The Two Kinds of Evidence in Murder Mysteries

In murder mysteries, there are two kinds of evidence: evidence which tells the detective what happened, and evidence which can get a practical result from society. The practical result is often a criminal conviction, but it need not be; a wedding being called off, the payment of an insurance policy, or the settling of a will all require similar sorts of evidence.

Of the two, it is the former type of evidence, not the latter type, which is of interest to the reader.

The main distinction between the two types of evidence is not really one of the strength of the evidence, that is, of the level of certainty which it conveys. In fact, one of the common features of murder mysteries is the early presence of highly convincing evidence which will convict an innocent person unless the detective uncovers the truth. No, the distinction is not in certainty. The distinction is, rather, what is required knowledge and understanding is required to apprehend the true meaning of the evidence.

Convenient names for the sorts of evidence of which we are speaking might be complex evidence and simple evidence. Complex evidence requires extensive background knowledge and understanding of human nature. Simple evidence does not; it tells its story plainly. (Using this terminology, we can say that it is common for murder mysteries to, early on, have complex evidence which appears to be simple evidence.)

In order to achieve societal action, such as convicting the murderer in a court of law or getting some other legal effect, one must have simple evidence. However, simple evidence is, in murder mysteries, hard to come by. This is, of course, a selective effect. In the case where the murderer’s fingerprints are on the murder weapon, and the murderer was seen killing the victim by multiple witnesses who know the murderer personally, and the murderer was caught immediately afterwards—these are not the stuff of murder mysteries.

Detective stories have the structure of story-within-a-story: the murder is the interior story while its detection is the outer story. Within the outer story, it is frequently the detective’s main purpose in his investigation to try to uncover simple evidence about the inner story. This makes it curious that his success or failure at achieving this goal is (almost) irrelevant to whether the story is a good story.

An excellent example of this is the Poirot story Five Little Pigs. In it, the daughter of a woman who was hanged for murdering her father, seventeen years ago, comes to Poirot asking him to uncover the truth. She just received a letter from her mother, written immediately prior to her execution but entrusted by lawyers to be delivered on her daughter’s 25th birthday, telling her that her mother was innocent.

Poirot undertakes the investigation and interviews all of the people principally concerned. At the end, he explains how all of the evidence which had pointed to the guilt of the woman’s mother actually pointed to the guilt of someone else. That person speaks alone with Poirot, afterwards, and asks him what he intends to do. Poirot says that he will give his conclusions to the authorities, but that it is unlikely that they will pursue it and very unlikely that they will get a conviction. And that’s fine. It’s a very satisfying ending to the story.

But why?

I suspect that the answer (which may be obvious) is that complex evidence is fun, while simple evidence is not fun. It takes brainwork to understand complex evidence, while simple evidence is too easy to be interesting. What matters in a murder mystery is being interesting, not achieving results. Achieving results is, really, the domain of an action story, or possibly a drama. This is why detectives tend to hand their cases off to the police at the end of the story. It’s best if the tedious work happens off-screen.

But there’s an interesting complication to this.

It is not a good story if the detective (and hence the reader) merely finds out what happens without anyone else learning it. Why this is so relates to the detective’s role within the story. As I’ve said before, the detective is Christ figure: the world has been corrupted by the misuse of reason, and the detective enters it in order to restore order to the world through the proper use of reason. So while society need not act, something must be put right. That is, someone beside the detective must learn the truth and be better off for it.

A good example of this is the Sherlock Holmes story The Blue Carbuncle*. It begins with a curious set of coincidences which place the key evidence in front of Sherlock Holmes, and with some investigation he discovers who it was who stole the gemtsone. He invites the man to his room, and he comes. After Holmes confronts him with the evidence, he falls apart and confesses, sobbing. I’ll quote just the last part:

“Get out!” said he.

“What, sir! Oh, Heaven bless you!”

“No more words. Get out!”

And no more words were needed. There was a rush, a clatter upon the stairs, the bang of a door, and the crisp rattle of running footfalls from the street.

“After all, Watson,” said Holmes, reaching up his hand for his clay pipe, “I am not retained by the police to supply their deficiencies. If Horner were in danger it would be another thing; but this fellow will not appear against him, and the case must collapse. I suppose that I am commuting a felony, but it is just possible that I am saving a soul. This fellow will not go wrong again; he is too terribly frightened. Send him to jail now, and you make him a jail-bird for life. Besides, it is the season of forgiveness. Chance has put in our way a most singular and whimsical problem, and its solution is its own reward. If you will have the goodness to touch the bell, Doctor, we will begin another investigation, in which, also a bird will be the chief feature.”

Here there are two aspects to the world which Sherlock Holmes has put right, even though he has produced no evidence for a jury. In the first, the man wrongly accused of the crime will not be convicted of it because the principle witness against him has fled. The second is less certain, but the true criminal may well repent of his crime since he’s seen what evil he’s capable of and still has a chance to make his way in society honestly.

This satisfies the role of the detective as Christ figure. In fact, it even has a curious echo (perhaps intentionally) of the story of Christ and the woman caught in adultery, and how he releases her from the punishment for her crime on the condition that she sins no more. Neither is, strictly speaking, a satisfying story, but they have something else to them—the idea that there is something better than justice. That’s a very tricky notion, because mercy should never be unjust—but at least in the story of the Blue Carbuncle, what was stolen is returned, and so justice is at least mostly satisfied in restitution.

Be that as it may, the primary point under discussion is satisfied. Holmes collects complex evidence which tells him (and thus the reader) the tale, and this is the interesting part. Achieving a practical effect from society is of minor concern.

(I suspect that part of the reason why The Adventure of the Blue Carbuncle ends as it does is that, though it predates Fr. Knox’s decalogue, it violates rule #6 (“no accident must ever help the detective, nor must he ever have an unaccountable intuition which proves to be right”). The Adventure of the Blue Carbuncle is predicated upon a series of accidents, all of which help the detective. If he achieved a practical societal effect, his reputation would benefit by pure chance. By letting the criminal go, he remains in anonymity and so the accidents which help him produce only an interesting set of circumstances.)


*A carbuncle is a red gemstone, most often a garnet, so a blue carbuncle is something of a contradiction in terms. The story suggests it is a blue diamond, though it could be a blue garnet or even a blue sapphire.

Acquiring Murder Weapons

Much of what drives the plot of a murder mystery are the problems with a murderer who wishes to avoid capture has to solve. In no particular order, the big ones are:

  • Finding an opportunity to commit the murder
  • Planning how to commit the murder
  • Acquiring the tools necessary for the murder
  • Committing the murder without (living) witnesses
  • If possible, having an alibi for the time of the murder
  • Disposing of the murder weapon
  • Disposing of the corpse (I’m including making it look like accident/suicide here)

(Some of these problems are really alternatives to each other; for example if one has an alibi for the time of the murder, one doesn’t need to bother with disposing of the body. If one can acquire the murder weapon in an untraceable way, one can leave it next to the corpse with a label “murder weapon” helpfully attached to it. Etc.)

In writing a murder mystery, these are the questions the writer needs to answer and—at least the way I work—before writing the actual mystery. Actually, let me back up for a second and explain by way of my theory of murder mysteries:

A murder mysteries is actually two stories: a drama told backwards, and a detective story told forwards.

I think it’s fine to write a murder mystery by the seat of one’s pants as long as it’s only the detective story one is “pantsing”. If you’ve written the drama (the murderer’s story) ahead of time, you can’t get into too much of a mess with the detective story. Where people go wrong is in not having first figured out the drama in its natural order so that they can gradually tell it backwards. (NOTE: there will be people who can successfully write murder mysteries in a completely different way. I’m not trying to lay down laws everyone must follow.)

That’s what I mean by needing to figure out the murderer’s solution to his intrinsic problems first. Once you’ve figured out how the murderer has planned his murder, you can then work out how the murder actually happened, at which point you now have a coherent story for the detective to detect. I find this order of doing things very helpful for two main reasons:

  1. This mirrors reality; it’s the order in which murders actually do happen. This means that there’s precedent for it being a workable system.
  2. One has complete freedom for the first decisions one makes. And it’s in the murder itself where plot holds are the most damaging to a murder mystery. Thus starting here gives one the fewest temptations to try to hide a plot hole in order to preserve what’s already been done.

This also lets you decide ahead of time, in a coherent way, what mistakes the murderer made (i.e. what evidence they left), so that you know where to direct your detective to look.

Types of Murder Weapons

So, the question I want to consider is how the murderer acquires the murder weapon. There are several categories which each have their advantages and their shortcomings:

  1. Ready to hand
  2. Store-bought
  3. Home-made

A ready-to-hand murder weapon, such as a kitchen knife used to murder someone in the kitchen, has the benefit of being untraceable, since there is an obvious explanation for what the kitchen knife was doing there. The downside is that such things are unreliable and so unlikely to be in a planned murder. (Of course, one can always abscond with the knife beforehand then bring it so that it looks like it was snatched up in the heat of the moment.) Ready-to-hand weapons are also very likely to be simple—knives, pokers, hammers, etc. People very rarely leave loaded guns, crossbows and bolts, etc. lying about. This means that ready-to-hand weapons will require one to get very close to the victim and use a great deal of physical force. This eliminates squeamish murderers. And most modern people are pretty squeamish.

A store-bought murder weapon is likely to be in good working order and quite possibly useful for killing at a distance and without much force. The downside is that they are—at least in theory—traceable. Guns, crossbows, etc. have serial numbers. Shop keepers have memories and many big box stores have comprehensive video surveillance. Acquiring these things in an untraceable way can be done, but it takes far more work and premeditation, since the best bet here will be to buy them on the secondary market, in cash, months or better yet years before the murder. This requires a very patient—or lucky—murderer.

Home-made murder weapons offer a good compromise between ready-to-hand and store-bought. Building supplies like plywood, hinges, rubber bands, and so on are basically untraceable. And there are a ton of very deadly weapons one could very realistically make. If you don’t believe me, just watch a few episodes of The Slingshot Channel. The downside here is that one requires a fairly competent murderer with at least a few tools. The problem this introduces is one of personality: people who are patient and good at problem solving just don’t seem like the murdering sort. However much of a problem the victim is, there’s probably also a non-lethal way around the problem they pose.

Accomplices Make Everything Easier

One practical way around most of the problems brought up in the murder weapon section is an accomplice. Even better for the problem of murder weapon acquisition is the anonymous accomplice. As long as the murder weapon isn’t directly traceable to a name (i.e. isn’t a new gun/crossbow recently bought at a gun/bow shop), this will greatly obfuscate the trail of the weapon. If the detective doesn’t know whose footsteps to retrace, he will have a very hard time retracing them.

An anonymous accomplice can also go a long way in solving the problems of personality introduced by the choice of a murder weapon. Two people can be more patient than one, an accomplice who wouldn’t murder anyone himself might still help a lover or friend to buy or make a weapon, and so on.

(In fact, the ability for two people to have two personalities is sometimes revealed in just this way—the detective is talking about the contradictions in the murderer’s behavior and someone says, “It’s like our murderer is two different people… wait a minute!”)

Of course, this introduces its own set of problems since now the relationship must be explained. I think that the most typical motivations for the assistance are:

  1. Romantic interest
  2. Financial interest

Though this makes sense since they are the two big motivations for nearly everything and especially for nearly everything which is really bad. The other big motivation in human life is religious zeal, and while you probably could come up with a story in which a radical Muslim murders a Jew out of religious zeal, it would be hard to come up with a story in which he wanted to conceal it (though you could have friends do that over his objections) and in the current environment I doubt that such a story would be well received.

There is another way in which accomplices make things easier, though: being twice as many people they make twice as many mistakes—that is, they leave twice as many clues. Worse yet from the perspective of the murderer and better yet from the perspective of the detective, they can have the motivation to double-cross each other.

As with all solutions to a problem in a murder mystery, an accomplice solves some problems and causes others. This is why it’s better in real life to always do right, of course, because then all of your problems will at least be good problems to have, but in fiction it’s what has made mystery such an enduring genre.


If you enjoy murder mysteries, please consider checking out my murder mystery, The Dean Died Over Winter Break.

tddowb

The Butler Did It Again

(This is a follow-up to a series of blog posts on the subject, the most recent being here.) As I was reading another article on the origin of the phrase, “the butler did it,” my attention was drawn to the story The Strange Case of Mr. Challoner, by Herbert Jenkins. Published in 1921, it preceded The Door by nine years. (Interestingly, Herbert Jenkins owned the publishing house which published P.G. Wodehouse’s books, most famously the stories of Jeeves and Wooster.) I tracked down a copy and read it. (There’s a free ebook version of the book Malcolm Sage, Detective on kindle, which collects all of Jenkins detective stories—if you want to read it I suggest you do it now because there will be spoilers below).

Jenkins’ detective was Malcolm Sage, who was at least vaguely in the mold of Sherlock Holmes and Hercule Poirot, by which I mean that he was both very observant of physical details and very eccentric. All of  the stories about Malcolm Sage were short stories, which is very significant to understanding the relationship of this story to the phrase, “the butler did it”.

Novels and short stories are very different things in any genre, but this is especially true of murder mysteries. Novels tend to focus on the unraveling of intertwining mysteries, which is to say the elimination of red herrings. This is somewhat necessitated by the length of a novel; each red herring forms a sort of sub-mystery, which allows one to enjoy the solving of mysteries over and over throughout the course of a novel. There are exceptions, of course. It is possible to combine a mystery with some other genre where the other genre takes up most of the page count. Adventure is the obvious example; a mystery/adventure works well where each clue is the reward at the end of an adventure. To some degree the Hardy Boys books were like this, and to a lesser extent this is often true of the Cadfael stories. The Virgin in the Ice and The Summer of the Danes are both great examples of where the adventure takes up more pages than the mystery. (Both are excellent novels.)

For related reasons—though there are notable exceptions—murder mystery novels don’t tend to focus on figuring out a single ingenious mechanism for concealing the murder(er) for which the evidence was present at the crime scene. By contrast, this is extremely common in short stories. Among other things, they don’t have the space for disentangling red herrings. Short stories which were printed in magazines tended to be extremely short, sometimes only a few thousand words. It also is simply the right size for that sort of game.

The Strange Case of Mr. Challoner is a locked-room mystery. There is one obvious suspect: a nephew of whose impending marriage the deceased disapproves and who will be disinherited on the morrow. The butler was the last to see the deceased alive, and the body was discovered in the library, with all of the doors and windows locked from within. The deceased was staged to look like suicide, and the local police take it at face value. Malcolm Sage makes numerous measurements and observations, and also directs that the photographer attached to his detective agency take a number of photographs. Malcolm Sage is so fond of photographs as evidence that he gives a lecture on their importance to the local police detective inspector. Eventually he reveals that the butler, who had only been working in his position for six months and was highly praised for the excellence of his work, is the culprit. Sage had taken supposedly exclusionary fingerprints from everyone, and used those to find out that the butler had a criminal record and was still wanted. Further, he explained that the butler had put a small metal rod through the hole in the key’s handle and using a string attached to it turned the lock by pulling on the string with the door closed. Once the key turned far enough, the metal rod fell out of the hole in the key’s handle, and he used the string to pull the rod under the door and retrieve it.

Unlike the butler in The Door, this time at least the butler was actually taking advantage of his role as butler in committing the murder. His master didn’t think anything about his coming from behind because it’s the sort of thing that butlers do, and moreover he had an excuse for being in the house after the rest of the household had gone to sleep because he lived there. So at least in this case butling was relevant to the butler’s commission of the crime.

None of the articles I’ve seen so far have cited The Strange Case of Mr. Challoner as having had any influence on the phrase, but then again none of them have cited any evidence for why The Door did have influence, either. It leaves me wondering whether any of this is actually relevant to the phrase I’ve been considering. It might well not be. With murder mysteries having been quite popular ever since Sherlock Holmes first studied scarlet, I assume that there were a great many short stories in the weekly and monthly publications of the early 1900s which have largely been lost to the sands of time. In the days before television and even before radio plays were particularly popular, theatrical plays were quite popular. Wherever there is a maw gaping for novelty, there will be people trying to fill it. Certainly this is the source that the character Broadway cited as his authority that all murders were committed by butlers in the 1933 short story, What, No Butler? I’m disinclined to think that much of the source was movies, though I don’t have any hard evidence for that. Murder mysteries don’t lend themselves well to silent films, though I have no doubt that somebody tried it at least once. The Jazz Singer was the first talkie, in 1927. Talkies took over quite quickly, as I gather, dominating film no later than the mid-1930s and probably in the early 1930s, but that’s rather close to when What, No Butler? was written to have embedded itself in the culture as a common trope by then.

I’m left where I was before, wondering where this trope came from. Perhaps I’ll be successful in tracking down contemporary reviews of The Door, which might be illuminating, but unfortunately a quick google search didn’t turn up anything. I might have to resort to going to the library!

So, The Butler Did It

I’ve been reading Mary Roberts Rinehart’s murder mystery The Door, which I talked about here and here, at five and twenty two chapters in, respectively. This was started off by my wondering about the phrase, “the butler did it”. I’ve finally finished the book, so this post will finish off my review of The Door, and also discuss the idea of the butler being the murderer. I’d warn you about spoilers, but, well, I think that you already know that the butler did it. I might spoil a few side-mysteries too, though, so caveat lector.

The book was in its entirety written in the style of the memoirs of someone who observed a very strange situation. I am used to murder mysteries and detective fiction being, roughly, synonyms, but The Door is very clearly a murder mystery while it is not at all detective fiction. There is a police detective—who does solve the case—but almost entirely outside of the narrative. Several members of the family play at a little detecting, but only occasionally. Only one of them does anything which does not simply anticipate a later discovery, and that was to effect a useful introduction, rather than any actual detection.

The story also maintains the style of foreshadowing hints until the end, abandoning it only as the police detective explains the solution, which is the last thing that happens in the book. I’ve concluded that I don’t like this style. It feels at best overwrought, and at worst like an attempt to spice up a dull narrative with chopped up bits of other parts of the same narrative. I don’t mean that all foreshadowing is bad, of course, but The Door seemed to use foreshadowing in place of a compelling plot.

There is also the very strange question of the narrator, Elizabeth Jane Bell, who narrates the story in a very personal way. Throughout the story alternately laments the tragedy, investigates it, and destroys evidence to try to protect the family. It’s that last part which is especially hard to reconcile with the narration; why on earth would she be narrating all of these scandalous details in a memoir when the character of herself within the memoirs would want all such scandal wiped out? Whether you take the inconsistency between herself in the story and herself as narrator to be a problem with the character or a problem with the narrator (I took it as the former), it is still an unsettling problem.

There is also the problem of the family which Elizabeth Jane was trying to protect. Her niece Judy was never really under any suspicion having, as I recall, an alibi from the beginning. She was the only really sympathetic member of the whole family other than Elizabeth Jane herself, and she mostly from a general pleasantness which seemed to be a combination of decent manners, comfortable circumstances, and little ambition. The rest were detestable. Towards the end I was hoping that the murder would be solved after the good-for-nothing Jim was executed, just so the wretch would be out of the story. The other characters were similarly unpleasant, which left me very unsympathetic to the family’s desire to avoid scandal, which was to a fair degree their only major motivation in anything that they did. But this brings up an interesting point in murder mysteries in general: it’s hard for likable characters to be suspects.

The mystery in a murder mystery obviously depends on there being more than one suspect. More properly, on there being more than one credible suspect. The problem is that a character can fail to be credible as a suspect by being too likable. It’s very difficult to write an enjoyable story about a good person who stoops to murder but then cheerfully covers it up. It’s that much harder to write several characters who are all credible in that way; to pull it off one must write good characters with depth, rather than the common approach of paper-thin automatons who are good merely because they’re not tempted by ordinary temptations. It’s much easier to make suspects credible by simply making there be nothing to which they won’t do for gain.

Another important distinction between suspects in a mystery is between those with an obvious motive and those without an obvious motive. Very often this does not line up well with the moral probity of the characters. In order to put an innocent person in peril (to heighten the tension) a morally upright person will get an obvious motive, while a moral degenerate will get none. This helps to spread the doubtfulness around, to be sure, but because both of these suspects have something obviously going for them as suspects, it is especially common to make the culprit someone who is not very morally offensive (apart from their murders) who has a hidden motive. Which brings us to the butler.

How much was the butler a character and therefore a potential suspect? It’s hard for me to say fairly because I already knew that he did it, of course, but doing my best to be fair, I would say somewhat, but not much. Joseph (the butler) gets progressively more tired, worn out, and on edge as the story progresses, which certainly was a clue (that he was running around doing things while everyone else was asleep). He had originally come from one of the victim’s household’s, which should have been a clue but actually wasn’t—his prior connection to the rich victim had no significance as far that was revealed in the story. Nothing was ever made of him having the opportunity for the murders, because they happened at times when everyone had opportunity, and the house was small enough that a butler’s ability to be unnoticed had no significance. In fact, all three murders happened outside of the house, so his position as butler was—if anything—a disadvantage. He had to sneak off to commit them, or commit them while he was off-duty. The one time his being a butler was an advantage was when he answered the door when one of the victims came to see Elizabeth Jane but he turned her away because Elizabeth Jane was sleeping. Any butler might have turned her away, and any murderer might have learned of her coming and consequently resolved to kill her before she could tell what she knew.

On balance, the disadvantages of Joseph’s being a butler far outweighing the advantages makes Joseph’s being a butler fairly irrelevant to his being a murderer. It’s really just his profession. Most murderers have a day-job and there’s no particular reason it shouldn’t be butling. In this case his being the butler of the narrator was something of a camouflage; it meant that she didn’t notice him. Also his many years of loyal service made her affectionate of him, and this combined with the murders happening nowhere he was supposed to be and her always thinking of him as having no existence past being her butler disguised him as a suspect. But it didn’t disguise him totally. One of the themes of the book is how little one really knows of the people one thinks one knows, and the fact that Joseph had a wife somewhere but Elizabeth Jane had no idea where does actually highlight this blindness in a way that makes it fair game for the reader to not be so blind. In fact, I would argue that line by Jane Elizabeth is a well crafted notice to the reader that Joseph is a potential suspect.

Further, if the test of victory in the contest between the reader and the writer of a murder mystery is that the writer wins if the reader doesn’t guess who the murderer is but blames himself rather than the writer for it, then I believe that The Door has the potential for victory. Reading it through while knowing what to look for, I think that Rinehart did play fair with the reader. Certainly it seems possible she knew who the murderer was from the first, and did not merely cast about for someone she hadn’t already ruled out when she came to the ending. So I don’t think that there’s any cogent criticism to be made of her choice of murderer. (Except, perhaps, that it’s a little odd for someone who engages in fraud, forgery, and conspiracy—which eventually leads to multiple murders to cover those up—to have no criminal history, but instead a long and unmarred career in positions of significant trust.)

So when we come to the question of whether it is legitimate that, as Wikipedia puts it (as of the time of this writing), “Rinehart is considered the source of the phrase “The butler did it” from her novel The Door (1930), although the novel does not use the exact phrase.” Not only does the novel not use that exact phrase, it doesn’t use any even somewhat similar phrase. I’m going to quote the reveal in the novel, but I need to mention a little context first. Joseph had been mysteriously shot in the collar bone about a week before, but he was not killed and recovered enough to come back to his duties, though with his arm in a sling. Elizabeth Jane had, therefore, given him leave to go on holiday to recover. We have not learned up to this point who Joseph’s wife is, but we can mostly guess it was a woman who figured into the plot somewhere else, who we knew to be dying of inoperable cancer. We’re picking up with the tail-end of the explanation given privately to Elizabeth Jane by the police detective. During the explanation he had been calling the murderer “James C. Norton”, which he told her was the pseudonym the murderer had used to procure a safe deposit box. So, with that said, here is the reveal in the novel:

“So we got him. We’d had his house surrounded, and he hadn’t a chance. He walked out of that house tonight in a driving storm, and got into a car, the same car he had been using all along; the car he used to visit Howard Somers and the car in which he had carried Florence Gunther to her death, under pretext of bringing her here to you.

“But he was too quick for us, Miss Bell. That’s why I say I bungled the job. He had some cyanide ready. He looked at the car, saw the men in and around it, said, “Well Gentlemen, I see I am not to have my holiday—”

“Holiday! You’re not telling me—”

“Quietly, Miss Bell! Why should you be grieved or shocked? What pity have you for this monster, whose very wife crawled out of her deathbed to end his wickedness?”

“He is dead?”

“Yes,” he said, “Joseph Holmes is dead.”

And with that I believe that I fainted. [that’s the last line in the book]

There is nothing there remotely similar to the exact phrase, “the butler did it.” As you can see, there was nothing there even related to him being a butler. There were a few things which happened in the house that his living in the house enabled, but much of the criminal activity actually in the house was not in fact Joseph’s doing. The door referred to in the title was a hotel door where a fraud was performed, and was not in the house in which Joseph was a butler. It was not even in the same city as the house in which Joseph buttled. Except possibly as a violation of the tacit convention that the butler is the one person who never, ever commits the murder(s) in a murder mystery, his being a butler is utterly irrelevant either to the murders or to whether one suspects him of those murders.

After a bit of research, I found what seems like evidence that Damon Runyon’s What, No Butler? was first published in Collier’s Weekly, August 5th, 1933. That is not so early that the joke that the butler always does it was necessarily common by the time that The Door was published, three years earlier, but I think it does suggest it. Given what the book actually is, and the timing of it relative to jokes about the butler always being the culprit, I really doubt that The Door was in any way the origin of the phrase. It’s not impossible, but I’d really like to see better evidence for it besides this being the first (and nearly only) book which anyone can find in which a butler actually did it.