This species of translation is generally called 'Composition,' a title seemingly appropriated from what is now specially designated as 'free' or 'original' composition, a linguistic exercise which in Latin and Greek has fallen into unmerited neglect.

The motives of the two species of Translation are different. The sole design of the one is to impart a knowledge of an original to those to whom it would otherwise be unknown or, in the case of exercises, to show the possession of that knowledge by the translator. It is therefore in its essence Retrospective. The other includes in its design the embodiment and exhibition in the translation of a knowledge of the language into which the translation is made. It is accordingly Prospective in exact proportion as its care is for the copy rather than the original.

For the present this distinction must serve. But it would be better if we could turn the difference in the Latin words which we have already noted on p. I to further account and called renderings of the first kind 'translations' and those of the second kind ' versions,' and, taking yet another step forward, distinguished their makers as 'translators' and 'verters' and their work as 'translating' and 'verting.'

In perfect bilingualism, as already said, it is indifferent from which direction approach is made. The aspect of a straight line is the same from either side. of it. Not so with the concave and convex aspects of a curve. In Retrospective translation ascertainment and comprehension come first, and expression follows in their train. But in Prospective translation comprehension is assumed to be already attained, and the whole mental effort may be concentrated on Expression.

Nor is this all. In Retrospective translation the translator's better knowledge of his native language gives him a wider field of choice and enables him to select with greater confidence the turn most appropriate to the occasion. In Prospective translation the 'composer' must often use an inadequate phrase, because it is a common one, or acquiesce in a looser rendering because his knowledge is not sure enough to attempt a closer. Hence the pithy saying attributed to a schoolboy: There are some things I may not write but Master may-and some that neither of us may write but Horace and Vergil may.'

We can now understand why good 'translators' are not necessarily good 'composers' nor 'good composers' necessarily good translators, as indeed all teachers and examiners know. There is no paradox here. The idea that there is seems to spring from the fallacy that, if A is a good translation of B, therefore B is a good translation of A. This led Mr S. G. Tremenheere to write in the preface to his interesting rendering of the Cynthia of Propertius (1899): 'I shall be satisfied if the reader considers that, supposing my lines were the original, the Latin of Propertius is a just rendering of them. That is the criterion which I have applied to myself.' The contrary view is implied in what a former colleague at Trinity, Mr F. M. Cornford, himself a deft translator both ways, wrote of a translation from Cicero which he made for the use of our pupils in Latin prose: 'The English is designedly not a good translation of the Latin; but the Latin is a good version of the English.'

We may here deal with a difficulty raised by Professor Naylor in his interesting little book

'Latin and English idiom' (Cambridge 1909). Professor Naylor, whose aim is a practical one', in an attack on 'accurate translation' which, like Professor Wilamowitz, above, p. 7, he regards as an 'unmixed evil,' puts the following dilemma:

The methods of expression found in the two dead languages are often so utterly different from those of modern times that we allow the impossibility of word-for-word renderings from English but make no such concession when the position is reversed. Thus were I asked to put into Greek, 'In this way the myth was preserved, I write οὕτως ὁ μῦθος ἐσώθη καὶ οὐκ áróλero; but your 'translator' says (Davies and Vaughan, Plato, Rep. 621 B): 'thus...the tale was preserved and did not perish2'

Mr Naylor assumes that the processes employed in Retrospective and Prospective translation are identical. But we have seen that they are not. It is proper to tell the student who is turning, say, English into Latin, to put 'safety first' and to aim not at the nearest idiomatic translation but at the most idiomatic that he can find. For his object is to write Latin, not to render English, and to him the borderland of doubt and possible error is much larger in the use of Latin than in that of English expression. All know that liberties are allowed to renderers of English prose and verse into the Classical languages which the most 'adaptative' of translators would not now take when translating the Classics into

1 'At a time when Classics are on their defence it might be well to sacrifice 'Composition' altogether and to ask our students to 'Anglicize' as well as to 'translate' the passages set before them' (p. 4).

2 To me neither of these renderings seems adequate, and so neither of them 'accurate.' The English of the second is not quite idiomatic ; but the first sacrifices the emphasis of the Greek, giving us one verb only instead of two. 'Was saved from perishing' would seem better than either.

English. I have before my mind an excellent version of a passage from Macaulay by a master of Latin and English, my friend Sir James Frazer, in which he transformed completely a description of Monmouth's movements before the battle of Sedgmoor by transferring it to the soil of Italy and the civil warfare of A.D. 69.

In matters of smaller detail the point may be illustrated from Prospective translations of my own. In the version from Gray no. 70 I had originally given Ρουβέλλιος for Rubellius and Σύλλας for 'Sylla,' as I should have done had I been translating Gray's tragedy for the benefit of an ancient Greek. But in a Prospective translation into Greek tragic trimeters these Latin names seemed incongruous and hence they now appear as Πύρρος and Δίων.

When the author of the Hebrew Melodies needed a lake for a simile it was 'deep Galilee' that occurred to him (no. 42) and Retrospective translators of Byron into French, German, and so forth must of course leave this name untouched. But a Prospective translator into Latin will at least be excused if he substitutes as more appropriate the name of some Italian lake.

As divisions of the year uer is undoubtedly 'spring' and aestas 'summer.' But the differences of a Northern and a Southern climate are reflected in the associations of the words and a Prospective translator into Latin Verse must often represent the idea of 'summer' by uer or uernus (nos. 27, 40, 42).

Xíμvn is the nearest word in Greek to the English 'tarn,' but its associations unfit it for rendering the 'unsunned tarn' of Browning no. 72.

Rossetti, in disregard of Classic tradition, makes Cassandra 'wring her hands,' no. 31. But the translator into ancient hexameters must eschew the anachronism. Scott's dedicatory poem no. 46 was of course addressed to a woman. But in Roman times it would have been addressed to a man; and deus is the only Latin word which carries any of the suggestions of 'angel.' The human 'goddesses' of Catullus and his like had nothing 'angelic' about them. In all such cases as I said in 'Sermo Latinus' p. 53, what is aimed at is not 'verity' but 'verisimilitude.'

Nor is it otherwise with differences that concern the 'genius' of a language. The signal indirectness of English speech-its habit of leaving out the essential part of an expression—will not be reproduced by the Prospective translator. Byron in the Giaour, using metaphor within a simile, writes of 'the insect-queen of Eastern spring,' no. 43, meaning thereby a butterfly (as his commentators explain). A reader who does not recognise the queenliness of this insect will not understand the simile; and a translator into Latin Verse must, by calling a butterfly a 'butterfly,' ensure that he shall. The same writer makes his wolf come down on 'the fold' no. 42. The Latin for a 'sheepfold' is ouile; but the Prospective translator must put the sheep in it 'plenum ouile' Aen. IX 59, 339. In Tennyson's lines on the eagle about to swoop (no. 57) the bird is said to be watching, but what he 'watches' we are not told. Not thus Apuleius (Florida II p. 146 de Vliet) from whom the English poet is drawing 'quaerit quorsus potissimum in praedam superne sese ruat fulminis uicem de caelo inprouisa.'

In the same writer Iphigenia thus describes her end