Six strands of theory bear on the method, each answering a question a writer asks in revision. Like every account of structure in this book, they describe conventions of a period as well as properties of narrative, and the last strand says how far the present suspicion of subplots is such a convention.
Why does a second story cost the reader anything?
Because the reader keeps a model of the story, and a change of thread changes most of it at once. Rolf Zwaan and Gabriel Radvansky’s review of research on situation models (1998) holds that readers track a few dimensions, among them protagonist, place, time, cause and intention, and that a shift in any of them costs processing time. A crossing between threads usually shifts all of them together. Catherine Emmott’s Narrative Comprehension (1997) supplies the mechanism a writer can use: a scene’s contextual frame, the reader’s running record of who is present and where, is not deleted when the narrative leaves it but held in a recoverable state and recalled on return. A thread is recalled rather than re-read, and recall decays with the interval. Two inferences follow, as tendencies rather than findings. A thread with a distinctive cast, place and register is recalled from fewer cues. And a reader resuming a cold thread usually manages it but resents being asked, so the fault registers as a complaint about pace rather than about clarity. No experiment has measured how long a novel’s secondary thread may lie dormant, and the intervals given in The Mechanism are conventions informed by this research, not derived from it.
Isn’t unity the rule?
It was a rule for tragedy. Aristotle’s Poetics asks that a plot imitate one action whose parts are so connected that removing one disjoints the whole, and calls the episodic plot, in which episodes follow one another without probability or necessity, the worst kind. The same work allows the longer epic to take on episodes that extend it, and the novel is epic in scale; much weak advice about subplots applies tragedy’s rule to a form Aristotle would have classed with epic. E. M. Forster’s Aspects of the Novel (1927) gives the oldest defence: a novel may be held together by pattern and by rhythm, repetition with variation, as well as by cause, so a second chain may be integrated by rhyme rather than by consequence. Forster also gives the caution, that pattern pursued rigorously shuts out life.
How can two plots that never meet belong to one book?
Richard Levin’s The Multiple Plot in English Renaissance Drama (1971) answers from a theatre in which double and triple actions were the norm. Levin names three relations between actions. In direct linkage the secondary action changes what is possible in the primary one. In analogy it repeats the primary action’s pattern in another key, at another social level or among other people. In contrast (Levin), so named here to keep it apart from The Viewpoint Shift’s viewpoint domain (Chapter 13), it inverts the primary action: the same pressure, the opposite result. Levin’s central argument is that analogy can integrate a work whose actions never meet, because the relation is the work’s argument; and relations change, so that when an analogy becomes a causal link the change has a name, convergence.
William Empson’s essay “Double Plots” (1935) is the strongest defence of the second chain. Set two actions side by side and each comments on the other, so that the juxtaposition produces a meaning that belongs to neither alone; the effect may be felt without being statable. Its weakness is that an ingenious reader can find such a comment between almost any two things, so a writer who defends a thread this way should be able to say where the comment is produced. The transfer from stage to page also costs something. A play is seen in one sitting, and an analogy perceptible across twenty minutes may be lost across eighty pages, so a novelist must mark analogies a dramatist could leave implicit.
What makes a thread a subplot, and where does it join?
Seymour Chatman’s Story and Discourse (1978), developing Roland Barthes, divides events into kernels, moments that open an alternative and cannot be deleted without breaking the story’s logic, and satellites, which elaborate and can be deleted at a cost only to texture. Applied to threads, the distinction gives the chapter its definition in working form: a subplot is a chain of events that are kernels in its own system and satellites in the main plot’s, until convergence, the point at which its kernels become the main plot’s kernels too. Convergence is literal when the chains meet in event and thematic when they never meet and comment on one another. Before that point, deleting the thread thins the main story; after it, deletion breaks it. The question this yields, where would deleting this thread break the main plot?, finds convergence where it is rather than where the writer believes it to be. Chatman concedes that the line between kernel and satellite is often arguable, so the instrument diagnoses rather than decides.
Who gets a plot?
Alex Woloch’s The One vs. the Many (2003) treats a novel’s characters as a character-system in which attention given to one is withheld from another, and The Viewpoint Shift (Chapter 13) uses it for the allocation of viewpoint. Read from the plot side, the same arrangement is a plot-system: some characters are granted goals, obstacles and reversals, and others only appearances. A plot is the stronger grant. Viewpoint confers access and a plot confers consequence, and a character can have either without the other: Dov is entered only once, near the climax, and has a complete chain throughout. The distribution of chains also argues something. A novel in which the servants appear constantly and never want anything makes a claim about servants, whether or not its author intended one.
Is the suspicion of subplots a law?
Largely it is a product of book length. The nineteenth-century English multi-plot novel was underwritten by publication in monthly parts, which rewarded alternation between threads, and by circulating libraries that favoured the three-volume novel. Bleak House came out in monthly numbers from 1852 to 1853, and Middlemarch in eight parts in 1871 and 1872. The three-volume format collapsed in the 1890s. A literary novel today commonly runs to between 80,000 and 110,000 words, a quarter to a third of the length of Middlemarch, so a subplot taking a fifth of it is some 16,000 to 22,000 words, a heavier cost than the same share was for Dickens. The braided form survived where length survived: epic fantasy, the family saga and long-form television, whose season reinvented the monthly number. The present suspicion is best learned as a convention of the current market, applied with full force to a book with one investment and set aside for a book built on the relation between several.
See also: the Reference theory, with the Victorian publishing economy in a table, Levin’s three-level play, and the case for and against the attention tax at full strength.