Magazine “History” · Reading time approx. 10 min

A father, a son, a sister, a brother, a granddaughter: five suicides in four generations. It is tempting to call it a “curse.” A genogram asks more precisely: What gets passed down — and what doesn't?

Genogram of the Hemingway family across four generations: Clarence, his children Ernest, Ursula and Leicester, Ernest's sons Jack and Patrick, and Jack's daughters Margaux and Mariel
The Hemingway family as a genogram following McGoldrick & Gerson: square = man, circle = woman, ✕ = deceased. There is no separate symbol for suicide — the manner of death belongs in the text, not in the sign. What the chart does show is the distribution: three suicides in the second row, none in the third, one in the fourth. Only the people named in the article are shown; Ernest Hemingway had six siblings and three sons.

A Word Older Than Its Reputation

“The Hemingway curse” has been a fixture of English-language cultural journalism for decades. It surfaces whenever the family comes up in conversation. It sounds like a literary bon mot — fitting for a writer whose own work is full of violence, death, and stoic toughness. In fact, the phrase rests on a real, documented foundation: Clarence Hemingway, Ernest Hemingway's father, took his own life on December 6, 1928, in Oak Park, Illinois. Ernest was 29 years old at the time. Weeks later, he wrote a letter to his mother-in-law, Mary Pfeiffer. In it he set down a sentence that, in hindsight, reads like a premonition: “I'll probably go the same way.” 33 years would pass before that sentence came true.

For therapists and counselors who work with genograms, this letter is precisely the interesting point — not the prediction itself, but the fact that it was put into words at all. A family history that inscribes itself as an expectation into a 29-year-old is not a question of inheritance. It is a question of the narrative a family tells about itself. This is exactly what the Hemingway genogram can show, without reducing the family to a fate.

Five Deaths, Four Generations — The Data

Before looking for patterns, the sober facts come first. Five suicides across four generations are documented in the Hemingway family:

  • Clarence Hemingway, Ernest Hemingway's father, died on December 6, 1928, at the age of 57. He suffered from diabetes and, according to the consistent account of his biographers, from depression.
  • Ernest Hemingway himself died on July 2, 1961, at the age of 61, in his home in Ketchum, Idaho — after years of alcohol dependence, recurring severe depressive episodes, and a series of physical illnesses that intensified dramatically in his final year.
  • Ursula Hemingway, his younger sister, died in 1966 at the age of 64. She had cancer at the time and, like her brother, suffered from depression.
  • Leicester Hemingway, his younger brother, died in 1982 at the age of 67. He too suffered from diabetes and severe depression — the same combination that had already claimed their father's life 54 years earlier.
  • Margaux Hemingway, Ernest Hemingway's granddaughter (daughter of his eldest son, Jack), died on July 1, 1996, at the age of 42 — one day before the 35th anniversary of her grandfather's death, a coincidence widely noted in the press.

Five people, four generations of one family, a recurring pattern of depression and — in three of the five — an additional diagnosed physical illness. That is the established factual record. What follows from it is the real work. And this is exactly where the question begins that distinguishes this article from a mere list: How does one even record something like this in a genogram?

The Notation Question: How Does Suicide Appear in a Genogram?

Anyone drawing a genogram for the first time according to the standard developed by Monica McGoldrick and Randy Gerson often looks for a dedicated symbol for suicide — a cross in a special color, perhaps, or a lightning bolt, something that sets this cause of death apart from others. There is none. The core symbol set has exactly one sign for “deceased”: an ✕ placed over the person's square or circle, accompanied by the date of death. Whether someone died of old age, an accident, an illness, or suicide cannot be read from the symbol itself. The cause of death, if it is recorded at all, is a text note next to the mark — not a graphic element, but an annotation.

This is not a gap in the standard but a deliberate restraint, and it has a good reason. A symbol suggests sameness. Anyone who introduces a special sign for suicide implicitly treats every one of these deaths as the same kind of event — regardless of age, circumstances, or history. A text note, by contrast, forces specificity. It requires writing down each individual case as it actually was — with its own age, its own context, its own open questions. For the Hemingways, that means: five ✕ marks, five dates of death, five different notes — not one repeated special symbol that visually fuses the five deaths into a series before anyone has even looked closely. Current genogram software follows the same convention. There is a field for the date of death, but none for the cause of death. As in the paper original, it belongs as an annotation to the person, not as its own graphic element.

For practice, this has two implications. First: the decision whether to note a cause of death at all lies with the practitioner who draws up the genogram together with a client. Second: often the decision lies even earlier, with the family itself, which determines whether and how openly a suicide is discussed. This is precisely what makes the notation question more than a technical footnote. A genogram that consistently omits a cause of death is, in doing so, also recording a family's silence. A genogram that names it makes visible what otherwise only resonates between the lines of a family's narrative.

Four Generations, One Line

When the Hemingways are drawn as a genogram, with generations stacked in rows, a structure emerges that a chronological narrative easily obscures. The five suicides are not distributed evenly across the generations; they jump. Clarence Hemingway stands in the first row. In the second row, his children, three appear at once: Ernest, Ursula, and Leicester — out of six siblings in total, the remaining three daughters were spared. In the third row, Ernest Hemingway's own children, none appear. Neither his eldest son, Jack, nor his two other sons, Patrick and Gregory, died by suicide. Only in the fourth row, with Jack's daughter Margaux, does the pattern reappear.

This inconspicuous third generation is the real discovery in the genogram. A purely chronological account — “1928, 1961, 1966, 1982, 1996” — reads like an unbroken chain. In the genogram, where generations sit side by side, it becomes visible that the chain in fact has a gap. An entire generation carried the burden, the pressure of expectation, the knowledge of what came before — and did not pass it on in the same form as the generation before it. What this generation did differently cannot be conclusively answered from the publicly available sources. But that it was different is legible in the genogram before anyone even asks why.

What can be clearly identified, by contrast, is the role of the family narrative itself. Ernest Hemingway's letter to his mother-in-law was written weeks after his father's death. It shows how early and how directly such an expectation can be put into words. It does not happen as a diagnosis, but as a sentence a 29-year-old writes about himself. Family research distinguishes two mechanisms here. One is modeling: a child watches how a parent handles crisis and adopts this behavioral repertoire without it ever being spoken aloud. The other is the explicit family narrative, the sentence that is actually said: “That's how it ends for us.” Neither is directly visible in the genogram. But both can be suspected at the points where a pattern persists across generations even though the circumstances have completely changed. Oak Park in 1928 and Ketchum in 1961 have almost nothing in common besides the family.

What Belongs Alongside the Family History

Anyone looking at Ernest Hemingway's final year in isolation finds a separate, independent backstory that makes any simple explanation of “suicide because of a family pattern” fall short. In November 1960, he was admitted to the Mayo Clinic in Rochester, Minnesota, under a false name — officially for high blood pressure, in fact for a severe depressive episode. There he received electroconvulsive treatments repeatedly over the course of two months. In April 1961, after a further crisis, he was admitted to the Mayo Clinic again and treated a second time. By his own account, given in letters, he experienced the treatment as devastating for his memory. For a writer whose work depended on precise recall of his own experiences, this was a severe loss. He himself described it as the end of his ability to write. He died only a week after his discharge from the second course of treatment.

In this context, another hypothesis keeps recurring in the literature on Hemingway. In his final year, Ernest Hemingway was diagnosed with hemochromatosis — a metabolic disorder in which iron accumulates in the body, damaging the liver, pancreas, and other organs. Some authors have built on this to argue that a hereditary form of the condition might also help explain the psychiatric symptoms in him and several of his relatives. This claim remains entirely unsubstantiated and is explicitly classified as unproven speculation by the majority of experts who have studied the case. The medical means to test it in someone who died in 1961 simply do not exist. The diagnosis itself is documented. It nonetheless does not belong in this article as an explanation. Rather, it shows how quickly a narrative can grow out of a real finding, one that reaches far beyond what the finding actually supports. This very temptation — turning a single factor into an all-encompassing explanation — is the same temptation behind the word “curse.”

Repetition Is Not a Law of Nature

A genogram that shows only the suicides paints an incomplete picture. Just as important are the family members who carried the same legacy and took a different path.

Patrick Hemingway, Ernest Hemingway's second son and Jack's brother, grew up with the same history as his father, his brother, and his niece Margaux. He died in September 2025 at the age of 97 — a life that lasted more than six decades longer than his father's. A genogram that carefully places birth dates alongside death dates makes this contrast visible automatically: the same origin does not produce a uniform lifespan.

Mariel Hemingway, Margaux's younger sister and Jack's third daughter, has chosen a different, equally remarkable path: that of open engagement. In the 2013 documentary “Running from Crazy” by Barbara Kopple, she speaks publicly about the history of suicide and mental illness in her family. She speaks just as openly about her own experience with depression and suicidal thoughts. In it, she explicitly describes how important it was to her not to carry this legacy in silence but to actively work through it — with therapeutic support and in public. In the terms of genogram work, this is the counterpart to the silence that earlier generations of this family kept around Clarence's death: not repetition, but a deliberate break.

Both examples necessarily belong in the same genogram as the five suicides — not to gloss over the history, but because a genogram that shows only the repetition reduces the family to an expectation that the facts themselves disprove.

What Practitioners Can Take from This

Clients who are themselves confronted with a familial clustering of suicide or depression know a particular worry. They often phrase it in exactly this pattern: “In our family, it's hereditary.” The sentence carries a real fear and deserves to be taken seriously. Mental illness can indeed show familial clustering. But it also deserves the kind of differentiation a genogram can provide — distinguishing between what may be passed on as vulnerability and what is added through expectation, modeling, and narrative. The second part is changeable.

Hypothetical case example: A woman comes to counseling because her uncle took his own life years ago. She says right at the start: “I'm afraid this runs in our family.” On the genogram drawn together in session, it becomes apparent that no one in her grandparents' generation ever spoke about this suicide — there were hints, but never a named fact. The counselor does not ask about the act itself, but about how the family handled the silence. What becomes clear: it is not necessarily the diagnosis that repeats, but the pattern of not talking about it — and that is exactly where the further work begins.

The Hemingway genogram shows this twice over. First, in the generation that broke the chain without anyone being able to say exactly why. Second, in Mariel Hemingway, who did not leave the break to chance but actively brought it about. For work with genograms, this means: document repetition, yes — but never without, in the same breath, looking for the places where it did not occur.

If this is difficult for you: Reading about suicide can be distressing — for you personally, or for clients you share this text with. In Germany, the telephone counseling service (Telefonseelsorge) is reachable around the clock, free of charge, at 0800 111 0 111 and 0800 111 0 222.

Sources