Clayton Rawson wasn't super prolific, but is a pretty big name in locked room mystery circles. On Hoch’s famous list, this novel is identified as the seventh best locked room mystery of all time (although that list is surely pretty creaky by now; it was produced in 1981 and only includes one novel that wasn’t written in English). His novels are all available in ebook format, and this book in paperback at time of writing.
This novel is upfront with its intentions. In the first chapter, we’re told that it recounts a seemingly impossible crime, and that after Merlini solved it, the police inspector had no objections except that “he couldn’t understand why he hadn’t seen it all along.” In the second chapter, the narrator is writing an essay on mystery novels, where he argues that all of the possible variations of suspects and clues have been used, that nobody could create an original detective, and that people should therefore stop writing them. In the third chapter, a scholar of black magic is found dead, strangled in a room that was sealed shut from the inside, lying inside of a pentagram. But how? And whodunnit?
Because it’s upfront, one forgives this novel for things one would criticize others for. The suspects, on the whole, are very flat. But what’s the use of complaining? The book looked you in the eye and told you that it doesn’t care about any of that. Merlini is … fine. He sometimes tells puns and recites limericks? Is that a personality? But again, lots of Great Detectives are basically collections of quirks, and since this novel positions itself as a love-letter to that type of novel, it’s pointless to kvetch. Here’s a seemingly impossible puzzle. Can you solve it or not?
That said, there are pacing issues that it being a puzzle novel doesn’t really excuse. The first half of the book takes place entirely in two rooms, and feels more leaden than miraculous. When we should be marveling over the impossible death, we are instead suffering through a chapter-long summery of the dead man’s occult library, complete with footnotes.
And as for the solution, hm. I do like the central "twist" to the first murder. The second murder is silly, in a way you might or might not be willing to excuse on account of the novel's age. More than anything else though, I ended the novel with the feeling that the killer had done a bunch of nonsense and taken an absolutely hideous risk for no reason. The novel gestures at the idea that creating a locked room would make a murder impossible to prosecute, because you couldn’t explain to the jury how the crime was committed. The problem with presenting that as a motive for an impossible crime (Carr did it in The Ten Teacups as well) is always the same—the writer refuses to commit to the bit and let the police in on who the killer is. It's not true that the killer can't be prosecuted because nobody can explain their trick; the killer can’t be prosecuted because the police don’t know who they are, which is the exact same situation that would have arisen if the killer had walked into the first victim’s room, shot him, and then walked out without all the rigmarole.
Another complaint, something that I see more than I'd like to in this kind of novel. I said that I'm fine with the characters being flat, because the author was trying to do something else, but there's one thing I can't abide, when, after a killer is revealed, the detective makes sweeping statements about their personality in ways that are supposed to make sense of the novel, but the killer has never demonstrated for the reader any of the traits they're supposed to have. That is, Rawson realizes that the killer behaves pretty nonsensically; but, Merlini explains, they're an "egomaniac" so they don't act the way a regular person would act. I maintain that if you claim you're writing fair-play detection, and a character being an egomaniac is a necessary piece of the solution, you have to portray them as such, or at least give some hint that they have that quality.