May I say how much I like this book before I make it clear that I cannot say how much I like this book? Rhetorics of Fantasy, most of which is superbly thought through, is perhaps the first full-length study of the vast fuzzy genre of fantasy to have been written as though the genre exists. I do not except from this assertion Brian Attebery's two singularly sharp full-length studies, The Fantasy Tradition in American Literature: From Irving to Le Guin (1980) and Strategies of Fantasy (1992); or The Encyclopedia of Fantasy (1997), edited by me and John Grant, which is to say effectively written by me, Grant, Mike Ashley, Roz Kaveney, David Langford, Brian Stableford and a few others. Certainly for its authors, the Encyclopedia, which I'm going to have to mention again a few times, was a claim-staking volume; it constituted an assertion that, within very wide Water Margins, there was indeed something not foolishly called Fantasy which could in some not foolish and indeed usable sense be "mapped." Rhetorics of Fantasy takes that mapping for granted, and works from within. It is a combine harvester in the heart of the Land.

So far so good. Farah Mendlesohn makes it clear in her Introduction, which may be the twelve most important pages in the volume, that her "taxonomical" division of fantasy into four categories that can be understood as rhetorics is meant to work heuristically, as a tool to let the light in; though her tone can become a bit commanding at points, Rhetorics was not hammered together into a table of laws. The reader who forgets this will bang her forehead against almost everything Mendlesohn has to utter; the reader who remembers that the book has been woven from strobes not stone will find it an immensely useful opening of the way. So far so good.

My problem with the book comes primarily from its use as "givens", needing no elucidation, of some arguments and terms and models developed in the Encyclopedia: it would be foolish for me to say I think she is wrong to do so, because I do continue to think that the four-part modelling of the grammar of discourse of fantasy expressed in the book (see here; Mendlesohn aptly calls it a "four-note bar"), as well as some less assertive descriptive terms (like Polder or Edifice), still work towards a valid mapping. So I guess my problem is in fact a problem for others: it will have to be for other reviewers of the book to fault her for taking as given arguments which for many critics are not givens at all, and which she is indeed among the first (this may be a generational thing) to use as parts of speech.

More generally, there are a few things I can perhaps mention. Some mannerisms detectable in Rhetorics I find dubious, certain distortions in expositional decorum that other critics (I include myself, and Gary Westfahl, and the very much more senior Harold Bloom) are prone to when hot on the trail: a tendency to define the middle through the extreme; a sweet-tooth for the faux-apodictic witticism; a habit of transforming last paragrah's Unexamined Apodictic into the next paragraph's Given.

But the bulk and essence of Rhetorics of Fantasy is, of course, entirely down to its author. Her four categories—Portal-Quest Fantasy; Immersive Fantasy; Intrusion Fantasy; Liminal Fantasy—neatly quarter the field as a whole, though her ill-hidden dis-ease with Portal/Quest Fantasy does lead her to avoid giving Sword and Sorcery much serious attention. The first three of these categories are fairly close to being obvious at a glance.

Portal/Quest—Mendlesohn's arguments for conflating the two are strong, though a "weaker" assertion that they are cognate (rather than aspects of the same movement) might have been less taxing to grapple with—should be familiar to all of us, as tales whose protagonist pass through a portal from this world to another have been conspicuously central to the field ever since Alice. Mendlesohn's argument, that Portal/Quest tales are (vulgarly) overcontrolled and imperialist, is more convincing than one might wish to believe, though I did find that the chapter on Portal/Quest, dealing though it does with the most obvious of the four categories, was the hardest to cope with.

Over and above the slightly over-stressed conflation of portal and quest, the main problem for readers of the chapter, which is structured chronologically, may be the selection of exemplary texts. It needs some fairly agile footwork, right at the get-go, to get around the reader's not entirely naive presumption that John Bunyan's Pilgrim's Progress (1678, 1684) should primarily be understood as an allegory couched in the form of a dream. Readers like myself—who are in any case uncomfortable with Mendlesohn's lack of interest in (and Brian Stableford's Roundhead disavowal of) the argument that the late 18th century marks the point when the fantastic becomes conscious of itself as fantastic, and therefore begins to take on the form and feel of the fantastic we know today—may find a 17th century text like Bunyan's a bit hard to swallow as an incipit to get into portals with. The next two texts examined at length, George MacDonald's Lilith (1895) and David Lindsay's A Voyage to Arcturus (1920), provide a different challenge, in that they occupy what one might call the far rims of portal, and though Mendlesohn's arguments for including them are tellingly put, it might have been more reassuring had she selected a title at this introductory point—A. Merritt's paradigmatic The Ship of Ishtar (1924 Argosy; 1926) comes to mind—more shaped to the task, and more influential in the chronology of the form. In the end, Mendlesohn gets away with this testing of the far reaches of the form before addressing clearly the default workings of the form itself; but this may be because we do intuitively recognize a Portal when we run into one.

Immersive Fantasy may be roughly understood as describing that central body of texts set in secondary worlds which—once you are inside one—closely resemble the realized or "completed" worlds typical of modern SF, though the marked tendency of their authors to set texts in Urban Fantasy contexts marks the form off from much SF. The texts Mendlesohn chooses here are more easily understood as exemplary—the only work stretching the category to its uttermost being Mervyn Peake's first two Titus Groan books (1946 and 1950), which she is only able (with difficulty) to treat as Immersive by ignoring Titus Alone (1959), a text whose inclusion might have caused her to shift Peake into her extremely sharp and mature final chapter, "The Irregulars: Subverting the Taxonomy", which features a brilliant reading of Steve Cockayne's The Legends of the Land sequence.

Intrusion Fantasy (for historians of taxonomy, this form shares some characteristics with "Supernatural Fiction" as defined in the Encyclopedia of Fantasy) is again exemplary. I found Mendlesohn's reading of Robert Holdstock's Mythago Wood (1984) as Intrusion rather than Portal instantly unconvincing, but sufficiently persuasive, in the end, to live with; some of the other readings are also arduous, and again the play of format against/with example generates a kind of post hoc heuristic frisson, which may be fun, but which tends to strew sparklers in the mind's eye.

The final quartering, the Liminal Fantasy, is extremely difficult to describe, partly because it is precisely based on what the reader complexly perceives in texts constantly equipoisal as regards various ontologies, outcomes, settlements. The understanding of a Liminal Fantasy is gained WYSIWIG: what you see is the case of what you get. Given Mendlesohn's slightly tentative (but arousing) hint that the Liminal Fantasy might underpin/underwrite all other forms of fantasy, I would have almost preferred to describe this category as Literal Fantasy: a term particularly fitting, it might be, to Mendlesohn's most telling exemplar, Hope Mirrlees's Lud-in-the-Mist (1926), where every word means exactly what it says. But labels acquire gravity with use, and Mendlesohn's quartering works fine as an ensemble.

As already claimed, there are a few problems with the selection of texts, too many of which are too little known for readers to make much of Mendlesohn's sometimes hazy synopsizing; and even well-known texts are sometimes blurrily rendered, so that the thread of argument is occasionally lost. Mendlesohn makes clear that Rhetorics of Fantasy is not a survey, and that her exemplary texts have not been chosen to create a conspectus; but her relatively casual selection strategy (one might call it WYSIWIG) has occasioned some inevitable losses. Authors of Sword and Sorcery/Heroic Fantasy are not seen much of—no Merritt (see above), or Sprague de Camp, or Fletcher Pratt, or John Myers Myers, or Poul Anderson. No Slick Fantasy writers are included—no Saki, or John Collier, or Thorne Smith, or Roald Dahl. And it would have been salutary to see how Mendlesohn might have netted Gene Wolfe, who has written fantasies both exemplary of and challenging to all four quarterings of Rhetoric's catcher.

But then the point of Rhetorics is not really its cache of texts, some of which are in any case subject to some fairly strained readings. The point is the comprehensive grasp of the catcher, the joy of grasping the world. The structure of our reading of fantasy will never be the same again.

At least once before in Strange Horizons, I've spoken about the EngLit industry, and the appalling ascription practices sanctioned by the Modern Language Association. As a scholar in the context-dependent literatures of the fantastic—where time and place of publication are of the essence—Mendlesohn properly sidesteps as many MLA barbarisms as she can, though a few creep in. Her primary ascription of texts, whenever possible, cites original date and edition, which is against MLA practice; and there are relatively few errors, though—because the whole is so reponsibly done—these do stick out here in the real world of Strange Horizons (an MLA-trained scholar wouldn't notice them). But in the end the buck for mistakes necessarily rests with the author, though Wesleyan doesn't seem to have given its proofing sufficient emphasis (I know one-person small presses with lower error rates than I found here), and some of them should be mentioned.

But I'll list now only the substantive dating errors I've noted in the checklist, as wrong dates are damaging to anyone's reading of any text of the fantastic; readers of Rhetorics of Fantasy might wish to correct them in their copies: Lady Asquith's Third Ghost Book is 1955 not 1957; Andrew Greig's When They Lay Bare is 1999 not 1992; Shirley Jackson's The Haunting of Hill House is 1959 not 1962 (in the body of the text Mendlesohn gives 1957 and 1959); Yann Martel's Life of Pi is 2001 not 2002; Jeanette Winterson's Oranges Are Not the Only Fruit is 1985 not 1990; and Gary K. Wolfe's Critical Terms for Science Fiction and Fantasy is 1986 not 1996. For a volume whose checklist occupies twelve packed pages, this is good going.

John Clute (jclute@gmail.com) has been writing SF and fantasy criticism since the 1960s, and has been involved in writing encyclopedias since the 1970s. His novel Appleseed (2001) was a New York Times Notable Book in 2002. His most recent book is The Darkening Garden: A Short Lexicon of Horror. He is currently working on a third edition of The Encyclopedia of Science Fiction, and preparing a volume of reviews, Houston Do You Read.