Though I’m only about halfway through writing Wedding Flower Will Do for a Funeral (the second Chronicle of Brother Thomas), it’s good to put some thought into when the next book will be—murder mystery plots are the sort of thing it’s nice to kick around for a while instead of having to come up with in a moment. Accordingly, I bought a book on poisons for mystery writers. I hope to review when I’m done with it, but I wanted to talk about the subject of poison as the murder weapon.

(I should note that this is thinking out a variety of alternatives, and is not aiming at any single conclusion; it’s more like a walk through a workshop than an essay.)

Poisons kill, but they do so in a variety of ways. Some kill quickly, others slowly. Some kill very painfully and some just put the victim to sleep and a short time later into eternal sleep. And as I was reading over descriptions of the effects of various poisons it occurred to me that some poisons have greater differences between them than there are between some of those poisons and more conventional weapons such as knives and bullets.

One of the great differences in poisons is a question of detection. That is, how hard is it to discover that the victim was poisoned? Poisons which cause the victim to writhe in agonizing muscle spasms for days before finally killing them, for example, are not likely to be mistaken for death by natural causes. So why use such a poison?

(Before answering that question I should note, in passing, that these tend not to be popular poisons in television mysteries because they don’t give the opportunity for the detective to spot the clues which indicate poison that most people have missed. If the detective confidentially whispers to the police that a victim taken suddenly ill right after dinner and who thrashes about for several days before finally was probably poisoned, he’d be liked to get a sarcastic, “How did you work that out, then?” if it’s a British show or, “Thank you captain obvious” if it’s an American show. Good television, this does not make.)

I think that the best reason to use obvious poisons—except in the case of pure malice, that is, to want to see the victim suffer—is in order to frame somebody. The big problem that murder mysteries have is that of motive. Cui bono? Whose good? Who is it who benefits so much from someone’s death that they’d commit it in cold blood (and murder by poison almost certainly has to be in cold blood since the poison must be procured beforehand). There are generally only a few people who will benefit to any great degree from the death of a person; this narrows the field of suspects down quite considerably. Good for the detective, not nearly so good for the murderer. (And to actually go through with a plan for murder, one must expect to get away with it.)

There is still a problem with the frame-up: if the case against the person being framed isn’t air-tight the field of suspects will become very small indeed. This can certainly be made to work, but poisons introduce the problem that the murderer doesn’t have to be present when the victim takes the poison. While convenient, it renders alibis useless. (This can to some degree be worked around by contriving to make it seem like the time the victim took the poison was known.)

If the murderer is not trying to frame one of the few other people who will benefit from the death of the victim, an obvious murder which does not readily admit of an alibi seems very unlikely to appear a good idea. I suspect, then, that this is probably best used in revenge killings, and in particular those where the relationship between the killer and the victim is not generally known. In English cozies this is the classic case of the killer being the grandson of someone who the victim murdered forty years ago in Australia.

This can be done extremely well; I think most of the interest is going to lie in establishing the backstory and solving a 40 year old mystery in order to unravel the present mystery.

The other sort of poisoning—the gentle kind—results, I think, in a very different sort of murder investigation. Probably the most notable aspect of this is going to be the overturning of an initial conclusion that the victim died from natural causes. The most classic example of this is, I think, the elderly rich relative.

In many stories the climax of the investigation is the digging up of the body and testing it for poison, which is then found. There’s nothing wrong with that plot, but things get very hard if a monkey wrench is thrown into it. The obvious monkey wrench is the undetectable poison—and there are a few—but it’s interesting to consider the approach that Dorothy L. Sayers used in Unnatural Death. It suffered from the minor problem that the effect she relied on was exaggerated about 100-fold; as one reviewer put it the method would work but the apparatus used would be comically large. But that aside, since a poison wasn’t used none could be found. And the rest of the story tells us, I think, how stories about undetectable poisons have to go.

If the first murder was undetectable, the only real solutions is for there to be more murders, this time imperfect. The murderer had ample time and opportunity to plot the first murder, but latter ones will either be rushed or the murderer will relax because of the overconfidence created by success.

The murderer can be pushed into subsequent, rushed murders either by the detective—who seems to be getting too close—or by someone who witnessed an incriminating part of the murder and is now blackmailing the murderer. (It’s convenient for detectives how few fictional people realize that blackmailing a murderer is a very dangerous way to make money.)

In the former case, this can be done by way of the murderer having an unwitting accomplice—somebody who didn’t understand the significance of an action they knew the murderer did or may have even done at the murderer’s request. The impetus comes when the detective is starting to ask questions which might make the unwitting accomplice realize the significance of what they know. The tricky part about this is that the detective can’t do this on purpose or he’s guilty of the unwitting accomplices’s death. It’s not easy to pull this off even unintentionally, though, since the brilliant detective should—because of his brilliance—foresee the probable outcome of asking the questions he’s asking.

All things considered, I think the cleaner way is for the second victim to blackmail the murderer. The downside is that the detective is thus being handed a piece of luck outside of his control, which isn’t satisfying. On the other hand, this is true of (basically) all possible clues. The murderer’s bad luck is the detective’s good luck. If the murderer committed the perfect murder, the detective couldn’t solve it.

On the one hand, this feels like a cheat. On the other hand, it is appropriate; to murder is imperfect and imperfect people do not do things perfectly. Murder is a sort of short-cut, and people who take one short-cut will take others, too. The real trick is to keep the sort of short-cuts taken that help the detective in-character with the murder itself.