No Autopsy, Two Rulings, One Legend: The Record Behind Houdini’s Detroit Death

Harry Houdini died in Detroit 100 years ago today. Nobody performed an autopsy. The question of what killed him was answered for most of a century by an insurance company, and it answered twice.

One Hundred Years After Room 401

I love a Halloween story as much as anyone, and Houdini dying in Detroit on October 31 is about as good as they come. The greatest escape artist in the world, stopped by his own appendix, on the one night of the year built for ghosts. His widow held a seance every Halloween for a decade. People are still holding them, including a sold-out one tonight at the American Museum of Magic in Marshall.

Here is what I notice when I set the legend down and look at the paperwork. The most famous death in Detroit’s history was never formally investigated. There was no autopsy. There was no inquest. The man who threw the punch was never charged and never cleared. The closest thing to an official finding on causation came from a life insurance company deciding whether to pay $25,000 or $50,000, and that company came down on both sides within six months.

That is not a magic trick. It is a records problem, and it is the same records problem I write about every week in Michigan courtrooms and county offices. When the institution responsible for answering a question never answers it, somebody else will. Usually somebody with a stake in the answer.

What the Record Actually Contains

Houdini’s last tour was already going badly before Montreal. On October 11, 1926, a piece of equipment struck his ankle while he was being locked into the Water Torture Cell in Albany, New York, and he kept performing on a fractured ankle. On October 22, resting on a couch in his dressing room at the Princess Theatre in Montreal, he was visited by three students. One of them, J. Gordon Whitehead, asked whether it was true he could take any punch to the stomach. Before Houdini could brace, Whitehead hit him several times.

He finished his Montreal run in pain and took an overnight train to Detroit. By the time he reached the Garrick Theatre on October 24, he was running a fever reported at 104 degrees. A local physician, Dr. Leo Dretzka, examined him in his dressing room. Houdini performed anyway and collapsed afterward. He went back to the Statler Hotel instead of a hospital.

At the Statler, the substitute house doctor, 25-year-old Daniel Cohn, examined him and called Dr. Charles Kennedy, a Grace Hospital surgeon. Kennedy later wrote that he arrived around 3 a.m., heard about the blow, and suspected a ruptured intestine or a clot in a blood vessel caused by it. Houdini refused the hospital until his own New York physician told him to go. He was operated on at 3 p.m. on October 25. The doctors’ public statement that night described an obscure abdominal condition that proved to be an anomalous appendicitis. Peritonitis had already set in.

For the next six days, Grace Hospital issued signed bulletins on his temperature, pulse, and respiration, and newspapers across the country printed them. He rallied, then declined. He died on the afternoon of October 31 with his wife, Bess, and his brothers present. His body went to the W.R. Hamilton funeral home on Cass Avenue and then by train to New York.

Then the record stops. No autopsy was ordered. No coroner’s inquest examined whether a man who had been struck nine days earlier died of that blow. The appendix was the answer, and nobody with the authority to test it ever did.

The Insurer Ruled Twice

Houdini carried a $25,000 life insurance policy with New York Life. Like many policies of the era, it carried a double indemnity clause: if his death was accidental, the company paid another $25,000. A death from disease paid the base amount. A death caused by an injury paid double. That single contract term made the Montreal punch worth exactly $25,000.

The company investigated. On December 22, 1926, the Associated Press reported from Montreal that H.J.J. McKeon, the local manager for a New York insurer, had finished his inquiry. The New York Times headlined it as the insurance company confirming the denial that a blow killed the magician. At that point, the payout stood at $25,000.

Bess Houdini challenged the finding in early 1927. On June 13, 1927, New York Life’s main office in Manhattan announced the opposite conclusion. Expert testimony, the company said, had established that the blow caused his death. She received $50,000. Accounts of the investigation say the company took statements from Jacques Price and Sam Smilovitz, the students who witnessed the punch.

Look at what that means. The determination that turned a dressing room incident into the cause of death of the most famous entertainer in America was made in a claims file. The decision maker had a direct financial interest in the answer. It went one way in December and the other way in June. The expert testimony behind the reversal was never published, and no court reviewed it. And that is the version that went into the history books, the Tony Curtis movie aside.

The Investigator Who Was Investigated

Houdini spent the last years of his life doing something most people do not associate with a stage magician. He investigated fraud. Spiritualist mediums were charging grieving families to contact the dead, and Houdini set out to show, seance by seance, how the tricks worked. He hired undercover investigators, most famously Rose Mackenberg, who attended readings under false names and kept records of what she was told and what she paid.

In 1926, he took that work to Congress. Representative Sol Bloom of New York sponsored H.R. 8989, the Copeland-Bloom bill, which would have made it illegal to tell fortunes for pay in the District of Columbia. Houdini testified on February 26 and again during hearings on May 18, 20, and 21. Mediums and astrologers packed the room. Order collapsed repeatedly. Mackenberg testified that a Washington medium told her senators had come to her for readings. Houdini argued that mediums were fortune tellers using a religious label to avoid the law. The Spiritualists argued the bill attacked their faith. The bill died in committee.

I want to be direct about why this matters to his death. Houdini’s method was the method of an accountability investigator. Verify claims independently. Go undercover when the subject controls the story. Write everything down. Do not accept the explanation offered by the person with something to gain. He applied that standard to everyone who claimed powers they could not prove.

Eight months later, nobody applied it to him. His death was explained by his doctors, adopted by the newspapers, and settled by his insurer. The one man in America most likely to demand an autopsy was the one in the casket.

His crusade also supplied the motive for the most persistent theory about his death. He had made enemies, publicly and by name. In their 2006 biography, William Kalush and Larry Sloman pointed to a November 1924 letter from Arthur Conan Doyle, a devoted Spiritualist, predicting Houdini would get what was coming to him. They identified Boston surgeon Le Roi Crandon, husband of the medium Houdini exposed as Margery, as a possible organizer. They also pointed to an experimental serum given to Houdini at Grace Hospital.

Three Theories, One Missing Autopsy

Every theory about Houdini’s death is really an argument about evidence that was never collected. Here is what each one rests on.

The punch. The blows are documented by witnesses and by Kennedy’s own account of what he was told that first night. What is not documented is that they caused anything. A blow to the abdomen does not ordinarily cause appendicitis, and many physicians and historians believe Houdini’s appendix was already inflamed before Montreal. The punch may have masked symptoms he would otherwise have noticed. The record supports that he was punched and that he died. It does not support that one caused the other, and the only institution that said otherwise was paid to decide.

Poison. In March 2007, a Houdini great-nephew, George Hardeen, joined Kalush and Sloman in calling for an exhumation to test Houdini’s remains for arsenic. Forensic pathologist Dr. Michael Baden and forensic scientist James Starrs were named to the team, and attorney Joseph Tacopina said paperwork would be filed. Relatives on Bess Houdini’s side objected, as did parts of the magic community. When pressed by the Washington Post, Kalush acknowledged there was no known link between Crandon and the doctor who gave the serum. By 2009, the Houdini historian John Cox reported that the legal papers were never filed and Hardeen had stepped back. The poison theory has no physical evidence behind it. It also has never been tested, because the one test that could answer it never happened.

The death certificate. This summer, retired urologic surgeon Larrian Gillespie published Houdini: The Man Who Died Twice through Cipher House Books, a press she owns. According to her account to the New York Times, the Detroit city death certificate shows scratched-out information and inaccurate names, and was signed by a physician who was not present when Houdini died. She says three certificates existed, one with the state, one with the city, and one filed with the insurer, and that a fraud official at Michigan’s vital statistics office later found the city copy invalid. Her medical theory is that a bowel obstruction, not the appendix, killed him after an emergency operation on October 29, that a second operation followed on October 30, that he died that day rather than on Halloween, and that the later surgeries were left off the death records. She argues an autopsy was avoided because the obstruction could have raised malpractice questions.

Reaction has split. Houdini author Patrick Culliton told the Times he thinks she has cleared Whitehead. Hardeen called her case persuasive. Cox, who helped with her research, praised the detective work but concluded in his review that “the mystery remains.” Gillespie also paid for a headstone on Whitehead’s unmarked Montreal grave. Whitehead had lived as a recluse and died of malnutrition in 1954. The stone reads “Falsely Accused No More.”

Why This Matters Beyond Houdini

It is easy to treat this as a century-old curiosity. It is not. The structure of what happened in Detroit in 1926 shows up in Michigan death investigations today.

Under Michigan’s current medical examiner law, deaths that are sudden, unexpected, or follow violence generally fall to county medical examiners to investigate. Whether a full autopsy follows is a judgment call made county by county, under real budget pressure. When that call goes the wrong way, the family is left with the same thing Bess Houdini had: a certificate, a hospital’s account, and whatever an insurer decides.

Insurers still make causation calls that function as findings. Accidental death riders, workers’ compensation, and life insurance contestability reviews all require a company to decide how someone died. Those files are private. The reasoning is rarely disclosed. And when no public agency has done the work, the claims decision becomes the only answer anyone has.

Death certificates still get amended, signed by physicians who were not present, and filed with errors that nobody catches until a family or a researcher goes looking. The certificate is a legal document. It is also, very often, the only official record of a death that was never examined.

And theories still grow fastest where evidence is thinnest. I see it in every cold case I cover. When an agency closes a file without doing the work, it does not end the questions. It hands them to whoever is willing to speculate, and it lets an innocent person carry a reputation for decades. J. Gordon Whitehead lived and died as the man who killed Houdini, on the strength of an insurance claim.

Bess Houdini held her last official seance in 1936. Ten years was long enough, she reportedly said, to wait for any man. A hundred years is long enough to wait for an answer. The certificate is in Lansing. Somebody should pull it.