There was also a time when the pages and pockets of the Book remained in place while they shed words and sentences, leaving a blank space on a page or even a whole page which was blank. This disappearance was quite different in kind from that given by the pulling out of pages and insertions, and the Book did not become any thinner. The whiteness spread rapidly and it seemed possible that soon the reader would have only white pages to leaf through, that he would open pocket after pocket to find nothing but blank pages. As you can see, the Book led a rich existence thanks to the several ways by which it could approach nothingness.
Looking for the beginning
For a time I tried to find the Book’s original layer, the first passage out of which the Book grew by insertion, transcription and erasure. It may be that such an undertaking was within the bounds of realizability, but it proved so difficult that I gave it up. It certainly was impossible to identify the oldest layer by its content; the constant modifying of the text meant that it was far from out of the question that the oldest passages were those which told of the latest technical discoveries, about which the islanders had learned from sailors. To begin with I imagined it would be possible to identify the oldest text by the level of insertion at which it appeared. The fact that the Book grew mainly through insertion encouraged the assumption that the manuscript was like an inverted Troy, where the oldest layers were on the surface (in the Book’s larger foldouts) and those which had been written and stuck in only yesterday were at the deepest level of insertion. The task of a textual archaeologist would be to monitor closely the surface while resisting the temptation to go beneath it.
But suddenly it hit me that this assumption was false. Certainly, the making of insertions was the main activity demanded by the Book, but there was no guarantee that individual passages of text would settle in one place; indeed, passages went up and down in the hierarchy of levels — they might fall sharply before embarking on a steep climb, as if riding a Ferris wheel. It was not uncommon for a passage to disappear and then reappear in modified form as an insertion on the second, fifth or sixth level, to be lost again without trace in the bowels of the Book before bobbing up on the first level in white pages which had lost their text in one of those periods when hatred of words was paramount. The Book was really a cyclical entity, a demented structure whose foundations were on its roof. It was probable that the beginning of the Book had long ago been lost through the regular batterings it took as it was transformed, yet it may have been the case that the beginning remained hidden at the bottom of the deepest pocket — a pocket that was so distant that the motions of constant change could not reach it, which was buried so deep that it would exceed the patience of the most persistent of insertion speleologists.
Once a foreigner (myself, for example) had grown used to the Book’s revolving-wheel nature, he was driven to distraction by the Book’s other whims. The Book constantly violated the seemingly obvious rule that the number of flights taken in its descent should correspond — if the action were to be returned to the original level — to the number of flights taken in its ascent. It might happen, for example, that the hero of Story A was also the narrator of Story B, while in Story B the narrator of Story C made an appearance; but the action would not, as one would expect, return to Level B once Story C was finished — instead it was Story A which continued on the pasted-in strip. It was as if by some dark magic Story B and the world it described had disappeared from the Book entirely; but later, once we had recovered from it, Story B would continue in some wholly unreasonable place, say at Level F or Level G. The hero of a story on Level A wrote the novel Silver Cloud, which was played out on Level B. Some way into the story a character from Level B stepped into a bar on Level A, where he struck up a conversation with the cousin of the author of Silver Cloud. I didn’t know whether such offences against logic occurred by intention or through negligence, but to begin with they so exasperated me that I felt inclined to seek out their authors so I could throttle them. Fortunately the authors of individual parts of the Book were so entirely anonymous, there was no chance of finding out their names. Besides, to ask around after the identity of an individual author was socially unacceptable. The sense of shame it would provoke might have something in common with that known by the island’s kings, who would have liked to remain every bit as anonymous as the authors of the Book.
But this was not all. At certain points the Book would bite itself in the tail, so that it was barely possible to ascertain the level on which a given passage was located. (When counting off the levels, one could start wherever one chose.) Let us say that a character appears in Story A who is the narrator of Story B, while one of the characters in Story B is the narrator of Story C; then one of the characters in Story C begins to narrate not Story D (as one would expect) but Story A. (Escher’s lithograph Print Gallery is folded into itself in a similar way (although it has a simpler A (A) form. (In a gallery a visitor is looking at an art print which portrays a town which contains the gallery in which the visitor is standing. (I would like the reader to consider what this lithograph would look like if — like a story in the Book—it had the form A (B(C(A)))?))).) (If that last sentence was written by one of the authors of the Book it is quite possible, dear reader, that you would count in it three left-hand and thirteen right-hand parentheses.)
Karael and I once sat on the jetty in the lower town discussing the way the Book folded in on itself. We had been bathing in the harbour and had taken the Book along with us. (As I have mentioned, it was not necessary to treat the Book with any great care; smudged letters and marks were taken as part and parcel of the Book’s transformation.) I argued that the form the Book took did not correspond to the arrangement of its contents, complaining that the rule was abused which stated that an insertion at the lowest level should be in an interior pocket; I also suggested how this might be corrected. It would be necessary to cut an opening into a pocket pasted on to Foldout A — the biggest, which contained Text B — and another into the pocket containing Text C which was inserted in the first pocket; a pocket could then be pasted on to the appropriate place on Text C. Where this new pocket was at its narrowest, it would be elongated so as to pass through both openings, only widening out once it had escaped the physical confines of the Book. The new pocket would need to be big enough to ensure that it could be turned back in towards Level A, closing this (and the whole Book along with it) inside itself. It would also be necessary to cut into this pocket an opening by which its beginning — which coiled out from pocket C — would be able to exit.