This book is a labour of love, and nearly perfect in what it chooses to do. The only real questions are about those choices, about the boundaries they draw. John Berlyne makes clear in his Foreword that he's been fascinated by Tim Powers's work ever since he first encountered The Anubis Gates (1983). This book springs from that impulse: to provide people who share his enthusiasm with Powers marginalia they can't find elsewhere. It's clearly had Tim Powers's co-operation—he contributes a number of commentaries, often substantial and interesting—and also includes brief pieces by other authors, largely personal reminiscences of Powers.

The first 170-odd pages are mainly taken up with a Powers bibliography, divided into three sections: novels, short fiction, and material by "William Ashbless," the fictional poet who appears in The Anubis Gates and has been used as a pseudonym by Powers and James Blaylock. Each bibliography entry for a given work starts with its first edition, and then proceeds in chronological order through other editions, including non-English language ones. Many of these are represented by their cover art, nicely reproduced in colour. The bibliographic listing itself is astonishingly accurate—I could find no errors or gaps in it—and will surely be a basic reference for Powers scholars. Berlyne is especially assiduous at noting when states of a particular book are rare, and he clearly has collectors in mind for much of his approach. My one cavil is that it's more concerned with the exterior of the books than the interior: the listing of the collection Strange Itineraries (2005), for instance, really needs a contents list of the stories it contains. Instead, it says that "it collects together the same stories as B7, with the addition of B9/1, B9/2 and B/10" (p. 134). Berlyne's cross-refererencing system makes it perfectly easy to track down what those latter three stories are, with some flicking through pages. But the listing for B7 (the earlier collection Night Moves (2001)) doesn't itself contain a contents list. Elsewhere, variant texts of the books aren't always noted. For instance, Powers's second novel was first published as Epitaph in Rust in a 1976 Laser Books edition substantially altered by the publisher (p. 14). A revised edition, An Epitaph in Rust, appeared in 1989 and was more in line with the author's intentions. Berlyne doesn't say, however, whether the two subsequent editions (in 1997 and 2004) used the 1976 or 1989 text.

The more serious question about the bibliography is whether, simply, a book is the best form for an enterprise of this kind. For a writer like Powers, who's in the middle of his career, this is at best an interim report, a snapshot of a moving wave. New books, and new editions of old books, will keep emerging. Would an online bibliography not be a better way to do the job and keep it up to date? (I'd note that Berlyne has some supplementary material to the book's bibliography on his own website.)

The same objections don't apply to the second half of Powers, which gathers together juvenilia and early material for Powers's novels. The greatest amount of text is on The Anubis Gates, a book where it seems the author went down a number of blind alleys before arriving at its final form. Powers presents both a substantial excerpt from a first pass at this material, "To Serve in Hell" (pp. 223-277) and then a chunk of the "proper" first draft of the novel (pp. 285-304). There's also a full outline of the book and reproduced manuscript pages. Other books get similar treatment, though not at such length. The text is often enlivened with Powers's drawings, which he seems to use to give himself an early visual sense what characters look like. Throughout, Powers is thoughtfully designed, presenting text and illustrations so that each reinforces the other's argument.

A couple of things about Powers's working methods emerge very clearly from all the material in this section. The first is that the main problem he works on before he starts drafting a novel is structure in its broadest sense. Time and again in the early versions, he's asking and answering questions about plot and how it will unwind itself. These structural issues often shade into issues of chronology: there are numerous, and astonishingly detailed, timelines for several of the books here. The most extensive of these is for the trio of novels beginning with Last Call (1992), running to fifteen large pages (pp. 526-540)

The second thing these extracts make clear is that, at least when he's preparing a novel, Powers is very conscious of the influences bearing on him. For instance, with his Cold War spy novel Declare (2000), he originally referred to the central character Hale as Guillam (548)—a character from John Le Carré's books such as Tinker Tailor Soldier Spy (1974). In a typescript page of Declare notes, he reproduces the line "I'm free. Rick is dead." (p. 556), the pivotal emotional moment in Le Carré's A Perfect Spy (1987). Elsewhere, while figuring out the structure of The Stress of Her Regard (1989), he notes, "You know who did this sort of thing and got away with it? George R R Martin, that's who, in Fevre Dream. Remember?" (420)

The last thing to be said is that there is a whole secret history of Powers that's not represented in this book. A whole set of less easily answerable questions than influence or structure crop up periodically and go unanswered. Why does Powers write the books he does, and not others? What informs his peculiar view of history, his desire to infiltrate the fantastic into the gaps between what is known? How does his Catholicism inform his view of "magic"? There are, as I say, hints at answers to these questions but nothing definitive. On the last, for instance, he says, "I've always been impatient with attempts to disassemble Christianity into a sort of tepid philosophy, rather than taking it as the supernatural Grand Unified Theory it claims to be, and that attitude especially annoys me when it shows up in clergymen, whose job is to present the thing whole" (p. 129). On questions of his career, Powers is disarmingly direct about his desire for a mainstream success with Last Call (p. 70). But when China Miéville shows up and contributes an appreciation of The Anubis Gates (pp. 282-4), he's speaking from an entirely different world. He makes the point that the book starts from an axiom that ought to render it undramatic but somehow doesn't. If "the past" is fixed and known by the main characters who travel back in time to it, then how can their choices make any difference? That sort of question—stemming from the basic critical job of standing outside a work and the way the author presents it—is almost entirely absent from Powers. That's not a problem per se, but it's worth being clear about what this book does and doesn't do. Powers is marginalia, not commentary—and, for the moment, a fine and beautiful collection of marginalia.

Graham Sleight lives in London, U.K. He is editor of Foundation, and writes for The New York Review of Science Fiction, Locus, and Science Fiction Studies.