Scattered strip farming

Why did villagers work many scattered strips instead of one consolidated field?

Across much of medieval and early modern Europe, a household's land was not a field. It was a scatter of narrow strips, often dozens of them, spread through the village's open fields — one in the valley bottom, one on the mid slope, one at the frosty upper edge — and the farmer walked between them all season. Consolidation looks so obviously better that reformers spent centuries saying so. So either these villagers were irrational for a very long time, or the scatter bought something the reformers were not counting.

Begin by taking the cost seriously, because a weak case for the practice is worse than none. Walking between strips consumed working time and cart movements. Turning a plough team needed headlands, and boundary baulks stood uncultivated, so land was lost outright. No strip could be fenced, so crops and rotations had to be timed with everyone else's, since the village grazed its animals on the stubble after harvest. And encroachment — a ploughman taking a furrow's width from his neighbour — was a permanent source of dispute. These losses are real and they recur every year.

So what could be worth that? Ask what a bad harvest actually meant. Not a poor return — hunger, eating the seed corn, borrowing against next year, at worst losing the holding or a child. Which tells you the objective was not to maximise average yield but to keep the probability of falling below the subsistence line small. Those are different problems, and they routinely recommend opposite actions.

Now the step the whole account depends on: for scattering to reduce that probability, harvest failures must be patchy within the village. Are they? Some are. Hail falls in swathes a few hundred metres wide. Late frost drains into hollows and leaves the slope untouched. A wet spring drowns the valley bottom while well-drained ground does fine, and a dry summer scorches the thin upper soils first. Other shocks are not patchy at all — a general drought, a war, a village-wide epizootic in the plough oxen — and against those, scattering buys precisely nothing.

The arithmetic constrains the answer, so it is worth doing. If a household holds n plots whose failures are independent and equally variable, the variance of its average yield falls as 1 divided by n, so the spread falls as 1 over the square root of n: four plots halve it, nine cut it to a third. Two consequences follow. The gain from a second plot is large and from a twentieth nearly nothing, while the walking cost of each extra plot is roughly constant — so the model predicts moderate scattering, not maximal, and more of it where travel is cheap and the margin above subsistence thin. And independence is far too generous: with any positive correlation between plots the variance stops falling at a floor set by that correlation, however many strips you hold, so the real reduction was much smaller than the square-root rule suggests.

Two refinements, then a serious challenge to the whole account.

The first is that a strip system spreads local shocks only, which is why villages ran storage, common rights and mutual credit as well. Scattering is one layer of a stack, not the strategy. The second is that it was not a private choice: the open field was a set of collective rules, and a household could not consolidate alone, because its block would still be crossed and grazed. Hence enclosure had to be a village-level or parliamentary act.

What should refute the risk account? It predicts that yields on plots within a village are imperfectly correlated, measurably so; that scattering should be heaviest among households closest to subsistence; and that it should be abandoned as substitutes arrive — grain markets, storage, credit, wage work. The refuting observations are therefore available: heavy scattering on uniform terrain where shocks are village-wide, or scattering uncorrelated with how vulnerable a household is, or scattering that survives untouched once markets and storage make it redundant.

And now the challenge. The risk explanation, argued most forcefully by McCloskey, is not the only one on the table and the debate is genuinely unsettled. Scatter is also produced without anyone choosing it, by partible inheritance and an active land market subdividing and exchanging parcels over generations. It is also produced by the plough: a heavy team was assembled from several households' oxen, and a day's collective ploughing distributed among the contributors comes out as strips — Fenoaltea pressed this labour-and-asset-sharing reading against the risk one. And there is a plain equity argument, that giving every household a share of each soil quality is what a village dividing land would naturally produce. These are not mutually exclusive, and I cannot tell you which dominates. Nor will I give a figure for how much variance scattering removed or what it cost in yield: published estimates vary enough with the assumed correlation structure that quoting one would report a modelling choice as a finding.

• How enclosure actually changed yields and employment, as against what its advocates claimed it would do.
• Why storage, common grazing and village credit each cover a different part of the risk scattering leaves untouched.

Every term the collection defines is gathered in the glossary.