What the Drafts Know: Manuscript Evidence and the Big Theories

The fantasy of the settling draft

Every reader who finishes Infinite Jest wants the same impossible thing: one more document. A page somewhere that says, in Wallace's own hand, who dosed Hal, whether the wraith is real, what happens after Gately wakes on the beach. And the extraordinary thing about this novel is that the document almost exists. There is a real archive — draft typescripts, margin notes, editorial letters, whole scenes that were written and cut and whole chapters that were planned and never written — held at the Harry Ransom Center and read firsthand by a handful of scholars. Steven Moore owns a photocopy of the 1993 draft; David Hering spent two research visits inside the containers; Stephen Burn quotes the manuscript notes; the biographer D.T. Max built his account from the papers and the editor's files. A large body of considered-and-abandoned material sits behind the published book — how large is a question the editing history settles, and this essay leaves the page-count arguments to the editing of Infinite Jest and to the last hands on the text. What matters here is the temptation the archive creates: the hope that somewhere in that abandoned material a draft settles it.

It does not, and learning why is the point. A draft is a record of what Wallace was considering, not a description of what the finished book does. The strongest thing a manuscript can prove is not that an answer is true but that Wallace saw a possibility, held it, and let it go. The archive is spectacular precisely as a museum of roads not taken — and the roads not taken are evidence about the road that was. What follows walks the marquee exhibits, and for each asks the disciplined question: what does it actually establish about the big theories, and what does it only tempt us to believe?

The margin note that names the doser

Start with the exhibit that looks most like a smoking gun. In the Ransom Center notes Burn reproduces, Wallace sketched, in his own planning hand, the mechanism behind Hal's collapse: "Pemulis expelled, puts DMZ in Hal's chew," and, in a fuller variant, "Pemulis, kicked out, administers DMZ in Tylenol-like fashion in food, anonymously" (burn-ij-readers-guide). Here, apparently, is the answer to the most argued question in the book — Hal was dosed with the DMZ, and the agent was Pemulis. Readers who have spent years on this crux feel the pull of it like gravity.

Two disciplines have to be applied at once. The first is Burn's own, stated where he prints the note: these are "artistic possibilities rather than definite plans" (burn-ij-readers-guide). A note is a thing you write to keep an option open, not a thing you write to record a decision. The second discipline is subtler and more damaging to the fantasy: the note does not even name the doser the popular theories name. The famous fan reconstruction has the wraith dosing Hal's toothbrush; the manuscript note has Pemulis dosing his chewing tobacco. The draft points at a different culprit than the theory it is usually enlisted to support — which is exactly what you would expect if the draft is a discarded route rather than a hidden solution.

And the finished book declines both. The archive's own weighing of the DMZ question is that no dose is narrated anywhere in the published text; the reading the evidence most supports is the plainest one — that Hal's Year of Glad state is withdrawal, not a poisoning — while the wraith-toothbrush reconstruction rests on off-page inference and the mold-origin variant stays genuinely open. The question is one the book is built to keep unsettled. We know why it is built that way, because the biographer read the letters: the Year of Glad opening was written late, in 1994, and Max's judgment is that the new frame deliberately piled up the candidate explanations rather than choosing one. "Hints for all three of these plot twists are in the text," and Wallace "was never going to let the reader settle on one" (dtmax-every-love-story). The Pemulis note is real, and it tells us something true and important — that Wallace once considered making the dose literal and assigning it. Then he built an opening whose whole design is to refuse that assignment. The margin note is the option; the published ambiguity is the choice.

The archive holds a second note of the same species, on the book's other great crux. Beside Gately's Demerol craving in an intermediary typescript Wallace wrote "Processing of Demerol in body?" — and the scholar who has made the most of it, Casey Henry, reads it as very nearly settling the relapse question: "an endorsed and hence highly convincing 'strong' hypothesis" that the drug is already in Gately's system, so that the hospital ordeal tests intent rather than fact (henry-epiphanic-structure-ij). It is the strongest margin-note argument in the literature precisely because it has the same shape as the Pemulis note: a planning hand keeping an option open, read years later as a verdict. The same disciplines apply. A question mark in a margin is not a decision, and the published beach scene, like the published opening, withholds exactly what the note would settle. What the note proves is narrower and still remarkable — that here too Wallace weighed making the question answerable, and published a page that doesn't answer it. Henry's own larger argument actually strengthens the pattern this note keeps finding: his draft work shows Wallace cutting the clinical "jamais vu" explanation for Hal's blizzard panic, striking the William James chapter-citations that would have labeled Lenz's sickness, and revising Gately's heroism toward ambiguity — the epiphanies, in Henry's phrase, made "intentionally… unclear" and progressively obscured as the drafts matured.

The wraith and the "I"

The narrator question runs the same course. On the title page of the 1993 draft Moore found a margin note that reads like a schematic of the book's darkest machinery: "no speech--Death[.] See Wraith & Gately" (moore-first-draft-ij) — the DMZ's mutism, death, and the wraith-Gately nexus wired together in Wallace's planning hand before the Year of Glad frame existed. And the wraith was, at the draft stage, far less deniable than it is in print. Where the finished book gives us the haunted, over-determined player Ortho "The Darkness" Stice, the 1993 draft called Stice's nickname "the Wraith" outright, and had Hal address him as "Wraithster" (moore-first-draft-ij). The identification that the published book buries in hint and suggestion was, once, sitting on the surface of the sentence.

There is a first-person thread too, and it was noticed early by the sharpest reader in the room. Pietsch, responding to the 1993 manuscript, saw a voice surfacing through the fragments — "there surfaces through the stories an 'I' who may be the one trying to put everything together" (dtmax-every-love-story) — an independent, pre-publication witness that the narrator ambiguity is not an accident of the book's late assembly but was legible in the material from the start. He was still chasing that "I" a year later; his December 1994 notes on a late chapter still ask, flatly, "who's the I who speaks at the end?" (pietsch-letter-dec-1994). The manuscript, in short, contains exactly the evidence the wraith-narrates-the-book theorists want: an explicit wraith, an explicit first person, an editor who felt both.

And again the finished book steps back from all of it. The published wraith is taken by most readers, and by the archive's weighing, to be literally real — but the further claim, that the dead father is the narrating consciousness of the whole novel, is the strained one, not the settled one; the book's resting state is a plural, unattributable narration that no single "I" governs. The drafts do not close this question. They do something more interesting: they show that Wallace could have made the wraith the plain narrator, that he wrote toward that and had it caught by his editor, and that he then withdrew the identification into the sub-surface where it now lives. Who narrates the published book is a question for the reading of the published text — a different job, and its own essay's. What the drafts add is the knowledge that the ambiguity was manufactured out of something that had once been explicit.

The chapters he did not write

The most literal version of the archive fantasy is the belief that Wallace somewhere wrote down the missing year — the span between Gately on the beach and Hal's collapse that the book pointedly omits. He did not write it. But he outlined it, and the outline is the most poignant object in the whole record. After the manuscript's final line, Moore found sketched plans for chapters that were never drafted: one headed with the contents "Hal and Helen, Hal and Exhibition, Pemulis and DMZ," the next simply "Gately and Joelle," and, attached, an image out of a nightmare — a captive whose captors "sewed his eyes open and Superglued the back of his head to the wall and propped the TP viewer in front of him," with "mirrors all around him so he couldn't look away" (moore-first-draft-ij). There, in a list, are the events readers reconstruct for years: Hal and Helen Steeply, the Exhibition, Pemulis and the DMZ, Gately and Joelle, and the viewing-torture that the published book renders only as dread and dream.

It is overwhelming to read, and it settles nothing about what happens after page 981 — because these are notes for chapters that three more years of work chose not to write. That is the crucial move. The manuscript did not lack an ending; its ending is the ending we have, defended scene by scene against an editor who pushed back on it. What it lacked was the year, and the year was contemplated and declined. A plan that was abandoned is evidence of a withholding, not a defeat of it — the strongest thing the outline proves is that Wallace looked directly at writing the missing year and did not.

The same logic governs the book's abandoned skeleton. In manuscript, Hering reports, the novel was built differently: it "was originally divided into two parts in a similar fashion to The Broom of the System" (hering-dfw-fiction-and-form), with an early draft carrying the handwritten heading "Pt. II open?" and a pair of mirrored Mario-and-Hal night scenes both titled "Nocturne," as if the book were meant to fold along a central seam. The web-like, recursive structure everyone treats as the essence of Infinite Jest was not the starting point. It was arrived at, by throwing away a cleaner architecture. Here the abandonment is the whole finding: the connectedness that the finished book asks its reader to reassemble is a thing Wallace built by demolishing the two-part frame that would have made the reassembly unnecessary.

The people who were other people

The loosest layer of all is names, and it is worth walking because it is where the fantasy of the settling draft finally exhausts itself. In the 1993 draft the cast is half-recognizable: "Orin was 'Cully'"; Avril was "Constance Bilodeau," and — a detail with real consequences — she was "from New Brunswick rather than Quebec"; "Joelle was 'Joelle ver Nooy,'" an Arkansas pipe-fitter's daughter and not yet the most beautiful girl of all time; "Poor Tony was 'Queer Tony'"; and "the Year of the Whopper was the Year of the Twinkie" (moore-first-draft-ij). A character with a different name is barely evidence of anything except that the surface of the book was still liquid. The one place it tightens into an argument is Avril: that her separatist connections trace, in draft, to New Brunswick rather than Quebec shows the whole Quebec apparatus is a later overlay — which bears on the questions about her allegiances without deciding them, and is offered here as texture, not verdict.

Two draft names cut deeper than the rest, because they are Wallace's own. Hering reports that "Two Davids appear in the drafts of Infinite Jest, and both are subsequently erased" (hering-dfw-fiction-and-form). In the earliest version of the professional-conversationalist scene — the one that opened the book for years before the Year of Glad displaced it (scene 005, pp. 27-32) — "Hal is called David." And, later, "In a subsequent draft the character later named as Marlon Bain is called David Foster Wallace" (hering-dfw-fiction-and-form). Two characters carried the author's name into the drafts, and both had it stripped out before print. The autobiographical signal that theorists reach for — Hal as Wallace, Bain as Wallace, the book as coded self-portrait — was not imagined by readers. It was written in, twice, in the plainest possible form, and then deliberately pruned. That the surrogacy survives at all in the finished book, as atmosphere rather than a name tag, is the residue of a choice to hide it.

Even the odder theories have a draft to point to, and the drafts oblige without confirming. On the pages Wallace drafted for the terrorists' initiation, Burn found the note "Wayne's Brother?--Wayne violently anti-AFR" (burn-ij-readers-guide) — a floated sibling and a whole political motive for John Wayne that the published book keeps only as a rumor and a buried kinship clue. It is the pattern in miniature: the question mark is Wallace's. The draft records him wondering, not deciding.

What the drafts know

So the honest yield of the archive is not a set of answers. It is a demonstration that the answers were available and were refused. Wallace considered making the dose literal and naming Pemulis; he considered a wraith you could point to and an "I" you could locate; he outlined the missing year down to its torture image; he built the book in two clean halves and gave his own name to two of its characters. Then, across years of revision, he took the toothbrush out of the narration, sent the wraith under the surface, declined to write the year, broke the two-part frame into a web, and crossed out the Davids. The finished book's ambiguities are not gaps he never got around to filling. They are load-bearing, and we can prove they are load-bearing precisely because the alternatives exist, in his hand, in a box in Austin, marked considered and dropped.

The drafts know the book was chosen. A reader holding the published novel and wishing for the settling draft has the relationship exactly backwards: the draft is not the truer version that the book obscures, it is the rejected version that the book improves upon by refusing. The manuscript record is the best evidence in the whole field of Wallace studies for the one thing his readers find hardest to accept — that the difficulty is deliberate, the holes are designed, and the man who could have told us who dosed Hal, again and again, chose not to. The published book is, as his editor came to call it, the version he wanted; the archive is the enormous, meticulous proof of everything he wanted it not to be.

Sources

• moore-first-draft-ij — Steven Moore's page-by-page collation of the 1993 draft he owns: the title-page "See Wraith & Gately" note, the Stice-as-"the Wraith" naming and "Wraithster," the after-the-last-line outline for the unwritten chapters (Hal and Helen; Pemulis and DMZ; Gately and Joelle; the eyes-sewn-open image), and the roster of draft names (Cully, Constance Bilodeau of New Brunswick, Joelle ver Nooy, Queer Tony, Year of the Twinkie).
• burn-ij-readers-guide — Burn's Reader's Guide (2nd ed.), quoting the Ransom Center manuscript notes directly: the "Pemulis expelled, puts DMZ in Hal's chew" planning notes with Burn's own caution that they are "artistic possibilities rather than definite plans," and the "Wayne's Brother?" AFR-initiation note.
• hering-dfw-fiction-and-form — David Hering's monograph, built on two Ransom Center visits: the two erased "David" characters (Hal-as-David; Bain-as-"David Foster Wallace"), and the abandoned two-part structure ("Pt. II open?"; the mirrored "Nocturne" scenes).
• dtmax-every-love-story — the composition biography built from the archive and Pietsch's files: Pietsch's 1993 sighting of the surfacing "I," and Max's reading that the late-written Year of Glad opening was designed to multiply Hal's explanations so no reader could settle on one.
• iw-pietsch-2016-full — Michael Pietsch's 2016 crowd-sourced interview, in which the editor calls the published book "the version he wanted, not something imposed on him... the Writer's Cut" (the source for "as his editor came to call it" above).
• pietsch-letter-dec-1994 — Pietsch's December 1994 second-read notes, in scans of Wallace's marked copy: the editor still asking, of a late chapter, "who's the I who speaks at the end?"
• henry-epiphanic-structure-ij — Casey Henry's genetic reading of the epiphanic structure, worked directly from the Ransom Center typescripts: the "Processing of Demerol in body?" margin note beside Gately's craving, the deleted "jamais vu" thread, the struck William James chapter-citations, the deleted Gately water dream, the Mario-as-eunuch first draft, and the invented "Catharsis of resolving" added late — the drafts read as Wallace progressively obscuring his epiphanies.