When Memory Became a System

01 / WHEN MEMORY ISN’T ENOUGH

Someone pressed a sign into wet clay. Someone else had to understand it.

Imagine a delivery arriving in a crowded settlement more than five thousand years ago. Grain is unloaded, some to be stored, some to be processed, some eventually to be distributed again. Someone needs to know how much arrived and perhaps where it should go. For a household or a small group, much of that information can live in people’s memories. But what happens when deliveries multiply, goods move in and out of storage, and the people who need the information were not there when the transaction took place?

By the late fourth millennium BCE, this was becoming a practical problem in southern Mesopotamia. Uruk had grown into an urban centre on a scale that demanded extraordinary levels of coordination. Cultivation, storage, craft production, construction and the movement of goods connected people and activities across the city.

Chapter 1 followed one consequence of that transformation: feeding people and organising labour required systems reaching beyond individual households. But coordination generates information. If grain arrives, someone needs to know how much. If supplies leave a store, someone needs to know where they went. If goods or labour are allocated, that allocation may still matter after the people involved have walked away.

A conversation can carry such information for a while. Then memories diverge, people leave and yesterday’s agreement becomes increasingly difficult to reconstruct. So the problem is not simply how to remember more. It is how to make information survive the people who first remembered it.

Writing did not appear in a world without earlier ways of storing and communicating administrative information. Across Southwest Asia, people had already used material devices such as seals, numerical notation and other clay-based accounting practices. How exactly these traditions contributed to proto-cuneiform remains debated; no single straight line explains all the evidence.

By the later fourth millennium BCE, proto-cuneiform appears on tablets at Uruk. The earliest writing was therefore not born from nothing. It emerged from a world already experimenting with ways to make information material.

02 / READ THE OBJECT FIRST

Start with what survives.

A small clay tablet now held by the Metropolitan Museum of Art offers a way into that world. At just 6.85 centimetres across its longest dimension, it fits easily in a hand. Across its surface are numerical notations and other signs.¹

The museum identifies entries concerning malt and barley groats and interprets the tablet as an administrative account, most likely documenting grain distributed by a large institution. It dates to roughly 3100–2900 BCE. Its precise archaeological origin is uncertain: the Met lists it as coming from Mesopotamia, probably from Uruk.¹

We can observe	We can interpret	We cannot recover with certainty
Numerical and other signs were placed on clay.	Entries concern malt and barley groats.	The exact circumstances in which the tablet was made.
Different types of signs occur together.	It belongs to an early administrative writing tradition.	Who stood nearby or what was said.
The tablet dates to around 3100–2900 BCE.	It most likely documents grain distribution.	Its precise archaeological findspot.

That distinction will follow us through the article. The tablet is evidence, but evidence is not the same thing as a complete event. Early proto-cuneiform relied heavily on numerical notation and signs representing commodities and other categories. Grammatical information was extremely limited. The system did not have to record everything someone could say. It had to preserve what someone needed to know later.

03 / TURNING AN EVENT INTO INFORMATION

A useful record leaves most of reality out.

Imagine that you are responsible for recording the grain delivery. You could preserve everything: who carried it, how long the journey took, whether the sacks were damaged, who helped unload them, what the weather was like and what everyone said while working. You would have a vivid story. You would also have a terrible account.

Administration depends on selection. A complicated event has to be reduced to information that can be recorded and recognised again. For a transaction, that might include:

• what was involved;
• how much there was;
• which person, group or office mattered to the record;
• what kind of transaction was being recorded.

Individual proto-cuneiform texts vary, so we should not imagine every tablet as a form containing the same fields. The larger point is the act of abstraction itself. A heap of barley is physical. A numerical notation associated with a commodity sign is information about that barley.

That information can sit in another room. It can remain after the grain has been consumed. It can be examined by someone who never saw the original delivery. And once many events can be represented through recurring conventions, something else becomes possible: comparison.

04 / A SIGN IS NOT YET A SYSTEM

A mark becomes useful when other people can understand it.

Draw a triangle on clay and decide that it means grain. You have created a sign for yourself. Now leave the room. Can someone else read it?

For information to survive its original writer, the system needs more than marks. Several things have to remain sufficiently stable:

• signs must be recognisable to other trained users;
• numerical conventions must be shared;
• administrative categories must remain understandable;
• new users must learn how records are produced and interpreted;
• that knowledge must survive when individual writers leave.

This is where another group of early texts becomes especially revealing. Alongside administrative accounts, scribes produced what modern scholars call lexical lists. They arranged signs and terms into recurring categories. One of the best-known is conventionally called Archaic Lu A, a list concerned largely with professions and titles.

At first glance, such lists seem to solve an obvious problem: a shared writing system needs shared knowledge. But the lexical tradition also points toward something larger. These texts organised parts of the world into ordered sets—professions, vessels, animals, plants and other categories. Scribal knowledge was becoming something that could itself be structured and preserved.

Writing had therefore created a second memory problem. It was no longer enough to preserve information. People also had to preserve the knowledge required to understand the information. The tablet could outlast its writer, but its meaning still depended on human expertise.

05 / THE ARCHIVE HAS A POINT OF VIEW

What gets recorded becomes visible to us. What does not can disappear.

An account concerned with grain has little reason to preserve someone’s feelings about receiving it. A record may preserve a quantity without preserving the conversation surrounding it. Administrative writing can therefore make institutions unusually visible while leaving many individual experiences almost completely silent.

The record may preserve…	But not necessarily…
a quantity entered by an administrator	whether everyone agreed that it was correct
a commodity or administrative category	what that commodity meant to the people receiving it
something an institution considered worth tracking	everything the people involved considered important
evidence of a transaction or allocation	whether the arrangement was considered fair

This is not a reason to distrust the tablet. It is a reason to read it more carefully. Administrative records are not transparent windows through which we can simply watch ancient society happening. They are objects produced inside that society for particular purposes.

The tablet tells us something about malt and barley groats. But its very existence also tells us that someone considered those commodities and quantities worth recording, that conventions existed for doing so and that people possessed the expertise to make the resulting record useful.

06 / WRITTEN DOESN’T MEAN TRUE

A written number answers one question better than several others.

Suppose a tablet records ten units. What exactly have we learned? There are three different claims hiding inside that apparently simple fact:

1. Ten units were recorded.
2. Ten units actually changed hands.
3. Everyone involved considered the transaction correct or fair.

Those claims are not equivalent. The tablet itself may give us strong evidence for the first. Establishing the second could require other evidence. The third asks a question the administrative record may never have been designed to answer.

Writing created new possibilities for checking and continuity, but it did not remove people from the process. Someone decided what to record. Someone entered the quantity. Someone used the categories. Someone may have had the authority to accept the result.

This is where writing intersects with power—not because every scribe controlled the city, and not because every administrative tablet records coercion. The more defensible point is simpler: durable information expands what an organisation can remember, compare and coordinate.

07 / THE CITY COULD REMEMBER WITHOUT YOU

A transaction ends. Selected information can remain.

Return to the grain delivery from the beginning. The grain arrives. Someone receives it. People carry it away. By evening, the physical event is already over.

Without a durable record, knowledge of what happened remains heavily tied to people. Now place a tablet between them. Days later, the people may be somewhere else. The grain itself may already have been processed or distributed. Yet selected information about the transaction can remain.

• another trained person can recover information about an earlier event;
• separate transactions can be compared;
• records can remain when personnel change;
• an organisation can build continuity across many individual encounters.

Writing did not replace human memory, and societies without writing developed sophisticated ways of preserving knowledge. What the tablet added was something more specific: selected information could remain available within an institution after the original encounter had ended and the people involved had moved on.

In that limited but important sense, the city had begun to acquire a memory of its own.

08 / WHAT SURVIVED

The grain disappeared. The information did not.

More than five thousand years later, the tablet still exists. The grain does not.

We do not know the people who delivered it, who received it or who pressed the signs into the clay. We cannot hear the conversation that surrounded the transaction. Even the tablet’s precise archaeological origin remains uncertain. Yet selected information survived.

That is what makes this unassuming object so important to the story of the first cities. Its marks belong to a world in which information was becoming something that could be separated from an event, organised according to shared conventions and recovered later by someone who had not been there.

The uses of writing would expand enormously. By the middle of the third millennium BCE, cuneiform was being used for economic, religious, political, literary and scholarly texts.² But none of that is necessary to make the earlier administrative technology remarkable.

Return to the tablet one last time: malt, barley groats, quantities, signs. Almost everything about the original moment has disappeared. The information did not.

FAQ

Questions that follow from the evidence

Was writing invented in Uruk?

Proto-cuneiform is first attested on clay tablets at Uruk in the late fourth millennium BCE, making the city central to the history of early writing. But it is misleading to imagine a single inventor or moment of invention. Earlier accounting and communication practices formed part of the wider environment from which proto-cuneiform emerged.

What was the earliest writing used for?

Much of the surviving proto-cuneiform material is administrative. Early tablets record economic information using numerical notations and signs associated with commodities and other categories. Lexical texts also appear very early, showing that organising and transmitting scribal knowledge became important in its own right.

Could people read proto-cuneiform like a modern sentence?

Not in the way we read ordinary prose. Early proto-cuneiform relied heavily on numerical and lexical information and included little grammatical information. The relationship between signs and spoken language developed substantially during the following centuries.

Is proto-cuneiform the same as cuneiform?

Proto-cuneiform is the conventional name for the earliest phase of the writing tradition from which later cuneiform developed. Sign forms, writing techniques and the ability of the script to represent language all changed over time.

What can’t these tablets tell us?

Administrative tablets rarely preserve the full human context behind the transactions they record. They may tell us about quantities or categories without revealing whether an entry was disputed, how participants experienced an arrangement or what aspects of the event mattered outside the administrative system.