ture of this epic, Virgil had not so much to invent as to select and harmonize from an enormous mass of fragmentary and often conflicting legend, both Greek and Latin. He had definitely rejected two plans: that on the one hand of a chronicle-poem on the lines followed by his Latin predecessors; that on the other hand which he had been urged to take both by court-pressure and by the prevalent tendency of fashion, of a poem the main scenes and action of which should be contemporary. But here his real difficulties only began. He had to create a world in which to place his action. He had to fix and observe the limits demanded in order to secure, within his chosen frame-work, epic unity and dramatic tension. He had to make his story credible, interesting, human. Yet he had so to construct it as to give scope for all the motives mentioned above. He had to preserve, through a fabric of unexampled complexity, a single large pattern, a single dominating tone. The artist must not be overwhelmed by the material, or lose breadth of handling in laborious refinement of workmanship.

Virgil then had to lay his foundations deliberately and carefully. There is no reason to

VIRGIL AND HIS MEANING

question, on theoretic grounds, the statement of Donatus 20 that he drew out, in prose, a detailed sketch-plan of the contents of his Aeneid, divided into books, and that he kept this by him for reference and guidance in working on one or another part of the poem. This bears out what from internal evidence alone is sufficiently clear, that the composition of the Aeneid was not continuous. Portions which come later were composed before others which come earlier. The scheme underwent modifications, during the progress of composition, in important particulars. As published after his death it was not yet what he meant to make it. The editors performed their difficult task with great skill and judgment, but they could not do what Virgil himself had left undone. Of what they actually did they do not appear to have left any written record. But the general principle on which they proceeded is expressly stated: it was to cancel where necessary, but to add nothing— hac lege ut superflua demerent, nihil adderent tamen.21 The Aeneid as they found and left it contains a certain number of inconsistencies, repetitions, and awkwardnesses of handling or transition. There were in it several gaps, and quite a considerable number of passages

centatively, perhaps marginally, inserted by Virgil's hand, and not yet decisively adopted nor fully incorporated. Some of these, we are told, were struck out by the editors, and four such have been actually preserved. It was clearly Virgil's object to keep the poem fluid, as far as possible, up to the last moment. The Aeneid, as we possess it, is the product of an immense mass of raw material, all of it worked upon, but still in process of undergoing large revision; yet in the main so nearly approaching to its final shape that it has been for the world at large a completed work of art. Careful study and minute analysis enable us to distinguish in it what may be called strata of composition, and to surmise, with greater or less probability, a good deal of Virgil's actual processes. We can assign the relative date of certain parts; we can trace the insertion or expansion, the re-casting or cancellation of certain episodes or passages. That enquiry is perilous to pursue too far, but is fascinatingly interesting. It can be pursued effectively only by highly trained investigators; and only by those among them whose scholarship is combined with delicate artistic sense and with some faculty of imaginative divination.

VIRGIL AND HIS MEANING

One instance may be cited as illustrative, because it is on a large scale and involves comparatively simple problems. The third book of the Aeneid differs in style, and is at a different stage of structure, from all the others. It gives the story of Aeneas' adventures, from the departure from Troy up to the arrival in Sicily, as related by him to Dido at Carthage, in continuation of his narrative of the fall of Troy. Its construction is clear evidence not only of its unfinished state, but of its early composition. It contains passages, particularly at the beginning and end, in the dry, annalistic manner of Virgil's predecessors, which we may say with confidence are youthful work, anterior to the Georgics. Inserted in it are lumps of rather laboured material, easily detachable from their context (e.g., 11. 414-428, 445-452, 575-582). The substance, and even the wording, of the episode of Achaemenides (11. 588-654) has has been largely drawn upon, and used to better effect, for the episode of Laocoön in the second book; and it is not impossible that Virgil would have struck it out entirely. Throughout, except in the opening and concluding lines, and in the beautifully managed episode of the meeting

with Helenus and Andromache, the narrative is curiously impersonal, as though told by a chronicler, and in sharp contrast to Book II, which is saturated from beginning to end with the personality and emotion of the narrator. In ten or a dozen places at least there are lines or phrases quite unsuitably put in the mouth of Aeneas, scraps of history or archaeology, and even references to the subsequent history of places mentioned. There are also marked, if minor, discrepancies from the rest of the poem, as when Helenus prophesies to Aeneas that the Sibyl will tell him what in fact is told him by the glorified spirit of Anchises.

Recent research, with the more delicate analysis possible to modern scholarship, has found the clue. It has established beyond reasonable doubt that the substance of this book was originally written as direct narrative, and was in fact the first book of the poem. Twothirds of it at least can be reinstated in that form with slight verbal changes, most of which can be made even without disturbing the metre. The clue, once grasped, enables us not only to explain what had always been perplexing difficulties, but to get, as it were, inside Virgil's