^{page 33 note 1} We know also why this criterion would be of no use for dating verses of Cicero's. Probably the fashion had not come in when he wrote his Aratea, admodunt adulescentulus. When it did come in, he thought it an affectation, as appears from the well-known passage in his Letters (ad Att. vii, 2, I: ita belle nobis flauit ab Epiro lenissimus Onchesmites; hunc σπoνδειáζoντa si cui uoles των νεωτέρων pro tuo uendito).

^{page 34 note 1} A Latin word forming a dispondeus, not a Greek proper name, as is usually the case in Augustan poetry (nobile Pallanteum).

^{page 35 note 1} It is pointed out by Drachmann, in an article which I shall discuss later, that we should have to assume a writer who was unaffected by the revolution which Virgil had made in versification, and yet knew and admired Virgil so much as to introduce into his own poem whole lines and phrases from Virgil's writings. It is not a supposition that can be taken seriously.

^{page 35 note 2} Or nearly so, if we exclude the non as belonging to the general construction of the whole passage.

^{page 37 note 1} In all the eleven cases the participial clause precedes the main verb.

^{page 37 note 2} This infrequency may be set down to the fact that the Eclogues are a dramatic form of composition. Participial clauses belong chiefly to description and narration (description in Cicero's Aratea: narration, in Lucr. VI., where they become more frequent in the narrative of the plague at Athens); they are less likely to occur in conversation. But description or narration, of course, does not compel their use: compare Germanicus with Cicero, the Aeneid with Catullus or the Ciris.

^{page 37 note 3} This solitary specimen (Ecl. X. 25) is an echo of Lucretius (IV. 587).

^{page 39 note 1} There are also half a dozen more like νoûσαν ανα στρατδν ωρσε κακην δλεκoντo δε λαoí or 'Aτρεíδα δε μáλιστα δνω, κoσμητoρε λαων, which I count as hephthemimeral and not as ‘trochaic.’.

^{page 39 note 2} Hexameters which fall into two equal parts, like Ennius' spernitur orator bonus, horridus miles amatur, are so very rare as to be altogether negligible for this inquiry.

^{page 39 note 3} Of course it is not a fact of absolutely no importance at all. It means that the line is not exactly like 'obstupuit simul Aeneas ‥.

^{page 39 note 4} From an article by Drachmann, A. B. in Hermes (vol. 43, p. 412)Google Scholar I learn that Meyer was the originator of the method. Drachmann himself adopts Meyer's rules, ‘though I am not quite convinced that they are right’.

^{page 42 note 1} This result I believe to be substantially sound, and as accurate as can be looked for. Substantially sound, because there is no reason to think that Virgil's manner in the 1,000 lines selected was different from his manner in other parts of the Aeneid. He had already written the Georgics, and his style was mature. As accurate as can be expected, because the attempt to count all the lines in the Aeneid would not result in absolute precision and certainty. Some lines marked ph might seem to be p on a second scrutiny or if considered by another critic, and some marked p turn out to be ph; some marked th might be finally relegated to h. Callimachus maxim must be kept in mind:μη μερεν σχoινψ πεσιδιτην σoϕμ—if the words meant that it is unsafe to apply a. footrule to poetry.

^{page 43 note 1} In Catullus LXIV., on my method of reckoning, p amounts to 827 per cent. In the Ciris and Laus Pisonis it is about 80 per cent. In other Latin poems the percentage does not depart widely from the Virgilian one, i.e. it is about 64. Catullus had a very strong preference for the rhythm ‘prognátae vértice pínus’, which excludes h.

^{page 44 note 1} The adjective εμπνρα naturally goes closely with σηματα. Between it and σηματα there is slightly less severance than between σηματα and ιδεσθαι. It may be thought that elision has a connecting effect, and makes σηματ ιδεσθαι. a group; but that effect is doubtful, and I think, if it exists, balanced by the close connection between εμπνρα and σηματα.

^{page 44 note 2} Ecl. III. 15 is one of the rare exceptions. The line quoted above is in the Fourth a Eclogue. In that Eclogue there is not a single example of b, as I reckon b's. In this I see a further confirmation of my position. For Ecl. IV. ought to be very different from the others: it is not a pastoral or dramatic idyll at all.

^{page 47 note 1} What Drachmann surveys is a tract of two or three hundred lines in each case, not necessarily consecutive, for he limits himself to narrative or description, avoiding speeches. This seems a sound precaution. It is specially in regard to the relation of sentences to verses that dramatic parts would differ from narrative. Of course some poets show greater still than others in differentiating them. It is not, I think, a very important consideration for the texts I have dealt with above. I have been content to take rather larger tracts of text (usually not less than 500 lines).

^{page 48 note 1} Drachmann deals briefly svith a similar thing, lines in which two adjectives and two substantives are symmetrically placed, such as ‘Gnosia Cecropiae tetigissent litora puppes’ (ab A B) and ‘indomito nee dira ferens stipendia tauro’ (ab B A). His results are: Ciris, 15'5 per cent.; Catullus LXIV., I4'5; Eclogues, 47; Georg. I. and IV., 6'I (Hermes 43, p. 418). My results for A so far agree with this that they put the Ciris and Catullus in one group and the Eclogues and Georgics in another. They also, it may be note d, associate the Culex with the Eclogues, and would put it slightly before them in time—if we limit our view to this sequence of five, Cat., Ciris, Culex, Ecl.. Georgics.

^{page 48 note 2} As it does in Homer.