Poisons were a common method of killing people in golden age detective stories. The two primary ones were arsenic and cyanide. I believe that this was the case primarily because of availability. (It seemed that they were commonly sold as weed killer and insect killer.) I’ve seen more than a few references, however, to people using exotic, undetectable poisons (often from South America) in murder mysteries, though I’ve never seen its actual use in them.
I’ve recently been reading some Miss Marple stories, and while the Miss Marple short stories began in 1927 and the first Miss Marple novel was in 1930, the bulk of the Miss Marple novels were in the 1950s and 1960s. Times had changed, especially with regard to poisons.
I found the description of a poison in The Mirror Crack’d From Side to Side (1962) extremely interesting:
Heather Badcock had died as a result of four grains of hy-ethyl-dexyl-barbo-quinde-lorytate, or, let us be frank, some such name.
This drug turns out to be a (fictional) anti-anxiety medication which goes by the brand name “Calmo.” It is by this name that it is typically referred to throughout the rest of the book.
A similar device was used in A Caribbean Mystery (1964):
“They found he’d had a lethal dose of something that only a doctor could pronounce properly. As far as I remember it sounds vaguely like di-flor, hexagonal-ethylearbenzol. That’s not the right name. But that’s roughly what it sounds like. The police doctor put it that way so that nobody should know, I suppose, what it really was. The stuff’s probably got some quite simple nice easy name like Evipan or Veonal or Easton’s Syrup or something of that kind. This is the official name to baffle laymen with. Anyway, a sizeable dose of it, I gather, would produce death, and the signs would be much the same as those of high blood pressure aggravated by over-indulgence in alcohol on a gay evening.”
I strongly suspect, though I can’t prove, that Mrs. Christie had no real medication in mind, in either case, and I must say that this does greatly simplify the job of the mystery writer. There is the question of whether this violates the fourth commandment of Fr. Knox’s ten commandments of detective fiction, but I think that it does not. The point of the commandment is not whether the poison is known to the reader, but whether it is known to the medical science of the people in the stories. The point is that the big reveal at the end may not be a completely made-up poison, since if it is the solution becomes completely fictional. If the poison, though fictional, is known in the middle of the story, then the reveal at the end will rely on real human things such as motive and opportunity.
This approach does leave off a few issues of verisimilitude, though. One of the great problems with a poison is the dose that is needed to kill. It’s not that hard to make people feel sick, but actually killing—especially in a reasonably short time frame—requires an accurate dose. This will vary considerably with the individual chemical, and the LD50 of a particular drug is not always easy to come across. You can often find information with google on the LD50 (lethal dose for 50% of the population) of medications for rats and mice, but human LD50s are not always available, and the values for rats and mice can vary considerably. This is not just about making the dosage appropriate in the book, for the murderer to use the poison, he probably needs to have some idea of how much he should use in order to feel confident at the attempt.
To pick an example at semi-random (I had to google about five different medications before I found one with human info), Citalopram, sold under the brand name Celexa, has a human LD50 of 56mg/kg. For 110 pound people, this means that a dose of 2,800mg would kill half of the people you gave it to. To put this into perspective, the dosing is usually about 30mg/day, though possibly up to 40mg/day. Let’s assume the source of the pills are 40mg pills, this would mean having to give a 110 pound person 70 pills in order to have a 50/50 shot at killing them. You’re not going to manage putting 70 ground up pills into someone’s coffee. You’re probably going to need something like a stew, but it’s going to have to be one heck of a stew to cover over 70 pills worth of magnesium stearate or whatever chalk-like substance makes up most of the pills. (This, incidentally, is why pills tend to be mostly inert ingredients—it is extremely effective at preventing accidental overdose in significant quantities.)
Now, I doubt that I happened to select the most dangerous drug with human LD50 information on my first few tries, but it would be quite a coincidence if the murderer did, too. This means that trying to kill someone with medicine would be very unlikely to be a spur-of-the-moment thing, and would need to be researched. (To be fair to Mrs. Christie, btw, medicine has mostly gotten safer since the 1960s, so things weren’t quite as bad for murderers in her day.)
I also suspect that, if one were doing a properly researched murder, it would be a better idea to try to play off of drug interactions. There are a lot of combinations of drugs that are far more dangerous than simply a larger dose of the one, and of course many of these also interact with alcohol which is a lot easier to get into somebody’s blood in large quantities than most medicines are, since the victim might well put the alcohol there on purpose. Though it might be more effective, this approach is, perhaps, even less of a spur-of-the-moment weapon than a simple overdose is.
There is also a curious problem on the other end of the murder mystery—identifying the stuff in a corpse. Chemicals are identified by tests with reagents, which means that they must be specifically tested for. The police lab must, therefore, have some reason to test for the medicine in order to find it. Merely having the police report that a high level of an unpronounceable poison was found is cheating. And that is, of course, supposing that the medicine would even still be in the blood to test for. All sorts of chemicals break down in the body over the course of a few hours, many medicines among them, and cannot be found even if you know to test for them unless you run the test immediately. I suppose that this later part can be hand-waved away by simply not giving the made-up medicine this property, but there’s no real way around the test needing to be specific.
Ultimately, I think that this approach to medicine-as-poisons (just making them up) is fine, but doing it right would be so much work that it probably would be easier to use a real medicine. I think better than medicines, though, are recreational drugs. LD50 information tends to be readily available on these, and owing to being illegal they are generally available (to the degree that they are available) in pure forms quite in excess of a lethal dose. Further, since they are illegal, anonymous procurement does not greatly stretch the imagination.
Of course, there would be no harm in mixing recreational drugs with other drugs for synergistic effect. As long as it was planned well ahead of time.