You ask a neighbour to take in a parcel for you. Try to press twenty on him and he will probably be annoyed — not because it is too little, but because paying closes the matter, while what he was holding was a small unfinished obligation of yours.
Economics textbooks open with a different story: long ago people bartered, it was clumsy, so money was invented. Anthropologists have looked for over a hundred years and have not found that society. This page takes apart three things: where that story goes wrong, why still owing holds a relationship together better than settling up, and what the earliest money was actually for.
The textbook story runs roughly like this: the baker wants meat, the butcher wants shoes, the two do not match up, so everyone settles on some commonly wanted thing as an intermediary. Adam Smith wrote a version of it in 1776 and it has been retold for more than two centuries since — yet that society has never been found. The Cambridge anthropologist Caroline Humphrey put it bluntly in 1985: no example of a barter economy, pure and simple, has ever been described, let alone the emergence of money from it. Her paper was titled Barter and Economic Disintegration, and her observation ran the other way: large-scale barter typically appears after a monetary system has collapsed. It is a product of breakdown, not a starting point.
The mechanism is not hard to see. Barter requires a double coincidence of wants: I have what you want, you happen to have what I want, and it has to be at the same moment. But people who know each other do not need that coincidence at all — take the meat, we will remember. Credit does not need a double coincidence, only that the two of you will meet again.
The archaeology gives the same order. Mesopotamian temples and palaces were keeping accounts on clay from around 3000 BCE: the shekel, a weight of silver, served as the unit of reckoning, grain and silver had conventional conversion rates, and loans and interest were routine. The earliest struck coins appear only in late seventh-century BCE Lydia, in what is now western Turkey, made of a gold-silver alloy. Written debt precedes minted coin by more than two thousand years, and many of those tablets involved no metal passing hands at all, only a record of who owed whom how much.
Evidence strength, stated plainly: Humphrey's claim is a negative one, and not found is not the same as does not exist. David Graeber's Debt: The First 5000 Years (2011) pushed the argument very broadly, and specialists have faulted a good many of its particular historical claims, so the two should be read apart; but the main line — that the barter myth does not hold and credit precedes coinage — is not much contested.
Back to the parcel. You offer the neighbour twenty and he is annoyed. He was holding a small obligation of yours, and that implied there would be a next time; pay, and the thread is cut. This is the most durable line in Graeber's book: a debt is an exchange not yet completed, and it is precisely the incompleteness that keeps the relationship running. The piece on gifts and reciprocity covered why giving outright binds harder than trading does; this is the same mechanism from the other end — anything that can be settled in one stroke is, once settled, no longer owed, and there is no longer any need to meet.
There is a weighty field case. In Evans-Pritchard's 1940 ethnography of the Nuer of South Sudan, a killing is settled in cattle — a substantial number, typically around forty head — mediated by a leopard-skin chief who belongs to neither side. The point is not what a life converts to in cattle. Those cattle were commonly used afterwards as bridewealth, so two families who would otherwise have gone on killing each other became affines instead. What compensation does is not conclude anything; it converts a relation that would keep killing into a relation that will keep circulating.
So when does a debt turn into something else? This is Graeber's decisive step: for an obligation to become a debt, three things have to be in place — a quantified amount, a determinate creditor, and a moment at which it can be discharged. With all three, it can be transferred, sold, and handed to a third party for enforcement. The favour you owe your neighbour cannot be assigned to someone else to collect; it is bound up with its being him. A sum written down can be, and whoever holds the note is the creditor. Quantification does not merely record the obligation clearly, it detaches it from this particular relationship.
The criticism this most deserves is that it slides easily into treating measurement itself as corruption. The other side is just as strong: a vague ledger of favours is exactly what can be drawn on without limit, because no figure is there to say enough; a written amount is, for the one who owes it, a protection, because it supplies an endpoint. And historical debt cancellations — Mesopotamian clean-slate edicts, the Jubilee in the Jewish tradition — work only because amounts are countable. Each reading is half right, and no side is taken here. Evidence strength: ethnographic cases plus historical textual interpretation, with no available experiment.
What is a finger worth? The question sounds offensive, yet nearly every early law code answered it in earnest. The Law of Æthelberht of Kent, from around 600 CE and the earliest written law surviving in English, tariffs the items one by one: so much for a knocked-out tooth, so much for a severed finger, so much for striking an ear deaf. Germanic law even had a word for the whole idea, wergild, which means literally man-price.
The numismatist Philip Grierson argued in 1977 that compensation tariffs of this kind are a likely early source of the very idea of a scale of value. Their business is not trade but stopping a feud — for two families to negotiate at all, there has to be a schedule both accept. Anthropologists call such means of payment social currencies: used for marriage, compensation and fines, but never for buying tonight's dinner.
A fuller field case comes from central Nigeria. Working among the Tiv in the 1950s, Paul Bohannan described a set of ranked spheres of exchange: at the bottom, subsistence — yams, grain, small livestock, tools; in the middle, prestige goods — brass rods, cattle, white cloth, and historically people held as property; at the top, rights in marriage. Exchange within a sphere was everyday, converting upward was an achievement and converting downward a disgrace. Between spheres there was no rate at all — and precisely because there was none, nobody could sell enough grain to buy the right to marry.
When general-purpose money arrived under colonial rule, the walls came down. Bohannan's 1959 paper is about the consequences: everything landed on one scale, grain converted to money, money to brass rods, brass rods toward the marriage sphere. The young gained a route upward, and elders whose standing rested on controlling the upper goods were hollowed out. The decisive property of general-purpose money is that it is general, not that it is convenient — it places classes of things that were incommensurable onto a single axis, and that step is social rather than technical.
The criticism belongs here too: the three-sphere model has since been faulted as too tidy, with later work holding that Tiv conversions were in practice more flexible and that monetisation partly predated colonial rule. It is powerful as a schema and should be discounted as history. The same goes for reading it forward: you do not transfer money to a friend who helped you move, nor put a price on a child's grades — those walls are still standing; converting attention, time spent queuing, or a recommendation into points is among the ones just taken down. Which walls ought to stand is not answered on this page. It states only something observable: remove a wall and flows that were blocked really do happen, and whoever was standing on that wall really does lose their footing.