By Serial Killer Magazine
In September 1803 a Berlin court convicted Sophie Ursinus of attempting to poison her servant, and sent her to the fortress at Glatz for life. She had been investigated over four people in her household: her husband, her lover, an elderly aunt whose money she inherited, and a servant called Benjamin Klein, who became violently ill after taking food and medicine from her and who survived. The court convicted her of the attempt on Klein. Of the three people who were actually dead she was convicted of nothing at all, and she was never tried for any of them again. Those deaths had been written down at the time as consumption, a stroke and old age, and no court ever replaced those words with different ones. Two hundred and twenty years later that is still what the record says.
Thirty-three entries in this archive describe a death that was certified as natural and later came to be treated as a killing. That is a large number for one collection, and the obvious reading of it is that doctors and coroners made a lot of mistakes. The cases themselves say something narrower and more uncomfortable. In most of them the certificate was not a passive failure of observation. It was the thing that made the next death possible, and in a surprising number of cases it outlived the trial that was supposed to correct it.
A death certificate is not an opinion. It is an administrative permission. It releases a body for burial or cremation, it closes the file, it satisfies an insurer, and it removes the one occasion on which anybody official is obliged to look. Once it is signed, the practical question stops being whether the death was suspicious and becomes whether anyone is prepared to go back and reopen something already settled. That threshold is high, and it is highest of all for exactly the deaths in these cases: the very old, the very young, the chronically ill, and the poor.
Amelia Dyer's case shows what the document was worth to the people using it. Dyer took in unwanted infants for money in Victorian England for around thirty years. In 1879 she was convicted of neglect and served six months' hard labour, and the complaint that led to the conviction did not come from a bereaved parent. It came from a doctor, who had noticed how many death certificates he was being asked to sign for one woman. She learned the correct lesson from her own prosecution. Afterwards she stopped calling doctors and stopped registering the deaths at all. When she was finally arrested in 1896 it was because a baby's body had been taken out of the Thames.
Harold Shipman solved the same problem from the other side of the desk. A general practitioner in Hyde, he injected elderly patients with diamorphine, mostly women living alone, usually in their own front rooms in the afternoon. He then signed the certificates himself, recorded old age or heart disease, discouraged post-mortems, and edited the medical records afterwards so that the history supported what the certificate said. He was convicted in 2000 of fifteen murders. The public inquiry that followed concluded he had killed at least 215 people over twenty-three years. The instrument Dyer had to evade was simply part of his job.
Miyuki Ishikawa's maternity hospital in wartime Tokyo shows what the paperwork requires when one person cannot supply it alone. More than a hundred infants died there of deliberate neglect while she collected fees from the parents for taking them in. That needed a working chain of adults and the chain held for nearly four years: a doctor named Shiro Nakayama wrote the death certificates, an undertaker was paid to remove the bodies, a mortician accepted them for burial. Nobody in that chain had to be told what was happening. Each of them only had to do the ordinary part of their own job and not ask about the others.
The part that is easy to miss is what happened afterwards, because in several of these cases the certificate was never actually overturned. Daisy de Melker was tried in South Africa for three deaths and convicted of one. The court acquitted her of poisoning both of her husbands, whose deaths had been certified as cerebral haemorrhage and chronic nephritis, and convicted her only of the murder of her son Rhodes. Martha Ann Johnson was convicted in Georgia of the murder of her eleven-year-old daughter Jenny Ann Wright. Her three younger children, who died in 1977, 1980 and 1981, had been recorded as sudden infant death syndrome, and no conviction ever followed for any of them. Their deaths are still recorded that way. Anna Marie Hahn was convicted of one murder, that of Jacob Wagner, while five or more deaths among the elderly men she had worked for remain attributed rather than proved.
Diana Lumbrera's case shows how unstable the legal ground is even when a prosecution succeeds. Seven children died around her in Texas and Kansas between 1976 and 1990, each death recorded as natural until the last one, and she collected insurance on several. Three of those deaths are covered by convictions and none of the three is straightforward: the Kansas conviction for her son Jose Antonio was reversed by that state's Supreme Court in 1992 for trial errors before a retrial convicted her again, one Texas count was a guilty plea, and another was a plea of no contest, which is not an admission of guilt. Charges over two further children were dropped and one county declined to prosecute at all.
Where the certificate had already led to burial, there was sometimes nothing left to prosecute with. Kaspars Petrovs strangled elderly women in their flats in Riga between 2000 and 2003. He confessed to thirty-eight killings and was convicted of thirteen. The reason for the gap was not doubt about what he had done. At many of the scenes there had been no sign of a struggle, the deaths of elderly women had been recorded as natural, and the bodies had been buried without an autopsy. The evidence that would have supported the other charges had been lawfully disposed of, on the authority of the paperwork.
In one case the document did not just conceal a death, it produced the next one directly. Anthony Hardy killed three women in Camden between 2002 and 2003. The first death was recorded as natural causes and he was released. He killed two more, and dismembered them, before pleading guilty in 2003 and receiving a whole-life order.
Set against all this, look at how these cases actually ended, because it is almost never medicine that ends them. Olaf Däter, a geriatric nurse, killed five women in Bremerhaven in ten days in June 2001 by letting himself into their homes, and the first four deaths were certified as natural. He was caught because a sixth woman survived and gave the police his name. Gwendolyn Graham and Catherine Wood smothered five elderly patients in a Michigan care home in 1987, and none of the deaths was questioned at the time; the case opened in 1988 when Wood told her ex-husband what they had done and he reported it. Sararat Rangsiwuthaporn's cyanide poisonings in Thailand passed as natural for eight years, and police say fifteen people were poisoned in seven provinces between 2015 and 2023. What broke that case was one victim collapsing in a public place beside the accused, in daylight, in front of other people.
So the thing that convicts, over and over, is the sequence rather than the death. It is the eighth certificate from the same doctor, the fourth cot death in one family, the sixth woman in ten days, the friend who does not come home from the river. And the sequence is precisely what no single certificate can register, because each one is signed by somebody looking at one body in isolation and asking a reasonable question about that body alone. The medical evidence in the first death was usually no better after the arrest than it had been before it. What changed was that somebody finally had a reason to count.
Jeanne Weber is the case that stops this being a comfortable argument about carelessness. Children died in her care in France between 1905 and 1908, including her own. She was twice cleared, and cleared on the evidence of a leading forensic expert who examined the bodies and insisted that the marks on their throats were natural. This was not a country doctor signing a form he had not thought about. It was the best available expertise, applied at length, reaching the wrong answer in public. Her one conviction was later quashed, up to ten deaths remain disputed, and most of them were formally found to be natural. She was declared insane in 1908 and died in an asylum ten years later.
That is worth holding onto, because the reasoning that exposes a series can also manufacture one. A cluster of deaths around a single person is powerful evidence and it is also the kind of evidence that looks the same whether or not anything was done. The archive contains contested cases where exactly that argument is now being fought in the other direction, with statisticians and clinicians arguing that a run of collapses had natural explanations and that the pattern was assembled after the fact. Both readings cannot be right in any given case, and this collection is not the place where that gets settled. What these thirty-three entries do establish is narrower and firm enough on its own: for a very long time, in a great many countries, the document that was supposed to be the last check on a suspicious death was instead the first thing that made the next one easy.