In 1667, when Paradise Lost first reached print, its title page proclaimed it "a poem written in ten books." In 1674, however, Milton brought out a revised edition, now in twelve books. This twelve-book version became canonical; and yet his changes have often seemed so few and small, and the poem itself in most ways already so finished, that its need for restructuring remains mysterious. What did Milton find less good about the first version? and better about the second one?
Although this double-yoked question has of course been addressed many times by Milton scholarship, it has never been settled, for several reasons. One is the lack of decisive evidence. A related reason is that the issue is so intriguing that all Miltonists want to answer it for themselves. (And so do 1, here.) But thirdly the inquiry needs regular revival because it raises wider questions of method -- what counts as evidence, and how best to use it.
My main aim is to answer both parts of the question with new or neglected evidence. But to prepare the ground for that, I begin with some previous answers and their methods of using evidence.
The traditional answer relied on the poem's declared epic ancestry. Homer's epics had twenty-four each, and Virgil's had twelve. Both, obviously, were built on a base of twelve. But whereas Homer's twenty-four derived first and foremost from there being twenty-four letters in the Greek alphabet, Virgil's twelve proclaims another intention -- to recall Homer, yet differ from him. As for Milton, then, following both, the fact that like Homer he suffered blindness and makes blindness a motif, assured that readers felt Homer as exemplar. But if so, then the twelve books of 1674 assured the other link, to the Aeneid as greatest of secondary epics.
Nevertheless, although this inference remains persuasive, it has drawbacks. It is only inference. It may be putting the cart before the horse, the contingent before the necessary, as Addison urged.(1) (The redivision of the books "was not done for the sake of a Chimerical Beauty, but for the more just and regular Disposition of this great Work.") Indeed, the more strongly the case for twelve books is expressed, the harder it becomes to explain why Milton first had his poem in ten.
An essay by Arthur Barker has the great merit of answering both parts of the question -- both at once, too.(2) He reasons that although the poem was Virgilian from the outset in many respects, it was not so in its book-divisions, because these at first sought a five-act structure. This structure was of two books per "act," just as in Davenant's Gondibert of 1650-52 and in "a long series of abortive five-act epic experiments." Barker reasons that subsequent reflection showed Milton his structuring had put excessive weight on his fourth act," the old Books VII and VIII, corresponding to new Books 7-9.(3) In other words, structure overstressed the Fall itself, and Satan's success there; it emphasized humanity's loss at the expense of gain. The new structuring, says Barker, "by suggesting a different structural pattern bring[s] out implications muted in the earlier division" (p. 23).
Barker's essay has never, to my knowledge, been fully answered. Not only is it marvellously persuasive, it ramifies adventurously. It helps explain how Milton was "of the devil's party," Satan-centred by a structural inadvertence, which in 1674 he took his chance to put right. It shows him articulating his Virgilian debt in 1674, not only for structural reasons but in order to dwell on the central paradoxes of loss and gain.
Nevertheless, Barker's reasoning did not win general allegiance, because of some natural misgivings; and after 1949 it underwent virtual eclipse because Milton scholarship had discovered numerology.
To summarize the misgivings, first. That Milton would be taking any notice of Davenant's theory or practice during composition of 1667, which goes back to the 1640s, is a dubious assumption, lacking much supporting evidence. And that he wrote VII-VIII (= 7-9) as a single unit is just not plausible. The Fall itself always began with a fresh invocation, ( "No more of talk where God or angel guest . . ." VIII.1 = 9.1), sundering the two supposed parts of Act Four. Contrariwise, the Fall always rolled on into its expanding repercussions in the next book (IX = 10). I return to these matters later.
As for the numerological onslaught, Alastair Fowler's edition of the poem (1968) rounds up work by others and himself to argue that an eloquent number-symbolism -- having nothing to do with five-act structure nor yet with Homer and Virgil -- exists in both versions.(4) Briefly, line-counting reveals that 1667 hinges on VI.761-62, said to be thematically central lines about Christ's victory:
[He in celestial panoply all armed]
Of radiant urim, work divinely wrought
Ascended, at his right hand victory
[Sat, eagle-winged, beside him hung his bowl.
In 1674, however, not lines but Books are to be counted, and the poem hinges on the third of its four invocations when it declares that "half yet remains unsung" (7.21).(5)
Numerologists continue to make the main new contribution to the debate. I myself heard a scintillating paper in 1988 by the Norwegian scholar Maren-Sofie Rostvig.(6) She was arguing that a symmetry of books which worked by proportioning was superior to one which merely balanced its books one-to-one, whether by linear or palindromic means. She compared the Confessions of Augustine and his preference for a proportioning of 2:1 to argue that the elegant balancing of that work (known to Milton) is matched in 1667 -- and not in 1674 -- which makes the first edition superior. To urge that Milton mismanaged the revision we are examining is the logical bold next step, which Rostvig duly takes. She advocates that editors should "return to the first edition as the authoritative text."
Bibliography is implicitly called in support here, and certainly bibliography has been moving away from the old automatic preference of an author's latest revision as copy-text.(7) Some editors now prefer an author's first thoughts; some like a composite text; some view each text of each author as a case by itself. Some might invoke senility, or the late-Henry-James syndrome. Our present question does interact with these editorial issues. As for me, however, I favor the pragmatic position, of taking each case on its merits, since it is precisely the rival merits -- for Milton …