Herd dispersal among kin

Why do herders scatter their animals among distant kin instead of keeping the herd together where they can watch it?

A herder's cattle are his entire capital: his food, his bride payments, his claim to be somebody. You would expect him to keep them where he can see them. Instead, across the East African savanna and the Sahel, ethnographers keep finding the opposite — a man's animals parcelled out in twos and threes to households a day's walk away or several, and the beasts in his own kraal partly belonging to somebody else. He has given up daily control over most of his wealth and taken on the daily work of somebody else's. What is being bought?

Start with the thing that can actually destroy a herd. Not slow attrition — that you plan for. The dangers are lumpy: a raid takes the animals standing in one place on one night; a rinderpest or contagious pleuropneumonia outbreak sweeps the animals sharing one watering point; a failed rain scorches one stretch of range while a neighbouring stretch, fifty kilometres off, greens up.

Every one of those is local. And that is the hinge of the whole account, so it deserves testing rather than assuming: if losses were general — a shock that flattens the whole region at once — then splitting the herd across it would change nothing whatever. Scattering only pays where the map is patchy.

Is it? Partly, and honestly not always. Rainfall on semi-arid rangeland is incoherent over short distances; one valley gets a storm and the next does not. Raiding is aimed at one homestead. Disease spreads along contact, which is spatial. But a continental drought or a rinderpest panzootic hits everyone, and against those the practice buys precisely nothing.

Now notice something a farmer scattering his fields cannot do, which is where this stops being a variation on that theme. Land cannot move. Animals can, they keep producing while they are away, and — the decisive point — they can be lent without being given. The owner keeps ownership. The holder gets the milk, the traction, the manure, the standing of a larger visible herd. The owner gets his animals grazing on somebody else's range, watched by somebody with an interest in them, out of the path of whatever comes for his own.

So the arrangement is doing two things at once, and they are easy to confuse. The first is spreading exposure before the shock. The second is the claim itself: after a wipeout, the man with nothing left walks his network and asks for the animals he placed and the calves born of them. He is not begging; he is calling in something that was always his.

Why, then, are the partners so often distant — other clans, other sections, men met at a market? Because a neighbour's herd dies in the same drought as yours. A partner whose fortunes move independently of yours is worth more than a close one whose fortunes do not, and building such a tie takes years of small gifts and visits precisely because there is no kinship to lean on.

I should not oversell the credit reading. The Maasai relationship osotua — an "umbilical cord" partnership — involves requests made in genuine need, given if the giver can, and, on the ethnography, not accounted or repaid on a schedule; Lee Cronk and colleagues argue this is need-based transfer rather than debt, and that modelling it as reciprocal exchange gets its logic wrong. Elsewhere — Tswana mafisa, Fulbe and Somali arrangements — the loan looks more like an asset a patron can recall, and shades into clientage. Different institutions in similar clothes, and I cannot tell you which is the general case.

Two things about the evidence itself. Nearly all of it comes from a handful of twentieth-century fieldworkers among a handful of peoples, so the "cross-cultural pattern" is thinner than it sounds. And the herd counts were gathered from owners with excellent reasons to understate what they held and to whom — taxation, raiding, in-laws with claims — so the record holds the numbers people were willing to say aloud.

The word "scatter" makes this sound like diversification and nothing more. Diversification is only the first half. The second half is that a placed animal is a relationship kept warm — a reason to visit, a stake in each other's survival, a standing obligation that can be activated in a year when it matters. A household that placed nothing has nowhere to ask. That is also why the practice does not melt away once a herder is rich enough to absorb a loss himself: withdrawing from the network means losing it, and the network is what he will need on the occasion his own wealth is not enough.

What would refute the account? It predicts that placements go preferentially where fortunes are uncorrelated with the owner's, that they are heaviest among households nearest the herd size below which a family cannot rebuild, and that they thin as veterinary services, fenced ranches and livestock markets substitute for them. The refuting observations are therefore available: dispersal concentrated among immediate neighbours grazing the same water, whose herds all fail together; or placements unrelated to a household's vulnerability; or dispersal surviving unchanged where reliable sale and restocking markets have arrived. On the record I know, it does thin under commercialisation — but that record is qualitative, and I would not put a figure on it.

• How these placement networks behave when a cash market for livestock arrives and the animals become sellable at a moment's notice.
• Why some of these relationships harden into clientage while others stay between equals.

Every term the collection defines is gathered in the glossary.