RogerBW's Blog

The mystery story when done well is an extreme form of the story problem, which one can enjoy both on the level of normal fiction with characters and plot and setting and so on and on the level of working out whodunnit before the author reveals the solution.

I'm not a strict follower of the van Dine or Knox rules (particularly as some of them were pretty clearly written as direct criticisms to certain books by Agatha Christie). I'm more inclined to favour the Detection Club approach, a compressed version of the Knox rules, summarised thus by Simon Brett:

that the author pledges himself to play the game with the public and with his fellow-authors. His detectives must detect by their wits, without the help of accident or coincidence; he must not invent impossible death-rays and poisons to produce solutions which no living person could expect; he must write as good English as he can. He must preserve inviolable secrecy concerning his fellow-members' forthcoming plots and titles, and he must give any assistance in his power to members who need advice on technical points.

But in any case, as with rules of grammar and perspective I think the artist is better off knowing what they are, and perhaps breaking them deliberately, than breaking them by accident. Van Dine in particular is very impatient with the rest of the book, seeing descriptive or character-analysing passages as mere wrapping round the meat of the puzzle; I'm more prone to try to enjoy the book both as a puzzle and as a story. (Apart from anything else, this means I can read it again later even if I remember the plot.)

It's unfortunate when one can solve the mystery by the shape of the story: when a key question is raised but never answered and nobody ever chases it up, or when the detective mulls over the credentials of the five suspects while the reader remembers that there were six of them. This wouldn't necessarily be a problem for a new reader, but one does start to notice these patterns after a bit, and it's a shame when writers fall into them.

The puzzle part of the story is traditionally regarded as a triad of means, motive and opportunity:

(T-shirt design from Fo'Paws Productions.)

Means were early gimmicks: the ice dagger or bullet that melts in the victim, the undetectable poison, and so on. This sort of thing has mostly faded away now, along with opportunity gimmicks like the traditional locked-room mysteries (which I think it's fair to say that John Dickson Carr examined so exhaustively as to leave little space for innovation), and those puzzles that involve detailed reading of railway timetables. (I seem to remember at least one Agatha Christie short story that hinges on the complete impossibility of a train running five minutes late.)

What's left is motive, and that's where the puzzle overlaps with the good story: of these people, all of whom physically could have dunnit (and probably all of whom have some sort of reason), which is the one who actually did? At that point all those character-analytical passages come into their own, and if the people are well-enough drawn that will give clues that can be blended with conventional means and opportunity information to come up with the result.

As Dorothy Sayers put into the mouth of Harriet Vane, constructing a detective story in Strong Poison: "She is a person with a monomania—no, no—not a homicidal one. That's dull, and not really fair to the reader." I regard it as a failure mode of a mystery story to say "a loony did it for his own reasons which don't make sense to a rational observer". Yes, clearly murderers are not thinking quite like normal people, or we'd be up to our ears in corpses. But removing both character and motivation (because the loony has to appear something like a reasonable person, so most of what you learn about their motivations has been faked) removes for me the enjoyable part of the puzzle, and throws one back on exactly who was where when.