Overview
Harry Houdini died in Detroit on the afternoon of 31 October 1926, of diffuse peritonitis following a ruptured appendix. He was fifty-two. That he died on Halloween is a coincidence that has done more for the story's longevity than any piece of evidence.
Nine days earlier, in his dressing room at the Princess Theatre in Montreal, a McGill student named J. Gordon Whitehead had asked whether Houdini could withstand blows to the abdomen and then delivered several before Houdini could brace himself. Houdini was reclining on a couch at the time, recovering from a broken ankle.
The accepted account connects those two events. The conspiracy account connects them differently: that the punch was not an over-enthusiastic student's stunt but an attack, and that the man who had spent years destroying the livelihoods of fraudulent mediums was killed for it.
The medical evidence, examined since, complicates the accepted account as much as the conspiracy one.
The campaign that supplies the motive
Houdini's last years were given over to a public war on Spiritualism. He testified before Congress in support of anti-fortune-telling legislation, attended séances in disguise, and staged demonstrations showing how the effects were produced.
The most bitter of these confrontations was with the Boston medium Mina Crandon, known as Margery, whose case had divided a Scientific American investigating committee. Houdini's intervention was decisive and public, and it made him enemies inside a movement that was then substantial, organised and wealthy.
It also cost him a friendship. Arthur Conan Doyle, a committed Spiritualist, had been close to Houdini, and the relationship broke over it. That two of the most famous men of the era fell out publicly over whether the dead could be contacted is part of why the death that followed acquired narrative weight.
The motive, in other words, is real. Houdini had genuinely damaged people with money and belief at stake.
The medical problem with the punch
The difficulty is that the accepted story may also be wrong — not about murder, but about the punch.
Surgeons reviewing the case have made two points. The first is that no causal link between abdominal trauma and appendicitis has been established; the mechanism the story requires is not one medicine recognises. The second is the timeline, which is the harder objection. Houdini performed for a further week, travelling and appearing on stage, before collapsing. A ruptured appendix spilling infection into the abdominal cavity does not permit nine days of that. He could not have survived the interval if the blow had caused the rupture.
The likeliest reading is that appendicitis was already developing when Whitehead walked into the dressing room, that Houdini attributed his worsening pain to the blow, and that this misattribution is precisely why he did not seek treatment until the infection was overwhelming. On that account the punch did contribute to his death — by supplying a wrong explanation, not by causing the injury.
What the surgeons found is also on the record. Houdini was operated on twice at Grace Hospital in Detroit, and the appendix had already ruptured; the infection that killed him was streptococcal peritonitis, which in 1926 had no reliable treatment. He was given an experimental serum as a last measure and it did not work. Penicillin was more than a decade away from clinical use. A man in his position today would almost certainly survive, which is part of why the death seems to demand a bigger explanation than it has.
What the insurer did
One fact is usually left out of both versions, and it is the most concrete thing in the case.
Houdini's life insurance policy paid his widow a double indemnity for accidental death. An insurance company had a direct financial interest in disputing that his death was accidental, examined the circumstances, and paid the higher sum anyway.
That is not proof of the punch's role — it reflects what was believed at the time on the information available, and insurers settle for many reasons. But it is a decision made by a party with money at stake and no interest in the legend, and it is the closest thing to a contemporary adjudication that exists.
The murder case
The strongest version of the murder argument does not claim Whitehead was an assassin. It proposes poison.
The argument, developed most fully in a 2006 biography by William Kalush and Larry Sloman, is that Spiritualist circles of the period had a documented history of threatening and in some cases harming their opponents, that Houdini had received threats, and that his symptoms are compatible with poisoning as well as with appendicitis. Its proponents note that no autopsy was performed.
In 2007 Houdini's great-nephew announced an intention to seek exhumation to test for poison. The proposal attracted substantial attention and was not carried out; other family members opposed it, and the practical obstacles to detecting most poisons in eighty-year-old remains are considerable.
The test he left behind
Houdini did something before he died that bears directly on the theory, and it is usually omitted.
He and his wife Bess agreed a coded message that only the two of them knew, so that if anyone claimed to be channelling him after his death she could test it. Bess then held a séance on the anniversary of his death every year for a decade. Mediums came forward, and none produced the code. She ended the practice in 1936, saying she had waited long enough.
This matters to the murder theory in a way that cuts against it. If Spiritualists had killed Houdini to stop his campaign, the campaign did not stop — it continued for ten years, conducted by his widow, in the most public form available, and it went on failing to produce a result in their favour. Nobody interfered with her.
The annual Halloween séance has continued in various forms since, which is one reason the date has stayed attached to the story.
Weighing it
The factual core is agreed. Houdini was struck, he continued working, he died of peritonitis from a burst appendix nine days later, and no autopsy was conducted.
What the murder theory adds is a motive that genuinely existed and an absence of evidence that genuinely exists. What it lacks is a mechanism, a suspect, or any contemporary indication that anyone involved in his care suspected anything but a burst appendix — including surgeons who opened him up and found exactly what an untreated appendicitis produces.
The more interesting revision is not the murder theory at all. It is that the standard story — that a student's punch killed Houdini — is itself unlikely to be true in the way it is usually told, and has survived for a century largely because it is a better ending than an ordinary infection that a busy man ignored for a week.