To claim territory for a new discipline, Durkheim picked the least sociological subject available: suicide.
That was not coldness, it was tactics. If even this — the most private thing, the thing that looks most like something going wrong inside one person — can be systematically explained by social conditions, then the claim he actually wanted to establish will hold: there is a class of facts that is neither in your head nor in your body, and that still holds your hand down.
Psychology had already taken individual consciousness, biology had taken the body. What was left for sociology?
Durkheim's answer: there is a class of phenomena whose carrier is not any person but the relations between people — language, law, money, marriage custom, fashion. He called them social facts (faits sociaux) and gave two testable marks: exteriority (they existed before you) and constraint (breaking them meets resistance, from legal sanction to an awkward look).
Language is the cleanest case. You did not help write the grammar of your language; it was there before you were born. You may of course break it, and the cost is that nobody understands what you said — that cost is the entire content of "constraint", and it needs no enforcer. Money is the same: the purchasing power of a piece of paper is not in the paper but in the state of everyone believing that everyone else believes in it.
The step that really matters is not "society influences individuals", which everyone agrees with. It is the methodological move that follows: study social facts as things — do not ask people why they say they do it, ask under what conditions this practice becomes more common and under what conditions it becomes rarer.
How can something be exterior and constraining without anyone giving orders? Because a social fact is a self-sustaining equilibrium: each person conforms because they expect the others to, and each person's conformity constitutes everyone else's expectation. The rule is in no one's head; it lives in the loop of heads pointing at one another. Which also explains why it is so hard to change — the cost of deviating falls on the deviator alone, while the benefit only arrives if everyone moves together.
What a company calls its "culture" is mostly this: written into no policy document, yet a new hire learns within weeks what hour it is acceptable to message and what must not be said in a meeting. It is exterior and it constrains, which is why changing it is far harder than changing a process.
On evidence: this is a disciplinary programme, not an empirical finding. Its empirical footing was nineteenth-century moral statistics — people had already noticed that crime rates and marriage rates were startlingly stable year to year, and Durkheim set out from precisely that contrast between unpredictable individuals and very steady totals. The most persistent criticism is that it treats society as an entity standing above individuals, which slides easily into circular explanation. Most people now take a middle position: the emergent pattern is real, but you owe an account of which individual interactions generate it — irreducible does not mean unexplainable.
Durkheim never argued that society causes a particular person's suicide. What he set out to show was different: under different social conditions, suicide rates differ stably.
From official statistics across Europe he reported several differences: Protestant regions above Catholic ones; unmarried above married, and those with children lower still; rates falling during wars and major political crises. His explanation was consistent throughout — differences in integration. Catholicism wove believers into denser shared ritual and doctrine, while Protestantism handed interpretation back to the individual; marriage and child-rearing lock a person into a web of obligations; war manufactures a temporary but intense collective belonging.
How much of this survives? It has to be read in three layers, and the layers do not agree.
The methodological legacy: using group-level rates for causal inference. Enormously influential, and uncontested.
The original data: substantially challenged. One study used individual death registrations from the Netherlands for 1905–1910, which recorded the deceased's religion, and found that Catholic deaths were far more often filed under "sudden death", "accident" or "cause unknown" — suggesting the gap between the denominations was to a considerable degree an artefact of registration practice. That challenge was itself rebutted, and the argument is unresolved. A stronger version says official suicide statistics are products of coroners' classification work in the first place, so using them to study social factors risks circularity.
The core correlations: the associations between marital status, religious participation and suicide rates have recurred across more than a century and dozens of countries, and the direction is robust. But moving from the correlation to "integration is what is doing the work" is far less secure than the correlation itself.
One more general lesson grows out of this, called the ecological fallacy: a correlation at group level cannot be carried down to individuals. "Regions with more Protestants have higher suicide rates" does not imply "most people who die by suicide are Protestant" — the high rate could come from the Catholic minority in those regions. Durkheim's reasoning is the most famous case of this inference. It does not refute his conclusions; it draws a clear line around what was actually shown.
This is the most counterintuitive finding in the book, and the doorway to his second concept.
That rates rise when an economy collapses is easy to understand. But he noticed that rates rise in sudden booms as well. If the suffering came from poverty, that makes no sense.
His explanation is anomie (anomie) — literally "without norm", though it does not mean "no rules". It means this: desire has no internal stopping point, and what stops it is always an external yardstick — how much counts as enough in the position you occupy, what counts as respectable, who it is reasonable to compare yourself with. That yardstick is maintained collectively by the people around you, and in ordinary times you have no idea it is there.
Violent change in either direction shatters it. Your position moves suddenly, the old frame of reference stops working and no new one has grown; desire loses its scale, becomes unbounded, and can therefore never be satisfied. What breaks people is not being poor, it is losing the means of measuring "enough".
The easiest place to recognise this today is any setting where income or feedback has no ceiling and no shared reference: the six months after a reorganisation, when seniority, career paths and the meaning of "doing well" have all been reshuffled; or work whose pay is set entirely by opaque distribution rules, so that neither a rise nor a fall comes with an explanation. What hurts is often not the number but that nobody can tell you what counts as normal.
Be explicit about evidence: this layer is weaker than the previous one. The original evidence for booms raising suicide rates comes from nineteenth-century statistics, coarsely defined and poorly comparable; and anomie is extremely hard to measure independently — how to quantify integration and regulation separately has no agreed answer, and the four suicide types are often inoperable on real data. The concept earns its keep through the question it poses — what gives desire its scale — rather than through the numbers it came with.