Entry 1 of 10 · Late fourth millennium BC

The mud trail: when a number did not mean a number

Ancient clay accounts expose a very modern audit risk: checking the arithmetic before understanding the data.

The spreadsheet looks immaculate. Every column adds up. Someone has checked the formulas. Unfortunately, half the quantities are in kilograms and the other half are in cases.

This is an invented modern embarrassment. Its ancient cousin is considerably more interesting.

In the earliest written accounts from Mesopotamia, the meaning of a numerical sign could depend on what was being counted. A mark that looked familiar did not necessarily carry the value a modern reader expected. Before you could check the sum, you had to understand the system.

Some later scholars got that order wrong. They treated apparent contradictions as mistakes by ancient scribes, when the mistake lay in their own interpretation.^{Note 1} There is something pleasing about an accountant being vindicated several thousand years after a difficult review meeting.

The accounts before the alphabet

The late fourth millennium BC tablets studied by Hans Nissen, Peter Damerow and Robert Englund record a world concerned with quantities, commodities and administration. Individual entries could be organised into compartments, with totals elsewhere on the tablet. The authors also describe tablets where checking apparently concerned the number of transactions rather than the total quantity of products.^{Note 2}

These were structured records, with conventions governing what went where. Their makers were doing more than doodling sheep in wet clay.

Yet the evidence needs careful handling. The surviving tablets do not give us a complete job description for an ancient auditor. A total, an administrative mark or a repeated entry is not enough to reconstruct an independent audit profession. We can see practices of recording and checking without pretending that Uruk had a nineteenth-century accounting partnership waiting for its royal charter.

The uncertainty is part of the story. These records have survived better than the explanations their users took for granted.

The reviewer was wrong

Nissen, Damerow and Englund describe how researchers had assigned familiar numerical values to signs, then carried those assumptions into other contexts. Their own analysis instead examined the rules governing how signs appeared together across the surviving corpus.

The results pointed to several numerical systems with different uses. Identical-looking signs could acquire different values within those systems. Cereals and field measurements, for example, did not simply share one universal numerical vocabulary.^{Note 1}

That changes the question from “Did the scribe add this correctly?” to “Have we understood what the scribe recorded?”

It also changes where professional scepticism should be directed. The preparer may be wrong. So may the reviewer’s model of the preparer’s world.

A very old problem in a very new data export

Imagine an audit team combining sales files from several subsidiaries. One system records gross orders; another records shipments after cancellations. Both have a column called “sales”. A third file expresses amounts in thousands. All three exports are genuine. Combining them without understanding those differences can produce a convincing answer to a question nobody meant to ask.

The ancient comparison is useful precisely because it is limited. Clay tablets were not databases. But both require readers to recover meaning from a recording system, and neither explains every convention merely by existing.

For modern audit analytics, that suggests a practical order of work: establish what a field means, how it was produced and what population it represents before applying an impressive test to it. A reconciliation can establish agreement between two files while leaving their shared misunderstanding untouched.

The same applies when a machine reads the records. Recognising a symbol or extracting a number is only the beginning. The tool still needs the right unit, category, context and relationship to the underlying event.

The Mesopotamian contribution to auditing history is therefore more interesting than a claim to the first tick mark. It reminds us that evidence has a language. The earliest task of the auditor is learning enough of it to avoid correcting a perfectly competent scribe.

Next: Domesday and the danger of a complete-looking dataset.

Sources

Hans J. Nissen, Peter Damerow and Robert K. Englund, Archaic Bookkeeping: Early Writing and Techniques of Economic Administration in the Ancient Near East (University of Chicago Press, 1993), chapter 6, especially pp. 25–27. The account of scholars’ mistaken assumptions and the analysis of numerical systems comes from this study; the spreadsheet examples are modern illustrations. ↩ ↩
Nissen, Damerow and Englund, Archaic Bookkeeping, chapter 5, especially pp. 20–21. The authors explicitly qualify what can be inferred about the procedures behind the surviving records. ↩