What Actually Falls When a Civilisation Falls

Over two hundred causes have been proposed for the fall of Rome. That number is not evidence of thorough scholarship. It is evidence that the question is badly posed.

The trouble with single-cause explanations is easy to state and hard to get round: the proposed cause is almost always present in periods when nothing collapsed. Droughts, plagues, invasions, currency debasement, inequality, an oversupply of ambitious elites: all of these appear repeatedly in the histories of societies that carried on regardless. A factor present in the survivors as well as the casualties is not doing the explanatory work, however satisfying it is to name.

Joseph Tainter's reframing is the most useful thing written on the subject, and it starts by defining the term. Collapse, he says, is a rapid loss of an established level of sociopolitical complexity. Not destruction, not depopulation, not the end of a culture: a simplification of structure, happening fast.

Then he asks what complexity is for. Societies add it to solve problems: more administrative layers, more specialists, more infrastructure, a larger army, a more elaborate legal system. Each addition is a rational response to something real. But complexity is subject to diminishing returns like anything else (the first layer of bureaucracy converts chaos into order, the fifth coordinates the fourth) and it is never free. Costs accumulate and are rarely shed when the original problem passes.

Eventually the marginal return on further complexity goes negative. At that point a simpler arrangement is genuinely better for most people living inside the system, and collapse is not only a catastrophe visited from outside but sometimes a rational choice made from within, by people declining to keep paying for a structure that has stopped delivering.

This is an economic argument and it is testable against fiscal records, which is what distinguishes it from moralising about decadence.

It also changes who you think collapse happens to. The evidence on ordinary people is genuinely mixed, and where it is clear it often runs against intuition. After the western Roman state receded, some rural populations in Britain and Gaul show improved skeletal indicators (taller, better nourished) as tax obligations, rents and conscription lapsed. James Scott's work documents populations across Southeast Asia deliberately moving beyond the reach of states, adopting crops and settlement patterns designed to be hard to tax.

What collapses first and most visibly is the elite superstructure: the monuments, the inscriptions, the imported luxuries, the writing. That is also precisely what archaeology detects best. So the record is systematically biased toward the experience of the people who lost most, and the phrase "dark age" describes the quantity of surviving documentation at least as much as the quality of anybody's life.

The standing example of genuine, interlocking collapse is the Late Bronze Age. Within about fifty years around 1200 BCE, the Mycenaean palaces, the Hittite empire, Ugarit and the whole eastern Mediterranean palace system ended. Linear B literacy disappeared from Greece for four centuries.

No single cause holds. There is now solid pollen and isotope evidence for prolonged drought. Egyptian sources describe the Sea Peoples. There are earthquakes in the record. But drought, migration and earthquakes happen elsewhere without this result, and what made the difference was probably the coupling: these were palace redistribution economies, each dependent on the others, running a bronze technology whose tin came from thousands of kilometres away along routes any one of these shocks could cut. Tight coupling turns a local failure into a general one. That is a claim about structure rather than about any particular disaster.

And then there is the critique that has reshaped the field, which is that collapse is frequently reorganisation misread.

The Maya are the case. The southern lowland cities were abandoned over roughly a century and a half, monumental construction stopped, and the inscriptions ceased. In the popular telling the Maya vanished. There are around seven million Maya people alive today, speaking some thirty Mayan languages. What ended was a particular political form (the divine-king city-state) and populations moved north, where Maya cities were still flourishing when the Spanish arrived.

The archaeological signal reads as disappearance because the things that stop are the things that leave durable traces. Migration, reorganisation and life at a smaller scale leave fainter ones. Which is worth holding on to as a general rule about this evidence: when a record goes quiet, the first question is whether the people went away, or whether only the writing did.