Suicide

In 1897 Durkheim published Le Suicide. It is generally regarded as the first major work of empirical sociology — the moment the discipline stopped arguing that it was possible and demonstrated it.

He chose his case deliberately. If a social fact could explain this, it could explain anything.

A pattern that should not exist.

Suppose you were handed the death records of a European country for the year 1880, and asked to guess how many suicides there would be in 1881.

Everything about the act says the number should be unpredictable. Each one is a private catastrophe with its own particular history — an illness, a ruin, a bereavement, a despair. These are not coordinated. The people involved do not know one another. Nothing connects the man in Bremen to the woman in Lyon.

And yet the number is one of the most stable statistics a country produces. Durkheim laid the figures side by side and found that national suicide rates varied only slightly from year to year, while differing enormously between countries — and those differences persisted, decade after decade, through changes of government, harvest and trade.

More than that, the rates sorted themselves by characteristics no theory of individual despair could explain.

Protestants killed themselves at markedly higher rates than Catholics — and this held within countries, so it was not a matter of national temperament. Prussian provinces with more Protestants had higher rates than those with fewer.

Unmarried people at higher rates than married. Married people without children at higher rates than parents.

Rates fell during wars and political crises, when one might expect the opposite.

Rates rose during economic booms as well as depressions (4.2.3).

And soldiers killed themselves at higher rates than civilians — despite being young, fit, fed and employed.

Something is producing these regularities, and it is not located in any individual.

Durkheim's argument, in the form it takes at the start of the book.

Every suicide has an individual explanation. He does not deny this. Each person who dies has reasons, a history, a state of mind.

But individual explanations cannot account for the rate. If the causes were purely individual and randomly distributed, rates would fluctuate randomly. They do not; they are stable within societies and vary systematically between them.

Therefore the rate is a social fact (4.2.1) — a property of a collective, external to individuals, existing in its own right — and requires explanation by other social facts.

His formulation: "Each society is predisposed to contribute a definite quota of voluntary deaths."

And he clears the ground first, in a long opening section, by systematically ruling out the non-social explanations available in 1897: mental illness alone (which does not track the rates by religion or marital status), heredity, race, climate, temperature, and imitation — the last a direct demolition of his rival Gabriel Tarde, whose whole system rested on imitation.

That negative section matters as much as the positive one. It is Durkheim demonstrating what "eliminate rival explanations" looks like in practice, and it is why the book reads as science rather than assertion.

The idea, in plain words

	Too little	Too much
Integration	Egoistic	Altruistic
Regulation	Anomic	Fatalistic

Egoistic suicide — insufficient integration

The individual is insufficiently bound into a group; the "excessive individuation" leaves them without a collective purpose to hold on to and without others whose claims on them make life obligatory.

This is Durkheim's explanation for the religious finding, and the reasoning is subtle and worth following. He does not argue that Catholic doctrine forbids suicide more strictly — both faiths condemn it absolutely. He argues that the two produce different degrees of integration.

Catholicism binds its members into a dense collective: a shared and mandatory ritual life, an authoritative interpretation of scripture, a priesthood mediating between believer and God, an extensive framework of common practice. Protestantism, particularly in its Reformed forms, places the individual alone before God, with private reading and private conscience as the route to salvation. It is, he says, "a less strongly integrated church."

The mechanism is integration, not belief. And the same mechanism explains the other findings: married people are more integrated than single ones; parents more than the childless; and — crucially — rates fall during wars and political upheavals, because national crisis fuses a population into a collective with a shared purpose. That last finding is the one that convinced many contemporaries, because no individualistic theory predicts it.

Altruistic suicide — excessive integration

The mirror image. Where the individual is so thoroughly absorbed into the group that their own life has little independent value, and the group's demands can override the instinct for self-preservation.

Durkheim's cases: military suicide (his explanation for the soldier finding — military honour subordinates the person to the corps); the ritual suicides he read about in ethnographic accounts, including seppuku and the practice of sati; and the self-sacrifice of the old or ill in societies where dependency is dishonourable.

He distinguishes obligatory altruistic suicide (duty demands it), optional (honour permits and approves it), and acute (undertaken for the joy of sacrifice itself, as in some forms of religious martyrdom).

A serious caution on Durkheim's sources here. His accounts of non-European practices come from nineteenth-century travellers' reports, colonial administrators and missionary ethnography, and they are frequently inaccurate, decontextualised and shaped by the imperial framing 3.2.7 examined.

Sati, for example, appears in his text as a straightforward case of collective duty overriding individual life. The historical scholarship — Lata Mani's Contentious Traditions (1998) is the standard account — shows a far more complicated picture in which colonial legal debate, competing scriptural interpretations, property and inheritance interests, and outright coercion were all in play, and in which the practice was geographically limited and contested within Indian society itself.

The analytical category — that excessive integration can produce self-destruction — is defensible and has been productively applied. The examples are a nineteenth-century European reading of societies he never visited, and should be treated as such. This is Part 6's territory, and it arrives inside the canon's founding text.

Anomic suicide — insufficient regulation

The condition of 4.2.3: aspiration without limit, a scale dissolved. The evidence is the rate rising in booms as well as slumps, and the higher rates among the commercial and industrial classes and the divorced.

Durkheim treats divorce as a case of chronic anomie in domestic life: marriage regulates desire, and its dissolution removes that regulation. His analysis here contains one of the book's most-criticised passages — he argues that marriage benefits men (through regulation) more than women, and that divorce law liberalisation therefore raises male suicide while lowering female. The empirical claim is contested; the framing, which treats women's aspirations as naturally more limited and therefore in less need of social regulation, is straightforwardly of its period and wrong.

Fatalistic suicide — excessive regulation

The fourth type appears in a footnote, and Durkheim says it has "little contemporary importance." It is the suicide of those whose "futures are pitilessly blocked and passions violently choked by oppressive discipline" — his examples are the very young husband, the childless married woman, and, in a phrase, slaves.

This footnote has had a much larger career than its author gave it. It is the type that best fits prisons, closed institutions, and situations of total control — Goffman's total institutions (2.3.7). Its brevity is itself a finding about Durkheim's preoccupations: a theorist worried about modern society dissolving its bonds had little use for a category describing people crushed by them.

Take the fact that most surprised Durkheim's readers: nineteenth-century European soldiers killed themselves at rates several times higher than comparable civilians.

What individual explanations predict. Soldiers are young, physically fit, fed, housed, employed and in company. On any account of suicide as the product of illness, poverty or isolation, they should be at low risk.

What Durkheim checks. The rate rises with length of service, is higher among volunteers than conscripts, higher among officers and NCOs than privates, and higher in elite units than ordinary ones. Each of these is the opposite of what a "military life is miserable" explanation predicts — the people most committed to and most successful in military life have the highest rates.

His conclusion. The mechanism is not misery but integration too complete: a moral order in which the individual is wholly subordinate to the corps, personal life is minimal, and honour outranks self-preservation. The more thoroughly a man is a soldier, the more this holds.

Notice the form of the argument, because it is the book's real lesson. Durkheim does not merely find a correlation. He derives predictions from his mechanism, checks whether the internal gradients run the way the mechanism requires, and shows that the rival explanation predicts the reverse. That is a genuine test, and it is why this book, rather than any of the grand systems before it, is called the first work of empirical sociology.

Go deeper

The great criticism: Douglas, Atkinson, and the problem of the data

The most serious attack on Suicide is not about theory. It is about whether the numbers mean what Durkheim assumed.

Jack Douglas, The Social Meanings of Suicide (1967), and J. Maxwell Atkinson, Discovering Suicide (1978), made the argument that has shadowed the book ever since.

A suicide statistic is not a natural fact waiting to be counted. It is the outcome of a social process: someone dies; a doctor, police officer or coroner investigates; a decision is made about the person's intention; and a category is recorded.

That process varies enormously — between countries, over time, and between individual officials.

And here is the fatal point for Durkheim's central finding. In Catholic societies, suicide was a mortal sin with severe consequences: burial in unconsecrated ground, shame on the family, in some places forfeiture of property. Coroners, doctors and priests in such societies had powerful reasons to record an ambiguous death as an accident.

So the Protestant–Catholic difference may be partly a difference in recording practice rather than in dying.

Atkinson took this further, sitting with coroners and observing how they worked. He found they operated with informal "common-sense theories" of what a typical suicide looks like — a certain kind of death, a certain kind of person, a certain kind of biography — and read ambiguous cases against that template. The statistics record the coroners' theories as much as the deaths.

How serious is this? Genuinely serious, and honest teaching should not soften it.

What survives it. First, some of Durkheim's findings are hard to explain away as recording artefacts — the marital status gradient, the soldier gradients, the boom-and-slump pattern, the fall during wars. Second, cross-national differences in rates persist today, in countries with modern, standardised, medically certified death registration; the Baltic states and South Korea have rates several times those of Mediterranean countries, and no plausible recording difference accounts for the gap. Third, and most decisively, the finding that suicide rates vary systematically with social conditions has been replicated with vastly better data for over a century.

What does not survive. Confidence in Durkheim's specific nineteenth-century numbers, and any assumption that official statistics are transparent windows onto the world. Douglas and Atkinson's real contribution is methodological and permanent: every official statistic is produced by an institution, and how it is produced is itself a sociological question. That insight applies to crime figures, unemployment figures, hate-crime figures and disease registries, and it is now standard practice to ask it. Anyone using administrative data without asking who recorded it and under what incentives is being careless.

Other criticisms worth knowing

The ecological fallacy. Durkheim infers about individuals ("Protestants are more likely to kill themselves") from data about areas ("provinces with more Protestants have higher rates"). This inference is invalid in general — it is entirely possible that the excess deaths in mixed provinces were Catholics — and it was named as a formal error by W. S. Robinson in 1950, in a paper that used Durkheim as an example. Later work with individual-level data has generally supported a religion effect, but the logical error stands and is now taught as a standard warning (7.2).

Circularity. How do we know a society is poorly integrated? Partly because its suicide rate is high. How do we explain the high rate? Poor integration. Durkheim does offer independent indicators — religious affiliation, marital status, family size, political crisis — but the danger is real and recurs throughout functionalist explanation.

Neglect of meaning. The whole interpretive tradition (4.3.1, 5.4) objects that Durkheim explains rates while ignoring what the act means to those involved — and that a suicide understood by its performer as an act of protest, of honour, or of escape are different phenomena being counted as one. Douglas's title makes exactly this point.

And gender. Women attempt suicide at considerably higher rates than men in most countries, while men die by suicide at higher rates — a pattern with substantial cross-national variation and no single accepted explanation, in which method availability, help-seeking norms and diagnostic practice all figure. Durkheim's framework, which treats women's social integration as naturally thinner and their aspirations as naturally more limited, has essentially nothing useful to say about it.

What modern research finds

The remarkable thing is how much of the shape of Durkheim's argument has survived, on far better evidence.

Social integration protects. Marriage, parenthood, religious participation and dense social ties are consistently associated with lower rates in modern studies. The effect sizes are moderate and the causal direction is not always clean — people at higher risk may be less likely to marry — but the association is robust.

Rapid social disruption raises rates. Post-Soviet Russia, the Asian financial crisis of 1997–98, and the 2008 recession all produced measurable increases. The recession studies are careful and consistent: rises concentrated among working-age men, larger where unemployment rose faster, and smaller where active labour-market policy and social protection were stronger. That last finding is Durkheim's argument about regulative institutions, tested.

Wars. The wartime decline Durkheim identified has been observed repeatedly since, including in both world wars.

And the additions he did not have. Means restriction is one of the best-evidenced interventions in the field — the removal of coal gas from British domestic supply in the 1960s and 70s, barriers on bridges, restrictions on pesticides in South and East Asia. Media reporting guidelines have measurable effects in both directions. Treatment of depression and reduction of alcohol consumption matter substantially. None of this is in Durkheim, and all of it is real. A complete account needs the individual, clinical and situational levels alongside the social one — which is a correction to Durkheim's own insistence that psychology be excluded, and a reminder that a founder's methodological rules can be too strict for their own subject.

"Durkheim proved suicide is caused by society." He showed that rates vary systematically with social conditions. Individual deaths have individual causes; the two levels are different questions.

"Durkheim's numbers were sound." His data came from nineteenth-century official records produced by processes that varied by religion, jurisdiction and period. Douglas and Atkinson's critique is serious.

"Protestants are more suicidal than Catholics." Durkheim's inference from area data to individuals is the ecological fallacy. And the difference may partly reflect recording practice.

"Altruistic suicide is a thing of the past." The category has been applied to military contexts and to suicide attacks, though that application is contested and the psychological evidence on perpetrators is mixed.

"Fatalistic suicide was a major Durkheimian type." It occupies a footnote and he thought it unimportant. Later scholars have found it far more useful than he did.

"Sociology has replaced psychiatry here." No. Means restriction, treatment of depression, alcohol policy and crisis support all have strong evidence behind them. The social level explains rates; it does not exhaust the subject.

Why it matters

It is the proof of concept for the entire discipline. Take the most private act imaginable, show that its rate is patterned, show the patterns track social conditions, eliminate the rival explanations, and you have demonstrated that social facts exist and can be studied scientifically. Every subsequent claim that "this apparently personal thing has social causes" (1.1.1) descends from this book.

It established the template for quantitative sociology: a rate as the dependent variable, systematic elimination of rival explanations, comparison across groups and periods, and a theory that generates checkable predictions rather than post-hoc interpretations.

Its criticism is as valuable as the work. Douglas and Atkinson's demonstration that official statistics are institutional products is one of the most useful things any sociologist has established, and it applies to every dataset anyone will ever use.

And it shows that a founding text can be both flawed and indispensable. The data are questionable, the ecological inference is invalid, the ethnography is colonial, and the gender analysis is bad. It is still the book that made the discipline possible, and its central finding has been confirmed many times over with better evidence. Holding those facts together — without either dismissing or defending — is what mature reading of a classic looks like.

Le Suicide (1897) is the first major work of empirical sociology. Its argument: individual suicides have individual causes, but the rate is stable within societies and varies systematically between them, and therefore is a social fact requiring social explanation.

Durkheim first eliminates the rivals — mental illness, heredity, race, climate, imitation — then explains the rate by two variables, each with an optimum:

Egoistic (too little integration) — explains the Protestant/Catholic, married/single and parent/childless gradients, and the fall in rates during wars. The mechanism is degree of integration, not doctrine. Altruistic (too much integration) — military suicide and ritual sacrifice, though his non-European examples come from unreliable colonial sources. Anomic (too little regulation) — rates rising in booms as well as slumps. Fatalistic (too much regulation) — a footnote he thought unimportant, later found far more useful.

The great criticism: Douglas and Atkinson showed that a suicide statistic is the product of an institutional process — and that Catholic societies had strong reasons to record ambiguous deaths as accidents. This is serious. What survives it: gradients that recording cannot explain, and modern cross-national differences in well-registered data.

Also: the ecological fallacy (inferring about individuals from area data — named using this very book), circularity, and neglect of meaning.

Modern evidence supports the shape of the argument — integration protects, disruption raises rates, wars lower them, and social protection moderates recession effects — while adding what Durkheim excluded: means restriction, treatment, alcohol policy and media guidelines.

Suicide rate — the number of suicides per unit of population; a social fact, not a tally of individual decisions.

Integration — the strength of an individual's bonds to a group.

Regulation — the extent to which a society sets limits on conduct and aspiration.

Egoistic suicide — arising from insufficient integration.

Altruistic suicide — arising from excessive integration; obligatory, optional or acute.

Anomic suicide — arising from insufficient regulation.

Fatalistic suicide — arising from excessive regulation and blocked futures.

Ecological fallacy — inferring properties of individuals from data about groups or areas.

Social construction of statistics — Douglas's and Atkinson's argument that official figures are products of institutional decision processes.

Means restriction — reducing access to methods; among the best-evidenced modern preventive interventions.

One — check a cross-national table. Find current WHO age-standardised suicide rates by country. The variation is severalfold and persistent. Ask what social differences could produce it — and what recording differences might still remain.

Two — apply the two variables. Take any group you know — a workplace, a congregation, a team, an online community — and rate it on integration and regulation. Note that Durkheim says both extremes are dangerous, which is a more interesting claim than "community is good."

Three — interrogate a statistic. Pick any official figure (crime, unemployment, hate crime, waiting lists) and find out exactly who records it, when, under what definition and with what incentives. This is Atkinson's method, and it is the single most useful habit in quantitative work.

Four — spot the ecological fallacy. Find a news story inferring about individuals from area-level data ("areas with more X have more Y, so X-people do Y"). They are extremely common once you look.

Five — hold both readings. Write one paragraph on why Suicide deserves its reputation and one on why its evidence cannot be trusted as it stands. Being able to write both is the skill this lesson is really teaching.