibly parts of trochaic tetrameters : 108 lines of Ennius, 284 of Accius. * Some words of two syllables had perhaps, in certain positions, no appreciable accent at all. They were of the nature of enclitics in phrases like * operam dabit * or * voluptas mea '. These belong to colloquial Latin, and it may be due to their frequency in comedy that Plautus more often ends a line with a word of two syllables than the tragic poets do. AND SENARIUS 89 came together it differentiated them. Thus an iambic effect could be obtained without the quantitative strictness of the Greek trimeter. In shaping the verse as they did Andronicus and Naevius undoubtedly created something which suited the genius of the Latin language. Barbaric looseness and irregularity is the impression which this verse gives when we first turn to it from the verse of the Greeks. But the impression is largely a mistaken one. The verse is an artistic thing, and has its own laws (compare what is said above, p. 82, regarding 'abduxi negotiis' — the com- pensation provided).^ The proof that it suited the Latin language lies in the fact that the admission of spondees in the second and fourth places was a thing which gained ground. It did not gradually die out. This can be roughly tested by counting in the three great tragedians the number of lines which- satisfy Greek rule— which have a pure iambus both in the second and in the fourth place. We may call these ' trimeters ' and the other Hnes ' senarii '. In Ennius the proportion is about 1:3 (i. e. one line in four is a 'trimeter'). I have surveyed 96 lines, practically all that can usefully be taken into account : 70 are senarii, 26 trimeters ; but if we allow final s to make quantity by position, only 22 are trimeters. In Pacuvius the proportion is about i : 4^. In Accius it is about i : 6| (273 lines surveyed : 37 trimeters, or 33 if -s is allowed to lengthen a syllable). Accius survived into the first century B. c, and in that century a revolt or new movement took place. Catullus writes not only ' trimeters ', but also the extremely strict and artificial form of iambic verse which admits no spondees at all. What, then, is Cicero's position ? Cicero had a great admiration for the old poets, and could imitate their style admirably, as we have seen. But he was not unaffected by the new movements, and he knew the Greek poets very well. His ^ Terentianus says that comic poets aimed at a resemblance to real speech : ut quae loquuntur sumpta de vita putes, they were deliberate artists in their handling of verse : in metra peccant arte, non inscitia, and he adds that Roman poets went further in this direction than Greek : magis ista nostri. 90 IAMBIC TRIMETER imitations of Pacuvius and Accius are akin to Horace's line (see infra, p. 233, footnote) : alitibus atque canibiis homicidam Hectorem. The effect is frequently obtained withoict spondees in the second or fourth places. Of in extant iambic lines of Cicero 35 are trimeters, i. e. the ratio in his case approaches i : 3. Whether Pollio wrote 'trimeters' only is perhaps doubtful, but it is highly probable that Varius and Ovid did. It was not only in this matter of the admission of spondees that the verse of the early drama was in agreement with the genius of the Latin language and the nature of Latin accentuation and pronunciation. In its prosody, as well as in metre, it differed widely from the verse of the Augustan age. It shortened many syllables which were long by nature or position, and which were always long for the hellenizing metricians. For them vohtptas and domi are always a bacchius and an iambus, <^ and ^ — . But the reader of Plautus and Terence will again and again find such syllables shortened— treated as short for the metre, that is, though whether the second syllable of voluptas was ever in fact exactly as short as the first is another question. Most of these shortenings come under a rule or principle known as the 'Law of Breves Breviantes'. There is not precise agreement about its formulation, but the essence of it is that before an accented syllable the second syllable of an iambus is shortened ; as in voluptdtein inesse tantum^ do?m mdnsit} These shortenings, it should be ^ For a full statement of facts and details the reader should consult Professor Lindsay's account of Plautine verse in the Introduction to his larger edition of the Captivi. ' Brevis brevians' was a name given to the first syllable of an iambus, which was supposed to shorten the syllable that followed. Its shortness was clearly a contributory cause : 'frlgefacio' does not shorten its e, but 'calefacio* does ; 'mo]6stc3rum' is admissible in Plautine verse, but the second syllable oi funestus cannot be shortened. The accent that was the other contributory cause was the accent on the third syllable of the group, or the accent on the first, or both. The result is found without an accent on the third syllable, as in Terence's line : ex Graecis bonis Latinas fecit non bonas. AND SENARIUS 91 added, do not belong only to iambic verse, but are found in the verse of the drama generally, e.g. in anapaests : nobis datiir bona | paiisa loquendi. (Accius.) It cannot be doubted that these phenomena in prosody answer to the actual pronunciation of Latin in the living speech of the time. They were not ' licences ', or things done inetri gratia— z.w explanation which is seldom really tenable, for all but the most incompetent poets could easily have written a line or a passage quite differently if they had chosen. The old poets were neither metrical blunderers nor deliberate and laborious metricians. They wrote fluent verse by the ear, verse which was to be delivered in a theatre, and have its rhythm for the audience (whether that audience knew anything of metre or not). The feeling for this old verse was perhaps beginning to be impaired when Cicero wrote that some of the senarii of the comic poets are so abiecti or debased that no ?tumerus can be found in them. But what seems to prove conclusively that the shortenings answered to real pronunciation is the fact that the principle operates upon the form of Latin words. Only something that existed in current speech could do that. In some words of iambic form the last syllable has become ambiguous, short, but capable of resuming its length when the ictus of verse fell upon it ('mihi,tibi, sibi : ibi,ubi: vide, cave' — so that 'cavg' can find admission to the hexameter of Horatian satire) ; in others the second syllable definitely ceased to be long ('bene, male: modo' when it is an adverb). In compounds 'patefacio, calefacio' supplant * patefacio, calefacio'. {d) The length of the first syllable in an impure or heavy iambus. — The iambic dipody w— can hardly have consisted of two parts in the ratio of 4 : 3, making a total of seven times (y\u w w I \j w w). Aristoxenus speaks of a syllable that is oKoyo^^ and a heavy trochee seems to have been called x'^P^^^^^ aXoyos. The syllable was not a simple multiple of the short syllable or of the XP°^°^ Trpcoros, and the ratio of the two parts of the foot was fiera^v dvolv \6yoiv yvMpijjLOLv (2 : 2 and 2 : I being yviopnioi \6yoL). What it was has been a subject of much discussion. The various views are collected and summarized by Mr. White {Verse of Greek Comedy, p. 5, § 16). Voss and Lehrs did give the 92 IAMBIC TRIMETER syllable the time of a normal long. Westphal, taking Aristoxenus's fiera^v to mean ' midway between ', made the syllables i| : 2 (ratio 3 : 4, not ypapifjLos). It has also been supposed that metrically — perhaps it would have been safer to say ' for music or singing ' — the syllable counted as short (Schmidt-Jebb : Mr. White does not mention this view). Bockh gave the foot only the time of three shorts, but made the ratio between the two syllables 3 : 4 (i. e. f + VS instead of ^ + V). Schmidt, followed by Jebb and others, uses a special symbol for an ' irrational ' long (>). The symbol has some advantages, but it has not been generally adopted. The question is not one of very great importance for the ordinary reader of iambic and trochaic verse. II. The Iambic Tetrameter Catalectic or ' Septenarius ' The iambic tetrameter shortened by a syllable was a favourite verse of comedy : it was written by Hipponax and Simonides, and so far as we know it had no place in tragedy. It was vivacious or frivolous, abusive or argumentative. Terentianus Maurus says of it : sonum ministrat congruentem motibus iocosis. The lyric use of it, with the ' scoptic ' tone and the manner of a music-hall song, is well illustrated by Arist. Plutus, 290-321 : KoX iir]v kyoa ^ov\.r](Toyiai 6p€TTave\b tov KvKXcoTra kt\. (a system of five parts : strophe, antistrophe, strophe^*, anti- strophe^, epode). Terentianus's line departs from the normal structure of the verse, the line just quoted from Aristophanes illustrates it ; there is usually diaeresis after the eighth syllable, at the end of the fourth iambus. The line or a-Ti\o^ is made up of two /cooAa, which are dimeters, each of these consisting of two dipodies. AND SENARIUS 93 The closing cadence has been described by some metricians as : . . . . — v^ — v^ I — — A and this modern or trochaic method of scansion perhaps represents the effect of the verse. At all events it brings it into line with the ending of a trochaic tetrameter scazon. Taken as strictly iambic it would end with ^^ — j _ 1. Neither of these effects would be given exactly when the verse was not sung but read, and for reading it is not very important to decide between them. The penultimate syllable is never resolved by the Greek poets. Plautus not unfrequently allows it to be represented by two shorts, from which it may" be inferred that he did not think of the syllable as rpi^povo^. Roman poets sometimes misunderstood, or at all events treated in a new way, the metres which they adopted. The practice of the Greeks in this instance, though it does not amount to proof, at least affords a presumption that the syllable was felt as a notably long one. The structure of the verse is affected by its place and purpose. It may be ' melic ' outright, a lyric that is sung to music. In other parts it was given in recitative (still with some musical accompaniment). Or it might be merely spoken (yfriXr} Xi^is). Mr. White ( Verse of Greek Co7?iedy, § 59) distinguishes four grades or types : (i) melic ; (2) recitative ; (3) melodramatic ; (4) spoken. By ' melodramatic ' he means the method attributed to Archilochus, irapaKaTaXoyrf, in which ordinary speech was supported by some musical notes. But it is not quite clear that in TrapaKaraXoyri the delivery was quite ordinary speech ; and, if it was not, TrapaKaraXoyri must be identified with (2) and there is only one grade below it, not two. This doubt, however, does not affect Mr. White's statistics, for he deals with only three grades of iambic ' septenarius ', calling the third ' melodramatic '. Whether the ordinary speaking voice in these verses was or was not sup- 94 IAMBIC TRIMETER ported by notes of a musical instrument is not a very important question. In trochaic verse an 'anapaest' is admitted with some freedom, for it is only a heavy trochee with the first syllable resolved (O O -) ; a dactyl is very rare. Conversely, in iambic verse a 'dactyl' finds free admission, being a heavy iambus with its second syllable resolved (- O O), but the parallel does not hold further : in comic iambic verse an anapaest is not rare, but extremely frequent. These general considerations must be kept in view in considering Mr. White's statistics, which show some curious facts. His ' melodramatic ' (spoken) tetrameters closely resemble the trimeter of comedy \ there are many dactyls, many anapaests, and of course very many tribrachs [Verse of Greek Comedy, p. 64, §§ i75~7)- Diaeresis is neglected (i.e. the fourth foot does not end with the end of a word) more frequently than in the 'recitative' tetrameter (once in 4*8 lines as compared with i in 7*8). The ' recitative ' tetrameter is characterized by extreme severity as regards tribrach and dactyl, and it rejects the anapaest entirely (there is only one doubtful example in 155 lines). In these respects it is far more strict than the 'melic' tetrameter, which, however, is considerably stricter than the ' melodramatic ' as regards anapaest and dactyl. The ' melic ' tetrameter is other- wise very like the ' melodramatic '. All three have irrational feet very frequently (i. e. the first syllable of a dipody is long) : the percentage of irrational dipodies is 70 in the recitative tetrameter, and very nearly as large in the other two (melic 69, 'melodramatic' 67). This is a larger percentage than that of the trochaic tetrameter (65). The trochaic tetrameter rejects the dactyl as persistently as the iambic tetrameter rejects the anapaest. Mr. White's figures for the dactyl in trochaic tetra- meters are 6 in 776, but if we disallow two of these (p. 105) the proportion is about i in 200. In the early Roman drama the iambic ' septenarius ' was of AND SENARIUS 95 course written with un-Greek variations which are found in other metres. Apart from these Plautus's treatment of it is marked by great regularity in the observance of diaeresis. There is almost always the end of a word at the end of the fourth foot. The two parts of the verse are separated more than they are in Greek. It is in keeping with this that syllaba anceps and hiatus are readily admitted at this point : hinc med amantem ex aedibus | eiecit huiu' mater. argenti viginti minae | ad mortem me adpulerunt. If the first part of the line is to be thought of as a separate verse, its last foot should be a pure iambus. This also is the rule in Roman comedy. The exceptions are so few as to raise a suspicion about the soundness of the text where they occur. The metre has practically no further history in ' classical ' times, for no comedy of the imperial age is extant. But it had great vogue in Byzantine times, invaded by accent, and becoming in the end almost wholly accentual : KXavSiov 8e (papfxa^Oevros Nepcou 6 /jLrjrpoKTOi'o? 7rapeLO-€(j)0dpT] KccKLo-ra rots Ta>v 'Pco/xaicou (TKrJTTTpoL^. (In (j)apiid\6evTGs and in Nepchu accent and ictus diverge, but otherwise there is coincidence ; KccKLcrTa has a slight secondary accent on the last syllable, which carries the iambic ictus^. In later European literature a verse that closely resembles it has wide currency, a line of six iambi plus a syllable. It is a verse that has hardly any place in ancient poetry. Christ quotes for it only Aes. Choeph. 323 : rkKvov, cj)p6vr]iia rod Oavovros ov Sa/xd^ei. The words, he adds, are written in the Codex Laurentianus as two lyric /coiXa : rkKvov (j)p6vrjixa rod. OavovTos ov Safid^eL. No doubt they are rightly written so. The effect is almost the same as would be given by : 96 IAMBIC TRIMETER T€Kvov, v Xiyeiv, and by synizesis in jEcc/. 1156 : T0L9 yeXcocTL 5' rjSicos 8ia tov yeXcou Kpiueiv e/xe. But this solution has not recently found favour, and it certainly seems untenable. There are more instances than Christ sup- posed. There are two more in Aristophanes, prj Ala Ka/xe ravT eSpao-e, Eq. 319, and Tot's d SpdjxaTL (leg. Spa/xij/jLaTL ?).^ The metre is first used by Archilochus, in very much the same way as iambic verse, for lampoon or invective, personal or ' scoptic ' effusions. For this purpose we should expect that celerity would be imparted to it by keeping the foot pure or normal, -w rather than . (Compare the 'scoptic' trimeter discussed above, p. 69). This is very clearly the case tn the seco7id half of the line^ if we compare Archilochus with the dramatic poets. There are about sixty tetrameters ^ There is no extant example of tetrameters in the parodos of a tragedy, but O. Tyy. 1515 is a survival of its use for the e^odos. The Aristotelian definition of a aTacrifxov in tragedy as to dvev dvanaiaTov Kal rpoxaiov {Poef., c. xii) implies its use in parodos and exodos. 112 THE TROCHAIC TETRAMETER of Archilochus extant (a few more were added by the discovery of the monument to Archilochus in Paros, but these may be left out of account). In the second part of the line there are twelve spondees (and one anapaest). In fifty-six lines of Aeschylus {Fersae 703-58) there are nineteen spondees, or one in every three lines as compared with one in every five in Archilochus. In Euripides there are rather more spondees : in the Phoenissae 588-637 (the altercation of Eteocles and Polynices, fifty lines) there are nineteen (and two anapaests), or one in every two and a half lines. In Aristophanes the occurrence of spondees is much more frequent, as Mr. White's statistics show, one and three-quarters to a line, or two to a line if anapaests are included. Aeschylus, it would seem, retained something of the Archilochian rapidity and precision in the second part of the line, though making it distinctly more spondaic than Archilochus ; while in Euripides it is more heavily weighted with spondees, and in Aristophanes — a result which one would hardly have predicted — more heavily still. In the first part of the line there is no difference between the four poets that would be perceptible to a reader. Here, of course, there are two places where a spondee may occur, not only one, and the numbers are greater, and in fact more than twice as great ; roughly there are five spondees to every four lines or one and a quarter to a line. In Menander's tetrameters {^Samia and UepLKeLpo/jiiurj) the Aristophanic proportion of spondees is maintained. Thus the preference of the Roman drama for spondees (which becomes a rigid rule for the fifth foot in Seneca) only carried further a tendency which had shown itself in the drama of Athens. II. The Tetrameter Scazon Like the iambic trimeter the tetrameter was converted by the pugnacious Hipponax into a heavy-headed club or cudgel by making the penultimate syllable long. AND 'SEPTENARIUS' T13 Xd^ere /lov Oalfidria, Koylrco BovwaXov tov 6(f)6a\/i6u' dficpcSe^io^ ydp €i/jLi Kov)( d/jLapTduco kotttcou. The second part of the Hne was no doubt — v^ | — ^ | l_ | — ^ . The history of this verse precisely resembles that of the choliam- bus. Hipponax and Ananius do ?/^/make the sixth foot invariably a pure trochee (as Martial would have done if he had written tetrameters) : dXXd irdaLV l^OveaaLV e/jL7rp€7rr]^ kv fivTTcorcp. The verse was written by Varro at Rome in his Saiurae. He has a line of the strictest type : hunc Ceres cibi ministra frugibus suis porcet, but other lines show that he did not bind himself by the restrictions seen in this one : hunc vocasse ex liquida vita in curiae vostrae faecem ! sic canis fit e catello, sic e tritico spica. 'Liquida' (w w — ) in the third place in the first of these lines, and 'sic e ' ( ) in the fifth place in the second line/ show that Varro is not even following Hipponax ; he is making a Latin septenarius limp, heavy though its gait already is through the admission of spondees and anapaests in the first, third, and fifth, as well as in the other places. ^ Like fit I'sj't in the Pompeian line quoted above (p. 103, footnote). 1887 CHAPTER VIII HENDECASYLLABICS OR PHALAECEI The lyric art of Pindar or Bacchylides with its varied cadence was forgotten or neglected in the Alexandrian age, and the tendency of poets was to take some one simple form of line and repeat it unchanged throughout the whole piece. To use a metre in this way was to use it Kara (ttl\ov (' iugiter uti ' is Terentianus's phrase). This Alexandrian tendency is dominant in Catullus, but not in Horace. The poet who so used a verse for the first time, or who used it conspicuously in this manner, sometimes gave his name to it. Thus a form of verse which had been written by Alcaeus — and written by him Kara (TTiyov — came to bear the name Asclepiadean. So, too, the verse with which we are now concerned, though it had been written by Sappho and called ^a7v(^iKdv^ came to bear the name of an obscure and presumably Alexandrian poet, Phalaecus. The Romans, however, commonly designated it by the number of syllables it contained, 'hendecasyllabi' (sc. 'versus'). There were other forms of line that contained eleven syllables, but this was the only one to be frequently employed Karh (rriyov. Mainly lyric in origin, and used occasionally in the lyric parts of tragedy (e. g. Eur. Orestes 833), it became with the Romans only semi-lyric in character. It lent itself in Latin to light and casual composition on familiar topics. Like most metres when they passed to Latin, it underwent, after Catullus's time, very strict regulation, and the old freedom of the first two syllables was entirely discarded. HENDECASYLLABICS OR PHALAECEI 115 It is this ' stichic ' and highly regulated use of it at Rome that brings it into the company of familiar and standard metres like the senarius and hexameter. Its earlier history asso- ciates it with lyrics, and problems about its structure arise which can hardly be discussed apart from various forms of lyric verse. The scheme of it in Greek is : — \j w — — V^W — KJ — W — — i. e. it begins with the ' Aeolic basis ', two syllables which are not yet stereotyped in one unvarying shape. Perhaps the most famous Greek example of it is to be found in the first two lines of the stanza in the scolion on the liberators of Athens, or supposed liberators, from the tyranny of Pisistratus : (ptXraO^ 'ApjjLoSC, ov tl ttov r^OvrjKas, u7]croi9 8' kv /xaKccpcoy ere (paariu eluai, (where the first line begins with — w and the second with ). It is obviously akin to Glyconic verse, and a Glyconic line results if we remove the last three syllables : cui dono lepidum librum ? CatuUus's first poem begins : cui dono lepidum novum libellum arida modo pumice expolitum? Cornell, tibi ; namque tu solebas meas esse aliquid putare nugas. As the second and fourth lines show, Catullus admits both — w and w — at the beginning. He does not admit ^ ^, though once he seems to have a tribrach (in a proper name : ' Camerium mihi, pessimae puellae ' — unless we are to take this as 'Cameryum '). In the same poem, where the tribrach occurs, he makes the curious experiment of admitting —-in place of the dactyl which follows the ' basis '. I 2 ii6 HENDECASYLLABICS Like several other metres the hendecasyllabic admitted of being interpreted as Ionic, and this view of it was taken at least as early as Varro.^ An ' lonicus a minori ' could take several forms : ww — — or — - — or ww — v^ or — \j —^ . Catullus's first line read in this way becomes : cui dono | lepidiim noivum libdllum. That the verse was commonly so read in classical times is not to be believed ; the rapid and turbulent or unhinged move- ment of Ionics is quite out of harmony with the tone and theme of most hendecasyllabics. But at a later time hymns were written in it with variations which show that the writer -thought of it as Ionic. Besides the Ionic theory of its nature there was the view that the line consisted of part of a hexameter and part of a senarius, two-and-a-half feet of each (TrevOrj/iL' fiepes SaKTvXiKov •{• irevOrjiiLfi^pks lafL^LKoy) : — — I — ww| — w — w — j — cui dono lepidum || novum libellum. Both these views of it would lead to the belief that the first syllable oug/tf to be long, and after Catullus's time it became the regular practice of Roman poets (so far as we know it) to make it so. Horace no doubt would have begun his lines with a spondee, if he had written any hendecasyllabics, and Martial invariably does so. Sidonius calls the verse triplicis metrum trochaei spondeo comitante dactyloque dulces hendecasyllabos. ' Terentianus Maurus, 2833-48 : ' Idcirco ' (because, if an anapaest be inserted after the first two syllables of a hendecasyllabic line, a Sotadean results, [w v-/ — ] — v^ v^ — ^w* — w— J=£) : genus hoc Phalaeciorum vir doctissimus undecumque Varro ad legem redigens lonicorum hinc natos ait esse, sed minores. Atilius Fortunatianus says : 'Varro in CynodidascalicoPhalaecion metrum ionicum trimetrum appellat, quidam ionicum minorem.' OR PHALAECEI 117 Terentianus Maurus analyses the verse in various ways, begin- ning with the division suggested by what was probably the prevalent theory of its origin : cui dono lepidum | novum libellum, and taking next the division I — WW II — w — w — — . In Statius's poems on the ' Via Domitiana ' {Silv. iv. 3) and on the ' Birthday of Lucan ' {Silv. ii. 7), the longest and most elaborate pieces of hendecasyllabics that we have from the first century of our era, the great majority of the lines are of one or other of these two types : Lucani proprium | diem frequentet quisquis coUibus | Isthmiae Diones pendentis bibit | ungulae liquorem. It is noticeable further in the practice of Statius and Martial that elisions are very few and very slight. There is a careful avoidance of elision which is foreign to Catullus. Coincidence between word and foot is not avoided. There are many lines in which only one foot does not end with the end of a word : Baetin, Mantua, provO|Care noli. Coincidence throughout, however, is quite rare {ViaDom. 91): Poenos Bagrada serpit inter agros (where spondee, dactyl, and the three trochees are all separate words). PART II CHAPTER I GREEK LYRIC VERSE -Mr I. Archilochus, Alcman, and Sappho Whatever forms of lyric verse existed prior to the Homeric poems or contemporary with them — and it seems probable that shorter lines, more or less Aeolian, existed, out of which a hexameter could be formed — the first 'lyric' verse that presents itself in literature or to the view of the historian is the elegiac couplet. The ancients would not classify it as ' lyric ', but it is clearly a change in the direction of lyric poetry. Tradition made Callinus or Tyrtaeus its founder, along with Archilochus, but for our present purpose we may ignore the dubious question of priority and contemplate it as one of the various advances in metrical art made by the poet of Paros. Archilochus was looked upon by ancient critics as a poet of the same quality as Homer, though he attempted nothing on the scale of the epos,^ and in particular as an originator of new metrical and musical forms. Marius Vic- torinus called him ' parentem artis musicae, iuxta multiformem | metrorum seriem diversamque progeniem '. Slight as the surviving fragments of Archilochus's poetry are, it is possible * Quint. X. I. 60 : ' Summa in hoc vis elocutionis, cum validae turn breves vibrantesque sententiae, plurimum sanguinis atque nervorum, adeo ut videatur quibusdam, quod quoquam minor est, materiae esse non ingenii vitium.' GREEK LYRIC VERSE 119 to discover in them at least three important lines of advance in metrical art. (i) There is first the vein of composition which we may associate with the word ' epode ' as that term is applied to verses of Horace. A well-known and established line or (TTL\os has a different appendage attached to it to make a couplet. An elegiac couplet is an epode in this sense (the second line a sequel or following verse, kircoSos)- It was a further step, and a very important one, to make this appendage consist of feet of a different type, e. g. to make the alternating line an iambic dimeter : Sv(rT7]uos eyKeLfiai irodo), d-\^v\o^, y^aXerrrjcrL Oecov 68ijurj(TLV €Kr]TL ireirapiiivo^ SC ocrricou, or, conversely, a normal iambic line might be followed by something shorter that was dactylic : epeco TLV^ viXLV alvou, cb KrjpvKdSrj' d)(^uv/jLevr] aKvrdXi]' TTidrjKos fj^L Orjpicoy aTroKptOeh /J.OVUOS dv' k(r\aTLrjv' Besides these there was the combination of the longer and shorter iambic line, trimeter and dimeter, which Horace adopted in most of his epodes. The ' epodic ' vein is known to us chiefly in Horace. The recurrent couplet or short stanza has little place in the lyrics of the drama or in Pindar. Pindar or Sophocles set up every time a new fabric of 'manifold music', a fresh piece of metrical and musical architecture ; the elements combined were more or less familiar, but the combination of them was devised for the occasion. The simpler ' epodic ' vein is seldom seen in tragedy. The elegiac couplet is found only once, for a Oprjvos, in the Andromache of Euripides, a iiovcoSia ('iXico alTreLvd IldpLs ov ydfxov ktX., 1. 103 f.), and in the same play he has 120 GREEK LYRIC VERSE a dactylic hexameter followed by a trochaic KotXov (1. 117): S> yvuat, a OiriSos SdireSoi^ Kol duccKTopa Bd 'raipe, BdfivaraL nodos- This also may have been part of an ' epodic ' couplet, perhaps with an iambic trimeter before it, as in Horace : Petti, nihil me sicut antea iuvat scribere versiculos amore percussum gravi. Horace probably thought of the second part of the longer line as iambic, and divided the whole thus : — WW — WW — j — — w — — — w — . But Archilochus's syllables can be taken in another way and, whether he meant it or not, the line will serve as an introductory example of a large class of lyrics : — WW — WW I — w — w — w — A. Here we have a dactylic and a trochaic group ; and the dactylic group comes to a marked close, like a hexameter. It ends in a spondee (though we may expect to find occasionally a syllaba anceps, the last foot appearing for prosody as GREEK LYRIC VERSE 121 a trochee). Verse of this type (with the trochaic part usually shortened and weighted at the close, — w ) was wTitten very extensively by Pindar, Bacchylides, and others, and various names have been devised for it, of which 'dactylo-epitrite' was perhaps the most useful and the least ambiguous. (iii) Another extant fragment of Archilochus runs : ToTos yap (piXorrjTO^ epooy virb KapStrjv eXvadeh TToWrji/ Kar d)(Xvu o/jL/jLaTcou e^euei^. \^KJ KJXov ; according to Aristoxenus the longest dactylic kS>Xov is one of four feet, and even if we think of the dactyls as more rapid than ordinary dactyls and equivalent in time to trochees, the feet are still too numerous— for trochees Aristoxenus's limit was six. But if; while preserving the structure, w^e reduce the number of dactyls, or of trochees, or of both, we shall arrive at lines which are short enough to be single KooXa, and these will form a new type of metre — 122 GREEK LYRIC VERSE dactyl and trochee not in separate compartments, but in the same kcoXou. Clearly there will be several species : (a) JVif/i dactyls preponderating and trochees few : 09 [i^Ta M.aivd(TL ^OLKyjLOS oyLjiaaL SauTai (Ar. Lys. 1285.) (four dactyls, and trochaic dipody catalectic — ^ — a). 0) Slol tS>v OvpLScou KaXov e/j.^XeTTOKTa (three dactyls, and trochaic dipody or possibly — u^ l_ — a). The same with w w or — or w prefixed : dTeXeoTTara yap Kal dfid)(ava tovs Bavovras KXaUiv. (Stesichorus.) Sapov 8' dVeo) \p6vov rja-TO Td(p€L Treiraym. (Ibycus.) 'Avavpov vnep TroXv/Sorpvos e^ 'IcoXkov. (Simonides.) (b) JVitA dactyls reduced tfi number — reduced to one : Kal jjiiXi Kal TO, TepTTv' dvOe dcppoSLO-ia (this is still two KcoXa, — ww — ^ l__ ' — v^ — w —\^ — a). TIdu, TleXaayLKov 'Apyos i/J-ISaTevcov (Phalaecean, — ^ — WW — w — w — — or — w — WW — w — w I — — a). TTOLKiXoOpou' dOdvaT^ A(ppoSLTa (Sapphic,, — w — w — WW — w — — ). (c) Wi't/i both dactyls and trochees fewer in number : vdi (f)opriiJLe6a crvv fieXalva (—WW — WW — w — — , Alcaic decasyllableX r) TToXiv rjixerepav e^ef (Ar. Thesm. 1140.) (the same, catalectic). The latter is seen also in Ibycus, fr, 1 : GREEK LYRIC VERSE 123 r]pL fi€u ai T€ KvScouLac firjXiSes dpSo/ieuaL poav eK TToraficoi^, 'Iva irapQivoav (the next three lines here are complete dactylic tetrapodies, each ending in a dactyl and ' hypermetric ' in scansion : " t' olvauOiSe^ av^oii^vai ktX.). The survey of these metres has led us into regions which it is not known that Archilochus trod. The shortest kS>\ov of the type which has now come into view is -ww -v^ -a (ofxfia- (TLV kvSlKOLS, ' vulgus ct arceo ')or - \j\j -k^ -^ {Sv(rfj.a\a S' €(ttI KpTvaL, 'Lydia, die per omnes'). Verse like this is not ' Parian '. It has been called ' Logaoedic ' or ' Lesbian ' or ' Aeolian '. * Logaoedic ' is a word that was invented by metricians of Roman imperial times. It was applied to lines of the type o) Bloc, tcov OvpiSccv KaXov kfJLpXeTTOLaa (five or six feet, six if the ending is 1— — a ; the maximum number of feet that can form a single kcoXou in trochaic time). We have now met with two priiiciples on which composite metres can be constructed ; the principle of separate groups or compartments, dactyls in one group, trochees in another, and the principle of mixture, groups internally heterogeneous, dactyl and trochee in the same KchXov. I'he words which come nearest to expressing this difference of principle are the Greek adjectives kindvvQ^TOv and jiiktov {eTnavi^OeTou, ' out- wardly compounded ', compounded by juxtaposition of groups). But before tracing these two veins of composition further — both of them are extensively illustrated in Pindar and the drama — we must revert to early history, and it will also be desirable to survey first those lyrics in which there is no heterogeneity at all, metres composed of one type of foot only {/lirpa {loi^GeiSfj or 6/xo^iSrj). In the time of Archilochus or not long after it, in the 124 GREEK LYRIC VERSE { seventh and sixth centuries B.C., there were other develop- ments of music and metre in Greece : Dorian, represented by Alcman at Sparta and Stesichorus in Sicily, and Aeolian or Lesbian, represented by Sappho and Alcaeus. The Dorian : poetry is mainly choric, composed for a public occasion and i expressing the feelings of a people or of a large company. The Aeolian poetry expresses the feelings of the poet or poetess, and it is often addressed to a particular person. Stesichorus (or Tisias) of Himera was a poet of considerable versatility and importance.^ Quintilian describes him as ' epici carminis onera lyra sustinentem'. He related heroic stories in new shapes, for example making Clytaemnestra take the lead in the murder of Agamemnon (as she does not do in Homer); and for these heroic lays he seems to have used metres akin to that of the epos, numerous groups of dactyls, if we may judge . by a few extant passages, with only few and short groups of trochees interwoven with them : ovveKa TvvSdpeos pi^cov TTOTe 7rd(TL BeoLs fJLovvas Aa^er' rjirtoSSpov KvTTpiSos' KHva Se TvuSapiov Kopais \oX(0(Ta/j.iur) Scyd/iovs re Kal Tptyd/iovs rlBrja-Lv Kal XiTrecrdvopas ... — WW — WW — ^ — — V^\^ — WW — — — WW — WW — \^ _^_^ _v^_ W — WW— WW— W— WW— w— — — WW —WW ... (or —WW — w — a). (TvuSapeov is trisyllabic, by synizesis TvuSapeov. Two of the lines are preceded by — and w, a syllable for which * The versatility which we attribute to Stesichorus may perhaps be illusory, things written by various poets having attached themselves to the name 'Stesichorus'. His personality is obsCure. But the most sceptical view of him does not impugn the existence of an early vein of lyric poetry in Sicily in the metre here described'. GREEK LYRIC VERSE T25 modern metricians have used the useful term ■ anacrusis '. Here the syllables can be accounted for as belonging to the preceding kcoXov : ovveKa Tvv8dp€0$ p^\C^v irore irda-L 6eoh ktX., but this is not always the case. The last kcoXou in the fourth line is one that we have met with as the shortest Mogaoedic' group, -^w -w -^.) The longest extant passage is dactylic almost throughout : AeXLO^ S' 'TnepiopiSa^ Siwa^ kcTKaTe^aLvev )(pvcr€OV, 6(f)pa St' 'flKeayoio Trepdaa^ dcpLKoiO' Upas TTOTL fievOea vvktos epejii^d^ TTOTt fiaripa KovpiStav r' dXoyov TraiSd^ re ^lXovs' 6 8' ey dXa-09 ^(Sa Sd(f)vaL(TL KardaKLOv Troaal TraiV ^169. — WV^ — V^W — ^W — WV> — WV^ — — — (^V_/ — WW — WW — WW WW — WW — WW — WW — WW WW — WW — WW — WW — WW — /\ WW —WW — — — WW — w — — WW — w — . (Heracles landing at Gades from the Sungod's boat, to encounter Geryones. In the third line an ' anacrusis ' which cannot be attached to the preceding line. Ancient metricians had to make this line anapaestic, w w - | w w -, &c.; or they made the whole anapaestic and called the dactylic lines ' acephalous '). Tradition ascribes to Stesichorus the inven- tion of the triple or triadic arrangement by which a strophe was repeated and followed by an ' epode ' — eircoSSs here meaning a strophe that follows another duplicated one, not a line that follows other lines. Of Alcman's poetry many short fragments survive which show great finish and variety of metrical form, and we have also a considerable part of a TrapBevLov, which was mono- strophic — the same strophe recurs, without an epode. In this 126 GREEK LYRIC VERSE system trochees predominate, a few lines are ' logaoedic ', and the penultimate line is one of four dactyls and the last is the Alcaic decasyllabic (= ' virginibus puerisque canto ') : 77 oij-x opfj^ ; 6 /xeu KeXr}9 'EveTLKQs, OL (5e yjxira TOLs e/j.ds dvey^ids ^4.yr)(TL\6pas kiravOel ^pL'cro? 0)9 aKriparos TO r' dpyupiov 7rp6(T007rov SiacpdSau — tl tol Xiyco ; — Ayr](TL^6pa /leu avra. — a de SevTepa neS' AyiScou to fei^o? L7r7ro9 eiprjucp KoXa^alos SpafieiTat, Tal TreXeLaSes ydp dplv 'OpOia ^dpos (pepoLcrai? vvKTa 81' dfiPpocTLav aTe crrjpLOU do'Tpov dfeipofiij/ai p.d)(0VTai. The first line is -^ -w -w - (=' non ebur nee aureum '). For the second the preceding strophe has 6 5' oXpLos octtls €v(ppcoi^, w -V-/V-/ -^ , so that oA/Sioy answers to {'E)j^eTiK69, -^^ to ^^^, an indication that in a lyric which is on the whole trochaic we have to reckon with a dactyl which has the time of a trochee. The couplet is repeated four times : lines i, 2 = 3, 4 = 5, 6 = 7, 8. Then follow two longer trochaic lines, trimeters, the element repeated being - ^ - ^ . Next come two dimeters of the same type (- v^ - ^ I - ^ - ^), then the dactylic tetrapody (ending in a dactyl, and ' hypermetric ' or continuous in scansion with the line that follows), then the Alcaic decasyllabic, d/v^ -^Ky — and - ^ — )} A good example of what we are proposing to call by Alcman's name is seen in the lines (Ar. Av. 747 (■): eudeu axTTrepel jxeXLTra ^pvvL\os dfiPpo(TL(x)v €7ri(ou dnelSocrKeTO Kapirov del (pepoou yXvKelav ccSdv. — \J KJ — WW — WW — WW — WW — WW — w — w — w — — . A It has been proposed to call this ' simplified logaoedic V but the simplification, or separation of groups of similar feet, is carried so far that the term 'logaoedic' ceases to be appropriate. If the word ' logaoedic ' is to be retained, it would be most serviceable in the sense that dactyl and trochee occur within the same kcoXoi/ — a kS)Xov which includes both is ' logaoedic' ; and we may regard on Sid tcou OvptScou KaXov efijSXiiroicra — to which the ancients gave the name — as the upper limit or extreme case. Anything longer than that — with the dactyls grouped together, and the trochees in another group — is not 'logaoedic'. 'G*- Solvitur acris hiems grata vice , veris et Favoni ^ A different account of ' dactylo-epitrite' verse, which has found favour with recent metricians, is discussed on p. 177, 2 By Mr. White, in his Verse of Greek Comedy. 128 GREEK LYRIC VERSE will be excluded from the -category. Contrast with this the so-called * Priapean ' line : o colonia quae cupis \ ponte ludere longo — w— v-^w— v^— A — <^— WW ^or I — ~ a), which consists of two KcoXa, each palpably ' logaoedic ' or ^ /XLKTov, a ' Glyconic ' and a ' Pherecratean ' verse. Glyconic was one of the many measures written by Sappho. The poetry of Alcaeus and Sappho is known to us very imperfectly, in a few portions of poems, stanzas, or lines that have been more or less accidentally preserved. In quite recent years a few portions of poems of Sappho have been recovered, which are metrically instructive.^ The chief forms of verse written by the Lesbian or Aeolian poets are the following : (i) The Sapphic stanza : 7roLKi\60pop\ dOdvar^ A(ppoSLTa. — \^ — :=^ — \^v^ — <^ — ^=- — W W ^ fer (The fourth syllable may be short or long. Horace made it invariably long, and Horace also preferred the caesura seen in KOL yap at (pevyei \ Ta\e(o9 Sid>^€i. Sappho has no such preference, and very frequently ends a word with the fourth syllable, as in : (paLuerat fioi \ Krjuo9 lv dvefiodv (TTacnv' TO jxkv yap evOev KVjxa KvXivSeTai, ^ They have been made accessible in Diehl's Sitpplementmn Lyricum, one of the ' Kleine Texte' published at Bonn (by Marcus and Weber), which contains also parts of two poems of Corinna and Paeans of Pindar. GREEK LYRIC VERSE 129 TO euuew afx/ies au to /j.€(T(tou pal' (f)opr]fj.€Oa crvv jieXatva ii. — v_/ — ^ \J — V^ — v^ — ^ — — — v^ — i^ — ^1^ — wv-* — v-/ (Horace made the fifth syllable always long, and introduced a diaeresis after it in the first and second lines : see Chap. III.) (3) Metres of which some were afterwards called 'Ascle- piadean ', each line beginning with what has been called a * basis ' or ' basis Aeolica ', a kind of disyllabic anacrusis, syllables which may be — v^ or v^ — or «^ '^ or . The chief metres of this type are : Phalaecean : ^^ :_^w -w -^ -^ Glyconic (with Pherecratean^ which is shorter by a syllable, the shortest of the lines of the type) : ^^;_^^ _^_ (Pherecratean, ^'^\~^^^ - ^). Metres like the Glyconic, but with more than one dactyl : (a) ijL€iiuaL(rO\ oiaOa yap coy ere TreSrJTTO/j.^i'. (Sappho.) ^V^ Vw^V-/ — v^ — , (b) wPTjp ovTOs 6 /xaio/xepos to fiiya KpeT09 dpTpi'^eL Taya Tav ttoXlv, a 8' €)(€Tac poiras. (Alcaeus.) — \J ^ — WV^ — v^v,^ — v^ Asclepiadean metres : (a) TjXde^ €K TrepaTcov yds eXecpaPTiuap Xd(3au TOO ^L(f)^os \pv(To8eTav e\cou — V^v-/ — — v_/w — \^ — . (j8) ixr]8\v dXXo (pvTevcrrjs TrpoTepou SeuSpiou d/nreXco _V^V^ — — \^ ^_/ — — WW — V_/ — . 1887 130 GREEK LYRIC VERSE The last two metres are obtained from the Glyconic by the introduction of one or of two choriambi (- w w -). Xd^av Xpva-oSeTau e\coj/ is a Glyconic, as is firjSeu SeuSpiou d/jL7ri\co. So far we have been describing metres in terms of dactyls and trochees. It is now necessary to explain that a different view of them has recently become current and has been advocated by distinguished scholars. An upholder of it would say to us at this point : 'There are no dactyls in these metres ; what you take for a dactyl is really part of a choriambus ; primitive Aeolian verse, so far as it divided itself into fixed groups of syllables,' he would probably be cautious enough to make this reservation, ' fell into groups of four syllables. The apparent dactyl in a Glyconic line results simply from an iambus following a trochee : dpio-Tou fiku \ vScop 6 Si, w — w I N^ — w -.' Postponing the discussion of this theory, which is open to some serious objections, let us first see what can be done with the dactyl and trochee. The ' Aeolist ' or ' Indo-European ' or ' quadrisyllabist ' metricians themselves admit that there was verse in which a dactyl and a trochee came together. What metrical elements or phrases are wanted for the construction of the verses of Sappho and Alcaeus ? We want hardly anything beyond the two forms of the shortest kooXou of the mixed type, which have been mentioned above : (a) —WW -\j — , vulgus et arceo, oixfLacrLu €u8ikol9. (b) — w w - w Lydia, die per omnes, Svcrfia^a 8' ia-ri Kplvai} * Perhaps discoverable in the Linus-song (Schol. //. xviii. 57o\ At all events the lines ^oiPos Se KOTO) a' dvaipfi Movcai 5e ae dprjuiovaiv seem to be the least questionable part of the text of it ( vw — w —^). iK GREEK LYRIC VERSE 131 (b) is written Kara aTL\ov by Aristophanes (Tr. 10) : ovK kros, ft) yvuaiK€9, Trdcri KaKolcTLV -qfids criu eKacTToO' dvSpe^ ktX., and Eupolis wrote a stanza in it ending with the line (J^r. 163): /xijXa 8e \p€jjL7rT€Tai {— (a)). (a) coincides in syllables with a very frequent form of the dochmius, which sometimes has (b) attached to it as a clausula : — \j \^ — \_) — /ZT^re (re OvfioTrXr] Or)$ Soptfiapyos dja (pepirco KaKov 5' — V^ \_/ — <^ _ €K/3aX^ €pcoT09 dp\dv. (a), w^hich has been called, and may conveniently be called, an ' Aeolic tripody ', enters into various metres. It is the second part of the first and second lines of an Alcaic stanza {vulgus et arceo). It is also the close of an Asclepiadean line {sevens arboreni). With two syllables prefixed it is a Glyconic {cui flavam religas coma7?i\ and to prefix a ' basis ' of two syllables is a regular habit of Aeolian verse. With four syllables prefixed (a kind of double 'basis', in the regular shape of a ditrochaeus, -^-^) it is a Sapphic line {ttol- KiXoOpov dOduar' AcppoScTa).^ With the dactyl doubled it is the fourth line of an Alcaic stanza, and the effect of an Alcaic stanza depends on the fact that the third and fourth lines taken together exhibit on a larger scale the movement * In one of the recently recovered poems Sapplio makes a more curious experiment in construction than this would be, prefixing a cretic to a Glyconic line {Suppl. Lyr. 7) : vvv 5e ti.v\Za.iaiv luTpe-mrai ^vvai- where the Hne is followed, in a regular three-line stanza, by another Glyconic and a Phalaecean. 132 GREEK LYRIC VERSE seen In the first and second lines separately — the third line is a longer trochaic or iambic phrase than the first part of one of the preceding lines ; and dactylic movement, after being as it were longer staved off, comes when it does come in greater volume, two dactyls instead of one. In the first two lines what is prefixed to (a) is an iambic wei/Orj/iLfxepe^, the syllables ^ - w — — .^ Whether we call this an iambic TTei^dr]fjLi/i€ph or a ditrochaeus with anacrusis makes no difference to our reading of them ; if we adopt the latter description of odi profanum it does not mean that we propose to pronounce them 'o— di pro — fanum '. What words Alcaeus would use to describe the syllables we do not know ; possibly words that would describe them very imperfectly. Metrical, like grammatical, terminology was probably very scanty and in- complete in early times ; so, for example, in the line ydii/rei \^ipos iXcou diro irvpyov, Xvypov oXeOpov [oXeOpou constructed as pLyjni^ dXedpiav would be) Homer writes an exact specimen of the construction now called 'cognate' or 'internal accusative', but it is certain that Homer had no such phrase by which to describe it. In the third line of an Alcaic stanza the lap,pLKov TrevOrj- fjLL/iepi?, instead of being prefixed to (a), precedes a ditro- chaeus (— - w - i=^ I - w — i=^) ; or we may say that the line is a trochaic dimeter with anacrusis (— j - w | - *^ )j which is in some ways a more convenient description, and if we adopt it we can point to Seneca's use of the line without anacrusis : semper ingentes alumnos. {Agam. 8ro.) * These syllables are readily prefixed in various forms of composition. They precede an ' enoplius ' (— v^w —\j but it does not give an equally good account of the third — WjW— wwl— w — , and it becomes still less attractive when a third dactyl is added : oouTjp ovTos 6 fj.aLo/xei'o^ TO fxeya Kperos (Alcaeus — but Sappho also writes the line : fjLudcreaOaL tlvol (pafxi Kal vcTrepou d/jL/jLeooi^) ^ W j W — WW — WW— I w— . ^ iTed'f}Trofx(v — fieOeiirofXfv. '^ One of the recent volumes of Oxyrhynchus papyri contains a portion of a poem of Sappho in which this verse was written Kara otlxov. The subject was the wedding of Hector and Andromache : "EnTOJp Kal cvviraipoi dyoia^ kXiKoiiriSa @r](3as l£ lapas TIXaKias t an divvdco dfipav 'Avdpopdxo-V ^vl vavaiv Itt' dkpivpov novTOv ktX. Here the shortening awiraipoi is Homeric, and belongs to dactylic verse. The poems in this Oxyrhynchus volume show us both Sappho and Alcaeus (p. 55) interested in Homer, and recalling Homeric events and Homeric phraseology. 134 GREEK LYRIC VERSE It is difficult to believe that the longer lines are not dactylic ; and if they contain dactyls, is it credible that the syllables — w ^ in the shortest line are something quite different ? We have then a series or scale, in which a Glyconic grows by the accretion of dactyls : (a) ray 8' eyco rdS' dfJLeL^ofiav. (b) fMe/iuaia-O^ olaSa yap a>s (re TreS-qTroji^u. (c) dyTpeyjret Taya rdv ttoXlv d 8' e^^rai porra^. But there is also another series, starting from a Glyconic : (a) Tai^ 8' eyo) Td8' dp,€iP6fiav — cui flavam religas comam.^ (b) Xd/Sau tS> ^L(p€09 \pv(To8eTav 'i\(ov — Maecenas atavis edite regibus. (c) fxr)8\v dXXo (fyvrevcrrj^ irporepov 8iu8piou d/nriXco — nullam, Vare, sacra vite prius severis arborem.'^ The two series are so much alike that it is almost impossible to dissociate them. But if they are parallel and cognate the first two syllables in the second series are a ' basis ', and the line does not begin with a quadrisyllabic foot.^ The scansion 1 Horace made the * basis ' regularly . 2 It was a curious fancy of Prudentius's to make these three lines form a stanza for his preface to the Cathemerinott (not altogether unlike what Sappho sometimes does, and not so infelicitous in effect as might be expected) : num quid talia proderunt carnis post obitum vel bona vel mala, cum iam quicquid id est quod fueram mors aboleverit. 3 A similar beginning in a hexameter was recognized as Aeolic, XpoLKTixHu, (VT ttv TToXXot kt\, {supru , p. 25). It appears in Sappho's hexameters : 01 av Tav vaKivOov ev ovpecn iroifjiives dvdp(s, olov rb ykvKVfjiaKov kt\. Compare also the wedding song, where a hexameter is disrupted : lip 01 5i) TO fxeKaOpov "Tfir]yaov d(pp€T€ TiKTOvds dvbpes. GREEK LYRIC VERSE 135 fJiijSep dXXo ' (f)VT€V(Tr)^ 7rp6\Tepov SeuSpi ov a/ZTreXo) seems to be excluded. That the first two syllables stand apart from the rest is indicated by the fact that they alone are variable : they may be — v^ or w — or or w ^ . Every other syllable in an Alcaic or * Asclepiadean ' line of this type is absolutely fixed. If we may assume then dactylic movement or effect in what follows, the simplest description of what happens in our two series will be this : in the former the first three syllables of — w w — w — (a)y ere Tr^S-qrrofiev) are duplicated, in the latter it is the first four. The phrase coy ae ireSrjTTOiiev begins but is arrested (cw? o-e ireSr]- co? ere neSi]- TTOfieu), and in the longer line or 'greater Asclepiad' it is so arrested twice. It is this description or analysis of the verse that will enable a modern reader to find it pleasing. To read * antispasts ' with ease and pleasure (> — w — w — w). Apart from these the chief forms which it assumed were : (i) 0000 —WW— (the first four syllables unregulated), the ' polyschematist ' dimeter. (ii) 00 — WW — w— (two syllables still unregulated), 'Glyconic '• The choriambus has shifted its place and is now in the middle of the line. (iii) — WW— w — w — (all the syllables now regulated), ' choriambo. iambic '. The choriambus moves gradually back from the end of the line. On the other theory of the verse it is a dactyl that shifts its place, (ii) is normal ^Glyconic', in (i) the dactyl is nearer the end of the line, in (iii) it is at the beginning of it. Which description is to be preferred ? When stage (ii) is reached a possibility of compromise comes into view. We may say, 'whatever happened before this, in primitive GREEK LYRIC VERSE 137 times, about which we have no direct evidence, it was surely possible that when the "Glyconic proper" was evolved (rav 8' e^cb Tad' afi(i^nfjLnv) the syllables — *^ w in it could be taken as a dactyl, at a time when a dactylic epos was flourishing; and it was thus that it became possible for Sappho to lengthen the line by intro- ducing another dactyl, and to use this longer line in a three-line stanza with two Glyconics '. But recent metricians would not admit even as much as this. They regard 'quadrisyllable' scansion or structure as much more lasting than that, and they find verses in the sixth and fifth centuries which they define as partly ' syllabic ' or amorphous. One of them ^ even gives the scheme of a Sapphic line as 0000 — ^»^ — w , and the ' polyschematist ' dimeter is discovered frequently in the lyrics of the drama. It must be admitted that there are lyrics in Aristophanes which the Aeolic theory accounts for very neatly. It fits them exactly: TTerov, TTiTov, 'SiKod'iKrjf fw — v^ — ! —WW—, diiambus and Trpii' e/JLTreTrprjcrdai KaXvKrjv cnoriambus) ; or : es (Badv TTjs rjXiKias (— w w w w — , two choriambi) ; and it is possible that ancient forms of verse would survive in comedy. The theory suits also the second of the recently recovered poems of Corinna, and an ancient vein of verse might also be looked for in (TKoXia and carniina popnlaria. But when we turn to the eminently 'Aeolian' poets, the 'Aeolic' theory meets with difficulties. There is no question of four unregulated syllables. In the first four syllables of a Sapphic line, and in the first five of an Alcaic hendecasyllable, the variations are of the most innocent and commonplace description. The ditrochaeus with which a Sapphic line begins is regulated as much as any ditrochaeus ever was in the usual verse of early Greece ; and in the Alcaic line the variations are those which are familiar in any iambic Trci/^r^/xt/xepe?. ^ Schroder, in Horazens Versmasse fi'ir Anfdnger erkldrt. The 'beginner', when he turns to his text of Horace, of course finds the first four syllables absolutely fixed. In Sappho only one of them is variable (the fourth). 138 GREEK LYRIC VERSE Horace regulated the metres still further, making certain syllables invariably long ; he fixed all the variable syllables, except the first of an Alcaic line, which he sometimes allows to be short. His lyrics were written to be read, mainly, though the Carmen Saeculare was undoubtedly sung. Sappho's verses were things which could easily be sung to the lute, without elaborate musical training and special preparation. They must have been also read. They are addressed often to particular persons and they were sometimes of the nature of a poetical epistle, e.g. the now extant poem on her brother's return. So they are regular, like Horace's. The same stanza is repeated with only the slightest variations. The notion that a Sapphic line was in part unregulated or amorphous is the opposite of the truth. The first two syllables in certain forms of Aeolian verse may be said to be unregulated ; but it does not follow quite certainly that this part of the verse was ' syllabic ' or amorphous. Several different forms of it, it may be, were definitely recognized and permitted. The same may be the case with the ' polyschematist ' dimeter. If the first four syllables of it were entirely unregulated we should expect to find all the sixteen combinations that are arithmetically possible, but it is admitted that only nine of them are actually found. There are at least three quite definite arguments which we may advance against Aeolic or quadrisyllable scansion. (i) The first has been already anticipated. It consists in pointing out that Sappho could extend a Glyconic line by introducing a dactyl— on the dactylic theory, a second dactyl — and use this line in a three-line stanza with two Glyconics ; and further {supra, p. 134), that rising, as it were, out of the Glyconic, we seem to have two parallel scales or series, one obtained by the insertion of dactyls, the other by the insertion of what, on the dactylic theory, are only ' apparent ' choriambi (— w w — A), and that the dactylic expansion seems to show that the first two syllables are a disyllabic basis. But this is not the only form that this argument takes. There is perhaps also expansion or extension by the insertion of a single trochee, notably in the case of scolia. Scansion by means of choriambi, diiambi, and Ionics explains some Hues, and then breaks down in others. GREEK LYRIC VERSE 139 A common form of stanza in scolia began with two Phalaeceans, and the Phalaecean, as we have seen, is capable of Ionic or quadri- syllabic analysis. vyiaiveiu fiep opiaTOV (iv8p\ BvrjTto looks promising for the quadrisyllabist metrician but the next line is less tractable : devrepov de (f)vav koXov yeviuOui — \j — I v^w — >^ I — ^u J the third line : TO rpiTov de nXovrelu ddoXojs presents to us a five-syllable group, \u ^u — w — I — WW — , but no further difficulty, and the fourth looks like a perfect * Aeolic ' trimeter : Koi TO TCTapTOV i]^a.v pera Tcav (f)t\cov — WW — I w w I W — W — J choriamb, antispast, and diiambus. It occurs also in Aristophanes, and is scanned by Mr. White in that way : f'crrt dcKaiov el 8i]poKpaTovpe6n — >^w — w w w — w — But this line coincides with a double dochmius, and if it is that it is divided in the middle into two equal parts, like pTjre ere 6vpoTvkr]\6r]s bopipapyos njra kt\. Apart from this agreement with a double dochmius, the line also perhaps occurs with an additional trochee in it, though the text is not altogether certain : TvSeidrjv re (f)a ivaXicov ^opd, 1 Horace's Esquilinae alites is very exceptional. Reasons will be given (Chap. Ill, § ii) for thinking that the final syllable oi Esquilinae is not to be thought of as shortened. It is a case of unmitigated hiatus. It resembles the o of Glauco, not the ae oi Panopeae, in Virgil's line : Glauc5 I et Panopeae et Inoo Melicertae. GREEK LYRIC VERSE 141 where quadrisyllabic scansion sunders the two short syllables of K€i). The irrational trochee is not a thing that may occur anywhere ; the country in which it is to be found is quite definitely mapped out. It comes in the group - v^ - ^ (ditrochaeus), and it comes also, for example, at the beginning of a logaoedic line like Srjfjios TOL ere KaXei yvvai-} To assume it here would make the second line one oifive feet, and the last would be 3 + 4. Se^o/iai may be either - w - a or - ^ L_ . In the antistrophe we must assume protraction of a syllable rather than a pause within a word {SevSpOTrrfficou Se fjL7] TTi^ioL pXdpa). The remaining five lines are not so completely free from ^ Of course when this is taken as 'Aeolic' ( \j \ w— v-/— )it is not there (cf. sttpra, p. 136). 1887 146 GREEK LYRIC VERSE what is dubious. Is the third iambic ? (It is certainly not at all like a trimeter, for it is divided as no trimeter — outside conversation in comedy — could be ; so too in the antistrophe. Tpi I T^Tay [JLevco' \ yovo^ [5' del].) Or should the lines be written Oeairio-acTa Trpevfievoos e- 7ri(T(TVTOV9 i^iov Tv\a^ ovrjaipLOvs ? Or may we leave them as they stand, ivithout thinking of the second as ' iambic ' ? The last line but one presents a different problem. It consists of six long syllables. What was it ? Certainly not three irrational or heavy trochees. It is not quite so unlikely — but still improbable — that it was - > i— - > l_ . Without Aeschylus's music all we can do is to suppose that it was a line of six syncopated feet, l_ i_ ^ ^ l_ i_ . Having followed the structure of this lyric the learner should take as an exercise another lyric of the same type, and analyse it for himself, e.g. Persae 115 f. ravTOL \ioL ^e\ay\iTcov (pprjy dfiV(r(T€Tai iaTia^ (Tvdeh, 870 0pTjKL(ou eiravXcov). Dactylic verse differs from anapaestic in several ways. An anapaest may take dactylic shape (- o w), but a dactyl never takes anapaestic, it is never o w -. Resolution is extremely rare, rarer than in lyric anapaests. The ' dimeter ' or tetra- pody is very frequent^ but not dominant as in anapaests. A rule of Aristoxenus makes the maximum colon in this type of rhythm (Terpdo-rj/Mou, \j^ ^^) a group of four feet. Various shorter KcoXa are possible, besides, and some have more vogue than others. A few of the prevalent groups may be enumerated. L 2 148 GREEK LYRIC VERSE (a) Dimeter or tetrapody, ending either in — w w or — : TToWaKi S' kv KopvaiJLOs ioprd. (Alcman.) When the last foot is a dactyl the construction is 'hypermetric'; another dactylic line must follow, without hiatus.* (b) Tripody, complete or catalectic : a> Albs d8v€7r€9 (pdri, 719 TTore rd? woXvxpvo'ov. (O. T.) It may also begin with a spondee : na/xfiLKTcop r' ewiKovpooy. {Fers.) Catalectic ; dkvaoL vecpiXat. (Ar. JVu^.) dj^poTepas dXiov. {Supra, p. 146.) (c) Dipody or ' monometer ', usually ending in a spondee, and often making with a tetrapody the lyric counterpart of a 'bucolic ' hexameter, but found also in the form —^^ ~\^\j. One of the few wholly dactylic systems in tragedy occurs in the Phoenissae of Euripides (1. 784 f.) : © TToXvfio^Qos ^^p'HS, ri nod' ai/xaTi Kot Oavdrat Kareyjei Bpofiiov 7rapd/j,ov(ro9 eopraTs ; ovK kirl KaXXiyopoL^ (rrecpdpoKTL vedviBos copa9 p6crTpv\ov diiTrerdcras Xootov Kara TTPev/jiaTa fieXirei 5 jiovcrav, kv a \dpLTes xopoiroLot. O) ^adioou TrerdXcov noXvdrjpoTa- ' dvT. Tov vdiTos, 'ipTe/iLSos XLOVOTp6\a. The seventh and ninth lines are clearly tetrapodies. Line 4 is likely to be tetrapody + dipody (bucolic division suggested by the words — but we cannot be quite sure that the division was there) : line 8 may have the same division. Line 6 may be supposed to be 3 + 3 + 2. Lines 2 and 3 would perhaps be best taken as 2^+3-|, divided as hexameters (line 3 contains an effect that is not permitted in the epic hexameter, st^J>ra p. 21). The dactylic lyric in the Clouds is well known (275 f.) : dkvaoi ve(pe\aL, dpOcojiev (pauepal 8pocr€pdv (fevcriu evdyrjTou Trarpoy an' 'flKeauov ^apva^io^ vy\rri\S>v opecov Kopvcpd^ kwl 280 8ev8poK6fiovs ivoL Trj\€/jL€da KapTTOvs T dp8o/j,iuau 0' Updv ^Oova KOL TTOTafxcctv ^adicou K^\a8riiiaTa KOL TTOVTOV Ke\d8ovTa ^apvPpofjLoV 285 oixiia yap alOepo^ aKd/jLarou (reXayeiTai fiap/jLapeais kv avyal^. dXS! aTToaeLO'dfjLevai vecjyos o/i^pLou dOavdra^ I8ias e7rt8oofj,e6a 290 TTjXeaKOTrcp 6fi/A,aTi yaXav. I50 GREEK LYRIC VERSE The first line is catalectic, -^w — <^v^ -7^, with a very natural pause after the vocative. There is of course no reason against hiatus between this line and the next. The next is no doubt 4+2. Then follow two tetrapodies of the type first seen in Alcman, then a similar dipody {SeuSpoKofiov^ lvoC); then four dimeters or tetrapodies of the same hypermetric type ; next 2 + 3, probably (in the antistrophe ''eva-Te^avoi re Be\S)v OvcriaL BaXtaL re). Line 287 was not certainly dactylic. If we read fxapfiapiai^ kv avyais it is a logaoedic line, but it is easily made dactylic by reading it fiapfiapiaicrtv kv avyals (and in the antistrophe TravToSaTraicnv kv copaLs). The presence of a logaoedic line is not incredible, but it would be more likely if it were the last of the whole system. Here the final clausula is a paroemiac or catalectic anapaestic dimeter. The dactylic verse of lyrics has several marked charac- teristics as compared with the hexameter. Spondees are rare, occurring chiefly at the end or at the beginning of a colon.^ Trochaic caesura is avoided : only one melic hexameter in Aristophanes has it, but in recited hexameters, epic or mock- heroic, it is frequent.^ Shortening of a long vowel before another is not excluded (e.g. napOivoL 6/iPpo(f>6poL), but it is not frequent. * Bucolic ' division of a group of six feet (4+2) is now and again suggested by the words : ea-TL TL t5>v8* krvjicos ; ein, ® irdrep, | €l ti (piXcT^ /^e. {Pax 118.) elni fioL, S) \pvaias t^kvov kXmSo^, afjL^poTe (pdfjLa. (Soph. O. T. 158.) But that a lyric line of six dactyls was regularly divided in the 1 @r)l3ar eKTeranai (polSepav (ppeva heiimri rraWojv, O. T. In this lyric verse does resemble epic (xpato^A*«j/, eur' av noWoi, supra, p. 25). 2 Forty lines out of 142 have it, according to Mr. White's statistics (Verse of Greek Com., § 363). GREEK LYRIC VERSE 151 ratio 4 : 2 seems to be rather a theory than a thing that can be seen in actual texts. Sappho's divided hexameter : L^OL Srj TO fiiXadpou ^Tfirivaov d€pp€T€, TeKToves dvSpes, is against it. Division between words perhaps does not count for much as evidence of division in a lyric, and some forms of lyric verse even preferred to make a k5)\ov end in the middle of a word, but the forms in which this certainly happens are not as a rule purely dactylic. To impose a scheme of 4 + 2 gives results like these : rivLK &v o^vXdXov iraptSr} OrjyovTos oSovra. (J?an. 815.) dpyaXiooi/ t kv ottXol? ^vvoScoy. KXeo (f)6ou8e fid^eaOo). {lb. 1532.) In the former passage (where the same strophe is repeated four times) one of the lines consists of five dactyls, and in all four strophae the division suggested by the words is penthe- mimeral : pri^ara SaLOfi^vrj \ KaTaXeTrroXoyqaei. {828.y The dactylic system here closes with a trochaic dimeter catalectic as a clausula. So also at 883 a short trochaic line closes a system of dactyls. So, too, in Agam. 104-21 the penultimate line is trochaic {^Xa/Siura XoktOicov SpSficou). This is prepared for by two brief trochaic movements in 1. 108 and 1. 116; and it prepares the way for trochees in the next strophe, 160 f. So in Oed. C. 228-35 there is a trochaic (or iambic) ending. 1 There is an anapaestic pentapody in Ac/i. 285 (= 336) : opov \_l _— U/v^ w— w — — w— ^ yeycovqreov ottlv SiKatov ^iucoi/ w — — j w — — epeio-fx 'AKpdyavTos eifcouvficov re Trarepodv dcorou opOoTToXiv. The division of the third line is very doubtful. Between the first and second lines it will be observed that the syllables -ol TLva Oe- do make a paeonic foot. The last line is not certainly cretic throughout. Cretics or bacchii can be found in the first half of it, but it is quite likely that there was a non-cretic clausula, as there plainly is in the epode : {ev : w w is a dupHcated trochee, 4:2 for 2 : i, and whatever broken or anaclastic Ionics really were, it is not likely that their rhythm constituted a different yeuos or €l8o9. It is not clear that dochmii did not. They are described as an oKTacTj/jLos pvOfios, and there is the authority of Quintilian and of the Scholiast on Heph., c. 10. 3, for regarding the foot ^ Terentianus speaks of the cretic as a highly effective foot both for verse and for oratory : optimus pes et melodis et pedestri gloriae. plurimum orantes decebit, quando paene in ultimo obtinet sedem, beatam terminet si clausulam Mktv\os, cnovbfios imam, nee rpox^-iov respuo (i.e. it should be penultimate, followed by dactyl, spondee, or trochee. With spondee or trochee it is — w — | — k: . When a dactyl follows the ending is really in a double cretic, — w— | — wJsi, the last syllable being anceps. These are the first and second of Zielinski's preferred or favourite cadences). 156 GREEK LYRIC VERSE as divided in the ratio 5 : 3 or 3 : 5 {rpia^ TT/aoy irevTaSa). Dochmii are easy in one way and difficult in another. What they really were and how they were arrived at is a very obscure question. The forms they assume, in syllables, are quite definite, and it is generally easy to recognize and scan and read them. To do this is important for the study of tragedy. It is to tragedy that they belong (though they are of course found in comedy also, where many tragic effects are parodied), and they seem to have been devised to express the intense feelings of distress or suspense or despair which accompany the crisis of a tragic action. (It is a rhythm iTTLTjjSeLO^ TTpbs Oprjvovs Kal (TTevayfjiovs, Schol. on Aes. S. c. Th. 98.) The normal form of a dochmius is '^ w — . Each of the long syllables may be resolved, and both the short syllables may be irrational — actually long, if counting as short ( > ). An irrational first syllable, with resolution of the second, would give > w v^ - v^ - , but whether this form was arrived at in that way is uncertain. Of the many theories which have been advanced about the origin of the dochmius, one, the most recent, Mr. White's, is perhaps the most promising, or the least unpromising. He suggests that it is derived from an iambic tripody, — - w - w -, by total sup- pression of the third syllable.^ The short syllable, he thinks, was not represented by a pause or by protraction of an adjacent syllable (which would leave the rhythm still kvv^d- crrjfjLO^), but dropped entirely so that the total time is eight \p6voL and the foot is 'oblique' or 'askew ', 86)(fjiL09i twisted or distorted to express mental anxiety or agony.^ The usual ^ Compare Aes. Suppl. 911 : where three iambi precede a dochmius : w— v-/— v..^— | \y w— . 2 WilamowitZ; in his most recent book {Sappho und S., p. 183), throws out a suggestion which would undermine all such attempts to show that GREEK LYRIC VERSE 157 line consists of two dochmii, and a line of the same form occurs occasionally outside tragedy, e.g. Pindar, 01. i: but this may be an accidental coincidence in syllables. The rhythm of Pindar's line may have been quite different. Dochmii are abundant in the S. c. Th., where they express the alarm and anxiety of the chorus for the fate of their city, besieged by the Argive host ; or their agonized appeals to Eteocles (686 f.) to refrain from combat with his brother. Except for a few iambi and cretics' the whole of the first utterance of the chorus is dochmiac (78-180), and the opening of it will serve as an example of the metre : Opiofxai (polSepa /leydX' ciyr}' fiedeLTac crrpaTO^' crTpaTOTreSov Xlttoov p€L TToXvs 68e Aeo)? TrpoSpofxos iTnrora^' aidepLa kovls /jl€ ireideL (pai/eTa-' 5 duavSos cra(f)r}s erv/xos dyyeXos. — eri 8k yds kfids ireSC ottXoktvtt' (o- TL \pLjX7rT^L Podv TTOTaTaL, (Spk/iei 8* dijLa^krov 8iKau v8aTos opoTVirov. the foot was in itself * askew'. He thinks that it is a verse that comes in cross-wise, that runs aslant or athwart the structure of the lyric. It does this, he thinks, in the SkoUon of Simonides, which he is discussing. It is perhaps unnecessary to find the emotional effect of dochmii in the nature of the foot itself. —\j WW WW W — W WW — w — > WW — w— jw— — w — c w — — w — i W WW — w — \^ W*-/ — W — W WW — w — I v^_ — v^ — jv^ — — v^ — W WW — W — ;W WW WW W — . In some passages of this play the form - ww - w - is very frequent (219-21, 226-8, 692-3, 698-700, 705-7)- lom'c verse is in some of its forms very simple and regular, in others it presents very difficult problems. In its regular shape it is either ' a maiori ' (' faUing ' — ww ww) or 'a minori' ('rising' w w w w ). In its 'broken' shape {'looviKol di/aK\(ofji€i/OL or KeKXaa-jxevoL) it is most frequently ww-w -^-^ (Anacreontic). The r^Bos of the measure was one of excitement and turbulence, frenzy or ecstasis, of licence and effeminacy. euoe, recenti mens trepidat metu plenoque Bacchi pectore turbidum laetatur. euoe, parce. Liber, parce gravi metuende thyrso.^ It was dveifiivoy, ' mollissimum rhythmorum genus ', asso- ' Ionic verse would be an appropriate vehicle for this : not so appro- priate for the tranquil close of Horace's ode : et recedentis trilingui ore pedes tetigitque crura. Horace is giving us here something like a dithyramb on a small scale, in a different metrical medium. Closing in tranquillity the ode successfully depicts what Aristotle would call the KaQapais of Ivdovaiaoixos by means of an bpyiadTiKov /xf\os. GREEK LYRIC VERSE 159 ciated with the 'mollities' or d^pocrvur} of Ionia and with the orgiastic worship of Dionysus and of Cybele. It could express the abandonment of grief or despair (KeKXaa/xiuo^ Trpo? to OpTjuijTLKOu). Its intrusion in the realm of metre may be compared with the advent of Dionysus at Thebes, and it sometimes provoked an attitude of protest or resistance like that of Pentheus. Plato would have banished it from his Ideal State, and perhaps had it in mind when he spoke of v^pecos rj fiauias npiTrova-ai Pda-eis. The condemnation would apply most obviously to certain forms of dvaKX(o/jLeuoi, such as Galliambics. Regular Ionics are quoted from Sappho and Alcaeus : ' a maiori ', evfiopcporepa MuacnSLKa ray dnaXd^ TvpLvva>s ( v^w WW ^u\J — w . This is not quite regular throughout, -^ — taking the place of the last Ionic) ; 'a minori ', e/xe 8d\av, kjie iraddv KaKOTaToov 7re8i\0LpouTiSa9 [leOcaiiev, 57. i rov KeXaii/oxp'jjTa ^oTpw. If the third syllable is very strong in a group of four, the reduction of the first to a short is only a way of making the same effect a little more obvious. Thus from the ditrochaeus ~ \^ — "^ we should get both ^^ — ^ and WW — , But if w v^ l^ w has given the reader the clue, he will have no difficulty in reading the next ditiochaeus in an Ionic fashion. The first syllable of that need not be reduced. Hence the Anacreontic combination of two Ionic feet ww-w I -^^-. The metre which expresses a loose, unhinged, excited state of mind is obtained by disturbing the balance of a trochaic dipody. Whether this is what actually happened or not, the possibiHty of such an origin for Anacreontics is of some importance. It is a trochaic dipody that has to be disturbed. No similar manipulation, as far as I can see, will elicit any Ionic form from an iambic dipody, w - ^ -. If trochees and iambi were equally real, an iambic dipody ought to give birth somehow to an lonicus a maiori. The inference would seem to be that, for lyric verse, iambic scansion is illusory. Ancient metricians — GREEK LYRIC VERSE 163 in later times at all events — construed much lyric verse iambically. But the reading of it that really works — like Copernican as compared with Ptolemaic astronomy — is trochaic (involving the assumption of anacrusis in many metres). The ditrochaeus is an element in many forms of verse, and if we set down two 'Anacreontei' other syllabic coincidences are discoverable ; we can pick out a ' logaoedic ' group of syllables, an ' Ithyphallic ' and others : itr. log. KJ 'nfvoi), 363-4— a touch of Ionic rhythm, as in Aeschylus, where the subject suits it. i68 GREEK LYRIC VERSE a close in a spondee, and the trochaic phrase is the shortest possible, a dipody with the last syllable usually long. Either group may be preceded by a syllable (anacrusis) usually long (or, resolved, w v^), but sometimes short. Various names have been in use for this type of metre : * Dorian ' (partly from a hint given by Pindar himself : AcopL(p (pcoi/au kvapfJLO^aL Tre^/Xo), 01. iii. 6); ' dactylo-epitrite' (from ewLTpLTOs, the Greek adjective for . the ratio 4 to 3, the trochaic element being in syllables ^ www ' wwww, 3 + 4); ' dactylo-trochaic ' (which is too wide in meaning) ; and most recently ' prosodiac-enoplic '. Various notations or modes of analysis have also been in use for it. (a) When a syllable precedes the groups -ww-ww — and - w they can be described as anapaestic and iambic respectively. This was a method adopted by ancient metri- cians. For the trochaic group it presents little difficulty. Anapaestic scansion becomes questionable when the first syllable is short, as in aVo) noTa/ioop iepcov ; and in a famous opening of Pindar's, reproduced by Aristophanes in E^/. 1264, TL kccXXlov dp)(oiiivoL(TLv rj KaTaTravo/xivoicriv. lambo- anapaestic scansion becomes highly questionable when the second group is no^ catalectic, for the whole then becomes hypercatalectic , a term which has no clear or satisfactory meaning ( — ^~J — ww- ww-|— ). See infra^ Excursus, p. 177. (b) The syllable that precedes is regarded as an anacrusis, a method belonging to modern music which has many advantages. (c) ' Prosodiac-enoplic ' scansion, for which there is some ancient evidence, divides the dactylic group into two equal ^ Or, ' for mere prosody '. That the second syllable in the apparent spondee is not a real long is shown by the fact that it is not resolved. The first syllable is a fjatcpa hio-qp.os. GREEK LYRIC VERSE 169 parts, a choriambus and an ' lonicus a minori ' (- w v>^ - | v^ w ), each of six times, i^da-rj/xa, and each therefore equal to a ditrochaeus (which the epitritus is assumed to be). It has the advantage of avoiding the assumption of a group oi three feet, which is difficult for anything that may accompany a march or procession — a march-rhythm is naturally in multiples of two, dipodies and dimeters (as in anapaests) ; but the poems actually written in this type of metre seldom belong to a procession {TrpocroSo^) and seldom are ' enoplic ' in the sense of accompanying a march or war-dance. It is not clear that the choriambo-ionic division is more than a special musical setting or musical arrangement adopted sometimes and for some purposes and not certainly very ancient. . For the ordinary reader (b) is by far the most convenient method, for (besides anacrusis) he has to recognize only tivo recurring elements, — ww — ww and — w . The reader of Pindar would do well to begin with the fourth Pythian ode : — w— — j — \y ^ — wv^ — ^ a-d/iepou jikv XPl ^^ Trap' duSpl (piXco — \u — — — \J ^ — Kj y^ — — (TTa/jLev evLTTTTOV ^a(nXrji Kvpdva'S — v^— — — w w — v^v^— ^ o^pa Kcofid^ouTL (Tvv jipKea-iXa. Clearly, if we adopt simple symbols for the two elements, the metrical structure of this can be exhibited with great brevity : for example, e and E (the initial of ewLrpLTo? and the initial of kvonXios — kvorrXiO^ was an ancient name for the dactylic group, whether it was always thought of as dactylic or not). 6 — might stand for the catalectic form of the group, + e for the group with anacrusis. On this notation the preceding lines are € E— € E € E— . lyo GREEK LYRIC VERSE MoTaa Aaroi Saiaiv 6(peiX6fxeyov Tlv Ocovl t av^rjs ovpov VflVCOV. That is e E € e. ev6a wore \pv(Te(ov Aios alrjrSyv TrdpeSpo^ ovK ccTroSd/xov AttoXXcouos tv\6vto^ Ipia Xprjcrei^ olKKrrrjpa Bdrrov Kapno(f)6pov Ai^vas, Upav vdcrov 0)9 rj8r] Xlttchv KTL(rcr€iei' eudp/xaTOV ttoXlv €U dpyivoevTL fiacTTcp. — V^V^ — WW — WW i— w — — — WW — WW |~*~^ — w— w — — W i""*^ 1— WW — WW — WW — ^ www — w . In these lines there are variations which would render our brief notation more complicated. There are two groups of foi/r dactyls (E* ?), in the fourth line of the ode and the sixth, the second of them catalectic {E^—?)? In line 7 there is syncopation of a foot or protraction of a syllable in -aeiej^ ev-dpiiarov (whether the syllable ev should be thought of as exactly TpiarjfjLos is a difficult question which hardly con- cerns the ordinary reader ; on any method of scansion it must be prolonged, it is jiaKpds JjL€l(ciou). In the last line there is resolution of a trochee {ttoXlv ev, w w w). The devices or variations presented by this strophe provide the reader with the means of scanning about half of the odes of Pindar. Of the recently recovered poems the paean written for an Athenian ^ On the dactylic theory, a group of four dactyls raises no difficulty, except this slight question of brief notation. The ' prosodiac-enoplic ' or quadrisyllabic scansion has more trouble with it. According to Blass (Introduction to Bacchylides. p. xxxiv) there are two groups, one cata- lectic, which together ' tetrapodiae speciem falsam praebent ' {evOa -norl Xpvcreoju^ — WW — ww — 7\, Atos alrjTWu vapdhpos, ww — w — ^\ Bacch. ix, strophe \. 6, dvOiOi ^avdav avabt]aai.i(vos K((pa\av is — w — WW — , WW - , w w ' . This is not plausible. GREEK LYRIC VERSE 171 Oecopiato Delos (St/pJ>/. Lyr, v) is a simple example of 'enoplo- trochaic' metre (if we may add one more to the many attempts to find a name for this mode of composition). Each strophe began with the words lr\U AaXi 'AttoWov (a 'prosodiacus '; or we may say that the first kS)\ov — here it is the first kcoXov alone — has an anacrusis) : irj'ie AdXi 'AttoWov — WW— WW— — Kol (TTTOpaSas (Pepe/irjXovs — W— — '— WW — WW — — eKTiaau vdcrovs epiKvSia t 'ia^ov — w w — w w— — AdXov, enei (t(J)lv AttoXXcou * — WW — WW— ^ (or —WW - ?) ScoK€u 6 y^pvcroK6[JLas — W W— WW — — A(TT€pia9 SifJLaS 0LK€?P' Lrji'e AdXi AiroXXoV Aaroos 'ivda //e TroiSes — W — — ~ ^ ^■— WW— — evfxeyei Si^acrde v6(£> OepdirovTa vfjiirepoy K^XaSevvd (jvv fxeXtydpyi irai- duos dyaKXios ofMCpd. In the second strophe it appears that the third line is separated from the fourth {fi^pdirovToi, an ending in a syllaba anceps and hiatus). In the first strophe this is not revealed. Con- versely, the second strophe reveals also continuity between the penultimate line and the last. KJ \y — W — — W V^ . Looking at the other specimens of Anacreon's composition in this form of verse v*'e observe (a) the structure is ' hyper- metric ', the last syllable always long and followed by no hiatus^ (cf. a-vfJLPovXo^^ rov kjxov S' epcoT, co Aevuva-e. Se)(€cr6ai and k\a(p7](B6Xl \ ^av6rj, lengthening before the double consonant ^): (b) Anacreon has a strong preference for beginning with , the 'spondaic basis', which Horace made invariable in metres of this type : ^ (c) the last line is. catalectic or shorter by a syllable, in its rhythm very probably ^ ^ t— — A , and coinciding in syllables with a syncopated or con- ^ No Inatus, that is, within the stanza. There may be hiatus between one stanza and the next. 2 A few of the extant lines show an iambus, w— (€70; §' ovt^ av 'AfxaXOirji, Fr. 8;. 172 GREEK LYRIC VERSE It is possible that Anacreon, besides suggesting ' broken ' ; Ionics in Aeschylus, gave a fresh impulse towards the com - position of verse of the Glyconic and Pherecratean type. They are seen together in the lines which have been taken for a regular hymn to ArtemiS;, but which Wilamowitz has shown to be rather a compliment to the people of Magnesia {Sappho inid S., p. 113). In this poem the longer stanza consists of four Glyconics followed by a Pherecratean : Catullus in his hymn to Diana has three Glyconics, as Anacreon has in other poems. Here, as in a similar verse to Dionysus, Anacreon begins with a shorter stanza, in which there are only two Glyconics : yovvovfJLai a-', eXa(pr]^6X€, ^apOrj rral Aios, ay plow BecriTOLV 'Aprefii $r]pa>y. That is, if we assume that - w ^ is a dactyl : GREEK LYRIC VERSE 173 densed anapaestic line whicli Pherecrates, the comic poet, wrote Kara (ttl\ov {infra, p. 254, footnote).^ Verse of this type was no doubt known to all the early poets, though there is no extant specimen by the Ionian, Archilochus. We have seen Sappho writing a stanza which consisted of two Glyconics and a line in which there is an additional dactyl : TToWa Kol t6S' eeiTri fior ^ a)L fi (B? Selva 7r€7r6i^6afi€v, WdcTTCp', T) [xdv a diKoicr' aTrvXifjiTrdi^a)' (which is one of the chief arguments for dactylic scansion, and from which it seemed easy to arrive at the 'iAsclepiadean' type of ' Maecenas atavis edite regibus '). Besides being used in fixed stanzas by the Lesbian poets and by Anacreon^ Glyconic verse was very frequently written by Pindar and by the dramatists, with numerous variations, such as resolution of a syllable or the transference of the dactyl to a different place in the line. The first Olympian ode begins with Glyconic H- Pherecratean : dpL(TTOv iikv vScop, 6 8e ! xpvcrbs ocldofievou irvp. \_/ — — V^V^ — \J — j — \J — WW — — Aeschylus makes less use of it than the other two tragic poets. His lyric composition is characterized by a liking for other things, for long dactylic groups, or systems of trochees with frequent syncopation, or the broken Ionics which he is said to have learned from Anacreon. But he does write Glyconics and Pherecrateans, sometimes alternating : NTcrou dOavdras rpiyo^ vocrcpLaacr' dTrpo^ovXcos TTveovO' d Kvi^6(ppcop inrv(D. KLjydveL 8e viv 'Epjirj^. {Choeph. 619—22.) 1 It is strange to find Mr. White {V. of Gr. Com., § 547) giving Pherecrates' lines with 'Aeolic' scansion, when their purport is to call attention to themselves as crvfi-nrvKToi di'd-naiaTOt. 174 GREEK LYRIC VERSE Or as the closing cadences of a strophe : ofjLfidiTcoy 8' kv (X)(r]VLaLS epp€L TTCca-' A(ppo8iTa. Or T0V9 8' cLKpavTos ex^i i^v^. He shows also a liking for certain variations or kindred forms, e. g. : } yetvaro 7raL8' d/ie/Kpfj. 8v(Tixa\a 8' €(TTL Kpivat ^ (which may be described as Pherecratean with the dactyl in the first place instead of the second). Like Euripides {Her. Fur. 419-22) he has a sequence of Pherecratean lines, with a penultimate Glyconic : vovorcjiiv 8' eV/xoy octt' olcttcov i^OL Kparos drepTTij^' evfJLevrjs o AvK€i09 ecr- Tco irdcra veoXata. {Siippl. 684-7 = ^94~7') (So also in the two preceding strophes, 639-42 = 652-5, and 663-6 =: 674-7.) Euripides shows some inclination to revert to Aeschylean enoplo-trochaics and Ionics ; but apart from that the lyrics of the two younger dramatists are mainly of the Glyconic type. The following are passages where Glyconics are written with considerable regularity : Soph. Antig. 332 f. TToXXd Tcc 8eLvd kov8\v dv- OpcoTTOv 8€iu6Tepov TriXei' TovTo Kal ttoXlov rripau TTOVTOV ^eLjiepLco p6t(o Xft)p€l irepilBpvx^oL- (TLV irepcov vtt ol8yia(Tiv ^ Compare his liking for the Alcaic ending rav dvadkcurov 'ikr} ris dpxav ( — v^v^ — WW — w ), with two dactyls. — w w — w — \J 1 1 — — — w w — W — A — Tra ve/xovo'a klct- ^^ ~ — v^w — v^ u_ 675 o-bu Kal Tav dfSaTOv O^ov — ww— v^— a (pvXXdSa fivpLOKapiTOV dv-qXiov —^<^ — ^w — v^v^ — v^v^ dvqveixov re irdi^TCOv )(^eifj,coi'coi^, 'lv ^aK^m- Tas del Alovvq-os e/ipaTevei (or--) 680 Oeais d/jLCpLTToXcoi' TiOijuais ^ — —\^ y^ — ^ (or ^ -A Here hne 676 is a dactylic tetrapody, a verse not unfrequently interposed among lines of other types (e.g. it is the penultimate line of the strophe in Alcman's irapOkvLOv)^ and the line that follows— as if to restore the balance, or to make the two lines together a brief deviation towards neTpov kTnavvdeTov — has no dactyl at all. 1 A Phalaecean line, if this division of the lines is right. The pen- ultimate line also is Phalaecean. ^ = xP^<^«'''os 'Acppodira in the antistrophe, O^ais is one syllable. v_/ — v_/ — v_/ — v^ w — '— — v_/w —\^ — w — ■ 176 GREEK LYRIC VERSE il It will be observed that the line quite frequently ends in the middle of a word. There are passages where this occurs throughout or nearly so, e.g. Eur. Io7i. 184-7. A less regular group of lines is seen in Soph. AJax, 1217-22 : yevoL- fxav 'iv v- <^ l_ — ^ (_ XoL^v 'iirecTTL ttovt- — w w — w i_ J i, Of Trpo^Xrjfi aXiKkvcTTOV olk- —^ — ww — v^ i_ pau vno irXoLKa ^ovvtov, —^ — ^ ^ — w — TCCS Upas OTTC09 —WW — w — TTpoaeLTTOi/iep 'ABdvas ^ ^"^ (or i_ — a) where only the third and fourth lines are normal Glyconics. The combination of Glyconic with Pherecratean was some- times treated as a line or (ttl\os and written Kara («/'- - — w w — w i_ pe^? Kd^aTTaTd)[X€vos, irpos Tov 76 XkyovT del K€)(T]pa9, 6 I'ovs Se aov w irapcbv d7roS7]/jL€L — —WW — w — — — WW — w — w w — w — ^ -^^. _^1 ^ The strophe is repeated four times. The short syllable at the GREEK LYRIC VERSE 177 Of these lyric systems some lend themselves more readily than others to the now fashionable quadrisyllabic method of scansion. Any Glyconic line of course does so (^ ^ - w : v^ — v-^ — ), once we are prepared to regard two short syllables coming together as belonging to different sections of the verse. eviTTTTOV, ^eve, ToicrSe xcopa^ ktX. fits the scheme fairly well, beginning with a heavy 'antispast' (the first syllable long) ( ^ I v^ — w -, evLTTTTOv, ^e|^€ TccaSe X^")' ^^^ ^^ ^^ ^^^' turbing to find a line of four unmistakable dactyls (676) ; 0) 8rjfj.e, KaXrju y '^X^'-^ ^^ ' acephalous ', the first syllable of the antispast disappearing ; the clausula -wep dvSpa rvpavvov remains rather obscure, as does the first syllable short in Ke)(r}va9, 6 i^ov9 Si v pyd/xcop Kar* ivoirkiov x,u>7rolos av Kara daKTvXoi^. What exactly does it prove ? To argue, from a passage in comedy is never a very safe thing to do, and here the chances are that what Socrates thinks so important is some preposterous piece of hair-splitting comparable in value to the measurement of yl/'uWcov i;^^??. A dactylic lyric like « noXvfxoxdos "A/j^ys-, n' tto^' aluciTL in the Phoenissae could presumably be scanned kixto. haKxvkov or Kara fxouoTToBlai/, and so could a hexameter (though a heroic hexameter could also be taken as two KcbXa divided and linked together by caemra). But a line like (paivojjLevav S' op' €S nrav aTvevbev o/xiXof iKecrOai is neither of these things and not to be treated in either way. You must take it as— what? Something not dactylic at all .^ Not necessarily. It runs kgt ivoTvKiov. But Aristophanes would have written as he did even if an evoTrXios was a group of three dactyls (the" third spondaic in form). It is true that Ka6' eua daKTvXov or KnTo. daKTvXou eKaarov would have made a more logical contrast. That is the precise point at which it becomes unsafe to argue from a comic poet's phrase. The passage does not really prove anything about the internal constitution of an evoTvXios in the fifth century. The Scholiast on the passage and ancient writers on metre, taken together, give us every possible view of the group of syllables — WW— v^"^ : there are four disyllabic feet (— w | w — | w w ] ) or a choriamb and an Ionic, or the group is ro exov 6vo daKTvXovs Koi eva aTTovbelou. It is also said that two of them make a i^afxerpov Kar euorrXiov — the e^afxerpov, perhaps, which Strepsiades is to distin- guish from a different sort of hexameter. i82 DACTYLO-EPITRITES, The next witness is Plato, in a well-known passage of the Republic, 400 B : ol^ai ^e yue aKTjKOivnt ov ancfias ivoTrXiov re rivn ovo}xn^ovTos avTov ^vvOerov Kai dciKTvXov koi rjpQ>6u -ye, ovk oida ottms dtaKoa/xovPTos Kai icrov iivo) Kai kutco ridevrns, (Is jSpa^v re k(u fxaKpou yiyvo^ievov, Kcii, MS eyo) oijjLai, 'Lajji(3ov Kai tiu' aWou rpox^mov MVo/jLa^e, .^fJLTjKrj Se Ka\ l3paxvTy]Tas irpocrrjnTe. This passage, like that in Aristophanes, is not one from which it will be safe to draw very precise inferences. It is casual in tone ; the speaker is waiving aside the technicalities of the subject. So far as it does go, it does not seem to afford any very tangible support for the choriambo-ionic enoplius. ' 'i(Tov aiw Kai /cttro)' is a phrase that belongs to dactyls and anapaests, the yeuos 'icrov (w w = — ). The equality between a choriamb and an Ionic is not very likely to be in the speaker's mind (and which of these is avco, which /carco ?). It is simply dactyls (or dactyls and anapaests) that he is thinking of in the first part of the sentence. Then he' goes on to speak of iambi and trochees. An enoplius is something dactylic, one oi fhfee things. But we have already had three things in view : the dactjds of the Phoenissae w irokvp.oxOos "ApT]Tr]v Se;!^o/xeV/?s'. What exactly does this passage tell us ? Archilochus wrote an often-quoted line ^Epaapovi^q ktX. In Archi- lochus himself there was always a Topr} (caes^ira or diaeresis — the end of a word) before the last six syllables, which thus were trochaic (— >^ — w ). But later poets (a very vague phrase) were not careful to have the Tvpi) exactly there. Cratinus, for example, wrote the three lines x"'^p' ^ H-^y' '^^'^•j only the third of which is like Archilochus's line — lKpL(t>v ■^6— ) the symbol for it. The words with which the lyric of Seikelos (first century of the Christian era) begins, oaov Cfjs, are w 1, not v^ i_ — . This is not only shown by evidence. It is in itself reasonable, if iambic verse is to be separated from trochaic. The nineteenth- century metricians read iambic verse as trochaic with anacrusis, and consequently missed, or abolished, this effect. If a line really is iambic and begins with ^ — \^ — , we cannot take this as '^ — WL_. The second foot is an iambus; ^ i— is not intelligible at all. The syncope must be found in the following syllable. TTPoal §' OTTO '2Tpvixuvos jJioXovaai is (^ — w — — i w — v^ 1, it is not wj— ^-^i — \ — \j — \u i — — A. The first syllable of ^Tpvfxovos, not the last of dno, is prolonged. It is prolonged to the length of an iambus, so that the iambic movement can start again or proceed with -/jloi^os fxoXnv-, and then the last syllable is protracted (the Seikelos inscription and the hymns seem to attest its protraction rather than the completion of the time by a pause). Here then is a definite difference, and a quite important one. We cannot argue about it as we did about *odi profanum '. Iambic and trochaic scansion give a notably different effect ; fivn^ b' 6 TTfjea^vs \ rod' etVe (f)u)vu>i/ tiapfia pev Krjp \ ro pr) nLdeaOai. SYNCOPATION AND ICTUS 19 r It is certainly very different. It divides tiiese lines conspicuously into two equal halves. Again, in S. c. Th. 293 we should have : I — I — I .... vTTfp8idoiK€u Xexnicop dvaevvdTOjms. Must we accept the principle as final, and learn to apply it and realize it throughout in the text of Aeschylus ? And in all other poets who write anything that can be construed as iambic ? ^ Perhaps. But let us at least contemplate carefully the limits of our knowledge and run no risk of being too readily credulous. Aeschylus wrote about a century and a half before Aristoxenus. We' do not know what kind of metrical and musical notation, what terminology or ideas, existed in his day. They may have been very rudimentary. Grammatical terminology can hardly have existed at all. It is still incomplete in Aristotle. Yet Aeschylus writes correctly ; there is very seldom anything that is obscure or confused ; it is very seldom that we find anything like iBpe'^ar olKi]Trjp(is . . . ttkttovs ottcos yevoicrBe rrpos Xpf'o? To^e. Further, Aeschylus and other ancient writers sometimes have things of which modern grammar can give a better and clearer account than Greek grammarians could do.^ It is possible that if we could summon Aeschylus as his Persians summoned Darius, and put questions to him, showing him modern musical notation for some of his verses, he would say, ' That expresses very much the effect that I intended, and expresses it more neatly and completely than the notation that I knew on earth.' Tp/fierpoi/ and TerpajLterpoi/ would be current in his time, probably, as names for the iambic trimeter and trochaic tetrameter. He would think of the former as consisting of uifi(3oi. What words would he use to describe the structure of a lyric like nfoal 5' aKo ^rpvp-uvos poXoDo-m ? When he composed it with its music, would he divide it into bars and use symbols that indicated quantity, or only put musical notes above ^ Mr. White, if one may argue from his notation, is not an ardent propagandist. In his metrical schemes he either does not indicate the protraction, or he puts a dot to indicate the missing time or syllable : €;raux7cras Se rots aois \6yois ^ — • — • — w — • — \j — (§ 73). This nota- tion does not make it obvious to the eye which syllable, the preceding or the following one, is to be prolonged. ^ For an illustration see supra, p. 132. 192 DACTYLO-EPITRITES, syllables ? If he thought of it as consisting of ' feet ', it is possible that he would use the word x^^P^^^^i which is certainly an early word and would be a convenient word for short ph^s long in either order, su — or — w, or for lyric composition in alternate ' longs and shorts '. If we try in this way to detach ourselves from notation and to imagine Aeschylus's procedure, further questions begin to suggest themselves. If a line began w — w . . . (what we now call an 'iambic' beginning), did that necessitate 'iambic syncopation' throughout the verse and at the end of it ? To illustrate the point from Aristophanes, in the lines : Xpr](rifinv fxev ovhev r»A- Xcoj Se beiKov koX fxeya we have no doubt trochaic syncopation, but in ^}]rovvT€s i)uiK' av av m- Kas \eycov rhs diKns we are to recognize iambic syncopatior. It is quite certain that Aeschylus did not know the term anacrusis (which was invented by Hermann), but it is probable or all but equally certain that he did not know 'acephalous' either. Is it not possible that he would think of the rhythm of a line ' iambically ' in one place, ' trochaically ' in another, and thus be free to make eitJiei' of two contiguous long syllables a protracted one ? When this possibility has been stated it opens up a further view. Can it be the real explanation of things which the nineteenth- century metricians explained by applying the method of modern music, with its invariable anacrusis and descending scansion ? ^TpvfjLovos jjLoXovanL is a very familiar phrase — not iambic. Can it be that the line began as iambic and ended as trochaic ? The iambic trimeter could easily be regarded as doing that, with its caesura : evdyyeXos [xev, \ coarrfp t) napoip-in, €a>s yei'oiTO | prjrpos €v(f)povi]i irapn, ' coanep T] nnpoipia ' presents itself to the ear rather as trochaic than iambic.^ * So in the scazon a trochaic movement may be said to ' set in ' after a marked ' caesura ', leading to the ending — v^ ' - [ — i=i : Suffenus iste, Vare, | quem probe nosti. SYNCOPATION AND ICTUS 193 mea renidet | in domo lacunar was actually described in later times as an iambic followed by a trochaic group. 'in domo lacunar' is the same phrase as 'Erpvi-iouos fjioXova-ni. l( Tri'oai 8' u~6 answers to ' mea renidet', would not the structure of the line demand that the suppressed or absorbed time should be absorbed by the first comma or group of syllables and not by the second ? In a similar way the second part of a hexameter could be thought of as anapaestic. The paroemiac {p^X^kv hi re vr^TTios eyvw) coincides with a catalectic anapaestic dimeter. In later times at all events there was no shrinking from such analysis. When Aristophanes was composing anapaestic tetrameters, did he in bringing a line to a close stop to reflect whether it was dactylic or anapaestic ? tJkktt^ aXX' ovpdviai vecpeXai, fxcydXat, Oeai dphpaaip dpyols. It is a hexameter from ovpdviai onwards, and we shall be justified in waiting for very cogent evidence before we begin to contemplate the scansion : Xe^o) Toivvv TrjV dp^aiav rrniheiav o)? 8Uk.(ito (or ^ /\ ?) ot' e-yo) TCI dUaia Xiycov rjvQovv Kai acocfypocrvvri ^vevopLcrro. The provisional conclusion or contention which we seem to arrive at from this discussion is that, while iambic syncopation appears to be a proved fact, we are not yet prepared to admit that all verses which appear to have an iambic beginning were thought of as ' iambic ' by the poet — we do not know that Aeschylus would have called his lyrics either iambic or trochaic — or were thought of as iambic throughout, for this purpose of syncopation. The question of reading lyric verse involves the question of t'c^us, which we postponed. Mr. White follows various metricians of recent times in denying that there was any voice-ictus or stress-ictus — a voice-ictus can hardly have been anything but a slight stress — in ancient Greek verse {V. of Gr. Coi?i., Intr., pp. xxiii, xxiv, § 28). There is no ancient evidence for it, only for beating time with the hand or wand or foot. He quotes M. Kawczynski : ' Or il me parait inadmissible de faire executer aux anciens par la bouche ce qu'ils faisaient avec le pied.' But Greek poetry was far more often choric or supported by music than ours is. Mr. White has told us 1P87 Q T94 DACTYLO-EPITRITES, that no metre but the iambic trimeter was delivered in yj/iXr] Xe|ty. Such beating of time was far more familiar to them than to us. But what if there is no conductor's wand or dancer's foot available or obtainable? Must not sojnething be done 'par la bouche ' to replace it ? Does a long string of short syllables become necessarily amorphous, scopae solutae, unless we can beat the floor or move a hand, unless we can make a movement which the hearer can hear or see ? What if he is in the next room and cannot see us ? Does that make it impossible to read to him resolved anapaests : Ka and > O w, not equivalent in time to the true anapaest and dactyl, for the long is an irrational, not a normal long. Whether we mark both short syllables with the ictus or only one of them (o o — or O w — ) seems unimportant. For reading, it would suffice that the very slight stress required should be heard on the first syllable ; both syllables would accom- pany the downward stroke of a conductor's wand or of the foot. The structure of Greek verse was given by quantity, and when such verse was sung, with music, or gesture and movement (axfjpaTa), or both, there would seem to be no need for a stress of the voice also. In that sense we may admit it to be true that ' there was no ictus in Greek verse '. But when these adjuncts are SYNCOPATION AND ICTUS 195 removed, either icins is soniethhig audible or the structure of the verse must be lost.^ The bearing of this on the main subject of this excursus is to strengthen the contention that the lonic-choriambic-diiambic- ditrochaic scansion would be impossible for a reader, ancient or modern. 1 In Latin the whole question of ictus becomes a different one. The conditions were different in at least three ways: (1) The Latin accent was different from the Greek accent, stress not pitch, and so more like what ictus had to be when given by the voice ; (2) it was regular in its incidence on words, following a uniform and simple law; (3) the Romans more often wrote verse for readers. 2 CHAPTER II THE HISTORY OF METRE AT ROME The history of verse at Rome falls into three great phases or stages, of the first of which we have only a dim and distant view : (i) Saturnian verse, (2) verse of the Republican period, (3) verse of the Imperial age (some of its features disclose themselves before the end of the Republic). To these might be added a fourth, the later time when accent indubitably asserts itself as a principle of verse composition. But the latter of these periods is beyond the scope of this book ; it connects itself with the Middle Ages when verse could be written like ' dies irae, dies ilia ' {' dies ' replacing ' dies ') or like ad Mar6nis Mausoleum ductus fudit super ^um piae r6rem lacrimae (where ' -eum ' rhymes with ' eum ' and presumably had a similar sound). Ancient Greek verse is consistently quantitative throughout, as far as we know it (till accent begins to assert itself in the first or second century after Christ). At a quite early time it has shaken off alliteration. There are alliterative phrases occasionally in Homer, such as vrirjcras €V vfja? or to yap yepas ecrW yepovTcou, and proverbs show a tendency to it (kukov KopaKos KUKou coov), but there is nothing at all resembling the alliteration of Old Italian or Old English HISTORY OF METRE AT ROME 197 verse. Further, if there was a period when the less con- spicuous part of a foot was ' free ' and might be one syllable or two (— or v^ or ww), the earliest verse we know has regulated very strictly such variation. There is hardly any question of a trochee in an epic hexameter (though a beginning like TToXXa \i (> — , a spondee which could be combined with iambi ; — >, a heavy trochee or trochaic spondee). Or the time may have been i~ : i^. OF HORACE 233 iambus. In trochaic verse the converse is seen. It is the last syllable of the dipody that may be heavier ; it is the first foot that must be a pure trochee. If we try the experiment of inverting the order in either case, we shall see that this is inevitable. If the second iambus were allowed to be heavy, the dipody would present the appearance of v^ , a short syllable and three longs, an amorphous thing. So, if the first trochee were heavy, the syllables would be ^ , again a thing that has no obvious rhythm or easily recognizable structure. The trochaic dipody, with this option in the last syllable (- w —^) is an element in many forms of verse. We shall meet with it again in lyric measures, and shall there find that a further step was possible — to enact that the last syllable of the four f/ms^ be heavy or long. It remains to describe briefly the chief forms of Horatian iambi and their characteristics. In one poem (£j>. xvii) he uses the trimeter throughout, line after line {Kara (rTL)(^ov), with the normal option of a spondee in the first, third, and fifth places.^ In the first ten epodes he uses the same line as an alternate one, putting after it the shorter dimeter (of four feet, quater- narius). Such a line was called an €770)56?, whence the name epodi for the whole collection of pieces. (a) The caesura is usually conspicuous and highly normal, in the third or fourth foot : 1 In lines 11-12 of this poem Horace has a metrical effect which recalls the older tragedians, Ennius or Pacuvius : unxere matres Iliae addictum feris alitibus atque canibus homicidam Hectorem. Several of the old tragedies were on subjects taken from the stor}' of Troy. Though Horace speaks of them elsewhere with no great admira- tion, thinking their versification heavy and cumbrous, the mention of Hector recalls them to his mind, and in the second of these lines, without violating his own stricter canons, he writes a line in tlie manner of Ennius. 234 THE LYRIC METRES beatus ille | qui procul negotiis. minatus urbi vincla | quae detraxerat. (13) A tribrach is admitted, i.e. the long syllable may be resolved. When the long is so resolved, the two short syllables must be in the same word : pavidumque leporem at advenam laqueo gruem (observe that the ordinary accent on the first syllable of ' leporem ' helps the effect, coinciding with the idus of a resolved iambus ; ^ — becomes v^ ^ w). Thus in the line aut amite levi rara tendit retia metre at once determines that the words are * amite levi; for the scansion 'aut amite levi' would divide a tribrach wrongly. (y) There are a few instances of what seems to be an anapaest in the fifth place (so laqueo in a line just quoted) : priusque terra sidet inferius mari. nunc gloriantis quamlibet mulierculam. But all of them are open to a doubt : it is possible to pro- nounce ' inferyus ', ' mulyerculam '. It is not clear that Horace meant to admit an anapaest in the fifth foot at all. (5) Horace's iambi do not conform to the rule known as ' Porson's Canon '. The rule is that if the last word in a line is in form a ' cretic ' foot (- w - , Tuscuh), and if the preceding w^ord is one of more than one syllable, then the syllable immediately preceding the 'cretic' must be short. Thus the rule is not broken in the line : furorne caecus an rapit | vis acrior but it is broken in lines like : non ut superni villa | candens | Tusculi. OF HORACE 235 Such lines are very rare in the Greek drama, and where they do occur, attempts have been made to remove them by con- jectural emendation.^ When the rule is violated, the line is weighted towards the end, an effect which seems to have pleased the Romans. It is used by Horace in a special way, to weight the end of the couplet. In the first ten epodes there are 366 lines, 183 trimeters, and 183 dimeters. In the trimeters the Porsonian rule is broken seventeen times ; in the dimeters — that is, at the end of the couplet — seventy-four times. In the trimeters the effect occurs on the average once in eleven lines, in the dimeters nearly twice in every five lines. '^ (e) If we consider the number of lines in which the penulti- mate foot is a spondee, we find a similar result and a similar contrast between the trimeter and dimeter. Out of 183 trimeters 97 have that ending; of the 183 dimeters only r6 (or 15) have not a spondee in the third place; as many as 167 are spondaic. But even in trimeters Horace has 1 A famous instance is the first line of the Ion of Euripides : "ArXas o xaXKiOiai vwtols ovpavuv emended by writing vuitois xciA/f€ oto'i*'. In . . . 'Apiofxapdos ^a.p8€(Tiv, Aes. Pers. 321, a proper name justifies the deviation from rule. 2 As against these seventy- four cases, there are only six certain examples of the type : pernicis ux6r Apuli, where the line ends in a word of three syllables, and the Porsonian rule is not violated. Therefore in the line v. 100 : et Esquilinae alites the probability is quite 12 to i that Horace meant ' Esquilinae alites', i. e. there is no shortening (a thing which belonged to dactylic verse in Greek, not to iambic), but rather what is seen in Virgil's line Glauco et Panopeae et Inoo Melicertae (where Panopeae is an instance of what it is not safe to assume in Horatian or high!}' hellenized iambics). 236 THE LYRIC METRES a penultimate spondee more often than the Greek tragic poets ; the percentage in his verses is 53, in theirs about 40.^ Roman tragedy, we shall not be surprised to find, is heavier still in its penultimate feet. In Seneca's tragedies the spondee seems to be compulsory (with the alternative of an anapaest) ; examples of iambi in the fifth place are very rare and doubtful. One other form of iambic verse remains to be mentioned — the ' catalectic ' line, which is the second of the couplet in Odes ii. 1 8 : non ebur nee aureum mea renidet in domo lacunar. This couplet is a ' homogeneous ' metre, though the first line is a trochaic (-v^ -w -w -a); a trochee is a foot of the same kind or type as an iambus, its parts being in the same ratio of 2 : I. But it is not certain that the first line would be thought of as trochaic. It was called EvpiTrlSeiou (or Xt^kvOlov sometimes, from the scene in Aristophanes, Frogs 1200 f.). It was part of an iambic line, from the penthe- mimeral caesura onwards. Further, it is a question whether iambic or trochaic scansion is preferable for the second line. Taken as trochaic, with a preliminary syllable or ' anacrusis ', the line is : This scansion gives the more intelligible account of the closing cadence ; the penultimate syllable is one of three times (= v^«^v^ or = a whole trochee), and the last syllable is part of a trochee — a trochee catalectic.^ The line, how- ever, begins with an iambic movement, and it is not impossible that Horace thought of it as made up of two parts : ^ This is based on a survey of about 500 lines of tragedy. 2 The last S3'llable of lacunar is short in prosody, but the last s^'llable of a line is a syllaba anceps, i.e. may be reckoned metrically as long. OF HORACE 237 mea renidet | in domo lacunar.^ ' mea renidet ' is the first part of an iambic line, the two-and- a-half feet that come before one of the chief pauses, 'in domo lacunar ', — ^^ —k^ —^ or — w — w i_ — a is a well-known metrical group or phrase ; and it is a phrase which Horace appends to a group of dactyls in an ode (i. 4) in which the second line of the couplet is the line we are now considering. It is only in these two odes that Horace uses this metrical phrase ; the other ode is not 'homogeneous' and belongs to Section HI."^ iii. Io7iic, The only other ' homogeneous ' metre written by Horace is the Ionic verse of Odes iii. 12. This is so regular and simple in form that little need be said of it. The foot is the lonicus a mi?iore, w w , and the poem consists of four stanzas, each containing ten lonici. Within the stanza the grouping seems to be 4-f4-}-2, but this is more doubtful. The end of a foot is usually marked by the end of a w^ord, an arrange- ment which has a somewhat monotonous effect. Catullus used this metre in a very different way in the 'Attis' (see p. 163). ^ This view is found in Terentianus Maurus : sequens epodos | cum parte iambi tres habet trochaeos. ^ The pause after the fifth syllable is quite distinct in ever)' line of i. 4 (' regumque turres. o beate Sesti '). It is a curious fact that Horace's Hne wants only the addition of one iambus to make it a normal Saturnian • mea (quidem) renidet | in domo lacunar. dabunt malum Metelli | Naevio poetae. Perhaps the horridus numerus of the carmina Livtj enforced by the cane of Orbilius, made a deeper impression upon his mind than he cares to admit. It may be noted also that there are a few Saturnians which have as their first part syllables answering to non ebur nee aureuin (in Naevius, 'immolabat auream'; 'qui suis astutiis* (— v^ w — ) in t e verses o he 'Collegium Coquorum'). 238 THE LYRIC METRES III. Archilochian or Parian Metres {fxerpa kirLa-vvdera) Archilochus of Paros, in the seventh century B.C., wrote both iambic poems and elegies. Horace is a professed follower of his, and claims to have been the first to present his ' iambi ' to the Roman reader : Parios ego primus iambos ostendi Latio, numeros animosque secutus Archilochi. {^PP- i- i9- 23.) The claim at first sight seems a little unjust to Catullus, who certainly had something of the spirit or 'animi' of Archilochus. But as regards the 'numeri' it is strictly justified. Catullus writes the 'scazon' mainly (which belonged to Hipponax, not to Archilochus), very seldom the ordinary trimeter, in more than one poem the rigid form of verse which admitted no spondee, and nowhere the combination of trimeter with dimeter in a couplet, which Horace uses regularly in his Epodes. But Archilochus not only wrote iambic and elegiac poems separately ; he also combined the two forms of metre in the same verse or couplet, and here, too, Horace probably had a valid claim to be the first to introduce Archilochian 'numeri' at Rome. Greater variety, piquancy, or complexity was given to a verse by combining in it different metres. But what Archilochus did was not to fuse or mix up different types of feet. To put a dactyl and a trochee in the same group or KooXou was a thing which belonged not to him, but to the Aeolian or Lesbian school of poets. What Archilochus did was to set down a group of iambi or trochees beside a group OF HORACE 239 of dactyls. The groups were internally homogeneous ; it was only the whole verse or couplet that combined unlike elements. The simplest example of such an arrangement is a hexameter followed by an iambic trimeter [Epode xvi) : altera iam teritur bellis civilibus aetas, suis et ipsa Roma viribus ruit. But the groups combined may be shorter than these, and the transition from one movement to another more frequent : Kj — \j — — — \^ — nunc mare nunc siluae nivesque deducunt lovem. (iambic dimeter followed by a dactylic tripody catalectic). Here the reader may be disposed to ask : ' Is this last example a verse made up of two grotips or Ka>Xa, or is it two verses ? What do you mean by verse and by kcoXou ? ' One thing that can be said with some certainty in answer is that for a group, phrase, or kcoXov there was an upward limit of length ; e. g. for dactyls the limit was four. A series of more than four dactyls is not one phrase but two.' Is there also a downward limit for a ^ verse ' ? ' Verse ' is a looser term, and perhaps not much would be gained by making it a strict one. A simple kooXoj/ or phrase may be a ' verse ', e. g. it would be inconvenient to be debarred from speaking of an iambic dimeter as a ' verse '. The phrases nivesque deducunt lovem nunc mare nunc siluae are long enough to serve as separate verses. They may be written in one line if the poet or the editor wishes to suggest that they are to be taken very closely together, with only a slight pause between them. Another question is possible : 'Is there a downward limit for a kooXou?' e.g. do the Izvo ^ A hexameter consists of two tcojXa, not exactly equal in length, linked together by the device of the caesura. 240 THE LYRIC METRES dactyls in the last line of an Alcaic stanza make a kooXou ? But this question does not yet concern us ; for in his Archilochian metres Horace uses no phrase shorter than three dactyls or four iambi. The Archilochian metres present no great difficulties or problems. We may classify them by taking first those in which there is least complexity : (i) A dactylic hexameter of the rapid, lyric type described in Section H, followed by an iambic trimeter which also moves rapidly, having in it no spondees. Ej>ode xvi. There are no elisions in the hexameters^ and in the iambi they are very few and slight. (2) Epodes xiv and xv. Hexameter followed by iambic dimeter. The dimeter has the characteristics already noted as belonging to it when it follows an iambic trimeter ; a liking for a spondee in the fifth place, and ready admission of an ending like ' adhaerens bracchiis ', which does not conform to the Porsonian rule. (3) A hexameter followed by a line of which the second half is the second half of an elegiac ' pentameter ', while the first half is an iambic dimeter. They are ' halves ' in time or rhythm, for an iambic dimeter and a dactylic tripody alike have twelve morae or tempora (each is a ScoSeKda-Tjjjiou or 8(o8eKdxpovov fieyedos). We have seen that each is long enough to count as a separate verse. Further, they are written as separate verses; for at the end of the first (as is not the case in an elegiac pentameter) a syllaha anceps is admitted {Epode xiii) : reducet in sedem vice. I nunc et Achaemenio levare diris pectora | sollicitudinibus findunt Scamandri flumina | lubricus et Simois. Like the dactylic tripody catalectic in Odes iv. 7, the second OF HORACE 241 KcoXov shows no preference for the Ovidian disyllabic ending. Out of nine lines, five end in a word of three syllables ^'siliiae', 1. 2 ; 'genua', 1. 4, &c.). In one Archilochian metre the iambic part comes first : (4) An iambic trimeter (admitting spondees in the usual places) followed by a dactylic tripody catalectic and an iambic dimeter. Some of the lines can be construed as a complete dactylic tripody followed by a trochaic dimeter catalectic i^Epode xi) : — KJ \J — \J w— — — \^ — — — w — A fabula quanta fui, con viviorum et paenitet. But that this is not what Horace meant is shown both by syllaba anceps : Inachia furere | silvis honorem decutit and by hiatus :* fervidiore mero | arcana promorat loco. All the dimeters but one have a spondee in the third place, and several of them are also non-Porsonian. (5) In Odes i. 4, the first line consists of a dactylic tetra- meter, followed by a trochaic kS)\ov of three feet (originally no doubt of four, but we cannot be sure that Horace meant it to be read so) : — w— w or —^ — ^ \ — — A. The second Hne is one which has been alre§.dy discussed (p. 236). In the first line no hiatus occurs between the dactylic and the trochaic part ; the two parts are not ' asynar- tete ' or detached, as in the case in the compound lines of Epodes xi and xiii. 1887 R 242 THE LYRIC METRES IV. Metres of the 'Aeolian' or 'Lesbian' Type (jikrpa jXiKToi) When a dactyl and a trochee occur in juxtaposition within what is undoubtedly one phrase or kcoXou, we have a type of metre clearly different from the two preceding kinds. The KcoXou is now internally heterogeneous.^ Verse of this kind was probably as old as Archilochus, and perhaps much older ; for it would seem to belong to quite primitive minstrelsy to make the less conspicuous part of a foot consist indifferently of one short syllable or of two. But it is in the Lesbian poets, a little later than Archilochus, that it first appears in an artistic shape and finds a place in literature. This Lesbian or Aeolian verse perhaps retained one feature of its origin ; some forms of it began with two syllables which seem to be subject to no restriction ; they may be or w — or — w or w w — or even w w . This so-called ' basis ' or ' basis Aeolica ' is in Horace strictly regulated. It is uniformly a spondee.^ But, it may be asked, how are we to draw a line between this type of metre and the ' Parian ' type ? This is the question which we postponed — how skorf may a kS)\ov be ? If tivo consecutive feet are alike, are we to regard them as a kS)\ov aad so make the metre Parian ? Perhaps it is 1 The term ' Logaoedic ' for metre of this kind has fallen into dis- credit, but some designation is wanted for the class : juiktov fxerpov is a convenient and simple term. It is not to be understood as meaning that the feet were of different length. When a dactyl was used along with trochees it may have been more rapid, its time equivalent io kj \j \j. 2 With one exception in Odes i. 15. 36 ('ignis Iliacas domos': it is difficult to believe that the prosody of ignis could be ignis and not ignis). Perhaps an early ode, written before Horace had finally fixed his metrical scheme. OF HORACE 243 best to postpone the question again, or dismiss it with only an approximate answer ; it would be fairly accurate to say that at least three feet must be alike, to be treated as a separate KcoXov. We have seen that no kooXou in Horace's Parian metres is less than that. We may approach the matter in a different way by enu- merating some of the elements, or groups of syllables, which are met with in the non-Parian metres. (a) There is first the ' basis ' just mentioned, in Horace always — (found in Asclepiadean, Glyconic, and Pherecratean lines). That it should be separated in some \Yay from the rest of the line seems to be indicated by the occurrence of lines in which it is absent : Sevri vvv djSpat XdpLT€9 KaXXtKO/iot re Movaai (which is otherwise very like ' nullam, Vare, sacra ', &c. Odes i. 18). (b) Another important element is the trochaic dipody, — w — ^ or — v^ . The second of these forms, with the last syllable long, is, or coincides with, the Pindaric 'epitritus', and it has been suggested that in the third line of the Alcaic stanza Horace has a Pindaric effect in view, though the metre he is writing is a very different one.^ (c) With this we may associate — - w , that is, epitritus with anacrusis, or the first two and a half feet of an iambic trimeter {lafi^iKou 7rev6r]iJLLiiepes)' (d) The syllables -v^w-, sometimes reckoned as a 'chori- ambus', are also a dactylic dimeter catalectic. (e) — v^ v_/ — v^ — may be described as an Aeolic tripody. In Odes i. 8, it is longer by a syllable : 1 The ^epitritus' in Pindar is regularly used in metres of the * Parian ' type. The name ' epitritus ' comes from measuring the length of its component parts, — w and (y>Kj\j and v-/ o* w v.y), 3:4. Adjectives compounded with kiri- express the ratio of « + 1 to n {knoySooSf 9 : 8, &c.). R 2 244 THE LYRIC METRES Lydia die per omnes, a line or group which originally and musically was probably a tetrapody, — wv^ — w i— — a. Here it occurs at the begin- ning and end of a couplet, with (b) + (d) intervening hoc deos ve;re Sybarin, — v^ |~^ ^■~• These are nearly all the elements that are required for the construction of Horace's non-Parian metres, (e) may be thought of as a kooXov, but for other and shorter groups of syllables Caesius Bassus, who wrote on metre not long after Horace's time, supplies another convenient word, Ko/x/ia, incisum ; KOfM/xa is a block or cut-off portion — (c) is a portion of an iambic trimeter. It is probable that Horace himself thought of such commata and that he was guided in so doing by a metrical theory to which Varro had given currency at Rome. According to this theory the primary metres were the dactylic hexameter, the iambic trimeter, and the trochaic tetra- meter. Other metres were supposed to have been arrived at by combining portions of these. Thus the 'hendecasyllabic' line, which Catullus uses and Horace does not use, could be regarded as consisting of a part of a hexameter + part of a trimeter : cui dono lepidum novum libellum. But if this was its nature, the first two syllables should both be long, for an iambus or a trochee cannot form part of a hexameter. Hence, perhaps, Horace's regular spondaic 'basis'; so, too, in Martial and Statius a hendecasyllabic line always begins with two long syllables. It can hardly be doubted that the influence of this theory is one of the causes why a Horatian line is often so different in effect from the corresponding line in Sappho or Alcaeus. The Greek line runs lightly as a single whole, whereas Horace's line is made up of two co7?i??iata. OF HORACE 245 The metres that we have now to consider fall under the three heads of Alcaic, Sapphic, and Asclepiadean, the last of these carrying with it the similar but shorter Glyconic and Pherecratean lines. i. The Alcaic Stanza. The Alcaic stanza is constructed, out of the elements specified above, in the following way : The first two lines consist of (c) + (e); the third line of (c) + (b) ; while the last line is an extended form of (e), beginning with two dactyls instead of one, and completing its last foot, instead of being catalectic. The effect of the whole becomes more apparent if instead of (c) we write (b) with anacrusis. The stanza then runs ^ (b) + (e) ^ (b) + (e) ^ (b) + (b) E The fourth line is a larger form of (e), without anacrusis. Thus the third and fourth lines repeat on a larger scale the movement which is found in each of the first two lines ; using simpler symbols we may describe the stanza as ab ab AB. It is this structure that gives to the third line its peculiar weight and significance ; it has often been observed that the effect of a Horatian Alcaic stanza depends largely on that line. It is a weighty line, similar to Pindar's ' epitrite '^ with the fifth syllable always long. In the first two lines, we may say, this movement maintains itself for about half the line ; then a dactyl breaks in upon it, but without maintaining itself— a trochee follows ; in the third line the ' epitrite ' movement holds out twice as long, the dactylic invasion is staved of! longer ; then in the fourth line the dactyl rushes in in greater force — this time there are two dactyls instead of one, and then 246 THE LYRIC METRES trochees reappear in a somewhat fuller form than in the first and second lines. To recognize this as the real movement of Horace's stanzas involves ruling out certain views of it which have sometimes been held. We must refuse to believe that the first and second lines end in two dactyls.^ And we shall not be inclined to accept a ' choriambic ' or ' Ionic ' scansion, which makes the stanza end thus : sacerdos — W W I — v»/v_^— j w — — virginibus puerisque canto. The chief features of Horace's stanza, as compared with Alcaeus's, are that in the first two lines he has a diaeresis at a fixed place, and that he makes the fourth syllable of the group - w - J^ regularly long.^ The first two lines thus ^ If Horace himself thought that they did, which is perhaps not impossible, he must have held also that a long final syllable could count as short. But it is difficult to believe that he so thought or felt about ne forte credas interitura, quae longe sonantem natus ad Aufidum . . . A syllaha anceps in the sense of a short syllable that counted as long is common enough. The converse is more difficult and doubtful. A final dactyl belongs properly to lines which are * hypermetric ', e. g. : TToWaKL 5' ev KopvcpaTs opiojv oKci Oioiaiv ddrj iroXixpafios kopra kt\. (Alcman.) (Compare the verses of Ibycus quoted above, p. 123, where the final dactyl occurs in a group of lines meant to be hypermetric, and not in a contiguous group.) The view that a long could count as a short, put jn circulation by Heraclides Ponticus or some other theorist, perhaps misled Seneca into writing a cretic where he ought to have had a dactyl : Oed. 449 f, in a passage of seventeen lines he has five heavy endings — vivaces hederas remus tenet summa ligat vitis carchesia. 2 From this rule Horace departs only once, in Odes iii : si non perirSt inmiserabilis captiva pubes. So rare a thing has naturally caused editors to doubt about the text, and OF HORACE 247 become more complex structures ; and an air of Roman gravitas is lent to the whole, which makes the verse an appropriate vehicle for the moral and patriotic reflections of Odes iii. 1-4. The diaeresis gives to the first and second lines an effect like that of the elegiac pentameter (which does not necessarily have dactyls in its first part, only in the second). Horace introduces it with great regularity ; there are only two lines in which it is wholly absent (i. 37. 14 'mentemque lymphatam Mareotico'; iv. 14. 17 ' spectandus in certamine Martio'); in three others the diaeresis is marked by the slight pause or division between the parts of a compound word ('antehac nefas deipromere Caecubum, i. 37. 5). In certain types of verse, such as the elegiac pentameter, the end of a kcoXov or Ko/i/ia or metrical section is regularly marked by the end of a word. Apart from such structures, a line or kcoXov usually tends to avoid any marked agreement between words and metrical elements. Thus a hexameter has its marked pause {caesura) within a metrical foot, not at the end of it; and the parts of a pentameter illustrate the prin- ciple, though the whole does not (e. g. ' lucida | sidera | nox ' is inadmissible). Coincidence between word and foot is perhaps most frequent in trochaic tetrameters : quaeque silvas quaeque lucos quaeque montes incolunt. to suggest emendations (e. g. peri'res, followed by vocative). But it is Horace's way to deviate now and again from a self-imposed rule, as if he said to us, ' I know quite well that the metre as Alcaeus wrote it admits something different, but I choose to make certain long syllables regular because it suits the Latin language, and makes the verse more uniform and therefore easier for the reader to appreciate'. Seneca probably took the same view, and thought of this comma as ^ — v^ — i:^. He has the fifth syllable short more often than Horace, when he uses the phrase as a detached brick or tessera in piecing together a lyric : Oed. 752 'effudit arma'; Agam. Q61 'Aurora movit ad solitas vices'; ibid. 916 ' latravit orS'. 248 THE LYRIC METRES In the third and fourth lines of his stanza Horace usually avoids it. A line like : Alcaee | plectro | dura | navis (ii. ij. 27.) is very rare, and Mr. Page points out that in five of the eight instances the penultimate word is repeated at the beginning of the next line : dura navis ; dura fugae mala, dura belli, and that in one of the other three a special effect of sound seems to be intended ('pronos relabi posse rivos'). The three long syllables of the line are very frequently in one word : audita Musarum sacerdos. In the fourth line the prevalent caesura is shown in virginibus | puerisque canto. The ' trochaic ' division of the second dactyl is much rarer — w I w interiore nota Falerni. One other question requires an answer — that of what is called synapheia {crvj^d(p€La, from (rvua^rj^) or 'hypermetron ', continuity of scansion between one line and the next. Virgil admits this effect in hexameters, but not very frequently : turres et tecta Latinorum ardua cernebant iuvenes. It was regular in anapaests ; and naturally frequent in any group of short lyric lines, for a break coming at short intervals would give to the whole an effect of disconnexion/ A line is ^ The continuity of utterance involved caused ' hypermetric ' systems to have in Greek the name npiyos. A good example occurs in Eurip. Ion 184-7 • oiiK iv Tais ^a6eais 'A6a- vais ei/ftioves ^crav av- Xal Oca/v /.lovov^ ou5' d7i;i- driScs Oepaneiai. OF HORACE ' 249 detached from the line that follows it when it ends with a syllaba anceps or when there is no elision between them. On the other hand, the absence of elision and of a syllaba anceps does not prove that a hypermetric effect was intended, though, if they are absent throughout a whole poem, there is a strong presumption that the poet wrote the piece or the stanzas of it on the ' hypermetric ' principle. ' Hypermetron ' is posi- tively revealed when there is elision or when a word runs on into the second line. In the Alcaic stanza Horace does not bind himself to a hypermetric structure. Mr. Page collects eighteen instances of hiatus or the absence of elision (all in the first three books— none in the fourth) ; there are about as many instances of syllaba anceps in the first book alone, e.g.: hie tibi copia | manabit quanta laborabas Charybdi | digne puer, &c. Unmistakable 'hypermetron' occurs chiefly between the third and fourth lines of the stanza. There are only two instances of elision : nos in aetern|um exilium impositura (ii. 3. 27.) delabentis Etrusc^um in mare. (iii. 29. 35.) In one instance the third line ends in a preposition : retusum in | Massagetas. (i. '^^. 39.) In eight it ends with ei^ preceded by elision : barbarorum et j purpurei. (i- 35- n-) Similarly, but without elision (once) : depone sub lauru mea nee | parce cadis, (ii. 7. 19.) Thus there are twelve cases of unmistakable or all but certain continuity ; to set against them, only one instance of a similar thing between the first and second lines : fuge quaerere et | quem Fors dierum. (i. 9. 13.) Catullus has the similar ending : saltuumque reconditorum amniumque sonantum. (xxxiv. 11-12.) 2 50 THE LYRIC METRES ii. The Sapphic Stanza, The so-called 'longer Sapphic ' has already been explained i^Odes i. 8 ' Lydia, die, per omnes '; see p. 244). We are now concerned with the shorter and more familiar Sapphic, which Horace uses very frequently and which had been attempted by Catullus. The Aeolian line written by Sappho may be thought of as a line of five trochaic feet, light in its movement, in which the writer exercises once the option of making the lighter part of the foot consist of two syllables instead of one. This, as we have seen, was perhaps a feature of very primitive verse, but in Sappho's verse it is strictly regulated ; it occurs in the third foot, and nowhere else. Its occurrence in the second place gives a 'Phalaecean' line (ptXTaB' ApfioSi', ov TL TTOV Tedi'rjKas. The quadrisyllabic or choriambic scansion adopted by some recent metricians divides a Sapphic line thus : In either case it begins with a trochaic dipody, and here, as in the Alcaic stanza, Horace makes the fourth syllable invariably long, - w . At the end of the line also he obviously has a preference for this effect. Most lines end with a completed — w spondee. An ending like ' diuque ] laetus intersis ' is infre- quent.^ Sappho's verses now and again, but only now and again have a pause -after the fifth syllable : Kal yap at (pevyei \ radices Sido^ei. ^ And '-que' may be a syllaha anccps. Horace has no line like Sappho's rrvKi'a divevvres nrip utt' wpavoj aiOe- pos dia jxioacv, where the last foot can hardly be anything but a trochee. OF HORACE 251 For reasons which can only be conjectured, not certainly ascertained, Horace had a special liking for this type of line. In the third book he has it throughout; in the first occasionally, and in the second twice, he has a pause which answers to the 'trochaic' or 'Homeric' caesura in the hexameter, dividing a dactyl thus, - w | v^ : Mercuri, facunde | nepos Atlantis. In the fourth book and in the Carmen Saec. he admits it more freely. But even there the great majority of lines have a caesura resembling the chief caesura of the Latin hexameter. It may be that the hexameter was in his mind. The result was to make the line a more complex thing, constructed of two parts, just as the 'diaeresis' in the first two lines of the Alcaic stanza gave an effect like that of the elegiac pentameter. But, apart from the analogy of the hexameter, Horace was probably influenced by the metrical theory already mentioned (p. 244). According to this theory of the derivation of metres, a Sapphic line consisted of the beginning of a trochaic line (- v^ - ^ - , two and a half feet) and the beginning of an iambic one (y^- ^ , again two and a half feet ; an anapaest is of course admissible in the first place in a senarius). He may have meant to make this structure obvious. Catullus followed Sappho, and bound himself by no such rule. He has the Horatian caesura quite frequently, but he also has lines without either of Horace's pauses : sen Sacas sagittiferosque Persas.* After Horace, the Horatian caesura becomes normal and invariable. Sapphic lines are regularly written with a caesura after the fifth syllable. This can be seen in Statius {Silvae iv. 7) and in the lyrical parts of Seneca's tragedies (e.g. Medea 579-669). Further, in his freer lyric compositions, 1 Catullus, it will be observed, also admits a trochee in the second place, ' seu Sacas sa-', ' pauca nuntiate ' ; but he does not have it very often. 252 THE LYRIC METRES Seneca uses the two portions of a Horatian Sapphic separately; * for him the two 'commata' are detachable metrical units or , phrases, which can be used like bricks for building a different structure. Sometimes, for example^ he inverts their order : niveique lactis | candidos fontes. {Oed. 495.) Sometimes he repeats the first part : nomen alternis ; stella quae mutat, {Agam. 820.) or the second : yetuitque coUo j pereunte nasci. [Ibid. 836.) He deals with Alcaic commata in the same way, sometimes \ combining Alcaic and Sapphic : procella Fortunae | movet aut iniqui. i^Ibid. 594.) Further, the metrical theory seems to appear in the fact that in place of the two short syllables in a Sapphic he admits w or - : roscidae noctis iussitque Phoebum nullus hunc terror | nee impotentis praebuit saevis tinxitque crudos vicit acceptis | cum fulsit armis ; for an iambic senarius does not necessarily, or even very frequently, begin with an anapaest. When the place of the caesura was once fixed, in a Latin Sapphic, the incidence of accents was to a large extent deter- mined : quale portentum neque militaris Dailnias latis alit aesculetis nee lubae t^llus g^nerat le6num arida nutrix. 'quale portentum', 'Daiinias latis' — the words may vary in length, but the principles of Latin accentuation put the accent on the same syllable. The regularity is interfered with only when a word of one syllable occurs in certain positions, as is seen in the third of these lines, 'nee Iiibae' (in 'p6ne sub OF HORACE 253 ciirru ' there is no such disturbance). The ode to which the lines belong (i. 22 'Integer vitae') has been set to music in modern times on the basis of accent : integer vi'tae sc^lerisque purus, and, further, it has even been advanced as a theory that Horace wrote with this accentual scheme in view. Conscious observa- tion of accent in one of the many elaborate forms of Greek lyric which Horace cultivates is not in itself a very probable thing. Some feeling for Latin accent, it is true, seems to be required to explain the modifications which certain Greek metres underwent on being transferred to Latin. But it is difficult to believe that the attention to accent was deliberate. The line which closes a Sapphic stanza, known as Adoneus, requires little discussion. Its scheme is - ^ v^ - — , dactyl and spondee. It may be noted that its form resembles the end of a Virgilian or Augustan hexameter; it consists usually of words of two, or three, syllables (' terruit urbem', 'augur Apollo'). The exceptions are: ' Fabriciumque', 'Mercuriusque', 'mili- tiaeque', 'est hederae vis', 'Bellerophonten', 'seu Genitalis', and several lines of the type 'se quoque fugit', * te duce, Caesar', 'cum bove pagus'. In regard to synapheia or hypermetron, the facts are very much the same as in Alcaic verse. Horace does not bind himself to continuity of scansion, but has it not unfrequently. It appears that when there is hiatus between one line and the next, the final syllable is never a short vowel, but alwa)'s a long one, or 7(m or eni (' insecutae ' Orphea silvae'; 'leonum | arida nutrix'). Here we see again Horace's preference for - w as compared with -w -v^. Close connexion between the third and fourth line is indicated by effects like 'nigroquje invidet Oreo' (though this occurs also in other places of the stanza) and more clearly by division of a word between them, 'love non probante uxorius amnis' (not found in other places). 254 THE LYRIC METRES iii. Asclepiadean Metres. Glyconic and Pherecratean. The Lesser and Greater Asclepiadean. If we prefix to the 'Aeolic tripody' described above, the usual disyllabic ' basis ', the result is a ' Glyconic ' line : 5^^|-v>"^— W-. This may be interpreted as trochaic, with a rapid dactyl taking the place of one of the trochees (-»^ I —\jKj I -w I -a). On the other or quadrisyllable interpretation it is — — -^ | ^ - w- : w — — ^ w — w — VCCOp 0€ (Pindar, O/. i. i.) (antispast + diiambus). But it is difificult to believe that the two short syllables do not go together, and that they form part of a dactyl is indicated by the occurrence in Greek of shortenings which belong to dactylic verse (/ceiVo/xaf before a vowel). The 'Pherecratean' line is shorter by a syllable;' probably what happens is that a penultimate trochee takes the form of a single protracted syllable : This line was eminently suited to be the last of a group or stanza, and a good example of the arrangement is preserved from Anacreon : 1 The fifth-century comic poet Pherecrates devised a type of anapaestic metre which coincided in syllables with the lyric verse we are considering : dv5p(9, iTp6a(rx.^Te tov vovv e^€vpr)[.iaTi /caivu) avfXTTTVKTOis dvavaiaTois. This may have been ^ | — ' | '^ ^ — I ~ A- How the name came to be given to the lyric verse is unknown. OF HORACE 255 yovuovfjLaL a e\a(pr]^6\€ ^avOr) iroL Al6r]P6\€ \ ^ai^drj). This stanza was used by Catullus in his 'Hymn to Diana', and in his 'Epithalamium' (in the former there are three Glyconic lines in the stanza, in the latter four) : silvarumque virentium saltuumque reconditor- um] amniumque sonantum. The metre was also used by tragic and comic poets (e.g. the passage from Euripides quoted above, p. 248, note). But Horace would have none of it. Was he avoiding the tracks of Catullus ? libera per vacuum posui vestigia princeps, non aliena meo pressi pede. Not only does he avoid this stanza, but he uses the Phere- cratean line not as the last of a group, but as the third of four. With Horace, the Glyconic becomes a closing line ; in the verse of quis multa gracilis te puer in rosa, Pherecratean precedes it, in other odes it follows a group of three Asclepiadeans ; or, again (as in i. 3), it precedes, in a dipody, a line which is not shorter than itself, like the Pherecratean, but the lesser Asclepiadean, which is longer. In all metres of this type Horace made the 'basis ' consist regularly of two long syllables. Catullus did not. Following the greater freedom of the early Greek poets he admits a trochee or an iambus, and he gives rapidity of movement to his 'Epithalamium' by making the initial trochee very frequent. Horace's practice was probably due to a metrical theory which 256 THE LYRIC METRES elicited from a single hexameter both the Glyconic and Pherecratean lines : cui non dictus Hylas puer aut Latonia Delos (when Virgil's hexameter is thus cut into two separate verses, the last syllable of 'puer' can be taken as a syllaba anceps). The ' Asclepiadean ' metres may safely be regarded as extensions or expansions of the Glyconic.^ The x\eolic tripody — KjKj — w — , duke decus meum, may be thought of as postponed by the interposition of an incomplete form of it, thus : O et [dulce decus] dulce decus meum and the incompletion or postponement may occur twice : nullam, Vare, sacra vite prius severis arborem, that is nullam severis arb- severis arb- severis arborem. But this fictitious example introduces what Horace avoids, the severance of a word at the end of the ' choriambus ' or choriambic group of syllables, - ^ o- - . Horace's scheme is illustrated rather by quae nunc oppositis debilitat pumicibus mare converted into quae nunc pumicibus pumicibus pumicibus mare. In Alcaeus and Catullus the severance of a word is not avoided : fxrjSeu dXXo (j)VTev-(rrjs nporepou SevSpLov d/i7reX(o. nee facta impia fal-lacum hominum caelicolis placent. Whatever scheme of scansion Horace had in mind, he must 1 The name may be derived from Asclepiades of Samos, the elegiac poet contemporary with Philetas. But there seems to be no other evidence for his use of the verse. Presumably he, or another poet of the same name, used it extensively, Kara arixov. But extant passages of Alcaeus show that he also had written it Kara arixov. 3 1 V OF HORACE 257 have felt that to carry a word over this point would involve a prolongation of a syllable, a thing which is unnatural except when verse is actually sung. The lyric poet is of course in theory a singer throughout, but except in the Carmen Saeculare^ and perhaps in one or two other pieces, Horace is really writing for readers. So he makes the end of a ' choriambus ' coincide with the end of a word, so that instead of a prolongation there may be a short pause.^ If such a difficulty was felt the inference would seem to be that the choriambus, whatever may have been the description of it in ancient metrical or musical notation, was not simply — v^w — , but rather — v^v^ — or — w v^ — A . Thus we seem to have a definite justification for adopting the ' trochaic ' or modern or rationalizing method of scansion. But it is not safe to say, ' This was the ancient notation'. Probably it was not; quite possibly, however, if we could put this notation before an ancient metrician or musician he would say : ' I see ; that is a good method of notation : it was not ours, but it seems in many ways better, simpler, and more consistent.' Horace observes with considerable strictness the rule which he has adopted for the * choriambus ' in Asclepiadean verse. In a few instances he admits, instead of the end of a word, the end of a prefix or part of a compound word : arcanique fides prodiga per-lucidior vitro.'^ dum flagrantia de-torquet ad oscula. ' It seems fairly clear that it is for the same reason that the first half of an elegiac * pentameter' ends with the end of a word. Elegiac verse was a metre which at an earl}' date came to be merely read and not necessarily sung. 2 Pompeius (157. 5 k) declares that Horace's perlncidior vitro shows the separable compound Wx'Ca.per ' very', e.g. 'per polsaepe peccas' s^Plaut.). It is quite likely that this per was more detached or detachable than other prefixes. But it is hardly credible that Horace intended it here. To compare an adj. with the prefix per is a solecism. It is a virtual superlative ; />^rbeatiss«w«s belongs to vulgar Latin. 1 887 S 2S8 THE LYRIC METRES This resembles his deviations from the rule of the diaeresis in Alcaic lines : utrumque nostrum in|credibili modo {supra, p. 247). In Alcaics he twice goes further than this : mentemque lymphajtam Mareotico. In Asclepiadeans the only similar thing occurs in iv. 8 : non incendia Car|thaginis impiae. The text of this ode has been a subject of much discussion, but there is no good reason for denying this line to Horace. Metrical rules were often stretched or relaxed when a proper name had to be introduced. Further it may be said to be Horace's general practice to break once or twice a rule which he has imposed upon himself, as if he were saying, ' I reserve my freedom ; I write as I do quite deliberately and I know that the verse can be written otherwise '. V. Schemes of Horace's Lyric Metres (in the order in which they have been dealt with above). I. Merpa iiovoeiSfj. Dactylic. (a) i. 7 and 28. «^v^ I ^V_/ I KJ^^ I ^\J ' ^^ WO I -GG -^^ - _ W 1 (b) iv. 7. wo I wo i WW wo I '^^ — 00 OW - /v, ^ The last syllable may be marked '^ from the point of view of prosody. It is often actually a short syllable. Metrically the last foot is . A spondee in the fifth foot is so rare that it has not been indicated in the scheme. It is rare also in the third foot of the second line (' men- sorem cohibent, Archyta,' Odes i. 28. 2). OF HORACE 259 Iambic. (a) Epode xvii. (b) Epodes i-x. (c) ii. 18 (the first line perhaps 'trochaic'). — v^ — \j — \u — A (or ^ - I ^ - I ^ , - v^ 1 - v^ I ^ - a) I 111. 12. v^w W^ ^v_/ W^ ^(>_/ \j y^ I \^ \^ wv^ 2. Merpa kirLavi'd^ra. (' Parian.') [i) Epode xvi. (2) Epodes xiv and xv. v_/v_^ I v_;w I _ WW I _ v_/w I I v=i^^(,^_^_Uy_ (3) Epode xiii. _ WW I _ WW I _ WW • _ WW [ _ I 1 I I i I ' (4) Epode xi. — WW — WW ~A|~~^~i~~ 1 The fifth foot is twice a spondee (1. 17 and 1. 29) in Epode xvi, once a spondee in Epode xiii (1. 9). S 2 : 26o THE LYRIC METRES OF HORACE (5) Odes i. 4. ^ - 1 w - I !^ - I v^ - 1 ^ - I - (or^-|v^-|^|i -v^|-^|i-|-a) 3. Mir pa /ilkto,. i. Tke Alcaic Sta?iza. ^ — w — v^v^ — w — !^ — v_> — \j KJ — ^ — v^w — \J\J — w — — ii. 77^^ Sapphic Stanza. - v^ v^ - ^ (Adoneus) The longer Sapphic, Odes i. 8. — v^ v^ — v^ — \^ I — \^ <^ — — ' note on the lyre) : a term, unknown to the ancient poets, invented in modern times to designate what is very familiar in modern music, a note or syllable which precedes the first actual bar of the measure. Whether such a syllable should be recognized is not an important question for Latin versification ; in Horatian metre the question of anacrusis arises only in the case of the first three lines of an Alcaic stanza. 262 APPENDIX Theoretically, a verse with anacrusis is also catalectic, and the syllable which appears at the beginning is equal to the syllable lost at the end, e. g. a trochaic line should, in theory, have a short syllable as anacrusis: ^\—^—^ — ^ — /\. But there seem to be undoubted instances of verses with anacrusis which are nol catalectic, e. g. Aes. Profn. 135 : (= the fourth line of an Alcaic stanza, with a short syllable pre- fixed). If the third line of an Alcaic stanza has an anacrusis it has it without catalexis. The word, and the idea, seem to be required for certain forms of strictly lyrical Greek verse. It is a mistake to apply the notion to such verse as spoken senarii, and to construe these as trochees with anacrusis. (The notion of anacrusis is not out of harmony with ancient ideas. Trochaic and iambic verse were the same ' in rhythm', the ratio of the parts of the foot was the same, 2: i. What did occur to the ancients was the converse idea that an iambic verse might be ' acephalous ', or lack its first syllable.) Aphaeresis is used for the converse of elision : the following vowel, not the preceding one, is 'taken away' {a(f)nipi'iTni), and has no place in the metrical scheme, as in fxq '70) for fxi) iyu), Latin elided a long vowel freely, and had practically no use for aphaeresis. Arsis and Thesis. The Greeks meant by arsis the lifting of the foot in the march or dance, and by thesis the downward move- ment. In an anapaest the two short syllables accompany the former movement, the long syllable the latter : arsis thesis Hence for the Greeks the long syllable in a dactyl, anapaest, iambus, &c., is the thesis, the rest of the foot the arsis. A different use of the terms grew up in Roman imperial times, and after that yet another usage, which has been the prevalent one with modern metricians— the voice and not the foot came to be thought of, and hence arsis and thesis exchanged meanings. The long syllable in the anapaest came to be called the arsis. ' Rise * and 'fair in English, and ' Hebung ' and 'Senkung' in German, have been adopted as terms answering to this use of arsis and thesis. In view of these equivalents it would be a mistake to revive the Greek use of the words. Asynartete {aawapTrjros, awaprdv, fasten together) : Horace's line: w -w - ^^ reducet in sedem vice. | nunc et Achaemenio is asynartete ; it would have been ' synartete ' if he had made it a rule that the last syllable of the iambic part must always be an APPENDIX 263 actual long syllable, and that it may not be followed by a hiatus. An elegiac pentameter is * synartete *. In Horace's line there is a rupture of continuity ; the two parts or kmXu are not ' fastened together ' or closely attached. The short syllable of ' vic6 ' counts as a long, with a pause after it. Basis: (i) a step in march or dance, the lifting and lowering of the foot, arsis + thesis. The term belongs chiefly to anapaests. (2) Basis or 'basis Aeolica ' is used to designate what is something not unlike an anacrusis, except that it consists of two syllables or a whole foot. In Horace, and in poets after him who use metres to which it belongs (e.g. Statius and Martial in hendecasyliabics), it is always two long syllables or a spondee (one exception, Oi/es i. 15. 36). In the early Greek poets it may be — v^ or w— , and occasionally it is w v-/ — or >^ w (Catullus admits — ^ and ^ — ). Brachycatalectic : it is doubtful whether this term has any utility at all, and at all events it is of no use for Horace. Bucolic : belonging to herdsmen {(BovkoXoi) or to the songs of herdsmen (Pastorals, Bucolica). The hexameters of Theocritus very frequently have a division at the end of the fourth foot, and this was called the bucolic diaeresis : r] Kara Ili]V€i(o KoXa re/xTrea | ^ Kara Ilivdu) nam neque Parnassi vobis iuga, nam neque Pindi (the end of a word is not enough. ' Peliaco quondam prognatae vertice pinus ' is not a bucolic line. To make a line ' bucolic ' in any useful sense of the term there must be a more or less marked rhetorical pause or pause in the sense). Caesura (ro/Ltr}) : a cutting or severance, a break or division of a verse which does not coincide with the end of a metrical foot (some writers use the word to include also a division which does coincide with a metrical ending, but it seems expedient to call that by a different name, diaeresis). A hexameter must have a caesura, it may not fall into two equal parts. A line written by Ennius : spernitur orator bonus, horridu' miles amatur, is defective, and not to be imitated. A senarius also must have a caesura, a pause no^ at the end of the third foot, but in the third or fourth foot. Aeschylus's line : QpaKrjv nepdaavTes j /ioyiy TroXXca novco may have been meant to depict a slow and halting march. Catalexis, catalectic (/cnrnXi^yw, stop, come to an end, break off) : catalexis takes place, and the line is catalectic, when instead of a syllable or syllables uttered by the voice there is a pause of silence answering to them in duration. The last foot is incomplete 264 APPENDIX (in sound, not in time). Thus, when dactylic verse is catalectic, two short syllables (= one long) are dropped (or are unrepresented in sound) : — ww — \^ ^ — a. (See ' Pause.') Colon {kmXov) : a ' phrase ' or coherent group of feet, sometimes forming in itself a ' verse * {o-tIxos, versus), sometimes part of a verse. A hexameter and a pentameter are 'verses' consisting of two co/a. But sometimes the distinction between a ' colon ' and a * verse ' is rather arbitrary or conventional. An ' asynartete ' verse (reducet in sedem vice. | nunc et Achaemenio) might appear as two lines or versus, if the poet or his editor so determined : reducet in sedem vice, nunc et Achaemenio. Comma (Ko/^/xa, indsuvi) : a term used in rhetoric for a group of words shorter than a kmXou, and also used by Caesius Bassus in metre. It is convenient to use it for a group of syllables which have not the length and independence of a kwXov, but yet do form a separable group, e.g. the first five syllables of a Sapphic line, —y , which is a ' comma ' or severed portion of a trochaic line. But the parts into which a hexameter is divided by the 'caesura' should not be called commata. A dactylic hexameter is not one KwXov made up of two commata, but a line, versus or o-ti;^o$-, made up of two KooXa. Crasis (xpao-iy, Kepawvyn) takes place when a vowel at the end of a word is not elided before a following vowel, but fused with it. Thus if /cat could suffer elision, kcu e-yco would be k' iyw («-^ — ; ; but with Kai crasis is the rule, and the result is a spondee, not an iambus (/cdyco, ——). ' Cyclic ' dactyl : a term which rests on a very doubtful passage in Dionysius, where there is mention of a special kind of anapaest ' which they call kvkXios ' (if kvkXios is right, an anapaest belonging to the dithyramb and its kvkXios x"pos would be meant : ' cyclic ' answers to kvkXikos, and would connect the foot with epic poetry). It was used by J. H. H. Schmidt and other metricians of last century to denote a dactyl shorter and more rapid than a normal one, answering in time to a trochee, and capable of being combined with trochees. The symbol -v/ w was used for it, and the time was supposed to be J ^ J. The theory was that a normal long syllable could be replaced by slightly shorter syllable with a very short one attached to it, — w for — . (When a dactyl is divided by caesura, as in the Horatian Sapphic, J J^ seems more probable.) Such a dactyl could be described as a daKrvXos Tplxp'>vos or Tpiatjfxos. APPENDIX 265 Diaeresis: a break, word-ending, or division whicli coincides with the end of a foot or a metrical group. It is convenient to have a term other than caesura for such a division as that in the middle of an elegiac pentameter or a trochaic tetrameter : Horace's break in the first two lines of an Alcaic stanza is also to be described as a diaeresis. The syllables i=^ — w , with which the line begins^ were perhaps thought of as part of an iambic senarius ; but the break would be a ' caesura ' only if what followed were the rest of an iambic verse — whereas a quite different effect follows, beginning with a dactyl. The scansion * odi [ profan|um vulg|us et arcjeo ' (iambic, of a kind, throughout) is not to be thought of. Diiambus and Ditrochaeus (dU) are terms sometimes used for an iambic or trochaic dipody (^ — v^ — and — w— — ). (Cic. and Quint, used * dichoreus ' for — <^ — — , and meant by trochaeus what we call a ' tribrach'.) Dimeter: a term used for a group oi four feet (tetrapody) in certain kinds of verse : i. e. the unit of measurement is not one foot, but a pair or dipody. Iambic and trochaic verse were so divided and described by the Greeks (see p. 232). A line of four anapaests also was called a dimeter ; but a line of four dactyls was called a tetrameter. Dipody, tripody, tetrapody, pentapody, he\apody arc words used for a phrase or verse measured by the number of single feet that it contains : dipody a group of two TrdSf?, tripody one of three. Hexapody, if in use, would answer to senarius. Elision {eltdere, force out, thrust out) : a vowel ' elided ' before a following vowel was ' thrust out' of the metre or metrical scheme. For metre 'elision' is a term that expresses the fact better and more clearly than the Greek word awnXoKprj. The vowel did not count metrically, though it is not to be supposed that it was alto- gether inaudible or entirely omitted by the voice.^ In Greek it was chiefly the short vowels n, f, o that \yere treated in this way. Latin went much further, and regularly elided vowels long or short, and also final syllables in m. Thus in re^^io oberat the long has no place in the metre, it is the following that counts (— >^ w ^). See Hiatus. Epitritiis (ttous- emrpLTos, a foot divided in the ratio 4 : 3) : a name for the group of syllables — ^ (w v^ v^ | w w w v^), a trochaic dipody or ' ditrochaeus ', ending with the heavier syllable which was allowed in that place. Whether the time was really 3 + 4 and the 'foot' enTaarjixos is a very disputable matter. 1 Forms like animadve/io, niagnopcre, dotnitio ( = dontiini itio, home- coming seem to point to great attenuation of the vowel or oi' tint. 266 APPENDIX The term is chiefly used for an element in those odes of Pindar that are sometimes called * dactylo-epitrite ' (a form of metre used also by the tragedians), where it is combined with what on one theory is a dactylic KwXnv (usually — w w — w w ), and on another a choriambic and Ionic group (— w w — ] w v^ ). Hiatus : a verse /liaff there is a break or ' gap ' in it when a vowel is not elided before a following vowel : ' Glauco | et Panopeae 6t Inoo Melicertae.' ' Glauc' et, Panope' et' is the normal, continuous scansion in Latin, but here the vowels stand side by side ; between them, instead of a consonant, there is nothing, a ' gap ' or ' rift '. Hiatus is of two kinds, both exemplified in this line of Virgil. (1) The preceding vowel retains its length, (2) it is shortened. (2) is a regular practice in Greek dactylic verse, but it seems to have also had some root in Latin. ' Ita mS | dl ament, leplde accfplmur (Plautus), an qui amant ' (Virg. Eclogues\ probably answering to actual facts of pronunciation. Latin verse admitted both kinds very sparingly. Virgil's ' insulae lonio in magno ' is explained by the fact that a word of Cretic form (— w — , Insulae) was not allowed to suffer elision in heroic verse. In Greek epic verse hiatus is more frequent, and when short syllables follow a long it is admitted very freely (rir^Xj^iaSea) 'AxiX^oyj in //. i. l). Hypercatalectic. See ' Brachycatalectic ' and ' Anacrusis '. ' Acephalous ' and ' Hypercatalexis ' were ancient terms used to account for things which the modern notion of 'anacrusis' often explains in a simpler way. Hypermetron or synapheia : metrical continuity between one line and the next {avvd(p€ia, contact, absence of any metrical severance or interruption). This is found chiefly in anapaests and in some Glyconic stanzas (see p. 248). The last syllable is not a syllaba anceps, and elision takes place if the next line begins with a vowel. The end of a line may fall within a word. In Horace's lyrics, as in most metres, this effect is permissive, not imperative. In some forms of verse it is very rare ; Sophocles now and again has elision at the end of an iambic line, a-vv oh r I ov xpnv ofxtXcbv, and Laevius in iambic dimeters writes : saurae, inlices bicodulae, hin- nientium dulcedines. In an epigram of Simonides it is used to bring an intractable name into an elegiac couplet : ^ fjLfy ^AdrjvoLoicn (jydos y€ve0\ fjviK* ^Apiarro- ycLTCov "imrapxov Kveipe kgI 'Ap/LxoSio?. Horace has it occasionally in this, the most unmistakable form of it (' neque purpura velnale neque auro'). APPENDIX 267 It might be convenient to use ' hypermetron ' for metrical con- tinuity between one line and the next, and ' synapheia ' for con- tinuity between two kojXu which make up a line. Ictus, 'beat'. Horace uses the word when he imagines himself to be training a chorus : Lesbium servate pedem meique pollicis ictum. In this sense it is a movement of the hand or foot, accompanying the metre, but outside of it. In ' pavidumque leporem ' (p. 234) the ictus of a resolved iambus is on the first syllable of * leporem ' (w ow taking the place ofw— ). On that syllable falls also the ordinary accent of the Latin word ' leporem '. The ictus marks the * arsis' of a foot. But very often it falls on a syllable which is not accented. It is a matter of dispute what an ictus was. Was it made audible ? Was it of the nature of a slight stress accent ? In ' arma virumque cano ' the first syllable of ' ciino ' is the syllable on which the ordinary accent falls ; the ictus falls on the second. How then was ' cano ' pronounced here ? How was ' Italian! ' pronounced in the next line ? That is a question to which no very definite or certain answer can be given. To read Latin verse well is a difficult thing, at all events for us. Perhaps the following remarks may safely be made about it : [a) the metrical structure must not be obscured, it must be felt, and made sensible to the hearer, even if this sometimes involves dropping or attenuating the ordinary accent ; {b) when the ictus falls on a long syllable a reader or reciter whose native language is English, with its tendency to make a strong stress accent the chief form of emphasis, should endeavour to make the syllable really long (pro- longed, not made loud), without thinking about an * ictus ' apart from that ; [c] when the ictus falls on a short syllable he will find that, in Latin, it usually coincides with an ordinary accent (as in 'pavidumque leporem'), so that he need not think about it in this case either. Lyric. The word 'lyric' is used in two senses, a wider and a narrower one. In its wider sense (as in the title ' Bergk's Poetae Ly?-ici Graeci) a lyric poem is any poem which does not narrate, as an epic does, or put characters and action on the stage, as a drama does, but gives expression to the feelings or thoughts of the poet (or of the poet and his friends, or the poet and his countrymen, as in a lament or a song of victory). In this wide sense it includes elegiac and iambic (abusive or personal) poems. In the narrower sense, which is the ancient one, it means a poem to be sung to music, a song — for which ' melic ' poem is some- times used. Elegiac and iambic poems had at first a musical accompaniment, but at a quite early date they came to be merely spoken or recited. So they fall clearly within the sphere of ' metre ' or Mctrik ; no 268 APPENDIX syllable has more than the time of an ordinary long. It is otherwise with * melic ' poems. Some Greek lyrics were meant only to be sung, not to be read at all. In dealing with these Metrik is really encroaching upon, or borrowing ideas from, the province of rhythm or music. Metre, Metrik^ M^TpiKt] : fxtrpiKt] is a branch of pvBiiiKx], metre is rhythm manifested in a particular medium or vKr], namely, articulate syllables or speech. We may think of pvBpiKi) as a highly abstract, central inquiry, which provides empty forms or time-spaces, of a certain length and subdivided in one way or another, to be rendered visible or audible by some kind of movement or sound, e.g. opxn^T'TiKTj deals with rhythm as revealed by the movements of a dancer (axrjpnra). Metrik deals with rhythm as it is worked out in detail in spoken syllables. So Ouintilian remarks that dactyl and anapaest are the same in rhythm : ' for rhythm it does not matter whether the two short syllables precede or follow the long '. Metre has sometimes meant a rather mechanical way of describing certain groups of syllables, regardless of their rhythm. Thus a dochmius {y w — ) can be described as consisting €^ avn- (T7rd(TTov Ka\ avWa^rji, v_/ ^y \ — , but this is a purely metrical description. For rhythm or music it had to be regarded as a novs oKTacrrjpos {fiaivovrai §e ol pvdpoi, ra de perpa diaipelrai, ovxi (Baii'fTai). In modern times writers on classical verse have usually not been content with a merely descriptive or mechanical system. Pause (Aflju/io, K€v6s xpovos) : the space or interval which is not occupied by sound at the end of a catalectic verse. The symbol in common use for a pause is the initial letter of Aei/xwa. Without any addition this denotes a pause equal to a short syllable {Kevos ^piixvs). Longer pauses are denoted by adding a symbol of length : thus A is a pause equal to w w or — {kcvos dixpovo^, or, in the language of music, dicrrjpos), a a pause equal to w w w or — w (Kevos Tplxpovos). Only A and a are required for Horatian metres (a in the case of ' arboribusque comae '). It is difficult to avoid using the word ' pause ' for another thing, a slighter and less measurable pause, e. g. the ' bucolic ' pause in 'omnia vel medium fiant mare, vivite, silvae'. This is a rhetorical pause or pause required by the sense, and as such outside metre ; but it determines the metrical type of line — the line is one with a marked diaeresis after the fourth foot. Prosody is a word which in its origin is not well suited to express what is commonly meant by it ; the Greek word irpocrcodia meant accent, and the Greek accent was a musical or pitch accent [Trpos-adeiv). ' Prosody ' is now used in two ways : (i) Sometimes it is used very widely and loosely, to include much of what belongs to metre, Afeirik, or perpiKi]. (2) In relation to Latin and Greek it has a narrower meaning, and a quite useful one. Metrik tells us how short and long APPENDIX 269 syllables are arranged in the various kinds of verse ; prosody tells us what actual syllables are short and long, or competent to occupy these places. Thus the distinction between Mevis ' and Mevis' belongs to prosody. So, again, it is a matter of prosody that the final syllable of ' mihi, tibi, sibi' is either short or long. But if we proceed to ask, ' When is it long ? ' the answer is no longer strictly within the bounds of ' prosody '. Metre is involved, for the fact is that the syllable was originally long, and the old quantity as it were comes to life again when the metrical ictKS falls on the syllable. Senarii {sc. versus) are (iambic) lines consisting each of six feet, or which have se7ii pedes. A Greek trimeter does in fact contain six feet, but it is convenient to reserve the name seiiarius for the Roman type of verse, which does not, like the Greek, show dipodic structure by excluding spondees from the second and fourth places. o-iTov8eid^cov (sc. uTixf^s:) '. a hexameter which ends in two spondees : 'me quoque deAiexi rapidus comes Ononis' (Hor. Odes \. 28.21), * sententia Phocaeorum' {Epode xvi. 17). Homer has such lines, and the atticization of the Homeric text perhaps increased their number, Ar/jtiSao becoming 'Arpei'Sao, and r]6a dlnv, rjco d'lav. In Roman poetry this form of line was admitted by Catullus and his contemporaries: ' moenia vexarentur', 'et decus innuptarum '. Cicero rejects it, and thought it an affectation of the vtcjTfpm {Att. vii. 2. i). In Augustan poets it is found chiefly when the dispondeus is a Greek proper name [supra, p. 23). o-rrovSctos {novs), spondee : a foot of two long syllables or a bar of two long notes ( J J ), so called because it accompanied a libation (o-novdrj). The notes were those of a flute, and no doubt they were often slow or protracted ( li i_j or J J ) : Victorinus, 44. 22 K ' dictus- a tractu cantus eius qui per longas tibias in templis supplicantibus editur'. Syllaba anceps. A syllable which is in the peculiar position of standing at the end of a line is regarded as outside ordinary principles of quantity, it is dubious (anceps), and if short may count metrically as long (e. g. ' submovere litora ; reducet in sedem vicg ', at the end of a KaXov), or if long as short. Cic. Orator 217: ' postrema syllaba brevis an longa sit, ne in versu quidem refert ' (*ne in versu quidem ', i.e. still less in a closing cadence of prose). A short counting as a long is very common, and Terentianus describes it clearly : omnibus in metris hoc iam retinere memento, in fine non obesse pro longa brevem. Reckoning a long as a short is a more difficult and obscure matter (see p. 246 ;/.). 270 APPENDIX Synapheia or Synaphea (awdcf^nn) : see Hypermetron. Syncope : what happens when a foot is ' cut down ' to one syllable, or compressed into one syllable. Gr. tovt] : ' protraction ' of the syllable, is another name for the same thing. The foot as a rhythmical unit is not really ' cut down ' ; the syllable becomes longer than an ordinary long, equal to a trochee in trochaic verse (— «^ = I—), or to a dactyl in dactylic verse (— o w = i_i ). The effect belongs chiefly to Greek lyric verse. Roman poets did not use metres in which it occurs in any obvious or certain way. It is doubtful whether Horace anywhere intended it to be felt. Synizesis (crvinOja-is, a ' sinking ' or ' settling ' together of two vowels) is most conveniently used for what happens when two vowels within a word, which usually are separate syllables, are combined into one long, e.g. when 6e6s, Seal, is scanned as one syllable (vimv fieV Bfnl doieu, II. i. i8). So too in compounds: a word like Theodosius (v^ w w w ^ inadmissible in dactylic verse) becomes Theudosius. When the vowels are in separate words the term crasis is used. Thesis : see Arsis. (B) CHRONOLOGIA METRICA Pre-Homeric lays (Aeolian ?) Epic — Homeric hexameter : Ionian : a verse intended for recitation. Elegy— Callinus : Tyrtaeus (? date of Tyrtaeus uncertain). (650) Archilochns of Faros, elegiac verse and iambi. Dactylic and iambic cola combined. Music and poetry at Sparta. Terpander. Alcman. Lyric— Stesichorus, Dorian lyric in Sicily. ' Epici carminis onera lyra sustinens ' (Quint, x. c. i). Metre — many long dactylic cola. 600 Alcaeus, (TTaaiojTiKa and avyLiroTiKo. Sappho, kpojTurd Lesbos. Aeolian yric. APPENDIX . 271 Mimnermus, erotic elegy. Solon, political verse ^^elegiac : iambic and trochaic, Archilochian). Hipponax : scazon or choliambus. Thespis at Athens : beginning of tragedy : trochaic tetrameter (Ar. Poet. c. iv). Theognis of Megara : elegy, political and social. Anacreon comes to Athens : Aeschylus gets from him 'Iwvikoi dvaic\d>fxevoi. 500 Pratinas protests that the words are more important than the music : rav dotSav KariaTaai Iliepls ^acriKaav. « Pindar : elaborate and varied lyric, measures both Dorian and Aeolian. Bacch3'lides, a professional verse-maker : simpler and more mechanical stj'le. Simonides : epigram, 9p7Jvoi, &c. Tragedy : in metre, dochmii peculiar to it, for a tragic crisis, a moment of tension and suspense. Comedy : the iafx^iKT] Idea, abusive, personal and political f Ar. Poef. c. iv : Crates opened a new^ vein). Cratinus, Eupolis, Aristophanes : iambic verse much freer than that of Archilochus Gorgias and Thrasymachus begin to study rhythm in oratory. Euripides : new developments in music (parodied in Frogs) : music more complex and less subordinate to words. Tragic trimeter freer and more conversational (resolution more frequent). 400 Timotheos : author of a vuj^os now known, JJepaai. Philoxenus. dithyrambs. Pluius of Aristophanes : choric parts slight and simple. Rhetoric : Rhythm and prose elaborated by Isocrates and others. Tragedy becomes rhetorical. Rhetoric invades history (Theopompus and Ephorus). Decline of the chorus : choric part very slight in comedy. (fx^oXifJia in tragedy (Ar. Poef. c. xviii dp^avros 'AydOojvos'. Isyllus of Epidaurus (hexr., troch. tetr. and Ionics in a paean to Apollo). Menander :' New' Comedy (^tjOiktj Idea) mainly in iambic trimeters (with occasional passages in trochaic tetrameter). Philetas of Cos, early Alexandrian elegy. 272 APPENDIX GREECE 300 Asclepiades /• lyric metres writ- Ai J • ten Kard ariyov Alexandria J ■^ hexameter preva- I lent Callimachus — scazon for fables elegiac verse in AjTm (narrative^ Delphic hymn, cretics Philosophy again expressed in verse Pure iambi (Stoics — Clean- thes) Scazons (Phoenix — Cyni- cism and popular moraliz- ing) Herondas ] Mixed, prose with M.ifiiajxl3oi verse intermixed in Scazons ' (MenippusandBion) Aratus Lycophron and in Nicander plainer style harsh and obscure Rhinthon, iXaporpaycvdia iKapcvSiai and fxaycoblai (Grenfell's Fragment) 200 ROME Epitaph of Scipio (L. Corn. Scipio Barbatus — Gnaivod patre prognatus — consul in 298) Epitaph of L. Scipio, consul in 259 Odtsia of Livius Andronicus 240 plays produced by Livius 239 birth of Ennius 235 Comedy of Naevius Plautus writing pallmiae — can- tica from IXapq^blai? 204 Ennius comes to Rome Naevius's ' Bellum Punicum ' in his old age — Saturnians APPENDIX 273 GREECE [This section (Greece, 200-Too) was left by Prof. Hardie un- written.] 100 ROME Ennius flourished Annales (hexameter) Hedyphagetica (hexam. looser, like Lucilius) Saturae (elegiac and other metres, written with con- siderable finish and pre- cision) Tragedies Terence — less canticum than in Plautus (nearer Menander^ Togata - same metres as pal- liata Lucilius — iambi, troch., then hexam. in Saturae Volcatius Sedigitus Porcius Licinus — literary mat- ters discussed in verse Accius — tragedies also Sotadeans Saturnians ?) i- 274 APPENDIX ROME Hexameter Drama 63 Cicero consul Cicero, Aratea Lucretius (rather more archaic and more akin to Enn. than his contemporaries) Atellana popular Tragedy written as a literary composition, not for the stage (Q. Cicero wrote four tragedies in sixteen days) 48 Pharsalus 42 Philippi 31 Actium Augustus Mimus popular Laberius Syrus ' Cantores Euphorionis' ; hexameter smoother and more finished Ciris? I Culex ? ' 42-38 Ecloguesof Virgil , Varius, Th^^estes, 29 b.c. Varro Atacinus | Virgil (70-19 Horace : satires and epistles) ! by rhetoric"! 1 ' Trabeata ' written b\' Melissus - A.D. 14 d. of Augustus Manilhis, Astronomica Tiberius, 14-37 Caligula, 37-41 Claudius, 41-54 Nero, 54-68 Calpurnius, Pastorals Lucan Seneca, Tragedies (iambic verse, strict : 5th foot always a spondee Lyrics — Horatian verse,usually Vespasian Titus Valerius Flaccus Kara arixov) Octavia, a praetexta domitian Nerva, 96-98 Trajan, 98-117 Statius Silius Hadrian • APPENDIX ROME 75 Other forms of Poetry Catulus, erotic epigrams Laevius, Erotopaegnia : metrical experiments Saturae of Varro: many metres, written witli considerable exactness Elegy Catullus Catullus : Hendecas. (Phalaecei Kara arixov) : Scazon (5th foot always an iambus, : experiments in Sap- phics, Glyconics, Asclep., Galliambics Callus Tibullus Propertius Ovid Catalepton ? (or some of the pieces in it ?) Horace : metres of Archilochus Epodes), Alcaeus, and Sappho (Odes) Phaedrus, Fabulae i^in the senarius of comedy) Caesius Bassus, lyric and iambic poet (and writer on metre ? Persius, satirist (Stoic) Petronius, Satyricon (prose, with occasional verse in various metres) Seneca, Ludus de Morte Claudii (a Varronian or Menippean satire) Juvenal : satire in regular and sonorous hexameter (rhetoric now invades even satire) Hadrian : iambic and trochaic dimeters. Epitaph on his horse in Anacreontics Florus PRINTED IN ENGLAND AT THE OXFORD UNIVERSITY PRESS iMiiir"''"'^^ RETURN TO DESK FROM WHICH BORROWED CJ8CUIATION DEPA«TMINT ^ 5 .■v'll This book is due on the last date stamped below, or on the date to which renewed. Renewed books are subject to immediate recall. ^^^^ 01Q74 CIRC ms ^qis, APR Muu U;, i!EC,MOFF!TTJUL09'9l) ^cge ,^00 . ^EPT JUN12 74 5^ LD 21-32m-3,'74 (R7057sl0)476 — A-32 General Library University of California Berkeley Hi /^f; "■ffa*^ U.C.BERKELEY LIBRARIES CDM7D135b2