In David Morley's review of Mimi Khalvati's new collection he quotes Theodore Roethke:
Form is not regarded as a neat mould to be filled, but rather as a sieve to catch certain kinds of material.
And though I'd never thought of it like that, and it's more obviously relevant to poetry, I was struck by how true this is. My Dictionary of Literary Terms & Literary Theory defines form in a literary work as ‘its shape and structure and the manner in which it is made.’ In poetry form shows up as the shape on the page and patterns within the sense and sound of the words. In prose fiction it's something a bit harder to spot.
I've had funny looks when I've said that in planning a novel it's thinking about form and structure that I really enjoy. If structure is the detail of which parts of the plot happen when and for how long, form is the bigger shape: how these things interplay with the style and substance of what you're writing. I think it's probably the thing readers are least aware of in a novel. When I'm working on a novel, the things readers notice - characters and events - sort of arrive, some from thin air, others from bouts of what I can best (if rather pretentiously) call guided meditation. They define what the novel is. But for me ‘how does this novel work?’ is the core of the problem that I'm setting myself, and it's solving that problem that is the process of writing a novel. Not, of course, that you know everything (or even all that much) about the characters and what they're going to do and experience. So, as Roethke puts it, it isn't that you decide on a form and then pour into it all those characters and events. What happens is that you need to work out a form - a structure, a style, an interplay of narrators and chapters - which will catch the right things: even things you don't yet know. And this kind of thinking is more active and less meditative than most of the rest of the early thinking. The damn thing has to work: the cogs must mesh, and the gears turn at all their different speeds, the beat of the different plot-engines must make a pleasing counterpoint and the chains of image run smoothly through it all, and all in a system that's no larger or more complicated than necessary.
So the moment when I draw up the big chart, with a row for every chapter and a column for every thread of the novel (which may mean six or seven) and start working out where everything needs to go, is the moment when I know that the novel will happen. And even though I do the whole thing in pencil, and much rubbing-out and re-writing goes on, I do my damnedest to get it right before I start Chapter One. Because though everything in the first draft of a novel is negotiable, every word can be changed and characters, voices and events can be re-written, if you get the structure wrong then re-doing the novel may mean rewriting it totally. Because if you've got the sieve wrong, to go back to Roethke's metaphor, you've been catching the wrong things, and you're going to have to re-weave that sieve, and only then set out to find the right ones.