For some six weeks in early 1804 Wordsworth planned to complete his autobiographical poem to Coleridge—the poem that was to become The Prelude—in "five parts or books."  On March 6, he wrote to Coleridge that he would within a matter of days commence the fifth and final book of the poem. Plans had changed radically by March 12, and Wordsworth, now referring to the autobiographical poem as a much more extensive undertaking on which he was "advancing rapidly," reshaped his materials into the first five books of a new poem, the eventual length of which was probably as yet undecided.  By March 18, Coleridge had received as part of the manuscript of Wordsworth's poetry that he would take with him to Malta (MS. M; Dove Cottage MS. 44) these first five books of the poem that was to become the thirteen-book Prelude, and Wordsworth was hard at work on verse to be incorporated into later books.
How close did Wordsworth come to finishing his relatively modest, five-book poem before deciding to commit himself to the work that would grow to thirteen books by May of 1805? What shape would the finished poem have taken? What led to the decision to abandon the five-book poem, finished or not, and how much of what Wordsworth did compose for the five-book poem is recoverable from the extraordinarily difficult manuscripts that contain this work (MSS. W and WW, Dove Cottage MSS. 38 and 43)? These are tantalizing questions, as they take us to the heart of the creative process out of which a great poem was born.
Jonathan Wordsworth's 1977 article, "The Five-Book Prelude of Early Spring 1804," first made the case that a five-book poem was probably "either finished or within easy striking distance of completion" in March 1804, and that it could be reconstructed in considerable detail from the available manuscripts.  The poem that he describes through close attention to progress reports among the Wordsworth circle and to details in MSS. W and WW probably would have begun in a manner identical to the 1805, thirteen-book poem, with the first three books corresponding to books I-III of 1805 (I and II closely, III more generally). Book IV probably combined the contents of 1805, Books IV (vacation) and V (books). Book V, the most difficult of the books to reconstruct in detail, probably began with material that would eventually form the first third of 1805, Book XIII (the ascent of Snowdon and a version of XIII, 70-165), and included about the last two-thirds of 1805, Book XI (ll. 123-388), ending with the "spots of time." This poem, Jonathan Wordsworth argued, is "in many ways the most impressive of the Preludes, bringing together in a densely packed, unique, and formally satisfying unit the great poetry of Wordsworth's original inspiration at Goslar in 1798, and the new magnificent sequences of early 1804" (1). On the question of why this formally satisfying whole, which was either complete or nearly complete by ca. 10 March, was almost immediately cannibalized in the process of pursuing much larger plans, Jonathan Wordsworth suggests that the poet was motivated by dissatisfaction with having left out biographical information (particularly the 1790 visit to France) and by a more general "unwillingness to make an attempt at the central philosophical section of The Recluse" (24). The 1977 article stops short of the claim that MSS. W and WW provide sufficient evidence on which to base an edition of the poem: "The poem does not survive as a whole in fair copy and cannot be printed, as can 1799 and 1805," though it "can . . . be reconstructed with a fair degree of certainty and precision" (1-2). Two years later, Jonathan Wordsworth and his co-editors declined to include a five-book text in their Norton critical edition of The Prelude, 1799, 1805, 1850 because "too little survives in faircopy" (xi). 
In 1992, Mark L. Reed published the most extensive arguments to date for an alternative view of this 1804 work. In a closely argued, twenty-nine page section of the introduction to his Cornell edition of The Thirteen-Book Prelude, Reed agrees with Jonathan Wordsworth on Wordsworth's intention to complete a poem in five books and on the probable content of the first two books of that work. Reed's view of Books III and IV, however, differs significantly from Jonathan Wordsworth's. Book III, Reed argues, would have encompassed the calendar year beginning with the poet's arrival at Cambridge, including material from his return to Hawkshead; Book IV was "probably a close equivalent of the book about books" that would become Book V in 1805.  Reed also comes to very different conclusions about how close the poem came to completion and about why Wordsworth abandoned it. According to Reed, Wordsworth's plans to complete the five-book poem sustained him only through two of three phases of work on the fifth book that he told Coleridge would be his last. In the first stage, Wordsworth planned to begin with consideration of the "quasi-imaginative power of Nature to work on outward things so as to impress human minds" (Mount Snowdon), which was to be illustrated by various examples (the passage now known as the "analogy passage"). Stage two probably involved a decision to make "'higher minds,' as such—rather than natural effects analogous to those produced by higher minds" the main subject. At this stage, the analogy passage was rejected, and Wordsworth seems to have intended to devote his final book to discussions of "imaginative love and liberty and the enlargement of these spiritual dimensions of his own life," especially as fostered through his relationship with Dorothy, Mary, and (perhaps) Coleridge. He may also have intended some discussion of the foes of imaginative development and of the means employed by Nature to counter those foes, and he "certainly" intended to conclude with the Spots of Time sequence (I, 35).
It is at that point, Reed argues, that Wordsworth decided to expand the poem indefinitely, as he struggled in MS. W 45r-47r with the theme that would eventually be developed in 1805 Book XI—"Imagination, How Impaired and Restored":
The passage incorporating these lines, Reed argues, "almost certainly marks the point by which Wordsworth's intention to complete the poem in five books had succumbed before a paramount impulse to treat later events of the poet's progress to maturity" (I, 34). An adequate treatment of that "warfare" and those "foes" required immense extension of the region of Wordsworth's song and called him to "treat the years and revolutionary events that had not only distracted him but led to the organization of his own mature mind and the mind of the modern world" (I, 38-39). Whereas Jonathan Wordsworth's view of the evidence suggests that Wordsworth completed or very nearly completed a poem that he almost immediately found unsatisfactory, Reed argues that Wordsworth found the theme of the full-scale Prelude in the restless, revisionary activity of the "5th book" materials in MS. W. As there is no completed poem recoverable in the extant manuscripts, Reed's edition presents work on the five-book Prelude through photographs and full transcriptions of MSS. W and WW. These give unprecedented access to the best evidence we have about this key moment in Wordsworth's development, the development of the Prelude, and English literary history.
Duncan Wu, then, has gone where no Prelude editor has gone before. Concurring with Jonathan Wordsworth on all important matters regarding the contents, organization, degree of completion, and significance of the poem, Wu provides a reading text that attempts to capture the poem described in the 1977 article as it may have existed at the moment of its abandonment, "only one step from completion" (20). In his introduction, Wu acknowledges the extreme degree of uncertainty involved in editing a poem for which neither faircopy manuscript nor printed text exists, but argues for that uncertainty as an enabling factor. Embracing Jerome McGann's view of the editorial act as always a collaborative venture between editor and textual witnesses, a venture that necessarily produces new literary meanings, Wu forges ahead in "a new variation on the Barthean and Derridian jeu," in which "textual aporia may be the occasion not for deconstruction, but reconstruction" (21). "It hardly matters," he writes, "whether Wordsworth completed the poem or not":
The justification for reconstructing it and analysing it is simple: for six weeks in early 1804, the poet conceived of it as representing The Prelude in its ultimate form. That is to say, its structure and contents had an imaginative reality for him during that time. For that reason alone, it is vital to our understanding of the poem's evolution. A text, imperfect as it must be, is badly needed.20
Among the many questions that such a view raises is the very practical one of the need for a full text of a hypothesized poem, the "structure and contents" of which has been known to students of Wordsworth for twenty years. The reconstruction presented here will be familiar in broad outline and even in many specifics to anyone who has read Jonathan Wordsworth's work (both in 1977 and in The Borders of Vision) or Kenneth Johnston's critical treatment of the five-book poem in Wordsworth and the Recluse. (Wu reprints excerpts from The Borders of Vision and from Johnston's book in a useful selection of critical commentary.) The first three books of Wu's reconstruction are virtually identical to their counterparts in 1805, and, of the 1,113 lines that comprise the reconstructed books IV and V, only about half (570) are edited from MSS. W or WW; the remainder comes from manuscript work toward other and better known versions of the poem. Even before Reed published transcriptions of the whole of MSS. W and WW, much of the work therein toward what Wu calls books IV and V of the five-book stage of the poem was available in De Selincourt's Prelude (and Darbishire's revised edition) and in the Norton Prelude.
The case might be made, though Wu does not make it explicitly in his introduction, that it is a solid benefit for scholars to have available an easily readable text of one possible reconstruction of the poem, as a guide through the textual complexity of the five-book stage. It is, after all, a hard task for a reader of Jonathan Wordsworth's arguments even to piece together the complicated text his article describes, much less to experience the reconstructed poem as a poem, to follow its development and transitions, or grasp its complex tonalities. And Wu's text is indeed useful in this regard. Even in the first three books of his reconstruction—those books that are the closest to more familiar versions of the poem—Wu's text provides some shocks of mild surprise. Because he is committed to recovery of a text that stands as close as possible to the moment that the five-book scheme was abandoned, Wu edits books I-III (toward which no work survives in the five-book Prelude manuscripts) from MS. M, the manuscript sent to Coleridge by March 18, 1804. The "shapes and phantoms" to which Wordsworth hoped to give poetic life are described in 1805 as "floating loose so long"; in MS. M, Wordsworth says he "long had marshalled forth" those phantoms (I, 129). It is a "naked infant," not a "Savage," that stands "in the thunder shower" (I, 302); the playing cards that figure in the mock-heroic battle in Book I, we learn, were soiled through use—"All furnished out in chimney-sweeper garb" (I, 553); and the "roofless walls" of Furness Abbey impress the speaker with their "inward breathings" (II, 127), not their "respirations." In two or three places, printing from MS. M allows Wu to present details unavailable in any other version of the poem. Only the MS. M version includes reference to a Cambridge friend, "unsettled in the heart by cozenage / Of new affections" (III, 206-209), to whom Wordsworth says he paid visits. And only in MS. M do we get an extended attack on the liturgical practice to which Wordsworth was exposed at chapel (as compared to his attack on compulsory attendance itself), in a passage that condemns the "Dull thoughtless mummery, that brings disgrace / On the plain steeples of the English church" (III, 419-20). All of these variants are listed in an apparatus criticus in Reed's edition (I, 1253-76), but it is convenient to have them presented here in context.
Wu's commitment to MS. M readings as temporally closer to the period during which a five-book poem was envisioned than are readings from Prelude MSS. A or B, however, at times produces unfortunate results and undercuts his text's usefulness as a reading text. A good example of this is the MS. M passage in which Wordsworth defends his "sympathies" with Nature's power, which sometimes showed themselves "By outward gestures, and by visible looks" that "some called . . . madness." "Such it was," the passage continues, "if prophesy be madness":
III, 150-54; emphasis added
if things viewed
By poets of old time, and, higher up,
By the first men, earth's first inhabitants,
May in those tutored days no more be seen
With undisordered sight.
The MS. M reading, "those," obscures the obvious contrast of past and present, making nonsense of the passage. Wu provides no note explaining the confusion, though surely the reading in 1805 and subsequent versions, "May in these tutored days" is to be preferred. Another kind of confusing reading occurs when Wu reverts to MS. M in Book IV to supply a passage presumably missing from MS. W, including "There was a boy." Wu's text unaccountably divides the first two lines of the well-known opening of that passage—"There was a Boy, ye knew him well, ye Cliffs / And Islands of Winander! many a time," in 1805—into a tetrameter and a hexameter line: "There was a boy—ye knew him well, / Ye cliffs and islands of Winander! Many a time / At evening . . ." (IV, 472-4). The decision not to emend in these cases might have been understandable (had notes been provided) as part of a consistent bias in favor of all MS. M readings if it were not for Wu's willingness elsewhere to be flexible. In one case, he silently introduces a reading that has no manuscript authority before the 1818-1820, MS. C revision, substituting a stylistically and metrically preferable word, "weariness," for MS. M's "weakness" in "restoration came, / Like an intruder knocking at the door / Of unacknowledged weariness" (IV, 26-28). 
Beginning with Book IV, Wu's text begins to look and sound in significant ways different from more familiar versions of the Prelude. The book begins with the return to Hawkshead for the 1788-89 vacation (material that would eventually find its way into 1805 IV, 121-180, 222-246), and incorporates the discussion of books that would eventually form 1805 Book V. One benefit of Wu's text of this book is that it makes possible in a way that a merely narrative or descriptive reconstruction cannot an appreciation of the effect of omissions. It is just as important for understanding the state of the poem in March 1804 that we do not find, for example, the tribute to Ann Tyson or the famous "slow moving boat" simile (1805 IV, 247-64) as it is that we recognize differences in parallel passages or learn of the existence of passages rejected from later versions. There are also some delightful, if minor, variants of later readings in this MS. W work, as when the infant prodigy—a "Stripling" whose brain "Rank growth of propositions overruns" (in 1805), is handled with a measure of jocularity, through a homelier epithet and figure: "With propositions are the younker's brains / Filled to the brim" (IV, 406-7).
Much of the verse that Wu edits from MS. W in his books IV and V, however, is quite rough. The editor's interest in presenting as much material as possible from MS. W at times leads him to present as "Wordsworth" verse that is clearly not up to Wordsworth's own standards for publication, and which in fact is sometimes very difficult, if not impossible, to understand without familiarity with later, more polished, versions. A good example is Book IV, 341-50, in which Wordsworth's speaker muses on the transitory nature of the works of human intellect, the subject that was to become the introduction to the Arab Dream passage of 1805 Book V. Here, first, is the passage in its 1805 version (MS. M is identical):
Tremblings of the heart
It gives, to think that the immortal being
No more shall need such garments; and yet Man,
As long as he shall be the Child of Earth,
Might almost "weep to have" what he may lose,
Nor be himself extinguish'd; but survive
Abject, depress'd, forlorn, disconsolate.
Reed's transcription of the corresponding passage from MS. W 21r will give a sense of the difficulties with which Wu struggles:
Wu's version reads as follows:
Trembling of the heart
It gives to think that the immortal spirit
No more should need such garments; and yet man
Might sorrow for the creative earth,
And man, as long as he is the child of earth,
Might almost weep to know what he may lose,
And yet not die themselves, but still survive,
Mother and son yet living as before—
Helpless, depressed, forlorn, disconsolate,
Abject, and all his glory passed away.
The passage clearly is grammatically deficient: "man . . . / Might almost weep to know . . . /And yet not die themselves"? It is also metrically deficient. "Might sorrow for the creative earth" is hypometric, and the following line, "And man, as long as he is the child of earth" is metrical only if we suppose a contraction of "he is" to "he's." Such a contraction is not only uncharacteristic of Wordsworth's practice, but would disallow emphasis on the word that most requires emphasis in the phrase: "as long as he IS the child."
Instances of grammatical and metrical deficiency occur with some regularity in passages edited from MS. W. In Book V, for example, the passage following the ascent of Mount Snowdon that begins "Such minds are truly from the Deity," and that in 1805 XIII continues, "For they are powers; and hence the highest bliss / That can be known is theirs—the consciousness / Of whom they are, habitually infused / Through every image, and through every thought" appears in Wu's text thus:
For they are powers, and hence the highest bliss
That can be known on earth—in truth, a soul
Growing, and still to grow, a consciousness
Of whom they are, they habitually infused
Through every image and through every being
In the manuscript, the fourth line quoted above reads "In a [ ? ] they habitually infused" (Reed, II, 295). In filling the lacuna ("Of whom they are"), Wu creates a hypermetrical line in which the reading of "infused" as the main verb (instead of an adjective modifying "consciousness," as in 1805) makes for an unexplained and jarring shift in verb tense.
Such passages raise serious doubts about the desirability of printing as Wordsworth's work verse "within striking distance" of completion, when that verse includes obvious deficiencies of a kind that surely would have been supplied in the process of working the material of MS. W into a finished version (and which were in most cases supplied within days of the decision to recast the poem). In his introduction, Wu embraces a notion of authorship as a collaboration of author and editor, which is justified as a kind of continuation of the collaborative work among author, amanuenses, printers, proofreaders, friends, and critics that originally produced the body of work we call "Wordsworth." The passages that Wu prints as Book IV, 341-50, and Book V, 86-91 are deficient in ways that that collaborative entity never would have countenanced. What Stephen Gill fittingly calls the Dove Cottage "poetry factory" had strict notions of quality control when it came to grammatical and metrical correctness.  Wu's desire to include as much unique copy as possible from MSS. W and WW is perfectly understandable, given his goal of providing a poem marketable as a genuinely distinct version of The Prelude. Where that goal leads him to print technically faulty verse as "Wordsworth," however, it tends to undercut his own editorial theory, ironically bringing into textual being a "Wordsworth" detached from the contexts within which we know his texts were produced. On a more practical level, where it produces passages unintelligible without knowledge of later versions of the poem, it undermines this book's claim to be a self-sufficient version of the Prelude.
None of these difficulties would have been particularly objectionable had Wu claimed for his work the sort of modest ends in relation to which I have praised the book. As a guide through the complexities of one possible reconstruction of the manuscripts work of March 1804, Wu's work is quite useful. For all of Wu's editorial playfulness and open-endedness, however, his introduction (and Blackwell's marketing strategy) tends to overstate the authority of his text and to blur the provisional nature of his work: the "poem" that is "badly needed" by students of Wordsworth is, in fact, this poem. He would have been on firmer footing if he had published his book before 1992. Before the publication of Reed's Cornell volumes, scholars had challenged Jonathan Wordsworth's view of the work of March 1804, but only negatively. Robin Jarvis, for example, argued in 1981 that "there was never a completed version of the Prelude in five books and we cannot now manufacture one," but did not present a fully developed alternative reconstruction of the state of the poem at the moment of its abandonment, as Reed has done.  In his introduction, Wu dismisses Jarvis with relative ease, showing cleverly how his putative position of doubt is ironically grounded in an unstated faith in "the existence of a controlling, auctorial intelligence" (17). Wu does not refute Reed's arguments that the "foes" passage marks the point by which MS. W work became work toward the longer Prelude, even though he makes much of that very passage in his own reading of the significance of his reconstruction (10-11). Nor does he acknowledge Reed's arguments for a competing view of the organization of Books III and IV. Such omissions give the appearance that there is much greater scholarly agreement than there is on these issues and ironically work against Wu's own stated interest in bringing to light as many reconstructions as the manuscripts will allow.
Wu's Five-Book Prelude is being marketed as "the great work of Wordsworth's poetic maturity" (cover). The advertising copy claims for the poem advantages that make it appear tailor-made for the beleaguered teacher, torn between resorting to excerpts and listening to students groan under the weight of some 8,500 blank-verse lines in the 1805 or 1850 versions. The five-book poem, it says, contains most of the best bits of the long Prelude (and much more "great poetry than the 1799 two-part poem") in a "less discursive" whole. "[I]t is likely," the copy continues, "to become for many the 'Prelude' of choice." It would be unfortunate were that to be the case, given the absence of anything approaching scholarly consensus on the textual issues that underlie the project. A more salutary result would be if the book were to guide students in a fresh examination of the challenging, difficult, and exciting manuscripts themselves, now conveniently available in the Cornell series.
Letter to Francis Wrangham, in The Letters of William and Dorothy Wordsworth: The Early Years, 1787-1805, ed. Ernest de Selincourt (2nd ed., rev. Chester L. Shaver; Oxford, 1967) 436.
Letter to William Sotheby, in The Letters of William and Dorothy Wordsworth: The Early Years 456.
JEGP 76 (1977): 1-25; the quotation appears on p. 24.
Jonathan Wordsworth, M. H. Abrams, and Stephen Gill, eds. The Prelude, 1799, 1805, 1850 (New York: Norton, 1979). In a more recent discussion of the manuscript evidence of 1804 work, Jonathan Wordsworth hints that the textual witnesses may be rather closer to printable form than he argued for in either 1977 or 1979. A five-book poem "may or may not have been complete (we are talking about a few lines of transition, lacking in the fair copy of MS. W)," he writes. He includes the 1804 version among the "seventeen distinct versions" of the poem that "can be detected in the manuscripts" and says that "[o]ne of these days, I shall reconstruct the exciting (and at least nearly completed) five-book version of March 1804." "Revision as Making: The Prelude and Its Peers," in Robert Brinkley and Keith Hanley, eds. Romantic Revisions (Cambridge, 1992) 20-21.
Mark L. Reed, The Thirteen-Book Prelude, by William Wordsworth (Ithaca, New York: Cornell University Press, 1992) I, 37-38.
MSS. M, A and B all read "unacknowledged weakness." See Reed, I, 154 for the argument that "weariness" was the intended reading from the beginning, even if it was not entered into a Prelude manuscript before 1818.
Stephen Gill, William Wordsworth: A Life (Oxford: Clarendon, 1989) 231.
"The Five Book Prelude Reconsidered," JEGP 80 (1981): 528-51; the quotation appears on p. 550. Jarvis argues that "the state of the manuscripts is such as to throw into question any attempt to form a complete picture of the five-Book poem, which ceased to present itself as either practical or desirable while it was still in progress" (534).