- ^ 
 
PA , 
 
 (Snmll IRniwmtg ^ilrmg 
 
 THE GIFT OF 
 
 .^s^...u^^to:^. 
 
 A /3^^/6: ^^ 
 
 'Oo 
 
uornen universiiy Liorary 
 
 PA 2333.B47 
 
 What was ictus in Latin prosody? 
 
 3 1924 021 619 311 
 
 What Was Ictus in Latin Prosody ? 
 
 BY 
 
 CHARLES E. BENNETT, 
 
 Professor of I<atin in Corheli. University 
 
 Reprinted from the American Journai of Philoiogy, Vol. XIX, No. 76. 
 
 BALTIMORE 
 
 1899 
 
The original of this book is in 
 the Cornell University Library. 
 
 There are no known copyright restrictions in 
 the United States on the use of the text. 
 
 http://www.archive.org/details/cu31924021619311 
 
[Reprinted from the American Journal of Philology, Vol. XIX, No, 4.] 
 
 I.— WHAT WAS ICTUS IN LATIN PROSODY? 
 
 English poetry, as a rule, is based on stress, i. e. on a regular 
 succession of stressed and unstressed syllables grouped by twos 
 or threes. The versification of 
 
 This is the forest primeval, the murmuring pines and the hemlocks, 
 
 depends entirely upon this harmonious alternation of stressed 
 and unstressed syllables, and the same is true of all ordinary 
 English verse.^ This basis of English poetry, moreover, inheres 
 in the very nature of the English language. Like all languages 
 of the Germanic group, our English speech is strongly stressed ; 
 we pronounce our words with an energy typical of the race. 
 
 Latin verse, on the other hand, like Greek, was based on 
 quantity. Recent discussion, it is true, has tended to show that 
 the native Latin verse, as exemplified by the Saturnian measure, 
 was governed by stress ; but however that may have been, it is 
 certain that, from the time Greek metres began to be introduced 
 at Rome, — from the time of Ennius, — Latin verse was quantita- 
 tive like Greek ; a line of Latin poetry consisted of an orderly 
 and harmonious arrangement of long and short syllables, i. e. of 
 syllables which it took a long or short time to pronounce. This 
 basis of Latin poetry again, as in the case of English poetry, is 
 strictly in conformity with the character of the spoken language. 
 For Latin apparently, in the classical period, was not a strongly 
 
 ' Such exceptions to this principle as occur (see, e. g., Goodell, Transactions 
 Amer. Phil. Assoc. XVI, p. 78 fF.) hardly concern the purposes of the present 
 discussion. 
 
362 AMERICAN JOURNAL OF PHILOLOGY. 
 
 Stressed language.^ Had it been, it is quite inconceivable that ■ 
 the long environing vowels should not have been shortened in 
 such words as eviidbdtur and scores of others like it in which the 
 Latin language abounds. Cf., e. g., a Latin inevltabile with 
 English inevitable. Strong stress necessarily reduces every 
 long pre-tonic and post-tonic syllable to a short one. In other 
 words, strong stress is absolutely inconsistent with the quanti- 
 tative phenomena of the Latin language. 
 
 This view of the Latin accent is further confirmed by the status 
 of the Romance languages. So far as my information and obser- 
 vation go, no one of these languages is strongly stressed. In the 
 utterance of French and Italians, I often find it difficult to deter- 
 mine on what syllable the stress rests, — ^so relatively slight is it. 
 In fact the quantitatively monotonous character of Italian makes 
 a strong stress accent impossible. This consensus in accentual 
 character on the part of so many daughters of the Latin, while it 
 cannot be held to amount to a proof that Latin was but slightly 
 stressed, nevertheless seems to me to point distinctly in that 
 direction. In fact I feel at times inclined to go much further and 
 to maintain that the Latin language of the classical period was 
 absolutely unstressed. As such a thesis is easily defensible, and 
 as its consideration may throw light on the subject under discus- 
 sion, I shall venture to advocate this probably startling view. 
 
 The thesis is that the Latin language of the classical period was 
 unstressed. I do not say unaccented, for that would be to contra- 
 dict the express testimony of the whole apostohc succession of 
 grammarians from Varro to Priscian; but I do say unstressed. 
 The Roman grammarians tell us that a long penult was accented ; 
 they tell us, further, that the antepenult was accented when the • 
 penult was short. Now, what was this accent, this accentus of 
 the Roman grammarians? Or first, what is accent in general? 
 Accent is an elastic term. To a person familiar with English 
 only, accent seems a very clear and simple thing; naturally so 
 because such a person imagines that all languages are like his 
 own; hence he will answer that accent is exemplified by the 
 second syllable of potdto, or the final word of the phrase let her 
 gdl, i. e. by accent he understands stress. But the study of 
 comparative phonetics teaches us that there are languages in 
 which the stressing plays no such r61e as in English and in the 
 Germanic languages generally, and that in some of these other 
 
 ^ Cf. Sievers, Grundztige der Phonetik", §35, i, Anm. i. 
 
WBAT WAS ICTUS IN LATIN PROSODY? 363 
 
 languages the chief feature of the spoken idiom is the rising and 
 falling of the voice. Certain syllables are thus uttered at a higher 
 pitch than others, and it is this shifting of pitch that gives the 
 character to the speech. According to the commonly accepted 
 view, Greek was characterized by this kind of utterance, and the 
 syllables whose pitch varied from the ordinary were designated 
 in accordance with principles to which we give the name of 
 accent. Sanskrit also at one stage, it is thought, must have been 
 so pronounced. In these two languages, therefore, what we call 
 accent was primarily variation in pitch. 
 
 We have already recognized two senses in which the word 
 accent is used. Both, while radically different, have at least this 
 much in common: The 'accented' syllable is the one made 
 prominent in some way in oral utterance.^ When the word is 
 spoken, that syllable stands out conspicuous, either by virtue of 
 its stress, i. e. a definite expulsory effort of the lungs, or by virtue 
 of its pitch. To these two varieties of an accented syllable, i. e. 
 a syllable standing out prominently in an uttered word, — to these 
 two conceptions, I ask, may we not add a third? May not a 
 syllable be primarily prominent by virtue of its quantity ? That 
 is, in a word like amdvit, for example, may not the rule of the 
 grammarians, that such a word was accented on the penult, 
 simply mean that they felt the quantity of the long penult as 
 making that syllable prominent, without any stress on the one 
 hand or any elevation of pitch on the other ? And in words like 
 Idtuii, hdmines, etc., may not the rule that these words were 
 accented on the antepenult simply mean that, in consequence of 
 the short penult, that syllable did not possess any prominence, 
 and hence, after the establishment in Latin of the three-syllable 
 law, the syllable next preceding became the conspicuous one? 
 
 There are, of course, objections that at once suggest themselves 
 against so radical a view as the foregoing. I do not pretend to 
 have established it, but simply to have suggested a theory of 
 Latin accent which all must admit is possible and which to me 
 seems even^' probable. At all events, it is certainly of the first 
 importance in approaching so delicate a problem as the pronun- 
 ciation of ajlanguage whose data we can no longer fully control, 
 first to rid] ourselves as completely as possible of all preconceived 
 notions derived^from our own language which might mislead us, 
 and to take into__account the great divergence of human speech 
 
 'Cf. Seelmann, Die Aussprache des Latein, p. i6, 2, Accentarten, 
 
364 AMERICAN JOURNAL OF PHILOLOGY. 
 
 along with the often radically different character of spokert 
 languages. 
 
 The thesis, then, that classical Latin was absolutely unstressed 
 may not be proven; but that, if stressed, classical Latin was only 
 slightly so, was, I think, made fairly certain by the argument 
 advanced at the outset of this paper ^; and it is precisely this 
 slightly stressed (possibly absolutely unstressed) character of the 
 Latin language that explains the character of Latin poetry. 
 Stress was either absent or at most quite subordinate; hence 
 syllabic quantity came to be the natural basis of verse. Theo- 
 
 ' This seems to me all but universally admitted among those who hold to 
 the stress view of Latin accent. Cf. Ritschl, Prolegomena ad Trinummum', 
 p. 207 ; Madvig, Latin Grammar, §498, note ; W. Meyer, Ueber die Beachtung 
 des Wortaccentes in der altlateinischen Poesie, p. 5 f.; Lucian Miiller, De re 
 metrica poetarum Latinorum', p. 233. This conclusion rests not only upon 
 the internal evidence above adduced, but also upon the statements of Roman 
 writers of the best period. Cicero and Quintilian, in discussing the question 
 of metrical sequences in prose, particularly at the close of a sentence, nowhere 
 pay the least attention to word-accent, but do lay special stress upon quanti- 
 tative distinctions. Cf., e. g., Cicero, Orator, 55, 56 (§§183-90), .64 (§§215- 
 18) ; Quint., Inst. Or. IX 4, 47 ; 61 ; et passim. From the definitions of 
 accent given by the ancient grammarians, it is impossible to gather any 
 consistent conception of the phenomenon. See the testimonia gathered by 
 Scholl in his De accentu linguae Latinae, p. 73 ff. Of the formal definitions, 
 none antedates the fourth century, and many are much later. Most of these, 
 moreover, are extremely vague. Dositheus, for example (Keil, Gram. Lat. 
 VII, p. 377, 6), defines accent as "unius cuiusque syllabae proprius sonus." 
 Maximus Victorinus (Keil, VI 188, 15) and Audax (VII 322, 12) as "unius 
 cuiusque syllabae in sono pronuntiandi qualitas." This same conception of 
 the qualitas syllabarum appears also in Sergius, Explan. (Keil, IV 528, 28). 
 Ps.-Priscian (III 519, 25) is more precise. According to him, "accentus nam.* 
 que est certa lex et regula ad elevandam et deprimendam syllabam unius 
 cuiusque particulae orationis." The Codex Bernensis 16 (K. Sup. XLV) gives 
 " accentus est vox syllabae, quae in sermone plus sonat de ( than ') ceteris 
 syllabis." Similarly Servius (K. IV 426, 16) and Pompeius (K. V 126, 10). 
 Three grammarians define accent as anima vocis, viz. Diomedes (K. I 430, 30), 
 Pompeius (K. V 1 26, 27), and the Cod. Bern. 16 (K. Sup. XLV). Lastly Diomedes 
 (K. I 430, 29) says: "accentus est acutus vel gravis vel infiexa elatio orationis 
 vocisve intentio vel inclinatio acuto aut inflexo sono regens verba." From 
 this chaos one can hardly have the conscience to seek support .for a theory. 
 The testimony of Diomedes might be cited in support of a stress accent, just 
 as that of Ps.-Priscian might be cited in support of a musical accent. But 
 what with the lateness of these writers, their mutual contradictions, and the 
 suspicion that some of them at least are but echoing statements made by 
 Greek grammarians concerning the Greek language, it is safest to base no 
 positive view of the nature of accent upon their utterances. 
 
WHAT WAS ICTUS IN LATIN PROSODY? 36$ 
 
 retically, now, this quantitative Latin poetry may seem suflfi- 
 ciently simple, were it not for the so-called ictus, a feature to 
 which our traditional prosody uniformly gives a prominent place. 
 What was this ictus? It is usually defined as stress accent. 
 With a single exception to be noted below, it is invariably thus 
 defined, so far as I am aware. Yet I question whether there is a 
 particle of legitimate evidence, internal or external, in support of 
 this view. The conception of ictus as stress accent seems to me 
 to have its foundation solely in the practical assumption that 
 Latin poetry was, like English and German poetry, really accen- 
 tual.^ I say 'practical assumption.' It would, of course, be 
 absurd to maintain for a moment that theoretically the quanti- 
 tative character of Latin verse has ever been denied. Yet so 
 long as Latin is pronounced with absolute disregard of vowel 
 quantity, as it necessarily is by the so-called English method of 
 pronunciation, and as it habitually is in Germany to my certain 
 knowledge, or with disregard of syllabic quantity, as it usually is 
 even where the Roman pronunciation is nominally followed, so 
 long is it inevitable that any theoretical recognition of the truly 
 quantitative character of Latin verse should be totally clouded by 
 the impulse toward securing a rhythmical effect. By a pronunci- 
 ation which yields gero, tiro, ingenium and thousands more of 
 the same sort, on the one hand, and fllius, his, vis, etc., on the 
 other, a quantitative verse is as impossible as would be an accen- 
 tual verse in English, were we to misplace the regular word- 
 accent. It is no exaggeration to say that were we to accent 
 Longfellow's line as follows : 
 
 This is the forest primeval, the murmuring pines, 
 
 the result would be no whit worse than is inevitably necessary by 
 any system of Latin pronunciation which fails scrupulously to 
 observe the quantity of every vowel and of every syllable. A 
 neglectof quantity was inevitable under the English pronunciation 
 of Latin ; it is inevitable under the pronunciation of Latin current 
 in Germany. Neglect of quantity leaves nothing except accent 
 as a basis for a metrical effect, and naturally leads to an accentual 
 reading of Latin verse, which brings with it the conception of 
 ictus as a stressed syllable. Yet this conception seems to me 
 demonstrably false, for the following reasons : 
 
 'Cf., for example, Christ, Metrik der Griechen und Romer', p. 3, where the 
 strongly stressed character of German utterance is appealed to in support of 
 the stress theory of ictus for the classical languages. 
 
366 AMERICAN JOURNAL OF PHILOLOGY. 
 
 1. So far as we know, no language is ever forced to an artificial 
 pronunciation when adapted to the service of poetry. It is irra- 
 tional to conceive any such adaptation.' The poet simply takes 
 the choicer words of familiar speech and employs them in their 
 ordinary equivalence with their regular pronunciation. He must 
 do so. For his appeal is to the many, not to a select handful 
 who may have been initiated into the secret trick of his versifi- 
 cation ; hence he must use words in the pronunciation familiar to 
 his auditors or readers. Otherwise he can make no appeal. His 
 art consists, on the mechanical side at least, in arranging words 
 in such a way that the poetic form is obvious to the meanest 
 observer who knows the words by ear or eye. Can any poetry 
 be cited in any language of which this is not true ? Is it then not 
 absurd to assume that in Latin poetic form consisted in employing 
 words with gratuitous stress accents unknown in the ordinary 
 speech? Can we conceive of an atavis, a regibils, a Troiai, a 
 cand, or a thousand other equally grotesque hermaphrodites that 
 we are compelled to father by this theory ? And is it credible 
 that poetry so inconceivably artificial should have been tolerated, 
 not to say admired, by such sober-minded persons as the Romans ? 
 
 2. The view that ictus was stress is to be rejected because it 
 involves the assumption of a second basis for Latin verse. We 
 have already noted that Latin verse is quantitative, i. e. a dactyl 
 is a long time followed by two short times. But if ictus is stress, 
 then a dactyl is a stressed syllable followed by two unstressed 
 syllables. We should thus get two principles as the basis ol 
 Latin verse, quantity and accent (i. e. stress), and it seems to me 
 impossible that there should uniformly and regularly have been 
 two principles at the basis of Latin verse or any other. » 
 
 3. It is nowhere hinted or imphed in the ancient writers that 
 ictus was stress. To judge from the prominence assigned to ictus 
 
 ^Cf. Schell, De accentu linguae Latinae, p. 25, note: nihil poeta infert in 
 linguam, sed quae praebet elementa disciplina et humanitate excolit, perpolit, 
 adauget. This principle would not merely appear to be a rational necessity, 
 but is supported by distinct testimony from the best period. Cf. Cic. de Orat. 
 Ill 45, 177, where it is clearly asserted that the elements of prose and verse 
 are identical ; also Orator, 11, 36, where Ennius is praised for adhering to the 
 common speech in his poetry (communis mos verborum). Yet Corssen Aus- 
 sprache, Vok. u. Betonung, IP, p. 975, maintains that just such an artificial 
 adaptation in pronunciation did occur in Latin verse. Similarly Lucian 
 MuUer in both editions of his De re metrica (2d ed., p. 234). The same view 
 also is implied wherever ictus is defined as stress. 
 
WHAT WAS ICTUS IN LATIN PROSODY? 367 
 
 in our grammars and other works on prosody, one might expect 
 to find that the word was widely current as a technical term 
 among the ancients. Such, however, is not the case. Among all 
 the systematic discussions of prosody found in the Latin gram- 
 marians I have been able to discover no definition of the term, — 
 in fact no mention of it as a technical term of prosody.^ The 
 word does occur a few times in the classical period, but so rarely 
 and in such context that there is no justification for regarding it 
 as a terminus technicus. Thus we find it in the familiar passage 
 of Horace, ad Pisones, 253 : 
 
 unde etiam trimetris accrescere iussit 
 Nomen iambeis, cum senos redderet ictus 
 Primus ad extremum similis sibi. 
 
 More frequently we find ictus in this signification combined 
 with digitus, pollex, or pes. Thus Horace, Carm. IV 6, 36 
 poUicis ictum; Quint., Inst. Or. IX 4, 51 pedum et digitorum 
 ictu intervalla signant; Pliny, N. H. II 95, 96, 209 ad ictum 
 modulantium pedum. From these and similar instances (the 
 total number, however, is very small), the natural inference is 
 that ictus as a metrical term primarily designated taps of the feet 
 or fingers, and was then transferred to denote the rhythmical 
 beats of verse. Certainly there is no evidence either from the 
 etymology of the word or from its use in any citable case to 
 indicate that it designated vocal stress. 
 
 Scarcely more support of the stress theory can be derived 
 from the use of the words arsis and thesis as employed in the 
 systematic treatises on Latin prosody prepared by the ancient 
 grammarians. It is commonly held that the grammarians and 
 metricians used these designations with an inversion of the appli- 
 cation which obtained in Greek. But as Westphal has pointed 
 out (Griechische Rhythmik, p. 106), such is by no means the 
 case ; the Roman writers exhibit no little confusion in this matter, 
 it is true, but they rarely use arsis in the sense of Greek 6i<Tis. 
 Probably Marius Victorinus is the only metrician^ who does this,' 
 
 1 Professor Hale (Proceedings Amer. Phil. Assoc. 1895, vol. XXVI, p. xxx) 
 implies that the word ictus is employed as a technical metrical term by 
 Charisius. His view and the argument based upon it will be discussed later 
 in this paper. 
 
 2 On Ps.-Priscian, De accentibus (Keil, III 521, 24), see below. 
 
 *Cf. Westphal, 1. 1., p. 107 f.; Weil et Benloew, Theorie generale de I'accen- 
 tuation latine, p. 98 f. 
 
368 AMERICAN JOURNAL OF PHILOLOGY. 
 
 and he does it but once (Keil, VI 40, 16 f.). The actual usage of 
 the Romans in employing the terms arsis and thesis can be seen 
 only by an examination of the decisive passages in which these 
 words occur. The commonest definition explains arsis as elevatio 
 (jublatid), thesis as positio. This bare statement, without the 
 addition of any explanatory remarks, is found in Sergius, Expla- 
 nat. in Don. I (Keil, IV 523, 2); Cledonius (Keil, V 30, 10); 
 Atilius Fortunatianus, Ars (Keil, VI 281, 4). Similar is Teren- 
 tianus Maurus's explanation, de Metris, 1345 (K. VI 366) : 
 
 parte nam attollit (sc. pes) sonorem, parte reliqua deprimit. 
 
 Other writers add vocis as explanatory of elevatio and positio. 
 Thus, Martianus Capella, IX (365, 17 Eyssenhardt) : <arsis est 
 elevatio>, thesis depositio vocis ac remissio; Isidore, Orig. I 16, 
 21 arsis et thesis, id est vocis elevatio et positio; Commentum 
 Einsidlense in Don. Artem Mai. (Keil, Suppl. 228, 23) : arsis 
 elevatio sc. vocis, eo quod ibi vox elevetur. Thesis humiliatio 
 vel demissio quia ibi vox deponatur. 
 
 A new conception appears in the three following writers : 
 
 Juliani Excerpta (Keil, V 321, 12): quid est arsis? Elevatio, 
 id est inchoatio partis, quid est thesis ? positio, id est finis partis 
 ... In trisyllabis, si in prima habuerit accentum, ut puta domi- 
 nus, duas syllabas vindicat arsis et unam thesis, si paenultimo 
 loco habuerit accentum, ut puta beatus, arsis vindicat unam 
 syllabam et thesis duas. 
 
 Servius in Donatum (Keil, IV 425, 7) : Arsis dicitur elevatio, 
 thesis positio. quotienscumque contingit ut tres sint syllabae in 
 pede ... si in prima syllaba fuerit accentus, arsis duas syllabas 
 possidebit; si autem in media syllaba, thesi duas syllabas damug. 
 
 Pompeius Comm. in Donati artem (Keil, V 120, 29): arsis et 
 thesis dicitur elevatio et positio . . . Romulus quando dicimus, 
 prima syllaba habet accentum : dicimus duo in arsi, unum in thesi 
 ... si media syllaba accentum habuerit, ultimae syllabae iungis 
 plura tempora, ut arsis habeat unum, thesis duo. 
 
 Marius Victorinus, Art. Gram. I (Keil, VI 40, 14), evidently 
 impelled by the spirit of Goethe's maxim: "Besonders lass 
 genug geschehen," gives us a unique ragout: 
 
 Arsis igitur ac thesis quas Graeci dicunt, id est, sublatio et 
 positio, significant motum pedis, est enim arsis sublatio pedis 
 sine sono [sc. pedis],' thesis positio pedis cum sono (the Greek 
 
 ' Westphal, Gr. Rhythmik, p. 105. 
 
WHAT WAS ICTUS IN LATIN PROSODY? 369 
 
 conception) : item arsis elatio temporis, soni, vocis : thesis depo- 
 sitio et quaedam contractio syllabarum ... in dactylo vero toUitur 
 una longa, ponuntur duae breves (the reverse of the Greek 
 conception)'; while elsewhere (Keil, VI 45, 2) the same writer 
 evidently regards the arsis as ihe first syllable of the foot without 
 regard to quantity. The passage reads : horum [the trochee and 
 iambus] arsis et thesis alterna mutatione variatur, si quidem in 
 iambo arsis primam brevem, in trochaeo autem longam habeat 
 incipieniem, thesis vero contraria superioribus sumat. This con- 
 ception has already appeared in the passage of Julianus above 
 cited (Keil, V 321), and appears again in Sergius (Keil, IV 480, 
 13): scire etiam debemus quod unicuique pedi accidit, arsis et 
 thesis, hoc est elevatio et positio ; sed arsis in prima parte, thesis 
 in secunda ponenda est ; Diomedes (Keil, I 480, 10) : iambi enim 
 arsis unum tempus tarn in se habet et eius thesis duo quam 
 trochaei versa vice arsis duo habet et thesis unum ; Terentianus 
 Maurus, 1388 (Keil, VI 367) : 
 
 dpcTjf unum possidebit, quando iambum partior ; 
 fiat alternum necesse est, cum trochaeum divides. 
 
 It remains only to cite the testimony of Ps.-Priscian (Keil, III 
 521, 24) : nam in unaquaque parte orationis arsis et thesis sunt, 
 non in ordine syllabarum sed in prpnuntiatione : velut in hac 
 parte ; natura. quando dico natu, elevatur vox, et est arsis intus. 
 quando vero sequitur ra, vox deponitur, et est thesis deforis. 
 quantum autem suspenditur vox per arsin, tantum deprimitur per 
 thesin. sed ipsa vox, quae per dictiones formatur, donee accentus 
 perficiatur, in arsin deputatur; quae autem post accentum 
 sequitur, in thesin. 
 
 The foregoing are the essential specific expressions of opinion 
 on this subject which I have discovered among the ancient 
 Roman writers on metric. It has seemed worth while to cite 
 their statements in full in order that we may see exactly what 
 support they furnish for the traditional theory of ictus and arsis. 
 In my judgment they afford no confirmation whatever of the 
 view that ictus in the classical period was stress. In the multi- 
 tude of testimonies which I have cited it is impossible to find any 
 definite, coherent common doctrine. The witnesses not merely 
 contradict one another; many of them, as we have seen, contra- 
 dict themselves, till one involuntarily exclaims with Weil and 
 
 Westphal, Gr. Rhythmik, p. 105. 
 
370 AMERICAN JOURNAL OF PHILOLOGY. 
 
 Benloew: "Rien n'est plus difficile ^ expliquer qu'un auteur qui 
 ne sait pas lui mSme ce qu'il veut dire." ' Of the writers above 
 cited it seems most natural to believe that those who define arsis 
 and thesis as sublatio and positio are merely translating the terms 
 of the Greek writers without any serious attempt to understand 
 their actual application. Those who add to this definition the 
 statement that the arsis was the initial syllable of the foot, the 
 thesis the last part, are apparently guilty of attempting to com- 
 bine two irreconcilable conceptions. Both of these, strange to 
 say, are found among Greek writers. The late Greek metricians 
 applied the term apais indiscriminately to the initial syllable of 
 the foot, quite irrespective of its quantity.'' 
 
 As regards those writers who define arsis as elevatio vocis, the 
 earliest of these is Martianus Capella, who does not antedate 400. 
 Even conceding that elevatio vocis could by any possibility have 
 been deliberately intended to mean 'stress of voice,' we have to 
 bear in mind that a century and a half before the time of Marti- 
 anus Capella quantitative Latin poetry had begun to be sup- 
 planted by accentual poetry. Commodianus is usually cited as 
 the first versifier who exemplifies the transition.' He wrote about 
 250. It is an interesting fact also that Martianus Capella himself, 
 in those passages where he essays poetic form, repeatedly yields 
 to the spirit of the age and employs an accented short syllable 
 where the metre demands a long one.* If, therefore, Martianus 
 Capella, Isidore, and the author of the Commentum Einsidlense 
 really meant stress by elevatio vocis, the presumption is strong 
 that their testimony holds only for the accentual poetry of their 
 own day, not for the quantitative verse of the classical period. 
 
 Another difficulty confronts us. How are we to reconcile the 
 statements of Julianus (Keil, V321, 12), Servius (Keil, IV 425, 7), 
 and Pompeius (Keil, V 120, 29) with the theory of an accented 
 arsis ? These writers tell us that in words of the type of Romulus, 
 dominus, the arsis consists of two syllables. Certainly a stress 
 accent cannot stand simultaneously upon two successive syllables. 
 Probably any attempt to reconcile these last three statements 
 with those previously cited would prove futile. Julianus, Pom- 
 peius, and Servius are apparently concerned with the phenomena 
 of individual words rather than with metrical feet. With them 
 
 1 Theorie generale, p. 100. ^ Westphal, Gr. Rhythmik, p. 106 f. 
 
 'See Teuffel-Schwabe, Gesch. d. ram. Lit., §384. 
 * Teuffel-Schwabe, ibid., §452, 5. 
 
WHAT WAS ICTUS IN LATIN PROSODY? 37 1 
 
 arsis manifestly has some connection with the accented syllable, 
 though it is clearly more than that. A similar attitude appears 
 in Ps.-Priscian, De accentibus (Keil, III 521, 24 f.), where all of a 
 word preceding the accented syllable (and including that) is 
 reckoned as belonging to the arsis, the remainder to the thesis.^ 
 
 I have been thus minute in considering in detail the testimonies 
 of the Roman metricians, because it seemed to me that we could 
 in no way see so well how completely they fail to afford the 
 slightest support to the stress theory of Latin ictus or arsis.^ 
 
 I have already given three reasons why it seems to me erro- 
 neous to regard ictus as stress: i. Because it involves the impor- 
 tation of a stupendous artificiality into the reading of verse. 
 2. Because it involves a dual basis for versification, — stress as 
 well as quantity. 3. Because the view finds no support in any 
 ancient testimony. To these three reasons I wish to add as 
 4. There are excellent grounds for believing that ictus was some- 
 thing else than stress. If Latin poetry was quantitative, as its 
 internal structure and all external evidence seem to show, then a 
 /dactyl was a long time followed by two short times, and a trochee 
 a long time followed by one short time, absolutely without any 
 other parasitic accretion. When, now, we come to use dactyls 
 by the line, one part of every foot will inevitably be felt as 
 prominent, viz. the long syllable. The relative amount of time 
 given the long syllable of every dactyl naturally brings that long 
 syllable into consciousness, and especially must it have done so 
 to the minds of the Romans, whose nice quantitative sense is 
 proved by the very fact that they made quantity the basis of 
 their versification. Yet the long of the dactyl has no stress.' It 
 is natural for us to stress it, us whose only conception of verse is 
 accented verse. But in so doing I believe we are simply trans- 
 ferring to Latin verse our own inherited verse-sense. I define 
 ictus, therefore, not as stress, nor as accent, but simply as the 
 
 'Julianus (Keil, V 321) reckoned only the accented syllable of a trisyllable 
 as belonging to the arsis. The preceding and following syllables he reckoned 
 with the thesis. 
 
 'Sergius (Keil, IV 483, 14), cited by Christ (Metrik^ p. 59) as supporting the 
 view that arsis was stress, ought not to be quoted in defence of this view. 
 Sergius is not here speaking of metrical feet, but merely of individual words 
 that accidentally form feet. The context makes this perfectly clear. 
 
 'Of course it may incidentally have stress, if the word-accent fall on this 
 syllable. But this stress was at most relatively slight, as already explained. 
 On the r51e played by word-accent in reading verse, see belovir. 
 
372 AMERICAN JOURNAL OF PHILOLOGY. 
 
 quantitative prominence inherent in a long syllable. This defini- 
 tion applies primarily only to the four fundamental feet — the 
 dactyl, the anapaest, the trochee, and the iambus. It does not 
 apply to the spondee, for example, when substituted for the 
 dactyl in dactylic verse. In such cases the first long of the 
 spondee is felt as the quantitatively prominent thing in the foot, 
 even though the second syllable of the spondee is also long. In 
 dactylic verse, the dactylic character and feeling so dominate 
 the line that any spondee naturally takes on a dactylic character 
 and is felt to be quantitatively prominent in its first syllable, just 
 as in the case of the dactyl itself. So in iambic measures, where 
 the tribrach or dactyl is substituted for the iambus, the quantita- 
 tive prominence inherent in the long syllable of the iambus is felt 
 as transferred to the two final shorts of the tribrach or the 
 dactyl.^ I 
 
 This conception of thesis or ictus receives no little support from 
 the positive testimonies of the Roman grammarians. These 
 writers in their definitions of arsis and thesis repeatedly call 
 attention in unambiguous phrase to the essentially quantitative 
 character of these concepts. In this, their agreement is con- 
 spicuous. Thus : 
 
 Diomedes (Keil, I 474, 30) : pes est sublatio ac positio duarunt 
 aut trium ampliusve syllabarum spatio comprehensa. pes est 
 poeticae dictionis duarum ampliusve syllabarum cum certa tem- 
 porum observatione -modus recipiens arsin et thesin ; 
 
 Marius Victorinus (Keil, VI 41, 25) : nam rhythmus est pedum 
 temporumque iunctura velox divisa in arsin et thesin vel tempus 
 quo syllabus metimur; id. VI 43, 26: signa quaedam accentuum 
 . . . syllabis ad declaranda temporum spatia superponuntur . . . 
 sed et hoc non praetermiserim, eosdem [Graecos] figuras pedum 
 secundum spatia temporum per litteras ita designasse, ut brevis 
 syllabae loco, quae sit unius temporis, ponatur B [Ppaxv], longae 
 autem loco, quae sit temporum duum, M [iJ-axpSv'] : hoc ideo, ut 
 per litteras regula pedum facile intellegatur. 
 
 Atilius Fortunatianus (Keil, VI 281, 4) speaks of feet as things 
 " qui gresstbus alternatis quasi incedunt per versus et moventur"; 
 
 Commentum Einsidlense (Keil, Suppl. 228, 9) : his [sc. pedi- 
 bus] ... ad peragendos versus tempora syllabasque metimur . . . 
 nulla enim alia res dinumerat tempora et syllabas velut pedes . . . 
 
 ^ On the baselessness of the traditional pointing of dactyls and tribrachs in 
 iambic measures (~^ w, ^ 6 <j), see the latter part of this paper. 
 
WHAT WAS ICTUS IN LATIN PROSODY? 373 
 
 pes est certa dinumeratio syllabarum vel certa dinumeratio tetn- 
 porum. 
 
 All of these definitions and observations exhibit a striking 
 unanimity in emphasizing the purely quantitative character of 
 ancient poetry.' Even when signs were employed by the Greeks 
 to mark the syllables, these signs were abbreviations of words for 
 Mong' and 'shqrt,' not for 'stressed' and 'unstressed." 
 
 To those who may cherish a scepticism as to the tangible 
 reality of 'quantitative prominence,' I would only say that that 
 phrase need appear shadowy to no one who will actually read 
 one thousand lines of Latin aloud with absolute fidelity to vocalic 
 and syllabic quantity. My own revolt against the traditional 
 view of ictus has been purely and solely empirical. It was 
 simply because by faithful practice in accurate reading my ear 
 quickly grew sensitive to quantitative differences, that I was 
 forced to believe that, as quantity was the basis of Latin verse, 
 so ictus was only quantitative prominence. This conclusion, I 
 say, was first forced upon me empirically, and the theoretical 
 formulation was entirely subsequent to, and solely the result of, 
 my actual oral experience in reading Latin. No one, in my 
 judgment, can approach this subject in a candid spirit who has 
 not first taken the pains to acquire the habit of exact pronunci- 
 ation of Latin vowels and syllables. Even in this country, where 
 we have nominally adopted the quantitative pronunciation of 
 Latin, we have still much to learn in this matter. Our short- 
 comings are so pronounced, and bear so directly upon the theo- 
 retical aspect of the question at issue, that I shall here venture to 
 recapitulate some of them. 
 
 First, we habitually neglect vowel quantity. One cause of this 
 is the vehement stress which (in accordance with our English- 
 speaking instinct) we regularly put upon the accented syllable. 
 The word evltdbdtur, for example, contains four successive long 
 vowels. Yet in ninety-nine cases out of a hundred, the penulti- 
 mate syllable is so strongly stressed that the first three vowels are 
 pronounced short. In Latin poetry the result of such pronun- 
 
 'Cf. Aristoxenus's designation of apai^ and S^ffif as XP^""^, XP^i""' l>i>6/wcol, 
 Xpdvoi iroSiml, Westphal, Gr. Rhythmik, p. 103 ; also as & ava xp^'>">i, l> k&tu 
 Xp6vo(, ibid., p. 104. 
 
 ' The modern practice of using the acute accent to designate the thesis — a 
 practice which is in itself an assumption of the stress view of ictus and has 
 done so much to propagate it — goes back only to Bentley. 
 
374 AMERICAN JOURNAL OF PHILOLOGY. 
 
 ciation is to wreck the quantitative character of the verse as effec- 
 tively as if in English we were to misplace the accents on succes- 
 sive syllables. How much poetic form would appear in Milton's 
 opening line of Paradise Lost, were we to pronounce ' Of man's 
 first dis6bedience,' for instance ? Besides destroying vowel quan- 
 tity as a result of over-stressing the accented syllable, we habitu- 
 ally neglect it in hundreds of other instances where there is no 
 such disturbing factor. By some strange fatality the -is of the 
 genitive singular is commonly pronounced -is, while the -is of the 
 ablative plural as regularly is heard as -Is ; while the number of 
 such pronunciations as pater, ager, nisi, quod, quibus, ingenium. 
 is simply legion. No one who pronounces Latin in that way can 
 expect to feel the quantitative character of a Latin verse, and is 
 in no proper frame of mind to give the quantitative theory dispas- 
 sionate consideration ; for one or two false quantities destroy as 
 completely the quantitative character of a verse of Latin poetry 
 as would one or two misplaced accents any English verse. 
 
 Even more serious than our neglect of vowel quantity is our 
 neglect of syllabic quantity. The shipwreck resulting from 
 neglect of vowel quantity occurs chiefly in open syllables, i. e. in 
 syllables whose vowel is followed by a single consonant, which- 
 always belongs to the following vowel, thus leaving the preceding 
 syllable open. In such syllables the quantity of the vowel is 
 always identical with the quantity of the syllable ; so that a false 
 vowel quantity involves the quantity of the syllable as well. In 
 closed syllables, on the other hand (i. e. syllables ending in a 
 consonant), an error in vowel quantity does not affect the quantity 
 of the syllable. I may pronounce vendo or vendo. In either 
 case the syllable will be long.^ Hence in closed syllables an 
 
 ' All closed syllables are phonetically long. This is a principle universally 
 accepted by the phoneticians. Yet Professor Hale in Harvard Studies, VII,. 
 p. 267, n., contests it. I can only refer to such standard works as Sievers,. 
 Grundzilge der Phonetik^ §35, i : "In Wirklichkeit kennen nur solche Silben 
 fiir kurz gelten, welche auf einen kurzen Sonanten (= vovrel) ausgehen, also- 
 solche ■wi&ra,la,pra,fm,&tc. Alle geschhssenen Silben aber sind lang." In 
 the following note Sievers adds : " Die ubliche Definition der positionslangen 
 Silben, spricht allerdings von mehr als einem Consonanten hinter dem 
 Sonanten (= vowel); in Wirklichkeit aber geniigt der Ausgang der Silbe auf 
 einen Consonanten um sie lang zu machen." Professor Hale offers no argu- 
 ment against this current doctrine of the phoneticians. The chief objection 
 he adduces is that the initial syllable is short in English many and battle^ 
 But in each case the syllable is open (md-ny, bd-tl). Cf. also Havet in Memoires 
 de la Society de linguistique, IV 22 f., who points out that it is not the two 
 
WHAT WAS ICTUS IN LATIN PROSODY? 375 
 
 error in vowel quantity does not destroy the quantity of the 
 syllable, and so does not interrupt the quantitative character of a 
 Latin verse. But the syllable must be actually closed in pronun- 
 ciation ; else where the vowel is short, the syllable will be left open, 
 and will be metrically short, destroying the verse. It is precisely 
 here that we err so frequently and so fatally in our reading of 
 Latin verse. We do not close the syllables that ought to be 
 closed and were closed by the Romans. The commonest class 
 of words where we commit this error are those containing a 
 geminated consonant — words of the type of ges-serunt, ac-cipio, 
 at-tigerat, ter-rarum, ap-parabat, an-norum, ad-diderat, flam- 
 marum, excel-lentia, ag-gerimus. These words we habitually 
 pronounce in prose and verse alike, as ge-seruni, d-cipio, d-tige- 
 rat, te-rarum, d-parabat, d-norum, d-didit, fld-meus, exce-lentia, 
 d-gerimus. Words of this type are extremely frequent in Latin. 
 I have counted forty-five in the first hundred lines of Virgil's 
 Aeneid, i. e. the pronunciation described destroys the quantitative 
 character of the Latin verse at forty-five distinct points, often 
 twice in the same verse. 
 
 Nor is this all. In other combinations in the interior of words 
 we are often guilty of quite as serious errors. In English, besides 
 
 consonants which lengthen the syllable, but that the syllable is lengthened 
 because one consonant is joined to the previous vowel, while the other 
 consonant is joined to the following vowel. Professor Hale (1. c.) advances 
 the view that in iste, for example, the first syllable is long because the time of 
 the consonant s adds to the vowel (= one mora) an equal amount of time. 
 But I feel confident that Havet is correct when he says (1. t., foot-note, p. 24): 
 " It is not the s which takes time. It is the silence between the s and the t. 
 The duration of the consonants themselves is, if not nil, at least a negligible 
 quantity." This may be clearly seen at the beginning of words. For 
 example, the first syllable of scribas, despite its three consonants, takes appre- 
 ciably no longer time than the first syllable of vivas. What really makes the 
 long quantity is the closed syllable with its accompanying pause, not the 
 'obstructed consonant,' as Professor Hale would hold. A good English 
 example of how the closed syllable may be long even when no consonant 
 follows, may be seen in the phrase at all. Ordinarily we pronounce this as 
 one word, d-idl{l) ; but occasionally it is pronounced as two words, at al(J), the 
 first of which is closed and is phonetically long. Our English a tall (for at all) 
 explains too the short quantity of final syllables ending in a consonant before 
 a word beginning with a vowel. Such syllables are not actually closed, but 
 the final consonant combines by a natural liaison with the following vowel, 
 e. g. ama ialias {= amat alias). See Sievers, Grundzilge der Phonetik*, §658, 
 who cites rdv avr6v, "gesprochen to-nau-ton." At the end of a verse, however, 
 such syllables are actually long. 
 
376 AMERICAN JOURNAL OF PHILOLOGY. 
 
 muta cum liquida, there are many other consonant combinations 
 with which in stressed syllables we show a regular tendency to 
 begin the syllable. This is especially true of the combinations 
 sp, sc{K), si, squ; also scl, scr, sir. This tendency of our vernac- 
 ular speech naturally affects our pronunciation of Latin words in 
 which these combinations occur. The j of such combinations 
 properly belongs with the preceding vowel, in order that the 
 preceding syllable may be closed and so made phonetically long ; 
 yet we frequently (almost invariably, according to my observation) 
 join the i with the consonants of the tonic syllable. I refer to 
 such pronunciations as a-spirir^ima, l-stius^ tempe-stdtibus, corU- 
 scdbat, mi-scilerai, magi- sir drum, d-sclipias, d-scripsit, qulsqui- 
 liae.^ My own students often exhibit a tendency to combine ci, 
 pt, ps with a following accented vowel, and produce short syllables 
 in such words as volu-ptdtef d-spe-ctdrum, l-psius. Where the 
 accent rests on the vowel immediately preceding these combi- 
 nations, the liability to error is very slight. 
 
 There are yet other cases in which error is frequent, if not 
 habitual. Unstressed syllables whose vowel is followed by r-t- 
 any consonant are particularly liable to be made phonetically 
 short in those portions of the country where the r is neglected. 
 This is especially true in the eastern part of the United States, 
 where pd-(r)i&rum, ie-{r)minorum, etc., represent the prevailing 
 utterance.* The combination oi m or n also with a following 
 explosive in unstressed syllables frequently is so treated as to 
 shorten syllables phonetically long. The process by which this 
 is accomplished is not yet clear to me. Observation, however, 
 has taught me that in such words as imperaior, intendo the first 
 
 ^ In Early Latin this division was probably common. I should so explain 
 the metrical use of the word by Plautus and Terence. Cf. Humphreys, 
 Proceedings of the American Phil. Assoc. l8g5, vol. XXVI, p. xxxi. 
 
 ' Lest our traditional rules for syllabication be cited in support of the 
 division here criticized, I would say that the traditional rules, though laid 
 down by the ancient grammarians, can hardly have been more than practical 
 working directions for copyists and stone-cutters. It is impossible that they 
 indicate the actual phonetic division of the syllables. See Appendix to my 
 Latin Grammar, p. 31 f. Since the publication of the Appendix, Professor 
 Hale, in Harvard Studies, VII, p. 249 f., while expressing dissent from certain 
 slight details of my arguments as stated in the Appendix, has endorsed the 
 main proposition there laid down and has fortified it by additional data. 
 
 'So probably in Early Latin. Cf. note I, above. 
 
 *In stressed syllables, where the r is neglected, the vowel is regularly 
 lengthened, e. g. po-ta. 
 
WHAT WAS ICTUS IN LATIN PROSODY f 377 
 
 syllable is frequently made short; whether by omission of the 
 nasal, by pronouncing a short nasalized vowel, or by a short 
 nasalis sonans («), I do not undertake to say. The fact, I 
 believe, is beyond question. 
 
 There is only one other class of cases to which I shall call 
 attention, viz, the unconscious liaison of final 5 after a short vowel 
 with the initial consonant of the following word. Where the 
 following word begins with s, p, c, i, v, m, n, f, etc., and where 
 the connection of sense is close, this liaison is in my experience 
 frequent. It is not surprising that it should be, for we habitually 
 join a final j of an unstressed syllable' in our own speech with a 
 following s, c, t. Examples in Latin are : urbl sporta, capl sca- 
 nem, urbl svici? A case that puzzled me for a time was Juv. Ill 
 53 carus erit Verri, as read by a student. The fourth syllable 
 sounded short to my ear, and it was only after repeated readings 
 that I discovered that the reader was really dividing : carus erl- 
 tVerri^ I do not say that this liaison is invariable. It is cer- 
 tainly frequent, and, where it occurs, must vitiate the quantitative 
 effect of the verse.' 
 
 These common errors in reading Latin must be clearly under- 
 stood, if they are to be remedied. It is by no means an extremely 
 difficult matter to acquire an exact quantitative pronunciation. It 
 takes time and pains and considerable oral practice. I do not 
 believe that it requires a particularly sensitive ear. By practice 
 in rigidly exact reading, the quantitative sense is not slow in 
 coming ; but without that exactness it cannot come and cannot be 
 expected to come. He who has once developed the quantitative 
 sense will, I am confident, feel no need of any artificial stress. 
 
 The foregoing views as to the nature of ictus had long been 
 matured and had been presented to class after class of college 
 students when I stumbled on the following neglected remark of 
 Madvig (Latin Grammar, §498, N.): "We should also guard 
 against the opinion which is generally current; viz., that the 
 ancients accentuated the long syllable (in the arsis) and distin- 
 guished in this way the movement of the verse (by a so-called 
 
 ' In Latin the final syllable, of course, Is regularly unstressed. 
 
 ' Sv, i. e. sw, is a sufficiently common initial combination in English ; the 
 same is true of tv, i. e. iw. 
 
 ' It may be a question whether the so-called weak pronunciation of final s 
 in Early Latin is not, after all, merely a phenomenon of liaison, the s going 
 with the following consonant, e. g. in Ennius's plinii sfidii and misernmH 
 snunriu smprtis. 
 
378 AMERICAN JOURNAL OF PHILOLOGY. 
 
 verse-accent, ictus mefricus), and consequently often accentuated 
 the words in verse quite otherwise than in prose (e. g. Arma 
 virumque cano Troja6 qui primus ab oris ; Itaham fato profuglis 
 Lavinaque venit), which is impossible ; for the verse depends on 
 a certain prescribed order and form of movement being distin- 
 guishable, when the words are correctly^ pronounced. In our 
 own verses we do not accentuate the syllables for the sake of the 
 verse, but the syllables which are perceptibly distinguished by 
 the accentuation in prose form verse by being arranged to 
 succeed each other in this way. In Latin and Greek (where 
 even in prose pronunciation the accent was quite subordinate,, 
 and is never named in speaking of rhetorical euphony, while on 
 the other hand the difference of quantity was distinctly and 
 strongly marked), the verse was audibly distinguished by this 
 very alternation of the long and short syllables." So far my 
 assent with Madvig is complete. He goes on: " But as it is not 
 possible for us either in prose or in verse, to pronounce the words 
 according to the quantity in such a way as the ancients did, we 
 cannot recite their poetry correctly, but are forced in the delivery 
 to give a certain stress of voice to the arsis, and thus make their 
 verses somewhat resemble ours. It should, however, be under- 
 stood, that it was different with the ancients themselves (until the 
 last century of their history, when the pronunciation itself under- 
 went modifications)." These words of Madvig were written in 
 1847 — over half a century ago. At that time it is not strange 
 that he should have denied the possibility of our reading Latin 
 verse quantitatively with substantial accuracy. Even before the 
 end of his life, it is likely that Madvig relinquished this part of 
 his earlier opinion. * 
 
 As regards word-accent in the reading of Latin verse, I believe 
 that it retained its full value ; for as I have maintained that in 
 poetry words are used with their ordinary prose values, and are 
 pronounced without addition of foreign elements, so I believe 
 that they were pronounced without subtraction of any of their 
 elements.^ Herein I agree entirely with Professor Hale (Pro- 
 ceedings Am. Phil. Assoc, vol. XXVI, p. xxvii).^ But we have 
 
 ' The italics are Madvig's. 
 
 ''In support of this we have also the clear testimony of the ancients. See 
 the abundant references in Christ, Metrik^ p. 59. 
 
 ' I regret, however, that this scholar is not as consistent in refusing to admit 
 into verse what was not in prose (artificial stress), as he is in refusing to 
 relinquish what was in prose (word-accent). 
 
WHAT WAS ICTUS IN LATIN PROSODY? 279 
 
 already seen that the Latin accent was slight. It was precisely 
 that fact which led the Romans of the classical period to make 
 quantity the basis of their verse. Assuming, now, that the word- 
 accent was very slight, and possibly was even merely quantity or 
 absence of quantity in penults, what wonder that, with quantity 
 predominant in the verse and i7t the Roman consciousness, such 
 slight word-accent as existed was felt as no intrusion ? An anal- 
 ogous situation reveals itself in our Enghsh verse. Our verse is 
 primarily accentual, and yet each syllable has its quantity, and 
 shorts and longs mingle harmlessly with accented and unaccented 
 syllables. Why should not the reverse have occurred in Latin 
 just as simply and just as naturally ? 
 
 To sum up, then: Latin poetry is to be read exactly like 
 Latin prose.^ Latin was primarily a quantitative language in the 
 classical period and is to be read quantitatively. The Latin 
 word-accent was relatively slight as compared with that of our 
 strongly stressed English speech, and is therefore to be carefully 
 subordinated to quantity both in prose and poetry. Ictus was 
 not a metrical term current among the Romans, nor was there 
 anything corresponding to it in the quantitative poetry of the 
 Greeks. The term is purely modern. We first imported the 
 conception of stress from our modern speech into the quantitative 
 poetry of the Greeks and Romans, and then imported the term 
 ictus to cover it. But just as the conception of artificial stress in 
 Latin poetry is false, so the term ictus is superfluous.^ Qkais was 
 employed by the ancient Greek writers on metric to designate 
 the prominent part of every fundamental foot, and is still entirely 
 adequate to cover that conception. 
 
 It remains only, in conclusion, to meet certain criticisms which 
 have been made upon my conception of ictus (thesis). In the 
 Proceedings of the American Philological Assoc. XXVI, p. xxx, 
 Professor Hale has characterized my view as immature and has 
 advanced certain objections against it. These objections are four 
 in number : — 
 
 ^I forbear to enter into any discussion of the difficult matter of elision of 
 final syllables ending in vowels or in -m. I hesitate to believe that poetry 
 involved an artificial deviation from prose utterance ; yet, on the other hand, 
 I cannot regard the evidence sometimes cited in favor of ' slurring ' in prose 
 as in the least decisive. 
 
 'In my Latin Grammar I nevertheless retained the term (defining it as 
 quantitative prominence), but purely on practical grounds. 
 
380 AMERICAN JOURNAL OF PHILOLOGY. 
 
 1. Professor Hale first objects that my definition of ictus as 
 'the quantitative prominence inherent in a long syllable' will not 
 hold, because it will not apply to the second long syllable of the 
 spondee when the spondee is substituted for the dactyl in dactylic 
 verse. But the second long of the spondee in such cases is not 
 quantitatively prominent. As already pointed out above, the 
 spondee is not a fundamental foot ; when it is used as a substitute 
 for the dactyl, it naturally takes on in consciousness the dactylic 
 character, i. e. the quantitative prominence is felt as resting on 
 the first syllable. The second long of the spondee is just as 
 naturally felt to be not-prominent, because it is felt in conscious- 
 ness as corresponding to the two shorts of the dactyl, which are 
 not quantitatively prominent. In defining ictus as the quantita- 
 tive prominence inherent in the long syllable of fundamental feet, 
 I by no means say or imply that every long syllable is quantita- 
 tively prominent. The situation is precisely the same as in 
 English verse. There we define ictus as the accentual promi- 
 nence inherent in a stressed syllable. Yet not every stressed 
 syllable is accentually prominent in English verse. In English 
 iambic measures the foot often consists of two stressed syllables ; 
 yet the first of these is not felt as accentually prominent, simply 
 because the verse has enough pure iambi to gain a distinct iambic 
 character (^ ^) and an occasional spondee (-^ ^) naturally is felt 
 as prominent only in the second accented syllable. 
 
 2. Secondly Professor Hale objects that my view will not hold 
 because in Latin iambic verse the tribrach and dactyl, when 
 substituted for the iambus, take the ictus upon the first of the 
 two short syllables into which the long of the iambus is resolved : 
 •-j-^^, —■ii'-j. This is a clear begging of the question. If it 
 were true that in suph cases there was a definite stress on the 
 syllables indicated, there could hardly be further discussion. But 
 that is the very point in controversy. Not a shred of evidence 
 exists to support the theory that the tribrach and dactyl were 
 stressed upon their second syllable in iambic verse. This is 
 frankly acknowledged by Christ in his Metrik der Griechen und 
 Romer^ p. 55. Christ, to be sure, as well as other authors of 
 manuals of prosody, does accept the hypothesis that the tribrach 
 and dactyl were so stressed in iambic verse. But this view is 
 simply a corollary of the false hypothesis that ictus was stress. 
 Once we assume that the iambus was stressed upon its second 
 syllable, it is not only natural but practically necessary to find a 
 
WHAT WAS ICTUS IN LATIN PROSODY? 381 
 
 location for the stress in resolved feet like the tribrach, dactyl, 
 and proceleusmatic. But that the ancients put stress either on 
 the long of the iambus or on the syllables into which it was 
 resolved remains to be proved. In fact, to my mind one of the 
 strongest arguments against the stress theory of ictus is that the 
 ancient metricians never allude to the location of the ictus in 
 resolved feet. If ictus was stress and the second syllable of the 
 iambus was stressed in verse, then the location of this stress in 
 resolved feet would be one of the first questions to suggest itself 
 to the metricians. Its consideration would have been inevitable. 
 Yet they never once allude to it, though they enumerate fre- 
 quently the various possible resolutions of the iambus. 
 
 3. Thirdly Professor Hale adduces certain passages from Quin- 
 tilian which he regards as making for the stress theory of ictus. 
 I cite these in full, italicizing the words which Professor Hale 
 deems important : 
 
 IX 4, 51 Maior tamen illic (sc. in rhythmis) licentia est, ubi 
 tempora etiam animo metiuntur et pedum et digiiorum idu inter- 
 valla signant quibusdam notis atque aestimant quot breves illud 
 spatium habeat ; 
 
 IX 4, 55 oratio non descendet ad crepitum digiiorum; 
 
 IX 4, 75 sex enim pedes (sc. of the iambic trimeter) ires 
 percussiones habent ; 
 
 IX 4, 136 [iambi] frequ^ntiorem quasipulsum habent. 
 
 Professor Hale urges that " verse-pulse is characteristic of all 
 verse-systems of which we have any actual knowledge, and can 
 hardly have been absent from the system of men who by impli- 
 cation speak of verse as lending itself to taps of the fingers and 
 beats of the foot." If by 'verse-pulse' Professor Hale means 
 stress, I must urge again that he is begging the question ; if he 
 does not mean stress, I fail to see the point of his remark. Nor 
 does the fact that in the passages above cited Quintilian speaks of 
 verse as lending itself to the taps of fingers and beats of the foot, 
 seem to me to point any more distinctly to a stress rhythm than 
 to a quantitative one. In the case of either it would be perfectly 
 natural to keep the time by taps of the finger or beats of the 
 foot. Professor Hale continues : " not only does the word 'ictus,' 
 like our word 'beat,' naturally imply stress, but it is used as 
 synonymous with percussio in Quint. IX 4, 51 [above cited]; 
 while percussio is used instead of ictus in IX 4, 75 [above cited]." 
 In answer to Professor Hale's assertion that "ictus naturally 
 
382 AMERICAN JOURNAL OF PHILOLOGY. 
 
 implies stress," I must again urge that that is only a petitio 
 principii. Ictus naturally implies stress only to those who start 
 with the assumption that it is stress. As pointed out above, the 
 word ictus is not a terminus technicus of the Latin metricians ; so 
 far as I have been able to discover, it is used only twice by the 
 systematic writers on metric, and in these two instances the word 
 cannot denote stress. The passages are : 
 
 Terentianus Maurus de Metris, 1342 (Keil, VI 366) : 
 
 una longa non valebit edere ex sese pedem, 
 ictibus quia fit duobus, non gemello tempore. 
 
 Diomedes, de Pedibus, III (Keil, I 475, 3) : ergo una longa 
 pedem non valebit efficere, quia ictibus duobus arsis et thesis, 
 non gemello tempore perquirenda est. 
 
 Here, if ictus be taken in the sense of stress, we get the extra- 
 ordinary doctrine that it takes two stresses, an arsis and a thesis 
 to make a foot. Evidently the word has no such meaning. It 
 means simply beats, — let us say a strong one and a light one. 
 Beat, stroke is the proper meaning (^propria significatio) of ictus ; 
 its figurative meanings can be determined only on the basis ot 
 actual usage (as in the two passages just cited), not by a priori 
 methods. This applies equally to the word percussio as used by 
 Quintilian, IX 4, 75 (above cited). Evidently the word is here 
 figuratively used. Mr. Hale, however, strangely denies this. As 
 I have above quoted his words he says : "ictus is used as synon- 
 ymous with percussio in Quint. IX 4, 51 ; while percussio is used 
 instead of ictus in IX 4, 75." In IX 4, 51, however, ictus is used 
 in propria sensu (digitorum et pedum ictu), so that if percussio in 
 IX 4, 75 is used instead of ictus (as in IX 4, 51) it must mean 
 'taps.' Possibly it does. But 'taps' are not vocal stress.' In 
 Quintilian, IX 4, 136 (above cited). Professor Hale declares that 
 Quintilian uses the word pulsum in place of ictus. If such is the 
 case, I would only observe that, until it is shown that the word 
 ictus was used to denote stress, the circumstance that pulsus is 
 employed as a synonym of ictus is of no significance. Certainly 
 pulsus itself does not have that meaning. To me it seems far more 
 
 ' Westphal, Gr. Rhythmik, p. 104, has collected numerous instances of the 
 use of percussio by the metricians. He recognizes the word as occurring in 
 the sense of 'interval' and of 'beat,' but, though himself a pronounced 
 adherent of the stress theory of ictus, he is not bold enough to seek in this 
 word any confirmation of that view. 
 
WHAT WAS ICTUS IN LATIN PROSODY? 383 
 
 likely that Quintilian is using tlie word in the same sense in which 
 Marius Victorinus employs it in the following passage (Keil, VI 
 44> 4) ■ pes vocatur . . . quia in percussione metrica pedis pulsus 
 ponitur tolliturque. 
 
 4. Fourthly Professor Hale adduces a passage of Charisius in 
 support of the view that ictus was stress. The context in which 
 Charisius used the word is as follows (Keil, I 552, 9) ; He is 
 discussing gender and observes that some nouns which are 
 masculine in Latin correspond to Greek nouns that are feminine. 
 This observation is followed by a list of fifty or more illustrative 
 examples in alphabetical order : ingressus «o■^o(ns• ; ictus ■nkrp/ii ; 
 iuncus, o|u(r;i;oii/or, etc. This citation is seriously advanced by 
 Professor Hale in support of the view that ictus is stress. I fail 
 to see that it has any bearing upon the question at issue. 
 
 Charles E. Bennett. 
 

 ■■mm.