The Oldest Question

1.0 What this chapter gives you#

1. You will be able to explain why a payment is an edit to a record rather than the movement of an object, using Ada’s notebook as the worked example.
2. You will be able to state Caroline Humphrey’s claim precisely — that no barter economy, pure and simple, has ever been described — and say what it does not claim, so that you are not caught out by anyone who has read the source.
3. You will be able to describe how a tally stick worked, why splitting a notched stick lengthways made it hard to forge, and how the words stock and foil reached modern finance from a piece of hazel.
4. You will be able to place coinage correctly in time: administrative writing at Uruk from about 3100 BC, the first coins some two and a half thousand years later beneath the temple at Ephesus.
5. You will be able to explain why the earliest coins undermine the commodity theory of money rather than supporting it, given that Lydian electrum of roughly 54 per cent gold was tariffed as though it were 73 per cent.
6. You will be able to distinguish the unit in which an obligation is denominated from the asset in which it is settled, and recognise that distinction as four thousand years old.
7. You will be able to say what actually makes a record money — that an identifiable party is obliged to honour it, and that something makes the obligation stick — rather than repeating the slogan that money is information.
8. You will be able to name the counterparty behind any unit of money you hold: your bank for a deposit, the Bank of England for a banknote, the retailer for a gift card.
9. You will be able to quote the British proportions that matter: about £91.5 billion of notes across roughly 4.98 billion of them, against bank deposits making up 97 per cent of the money in circulation.
10. You will be able to read a hundred-character Bacs Standard 18 record and explain why three of its fields exist purely so that human beings on both sides can later agree what the record meant.

Ask a hundred people what happened when they last paid for something and ninety-nine will describe a movement. Money left my account and went to theirs. Money came out of the machine. Money changed hands.

Almost none of that is true, and the exceptions are rarer than you think. When you tap a card in a shop in Leeds, nothing physical travels between you and the shopkeeper. When your salary arrives on the twenty-fifth, no lorry of banknotes crosses the country. In every one of these cases, what changed was a record. Somebody’s obligation to somebody else was written down, cancelled, netted off, or handed on. The physical object, when there is one, is a token that stands for the record. The record is the money.

This is not a modern discovery about a modern system. It is the oldest fact about money there is, and the evidence for it is older than coins and older than writing. This chapter is about that evidence: the clay of Mesopotamia, the split hazel sticks of the English Exchequer, and the awkward, well-documented fact that the barter economy every schoolchild is taught about has never actually been found.

The reason this matters for a book about payments is simple. If money were a substance, moving it would be a logistics problem, and the interesting questions would be about vaults and armoured vans. Because money is a record, moving it is an information problem, and the interesting questions are about who keeps the record, how two records are reconciled, what happens when they disagree, who bears the loss when one is wrong, and how quickly the thing becomes final. That is what the rest of this book is about. This chapter is about why.

The plain version#

Imagine a village with one shop.

The shop is run by a woman called Ada. She sells bread, milk, nails, batteries and the sort of things you need at short notice. Nobody in the village has much cash, because the nearest bank is forty minutes away, so Ada keeps a notebook under the counter. When Tom the carpenter comes in for a loaf, Ada writes: Tom, 1 loaf, £1.20. When Tom finishes a set of shelves for Ada’s back room, she writes: Ada owes Tom, £150.

Nothing has moved. There are no coins in this story yet. But something real has happened. Tom is owed £150 by Ada. Tom owes Ada £1.20. If Tom keeps buying bread every day for four months, the two numbers will meet in the middle and cancel each other out, and the whole relationship will settle without a single coin ever appearing.

Now watch what happens when the village gets a little bigger.

Sam the mechanic fixes Tom’s van. The bill is £80. Tom has no cash. But Tom knows that Ada owes him £150, so he says to Sam: “Go to Ada. Tell her to take £80 off what she owes me and put it against your account instead.” Sam agrees, because Sam trusts Ada’s notebook, and because Sam owes Ada money anyway for a year of milk and nails. Ada opens the book and writes three lines. Tom’s credit falls from £150 to £70. Sam’s debt to Ada falls by £80. Tom’s debt to Sam is gone.

Look at what just occurred. A debt was paid. Nobody handed over anything. No object changed owners. A record was edited, and everyone accepted the edit because everyone accepted the record-keeper. This is a payment. It is the whole of payment, in miniature, and every payment system in the world is a more elaborate version of Ada’s notebook with more checks, more speed and more lawyers.

Now here is the part that surprises people. This is not a clever modern shortcut that people invented once coins became inconvenient. As far as anyone can tell from the archaeological and anthropological record, this is the original arrangement. Ada’s notebook came first. Coins came very much later.

The story you were probably told#

The version most of us learn goes like this. Long ago, people bartered. If you had shoes and wanted bread, you had to find a baker who happened to want shoes. This was maddening, so somebody hit on the idea of a thing everyone would accept — cattle, salt, shells, eventually lumps of silver — and money was born to fix the problem of barter.

It is a beautiful story. Adam Smith told a version of it in 1776 in The Wealth of Nations, using the example of a butcher with more meat than he can eat and a brewer and a baker who want some of it but have nothing the butcher happens to need. Smith’s conclusion was that every prudent person would start keeping a stock of “some one commodity or other, such as he imagined few people would be likely to refuse in exchange”, and that metals eventually won because they do not rot and can be cut into precise amounts.

The trouble with the story is that when anthropologists went looking for a society that actually worked this way, they could not find one. Not a single one. What they found instead, over and over, in every corner of the world, was some version of Ada’s notebook: people who knew each other, kept mental or written track of who owed what, and settled up occasionally or never. Where they found barter, they usually found it happening between strangers, or between people who had no shared record-keeper and no intention of building a relationship, or in places where a money system had recently fallen apart.

The reason is not hard to see once you look at Ada’s village. Barter is what you do when you cannot trust a record. If you and I are neighbours, a running tab is easier for both of us than lugging goods around and haggling over exchange rates. It is only when I do not expect to see you again, or cannot enforce anything against you, that we both insist on a simultaneous swap where neither of us has to trust the other for a single day.

Two copies of one fact#

The obvious weakness of Ada’s notebook is Ada. What if she writes the wrong number, or rubs a line out, or dies and leaves the book to a nephew who reads Tom’s £150 as £15?

Human beings solved this a very long time ago, and one of the neatest solutions was a stick.

Take a piece of hazel or willow, about the length of your hand, and cut notches across it. A wide notch means a large amount, a narrow notch a small amount. Now — and this is the clever part — split the stick lengthways, right through every notch, so that you end up with two halves that each carry the same notches, along the same grain, with the same knots and flaws.

Give one half to the person who is owed the money. Keep the other half yourself. Neither half can be altered without the alteration being obvious, because if either of you adds or removes a notch the two halves will no longer match. To settle the debt, the two halves are brought back together. If they fit, the record is proved. If they do not, somebody is lying, and the wood says so.

That is a tally stick. Two copies of one fact, made from a single piece of wood so that no forger can produce a matching partner. The English Exchequer — the government department that collected the king’s revenue — used them for something like seven hundred years, and the vocabulary has stuck to English so firmly that we no longer notice it. The longer half of the split stick, held by the person owed the money, was called the stock. That is why an investor holds stock, and why a person who lends to a company is a stockholder. The shorter half was the foil, held by the debtor, which is why getting a raw deal is getting the short end of the stick.

And here is the point that ought to stop you. A tally stick was not a receipt in the modern sense, a dead scrap of paper filed away. It circulated. If the Exchequer owed you £100 and gave you a tally saying so, you could pay your own creditor by handing over the tally. He would accept it because he, in turn, could hand it to the taxman and have it counted against his own bill. The stick moved from hand to hand carrying an obligation with it, and it did this for centuries. It was, in every practical sense, money. It was also, unmistakably, a record. Not gold, not silver, not a substance of any kind. A notched stick, whose entire value came from the fact that a specific institution had written down that it owed you something.

The clay before the wood#

Go back much further and you find the same idea in a different material.

In the cities of southern Mesopotamia — modern-day Iraq — around five thousand years ago, people were running large organised economies centred on temples and palaces. Grain came in from the fields. Rations went out to workers. Sheep were counted, beer was issued, labour was allocated. All of this had to be tracked, and the tracking was done in clay, because clay was free and it lasted.

The earliest of these records are not sentences. They are not literature or prayer or law. They are accounts. They say things like: this many measures of barley, from this store, to this person. One clay tablet in the British Museum, made in the very last centuries of the fourth millennium BC, is a record of beer, marked with five different kinds of number sign. It is about the size of a postcard. It is not a poem. It is a ledger entry.

This is worth sitting with, because it turns the usual story of civilisation upside down. Writing was not invented so that people could tell stories and later, as an afterthought, applied to business. On the best current evidence, it went the other way. The pressure that produced writing was the pressure of keeping track of who owed what to whom. The alphabet is a side effect of accountancy.

And how long after this did coins arrive? Roughly two and a half thousand years. The first objects most people would recognise as coins — small, standardised, stamped lumps of precious metal — appear in what is now western Turkey around the end of the seventh century BC. Ninety-three of them were dug up by British Museum archaeologists beneath the great temple of Artemis at Ephesus, left there as offerings.

Put that gap in terms you can feel. If the clay ledgers of Uruk had been written on the day of the Battle of Hastings, coins would not have been invented until about four hundred years from now. Writing down debts is not a step on the road to coins. Coins are a late refinement on a system of written obligation that was already ancient when they turned up.

What money is, in one sentence#

Every form of money you have ever handled is a record of an obligation before it is an object.

The five-pound note in your pocket says so on its face, in words most people have stopped reading: I promise to pay the bearer on demand the sum of five pounds. That is not decoration. It is a description of what the note is. The note is the Bank of England’s IOU. The bank owes you; the note is the evidence.

The number in your banking app is the same thing one step further along. It is not a store of your pounds sitting in a box with your name on it. It is your bank’s written admission that it owes you that amount, payable on demand. If you have ever wondered why bank failures are frightening in a way that losing a wallet is not, this is why. A wallet is a container. A bank balance is a promise.

And the proportions are not close. At the end of February 2026, all the Bank of England banknotes in existence — every fiver, tenner, twenty and fifty in every till, pocket and cash centre in the country — came to about £91.5 billion, spread across roughly 4.98 billion notes. Set that against the money that exists only as entries in bank ledgers and it is a rounding error. The Bank of England’s own analysis puts bank deposits at 97 per cent of the money in circulation. Ninety-seven pounds in every hundred is a line in a database. Three are paper.

Now look at what that money does in a year. In 2025, the Bacs system — the one that pays most British salaries and collects most British household bills — processed 6.86 billion payments worth £6.05 trillion. Faster Payments, the system behind the instant transfer you make from your phone, handled 5.55 billion payments worth £4.84 trillion. That is well over twelve billion instructions in a single year, moving nearly eleven trillion pounds, and at no point did a physical object representing value travel from a payer to a payee. What travelled was information: a short, structured message saying reduce this account by this much, increase that account by this much, and here is the reference so both sides can agree it happened.

Ada’s notebook, scaled to sixty-eight million people.

Where the plain version stops being true#

The plain version above is a good map. Like all good maps it lies by simplification, and four of those lies matter enough to correct before we build anything on top of them.

Barter did exist; what has never been found is a barter economy#

It is fashionable now to say flatly that barter is a myth. That overshoots, and a reader who repeats it at dinner will be caught out by anyone who has actually read the source.

The foundational statement is Caroline Humphrey’s, in the journal Man in 1985, and it is more careful than its popular version: “No example of a barter economy, pure and simple, has ever been described, let alone the emergence from it of money; all available ethnography suggests that there never has been such a thing.” Read the clauses. She is not saying that people never swap goods directly. Of course they do; it happens in playgrounds, at borders, between prisoners, and across the front lines of wars. She is saying two narrower and much stronger things: that no society has ever been observed running its whole economy on direct exchange without money, and that money has never been observed emerging out of such a society.

Humphrey’s own fieldwork makes the point sharply. She studied the Lhomi of north-eastern Nepal in 1979 and 1980, a community that did a great deal of barter. They did not do it because they had never heard of money. They did it because currency was scarce and their links to monetary markets had broken down. Her argument is that barter in the observed world is overwhelmingly a post-monetary phenomenon: it appears where a money economy has decayed, been cut off, or run short of the stuff. It coexists with money rather than preceding it.

So the correction is this. Barter is real. What is unsupported is the sequence — barter first, then money to fix barter — and the idea that money is fundamentally a solution to the inconvenience of swapping shoes for bread. The evidence points instead to credit and reckoning coming first, with barter as one of several things that happen at the edges: between strangers, across hostile boundaries, and in the wreckage of monetary systems.

There is a second, subtler correction hiding here. Societies without general-purpose money often do not have one undifferentiated market at all. Paul Bohannan’s study of the Tiv of central Nigeria, published in American Anthropologist in 1955, described distinct spheres of exchange, with goods in one sphere not freely convertible into goods in another, because the culture did not treat them as commensurable. General-purpose money did not merely lubricate exchange in such societies; it dissolved walls that people had wanted kept up. Money is not a neutral efficiency improvement laid over an existing market. It changes what counts as a market.

“Credit came first” does not mean neighbours writing IOUs#

The Mesopotamian evidence is real, but it is easy to over-read it in a modern direction.

The tablets from Uruk are overwhelmingly institutional accounts. They are the internal records of large temple and palace estates disbursing rations and tracking obligations owed to and by the institution. They are much closer to the payroll and stock-control ledgers of a very large organisation than to two neighbours agreeing a private loan. The obligations they record are frequently administrative — quotas, allocations, deliveries due — rather than freely negotiated credit between equals. Calling this “credit” is fair, but you should picture a state accounting department, not a friendly overdraft.

The second over-read concerns the tokens. Denise Schmandt-Besserat’s thesis — that small clay tokens used for counting from around the eighth millennium BC were sealed inside clay envelopes called bullae, that the shapes were then impressed on the outside of the envelope so you could see the contents without breaking it, and that the impressions eventually made the tokens redundant and became writing — is enormously influential and genuinely elegant. It is also contested in the detail. Not every specialist accepts that the token system maps cleanly onto the earliest sign inventory, or that a single evolutionary line runs from token to envelope to tablet. Treat the token-to-writing story as the leading hypothesis rather than a settled fact, and note that the broader claim — that the earliest writing was administrative and economic — rests on the tablets themselves and does not depend on the token thesis being right.

The tally stick was a receipt that became negotiable, and the sources disagree about the details#

Two cautions. First, the tally began life as proof of payment into the Exchequer, not as a freely circulating currency. Its negotiability was a consequence of what it was — a hard-to-forge, institution-backed record of a sum owed by the Crown — rather than its design purpose. A tally in a strongbox was a receipt. A tally passed to a creditor was money. Same object, different use. Money is a role played by a record, not a class of object, and the tally is the cleanest illustration in European history.

Second, be careful about the notch conventions, because sources genuinely conflict. The Dialogus de Scaccario, the treatise on Exchequer procedure written around 1179, gives them in body parts: a cut the thickness of the palm for a thousand pounds, the breadth of a thumb for a hundred, the breadth of the little finger for twenty, the width of a swollen barleycorn for a single pound, and a narrower cut for a shilling. The 1911 Encyclopædia Britannica gives them in inches, and does not match: one and a half inches for a thousand pounds, one inch for a hundred, three-eighths of an inch for ten, a half-notch of that size for a pound, three-sixteenths for a shilling, the smallest notch for a penny, and small drilled holes for halfpennies. Note that the two schemes do not even agree on which denominations existed — the Dialogus has a twenty-pound notch and the Britannica has a ten-pound notch. The honest position is that the conventions varied across seven centuries and that no single table describes them all.

Anyone who tells you the tally system was uniform for seven hundred years is telling you something the sources do not support.

Information is only money when someone is obliged to act on it#

This is the correction that matters most for everything that follows, and it is where the slogan “money is information” starts to mislead.

A record is not money because it is a record. It is money because a specific, identifiable party is obliged to honour it, and because there is a mechanism that makes that obligation stick. Ada’s notebook worked because Ada was solvent, present, honest and answerable to a village that would make her life difficult if she cheated. Take away any one of those and the notebook becomes a list of numbers.

Three consequences follow, and they run through the rest of this book.

The first is that every unit of money in a modern economy has a counterparty. Your bank balance is your bank’s liability. A banknote is the central bank’s liability. A gift card is the retailer’s liability. When people say a payment “settles”, what they mean precisely is that the obligation has been discharged and the record can no longer be reversed. Settlement is the moment a promise stops being a promise, and a great deal of the machinery in this book exists to define that moment exactly, because until it arrives somebody is carrying risk.

The second is that cash is genuinely different, and the plain version glosses over it. When I hand you a twenty-pound note, the transaction is complete at the instant of handover: no clearing, no settlement window, no counterparty left owing anything, no possibility of recall. That is a real and unusual property, and it is why cash remains stubbornly useful in exactly the situations where records are unreliable — power cuts, disasters, distrust and crime. Every electronic payment is an attempt to reproduce the finality of cash across distance and time, and none fully manages it.

The third is that Alfred Mitchell-Innes’s famous formulation — writing in the Banking Law Journal in 1913, “Money, then, is credit and nothing but credit. A’s money is B’s debt to him, and when B pays his debt, A’s money disappears” — is a position in a long-running argument, not a neutral summary of the facts. Innes was demolishing the commodity theory of money, and he was right about the Exchequer in ways that mainstream histories of his day had got badly wrong. He was also making a maximalist claim. Economists still disagree about whether commodity-money origins matter, about how far the state’s power to demand taxes explains a currency’s acceptance, and about whether “credit” is the right frame for physical cash. This book leans towards the credit view because the historical record supports it and because it is by far the more useful frame for understanding payment systems. You should know that it is a lean and not a law.

The technical version#

The anthropological critique, stated precisely#

The claim under examination is what economists call the double coincidence of wants narrative, given its canonical modern statement in Adam Smith, An Inquiry into the Nature and Causes of the Wealth of Nations (1776), Book I, Chapter IV, “Of the Origin and Use of Money”. Smith’s argument is deductive rather than evidential: he reasons from the division of labour to the necessity of a commonly accepted medium, then to the selection of metals on the grounds of durability and divisibility. He offers no ethnographic case, because in 1776 there was none to offer.

The empirical counter-position is set out in Caroline Humphrey, “Barter and Economic Disintegration”, Man (New Series), Volume 20, Issue 1 (1985), pages 48 to 72, whose ethnographic base is fieldwork among the Lhomi of the Upper Arun valley in north-eastern Nepal in 1979 and 1980. Her thesis is that observed barter is characteristically a phenomenon of economies “de-coupled from monetary markets” — that it coexists with money rather than preceding it, and that under severe currency scarcity money itself may become one of the items bartered rather than the index of value for the rest.

The argument was popularised, and considerably extended, in David Graeber, Debt: The First 5,000 Years (Melville House, 2011), which treats the barter narrative as an origin myth internal to economics rather than a historical finding. Graeber’s synthesis has been criticised by economists and by some historians on points of detail; his central negative claim — that no one has produced the barter economy the textbooks require — has held up.

A related strand concerns the non-fungibility of value categories. Paul Bohannan, “Some Principles of Exchange and Investment among the Tiv”, American Anthropologist, Volume 57, Number 1 (1955), pages 60 to 70, described goods segregated into distinct spheres, with movement within a sphere ordinary and movement between spheres exceptional and morally freighted. Marcel Mauss’s Essai sur le don (1925) had earlier established that obligation-creating gift exchange, rather than spot barter, dominates in many non-market societies. The cumulative picture is not that early societies had a primitive form of our market, but that they had systems of obligation whose logic differs from market exchange at the root.

The Mesopotamian record#

Proto-cuneiform, the earliest attested phase of the cuneiform writing system, appears at Uruk in southern Mesopotamia and is conventionally dated to approximately 3100 to 2900 BC. The Metropolitan Museum of Art’s account of the origins of writing notes that the site “was dominated by large temple estates whose need for accounting and disbursing of revenues led to the recording of economic data on clay tablets”, that roughly 6,000 proto-cuneiform tablets survive with more than 38,000 lines of text, and that among the earliest signs are pictures of “rations that needed to be counted, such as grain, fish, and various types of animals”.

A representative object is British Museum registration number 1989,0130.4, the so-called British Museum Society tablet, acquired at Christie’s in 1989 from the Erlenmeyer collection and displayed in Room 56. It is dated 3100 to 3000 BC, measures 9.4 by 6.87 by 2.3 centimetres, and is catalogued as a record of beer impressed with five different types of numerical symbol. The presence of multiple distinct numerical notations on a single tablet is itself informative: proto-cuneiform used different sign systems for different commodity classes, so that counting sheep and measuring grain did not share a notation.

The prehistory of these tablets is the token thesis associated with Denise Schmandt-Besserat, set out in Before Writing, Volume I: From Counting to Cuneiform (University of Texas Press, 1992). The proposed sequence runs: small geometric clay tokens used as counters from the Neolithic; tokens sealed inside hollow clay envelopes (bullae) to fix a transaction’s content; tokens impressed on the envelope’s exterior before sealing so the contents could be read without breaking it; and finally the impressions alone, on a flat tablet, the tokens having become redundant. The thesis remains influential and remains disputed in its details.

On units and rates, the Old Babylonian and earlier systems were sexagesimal. The mina was divided into 60 shekels; the shekel was a weight of silver of roughly 8 grams. Michael Hudson’s work on palatial credit records the equivalence of the shekel-weight of silver with a gur of grain of 300 sila, an equation set administratively by the palace rather than discovered in a market. Hudson’s argument on interest is that the canonical Mesopotamian rate was a product of the number system rather than of economic conditions: one shekel per mina per month, that is one sixtieth per month, giving 12 shekels per mina per year, or 20 per cent per annum in decimal terms, and doubling the principal in 60 months under simple accrual. He notes, importantly for our purposes, that a debt denominated in silver did not require silver in settlement — “What was called a ‘silver’ debt thus did not mean that actual silver had to be paid, but simply that the interest rate was 20 percent” — with barley an accepted means of discharge.

That distinction between the unit in which an obligation is denominated and the asset in which it is settled is not an antiquarian curiosity. It is the same distinction that separates a currency from a settlement asset in a modern payment system, and it is four thousand years old.

The legal architecture around these obligations is visible in the laws of Hammurabi, of the eighteenth century BC. The provisions on merchants and agents are unambiguous about documentation: an agent must give a receipt for money received, and one who “is careless, and does not take a receipt for the money which he gave” bears the consequence. On deposits, the code requires a party leaving goods for safekeeping to “show everything to some witness, draw up a contract, and then hand it over for safe keeping”, and denies remedy to one who deposits “without witness or contract”. A society that legislates in this much detail about receipts and witnesses is a society whose economy runs on records.

The English Exchequer and the tally#

The Exchequer tally is documented in the Dialogus de Scaccario, written about 1179 and attributed to Richard FitzNeal, Treasurer of England. The physical description in the 1911 Encyclopædia Britannica is of “a willow or hazel stick about one inch in depth and thickness, and roughly shaped like a thick knife-blade”, notched across its width and then split lengthways through the notches.

The two halves had names. The longer piece, retained by the party to whom the sum was owed, was the stock; the shorter, retained by the Exchequer as the debtor’s record, was the foil. Details of the sum were also written on the wood in ink before splitting, so that each half carried both a notched and a written record. Verification was by physical fit: if genuine, the halves “fitted so exactly that they appeared evidently to be parts the one of the other”. This is a mechanical integrity check with no cryptographic component and no central register, and it worked for centuries because the grain and flaw pattern of a split stick was effectively unforgeable by the technology of the period.

On notch values, as noted above, the sources conflict. Set them side by side.

The correct reading of that table is that the convention changed over time and that neither source describes the whole period.

The tally’s monetary character rests on its transferability. Mitchell-Innes, in his 1913 Banking Law Journal article “What is Money?”, put it directly: “Tallies were transferable, negotiable instruments, just like bills of exchange, bank-notes or coins.” His accompanying claim about the Exchequer has aged best: “The general belief that the Exchequer was a place where gold or silver was received, stored and paid out is wholly false. Practically the entire business of the English Exchequer consisted in the issuing and receiving of tallies.” From this he drew the conclusion that gives the credit theory its name: “Money, then, is credit and nothing but credit. A’s money is B’s debt to him, and when B pays his debt, A’s money disappears.”

The system’s end is precisely datable and unusually consequential. An Act of 1782 provided for the discontinuance of wooden tallies, but the change took effect only on the death of the last chamberlain, in 1826. The accumulated stock of obsolete tallies — two cart-loads of them — remained in the Palace of Westminster until 16 October 1834, when Richard Weobley, the Clerk of Works, acting on Treasury instruction, had them burned in the two underfloor stoves in the basement of the House of Lords. Two labourers, Joshua Cross and Patrick Furlong, fed the stoves through the day. The first flames were seen at six o’clock that evening. The fire destroyed the House of Lords, the House of Commons, St Stephen’s Chapel, the Painted Chamber, the libraries and committee rooms of both Houses, the Speaker’s House, most of the procedural records of the Commons going back to the late fifteenth century, and the British standard yard and pound of 1496. Westminster Hall and the Jewel Tower survived. Charles Dickens, addressing the Administrative Reform Association two decades later, remarked that in the matter of the tallies “all the red tape in the country grew redder at the bare mention of this bold and original conception”.

The Palace of Westminster you see today exists because a government disposed carelessly of an obsolete accounting medium. It is the most expensive records-management incident in British history.

Coinage as a late and specialised development#

The chronology matters, so state it plainly. Administrative writing in Mesopotamia is attested from approximately 3100 BC. Coinage — standardised, stamped, portable metal of certified weight — appears roughly two and a half millennia later.

The evidential anchor is the deposit beneath the Temple of Artemis at Ephesus, where British Museum excavators recovered 93 electrum coins deposited as offerings during the latter part of the seventh century BC. The Lydian royal issues bearing a lion’s head run from about 4.7 grams, one third of a Lydian weight stater, down through fractions to pieces of 0.29 grams (one forty-eighth) and 0.15 grams (one ninety-sixth) bearing a lion’s paw. Analysis of the metal is instructive: the alloy is around 54 per cent gold, 44 per cent silver and 1 to 2 per cent copper, but the coins were tariffed as if they were natural Lydian electrum of roughly 73 per cent gold. The issuer was, in other words, extracting a seigniorage from the gap between the metal’s intrinsic content and its official rating.

That fact deserves emphasis, because it undermines the commodity story at its supposed strongest point. Even the very first coins were not simply weighed metal. They were metal whose accepted value depended on an institution’s stamp rather than on its assay. The stamp is a record. Coinage did not replace institutional trust with intrinsic value; it packaged institutional trust into a portable, hard-to-counterfeit form.

Obligation on the modern balance sheet#

The Bank of England’s Quarterly Bulletin for 2014 Q1 carried the article “Money creation in the modern economy” by Michael McLeay, Amar Radia and Ryland Thomas. It is the single most useful primary source for a British reader on what money now is, and its central statements are worth quoting exactly.

On composition: “Broad money is made up of bank deposits — which are essentially IOUs from commercial banks to households and companies — and currency — mostly IOUs from the central bank.”

On proportion: “Of the two types of broad money, bank deposits make up the vast majority — 97% of the amount currently in circulation.”

On creation: “Whenever a bank makes a loan, it simultaneously creates a matching deposit in the borrower’s bank account, thereby creating new money.” And, correcting the textbook account: “Banks do not act simply as intermediaries, lending out deposits that savers place with them.”

Set the physical stock against that. Bank of England banknote statistics record over 4.98 billion notes in circulation with a total value of about £91.5 billion as at 28 February 2026 — three per cent of the money in the economy, held in the form everyone pictures when the word “money” is said.

The consequence for a payments practitioner is structural. If 97 per cent of money is a liability recorded in commercial bank ledgers, then a payment between customers of two different banks is a message that causes two banks to adjust two sets of liabilities, plus a settlement between those banks in a third asset — central bank reserves — that neither customer ever sees. Every layer of the payments stack in this book exists to make that three-way adjustment happen correctly, in the right order, with an unambiguous point of finality.

What actually moves: the United Kingdom in numbers#

Pay.UK’s annual statistics for calendar year 2025 give the retail picture.

The wholesale picture sits above it. The Bank of England reports that in 2024 CHAPS, the United Kingdom’s high-value real-time gross settlement system, settled over £87 trillion, averaging over £344 billion each working day across around 208,000 payments a day. Roughly 208,000 instructions a day carry more value than 27 million Bacs items, which tells you most of what you need to know about the difference between retail and wholesale payment.

Now look at the shape of a single Bacs instruction, because it is the moment where the argument of this chapter becomes concrete. The Bacs Standard 18 format, the submission format used across the United Kingdom for Direct Debits and Direct Credits, is a fixed-width record of 100 characters, extended to 106 characters for multi-processing files.

One hundred characters. Two account identifiers, a two-character verb, an amount in pence, and three eighteen-character strings that exist purely so that human beings on both sides can later agree about what the record meant. That is a payment. That is the whole of it.

The record then moves through the three-day cycle that Pay.UK’s system principles set out: day one, input, when payment instructions and Direct Debit Instructions are submitted to the Bacs system; day two, processing, when payment service providers process the items; day three, entry and settlement, when providers apply the payments to customer accounts and interbank settlement occurs. Input, processing and entry occur only on processing days; Saturdays, Sundays and English bank holidays are excluded. That three-day rhythm is not a technical limitation of any modern computer. It is a convention about when obligations become final, and conventions about finality are what payment systems fundamentally are.

The industry is now migrating that structure to a richer one. ISO 20022 replaces fixed-width, position-defined records with structured XML messages carrying a far larger and more explicitly typed set of fields. For cross-border payments, the coexistence period during which Swift’s legacy MT messages and ISO 20022 both ran ended on 22 November 2025, after which MT is no longer supported for payment instructions in the live environment, and in-flow translation became a chargeable contingency from 1 January 2026. The direction of travel is towards more information per payment, not less — remittance data, structured parties, purpose codes, legal entity identifiers. If money were a substance, none of this would matter. Because money is a record, the richness of the record is the product.

Why this framing organises the whole book#

Every question that follows in this book is a question about records.

Clearing is the process of working out, from two sets of records, what each party owes the other. Settlement is the moment at which an obligation is extinguished and the record becomes final. Reconciliation is the discipline of proving that two independently kept records of the same events agree — which is what the two halves of a tally stick did mechanically, in wood, in the twelfth century. Double-entry bookkeeping is the same insight applied inside a single organisation. A chargeback is a rule about when a settled record may be unwound and who bears the loss. Anti-money-laundering controls are rules about what the record must contain and who must be able to read it. A cryptographic ledger is a mechanism for making a shared record tamper-evident without a trusted keeper, which is to say a tally stick for people who cannot meet to fit the halves together.

The units and institutions change. The problem does not. Somebody must write down who owes what to whom, somebody must be trusted to keep that writing honest, and everybody must agree when a promise has stopped being a promise.

Volume I of this book asks what money is: obligation, denomination, settlement asset, legal tender, currency, and the several different things people mean when they say “cash”. The volumes that follow ask how the information moves: through schemes and rails, files and messages, banks and processors, borders and clearing houses, and the specific British machinery — Bacs, Faster Payments, CHAPS, the Image Clearing System, Open Banking — that most readers will actually be building against.

Start from the wrong premise and every one of those chapters becomes a list of arbitrary rules to memorise. Start from the right one and they become variations on a single, very old idea, already fully formed when somebody in Uruk pressed a number into wet clay to record how much beer was owed to a workman.

Money is information. This book is about how that information moves.

1.98 Common wrong ideas#

1. Wrong: money was invented to solve the inconvenience of barter. Right: no society has ever been observed running its whole economy on direct exchange, and money has never been observed emerging from one; credit and reckoning come first in the evidence we actually have.
2. Wrong: barter is a myth. Right: barter is real and well documented, but the observed cases are overwhelmingly post-monetary, appearing where currency is scarce or a money economy has broken down, as among the Lhomi of north-eastern Nepal.
3. Wrong: the Mesopotamian tablets show neighbours writing IOUs to one another. Right: they are overwhelmingly institutional accounts of temple and palace estates, far closer to payroll and stock control than to a friendly overdraft.
4. Wrong: the token-to-envelope-to-tablet account of the origin of writing is settled fact. Right: it is the leading hypothesis and is contested in its details, and the broader claim that the earliest writing was administrative rests on the tablets themselves rather than on the token thesis.
5. Wrong: the tally stick was designed as a circulating currency. Right: it began as proof of payment into the Exchequer, and its negotiability was a consequence of being a hard-to-forge, institution-backed record of a Crown debt — a receipt in a strongbox, money when passed to a creditor.
6. Wrong: the notch conventions were uniform across the seven centuries the tally was in use. Right: the Dialogus de Scaccario and the 1911 Encyclopædia Britannica conflict, and do not even agree on which denominations existed, so no single table describes the whole period.
7. Wrong: money is information. Right: a record becomes money only when a specific, identifiable party is obliged to honour it and a mechanism makes that obligation stick; remove Ada’s solvency, presence, honesty or answerability and the notebook is a list of numbers.
8. Wrong: cash is simply the physical version of a bank balance. Right: handing over a note completes the transaction at the instant of handover, with no clearing, no settlement window, no counterparty left owing anything and no possibility of recall, which is a property no electronic payment fully reproduces.
9. Wrong: “money is credit and nothing but credit” is the neutral summary of the historical evidence. Right: it is Mitchell-Innes’s position in a long-running argument, and this book leans towards it because the record supports it and it is the more useful frame, but a lean is not a law.
10. Wrong: the three-day Bacs cycle is a technical limitation of the computers involved. Right: it is a convention about when obligations become final, and conventions about finality are what payment systems fundamentally are.

1.99 Chapter summary in 20 lines#

1. Almost everyone describes a payment as a movement of money, and almost nobody is right.
2. What changes in a payment is a record: somebody’s obligation to somebody else is written down, cancelled, netted off or handed on.
3. Ada’s village notebook contains the whole of payment in miniature, because Tom’s debt to Sam is discharged by three edited lines and nothing changes hands.
4. The textbook story that money was invented to fix the awkwardness of barter comes from Adam Smith in 1776 and is deductive rather than evidential.
5. Caroline Humphrey’s careful 1985 statement is that no barter economy, pure and simple, has ever been described, let alone the emergence of money from one.
6. Barter itself is real, but observed barter is overwhelmingly a post-monetary phenomenon that coexists with money rather than preceding it.
7. Societies without general-purpose money often kept goods in separate spheres of exchange, so money did not merely lubricate a market but changed what counted as one.
8. The earliest writing at Uruk, around 3100 BC, is administrative: rations, grain, sheep and beer, recorded with several distinct numerical notations.
9. Those tablets are the internal accounts of large temple and palace estates, and should be read as a state accounting department rather than as private credit.
10. Coins appear roughly two and a half thousand years after those tablets, in what is now western Turkey at the end of the seventh century BC.
11. Even the first coins were valued by the issuer’s stamp rather than by assay, since electrum of about 54 per cent gold was tariffed as though it held 73 per cent.
12. The English Exchequer tally makes the same point in wood: one stick split lengthways through every notch, verified by physical fit, unforgeable by the technology of the day.
13. Tallies circulated from hand to hand because the taxman would accept them, which made a notched stick money without making it a substance.
14. The longer half was the stock and the shorter the foil, and the careless disposal of two cart-loads of obsolete tallies burned down the Palace of Westminster in 1834.
15. A record is money only when an identifiable party is obliged to honour it and there is a mechanism that makes the obligation stick.
16. Every unit of money in a modern economy therefore has a counterparty: your bank for a deposit, the central bank for a note, the retailer for a gift card.
17. Cash is the genuine exception, because handover is final and leaves nobody owing anything, which is why it survives where records are unreliable.
18. In Britain the proportions are stark: about £91.5 billion of banknotes exist, against bank deposits making up 97 per cent of the money in circulation.
19. In 2025 Bacs and Faster Payments together carried over twelve billion instructions worth nearly eleven trillion pounds, and not one of them moved an object.
20. Every question in the rest of the book — clearing, settlement, reconciliation, chargebacks, finality — is a question about who keeps the record and when a promise stops being a promise.

Sources used: Caroline Humphrey, “Barter and Economic Disintegration”, Man (NS) 20:1 (1985), 48–72; Adam Smith, The Wealth of Nations (1776), Book I Ch. IV; Paul Bohannan, American Anthropologist 57:1 (1955), 60–70; A. Mitchell Innes, “What is Money?”, Banking Law Journal (May 1913); The Metropolitan Museum of Art, “The Origins of Writing”; The British Museum, object 1989,0130.4; Denise Schmandt-Besserat, Before Writing (1992); Michael Hudson on palatial credit and Mesopotamian interest; the laws of Hammurabi (Harper translation); Dialogus de Scaccario (c. 1179) and the 1911 Encyclopædia Britannica entry “Tally”; UK Parliament, “The Great Fire of 1834” and “Tally Sticks”; John H. Kroll, “The Coins of Sardis” (Sardis Expedition); Bank of England, “Money creation in the modern economy”, Quarterly Bulletin 2014 Q1, and Banknote Statistics (28 February 2026); Bank of England, RTGS and CHAPS introduction (2024 figures); Pay.UK, Annual Summary of Payment Statistics 2025 and Bacs System Principles v18 (January 2026); Bacs Standard 18 Message Implementation Guide; Swift ISO 20022 implementation guidance.