1 
 
 UC-NRLF 
 
 
 
 
 
 I 
 
 SB 13 SfiM 
 
 
LIBRARY 
 
 University of California. 
 
 GIFT OF 
 
 Q^rS&*yWh Aj 
 
 Class 
 
 It' 
 

 THE BRITISH ACADEMY 
 
 Flaws in Classical Research 
 
 Rv 
 
 J. P, Postgate 
 
 Fellow of the Academy 
 
 [From the Proceedings of the British Academy, Vol, III] 
 
 
 SF\ 
 
 
 
 London 
 
 Published for the British Academy 
 
 By Henry Frowde, Oxford University Press 
 
 Amen Corner, E.C. 
 
 Price Three Shillings and Sixpence net 
 
aSTilii 
 
 FSviTlMl 
 
 '2007 
 
 ig fron 
 
UNIVERSITY 
 
 _ OF 
 
 FLAWS IN CLASSICAL RESEARCH. 
 
 By J. P. POSTGATE. 
 
 FELLOW OF THE ACADEMY 
 
 Read October 28, 1908. 
 
 When our Council honoured me with an invitation which I 
 esteemed as a command to read a paper to the Academy, I naturally 
 gave much thought to the choice of a subject. It seemed to me that 
 what might most fitly claim to occupy its attention would be the 
 communication of some discovery or novel theory of importance 
 within the regions over which its activities extend ; such would 
 be the new Indo-European language Tocharisch, the subject of 
 a memoir lately presented to our sister of Berlin. But that is not 
 possible to every one or in every season ; and I regret, without surprise, 
 that it has not been possible to me in this. In despair of offering 
 a positive contribution I turned to the other side ; and here I seemed 
 to myself to have found a larger if not a fairer field. The sur- 
 veillance, no less than the promotion of research, would appear to 
 fall within the functions of an Academy, and if the mischiefs to 
 which I shall advert exist, their recognition and their amendment 
 may be reasonably regarded as its concern. 
 
 The due performance of my task involves the criticism of the 
 utterances of contemporaries ; but inasmuch as my business is not 
 with individuals but with general types and tendencies of error, 
 I shall avoid citing names wherever this is avoidable. References 
 I must give in the interests of the argument ; and if any scholar who 
 desires to control my statements finds upon their verification de sese 
 fabidam narrari^ I trust that of my reticence at any rate he will not 
 feel reason to complain. I will only add that I have not hunted for 
 proofs of the positions, nor have I rejected illustrations that were 
 pertinent merely because they might possibly be regarded as trite. 
 
 The main differences between classical and scientific investigations, 
 technically so-called, are two. (1) The inquirer's self is implicated in 
 the classical investigation as it is not implicated in the scientific. 
 This is unavoidable. (2) The classical investigator does not correct 
 for this disturbing influence as does the scientific. This is not 
 unavoidable. 
 
 The astronomer, as a matter of course, allows for the visual 
 peculiarities of the particular observer; the physicist isolates his 
 
 Al 
 
 197969 
 
2 PROCEEDINGS OF THE BRITISH ACADEMY 
 
 personal magnetism from the sphere of his experimenting as a 
 matter of course. Does the classical inquirer commonly do any- 
 thing corresponding ? Does he ? Can we say — to touch in passing 
 upon what may be called mere human bias — that it will make no 
 more difference to the investigation of a classical problem than it 
 does to the investigation of a mathematical one if the investigator 
 has been reviewed, never mind in what terms, by the author of the 
 solution which he is considering. Has the truth about anything 
 in Homer and Vergil no better a chance of acceptance in England if 
 discovered by an Englishman, in Germany if by a German ? The 
 existence of such prejudices must not be ignored. We cannot indeed 
 hope to remove them ; but we should refrain both from palliating 
 and from inflaming them, in the hope, ere long, of establishing an 
 enlightened public opinion which shall decree that their indulgence 
 is what it is — an intellectual humiliation. 
 
 No poring upon modern superiorities can escape the chastening 
 reflection how often the recognized instructors of our public are in 
 profound and circumstantial disagreement ; how often with, presum- 
 ably, the same evidence before them they passionately or obstinately 
 maintain diametrically opposite conclusions. In textual criticism 
 this is notorious. If any ask for proof, let them compare the views 
 of Bernardakis and Wilamowitz on the Moralia of Plutarch ; or, to 
 take an instance nearer home, the Oxford and the Corpus texts 
 of Propertius. The classical, unlike the scientific inquirer, takes 
 small trouble to see that his chief instrument, his critical faculty, 
 is accommodated to his work. He passes from pure to corrupted 
 texts, or from corrupted to pure, with an unadjusted mind, cor- 
 recting what he should interpret and explaining where he should 
 amend. 
 
 Shall I say more about the idols of the textual critics ? I think 
 I will. First, then, I say that it is absurd for them to put forth as the 
 object of their activity the systematic restoration of ancient texts 'as 
 far as possible 1 to their original form, when it is notorious that, as far 
 as possible, they systematically neglect one of the means of this 
 restoration. Let it be admitted that a transposition of verses is 
 often troublesome to judge and inconvenient to adopt, and that it 
 is fair matter for consideration whether on other grounds it is expedient 
 to make the change. But let us drop the farce of pretending that 
 this has any bearing on its truth. 
 
 But on this I will not linger to-day, but proceed to what may 
 perhaps be called the Critic's Paradox. In the ordinary affairs 
 of life we aim at acting on each occasion as the balance of the 
 
FLAWS IN CLASSICAL RESEARCH 3 
 
 evidence, that is, the preponderance of probability shall determine. 
 We do not take account of the circumstance that we have chosen rightly 
 on a large number of previous similar occasions and that now it is our 
 turn to be wrong. A man of business does not refrain from taking 
 the train to the city because an accident is overdue. But a textual 
 critic of a certain school does allow his judgement upon a particular 
 passage to be discomposed by the fact that he has deviated from the 
 traditions a number of times already. I have called this habit of 
 mind the critic's paradox. But that was honouring it too much. 
 For it is but a special manifestation of the rage to make system out 
 of chance which fills the salons of Monte Carlo and makes a millionaire 
 of M. Blanc. 
 
 The dissensions of different departments are perhaps more in 
 evidence. Archaeologists, comparative mythologists, textual critics, 
 philologers and literary critics shake their fists at each other from 
 opposite sides of a channel, over which as a rule they do not adventure. 
 . . . They cross at times with disastrous results. . . . There is some- 
 thing wrong here. We are not entitled to assume that one set of in- 
 quirers is as a class intellectually less competent than another. The facts 
 of linguistics are facts just as much as the facts of archaeology, and 
 so forth ; and if the interpretation of facts tends inevitably to discord, 
 it is the mode of interpreting that must be blamed. Nor will it 
 escape the observant that the conclusions of the newer and less settled 
 branches of inquiry are not always expressed with a proper reserve, 
 when regard is had to the uncertainty of many of their data and 
 the inevitable crudeness of some of their methods. On two occasions 1 
 I have ventured over the strait which divides me from the my thologists, 
 and I have received the impression that their treatment of linguistic 
 evidence at least is not as rigorous as it might be. But I would 
 not make this a reproach against them ; for it may be conceded that, 
 even if they do not argue strictly, they argue as best they can. 
 
 Upon two sequelae of mythological inquiry I can here but briefly 
 touch — its percolations into historical research. The practice, fast 
 becoming a fashion, of treating the statements of sober historians 
 as though they were the figments of mythopoeic hallucination, and 
 that of discrediting an account of an event on the sole ground of its 
 similarity to something which has been recorded before, are two 
 
 1 In aii examination of ' The Alleged Confusion of Nymph-Names ', American 
 Journal of Philology, xvii. pp. 30-44, xviii. "4 sq., and in a criticism of current 
 misconceptions of the ' Heads of Cerberus ', Preface to the English translation of 
 Breal's Semantic* (pp. xvi-xxiv). where also some of the linguistic problems con- 
 sidered in the following pajes are touched upon. 
 
4 PROCEEDINGS OF THE BRITISH ACADEMY 
 
 procedures as likely to be as mischievous as they are illegitimate. 1 
 The former operates with a subtle and powerful solvent that will 
 destroy the fabric of ancient history : the latter challenges one of 
 the fundamental principles of all historical science. 
 
 I pass to a consideration of the difficulties which are thrown in the 
 way of the study of antiquity by the proclivities of modern life, speech 
 and thought. 
 
 To speak first of words. Our knowledge of the ancient languages 
 is and must be chiefly won through translation. On the imperfections 
 of this method it is needless to enlarge. Suffice it to cite the well- 
 known Italian proverb tradottori traditori, and to remind you of the 
 frequency with which syntactical controversy, especially among our 
 transatlantic cousins, is made to turn, not on the meaning of a 
 construction, but upon its imperfect modern renderings. 
 
 The modern languages into which Latin has entered so largely, 
 as an original or an accessory component, are full of traps for the 
 student of the ancient speech. In English, for example, corresponding 
 words no longer correspond. The dictionary translations are in 
 many cases obsolete, and their drastic revision is an urgent need. 
 Thus from subtlety, elegance, and tremendous there has evaporated or 
 is evaporating all the essential flavour of their originals. There remains 
 but the vain resemblance of sound to perplex our minds with a 
 phantom of identity. Not the least value of the recent reform in 
 Latin pronunciation is that it cuts away so many of these misleading 
 and tantalizing associations. 2 The drift in this direction since the 
 eighteenth century has been great and still increases. This struck 
 me with especial force when I saw how the last editor of the comedies 
 of Terence boggled over the play of words in the epitaph of Plautus 
 
 et numeri innumeri simul omnes collacrimarunt, 
 translating it 
 
 And Rhythms numberless all wept in concert. 
 
 The play would have caused no trouble in the days of Pope ( who 
 lisped in numbers ; for the numbers came \ The mischief is not 
 confined to derivatives from Latin. Hardly any so-called equivalent 
 of a Greek or Latin word but has its pitfall for the unwary. Because 
 1 old woman ' is slighting in English, we read in a recent note on 
 
 1 I have referred to them in a review in the Classical Quarterly for October, 
 1907, pp. 312-17. 
 
 a Teachers of Latin must ever bear this in mind. Only the other day I 
 asked a pupil of more than average intelligence why he had avoided uegetus 
 in a version. His answer was that he always associated it with vegetable. 
 
FLAWS IN CLASSICAL RESEARCH 5 
 
 Terence, Adelphoe 617 'matrona, an elderly lady, can be called anus 
 only in a slighting way'. This is not the case, as we can see from 
 Catullus ix. 4 or even from Hecyra 231 quoted by the annotator 
 himself, 4 cum puella anum suscepisse inimicitias non pudet ? ' which 
 means ' are you not ashamed at your age at quarrelling with a mere chit 
 of a girl ?' where, if anything is slighting, it is not anus but puella. 1 
 The strong temptation which besets us to give to a word the sense 
 that to us is the most familiar or impressive may be illustrated from 
 the Latin noun lacus. The modern limitation of this word, in the 
 sense of the French lac, the Italian logo, and the English lake, has 
 distorted our feeling for the Latin uses, of which this was only among 
 several. It has darkened a passage of Propertius of some literary and 
 antiquarian interest, iv. 1. 121 sqq. : 
 
 Vmbria te notis antiqua Penatibus edit 
 (mentior an patriae tangitur ora tuae ?) 
 
 qua nebulosa cauo rorat Meuania campo 
 et lacus aestiuis intepet Vmber aquis. 
 
 The straits into which an error of this kind may lead the commen- 
 tator will be obvious from a note which I will translate from the 
 German, ' We must understand the lacus Vmber which was probably 
 drained under Theodoric. At least in Cassiodor. Var. ii. 21. 2 we 
 hear of a plan for draining the " loca in Spoletino territorio caenosis 
 fluentibus 2 inutiliter occupata *\ In summer it would supply a 
 suitable swimming bath 1 (*Er wild im Sommer ein angenehmes 
 Schwimmbad geboten haben '). So disastrous to the critical vision 
 is the prepossession that lacus should denote a watery expanse that 
 marshy pools round Spoleto in the times of Theodoric have to be 
 at Bevagna some seven miles away in the time of Propertius. How 
 the muddy (caenosa) or the steaming {intepentid) waters of such 
 lagoons with their concomitants of mosquitoes and malaria would 
 be a suitable swimming bath in summer, the reader is left to divine. 
 To a Roman, however, locus was a pit, tank, or basin, with no 
 necessary connotation of extent. And here it has the sense of the 
 basin or cup from which a stream springs at its source, as in Verg. 
 Aen. viii. 74 sqq. : 
 
 quo te cumque lacus, miserantem incommoda nostra, 
 fonte tenet, quocumque solo pulcherrimus exis, 
 semper honore meo, semper celebrabere donis, 
 corniger Hesperidum fluuius regnator aquarum, 
 
 1 Similar observations might be made with regard to senex, yepav. The dis- 
 paragement of age is a privilege of the junior world. 
 
 2 Surely fluentis. Can either caenosis or fluentibus be a noun ? 
 
6 PROCEEDINGS OF THE BRITISH ACADEMY 
 
 where Servius has i dictus lacus quasi lacuna ex qua erumpens aqua 
 facit " fontem " qui cum fluere coeperit " alueum * facit \ The 
 word has the same sense in Georg. iv. 364 ' speluncisque lacus clausos '; 
 Lygdamus [Tib.] iii. 1. 16 ■ Castaliamque umbram Pieriosque lacus' ; 
 and Prop. iii. 3. 32 • tingunt Gorgoneo punica rostra lacu \ 
 
 The stream whose source is here regarded is the Clitumnus, the 
 famous Umbrian river which Propertius celebrates elsewhere, as is 
 clear from the well-known description in Pliny, Ep. viii. 8 ' Vidisti- 
 ne aliquando Clitumnum fontem ? — modicus collis adsurgit, antiqua 
 cupresso nemorosus et opacus. hunc subter exit fons et exprimitur 
 pluribus uenis sed imparibus eluctatusque quern facit gurgitem lato 
 gremio patescit purus et uitreus 1 . Further on Pliny has 'rigor 
 aquae certauerit niuibus nee color cedit \ which would be enough to 
 show that ' non tepet ' (Housman) should be restored for * irctepet ' 
 above, if the latter word were not already condemned by the parallel 
 which is adduced to support it: Stat. Theb. ii. 376 'qua Lernaea 
 palus ambustaque sontibus alte | intepet hydra uadis '. 
 
 Let me take a recent thesis and a not very ancient criticism, both 
 perhaps familiar. 'I maintain that some shall idea is the real key 
 to these [subjunctives]. If so, we English-speaking nations ought 
 to bless our stars that we have been provided by the accident of 
 language with a verb which seems to have been designed by 
 Providence to make Latin modal syntax intelligible to us \ l 
 Mr. VYs ' notion of the dative case is a case which he can translate 
 by "fur"\ 2 
 
 The perverting effect of the modern vocabulary is trifling com- 
 pared with that of the modern syntax. It is the great gap between 
 modern and ancient modes of connected expression, and the small 
 success of teaching in bridging it for the average mind that are 
 at the bottom of the outcry against Classics, which has been so 
 loud in recent years. Very few among the longer sentences of 
 modern languages would an ancient Greek or Roman have recognized 
 as sentences at all — hardly any in English, a few more in French, 
 and still a few more in German. And for a very obvious reason. 
 To them it was of the essence of a sentence that the structure and 
 the thought should be conterminous. Towards our 'sentences' he 
 would have much the same feelings as a self-respecting vertebrate 
 towards a worm or other similar creature, divisible, without injury 
 to its economy, at almost any part of its length. 
 
 One main principle which it takes some trouble to grasp, and still 
 
 1 Proceedings of the Classical Association, 1908, p. 29. 
 
 2 A. E. Housman, Preface to M. Manilii Astronomicon Lib. I, 1903, p. Ii. 
 
FLAWS IN CLASSICAL RESEARCH 7 
 
 more to apply with precision, is that, within certain wide limits, 
 order in modern sentences is syntactically essential and in ancient 
 sentences syntactically indifferent. The modern sentence, to put it 
 roughly, is an arrangement in line, the ancient one within a circle. 
 Now the lineal habit of mind, if I may call it so, is often at a loss 
 when it has to understand the circular ; it is devoid of the sense 
 of grouping ; it has not been trained to the necessary attention. 
 If the groups are small, the trouble thus caused is small ; but it 
 is not absent altogether. In the second half of the pentameter 
 Tibullus writes uir multerque (ii. 2. 2), Ovid femina uirque. The 
 difference of order is absolutely without significance. But the lineal 
 mind is apt to imagine that some subtle distinction between the 
 places of man and woman is intended, as though Ovid were a sort 
 of pro- and Tibullus an anti-suffragette. Terence, Hec. 315, has 
 rursvm prorsum, which judged by ' lineal ' standards is strictly inde- 
 fensible ; compare to and fro. It is only because ' we English- 
 speaking nations' happen to have a similar neglect of sequence in 
 backwards and forwards that this does not strike us as strange. 
 In a recent note on Ter. Hec. 159 sq. 'sed ut fit, postquam hunc 
 alienum ab sese uidet, | maligna multo et magis procax facta ilico 
 est \ it is said of the second line that the order is * capricious \ The 
 order is not capricious. It is not (that is true) the order of a 
 Latinist of the twentieth century a.d., who would doubtless prefer 
 1 multo magis maligna et procax \ But it is just as clear and far more 
 effective, if the sentence is taken as a whole and due heed be paid 
 to the binding alliteration (pp. 38 sqq. below). A good many years 
 ago Mr. T. E. Page 1 called attention to the irrationality of current 
 views of the figure called hysteron proteron, as in Eur. Hec. %66 
 kslvt] yap vXeo-iv viv h TpoCav t &y€i. To the lineal mind these 
 1 inversions ' are nonsense ; to the circular but legitimate variations. 
 I have lately 2 referred to hyperbata or dislocations of order, and 
 shown how in Catullus lxvi. 77 an hyperbaton has caused the greatest 
 trouble to a long succession of scholars 3 who attacked the passage 
 upon lineal theories. The real character of such arrangements is 
 seen in passages like Ter. Ad. 917 ' tu illas abi et traduce ; and Lucan, 
 viii. 342 sq. • quern captos ducere reges [ uidit ab Hyrcanis Indoque 
 a litore siluis \ which almost shriek at us the warning respice finem. 
 
 Not only does the lineal habit hinder our sight of real connexions 
 between the distant members of a sentence, but it causes us to find 
 
 1 Classical Review, viii. 1894, p. 204. 
 
 2 Classical Philology (July, 1908), iii. p. 259. 
 
 3 Not excepting the last editor of the poem (Teubner, 1908). 
 
8 PROCEEDINGS OF THE BRITISH ACADEMY 
 
 imaginary bonds between adjacent ones. In the article above referred 
 to (p. 260) I cited two passages where words in tempting juxtaposition 
 to the lineal mind have been taken together without regard to the 
 sentence as a whole. One of these, Ar. Lys. 628 koX btaWdrTetv irpbs 
 fifxas avbpdo-iv AaKMVLKois, has an adverb (npos) in a place where it 
 aggrieves us by not being a preposition. Sometimes the offended 
 lineal sense is soothed with a label on the offending order. At Ter. 
 Hec. 364 ' qua me propter exanimatum citius eduxi foras ' you will 
 find that qua me propter is a ' tmesis ' for quapropter me, 1 and that 
 at Hec. 58 ' per pol quam pannos ' per pol quam is again a ' tmesis ' for 
 pol perquam. The Romans had a way of putting per where we do 
 not expect or approve of it, and however many times we may have 
 met the Latin formula per te deos oro we settle with satisfaction on 
 a passage like Horace, Odes i. 8. 1 ' Lydia, die per omnes | te deos oro ', 
 because there the poet has happened to leave the per in front of an 
 accusative with which we can construe it. These 'inversions' or 
 * dislocations • are not confined to per or to Latin. Mr. Housman 
 on Manilius i. 245 has given an ample collection for other pre- 
 positions, and within the last few weeks I have had ocular demonstra- 
 tion of the havoc which may be wrought among translators by 
 Callimachus's inconsiderate arrangement of the words es be bdnpv \x 
 ijyayz as is bi /me bdicpv \ ijyaye Anih. Gr. vii. 80. For other examples 
 I may refer to my remarks in the Journal of Philology, vol. 17, 
 p. 260. I commented there on misunderstandings of the Greek 
 article for which juxtaposition was responsible. The warning of 
 twenty years ago is still by no means superfluous. 
 
 What the ancients called hyperbaton in syllables is from the 
 modern point of view so singular that it demands a separate mention. 
 Tryphon (Boisson. Anecd. iii. p. 274) has hioi be kch ev rats o-vWafials 
 
 VTtepfiaTCL 7T€TT0LlflKa<TLV 0)9 KOI 2LfJL(t)Vlbr)S €V €77 LypdfXfJLa(TL 
 
 'Epfxrjv Tovb* &viBr] ArjfxrjrpLos opOidbov km 2 
 €v TrpoOvpois 
 
 Besides this dviOr] — Kev of Simonides we may set the lines of Ennius — 
 
 saxo cere- comminuit -brum 
 and lagoenas 
 
 Massili- portabant iuuenes ad litora -tanas. 
 
 (For more examples see L. Mueller's de re metrica, pp. 457 sqq.) 
 
 1 There is an exact parallel to this in Hor. Sat. ii. 6. 95 ' quo, bone, circa', 
 ' tmesis rarissima ', say the annotators. It is fortunate that the students of 
 English are exempted from learning that e to us ward ' is a tmesis for s toward us '. 
 
 2 So acutely restored by the brilliant Greek scholar whose recent death we 
 deplore — W. G. Headlam, Journal of Philology, xxvi. p. 93. 
 
FLAWS IN CLASSICAL RESEARCH 9 
 
 These are anyhow strange licences or, if we must adopt the con- 
 demnatory tone of Headlam upon the Greek example, ' grave lapses ■ 
 of language. But to modern speech they are more ; they are sheer 
 impossibilities. And so they are not imitated and cannot even be 
 reproduced. 
 
 Hypallage, a figure which, wherever there is an opening, the 
 modern annotator is prone to misunderstand (for example you will 
 find the os trilingue ascribed by Horace to Cerberus, Carm. ii. 19. 31, 
 still explained as a ' three- tongued mouth ' instead of a ■ triple mouth- 
 and-tongue '), is intelligible as soon as the principle of the sentence's 
 totality has been grasped. There is no real disorder in Homer's 
 Topyeiriv K€(f)a\i}v beivoio neXtopov or Vergil's arma dei Volcania, and the 
 mobility of thought which enabled an ancient to say what to our 
 analysis is * the seven-walled exits ' ray knTOLTcix&s efo'Sous (Aesch.), 
 though it means ' the seven exits of walls ', is capable of still further 
 extensions. 
 
 In his Pharsalia, book viii. 542 sqq., Lucan, to convey the thought 
 'Gods, who would have thought there was such daring within the 
 bounds of Egypt ? ', writes : 
 
 O Superi, Nilusne et barbara Memphis 
 et Pelusiaci tam mollis turba Canopi 
 hos animos ? 
 
 Now Canopus was on the westernmost, but Pelusium on the eastern- 
 most arm of the Nile ; and less than eighty lines back the poet 
 has defined its position, ■ qua diuidui pars maxima Nili | in uada 
 decurrit Pelusia septimus amnis.' But since to us the phrase 
 ' Pelusian Canopos ' is as meaningless as ' Doverian Folkestone ' would 
 be, it is the fashion to say that ' Pelusian ' simply means Egyptian, 
 thus imputing to the author a gross, if concealed, tautology. But 
 a Roman reader would have understood without more ado that 
 Lucan meant the inhabitants of Egypt wherever they were to be 
 found, the population of the river-side to Memphis in the interior and 
 that of the Delta seaboardjfrora Canopies to Pelusium : and the censure 
 of Oudendorp ad loc. * eodem iure alicui dicere liceret Pelusium Cano- 
 picum cum sint duo extrema Nili ostia, Pelusium ad Orientem, 
 Canopos ad occidentem ' would have seemed to him its sole and best 
 defence. 1 
 
 1 The phrase ' caught on \ Not only does Statius quote it in his tribute to 
 the memory of Lucan, Situ. ii. 7. 70 ' Pelusiaci scelus Canopi ' ; but the wordy 
 paraphrast of Dionysius Periegetes substitutes it for the 'A/iu/cXat'oto Kavufiov of 
 his original e et Pelusiaci celebrantur templa Canopi ', Auien. iii. 24. Lastly we 
 meet it in Sidonius, Carm. ix. 27. 4. 
 
10 PROCEEDINGS OF THE BRITISH ACADEMY 
 
 Look out Aganippis in a well-known dictionary and you will find 
 6 Aganippis, -idis,f., that is sacred to the Muses : fontes Aganippidos, 
 Hippocrenes, Ov. F. 5. 7 ' ; refer to a commentary on the passage 
 quoted and you will see that it is to mean 'inspiring song 1 ( 4 be- 
 geisternd ') ; turn to a leading Latin lexicon 1 and you are presented 
 with both, ' der den Musen heiligen u. begeisternden \ As Aganippe 
 and Hippocrene were separate but adjoining springs, these efforts of 
 interpretation would correspond to explaining a phrase 'Buxtonian 
 Matlock' as either (1) 'anti-rheumatic Matlock' (viz. Matlock with 
 the properties of Buxton), or (2) ' Aesculapian Matlock ' (Matlock 
 sacred to the Buxton god of healing), or (3) as ■ Aesculapian and anti- 
 rheumatic Matlock V Now what are the reasons for such contortions 
 of interpretation ? There are two. The first is that the commen- 
 tators are aware that Ovid knew perfectly well that Aganippe and 
 Hippocrene were different springs, Met. v. 312 ' fonte Medusaeo (i. e. 
 Hippocrene) et Hyantea Aganippe ', and the second is that we are 
 by first and second nature constitutionally unfitted to conceive of 
 a state of mind to which ■ Aganippw? Hippocrene ' and ■ Aganippe 
 Hippocrenw ' could mean exactly the same thing, viz. the couple of 
 hippine springs, Aganippe-Hippocrene. For this was what Ovid 
 intended by the phrase which has puzzled us. 
 
 dicite quae fontes Aganippidos Hippocrenes 
 grata Medusaei signa tenetis equi. 
 
 Attentive consideration of these passages enables us to pronounce 
 upon a much canvassed line of Propertius, iii. 22. 15, where the 
 MSS. have 
 
 et si qua orige uisenda est ora Caystri. 
 
 Here Haupt saw the sense which was required when he emended 
 4 et sis quae Ortygia ( = Ephesus) et uisenda est ora Caystri '. But the 
 true correction had already been made by J. Voss, Ortygii — Caystri, 
 that is ' the Cayster with Ephesus '. In Ortigii, ti fell out before gi, 
 
 1 The great Thesaurus is silent. 
 
 2 I would apologize for these crudities of representation. But for the detection 
 of the impostures the impostors must be stripped of their ancient garb. A fourth 
 attempt at a solution I have passed over. Of Aganippe, says another dictionary, 
 1 on la confonde avec l'Hippocrene.' If 'on' means Ovid, the answer to the 
 calumny is below. If ' on ' is to have a wider reference, where, we may ask, 
 outside of an asylum can such a confusion be engendered ? Show us the place, 
 and we shall know it for one where ' Brighton-Hastings ' or ' Brightonian 
 Hastings ' may mean { Hastings, which the writer confounds with Brighton ', 
 and e fox-dog ' denote some chimaera of a fox and a dog. 
 
FLAWS IN CLASSICAL RESEARCH 11 
 
 and orgii easily became orige which with the copyist stood for a real 
 word aurigae. 
 
 The greater ease of what we might call the ' intertransience ' of two 
 ideas in the circular grouping is perhaps the reason of the curious 
 employment of abstracts in Latin which strikes as particularly 
 strange in a language with such a love for the concrete. Such as 
 fonthim gelidae perennitates, Cicero, corui deceptus stupor, Phaedrus, in 
 which the quality is not ■ applied ' to a subject, but is regarded, so to 
 speak, as blended with it, ' The cool fountain-flows,' • the cheated 
 crow-fool.' 
 
 It is an axiom of classical exegesis that single Greek and Latin 
 words must often be treated as the equivalent of a phrase, the modern 
 mind analysing much that the ancient viewed as integrities. But 
 this generality is apt to be neglected where its application cannot 
 be immediately perceived. Let me first cite an example from a 
 Latin writer in which the ancient habit has produced a variation of 
 expression at once evident and instructive, Silius i. 627 sqq. ' sic 
 thalami fug it omnis amor dulcesque marito | effluxere tori et subiere 
 obliuia taedae '. 
 
 In Lucan x. 184 sqq. Caesar is explaining to Acoreus, the priest 
 of Isis, his desire to become acquainted with the mysteries of Egypt, 
 and in particular with the solution of that scientific riddle, the origin 
 of the Nile. 
 
 fama quidem generi Pharias me duxit ad urbes, 
 
 sed tamen et uestri. media inter proelia semper 185 
 
 stellarum caelique plagis superisque uacaui. . . . 
 
 sed cum tanta meo uiuat sub pectore uirtus, 
 
 tantus amor ueri, nihil est quod noscere malim, 
 
 quam fluuii causas per saecula tanta latentis 190 
 
 ignotumque caput, spes sit mihi certa uidendi 
 
 Niliacos fontes; bellum ciuile relinquam. 
 
 irirtus does not mean 'bellica virtus' (Sulpicius), nor 'ardour' 
 simply (so the last translator of the Pharsalia into English verse) ; but, 
 as the correspondence of amor ueri shows, ' passion for excellence '. 
 Weise's ' tantus vigor ad maxima quaeque vel perficienda vel cogno- 
 scenda ' is for a paraphrase perfectly correct. 
 
 The modern bias is responsible for some part of the difficulty 
 presented by a well-known crux in the Satires of Horace. S. ii. 2. 
 9 sqq. 
 
 leporem sectatus equoue 
 lassus ab indomito uel, si Romana fatigat 10 
 
 militia adsuetum graecari, seu pila uelox 
 
12 PROCEEDINGS OF THE BRITISH ACADEMY 
 
 molliter austerum studio fallente laborem 
 seu te discus agit 9 [pete cedentem aera disco] 1 
 cum labor extuderit fastidia, siccus, inanis 
 sperne cibum uilem. 
 
 The propriety of agit here with discus has been questioned not 
 without reason ; but the doubt disappears when, following the cue 
 given by studio in the previous line, we observe that discus is to be 
 understood as ' studium disci ' ; cf. Propertius i. 13. 28 * te tuus 
 ardor aget \ discus is, in fact, used here just as aurum is in Propertius 
 iii. 5. 3 
 
 nee tamen inuiso pectus mihi carpitur auro, 
 
 a line which I have quoted in full in order to remove a prevalent 
 misunderstanding as to its meaning, in-uidere is ■ to cast an envious 
 eye on • a thing or person, and inuisus is used of that on which such 
 a glance is cast. An envied person comes naturally to be hated, 
 €7rC(j)0ovos ; and in Latin this sense supplanted the original one in the 
 case of persons, and was even transferred to things, to which it was 
 etymologically inappropriate. Of these, however, the participle 
 could be used in the old sense which Horace expresses by means of 
 inuidendus, e. g. Carm. ii. 10. 7 ■ caret inuidenda | sobrius aula \ 
 This is its meaning here, and in Ovid, Met. xi. 127 sqq. of the plight 
 of Midas when his greedy prayer was granted. 
 
 diuesque miserque 
 effugere optat opes et quae modo uouerat odit. 
 copia nulla famem releuat; sitis arida guttur 
 urit et inuiso meritus torquetur ab auro. 
 
 Here ■ hated ' gives no sense. Midas had not hated gold ; he had 
 loved it only too well ; and hence he was ' male ojptato — circumlitus 
 auro' (infra 136). What, however, he had done, was to cast an 
 envious eye upon it, as the avaricious man does ; and for this he is 
 rightly tormented (' meritus torquetur '). 
 
 Few inquirers are proof against the charms of the siren etymology, 
 and when a captivating derivation comes into conflict with ancient 
 authors and evidence, the ancient and the evidence must commonly 
 retire. A now innocuous discussion of Max Miiller 2 is so instructive 
 that I must put it first in my illustrations. Varro in his work on the 
 Latin language, vii. 73-5, in treating of the origin of septemtrio? 
 
 1 The words in brackets are not from Horace ; but I leave them in the text as 
 they do not affect the present question. 
 
 2 Max Miiller, Science of Language, 2nd Series, Lecture VIII, p. 364. 
 
 3 The remarkable ' tmesis ' of this word in Vergil (Georg. iii. 381) and Ovid 
 (Met. i. 68) may be noted in passing. 
 
FLAWS IN CLASSICAL RESEARCH 13 
 
 septemtriones, the Great Bear, tells us that in his time oxen were still 
 called triones by the countrymen, especially when ploughing, * triones 
 enim boues appellantur a bubulcis etiam nunc maxume quum arant 
 terram. , ■ If,' said Max Muller, ' we could quite depend on the fact 
 that oxen were ever called triones, we might accept the explanation of 
 Varro and should have to admit that at one time the seven stars 
 were conceived as seven oxen. But as a matter of fact trio is never 
 used in this sense, except by Varro, for the purpose of an etymology, 
 nor are the stars ever again spoken of as seven oxen, but only as " the 
 oxen and the shaft"", boues et temo, a much more appropriate name.' 
 This is said to pave the way for the acceptance of Max Muller's own 
 derivation from septem striones, ' the seven stars,' on which we may 
 follow his own example just so far as to observe that strio, ' a star,' is 
 never used in this sense, except by Max Muller, for the purpose of an 
 etymology. 
 
 In the last etymological dictionary of Latin we read 'proelium, 
 " Kampf, Treffen," richtiger praelium oder prelium \ This ■ incorrect- 
 ness ' oiproelium will seem strange to those who are aware that it is 
 vouched for by the Fasti Capitolini, the Monumentum Ancyranum, 
 the capital MSS. of Vergil, and so forth : nor will the suggestion that 
 prae or pre, either common enough as the initial syllable of Latin 
 words, has been here corrupted into an absolutely unique beginning 
 proe appear less remarkable. But the explanation is simple. The 
 compiler of the lexicon cannot derive proelium and thinks that he can 
 derive praelium or prelium. J I do not understand measles,' said the 
 doctor in the well-known cynical anecdote, ' I will give the child fits 
 and cure them.'' 
 
 Lexicography, and, in its train, the interpretation of literature, 
 do not escape. The lexicon of Latin used most in this country has, 
 under trames, * akin to trans and Gk. rippa, goal. I. Lit. a cross-way, 
 side-way, by-path, footpath. Transf. 1. Poet., in gen., a way, path, 
 road, course, flight. 2. Branches of a family, Gellius. II. Trop., 
 a way of life, way, course, method, manner, Lucretius. 1 ' 
 
 The entire presentation of this word has been perverted by 
 etymology, trames does not mean a ' cross- way ' or ' by-path ', nor 
 again a ' way, path in general ', nor again a ' branch of a family ', nor 
 a • way of life, or method, or manner '. It means a path on a slope, 
 a narrow mountain track or what may be compared thereto. In all 
 
 1 We know where all this comes from ; for if we turn to the article in 
 Freund's lexicon we read : ' trames, der Querweg, Seitenweg, Nebenweg . . . 
 £) Uebertr. 1. poet., im Allgem. fur Pfad, Weg, Gang, Flug u. dgl.— 2. fiir 
 Abzweigung der Familie. *II. Tropisch. Weg, Methode. (Lucr.) 
 
14 PROCEEDINGS OF THE BRITISH ACADEMY 
 
 the passages cited it may have this reference. All, except Hor. 
 S. ii. 3. 49, lose both in force and pertinence if it have it not. 
 Appennini tramites Cic. Ph. xii. 26. transuersis tramitibus transgressus } 
 Livy ii. 39. 3 (of a hill country, as a glance at the map will show). 
 Prop. iii. 13. 44 ' et quicumque meo tramite quaeris auem ', a translation 
 from the Greek of Leonidas Tarentinus, roO0' vttb hia-abv opos. Of Iris's 
 rainbow-path down the clouds, Verg. A. v. 610 'cito decurrit tramite 
 uirgo\ Of the lines of a stemma or pedigree by which a man 6 genus 
 deducit *, or traces his descent. In the passage of Lucretius, vi. 27, it is 
 used significantly of the narrow upward way ; Epicurus ' uiam mon- 
 strauit tramite paruo | qua possemus ad id recto contendere cursu', with 
 which compare Pers. 3. 51 i surgentem dextro monstrauit limite 
 callem \ In the simile of Hor. S. ii. 3. 49 the trames is in forests 
 (siluis), but there is no reason why these forests should not be on 
 a mountain side. 2 
 
 There is small excuse for this blundering. Many years ago 
 Quicherat in his Gradus ad Parnassum led the way towards a 
 correct explanation, though he could not by any means shake off all 
 the misleading associations of his ' chemin de traverse \ 
 
 Whether trames originally meant a ' cross-path ', it is not within my 
 present purpose to determine ; suffice it to say that this sense is 
 nowhere apparent in its usage. And it may be added that we are not 
 limited to this supposition. It may be that trames is directly derived 
 from trama, a synonym of subtemen. For such epithets of trames as 
 paruus and angustus might be quite appropriately connected with the 
 slender threads of the woof or weft ; and the trames, winding backwards 
 and forwards up or down the mountain -side, presented an obvious 
 similarity to this thread in its crossing and recrossing of the vertical 
 loom. 
 
 Archaeology and history suffer too. I will take pomerium as my 
 example. The tendency of which I speak has seriously impaired the 
 otherwise excellent investigation of Mommsen in his Romische For- 
 schwigen, II, p. 23 sqq. Mommsen, whose etymologizing was intuitive 
 rather than scientific, was dominated by the theory that pomerium 
 must have come from post and murus (or O.L. pos and rnoerus). 
 
 1 What a passage this for the cross-path interpretation, ( having crossed by 
 cross cross-ways ' ! 
 
 2 There can he no such doubt about Prop. i. 18. 27 ( pro quotdiuini fontest 
 et frigida rura | et datur inculto tramite dura quies ', where the last German 
 commentator writes, ( Trames ist hier nicht der Waldpfad, sondern der Wald 
 selbst ' — two errors in as many lines. This is outdone by an English comment 
 on iii. 22. 24 ' ab Umbro tramite', 'from its Umbrian path,' i.e. l from the 
 vales of Umbria '. 
 
■ 
 
 FLAWS IN CLASSICAL RESEARCH 15 
 
 4 Nothing remains, 1 he says on p. 28, ' but to return to that explana- 
 tion which etymologically is the sole one possible (welche etyraologisch 
 die allein mogliche ist), 1 and, lightly brushing aside the far from 
 trivial objections to which as a derivation it is open, 1 he decides that 
 the pomerium lay inside the walls. The procedure is illegitimate. For 
 etymology is not evidence ; it is inference, and, as experience shows, 
 often most erroneous inference. 
 
 The same derivation dominated antiquity ; and this must not be 
 lost sight of when we would appraise the value of the ancient 
 testimonies. There were two views held amongst the ancients as to 
 what the Roman pomerium really was. Varro, L. L. v. 143, and 
 Messalla, quoted by Gellius xiii. 14, are cited as witnesses that it 
 was a strip of land running round inside the city walls post muros ; 
 Livy i. 44 as a witness that it was one on both sides of the city wall 
 inside and out, circa muros. Now let us ask, Of two witnesses, one 
 believing and the other disbelieving, that pomerium was derived from 
 post and murus, which would be the better witness for the supposed 
 fact, that it was actually post murum ? Obviously the disbeliever. 
 For on him this theory of the origin of the word could have exerted 
 no disturbing influence. Varro, however, is a witness belonging to 
 the former class : * qui orbis, quod erat post murum, postmoerium 
 dictum.'' 
 
 Next let us turn to Livy. I will translate the Latin. He says 
 
 i pomerium is explained as postmoerium or ■ behind- wall \ regard 
 being had to the etymology alone ; but it is rather circamoerium 
 or 4 round-wall ', a space which the Etruscans, when founding 
 cities in old times, used to consecrate with augural rites along 
 a line, marked by a series of boundary-stones on both sides 
 of what was to be the course of their wall, that buildings might not 
 be erected contiguously to the walls on their inner side (nowadays 
 these are often in actual contact) and that outside them might- 
 be ground upon which human cultivation did not encroach. 
 This space, which it was forbidden both to plough and to 
 occupy, was called pomerium, ' behind-wall,' by the Romans, not 
 more because it was ■ behind wall ' as because • wall was behind it'. 
 
 The evidential value of such a passage for the actual character of 
 the pomerium is intrinsically of the highest order. For Livy, or his 
 authority, is so dominated by the view that pomerium comes from 
 post and murus that he endeavours to reconcile this with the fact that 
 the pomerium was circa muros by an etymological explanation which, 
 as he states it, must be admitted to be absurd. 
 
 1 p. 25, n. 9. Of his examples of ' irregular ' sound-change there quoted 
 most would now be otherwise explained. 
 
 > OF THE 
 
 UNIVERSITY 
 
 OF 
 
16 PROCEEDINGS OF THE BRITISH ACADEMY 
 
 Messalla alone remains. What he, or rather what the * augures 
 populi Romani qui libros de auspiciis scripserunt ', laid down is as 
 follows : — 
 
 pomerium est locus intra agrum effatum per totius urbis cir- 
 cuitum pone muros regionibus certis determinatus qui facit 
 Jinem urbani auspicii. 
 
 The last words give us the clue. When the inner ring of the 
 pomerium which lay pone muros was crossed, the urban auspices lost 
 their power, just as the military auspices (bellica auspicia) became 
 inoperative when the general crossed its outer ring. And since the 
 crossing of the inner ring was far more under public notice than the 
 crossing of the outer ring, the word was used with this special 
 application. We must therefore accept the view adopted by 
 O. Muller, Becker, and Schwegler that the pomerium, in the proper 
 sense, ran along both sides of the wall. 1 
 
 It might perhaps be deemed superfluous to remind the scholars of 
 to-day that no portion of the past can be understood unless we arrive at 
 it by the historical path and cease to view it as something out of 
 relation to what preceded and what ensued. Unfortunately it is not. 
 
 The Homeric Article is a well-worn theme. Its half demonstrative 
 character and the impediments which the associations of modern 
 languages and of later Greek throw in the way of our appreciation 
 of its more ancient usages are familiar topics. But quod quisque 
 uitet, numquam homini satis cautumst in horas, and the ap- 
 proaches of error are here especially insidious. Of the three 
 stages in the history of the 'article', 6 avOpanos, (1) he [i.e.] 
 man ; (2) that man, ille homo ; (3) the man, Vhomme, it is the second 
 and third which require most careful discrimination. For their 
 difference is rather quantitative than qualitative, and there is no 
 half-way house on the road between them. Unless we are pre- 
 pared to see in Iliad x. 408 ttw? 8' at t&v tiAAa>z> TpaW <f)v\aK.aC re 
 not evvaC; an extravagant emphasis, we must concede to Monro, 
 Homeric Grammar, § 261. 3, that it presents 'the defining Article 
 of later Greek 1 . But in the three examples which are given at 
 the close of the same paragraph that would be a less imperfect 
 representation than the; II. ii. 80 ei /xeV tls tov oveipov 'A^atwi; 
 aXXos €vkttt€, | \jfevbos kcv (fraifj.cv, vii. 412 a>? elirwv to crKrjiTTpov 
 <Wo-xe0e nao-i deoio-L (the to calls attention to the uplifted staff) t 
 xx. 147 o<ppa to ktjtos vireKirpo^vyaiv aXiatro, \ SinTOTt pip aevatro aii 
 
 1 I cannot go further into the matter, but I must add that Mommsen candidly 
 admits that the leaving of a strip of ground clear on both sides of the wall 
 was a very reasonable safeguard. 
 
FLAWS IN CLASSICAL RESEARCH 17 
 
 rjiovos ireblovbc (that sea monster, the great sea monster). And if 
 ever there was a passage in which solemn emphasis was expressed 
 by article or pronoun adjective, it is surely x. 330 fxr] pev toIs 
 Xttttolo-iv avr]p ZiroxweTai a\\os. What, again, has the article of xxiii. 75 
 done to be catalogued as ' quite anomalous ' ? The spirit of Patroclos 
 pleads piteously to his slumbering friend, beginning evSeis, avrap 
 tyeio \iXao-jx4vos eirktv, 'A\lW€V, | ov \xkv /xcv (coovtos aK-qbtLS aWa 
 OavovTOs . . . Kai jjlol bos rr\v X € V okoGbvpofAaf ov yap er' avrts \ viarofj.aL 
 ef 'Ai'8ao iirriv jue irvpos \e\axnrt. Are we to surrender to gram- 
 matical classification the natural and pathetic touch in rrjz/ x € W a ? 
 Let a poet answer : 
 
 And hands so often clasped in mine 
 Should toss with tangle and with shells. 
 
 Tennyson, In Memoriam, x. 
 
 Monro's exposition of the Homeric article is, however, in the main 
 just and sober, and contrasts strikingly with the treatment to which 
 it is subjected in a recent volume of acute and ingenious studies 
 on the Odyssey. The design of the critic is to restore the pristine 
 Homeric usage to that poem, and he effects this by the wholesale 
 removal of articles in which he detects the trail of the modernizer. 
 That the language of the Homeric poems has suffered, it is difficult to 
 say how much, from being brought up to date, is hardly a matter for 
 dispute. But the task of restoration requires great caution and much 
 self-control, and a suspected usage must be examined, so to say, both 
 from before and behind. There was no sudden and tropical transforma- 
 tion of non-articular night into articular day. Rather between the 
 two lay a long and uncertain twilight, and even after the illumina- 
 tion was general, in hollows and under heights there lingered patches 
 of pre-articular shade. This is utterly neglected in the book to which 
 I refer. Where 6, fj, to can be understood as unemphatic Attic 
 articles, so understood they are, and bidden to evacuate the text 
 for substitutes which often possess neither literary appropriateness 
 nor palaeographical probability. What we may perhaps call the 
 6 intermediate article , — the the which has lost some of its mobility but 
 still retains its independence — naturally suffers most. Od. xviii. 74 
 o1t)v ex paKiaiv o ytpuv iinyovvLba <paivei is a striking example. For it was 
 1 probably 1 (note the probably) 'modified for the better accommodation 
 of the article from an original : oh]v Iv paKUcrai yipuav kt\? Is the 
 6 yepwv so otiose that it must be got rid of at such a cost ? ■ Yet 
 who would have thought the old man to have had so much blood 
 in him V— Macbeth, Act V, Sc. 1. 
 
 A2 
 
18 PROCEEDINGS OF THE BRITISH ACADEMY 
 
 Nearly every one of the articles which are attacked on the ground 
 of their coincidence with the later unemphatic ones does more 
 than simply define, e. g. Od. xi. 4 tol p.rj\a, not ' the sheep ' but 
 * the (necessary) sheep' (x. 572), xx. 77 ras Kovpas, 'the luckless 
 maidens', xix. 535 tov oveipov, 'the dream / must narrate to 
 you\ on which we are told that 'this is the only passage in the 
 Odyssey in which oveipos is accommodated or encumbered with the 
 article ', and the only one, we may add, in which there is purposed 
 reference to a particular dream. A similarly crude appeal to numbers 
 is to condemn xxi. 113 kclI hi kcv au-rds eyo> toO to£ov Trtipijo-aLiMriv 
 (Telemachus), 305 &? kclI aol fxiya Trfjpt 7n^>avo-KO/xat at k€ to to£ov \ 
 ivTavvo-ps (Antinous), 378 tA 8e r6£a (jyipcov ava da>/aa avfitoTrjs \ kv 
 XeCpea-a 'Obvo-rji balcppovi 6rj<€ Trapaards. ' The twenty-first book of 
 the Odyssey has the doubtful distinction of possessing the only 
 three examples of to£ov with the later article.' * But in 305 the 
 minatory tone is clear, ' if you draw that bow ! ' At 113 we think 
 of 'Le sabre de mon pere'. In 378 the alteration 6 hi spoils the 
 grouping of the picture, the centre of which is not the mere con- 
 veyer of the bow, but the bow itself, whose destination had just 
 been the subject of an angry dispute ; see 359 sqq., 366, 369 sqq. 
 Places where the noun has an attribute fare no better : ii. 403 
 tt]v <rr)v TTonbiyixevoL opfxriv, ' ilium tuum cursum expectantes ', iii. 1 45 
 tov *A6i]vair]s beivbv x°^°Vi * terribilem illam Mineruae iram ', xi. 519 
 &AA' oiov rbv TrjXecptbrjv KarevripaTo x a ^ K< ?s ' a ^ qualem ilium Tele- 
 phiden interfecit ! ' 
 
 Other usages of the Homeric ' article ' are assailed by means of the 
 same statistical fallacy and without even the excuse of the modernizing 
 scribe. On ra cKaora (xii. 16 ; also 165, xiv. 375, and //. xi. 706) it 
 is observed that 'against these four we have to set twenty-five 
 instances of tKao-ra without article in the Odyssey alone '. This 
 article is said to be ' quite needless '. Which means that the construc- 
 tion can dispense with it, an engine of argument that could be used 
 to decimate the ranks of the demonstratives in many literatures. 
 But the To., ' needless ' though it may be for the syntax, is not quite 
 needless for the sense. €KaaTa means ' each thing ', ra eKaa-ra means 
 ' those things, each one of them ', and expresses at xii. 16 fjp.€ls \xev t& 
 $Kaora 8tet7ro/uey the scrupulous care with which the details of the 
 unfortunate Elpenor's funeral were carried out (supra, 11-15), and 
 
 1 A similar argument is used against ttjv in xviii. 380 ov& av fxoi tt\v yaarep' 
 6vei8i{a>v dyopevots, the only place where yaarrjp takes the article, because the 
 only place where it is appropriate. It points the allusion to the still rankling 
 insults of xvii. 228 and xviii. 364. 
 
FLAWS IN CLASSICAL RESEARCH 19 
 
 at 165 it impresses on the reader the particularity with which 
 Odysseus required his crew to attend to his directions when they 
 sailed past the dangerous coast of the Sirens (supra, 156-64). 
 
 Another peculiarity, treated with the same severity, is, if anomalous 
 in Homeric, not less anomalous, to say the least, in later Greek, 
 where its survival from the ancient times of freedom has, for example, 
 troubled much the commentators upon Sophocles. The instances are 
 ix. 378 a\X* ore 6t) tolx ° f*°X^s Adu'os eV Tivpi fiiWtv \ atyzaBai 
 (here we are told that ' 6 plox^os e\cuz/os condemns itself'. This 
 means from the Attic standpoint, the critic forgetting for the nonce 
 that from his point of view 6 Zhdivos juox^oy is equally objectionable), 
 xi. 492 aX\ y aye p.oi tou iraiSo? dyauou \ivQov €vlo"ne(s) (bidden to make 
 way for aXk J &ye \i au-riica iraibos), xvii. 10 rbv £elvov hvarrjvov ay h 
 tioXiv o(f) p av €K€i0i | halra 7jra>x€w? (gov is read). 1 To which we 
 may add xxiii. 223 sq. tV b' art\v ov TrpoaQev ea> ZyKarOeTo 0v/x(j> \ Xuyp^y, 
 e£ 17s TTp&ra /cat rjp.eas i*ero irevdos, and from the Iliad i. 338 sqq. tg> 5* 
 avTQ) fjidprvpot eorwv | irpos re Oe&v fxaKapav irpos re dwqT&v avOpcoiKav \ 
 Kal irpos tou PacriXfjos d-m^cos, ii. 275 os rbv XwprjTYJpa i-n€<rf&6\ov lo~x 
 ayopacnv (though to what extent e-neo-fioXov is adjectival is not easy to 
 determine), xxi. 316 sq. (j>r]pX yap ovre t3Cnv xP aL(T f Ji V (T€ f JL€V °^ T€ Tt 
 elbos I ovre ra reu'xca KaXd (ilia arma pulchrd). 
 
 From later Greek three examples or (omitting Theocritus xxvii. 59 
 for more than one reason) two examples of the possessive adjective 
 e/xos are generally cited, upon which we read in Gerth-Kiihner, 
 Gr. Gramm. ii. p. 614 * Die Reispiele fur eine abweichende Stellung 
 des Possessivums sind durch Konjektur beseitigt \ They are Sophocles, 
 Ajax 572 sq. 
 
 Kal Tapta Ttvxrj p.r)T ayotvapxaL rives 
 
 Orio-ovv 'Axcuot? /XTJfl' 6 XufA,€u>y i\t.6s, 
 
 and Euripides, Hippolytus 682 sqq. 
 
 a> TrayKaKLorn Kal <f>C\(ov btacpOopev, 
 
 oV etpyacrco /xe. Zev? <r 6 y€vvr\r<op cjxos 
 
 TTpSppiCoV €KTpC\j/€L€V OVTao~aS TTVpl. 
 
 There is evident likeness between these two passages. They are 
 both spoken under the stress of strong emotion, though it is not held 
 in both under the same control. The curse of Hippolytus, the 
 last solemn injunction 2 of the self-doomed Ajax, clothe themselves 
 
 1 6 £clvos is expelled from thirteen passages, rbv gelvov from eighteen, and 
 tg> lei'vo) and tovs gelvovs from one each. This is the faith that can remove 
 mountains. 
 
 2 (nKTKTjnTco (566) and the prjr . . Orja-ovai, not to be attenuated into a mere 
 dependent on onus, but reminiscent of the use of pfj in solemn utterances of 
 a speaker's desire ; cf. Od. x. 330 (above). 
 
20 PROCEEDINGS OF THE BRITISH ACADEMY 
 
 naturally in antique language. And the article in both is not the 
 slip of a scribe but the choice of the author. 
 
 Let me digress for a moment to call attention once more to the 
 superficiality of the criticism which has been dealt out to another 
 archaizing appeal in this very tragedy, 835 sqq. 
 
 KaAw 8' aptoyovs ras act re itapQevovs 
 
 ad 0' opv&as iravra rav /3porois TrdO-q 
 
 acfjLvas 'Epivvs ravviTobas p.a6dv e/xe* 
 
 /cat <T(f>as kclkovs Kcfotora Kal iravaikiOpovs 
 
 £vvapTTa<T€iav axrircp elaop&a e/ute 840 
 
 avrocT(f)ayjj ttCtttovtcl to>s avTOcrcjxiyeis 
 
 irpbs t&v <£iAiotcov €Ky6v<nv okoiaro. 
 
 It is usual to reject the two or the four last lines as an interpolation. 
 It may be admitted that the connexion in 1. 840 would be im- 
 proved by reading x^o-Trep, as any interpolator would most certainly 
 have seen. But if the two lines 841-2 are forged, the forger was 
 a criminal artist as remarkable and as unfortunate as the murderer of 
 the little old man in the well-known story of Gaboriau. His skill has 
 been his own undoing. * To obtain for the curse of Ajax the utmost 
 solemnity possible", he has made Sophocles ' clothe it in an ancient 
 Ionic and Epic form as is shown by rw?, <£iA.iotg>i/, okoiaro. This effect 
 could hardly have been obtained in any other way. Of $iA.io-roy, 
 to which chief exception has been taken, it may be observed that 
 though not found elsewhere in extant literature it is sufficiently 
 supported by the </>iA.iW of the Odyssey and by its use as a proper 
 name. As regards the ending -aro, we may note that it is not 
 without significance that tragedy confines its use to the optative, and, 
 as my friend Prof. Ridgeway pointed out to me a good many years 
 ago, to the optative of uncontracted verbs V 
 
 To make a conclusion, those who have liberated themselves from 
 the thraldom of grammatical conventions and classifications, and who 
 remember the freedom which other languages, such as those of the 
 Romance and the Teutonic stocks, 2 use in their employment of articles, 
 
 1 From Breal's Semantics, Preface to the English edition, p. lviii, note. 
 
 a Extirpators of the l intermediate ' or transitional article in Greek and 
 restricters of its movements should first attack the We of Latin and Romance. 
 Beginning with the Italian article, with its fluctuations of place where an 
 adjective is appended to a noun, its insertion or omission with proper names and 
 so forth, they may then next consider the post-classical Latin and deal with such 
 examples as 'occidit pater tuus uitulum ilium saginatum', ' ille iudex iustus 
 (6 dUaios KpiTT)s)\ 'reuelahit deus brachium suum Mud sanctum', 'gigantes 
 nominati Mi' (Ronsch, Itala u. Vulgata, pp. 419 sq.), and end up with the writers 
 of the classical literature : Juvenal, Sat. v. 147 sq, ■ quales (sc. boletos) Claudius 
 edit | ante ilium uxoris (tov rrjs ywautos) ' ; Horace, Sat. i. 1. 37 'non usquam 
 
FLAWS IN CLASSICAL RESEARCH 21 
 
 also developed from demonstratives, will regard with equanimity 
 eccentricities characteristic of a period of flux, or inherited therefrom. 
 They will be chary of limiting the possibilities of a stage of language, 
 in which on the one hand a^Spa tcV os *ce OeoZaiv dWxdqrat fxaxapco-o-iy, 
 Od. x. 74, is found, and on the other Toy (eum) Tpia-KaihtKarov \itkvqhta 
 dvfxdv awr\vpa> II. x. 495, and they will take a more lenient, because 
 a more enlightened, view of peculiarities in later Greek, such as the 
 following 1 : — 
 
 Sophocles, 0. T. 572 : 
 
 T&s tfxds 
 ovk av ttot €i7rc AaCov bia<f)dopds. 
 (non dixisset illam meam Laii caedem.) 
 
 Track. 1249 : 
 
 Tolyap 7roi7Jo-o), kovk ctactftro/uai, to <t6v * 
 
 0€oicri btLKvvs (pyov, 
 
 (illud factum dis pro tuo ostendens.) 
 
 775: 
 
 6 6' OvhiV CtfiwS bv(TfJLOpO$ TO CTOK ^.OVTJS 
 
 huip7]\JL lAefe. 
 
 (illud munus tuum solius dixit.) 
 
 Euripides (?) Fragm. (Weil, Papyr.) 1. 32 : 
 
 /xe'xpi ttoctov tV Trjs tv\ti$ 9 
 
 TTCLTCp, (TV A7J\/ret ITClpaV €V TtafXto /3t<i> ; 
 
 (quousque istud fortunae in uita mea facies periculum ?) 
 
 And in Euripides, Hippolytus 471 : 
 
 ak\ y €t tci irXetw xprjerra t&v kclk&v *X ets > 
 
 it seems more respectful to Euripides to suggest that his phrase is 
 modelled on the earlier freer pattern, ■ if thou hast more good on this 
 side (to.) than bad on that (t&v) \ Od. xx. 309 sq. oiba cKacrra | icrdkd re 
 /cat tci x e P eta (f° r which we are offered arra), than that he has 
 flown in the face of contemporary usage by writing tci 7rAeia> when 
 he should have written 7r\eico tcx. 
 
 prorepit et illi* utitur ante | quaesitis sapiens', ib. 115 sq. * instat equis auriga 
 suos uincentibus, ilium \ praeteritum temnens extremos inter euntem ' ; Cicero, 
 de nat. d. ii. 114 e hie Geminis est ille sub ipsis | ante Canem, TLpoKvav Graio 
 qui nomine fertur '. When they have fixed this Latin fleeter, they may return 
 to Greek. 
 
 1 That examples like these should be judged by the Epic and not the Attic 
 standard will not be contested by those who recall such obviously Epic arrange- 
 ments as Philoct. 371 6 5' eln 'O&vaaevs, Ajax 311 koi tov yxv rpro nktitTTov 
 a<f>doyyos XP° V0V - 
 
22 PROCEEDINGS OF THE BRITISH ACADEMY 
 
 To metre I must make some reference ; but it shall be as brief 
 as possible. Vol. 17 of the Classical Review 1 saw a controversy 
 on the Latin Sapphic which I trust nothing that I now say will 
 revive. The point at issue was whether in his Sapphic odes Horace 
 wrote in the measure of Canning's Needy knife-grinder, whither art 
 thou going ?, or in the measure of his avowed model Sappho, or, as 
 proposed in a compromise strongly reminiscent of the would-be 
 wary examinee in a well-known anecdote, ' Sometimes in one and 
 sometimes in the other'. This controversy was but a by-product 
 of the modern pronunciation, which makes havoc of quantity and 
 plumps an overpowering stress-accent on the syllables which by 
 the laws of the metre should be unstressed. 2 So potent is it that, 
 as a schoolmaster told me but a few weeks ago, even when boys 
 have been drilled in the proper reading of the Sapphic by instruction 
 and example, they fall back into * needy knife-grinders ' as soon as 
 they are left to themselves. Another metre in distress is the 
 anapaestic which is prevailingly, at least in England, read with a 
 dactylic rhythm to the stultification of the tragic systems. Those 
 who had the opportunity of comparing Mr. C. PlattV recitation of p S 
 the parabasis of the Birds with the rendering of the anapaestic 
 measures and their false musical setting in the last Greek play per- 
 formed at Cambridge will not need to be told what a difference 
 this makes. The neglect of quantity is deep-seated in our age ; 3 and 
 I doubt if there is any easy remedy when people are found to believe 
 
 1 pp. 252fF.,339fF.,456fF. 
 
 2 No one, that 1 know, has ever contended that the Greek Sapphic ought 
 to be read, even in portions, in the 'needy knife-grinder' fashion. And yet 
 it will be found that where it can be so read, as in the ode to Aphrodite, 6, 10 
 &<ees orpovdoi nep\ yas (xekaivas, 11, 18, 19, 21, also 23 (for making <£i'Xei into 
 fill-eye is a trifle), 25, 27 (eight or nine lines out of twenty-one), it is so read. 
 For in this part of the world Greek words are accented on the syllable which 
 would bear the accent in Latin. Since now, as Mr. E. S. Thompson pointed out, 
 CI. Rev.,1. c. p. 457 a, a double scansion of Greek Sapphics cannot for a moment 
 be entertained, the advocates, German and English, of the needy knife-grinder 
 rhythm have to explain why it is needed to scan Horace and not needed to 
 scan Sappho, except for the reason that they know the Greek accent was not 
 a stress accent, and that they assume that the Latin was. 
 
 3 On this I am now able to quote two sentences from a paper by Prof. 
 J. W. White, ' On the Origin and Form of Aeolic Verse' , to appear in the next 
 Classical Quarterly. e That the quantitative rhythms and metres of Greek 
 poetry should seem complicated to men^whose language is accentual is inevitable, 
 whereas modern metres and rhythms are notoriously simple.' ' The attempt to 
 conform Greek Lyrics to the elementary — and uncertain — rhythms of modern 
 poetry that is merely read or recited implies a fundamental misconception of 
 relations.' 
 
FLAWS IN CLASSICAL RESEARCH 23 
 
 that anapaests may appear in the last foot of an iambic line, dactyls 
 at the end of an hexameter, and spondees and tribrachs in the * pure 
 iambics ' of Catullus and others. 
 
 It is no business of the scientific inquirer after truth to sit in 
 judgement on the tastes and morals of antiquity. Sympathy with 
 his author is of some use to a student, but of none to a savant. 
 Dispassionateness and insight are all that he requires. The admira- 
 tion stirred in us by the greatness and splendour of an ancient 
 monument of genius is prone to pass into a sentiment which 
 dresses the figure of its worship in fictitious and anachronistic 
 excellences, and resents as profanation any fact or hypothesis that 
 would fasten upon the idol deeds, thoughts, or expressions of which 
 the idolater personally disapproves. How strong and prevalent the 
 sentiment is among us it is difficult to say, since its expression is 
 generally confined to protests in unsigned reviews and private ' letters 
 to the editor'. When the evidence opposed to it is overwhelming 
 and admitted, it shuts its eyes or runs away ; though it is up in arms 
 on every fresh occasion. My own experience is that it is very strong 
 indeed, and that there are but few who can be trusted to decide 
 with equanimity certain questions affecting the private life, say, of a 
 Sappho or a Tibullus. To the others my advice, if I might presume 
 to offer it, would be this. If a scholar finds that one of two neces- 
 sarily alternative conclusions is from its character repugnant to his 
 feelings, this is a hint from his personality that he should leave the 
 matter alone. 
 
 Irrelevant judgements of another kind may furnish a transition 
 to the next division of my subject — the blinding effects of modern 
 vanity. Those who have used a well-known French manual of Latin 
 syntax (perhaps the best and certainly one of the best books on the 
 subject) will have noticed, I trust not without offence, how frequently 
 constructions in Latin authors are described as 'incorrect' or 'un- 
 classicaP. These expressions with their implications are perhaps 
 excusable in the school classroom when addressed to a beginner in 
 the language. But what business have they in a scientific work 
 based ' on the principles of Historic Grammar ' ? and who are we 
 that we should accuse Livy and Nepos of solecisms ? 
 
 The contemplation of an a priori probability, real or imagined, 
 that we are right, or less in error than our predecessors, should be 
 excluded from all scientific inquiry. Our indulgence in this intel- 
 lectual vice Prof. W. Ridgeway ! in a passage with whose substance 
 
 1 The Relation of Archeology to Classical Studies, p. 17 (=p. 53 of the Proceedings 
 of the Classical Association of Scotland, 1907-8). 
 
24 PROCEEDINGS OF THE BRITISH ACADEMY 
 
 I am wholly in agreement ascribes to ' a pettifogging spirit of 
 scepticism 1 . But in this matter I am unwilling to pass even an 
 indirect censure upon scepticism. va<p€ kcll fxe^vaa aitLcrTziv' &p6pa 
 ravra t&v <pp€v&v. A proper habit of honest doubt is the prime pre- 
 requisite of fruitful investigation. Let us distrust always and every- 
 thing, but chiefly ourselves : tecum habita : noris — I need not finish 
 the quotation. 1 But the spirit I mean is akin to the (pdovos of which 
 Thucydides (a thinker whose intellect we are told to-day was obsessed 
 by hovering phantoms) writes that an uninstructed hearer believes 
 that a narrative is exaggerated €t tl vtr^p ttjv kavrov (pvaiv olkovoi 
 (ii. 35. 4). Put yv&criv for (pvaiv here, and you have the vainglorious 
 doubter exactly. It is a matter of common knowledge that seemingly 
 irreconcilable statements may be found to be perfectly compatible 
 when new facts have come to light. And a writer who bluntly 
 rejects the express statement of an ancient on a matter where 
 mistake and prejudice are unbelievable, because he cannot harmo- 
 nize this with other statements and his own inferences therefrom, 
 asserts, though he may not be aware of it, the universality of his 
 own knowledge and the infallibility of his own reasoning. 
 
 It is with reluctance that I touch upon Accent once more : but the 
 remarks of a writer in Classical Philology of April last year 2 are too 
 precious to be lost. His object is to show that the Latin accent 
 of the time of Cicero and Varro was a difference of force, and not 
 of pitch. To establish this it was necessary to discredit or extenuate 
 the evidence of contemporary Romans. This had been done before. 
 But seemingly it had not occurred to any one that it would help if 
 the testimony for the musical character of the Greek accent could 
 be similarly disposed of. An ancient scholar and critic of note 
 stood, however, in the way, to wit, Dionysius of Halicarnassus, in 
 a treatise 7rept o-vvdio-tais 6Vo/xdro>i>, composed not later than 7 b.c. 
 His witness (from Chap. 11) I will now give in English, not in 
 a translation by myself (for this might possibly be suspected of bias) 
 but, wherever possible, in the rendering of the late A. J. Ellis, 
 Quantitative Pronunciation of Latin, p. 27 n. 
 
 The art of public speaking is a musical one too ; 3 for it differs 
 from that used in songs and on instruments in quantity, not in 
 quality. For in the latter (public speaking) words have also melody, 
 rhythm, modulation, and propriety. In speaking then also the ear 
 is delighted with the melody, is impelled by the rhythm, and especi- 
 ally longs for propriety. The difference is merely one of degree. 
 
 1 Persius, S. iv. 52. 2 pp. 202-3. 
 
 3 MovatKr) yap tis tjv tool 17 tg>v 7roXiTtKcoi> \6ya>v eVio-Tj^q (Dionysius). 
 
FLAWS IN CLASSICAL RESEARCH 25 
 
 The melody of speech then (biaXeKTov /xe'Aos) is measured by a 
 single musical interval which is as nearly as possible that called 
 a Fifth. It does not rise in pitch beyond three tones and a half nor 
 is it depressed in pitch more than this amount. But every word 
 which constitutes a single unit of speech l is not spoken at the 
 same pitch, but one in an acute pitch, another in a grave pitch, and 
 another in both pitches. Of those words [of one syllable] which have 
 both pitches some have a low pitch imperceptibly blended with the 
 high, and these we call ' circumflexed , But others have both in 
 different places and apart, and keep its proper nature for each. 
 In dissyllables there is nothing interposed between high pitch and 
 low pitch. But in polysyllabic words, of all kinds, there is but 
 one syllable which has the high pitch among many which have the 
 low pitch. On the other hand the music of song and of instru- 
 ments uses a greater number of intervals and not only the Fifth but 
 beginning with the Octave it performs the Fifth and the Fourth, 
 the whole Tone and the Semitone, and, as some think, even the 
 Quarter-tone audibly. But this (vocal and instrumental) music does 
 not hesitate to subordinate words to the air instead of the air to 
 the words. This is especially evident in the airs of Euripides which 
 he has made Electra sing when speaking to the chorus in his Orestes 
 (vv. 140-3) :— 
 
 aiya criya Xcvkov ixv°$ ap(3v\T]s 
 
 TlOcItC fit} KTVTT€IT€ 
 
 anoTrpoficLT €V€i<r' hito'npoOi KoCras. 
 
 Ill these lines the words alya alya \cvkov are set to a single note although 
 each of the three words has both high and low pitches, and the 
 word apfivkrjs has the third syllable of the same pitch as the second, 
 though it is impossible that one word should have two high pitches. 
 In the word rifoire although the first syllable is made lower, the 
 two that follow have both the same high pitch. The circumflex 
 has vanished from Krvirelre ; for the two (last?) syllables are spoken 
 at the same pitch. And a-noirpo^ar does not receive the acute accent 
 belonging to its middle syllable, but the pitch of the third has 
 descended to the fourth syllable. Rhythms are treated in the same 
 manner. For prose neither forces nor interchanges the length of any 
 noun or verb but preserves short and long syllables, as it has received 
 them by nature. Yet rhythmical and musical art change them, 
 shortening and lengthening, till they are often reversed, for they do 
 not rectify the time by the syllables but the syllables by the times.'' 
 
 So far is the writer to whom I have referred from recognizing, 
 with the accomplished English phonetician, 'the great value and 
 
 1 ov (irjv aTTaad y 17 \e£is f] Kaff tv fiopiov \6yov ra-rrofievT] errl rfjs avrrjs \eyerat 
 rdafios, aXX' rj p*v errl ttjs o^cia? T] Se errt rrjs fiapeias i] 8e eV ap<f)otp' roav hi apcfrorepas 
 ras raaeis e)(Ovcr£>v at pev Kara fiiav (rvWafirjv o~vve(f)Bappepov e\ovai ra o^ii to ^apv 
 as df] rre plana fievas KaXovfjiev, ai 8e iv crepe* re /cat erepto X^P'* (Karepov, i(f> eavrov 
 tt]v oiKeiav <fjv\drrop <f>vo~iv. kol rat? pev 8icrv\\d^ois ovdcv to 81a necrov x<opiov 
 (OapvTTjros tc Kai ogvrqros k.t.€. 
 
26 PROCEEDINGS OF THE BRITISH ACADEMY 
 
 importance of this passage, its explicit identification of Greek accent 
 withpitch, and its clear assertion of the strict observance of quantity in 
 prosed that he comments upon it in the following strain : 1 
 
 If Greek was in the third of these phases when Dionysius lived 
 (and Kretschmers and Mayser's examples [of the confusion of long 
 and short vowels in papyri] seem to prove that stress was not 
 sporadic, but regular), the accent would be such as a modern phoneti- 
 cian would describe as predominantly stress. But the accented 
 syllable, being usually uttered at a higher pitch than its neighbors, 
 might conceivably appear to a man learned in the theory of earlier 
 days as distinguished from the unaccented one by the difference in 
 pitch alone. Moreover, we must not forget that Dionysius is not 
 here discussing accent, qua accent, 2 but the /xeAo? of speech, which he 
 contrasts with the jxe'Aos of song and instrumental music. To re- 
 concile his words with the supposition that the accent of his day 
 was a stress (among the educated as with the masses), we are compelled 
 to discredit his statement only so far as to regard him as mistaken in 
 thinking that pitch-elevation was invariably present in the accent. 
 It would seem, therefore, that there is no adequate reason for assuming 
 that the educated Greeks, with whom Cicero and Varro conversed, 
 used an accent materially different from that of the people who wrote 
 the papyri. 
 
 If this is the treatment to which ancient evidence of the authority 
 and the precision of the one before us is to be subjected, better give 
 up the study of antiquity altogether. 
 
 I pass to my second subdivision : the neglect of ancient testimony 
 when it is in conflict with the results or the inferences of modern 
 research. The motive here is the same. Suppose that by our labours 
 in collating, comparing, or deciphering classic MSS. or papyri we 
 have ascertained that a particular reading is supported by the line of 
 tradition, whereas an ancient author bears witness expressly, or by 
 implication, to something different, which testimony is likely to go 
 to the wall ? 
 
 Ask the editors of the Fourth Eclogue of Vergil, 1. 62. Here the 
 only lection which can be shown to have existed in classical times 
 appears in hardly a single modern text? either with or without an 
 obelus, while a corruption, the origin of which is obvious to a tiro 
 in textual criticism, usurps its place. The poem ends thus in MSS. : 
 
 cui non risere parentes 
 nee deus hunc mensa dea nee dignata cubili est. 
 
 1 I put in italics phrases which are characteristically modern. 
 
 2 This means that Dionysius had not hefore him the confusion of stress and 
 pitch inextricably and disastrously imbedded in the modern term accent. 
 
 3 One is Mr. F. A. Hirtzel's in the Oxford series of texts. 
 
UNIVERSITY 
 
 CF 
 
 FLAWS IN CLASSICAL RESEARCH 27 
 
 Quintilian, a better witness than fifty capital and uncial MSS., 
 records qui, and records nothing else. For him the dative did not 
 exist, and if we would give the facts as they are our critical note 
 should run as follows : 
 
 qui Quintilianus, cui nescio quis post Quintilianum et codices 
 qui nunc extant omnes. 
 
 But lest the collators and comparers of Vergilian codices should have 
 laboured in vain, cui must be pushed up first into a seat contiguous to 
 qui, and then to one in front of it. A quotation from a well-known 
 modern commentary will show how this may be done : 
 
 A remarkable various reading of v. 62 is preserved by Quintilian 
 (ix. 3) . . . We must suppose then with Voss that Quint, found 
 ' quoi ' in his copy, and read it ' qui ' rather than ■ cui '. 
 
 We are staggered at the outset by the description of what we should 
 naturally have considered a gross blunder of Quintilian as the ' preserva- 
 tion ' of a reading. But this is a trifle. According to the commentator 
 cui or quoi was the genuine reading, was also that current in the age of 
 Quintilian, and stood, moreover, in the grammarian's own text of Vergil. 
 But so unfortunate was this eminent scholar and teacher ('uagae 
 moderator summe iuuentae"' is what his friend Martial calls him) 
 that every time he read the conclusion of this most notable writing of 
 Vergil, his eyes were closed to the o which stood in his copy between 
 the u and the i, that neither the conversation of his friends nor the 
 instruction of his pupils (so viel gelehrt und so wenig gelernt !), to all 
 of whom the reading which he ignores was perfectly familiar, ever 
 succeeded in opening them, and finally that he selected this halluci- 
 nation of his own to provide from the works of the master an example 
 of a grammatical anomaly. Let the Quintilians of our day, who thus 
 deal with the evidence of a perhaps not less illustrious predecessor, 
 reflect what treatment may be justly theirs at the hands of the 
 Quintilians of posterity. 
 
 As I have discussed this passage elsewhere, Classical Review, xvi. 
 (1902), 36 sq., I content myself with transcribing the words of 
 Quintilian, 'est figura et in numero; uel cum singulari pluralis subiungi- 
 tur : gladio pugnacissima gens Romani ; uel ex diuerso : qui (cui the 
 MSS.) non risere parentes nee deus hunc mensa dea nee dignata cubili est ; 
 ex illis enim qui non risere, hie quem non dignata,"* and noting, as 
 a further example of the hinc which I there proposed to restore for hunc 
 in 1. 63, Hor. S. ii. 1. 79 ' nihil hinc diffingere possum 1 , that is nihil 
 
28 PROCEEDINGS OF THE BRITISH ACADEMY 
 
 horum or ex his : for hinc has no construction either with diffingere 
 or with the variant diffindere. 1 
 
 Here, so far as we know, the corruption which modern texts present 
 did not exist in the classical age. In the next instance it did, and 
 was duly noted by a professional scholar. 
 
 Catullus, xxvii. 3-4, are thus given by Aulus Gellius, N. A, vi. 20. 6 : 
 
 ut lex Postumiae iubet magistrae 
 ebria acina ebriosioris, 
 
 where editors have ebrioso acino or ebriosa acina. 
 A modern German comments as follows : 
 
 Demnach 2 wollte Gellius thorichterweise ebria acina gelesen 
 wissen. 
 
 On the thorichterweise something will be said anon. At present 
 I would draw attention to the misrepresentation of the facts. Gellius 
 did not merely ' wish to read ebria acina ' : he found this reading in 
 a copy or copies which he regarded as trustworthy. ebrioso and 
 ebriosa were extant in other copies which he stigmatized as corrupt. 
 Here are his words, quoted also by the commentator in part : 
 
 Qui ebriosa autem Catullum dixisse putant aut ebrioso (nam id 
 quoque temere scriptum inuenitur) in libros scilicet de corruptis 
 exemplaribus factos inciderunt. 
 
 Of the two readings which Gellius preferred, one, acina to wit, is 
 accepted from him by a number of scholars who reject the rest of his 
 witness. But I will continue to confine myself to the commentator 
 whose treatment of the evidence is at least consistent. His idea of 
 ebrius and ebriosus is that they are two synonyms, and thus that we 
 are free to choose the one which abolishes the hiatus and preserves 
 the credit of the manuscripts. This I gather from the otherwise 
 
 1 Mr. Warde Fowler, who has recently republished his study on the Fourth 
 Eclogue in Virgil's Messianic Eclogue (1907), pp. 49-85, accepts, I am glad to 
 see, Quintilian's qui, though he retains the hunc of the Vulgate. He observes 
 that he cannot follow me in my account of Quintilian's words, ' ex illis . . . dignata.' 
 But unless he means that c ex illis qui etc. hie quem non dignata ' expresses that 
 the word hunc (singular) is ( grammatically ' dependent or ' follows grammati- 
 cally ' upon the plural qui etc., I can see no difference between our two accounts, 
 and if he does, the main question, how Quintilian understood the passage, is left 
 untouched. With regard to hunc v. hinc, I must correct his statement that 
 I ( contend that Quintilian's copy of Vergil was a bad one', which may mislead 
 the unwary. What I said was that ' the text of Vergil, as vouched for by this witness 
 of the end of the first century a. n. (i. e. Quintilian), was corrupt '. This means 
 that the corruption hunc was then part of the current text. 
 
 2 ( Accordingly ' : because Gellius had shown that he was acquainted with 
 passages of Homer where hiatus occurred between repeated vowels. 
 
FLAWS OF CLASSICAL RESEARCH 29 
 
 quite superfluous annotation ■ ebriosus steht bei Cic. und Sen., ebrius 
 in Prosa und Poesie \ 
 
 The ' foolish ' Gellius knew better. He knew, as Seneca says, 
 Ep. lxxxiii. 11, ' plurimum interesse inter ebrium et ebriosum\ that 
 a ' grape ' which is ' mero plena ' 1 may well be said by a metaphor to be 
 ' tipsy ', but hardly a * toper ', and that, although the comparative and 
 the positive of the same adjective are often combined in Catullus, 2 
 this is not done when the combination is absurd. And, if we had 
 him here to ask, he would answer, I fancy, that he understood the 
 poet to mean that Postumia was ' more fond of her liquor ' (ebriosior) 
 than the ' ever-tipsy ' grape (ebria). Whether the wine which intoxi- 
 cates the grape is the juice within the grape-skin, or whether, as 
 we may perhaps suppose, there is an allusion to the practice mentioned 
 by Pliny, N. H. xiv. 17 ■ conduntur et musto uuae ipsaeque uino suo 
 inebriantur 1 is for the present purpose a matter of no importance. 3 
 
 The conception of the grape as, so to say, a tiny skinful of wine 
 recalls to me a much disputed passage in Theocritus, the situation in 
 which I have never seen properly explained. In Id. i. 45 sqq. we 
 have a charming scene, illustrating the insouciance of boyhood, 
 as carved on a fiadv kktctv^lov (27). 
 
 rvrdbv 8' oaaov aTrcoOev akiTpvToio yipovTos 
 
 Trvpvatais o-TCMpvkaicrL kclXov pifipidev dAwa, 
 
 tolv 6\iyos rt? /ccopos e<£' al/xao*iai(n <f)v\a<r(r€i 
 
 rjfievos' afj.(f)l 8e viv ov oWtoircKts h p*v av opyods 
 
 (pom] (TivopLtva tolv rpwftftoy h 8' eirl irrjpq 
 
 irdvra bokov K€v6ovo-a to iraibiov ov irplv airncrtiv 50 
 
 (f>a.TL TTplv Tj OlKpCLTLOTOV CTTl §T]pOl<7l KaOl^TJ. 
 
 avTap oy avOepUoio-i Kakav tt\€K€1 CLKptboO-qpav 
 oyoivv i<f>apfi6o-h(t}v' /ueAerat 8e ol ovt€ tl itripas 
 ovre (f)VTa>v too~q~t]vov oaov irepl 7rAey/xart yaOtl. 
 
 The boy, who has been set to watch the vineyard, neglects his 
 task for the more congenial occupation of plaiting a locust-trap, and, 
 
 1 Cf. Ov. A. A. ii. 316 ( plenaque purpureo subrubet uua mero '. So uinum of 
 the juice in the grape ; Plaut. Trin. 525 e uinum priii' quam coctumst pendet 
 putidum.' 
 
 3 ' Hat C(atullus) solche Wendungen zwar ofter, z. B. 9, 10 beatiorum beatius, 
 22, 14 infaceto infacetior ', etc. 
 
 3 This was written before the last German commentary (G. Friederich, 1908) 
 came into my hands. It is a relief to read there c Man sieht sofort, welches 
 Beiwort der Weinbeere zukommt ; ihrer Natur nach kann sie nur einmal voll 
 sein, aber das Gewohnheitsmassige ist ganz undenkbar. ebriosa acina ist eine 
 contradictio in adiecto.' 
 
30 PROCEEDINGS OF THE BRITISH ACADEMY 
 
 absorbed in this, fails to notice the two foxes, the one ravaging 
 the vines, and the other making an inroad on the grapes which are 
 to save his aK.paTicrp.6s 1 from being a mere frpocpayia. (Athenaeus 
 of a make of bread which could be eaten alone, lorai — €v/3pcoro9 
 irpbs £iipo(j)ayCav, 113 B; €7H ^poiai shows a use of €ttl c. dative 
 of viands which we know from Aristophanes, Ach. 835, Pax 123 and 
 elsewhere). Compare the menu of the vegetarian Valerius Cato as 
 given by Bibaculus ap. Suet, de grammaticis 11 : 
 
 quern tres cauliculi, selibra farris, 
 
 racemi duo tegula sub una 
 
 ad summam prope nutriant senectam. 
 
 In conclusion I give verbatim a recent critical comment on Terence, 
 Phormio 330, as the spirit of much modern editing could hardly be 
 better expressed : 
 
 tennitur is due to Donatus. MSS. tenditur ... It would seem 
 a matter of questionable propriety to set aside the testimony of 
 the MSS. for the opinion of a single grammarian. 
 
 In this discrediting of ancient witnesses two faults of method may 
 be detected. First, the intrinsic character of the rejected testimony 
 is disregarded. Mommsen, with others in his wake, poured scorn on 
 the saying, traditionally attributed to Appius Claudius, that in the 
 articulation of Zeta the teeth of the living were bared like those of 
 the dead. The five varieties of Greek accents which Varro tells us 
 Glaucus of Samos distinguished are dismissed as the refinements 
 of a musician or the figments of a grammarian. But the Appian 
 pronunciation of Z is found to be otherwise attested ; 2 and the five 
 accents of Glaucus correspond surprisingly with accentual varieties 
 recognized by the most recent phonetic science. 3 What is most 
 reprehensible here is not the rejection of the testimony, but the 
 failure of its rejectors to discern the marks of genuineness which it 
 bore upon its face. 
 
 Secondly, exaggerated stress is laid upon unessential inaccuracies 
 and inconsistencies. Who has not noticed the conflicting numbers 
 of killed and injured which the bills of different newspapers put out 
 after some mining or other accident ? If modern and ancient witnesses 
 are to be treated alike, why do we not now exclaim 'look at the 
 numerical discrepancies ! was there any accident at all ? ' 
 
 1 It is clear that aKparioTov must contain or conceal an allusion to the custom 
 of taking an early snack consisting of bread sopped in wine anparov, though the 
 right reading here remains uncertain. 
 
 2 Classical Review, xv (1901), pp. 218 sqq. 
 
 3 Classical Review, xix (1905), pp. 365 sqq. 
 
FLAWS IN CLASSICAL RESEARCH 31 
 
 The true, the fair, and the scientific course is to reject the evidence 
 to the exact extent that it is vitiated by proved inaccuracies. 
 The whole, if the errors are fundamental : otherwise only the part 
 affected. The professional scholars and the grammarians of the present 
 day are ever ready with accusations against the grammarians of 
 Greece and Rome. The Latin grammarian in particular is a favourite 
 mark. It is assumed, wherever convenient, either that parrot- 
 like he has repeated a predecessor, or that he has applied to Latin 
 what is true only of Greek. There is less truth, I think, in those 
 reproaches than is generally assumed, and by champions of the 
 ancients a damaging uos quoque might not unseldom be retorted upon 
 the heedless modern assailants. But the charges should anyhow be 
 limited to the specific issues, and not enlarged to foment a general 
 prejudice. 
 
 The classical grammarians were unacquainted with the part which 
 the vocal chords play in modifying consonantal sound, and consequently 
 to that extent they fail to express correctly the differences which 
 they heard in the speech around them. They did not understand 
 why a breathed r sounded differently from a voiced r. They heard 
 the breath in, say, primp, as contrasted with etp-qrai, but they could 
 not analyse it. And Varro's discussion of the questiou whether one 
 should write r or hr or rh is not ■ grounded on grammatical theories V 
 but is a humble groping after the truth. An exact parallel is the 
 double writing of the English breathed w by hw and wh. When this 
 division of 4 voiced ' and ' breathed ' was crossed by the further differ- 
 ence of strong and weak consonantal articulation, or of fortes and 
 lenes (as Sievers, their discoverer, called the varieties), their per- 
 plexity was increased. But here, too, it is not difficult to interpret. 
 Dionysius tells us that the only difference between kttt, ybfi, and y6($> 
 is that k, etc., are pronounced x/aAwy.x, e ^ c -» da-aem, y, etc., pcTpim kcu 
 p.eTa£v afx(f)oiv. Put into modern terms, this means that kttt were 
 breathed lenes, \0(f) breathed fortes, 2 and y5/3 voiced fortes. Now 
 breathed lenes and voiced fortes appear to be a somewhat unusual 
 combination, for the good reason that the approximation of the tense 
 vocal chords which is necessary for the production of ' voice ' tends 
 naturally to moderate the force with which the air is expelled from 
 the lungs. Yet the grammarian's account is confirmed by two circum- 
 stances which have not received a due attention. The first is the 
 not unfrequent, and at first sight astonishing, representation of Greek 
 
 1 As Blass, Gk. Pron. (p. 90, n. 1), says. 
 
 2 Every one knows that, if a breathed sound is strongly articulated before 
 a vowel, a breath or h creeps in. 
 
32 PROCEEDINGS OF THE BRITISH ACADEMY 
 
 breathed lenes k, t, and p (and especially k l ) by the Latin voiced lenes 
 g, d 9 b respectively, instead of by the Latin breathed fortes. The second 
 is the otherwise remarkable phenomenon that the mediae y(3b, like the 
 aspirates x^^> but unlike the tenues kttt, passed to fricatives (open 
 consonants) in later Greek. 
 
 Of the harm done in the province of history and archaeology 
 Professor Ridgeway, in the address already referred to, has given some 
 noteworthy illustrations. I will add two that have come under my 
 own notice recently. An attempt has been made to apply the 
 fashionable method of historical probability to the names of the 
 generals in the traditional account of the Samnite Wars. These 
 are alleged to be fictitious on the ground of their ' suspicious agree- 
 ment ' with the names of commanders in the Social War. The 
 theory is refuted by Professor Gaetano di Sanctis in the Rivista di 
 Filohgia for July, 1908, pp. 353 sqq. Again, it has been customary to 
 call Plutarch's words <£>riixn$ kcu K\nS6vo$ Upov (Camillus 80, and de 
 Fortuna Rom. 5) a blundering appellation of the shrine of Aius 
 Locutius, to which it unquestionably refers. But it would appear from 
 the indications furnished by a recently collated MS. of Juvenal (i. 115) 
 that in the time of Plutarch this shrine was popularly associated with 
 Fama. 2 
 
 If the statements of ancient witnesses are set aside, need we 
 wonder that they are also read with inattention ? To modern ears 
 there is doubtless a certain unpleasantness in the occurrence of the 
 same word or of the same syllable close together. Many ears are 
 offended by Vergil's ' Dorica castra ' A en. ii. 27 and ' Achaica castra ' 
 ib. 462, and the time and trouble of even competent scholars has been 
 expended in collecting these and similar passages with a view to 
 a supposed canon of Quintilian (ix. 4. 4) thought to enounce the same 
 aesthetic principle. 3 The carelessness of this proceeding is superb. 
 Quintilian says that the final syllablES of a preceding word should not 
 be the same as the initial oneS of the following : ' uidendum est 
 ne syllabae uerbi prioris ultimae sint primae sequentis'; and his 
 examples are ' inuisae uisae ' and the famous ' O fortunatam natam me 
 consule Romam.' 4 Servius, it is true, reprehends A en. ii. 27 for « mala 
 
 1 This is just what we should expect. For the earlier the consonantal check 
 is applied to the stream of outrushing air, the more noticeable is the difference 
 between a ferns and a fortis. 
 
 2 See now Classical Quarterly, iii. (1909), pp. 66 sqq. 
 
 3 American Journal of Philology, xxiv. 451. 
 
 4 Professor Mayor, on Juvenal x. 122, understands Quintilian correctly, of 
 course ; but most of the examples which he produces of the objectionable repetitions 
 are not strictly in point. The sounds of e. g. ' moles mSlestiarum/ De Or. i, § 2, 
 
FLAWS IN CLASSICAL RESEARCH 33 
 
 compositio \ but Servius may have understood the remark of 
 Quintilian no better than some moderns, or, again like them, may 
 have applied his own principles of euphony to the Latin of the 
 past. 
 
 Studies in which are so many pitfalls as in ours must allow no 
 openings to error. In the difficult task of estimating and realizing 
 antiquity there is no aid with which we can dispense. My scientific 
 friends have sometimes remarked to me on the classical man's in- 
 attention to details. The minute care and circumspection which 
 they expect from work in their own department they allege, and 
 I fear with justice, is too often absent from classical investigations. 
 The aesthetic and literary exponents of classics are to blame for 
 much of this. Because they want broad effects, the picture as a whole, 
 so they say, they stigmatize as pedantic the tracing of fine distinctions 
 and the pursuit of small details. This view is a false one. For the 
 picture is injured if its parts are blurred, and it is no pedantry to 
 wish to know. 
 
 Sometimes this indifference to consequences produces only practical 
 inconvenience. Index-makers, and the writers of specialistic treatises, 
 are entitled to use any abbreviations that will lighten their labours 
 and save their space. But that does not justify the writer of books 
 or works intended for the general classical reader in lettering the 
 books of Homer or numbering the speeches of Demosthenes. What 
 the eighteenth Iliad and the De Corona oration are about is known 
 to everybody. But to how many are they not disguised when they 
 are cited as 2 and Or. 18 ? 
 
 These abbreviations are not merely a nuisance to the general 
 reader ; but they may produce error of a kind which is very difficult 
 to track. A writer's or a printer's mistake in a single sign may 
 make an important reference entirely useless. Monro, whose 
 avoidance of the citation by letter is what we expect from his usual 
 good sense, has himself fallen a victim to the practice. In his 
 Homeric Grammar 2 , § 270*, he writes of clauses in Indirect Discourse, 
 after verbs of saying. i Of these, again, only three are in the Iliad 
 (16. 131, 17. 654, 22. 439).' But Iliad 16. 131 is no example 
 though Odyssey 16. 131 is. The mistake has come from some 
 confusion of -n and II. The third example from the Iliad may be 
 1. 109. 
 
 Another practice of thoughtlessness, or (should we rather say with 
 Mr. Housman, 1. c, below, ?) of vanity, is the wanton alteration of the 
 
 are not identical ; and intentional jingles like Ter. Eun. 236 f pannis annisque ' 
 must also be excluded from the count. 
 
 A3 
 
34 PROCEEDINGS OF THE BRITISH ACADEMY 
 
 signs by which classical MSS. have been denoted by their dis- 
 coverers. This tampering with the record is a very common offence, 
 and one deserving of the pillory upon each occasion. 1 I have only 
 noticed one instance in which there was a gain in such a change 
 of symbols for MSS. This was Brieger's substitution of O ( = Ob- 
 longus) and Q (= Quadratus) for the A and B which Munro had 
 used to denote the two chief MSS. of Lucretius. The gain was 
 slight, and provides no excuse for altering the O which symbolizes 
 a codex of Silius to Q because the name of our college and our 
 manuscript also begins with a Q. 2 
 
 I pass to cases where something more is involved than mere incon- 
 venience to ourselves. Who was the first to fly into the face of 
 all contemporary ancient testimony and misname the battle of 
 Palaepharsalus or Pharsalia, I do not know. But from Drumann 
 onwards the flaring torch of ignorance and unreason has been passed 
 along the line of our historians. 3 This battle had no connexion with 
 the town of Pharsalus, with which we have been forced to associate 
 it. It was fought in the open country in the district of Pharsalia 
 near a ruined or insignificant hamlet, the site of which has yet to be 
 discovered. 4 I am glad to find that some one has at last been found 
 to break with the modern fashion and to call the battle-field by 
 a name which belongs to it. 5 But I regret somewhat that Dr. Rice 
 Holmes's choice did not fall on Pharsalia rather than upon Old Phar- 
 salus. For with habit and sloth arrayed against it Old Pharsalus or 
 Palaepharsalus has small chance of ousting Pharsalus. , 6 
 
 We have seen that it is one of the chief failings of modern research 
 to neglect evidence, however explicit, which has not come down to us 
 
 1 For examples see Classical Review, xiii. 59 a : ' X Y (of Caesar's Bellum Ciuile, 
 here re-christened D and Z, with the disregard of convenience general among 
 foreign scholars)/ ib. xx. p. 349 b, The Apparatus Criticus of the Culex, by 
 A. E. Housman, in The Transactions of the Cambridge Philological Society, vol. vi, 
 p. 13. 
 
 2 Classical Review, xiii. 127 a, note. [I am glad now to be able to quote 
 a well considered utterance by Dr. Kenyon in ' The Numeration of New 
 Testament Manuscripts ', Church Quarterly Review, April 1909, p. 86. ' The 
 symbols xABCD and many more have acquired a definite connotation which 
 pervades the work of textual critics since textual criticism rose to importance. 
 It is no light thing for a scholar to claim the right to abolish all of these and to 
 make the writings of his predecessors unintelligible to coming generations.'] 
 
 8 Signor Ferrero is an exception. 
 
 4 See Classical Review, xix. pp. 257 sqq. 5 Classical Quarterly, Oct., 1908. 
 
 6 Things have reached such a pass that Messrs. Tyrrell and Purser are 
 rebuked for their resistance to the fashion in the following terms : ' Tastes 
 cannot be allowed to differ about "the battle of Pharsalia" which occurs 
 passim', Classical Review , ix.p. 44 a. 
 
FLAWS IN CLASSICAL RESEARCH 35 
 
 by the direct line of transmission. A striking example of this is to 
 be found in what some may consider the unimportant province of 
 Latin spelling. 
 
 If we were asked why we have surrendered such spellings of a 
 former generation as foemina, sylva, lachryma, bacca, the answer 
 would be, I suppose : to put in their places spellings which are either 
 known to have been those of classical usage, or, at least, are not 
 known to conflict with that usage. In other words our aim is to 
 restore the contemporary spellings in so far as this can be done with 
 certainty or at least a fair approach thereto. This was the only in- 
 telligible reason for the reforms in Latin orthography which we 
 associate especially with the name of Lachmann. Our goal (whether 
 we attain it or not) is, I repeat, the contemporary spelling, and this is 
 the sole justification for the change. To take a parallel from English, 
 there may be some excuse for printing Chaucer in the spelling of the 
 twentieth century : there is none for printing it in that of the 
 sixteenth. 
 
 Now there is no fact in the history of the Latin alphabet better 
 established or more universally admitted than that the pre-Ciceronian 
 orthography differed from the later, or let us say the Augustan one, in 
 important details. Y and Z were no part of the alphabet, V being 
 commonly employed for the first, and S or ^S* for the second ; H was 
 not employed after a consonant, consequently C, P, T, R appear as 
 representatives of the Greek aspirated mutes and the breathed P. 1 
 This was beyond all question the spelling of Plautus and Ennius. 
 
 What then is the practice of editors of the older Latin authors as 
 regards these clear and definite points ? Little better than a tissue of 
 inconsistencies. In Lucian Mueller's edition of the Fragments of 
 Ennius's Annales, fragment IV d (a) of Book VI, 1. 180 is ' numini 
 Pyrrus, uti memorant, a stirpe suprema ' ; the next fragment V (&)* is 
 a quotation from Cic. Or. 160 * Burrum semper Ennius, numquam 
 Pyrrhum — ipsius antiqui declarant libri \ The editors give r, not rh, 
 because it happens to be in the MSS. of Nonius, but there is no reason 
 
 1 It might be thought that we ought to add to these differences the non- 
 gemination of doubled consonants in writing. But for the present T exclude 
 their consideration on two grounds. In the first place the exact date of 
 the introduction into literary writing of the doubling, assigned with probability 
 to Ennius, is uncertain. In such matters inscriptions lag behind the custom 
 of the people, and we cannot be sure that Plautus did not adopt the improve- 
 ment in his later plays. And in the second place, so far as I know, the change 
 was purely graphical, or, in other words, the Roman syllabification in the case 
 of doubled consonantal sounds was the same before and after the innovation 
 in spelling. If the contrary could be shown, the matter would wear an entirely 
 different aspect. 
 
36 PROCEEDINGS OF THE BRITISH ACADEMY 
 
 for believing that any one ever wrote Pyrrus in classical times; 
 Burrus was the ancient form, Pyrrhus the modern ; Pyrrus is a 
 bastard of the copyists. So in a standard text of Plautus we find 
 zona appearing at True, 954, 955, Merc. 925, but sonam at Poen. 
 1008. In another edition we find sonarius at Aul. 516 but zonarius 
 at True. 862, and, to take a commoner word, we are presented with 
 Syri at Bacch. 649, Surus at Pseud. 636 sq., Syrum at Stick. 433, 
 Suras at Trwc. 541, *9yra at J/erc. 415. What is the explanation of 
 such vagaries? It is that the aim of the editors is not to edit 
 Ennius or Plautus but to edit the tradition of Ennius and Plautus ; and 
 that consequently their eyes are not fixed upon the evidence for the 
 text but on the evidence for part of that evidence, and Cicero, like 
 other ancients, counts for nothing unless the mediaeval copyists agree. 1 
 
 But it may be said, correctness, incorrectness, partial correctness — 
 after all, what difference do they make in a matter of this kind? 
 Why are we to be troubled with this pedantic trifling ? I propose 
 then to show in some detail that they do make a difference. 
 
 I begin with an illustration which lies outside the region in dispute. 
 In Poen. 728-9 Plautus is jesting on the ambiguity of pultem, the 
 subjunctive of pulto, and pultem, the accusative of puis. 
 
 Agorastocles. quid si recenti re aedis pultem ? Advocati. 
 
 censeo. 
 Ag. si pultem non recludat? Adv. panem frangito. 
 
 Write pultem in the Augustan form pulsem 9 and the passage is meaning- 
 less. 
 
 Even withpultem I fear that many find it unmeaning, and this obliges 
 me to do something for it by way of correction, interpretation, and 
 even defence. Syntax requires the change of 'recludet' to 're- 
 cludat *, which I have given in the text. Geppert proposed ' et non 
 recludet ? ' But this, though making sense, abolishes the pun, 
 which demands that there should be asyndeton between pultem (under- 
 stood as verb) and recludat. The verse has been rejected ; but there 
 is none more genuine in the whole of Plautus. In its opposition of 
 puis and panis we see the national dish of the ancient Romans 
 
 1 It is not my design to set out these agreements, though I do not think 
 them unimportant. But as I have mentioned sona, I will indicate in which 
 of the passages cited it has MS. support. They are Poen. 1008 sonam A, 
 onam B } erasure before the o. Merc. 925 sonam codd. True. 955 sona is 
 indicated in the corruptions of the MSS. where Ussing restores. f nunc meos nego 
 (non ego) ', ' non cum zona ego '. Aul. 516 semul sonarii (Leo), semisonarii 
 codd. — A larger number than we should have expected to escape through 
 the copying and correcting of so many centuries. 
 
FLAWS IN CLASSICAL RESEARCH 37 
 
 contrasted with the Greek staple of diet which was ultimately to 
 supplant it. See Val. Maximus ii. 5. 5 'erant adeo continentiae 
 adtenti ut frequentior apud eos pultis usus quam panis esset \ Pliny, 
 N. H. xviii. 83 i pulte autem, won pane uixisse longo tempore Romanos 
 manifestum'. Cf. Aus. Technopaegn. 618 and Juv. xiv. 171 with 
 Mayor's note. 1 The native domestic porridge had no chance against 
 the professional bakery and confectionery of Greece where puis was 
 a dish unknown (Pliny, lib. cit. 107). Compare the complaint of 
 Persius about ' these foreign fashions ', ' et Bestius urguet | doctores 
 Gratos : " ita fit postquam sapere urbi | cum pipere et palmis 
 uenit nostrum hoc maris expers, | faenisecae crasso uitiarunt un- 
 guine pultes ".' 2 This contrasted pair Plautus here twists into an 
 implement for suggesting that if the door is not opened, it is to. be 
 broken down. I despair of reproducing his artifice ; but I offer as 
 an approximating paraphrase the following. Agorastocles. * What 
 if he won't let me walk in-to the groats V Witnesses. 'Then break 
 into the roll ! ' 
 
 The same word puis may introduce our first example of misspelt 
 Plautine borrowings. This one happens to be the hybrid compound 
 in Mostellaria 828. 
 
 Tranio. non enim haec pultipagus opifex opera fecit barbarus. 
 uiden' coagmenta in foribus ? Theopr. uideo. Tran. specta 
 
 quam arte dormiunt. 
 Theopr. dormiunt? Tran. illud quidem, ut coniuent, uolui 
 
 dicere. 
 
 Here the MSS. give pultifagus, and the editors pultiphagus, 
 neither the spelling of Plautus, any more than pultophagonides in 
 Poen. Prol. 54. The primal sense is : ' This is not the crude work of 
 our porridge-eating natives, but that of artists of Greece.' But there 
 is a subsidiary jest on pultare and pag-, which is to recall pango, 
 pactum, compages, antepagmentum, etc., and summon up the image of 
 the ' hammer and nail it ' artisans whom we all of us know so well. 
 
 May I add here that Pagus, as the MSS. give it, and not Phagw, 
 is the true name of a lost comedy of Plautus ? 
 
 On Bacchides 362 :— 
 
 credo hercle adueniens nomen mutabit mihi 
 facietque extemplo Crucisalum me ex Chrysalo. 
 
 Mr. Lindsay, printing as above, says very rightly ■ pronuntiandum ex 
 
 1 Had Rome been a pauperized mob of sightseers in those days, its cry would 
 have been for pultem et Circenses. 2 S. vi. 37 sqq. 
 
38 PROCEEDINGS OF THE BRITISH ACADEMY 
 
 Crusalo \ The note is necessary with the vulgate, superfluous with 
 the Plautine spelling. 
 
 One of the most characteristic features of Plautus's style is his 
 love of alliteration. This is often lost if the Augustan spellings 
 are adopted. In the same comedy the same name Crusalus is used 
 in this alliterative play in which c, r, and u take part. 1 
 
 683. hunc suspicabar propter crimen, Ousale, 
 
 mi male consuluisse. 
 
 687. istoc dicto f dedissef hodie in cmciatum Crwsalum ; 
 
 nam ubi me aspiciet, ad carnificem rapiet continuo senex. 
 
 691. nunc hoc tibi cwrandumst, Crwsale. 
 
 - 922. numquam edepol quicquam temere credam Crusale. 
 
 and after the interval of a line 
 
 aeguomst tabellis consignatis credere. 
 
 In 129 Pistoclerus trifles thus with his old tutor's name Ludus— 
 Lydus : — 
 
 non omnis aetas, Lude, ludo conuenit ; 
 
 and in 138 the same jest or jingle is repeated : — 
 
 P. tace atque sequere, Lude, me. 
 
 L. illuc sis, uide ; 
 non paedagogum iam me, sed ludum uocat. 
 
 (With the writing Lydum the jest, such as it is, is incomprehensible.) 
 In 416 there is alliteration with lubido, and I have a shrewd 
 suspicion that when Plautus wrote in 467 
 
 quid sodalem meum castigas, Lude, discipulum tuom ? 
 
 he was thinking, as in 138, of ludus in the sense of * school'. 
 Another name, Archidemides, is turned to account in 284 : — 
 
 cum mi ipsum nomen eiius Arcidemides 
 clamaret dempturum esse si quid crederem. 
 
 The play upon demo every one sees, but that upon area, which is 
 pointed to in si quid crederem, is missed. For area in this connexion 
 see line 943 of the drama ' hie equos non in arcem uerum in arcam 
 faciet impetum \ 
 
 1 In our pronunciation, it is true, one of these alliterations is not lost. But 
 that is because we mispronounce the Latin ch. The Romans of the classical 
 period did not reform their transliteration of the Greek X simply to have the 
 satisfaction of writing a silent h. 
 
FLAWS IN CLASSICAL RESEARCH 39 
 
 In Captiui 274 it is the name of Plautus's type of a philosopher 
 that is jested on. To talk of buying 7%ales for a talent is pointless, 
 if any verbal play was intended, but Plautus wrote 
 eugepae ! Talem talento non emam Milesium, 
 
 and this the MSS. attest, though no editor gives it, and some ignore 
 the pun. 
 
 The representation of by t will excuse or explain the frequency 
 of At(Ji)enae Atticae, otherwise as useless a piece of verbiage as 
 1 London in England ' would be. The difference in sound between 
 ad and at was very slight. Hence in Epidicus 20 sqq., the noticeable 
 repetition of compounds, aduentu aJportas (atportas A) 21, at- 
 tulisti 23, may be mocking echoes of atfletice in 20. 
 
 In Men. 294-5 Culindrus must be written with Heinsius, as 295 
 with its punning reference to culleus (coriaceus) shows : — 
 sei tu Culindrus seu Coriendru's, perieris. 
 
 In 854 of the same play, 
 
 barbatum tremulum Titanum qui cluet Cucino (Ritschl) patre, 
 
 Menaechmus's pretended madness may have disorganized his mythology, 
 and so perhaps we should keep the MS. as Lindsay does. But the 
 change to Titonum, which the editors turn into Tithonum, is a very 
 slight one. 
 
 In the obscure passage Poen. 689 sqq. Mr. Lindsay's later 
 suggestion (Class. Rev., x. 333) seems to be the best yet proposed. 
 He supposes a play on nvaxos and musca. Lycus (the pandar), ad- 
 dressing his supposed victim Collybiscus, says 
 
 ita illi dixerunt quei hinc a me abierunt modo, 
 te quaeritare a muscis, 
 
 who replies, 
 
 minime gentium. 
 
 And to Lycus's ' quid ita ?' rejoins 
 
 quia a muscis si mi hospitium quaererem, 
 adueniens irem in carcerem recta uia. 
 
 With muscis (= /xwrxois) the pun is perfect. It is only injured by 
 changing the MS. reading to muschis. 
 
 I have given the pandar's name as Lycus, with the editors ; but 
 that Plautus wrote and pronounced it Lncus is shown by several 
 indications. 
 
 157. sed lenone istoc Luco 
 
 illius domino non fotumst Zwtulentius. 
 
 187. ita decipiemus fouea lenonem Lucum. 
 
40 PROCEEDINGS OF THE BRITISH ACADEMY 
 
 The play on lupus is manifest and occurs again in 648 : — 
 
 canes compellant in plagas lepide Zwpum, 
 where the Palatine MSS. have Lycum. Lindsay suggests Xvkov. But 
 it makes no difference to the jest whether the name is written as 
 Greek or transliterated and given a Latin termination, Lucum, and 
 this latter accounts for the variants better. 
 In Pseud. 99 sqq., where the vulgate text is 
 
 ut litterarum ego harum sermonem audio 
 
 nisi tu illi drachumis fleueris argenteis, 
 
 quod tu istis lacrumis te probare postulas, 
 
 non pluris refert quam si imbrim in cribrum geras, 
 
 more than one scholar has recognized that there is a play on dracuma 
 (hpaxnr\) and dacruma (lacruma). 1 This play is obscured by the 
 spelling drachuma. 
 
 At 228 Ballio, threatening Phoenicium, says 
 
 eras Pomicium poeniceo corio inuises pergulam. 
 Does it need argument or the citation of parallels like Poenus to show 
 that Plautus meant the beginning of the two words, which he has 
 pointedly contrasted, to be identical ? The scribe of Plautus saw this, 
 spelling them both with ph. The editors give p in one, and ph in 
 the other. 
 
 It is impossible to say how much of the Plautine dialogue appears 
 flat and tasteless because we have missed the double entente which he had 
 in view. 
 
 Pseudolus 636 and following seems a case in point. Harpax the 
 soldier's servant, is questioning Pseudolus, who pretends to be a slave 
 of Ballio to whom Harpax has been sent, with a sum of money, by 
 his master. 
 
 The dialogue proceeds. — 
 
 Ha. sed quid tibi est nomen ? Ps. {aside) seruos est huic 
 
 lenoni Surus, 
 eum esse me dicam. (to Harpax) Surus sum. Ha. Surus ? 
 
 Ps. id est nomen mihi. 
 Ha. uerba multa facimus. eru 1 si tuo domi est quin prouocas 
 ut id agam quod missus hue sum ? Ps. quidquid est nomen tibi, 
 si intus esset, euocarem. 
 
 1 The form dacruma is vouched for by the obvious alliteration in Ennius's 
 well-known epitaph, ' nemo medacrumis decoret neque/unera/letu |/axit. cur? 
 wolito wiuo' per ora wirum.' Professor J. S. Reid has doubted this on the ground 
 that if Ennius had thus written the fact would certainly have been recorded. 
 But it is a question of pronunciation rather than of writing. The sound was 
 neither an ordinary d nor an ordinary /. And if I am not mistaken there 
 was exactly the same fluctuation in odos {odor) olos (olor) in the time of Plautus ; 
 see Pseud. 841 sqq. , upon which I have commented in the forthcoming Brugmann 
 memorial volume of the Idg. Forschungen. 
 
FLAWS IN CLASSICAL RESEARCH 41 
 
 Most readers, I think, will consider this poor fun ; and, if asked 
 to analyse their impression, would probably reply that Harpax's 
 repetition of Surus is pointless and that there is no special force in 
 1 quidquid est nomen tibi ', though it is of course true that Pseudolus 
 is not at present supposed to know his interlocutor's name, which 
 is first given in 653. 1 The clue to the mystery is, I believe, to be 
 found in 1218, where Pseudolus's appearance is described. 
 
 mihi quoque iamdudum ille Surus 2 cor perfrigefacit, (1215) 
 sumbolum 2 qui ab hoc accepit. mira sunt ni Pseudolust, 
 eho tu, qua facie fuit dudum quoi dedisti sumbolum ? 
 Ha. rufus quidam, uentriosus, crassis suris, subniger, etc. 
 
 That is when Pseudolus says ' Surus sum ' Harpax glances at the 
 thick calves of Pseudolus and inquires 'Surua?* 9 Pseudolus, who 
 cannot retort by extracting a jest out of his opponent's name, shows 
 his petulance by calling him Mr. 'No name', sura occurs in yet 
 another passage where the commentators have missed a joke (1173sqq.). 
 Ballio and Simo are chaffing Harpax : — 
 
 Ba. quotumo die 
 
 ex Sicyone hue peruenisti ? Ha. altero ad meridie. 
 
 Ba. strenue mehercle iisti. Sim. quam uelis, pernio; homost: 
 
 ubi sursun aspicias, scias posse eum gerere crassas compedis. 
 
 The play on pernix, 6 swift,' and pernae, ■ hams,' leads up to ' calves \ 
 The same word is utilized in Captiui 850 ' pernulam atque optal- 
 
 miam 1 , as it should be written, the play being on ob and talus. 
 
 Mr. H. W. Prescott, Classical Philology, January, 1909, pp. 4, 5, 
 
 rightly defends the text : but the intrusive h of the convention has 
 
 blinded his eyes to its purport. 
 
 A little further on in the Pseudolus is a verse which obviously 
 
 gains in alliterative force if written and recited thus : — 
 
 ego deuortor extra portam hue in tabernam tertiam 
 
 apud anum illam doliarem 4 claudam crassam Crusidem (659), 
 
 1 I observe here in passing that if harpax is printed as Greek in 654 (Leo and 
 Lindsay), it should have its accent on the last apnai-, like other adverbs in -a£. 
 
 2 So the editors, doing justice to the alliteration ; but one hundred lines 
 earlier, where exactly the same collocation occurs, they give ' si ueniret 
 Syrus | quoi dedi swmbolum'. 
 
 3 The difference of quantity is not important in Plautus's puns. Cf. Pseud. 
 791 ■ /wrinum est forum ' ; True. 773 ' cura cor meum mouit '; Merc. 643 e mails 
 mihi dedit magnum malum' ; Bud. 12, 25 ' Hercules istum infelicet cum sua lie- 
 entia '. 
 
 4 doliarem is here usually taken as ' pot-bellied ', c alter Bottich ' (Georges). 
 But it may be doubted whether this is the sense, or, at any rate, the sole sense 
 intended. (Donatus's note on mffarcinatam, Ter. Andr. 770, is clearly 
 
42 PROCEEDINGS OF THE BRITISH ACADEMY 
 
 where the most negligent observer can hardly miss the ringing of the 
 changes on d, r, t, and c. 
 
 Another character in this play, C(h)arinus, has had his name 
 played on in Greek. 
 
 712. Cal. Carinus. Ps. euge ! iam x<*P iV tovtco ttolov. 
 It is submitted to the same treatment in Latin. 
 
 736. di immortales, non Carinus mihi hie quidem sed Copiast ! 
 i. e. with our friend there is no question of the ■ triste nomen carendi *, 
 Cic. Tusc. i. 87. 
 
 The last place I shall cite is Rudens 494 sqq., where the ship- 
 wrecked pair Labrax and Charmides are indulging in recriminations, 
 each accusing the other of being the Jonah of their voyage. The 
 burthen of the bandied complaints is 'You did it, You\* 
 
 494. La. utinam te priu** quam oculis uidissem meis 
 
 malo cruciatu in Sicilia perbiteres, 
 
 quem propter hoc mihi optigit misero mali. 
 
 Ch. utinam quom in aedis me ad te adduxisti (domum), 
 
 in carcere illo potius cubuissem die. 
 
 deosque immortalis quaeso, dum uiuas, uti 
 500. omnis tui similis hospitis habeas tibi. 
 
 La. Malam Fortunam in aedis te adduxi meas. 
 
 quid mihi scelesto tibi erat auscultatio ? 
 
 quidue hinc abitio ? quidue in nauem inscensio ? 
 
 ubi perdidi etiam plus boni quam mihi fuit. 
 505. Ch. pol minime miror nauis si fractast tibi, 
 
 scelu 1 te et sceleste parta quae uexit bona. 
 
 La. pessum dedisti me blandimentis tuis. 
 
 Ch. scelestiorem cenam cenaui tuam 
 
 quam quae TWstae quondam aut positast Tereo. 
 
 A parallel to this ' tutoyant ' passage is Ennius, Annals, 108 
 * o Tite, tute, Tati, tibi tanta, tfwranne, fa/listi. 1 
 
 The use of alliteration as an irritant which appears often to pass 
 without observation 2 may be exemplified from another Plautine scene, 
 now emasculated by the introduction of the more familiar spellings. 
 
 negligible.) For the formation doliaris, like molaris, ollaris, etc., would more 
 properly mean ( belonging to ' the dolium, and such is its use in the only other 
 phrase, doliare uinum, for which it is attested. Then it would refer to the old 
 lady's affection for the cellar. 
 
 1 The ■ owlish ' iteration tu tu, as the parasite of the Menaechmi calls it in the 
 stormy altercation, 646-54. 
 
 2 For example, I do not find it used to account for the somewhat odd expression 
 in Hor. S. ii. 6. 30 sq. ( <( tu pulses omne quod obstat | ad Jfaecenatem mewori si 
 rwente recurras ?" hoc iuuat et melli est : non rwentiar ' (Horace echoes the angry 
 m's of the last sentence), nor noted at Prop. iii. 12. 1 ( Postume, jolorantem 
 potuisti linquere Gallam ? ' The p's are taken up again in 3, 5, and 6). 
 
FLAWS IN CLASSICAL RESEARCH 43 
 
 The earliest forms of coquo and coqnm which we can infer for Latin 
 are, for the verb, quequo, and, for the noun, quoquos. In the time of 
 Plautus quequo had become quoquo. How much longer the verb 
 remained quoquo, does not concern us here. But the noun quoquos was 
 current, at least in popular speech, in the time of Cicero, as we know 
 from the jest of the orator * on quoque vocative and quoque conjunction, 
 preserved by Quintilian (vi. 3. 47). The pronunciation of this quo, 
 when initial, was undoubtedly (kwo), though in the second syllable it 
 may have been weaker, as Lindsay, 1. c, suggests. 
 
 This is shown by the jingle in four lines of the scene, which I will 
 now quote as I believe Plautus wrote them ; Pseud. 851 sqq. 
 
 Cook, an tu inuenire postulas quemqu&m quoquom 2 
 nisi miluinis aut a^wilinis ungulis? 
 Ba. an tu quoqumatum 3 te ire quoqu&m postulas 
 ^win ibi constrictis ungulis cenem quoquas ? 4 
 
 At the end of the scene we have another ebullition, 889 sqq. : 
 Ba. molestus ne sis ; nimium iam tinnis : tace. 
 em illic ego habito ; intro abi et cenam quoque. 5 
 propera. Boy. quin tu is accubitum et conuiuas cedo, 
 corrumpitur iam cena. Ba. em, subolem sis, uide. 
 iam hie quoque scelestus est quoqui 6 sublingulo. 
 May I digress for a moment to observe that there must be many 
 jests in Plautus which seem pointless to us, solely because their point 
 has still to be recovered ? In a paper, read before the Cambridge 
 Philological Society on March 16, 1905, and briefly mentioned in the 
 Proceedings of the same year, I made some suggestions for restoring 
 their force to certain expressions in a scene of the Amphitryo, Act I, 
 Sc. i, to one of which, as the paper has never been published, I may 
 here refer. Mercury, the false Sosia, says that he has, ere now, sent 
 four men to sleep without a night dress, that is, has stunned and stripped 
 them. The true Sosia overhearing this says, 1. 152 (304 of the play), 
 
 formido male 
 ne ego hie nomen meum commutem et qvintvs 7 fiam e Sosia. 
 
 1 Lindsay. Latin Language, p. 300. Mr. Lindsay seems somewhat to dis- 
 parage the value of this evidence, saying, c Puns are unsafe evidence of 
 pronunciation.' There are, however, perfect puns as well as imperfect ones ; 
 and it is obvious that the orator's witticism, ' ego quoque tibi fauebo ', must have 
 missed fire unless the assonance was absolutely perfect. 
 
 2 coquam the edd. 3 Preserved by A. 4 Preserved in P. 
 5 Preserved in P. 6 Preserved by A. 
 
 7 The word has to be printed in the capitals of the Romans to give the 
 requisite ambiguity. Compare Ovid, Met. xiv. 580 * nomen quoque mansit in 
 ilia | urbis et ipsa suis deplangitur ardea pennis ', with Breal, Semantics, Preface, 
 Eng. Ed. , p. xxxviii and note. 
 
44 PROCEEDINGS OF THE BRITISH ACADEMY 
 
 This would gain much in force if we supposed that Plautus is 
 glancing at the pretensions of his contemporary Ennius to be a rein- 
 carnate. The mention of sleep and the curious agreement of the phrase 
 with that of Pers. Sat. 6. 9 sqq. (in which also there is a scoffing 
 reference to the lines in the first book of the Annates where Ennius 
 asserted upon the principles of the Pythagorean philosophy his claim 
 to be Homer come to life again), constitute a coincidence too 
 striking to be wholly accidental. ' Lunai portum, est operae, cognoscite, 
 ciues: | cor iubet hoc Enni postquam destertuit esse | Maeonides 
 qvintvs pauone ex Pythagoreo.' On the literary relations of the two 
 contemporaries, Professor Vahlen has some observations in the Prae- 
 fatio to his second edition of Ennius, p. xxi. He notes there 
 a few coincidences in diction, to which may be added the remarkable 
 word dulcifer, Annals 264, Plaut. Pseud. 1262. 
 
 In Stichus 342 sq. 
 
 Pan. ecquem conuenisti ? Pi. multos. Pan. at uirum ? Pi. 
 
 equidem plurimos: 
 uerum ex multis nequiorem nullum quam hie est. 
 
 It is obvious at first sight that by uirum Panegyris, who is anxiously 
 expecting her husband, intends not f a man ! but ' her man ', but the 
 insipidity of the dialogue remains till, taking a hint from nequam 
 (Bacch. 195, Poen. 658), we observe that Plautus is playing on another 
 sense of multus {mollis, effeminate) which we find in a well-known 
 epigram of Catullus 112. 2 ' multus es et pathicus \ 
 
 In the Classical Review for 1901, xv. 305, 1 have commented on the 
 absurdity of inferring from the adjective in 'hirquinis follibus,' Hor. S.i. 
 4. 19, that the Romans made bellows from the hides of he-goats. 
 The ambiguity of hircus serves the same purpose of raising a laugh in 
 more than one passage of Plautus. I quote Pseud. 737 sq. ' Ps. 
 ecquid sapit ? Cha. hircum ab alis \ and Poen. 871 sqq. ' Sy. Sine 
 pinnis uolare hau facilest : meae aloe pinnas non habent. | Mi. nolito 
 edepol deuellisse : iam his duobus mensibus | uolucres tibi erunt tuae 
 hirquinaej for the purpose of suggesting that in the sham ravings of 
 Menaechmi, 837 sqq. 
 
 ita ilia me ab laeua rabiosa femina adseruat canes 
 poste autem illinc hircus alus qui saepe aetate in sua 
 perdidit ciuem innocentem falso testimonio, 
 
 the change of one letter from alus to ales is all that is required to 
 give the passage some meaning. 
 
 The scene of the Mostellaria, where the impudent slave Tranio is 
 fooling the unconscious Simo and Theopropides, is honeycombed 
 
FLAWS IN CLASSICAL RESEARCH 45 
 
 with doubles ententes, of which only one (in 816 a, b = 845 sq.) has 
 been noted. I have neither the time nor the inclination to set them 
 out at length. But I will indicate some parallels. The part oinequam 
 in the Stichus (1. c.) is played by improbiores here (824). 1 There are 
 equivoques in postis (818) suggesting posticus, Aus. Ep. 11 (70), 7 
 (a similar play on pone, Aid. 657). So ab inflmo, tarmes (tero, and 
 gurgulio, Pers. iv. 38), secat (Mart. vi. 37. 1), intempestiue excisos 
 (Ov. Fasti, iv. 361) 826, inducti pice (viromo-o-ovv, Ar. Plut. 1093) 827, 
 and coagmenta in foribus 829 (cf. Baehrens on Cat. xv. 12, Ellis on 
 ib. 18). 
 
 Lastly, may I use this opportunity to suggest that, in the amusing 
 colloquy of Persa, 316 sqq., 
 
 Sag. a ! a ! abi atque caue sis 
 a cornu. To. quid iam ? Sag. quia boues bini hie sunt in crumina. 
 To. emitte sodes, ne enices fame ; sine ire pastum. 
 Sag. enim metuo ut possiem in bubile reicere, ne uagentur. 
 To. ego reiciam. habe animum bonum. Sag. credetur, com- 
 
 modabo. 
 sequere hac sis. argentum hie inest quod mecum dudum 
 
 orasti. 
 To. quid tu ais ? Sag. dominus me boues mercatum Eretriam 
 
 misit. 
 nunc mi Eretria erit 2 haec tua domus. To. niim' tu facete 
 loquere, 
 324. atque ego omne argentum tibi hoc actutum incolume redigam ; 
 nam iam omnis sucopantias instruxi et comparaui, 
 quo pacto ab lenone auferam hoc argentum — Sag. tanto melior. 
 To. et mulier ut sit libera atque hoc det argentum. 
 
 armentum would do more justice to the poet's vein and to the usage 
 of redigam, so common in the sense of driving animals back, than 
 argentum ? Compare the phrases in 11. 317 sqq. which I have put 
 into italics. With three argentum's in the neighbourhood (321, 326, 
 327) corruption of one armentum in 324 was almost inevitable. 
 
 I can hardly hope to have collected all the evidence derivable 
 from plays upon proper names ; it is quite possible that I have over- 
 
 1 The audience were prepared to follow the comedian's meaning by the notable 
 comparison of a man and a house in the first act, sc. 2 (cf. 133 sq. ' probus 
 fui | in fabrorum potestate dum fui', 145 'ego sum in usu factus nequior'). 
 I may add that in arte dormiunt (829 quoted on p. 37) there is apparently no 
 latent impropriety. Tranio intends it for the audience who are to observe how 
 { fast asleep ' are the old men of whom he is making game, these two vultures that 
 an improba •' comix ' is plucking (832). About uecturam (823) I do not feel so sure. 
 
 ■ There is surely a pun in Eretria erit. Compare 'facete loquere '. 
 
46 PROCEEDINGS OF THE BRITISH ACADEMY 
 
 looked a good deal, because, as every one can see, it is not always 
 upon the surface. But what I have produced seems enough to show 
 that what our present practice loses us in detail is not inconsiderable. 
 But this is not all. The greater sensitiveness in respect of the trans- 
 literation of Greek words into Latin, which was developed not so 
 very long before the age of Cicero, is itself a fact of some significance. 
 It is a sign or concomitant of the tendency which was destined 
 first to polish and to sharpen, and ultimately to destroy the artistic 
 form of Latin poetry and prose — of the movement, the rise of which 
 can be traced for example, in Horace's only half outspoken distaste 
 for the ueteres, while its culmination is seen in Martial 'Accius et 
 quicquid Pacuuiusque uomunt' (xi. 90. 6). To the apostles and 
 devotees of the movement, the cultivated semi-Greeks of republican 
 and, still more, of imperial Rome, tursiis (Italian torso) and crupta 
 (Italian grottd) must have seemed as strange and boorish as the 
 Mounseer, which I have myself heard addressed to a Calais waiter, 
 or the pronunciation of Bordeaux as Bordoaks which Charles 
 James* Fox is said to have recommended, both by precept and 
 example, would appear to educated Englishmen since the time of 
 the Second Empire. The hybrid puns, which Plautus has in such pro- 
 fusion, were bound to disappear as soon as the domestic pronuncia- 
 tion of Greek words was felt to be inadequate. And it is noticeable 
 that the comedian of the Hellenizing circle of Scipio avoided them 
 altogether. 1 
 
 A disregard of the Plautine spelling in this respect is therefore 
 more than a total of petty errors. It amounts to the defacement 
 of an ancient monument, the removal of a landmark in Roman 
 literature, the destruction of one of the means by which we are 
 helped to a genuine appreciation of antiquity. But the spirit of 
 our age, in the praefatio of a leading editor of Plautus, has already 
 issued its denunciation of the man who should restore his proper 
 spelling to the dramatist whom he edits : ' qui " Bacidis " scribere ani- 
 mum inducat, merito rideatur ' — merito rideatur, my masters ! 
 
 It is curious that some of those who have not hitherto troubled 
 about this matter do, at least unconsciously, admit the principle that 
 the pre-classical literature should be differentiated from the classical 
 by its spelling. An attentive examination of the editing of different 
 Latin texts in the same series or by the same editor will reveal 
 
 1 Jt should be added that plays upon Latin words are not numerous in Terence. 
 All the clear ones that I have noticed are at Haut. 356, 372, 628 ; Eun. 236, 575 ; 
 Phorm. 374, 500, 842 sqq. ; Hec. Prol. 9 sq. ; Ad. 432 (575). Most of them are 
 put in the mouth of slaves. 
 
FLAWS IN CLASSICAL RESEARCH 47 
 
 a singular circumstance. Twenty-five characters are employed to 
 print the Augustans, but twenty-four are found enough for the Early 
 Republicans. In Propertius and Martial uua must be spelt with three 
 different letters ; in Plautus it has to be satisfied with two. In the 
 standard critical edition of Vergil via confronted us ; in the same 
 editor's fragments of ancient dramatic poetry it turned into nia. 
 Unless such editors imagined, what I will never impute to them, that 
 the Romans enlarged their alphabet by a distinction between the 
 vowel and consonantal sounds of u between the years 200 and 100 b.c. 
 this proceeding, if not to be explained as I have explained it, is as 
 motiveless as it is irrational. 
 
 Whenever men take upon themselves to abandon the plain, the 
 simple, and the practical custom of writing a language as it was 
 written by its only accredited employers, they fall into diverse errors 
 and inconsistencies: ille sinistrorsum, hie dextrorsum abit. Thus in 
 a well-known and valued Thesaurus Poeticus of Latin I find under the 
 letter U the following: 'V f. n. Lettre de V alphabet Subjecimus illam 
 cui nomen U dederunt, T. Maur. {Litt. 154), 1 and under the letter V 
 the following : ' V, n. f. Lettre de Valphabet Cecropiis ignota notis 
 ferale sonans V, Ausonius Id. \% Litt. 8. Hujus in locum videtur 
 V latina subdita ' T. Maur. (Syll. 93). 1 While in another Thesaurus, 
 the gigantic mausoleum now being erected for the remains of Latin 
 literature, you will search the headlines in vain for the distinction of 
 u and v which figures in the text below. 
 
 I can but touch upon the trouble which is caused in particular 
 passages when the genuine spelling is in itself ambiguous. At 
 Lucan vii. 658 the editors of the text are at issue whether the 
 text should give volxxit or volvit — a vain dispute. For what the 
 poet wrote was either, or rather neither, but uoluit. In the Revue de 
 Philologie for 1908, p. 54, the following is printed for an iambic 
 senarius of Plautus, Cos. 592 
 
 qui me atque uxorem ludificatust laraa. 
 
 The literary monuments of Plautine and pre-Plautine times are 
 unhappily few, and to but a small section of students of Latin is the 
 question how the Greek words contained therein should be written 
 a matter of much concern. But it is otherwise with u and v. This 
 late, 2 unnecessary and inconsistent 3 distinction affects the whole of 
 
 1 This gem of lexicography is still sparkling in the revised edition. 
 
 2 It is certainly not older than the seventeenth century. 
 
 3 Its inconsistency lies on the surface. If u and u (w) need distinguishing, 
 why not i and i (y). In Germany the paradox is greatest The German's j is 
 exactly the Latin i, and he does not use it : his is not the Latin u and he does. 
 
48 PROCEEDINGS OF THE BRITISH ACADEMY 
 
 spoken and written Latin in every period. It confuses the presentation 
 of the data both of palaeography and of philology, and tends directly 
 to keep alive a corrupt and misleading pronunciation. Is it then going 
 too far to say that it is high time that it should be condemned by the 
 sentence and discountenanced by the example of all the official 
 guardians of our studies, who should at last determine that what the 
 ancients have joined no modern has the right to put asunder ? 
 
 ADDENDA. 
 
 Page 20, Note 2. 
 
 Horace, S. i. 2. 120, affords a noteworthy parallel to the article in 
 Greek : illam ' post paulo ' ' sed pluris 1 ' si exierit uir '. What is this 
 but the Latin equivalent of is toutous tous ' oirx). irpobtocra) rbv 'AOrjvaiaiv 
 KoXoa-vprov.'' Aristophanes Vesp. 666 ? 
 
 Page 23. 
 
 I did not think it necessary to produce proof that moral con- 
 demnation of an act alleged was likely to affect our judgement upon 
 the evidence by which the allegation was supported. But since this 
 paper was in type I have lighted upon a passage in a well-known 
 book upon the social life of Rome under the Caesars, published in 
 1888, which I will quote, since it contains an admission far weightier 
 than any argument could be. The writer says (p. 73) : 
 
 'The dictates alike of feeling and reason forbid us to believe 
 the worst accounts that have reached us.' 
 
 (The italics are mine.) 
 
 Pages 24sqq. 
 
 To prevent all possible misconception, I would add that in these 
 pages I am contending for the credibility of ancient witnesses in 
 matters of fact alone. With their authority as grammatical theorists, 
 philologers, or textual critics I am in no wise concerned. Accordingly, 
 while I am bound to accept Quintilian's statement as to what stood 
 in the text of Vergil in his day, I own no compulsion to hold with 
 him that qui — risere could be followed after an interval of but three 
 short words by a hunc, for which a hos would have done every bit as 
 well. For this is a matter of grammatical theory. 
 
 V OF THE 
 
 UNIVERSITY 
 
 OF 
 
INDEX I. RERVM 
 
 Abstract substantives in Latin, 11. 
 Accent, Greek, 24. 
 
 modern, misleading effect of, 22. 
 alliteration in Plautus, 38. 
 
 in other writers, 40 n., 42. 
 anapaestic metres misread as dactylic, 
 
 22. 
 archaic language in solemn appeals, 
 
 19 sq. 
 Article, Homeric and later Greek, 16. 
 ,, survivals of, 21 . 
 in Italian, 20 n. 
 in Latin, 20 n. 
 
 three stages in development of, 16. 
 1 breathed ' and ' voiced ' r, etc., 31. 
 { circular ' and ' lineal ' construction in 
 
 the sentence, 7. 
 classical and scientific investigations, 
 
 differences of, 1. 
 Classical Research, flaws in modern, 
 
 passim. 
 Comparative Mythology and History, 
 
 3. 
 contemporary orthography to be re- 
 produced, 35. 
 Critic's Paradox, the, 2. 
 Demosthenes, citation of his speeches 
 
 by members, 33. 
 details, classical scholars' inattention 
 
 to, 33. 
 Dionysius of Halicarnassus on the 
 
 ancient Greek accent, 24 sq. 
 disparagements of ancient testimony, 
 
 24 sq. 
 dissensions of classical scholars, 3. 
 doubling of single syllables not avoided 
 
 by Romans, 32. 
 Ennius, allusions to, in Plautus, 44. 
 Ennius and Plautus, spelling of, 35. 
 etymology, perverting effect of, 12. 
 'fortes' and 'lenes' in Greek and 
 
 Latin, 31. 
 Glaucus of Samos, the five accents of, 
 
 30. 
 Greek words, spelling of, in early Latin, 
 
 35 sqq. 
 Homer, citation of books of Iliad and 
 
 Odyssey by letters, 33. 
 Homeric article in emphasis, 18. 
 ' hypallage ', 9. 
 hyperbaton, 7, 8. 
 
 e hyperbaton in syllables ', 8. 
 
 ' hysteron proteron', 7. 
 
 Latin words in modern languages, 4. 
 
 Martial's dislike of the ueteres, 46. 
 
 modern faults of method in weighing 
 ancient testimony, 30. 
 misconceptions of classical antiquity: 
 aesthetic and ethical, 23. 
 lexical, 4. 
 metrical, 22. 
 syntactical, 6. 
 
 Mommsen as an etymologist, 14. 
 
 Muller, Max, on septemtrio, 12. 
 
 mythological and historical research, 3. 
 
 orthography, ancient Latin, 35. 
 goal of, 35. 
 
 Pagus, a play of Plautus, 37. 
 
 Pharsalia or Palaepharsalus (not Phar- 
 salus), battle of, 34. 
 
 Plautine orthography, importance of, 
 35, 46 sqq. 
 puns, 36 sqq. 
 
 Plautus and Ennius, 44. 
 
 practical convenience, disregard of, by 
 classical scholars, 33. 
 
 puns in Plautus, 36 sq. 
 Terence, 46. 
 
 quantity disregarded in Plautine puns, 
 41 n. 
 modern neglect of, 22. 
 
 re-numeration of ancient MSS. , incon- 
 venience of, 33 sq. 
 
 Samnite wars, names of generals in, 
 32. 
 
 Sapphic, the, in Greek and Latin, 22. 
 
 sentences, differences of ancient and 
 modern, 7. 
 
 shortcomings of mythologists, 3. 
 of textual critics, 2. 
 
 single word in Greek and Latin equiva- 
 lent to phrase in modern lan- 
 guages, 11. 
 
 survivals of Homeric article in Attic, 
 19 sqq. 
 
 Terence, no puns on Greek words in, 
 46. 
 puns on Latin words, 46 n. 
 
 ' tmesis,' 8. 
 
 translation, mischiefs due to, 4. 
 
 transposition of verses, textual critics' 
 prejudice against, 2. 
 
 A4 
 
50 
 
 INDEX 
 
 ueteres, aversion of Homer and Martial 
 
 to the, 46. 
 vanity, modern, blinding effects of, 23. 
 Varro on hr, rh, or r, 31. 
 vocal chords, as modifiers of conso- 
 
 nantal sound, unknown to Greek 
 and Roman grammarians, 31. 
 
 u and v in Latin, an irrational modern 
 distinction, 46. 
 
 Zeta, Appius Claudius on, 30. 
 
 INDEX II. VERBORVM 
 
 Aganippis Hippocrene, 10. 
 
 anus not necessarily slighting, 5. 
 
 Arcidemides, 38. 
 
 Atenae Atticae, 39. 
 
 Carinus, 42. 
 
 Crusalus, 38. 
 
 CrusiSy 41. 
 
 Culindrus, 39. 
 
 dacruma, 40 n. 
 
 doliaris, 41 n. 
 
 dracuma, 40. 
 
 duicifer (Ennius and Plautus), 44. 
 
 ebriosus, ebrius, 28. 
 
 ebrius (of a grape), 29. 
 
 Eretria and erit, 45 n. 
 
 hinc = ex his, 27. 
 
 hircus ales, 44. 
 
 hirquini folks, 44. 
 
 ille as article, 20 n., 48. 
 
 intepere, 6. 
 
 inuisus, e envied,' 12. 
 
 Ulcus, 5. 
 
 Lucus, 39. 
 
 Ludus, 38. 
 
 multus mm mollis, 44. 
 
 muscus, 39. 
 
 o/os, odos, 40 n. 
 
 optalmia, 41. 
 
 Ortygius Cayster, 10. 
 
 jtxmw and ;>m& opposed, 37. 
 
 Pelusiacus Canopus, 9. 
 
 per, ' displacement ' of, 8. 
 
 pernio and perna, 41. 
 
 pernula, 41 . 
 
 Poenicium, 40. 
 
 ponwrium, 14. 
 
 proelium, 13. 
 
 pultem (pulto), 36. 
 
 pultipagus, 37. 
 
 quintus and Quintus, 43. 
 
 quoquo, quoquos (quoque), 43. 
 
 septemtrio, 12. 
 
 *owa (not sfona), in Plautus, 36 and n. 
 
 Swrws, 36, 40, 41. 
 
 Ta/e*, 39. 
 
 Titonus, 39. 
 
 frames, 13. 
 
 Tw^stae (dat.), 42. 
 
 aKparto-fios , 30 n. 
 
 dprra£, 41 n. 
 
 f^ios (with unusual art.), 19. 
 
 cVi, (c. dat.) of viands, 30. 
 
 {•rjpocfrayia, 30. 
 
 ra cKcurra (Horn.), 18. 
 
 TtBff, 20. 
 
 <f)l\t<TTOS, 20. 
 
 INDEX III. LOCORVM 
 
 Aristophanes, Lysistrata, 628 : 8. 
 
 Fc*j9ae, 666 : 48. 
 Auienus, iii. 24 : 9. 
 Bibaculus ap. Suet. : 30. 
 Cassiodorus, Var. ii. 21. 2 : 5 n. 
 Catullus, xxvii. 4 : 28. 
 
 lxvi. 77: 7. 
 Dionysius (Hal.), nepi crvvOeo-tas ovopd- 
 
 tow, c. xi. : 24. 
 Ennius, Annals, 108 : 42. 
 
 Epitaph, 40 n. 
 Euripides, Hippolytus, 471 : 21. 
 
 682 sqq. : 19. 
 Gellius, Noct. AtL, vi. 20 : 6. 
 
 xiii. 14 : 15. 
 Homer, Iliad, i. 109 : 34. 
 x. 330 : 17. 
 x. 495 : 21. 
 xxiii. 75 : 17. 
 
 Homer, Odyssey, xii. 16 : 18. 
 xvi. 131 : 34. 
 xviii. 74 : 17. 
 xx. 309 : 21. 
 xxi. 378 : 18. 
 Horace, S. i. 2. 120 : 48. 
 ii. 1. 79 : 27. 
 ii. 2. 9 sqq. : 11. 
 ii. 6. 30: 42 n. 
 Juvenal, i. 115 : 32. 
 Livy, i. 44 : 15. 
 ii. 39. 3 : 14. 
 Lucan, viii. 342 sq. : 7. 
 viii. 543 : 9. 
 x. 184 sqq. : 11. 
 Lucretius, vi. 27 : 14. 
 Ovid, Fasti, v. 7 : 10. 
 
 Metamorphoses, xi. 127 sqq. : 
 xiv. 580 : 43 n. 
 
 12. 
 
INDEX 
 
 51 
 
 Persius, vi. 9 sqq. : 44. 
 
 vi. 37 sqq. : 37. 
 Plautus, Amphitryo, 304 : 45. 
 Bacchides, 362 : 37. 
 
 129, 138 : 38. 
 416, 467 : 38. 
 683, 687, 922 : 38. 
 Captiui, 274 : 39. 
 850 : 41. 
 Epidieus, 20 sqq. : 39. 
 Menaechmi, 294-5 : 39. 
 
 646 sqq. : 42 n. 
 837 sqq. : 44. 
 854 : 39. 
 Mostellaria, 816 sqq.: 45. 
 
 828 : 37. 
 Persa, 324: 45. 
 Poenulus, 157, 187, 648 : 39. 
 689 sqq. : 39. 
 871 sqq. : 44. 
 Pseudolus, 99 sqq. : 40. 
 228 : 40. 
 636 sqq. : 40. 
 654 : 41 n. 1. 
 659 : 41. 
 712, 736 : 42. 
 841 sqq. : 40 n. 
 851 sqq. : 43. 
 
 Plautus, Pseudolus, 889 sqq. : 43 . 
 
 1173sqq.— 1215:41. 
 
 Rudens, 494 sqq. : 42. 
 
 Stichus, 342 sq. : 44. 
 
 Trinummus, 525: 29 n. 
 Pliny, Nat. Hist, xiv. 17: 29. 
 Plutarch, Camillus, 80 : 32. 
 Propertius, iii. 3. 32 : 6. 
 
 iii.5. 3: 12. 
 
 iii. 12. 1 : 42 n. 
 
 iii. 22. 15 : 10. 
 
 iv. 1. 124 : 5. 
 Quintilian, vi. 3. 47 : 43. 
 
 ix. 3 : 27. 
 
 ix. 4. 4 : 32. 
 Seneca, Ep. lxxxiii. 11 : 29. 
 Servius on Aen. ii. 27 : 32. 
 Silius Italicus, i. 627 sqq. : 11. 
 Statins, Siluae, ii. 7. 70 : 9 n. 
 Sophocles, Ajax, 572 sq. : 19. 
 835 : 20. 
 
 Oed. Tyr., 572 : 21. 
 
 Trachiniae, 775 : 21. 
 1249 : 21. 
 Theocritus, i. 45 sqq. : 29. 
 Terence, Hecyra, 159 sq. : 7. 
 [Tibullus] Lygdamus, iii. 1. 16 : 6. 
 Vergil, Eclogues, iv. 62 : 26. 
 
HOME USE 
 
 CIRCULATION DEPARTMENT 
 
 MAIN LIBRARY 
 
 „ . General Library 
 Un.vers.ty of California 
 Berkeley