The Earliest Precursor of Writing.

Scientific American. June 1977, Vol. 238, No. 6, p. 50-58.
Denise Schmandt-Besserat.

Long before the Sumerians invented writing, accounts in western Asia were kept with clay tokens of various distinctive shapes. It appears that the tokens gave rise to the Sumerian ideographs.

What led to writing? The art itself is a good example of what students of the past call independent invention, since systems of writing have evolved in isolation at different times in different parts of the world. For example, one system −the Chinese ideogram− can be traced to its origin in archaic signs engraved on the scapular bones of sheep or the shells of turtles in the second millennium B.C. as a means of asking questions of heaven. Roughly 1,000 years later an entirely independent system of writing arose halfway around the world in Mesoamerica. It combined a simple system of numerical notation with complex hieroglyphs and was principally used to indicate the dates of various events according to an elaborate calendrical system.

Both Chinese and Maya writing were relatively late inventions. Some one system of writing must have been the earliest, and it is from such an initial point that we may begin the search for the antecedents of the art. The credit for being the first to write texts is usually given to the Sumerians of Mesopotamia. By the last century of the fourth millennium B.C. the officials of such Sumerian city-states as Uruk had developed a system of recording numerals, pictographs and ideographs on specially prepared clay surfaces. (A pictograph is a more or less realistic portrayal of the object it is supposed to represent; an ideograph is an abstract sign.)

At Uruk a team of German archaeologists directed by Julius Jordan turned up many examples of these archaic records in 1929 and 1930. The texts, about 1,000 of them, were first analyzed by Adam Falkenstein and his students. Today additional discoveries have increased the total number of Uruk and Uruk-style texts to about 4,000, and Falkenstein's pioneer efforts are being continued mainly by Hans J. Nissen of the Free University of Berlin and his associate Margaret W. Green.

Although the clay blanks used by the Uruk scribes are universally referred to as tablets, a word with the connotation of flatness, they are actually convex. Individual characters were inscribed in the clay by means of a stylus made of wood, bone or ivory, with one end blunt and the other pointed. The characters were basically of two kinds. Numerical signs were impressed into the clay; all other signs, pictographs and ideographs alike, were incised with the pointed end of the stylus. The repertory of characters used by the Uruk scribes was large; it is estimated at no fewer than 1,500 separate signs.

Hypotheses about the origin of writing generally postulate an evolution from the concrete to the abstract: an initial pictographic stage that in the course of time and perhaps because of the carelessness of scribes becomes increasingly schematic. The Uruk tablets contradict this line of thought. Most of the 1,500 signs (Falkenstein compiled 950 of them) are totally abstract ideographs; the few pictographs represent such wild animals as the wolf and the fox or items of advanced technology such as the chariot and the sledge. Indeed, the Uruk texts remain largely undeciphered and an enigma to epigraphers. The few ideographic signs that have been identified are those that can be traced back stage by stage from a known cuneiform character of later times to an archaic Sumerian prototype. From the fragmentary textual contents that such identities allow it appears that the scribes of Uruk mainly recorded such matters as business transactions and land sales. Some of the terms that appear most frequently are those for bread, beer, sheep, cattle and clothing.

After Jordan's discovery at Uruk other archaeologists found similar texts elsewhere in Mesopotamia. More were found in Iran: at Susa, at Chogha Mish and as far off as Godin Tepe, some 350 kilometers north of Uruk. In recent years tablets in the Uruk style have been unearthed in Syria at Habuba Kabira and Jebel Aruda, nearly 800 kilometers to the northwest. At Uruk the tablets had been found in a temple complex; most of the others came to light in the ruins of private houses, where the presence of seals and the seal-marked clay stoppers for jars indicate some kind of mercantile activity.

The fact that the Uruk texts contradict the hypothesis that the earliest form of writing would be pictographic has inclined many epigraphers to the view that the tablets, even though they bear the earliest-known writing, must represent a stage in the evolution of the art that is already advanced. The pictographic hypothesis has been revived anew. The fact that no writing of this kind has yet appeared at sites of the fourth millennium B.C. and even earlier is explained away by postulating that the writing of earlier millenniums was recorded exclusively on perishable mediums that vanished long ago, such as parchment, papyrus or wood.

I have an alternative proposal. Research into the first uses of clay in the Near East over the past several years suggests that several characteristics of the Uruk material provide important clues to what kinds of visible symbols actually preceded the archaic Sumerian texts. These clues include the choice of clay as a material for documents, the convex profile of the Uruk tablets and the appearance of the characters recorded on them.

Nuzi, a city site of the second millennium B.C. in Iraq, was excavated by the American School of Oriental Research in Baghdad between 1927 and 1931. Nearly 30 years later, reviewing an analysis of the Nuzi palace archives, A. Leo Oppenheim of the Oriental Institute of the University of Chicago reported the existence of a recording system that made use of "counters," or tokens. According to the Nuzi texts, such tokens were used for accounting purposes; they were spoken of as being "deposited," "transferred" and "removed."

Oppenheim visualized a kind of dual bookkeeping system in the Nuzi texts; in addition to the scribes' elaborate cuneiform records the palace administration had parallel tangible accounts. For example, one token of a particular kind might represent each of the animals in the palace herds. When new animals were born in the spring, the appropriate number of new tokens would be added; when animals were slaughtered, the appropriate number of tokens would be withdrawn. The tokens were probably also moved from one shelf to another when animals were moved from one herder or pasture to another, when sheep were shorn and so forth.

The discovery of a hollow egg-shaped tablet in the palace ruins supported Oppenheim's hypothesis (see image 2). The inscription on the face of the tablet turned out to be a list of 48 animals. The hollow tablet rattled, and when one end of it was carefully opened, 48 tokens were found inside. Presumably the combination of a written list and countable tokens represented a transfer of animals from one palace service to another. Unfortunately we have no accurate description of the tokens; they were subsequently lost.

The Nuzi archives are dated to about 1500 B.C. The great Elamite site, Susa, has levels that are more than 1,500 years older. The digging at Susa, undertaken by French investigators, began in the 1880's and continues to this day. Six years after Oppenheim's 1958 report Pierre Amiet of the Musée du Louvre was able to confirm the existence of a similar accounting system at Susa. The token containers at Susa, unlike the container from Nuzi, were hollow clay spheres. Amiet called them "bullae" (see image 4); so far about 70 of them have been found. The tokens they contain are clay modeled in a variety of geometric forms, including spheres, disks, cylinders, cones and tetrahedrons.

Amiet's finding was one of great significance; not only did it demonstrate that bullae and tokens were in existence at least a millennium and a half before they appeared at Nuzi but also it showed that they were as old or older than the earliest written records at Uruk. Indeed, it later became clear that the tokens, at least, were very much older.

In 1969 I began a research project with the objective of discovering when and in what ways clay first carne to be used in the Near East. The making of pottery is of course the most familiar use of clay, but before the appearance of pottery man was making clay beads, modeling clay figurines, molding bricks out of clay and using clay for mortar. As a start on my project I visited museums in the U.S., in Europe and in various Near Eastern cities that had collections of clay artifacts dating back to the seventh, eighth and ninth millenniums B.C. This interval of time, beginning around 11,000 years ago and ending a little more than 8,000 years ago, saw the firm establishment of the first farming settlements in western Asia.

In the museum collections, along with the beads, bricks and figurines I had expected to find, I encountered what was to me an unforeseen category of objects: small clay artifacts of various forms. As I later came to realize, the forms were like those Amiet had found inside his Susa bullae: spheres, disks, cones, tetrahedrons, ovoids, triangles (or crescents), biconoids (double cones joined at the base), rectangles and other odd shapes difficult to describe. Could these artifacts, some of them 5,000 years older than the tokens from Susa, also have served as tokens?

I began to compile my own master catalogue of these oddities, listing each token that was known to have come from a specific site. In summary, I found that whereas all of them were small, measuring on the average from one centimeter to two centimeters in their greatest dimension, many were of two distinct sizes. For example, there were small cones about a centimeter high and large cones three to four centimeters high. There were also thin disks, only three millimeters thick, and thick ones, as much as two centimeters thick. Other variations were evident. For example, in addition to whole spheres I found quarter-, half- and three-quarter spheres. Some of the tokens had additional features. Many were incised with deep lines; some had small clay pellets or coils on them and others bore shallow circular punch marks.

The tokens had all been modeled by hand. Either a small lump of clay had been rolled between the palms of the hands or the lump had been pinched between the fingertips. The clay was of a fine texture but showed no sign of special preparation (such as the addition of tempering substances, a practice in pottery making that enhances hardness after firing). All the tokens had, however, been fired to ensure their durability. Most of them varied in color from buff to red, but some had become gray and even blackish.

I found that the tokens were present in virtually all museum collections of artifacts from the Neolithic period in western Asia. An extreme example of abundance is provided by the early village site of Jarmo in Iraq, first occupied some 8,500 years ago. Jarmo has yielded a total of 1,153 spheres, 206 disks and 106 cones. Reports generally indicate that the excavators found the tokens scattered over the floors of houses located in various parts of a site. If the tokens had once been kept in containers, such as baskets or pouches, these had disintegrated long ago. Nevertheless, there is evidence suggesting that the tokens were segregated from other artifacts and even implying what their function was. The reports indicate that many were found in clusters numbering 15 or more and that the clusters were located in storage areas within the houses.

As I reviewed the museum collections and the related site reports I became increasingly puzzled by the apparent omnipresence of the tokens. They had been found in sites from as far west as Beldibi in what is now southwestern Turkey to as far east as Chanhu Daro in what is now Pakistan. Tokens had even been unearthed at an eighth-millennium-B.C. site on the Nile near Khartoum.

At the same time I found that some site reports failed to take note of the tokens that had been collected, or mentioned them only casually. When the tokens were noted, the heading might read "objects of uncertain purpose," "children's playthings," "game pieces" or "amulets." As an example, the tokens from Tello in Iraq were interpreted by their discoverer, Henri de Genouillac, as amulets that expressed the residents' desire for "personal identification." Another example appears in Carleton S. Coon's report on Belt Cave in Iran: "From levels 11 and 12 come five mysterious ... clay objects, looking like nothing in the world but suppositories. What they were used for is anyone's guess."

The realization that the tokens were all artifacts of the same kind was also hampered because, when they were listed at all in the site reports, they usually appeared under not one heading but several headings depending on their shape. For example, cones have been described as schematic female figurines, as phallic symbols, as gaming pieces and as nails, and spheres were mostly interpreted as marbles or as sling missiles.

Having studied at the École du Louvre, I was familiar with the work of Amiet. Nevertheless, I had compiled a catalogue of hundreds of tokens before I at last realized how much like Amiet's tokens from Susa these far earlier clay artifacts were. At first it seemed impossible that the two groups could be related; a minimum of 5,000 years separated the tokens of Neolithic times from those of Bronze Age Susa. As I extended my investigations to include later clay artifacts, dating from the seventh millennium B.C. to the fourth millennium and later, I found to my surprise that similar clay tokens had been found in substantial numbers at sites representative of the entire time span. Evidently a system of accounting that made use of tokens was widely used not only at Nuzi and Susa but throughout western Asia from as long ago as the ninth millennium B.C. to as recently as the second millennium.

The system appears to have been much the same as many other early, and even not so early, methods of account keeping. Classical scholars are familiar with the Roman system of making "calculations" with pebbles (calculi in Latin). Up to the end of the 18th century the British treasury still worked with counters to calculate taxes. For that matter, the shepherds of Iraq to this day use pebbles to account for the animals in their flocks, and the abacus is still the standard calculator in the markets of Asia. The archaic token system of western Asia was if anything only somewhat more complex than its later counterparts.

Considered overall, the system had some 15 major classes of tokens, further divided into some 200 subclasses on the basis of size, marking or fractional variation, as in the case of the quarter-, half- and three-quarter spheres. Evidently each particular shape had a meaning of its own; a few appear to represent numerical values and others specific objects, commodities in particular.

It is not necessary to theorize about some of these meanings; a number of ideographs on the Uruk tablets almost exactly reproduce in two dimensions many of the tokens. For example, Uruk arbitrary signs for numerals, such as a small cone-shaped impression for the number one, a circular impression for the number 10 and a larger cone-shaped impression for the number 60 are matched by tokens: small cones, spheres and large cones. Further examples of ideographs that match tokens include, under the general heading of commodities, the Uruk symbol for sheep (a circle enclosing a cross), matched by disk-shaped tokens incised with a cross, and the Uruk symbol for a garment (a circle enclosing four parallel lines), matched by disk-shaped tokens incised with four parallel lines. Still other examples are ideographs for metal and oil and more clearly pictographic symbols for cattle, dogs and what are evidently vessels; each tablet sign can be matched with a similarly shaped and marked token. In addition, the forms of many still unread Sumerian ideographs appear to match other tokens.

Why did such a repertory of threedimensional symbols come into existence? It cannot simply be a coincidence that the first tokens appear early in the Neolithic period, a time of profound change in human society. It was then that an earlier subsistence pattern, based on hunting and gathering, was transformed by the impact of plant and animal domestication and the development of a farming way of life. The new agricultural economy, although it undoubtedly increased the production of food, would have been accompanied by new problems.

Perhaps the most crucial would have been food storage. Some portion of each annual yield had to be allocated for the farm family's own subsistence and some portion had to be set aside as seed for the next year's crop. Still another portion could have been reserved for barter with those who were ready to provide exotic products and raw materials in exchange for foodstuffs. It seems possible that the need to keep track of such allocations and transactions was enough to stimulate development of a recording system.

The earliest tokens now known are those from two sites in the Zagros region of Iran: Tepe Asiab and Ganj-i-Dareh Tepe. The people of both communities seem to have tended flocks and were possibly experimenting with crops around 8500 B.C., although at the same time they continued to hunt game and gather wild plants. The clay tokens they made were quite sophisticated in form. There were four basic types of token: spheres, disks, cones and cylinders. In addition there were tetrahedrons, ovoids, triangles, rectangles, bent coils and schematic animal forms. Subtypes included half-spheres and cones, spheres and disks with incisions and with punch marks. The set totaled 20 individual symbols.

The Neolithic period and the succeeding Chalcolithic period, or Copper Age, in western Asia lasted about 5,000 years. Over this substantial span one finds surprisingly few changes in the tokens, a fact that may indicate how well suited to the needs of an early agricultural economy this recording system was. In about 6500 B.C., 2,000 years after the rise of the first Zagros farming communities, another Iranian village, Tepe Sarab, began to flourish. The token inventory from excavations at Tepe Sarab shows no increase in the number of main types and an increase in subtypes from 20 only to 28, among them a four-sided pyramid and a stylized ox skull that is probably representative of cattle.

Perhaps it was during the Chalcolithic period that the agricultural surpluses of individual community members came to be pooled by means of taxes in kind, with the supervision of the surplus put into the hands of public officials such as temple attendants. If that is the case, the need to keep track of individual contributions evidently failed to bring any significant modification in the recording system. The tokens unearthed at four sites that flourished between 5500 and 4500 B.C.−Tell Arpachiyah and Tell as-Sawwan in Iraq and Chaga Sefid and Jaffarabad in Iran−reflect no more than minor developments. A new type of token, the biconoid, appears, and among some of the subtypes painted black lines and dots have taken the place of incisions and punch marks.

Early in the Bronze Age, between 3500 and 3100 B.C., there were significant changes in the recording system. This period saw an economic advance quite as remarkable in its own way as the rise of the farming economy that laid the foundation for it. The new development was the emergence of cities. Surveys of ancient sites in western Asia indicate a drastic increase in the population of Iraq and Iran; urban centers with many inhabitants begin to appear close to the earlier village settlements.

Craft specialization and the beginnings of mass production appear at this time. The bronze smithies and their products gave the age its name, but craftsmen other than smiths also arose, concentrated in various areas. The invention of the potter's wheel allowed the development of a pottery industry, and the output of various mass-production kilns came to be distributed over great distances. A similar trend is apparent in the manufacture of stone vessels, and the development of an expanded trade network is indicated by the appearance in Iraq of such exotic materials as lapis lazuli.

The development of an urban economy, rooted in trade, must have multiplied the demands on the traditional recording system. Not only production but also inventories, shipments and wage payments had to be noted, and merchants needed to preserve records of their transactions. By the last century of the fourth millennium B.C. the pressure of complex business accountancy on the token system becomes apparent both in the symbols and in how the tokens were used.

To consider the symbols first, six sites of the late fourth millennium B.C. in Iraq (Uruk, Tello and Fara), in Iran (Susa and Chogha Mish) and in Syria (Habuba Kabira) have yielded tokens representative of the full range of early shapes. In addition, some new shapes appear, among them parabolas, rhomboids and replicas of vessels. Even more significant than the appearance of new shapes, however, is the great proliferation of subtypes indicated by a variety of incised markings on the tokens. It is also now that a few of the tokens begin to have appliqué markings: added pellets or coils of clay.

The six sites have yielded a total of 660 tokens dating to about 3100 B.C. Of this number 363, or 55 percent, are marked with incisions. Most of the incisions are deep grooves made with the pointed end of a stylus; the grooves are placed conspicuously and with a clear concern for symmetry. On rounded tokens such as spheres, cones, ovoids and cylinders the incisions usually run around the equator and are thus visible from any aspect. On flat tokens such as disks, triangles and rectangles the incisions appear only on one face.

Most of the incisions present a pattern of parallel lines, although incised crosses and crisscross patterns are also found. The number of parallel lines would not seem to be random: there can be as many as 10 incisions, and the frequency of one-stroke, two-stroke, three-stroke and five-stroke patterns is conspicuous. It is noteworthy that with the exception of two-stroke patterns odd-numbered patterns are the most frequent.

Although incised patterns are by far the most abundant, 26 of the tokens (some 4 percent of the total) show circular impressions apparently made by punching the clay with the blunt end of a stylus. Some of the punched tokens bear a single impression. Others show a cluster of six punches, arranged either in a single row or in two rows with three impressions each.

As for changes in how the tokens were used, it is significant that 198 of them, or 30 percent of the total, are perforated. The perforated tokens run the gamut of types and include subtypes of the unmarked, incised and punched variety. In effect this means that tokens of any type or subtype were available in both unperforated and perforated forms. The perforations are so small that only a thin string could have passed through them. Of the explanations that come to mind one is that all 15 types of tokens and their 250 subtypes are nothing more than individual amulets that the early Bronze Age urban folk of western Asia wore on strings around their neck or wrist. I reject this explanation on two grounds. First, none of the perforated tokens that I have examined shows any evidence of being used as an amulet, such as wear polish or erosion around the string hole. Second, it seems preposterous that such a complex repertory of forms, so widespread in geographical distribution and manufactured with such remarkable uniformity, should have served as personal adornment in 30 percent of the cases and for some other purpose in the remaining 70 percent.

I prefer the hypothesis that some tokens representative of a specific transaction were strung together as a record. It seems at least plausible that the complexity of record keeping in an urban economy might have given rise to duplicate tokens suitable for stringing.

The stringing of tokens, if that is what the perforated tokens imply, would be only one change in how these symbolic bits of clay were used at the end of the fourth millennium B.C. A much more significant change is the first appearance at this time of clay bullae, or envelopes, such as those Amiet found as containers of tokens at Susa. The existence of a bulla is clear-cut direct evidence of the user's desire to segregate the tokens representing one or another transaction. The envelope could easily be made by pressing the fingers into a lump of clay about the size of a tennis ball, creating a cavity large enough to hold several tokens; the envelope could then be sealed with a patch of clay.

There is no doubt in my mind that such bullae were invented to provide the parties to a transaction with the kind of smooth clay surface that according to Sumerian custom could be marked by the personal seals of the individuals concerned as a validation of the event. The fact that most of the 350 bullae so far discovered bear the impressions of two different seals lends support to my conviction. Amiet has suggested that the Susa bullae may have served as bills of lading. In this view a rural producer of, say, textiles would consign a shipment of goods to an urban middleman, sending along with the shipment a bulla that contained a number of tokens descriptive of the kind and quantity of merchandise shipped. By breaking the bulla the recipient of the shipment could verify the makeup of the shipment; moreover, the need to deliver an intact bulla would inhibit the carrier from tampering with the merchandise in transit. This sealed transfer of tokens between trade partners represents an entirely new way of using the ancient recording system.

The innovation had one serious drawback. The seals impressed on the smooth exterior of the bulla served to validate each transmission, but if the seal impressions were to be preserved, the bulla had to remain intact. How, then, could one determine what tokens were enclosed and how many? A solution to the problem was soon found. The surface of the bulla was marked so that in addition to the validating seal impressions, it bore images of all the enclosed tokens.

The most striking example of this stratagem is a bulla that proved to contain six grooved ovoid tokens. Each of the six, tokens had been pressed into the surface of the bulla before being stored inside it; they fit the surface imprints exactly. This means of recording the contents of a bulla on its exterior was not, however, universally practiced. On most bullae the impression was made with a thumb or a stylus; a circular impression stood for a sphere or a disk, a semicircular or triangular impression stood for a cone, and so forth.

The bulla markings were clearly not invented to take the place of the token system of record keeping. Nevertheless, that is what happened. One can visualize the process. At first the innovation flourished because of its convenience; anyone could "read" what tokens a bulla contained and how many without destroying the envelope and its seal impressions. What then happened was virtually inevitable, and the substitution of two-dimensional portrayals of the tokens for the tokens themselves would seem to have been the crucial link between the archaic, recording system and writing. The hollow bullae with their enclosed tokens would have been replaced by inscribed solid clay objects: tablets. The strings, baskets and shelf loads of tokens in the archives would have given way to representative signs inscribed on tablets, that is, to written records.

The convex profile of the early Uruk tablets may well be a morphological feature inherited from the spherical bullae. Much the same may be true of the selection as a writing surface of a material as unsuitable as clay, a soft and easily smeared medium that must be dried or baked if it is to be preserved. There can be little doubt about the relation between the shapes and markings of the tokens and the supposed arbitrary forms of many Uruk ideographs. No fewer than 33 clear-cut identities exist between the ideographs and two-dimensional representations of tokens and more than twice that many are possible.

To summarize, the earliest examples of writing in Mesopotamia may not, as many have assumed, be the result of pure invention. Instead they appear to be a novel application late in the fourth millennium B.C. of a recording system that was indigenous to western Asia from early Neolithic times onward. In this view the appearance of writing in Mesopotamia represents a logical step in the evolution of a system of record keeping that originated some 11,000 years ago.

On this hypothesis the fact that the system was used without significant modification until late in the fourth millennium B.C. seems attributable to the comparatively simple record-keeping requirements of the preceding 5,000 years. With the rise of cities and the development of large-scale trade the system was pushed onto a new track. Images of the tokens soort supplanted the tokens themselves, and the evolution of symbolic objects into ideographs led to the rapid adoption of writing all across western Asia.