Writing and record-keeping in early cities

DANNY LAW, WANG HAICHENG, HANS J. NISSEN AND GARY URTON

Writing and other technologies for enhancing human memory and the reach of communication seem, in many instances in the ancient world, to have a special relationship with the rise of ancient urban centers.

Ancient cities could not operate without some sort of common system of representation to connect and allow coordination among disparate groups, which they normally realized under the stewardship of an administrative bureaucracy. At some point in the gathering and ordering of people in cities, the logistical and administrative coordination required by the close proxim­ity of so many interdependent people also must have outstripped the human mnemonic capacity, producing a demand for an additional specialization, record-keeping, to track and direct the flow of people and resources that made up and provisioned the city. With the emergence of cities, hierarchy and control had become fixed and elaborated. This required institutions responsible for marking distinctions and for surveillance. The greater size and density of urban populations (in comparison with those of villages), and, arguably, the increased technical specialization that cities made possible meant that more potential inventors were around to find solutions meeting these needs, while inter-city connections facilitated the diffusion of these solutions to other places with similar needs.

Writing, in the context of the emerging needs of major cities, was an invention of immediate value that seems to have originated independently in a relatively few places in the Old and New Worlds, all of which were urban centers, and then was adopted rapidly by many other contemporary and later cities. But not all major cities had writing and the precise way that writing evolved and was deployed across the ancient world differs greatly depending on where we look. In this chapter, we will compare the diverse instances, presented in the preceding chapters, in which writing and other forms of record-keeping developed apparently in close coordination with burgeoning cities. Cases from China, Mesopotamia, the Andes, and Mesoa­merica show that responses to the challenges and possibilities that emerged because of urbanization are almost as numerous as the cases investigated. In spite of the variation, parallels do emerge. These similarities, across so many contexts, provide a better picture of some of the less visible changes to society through the urban revolution, as well as the social nature of writing and other systematic forms of representation and record-keeping.

Types of communicative technologies

The cases surveyed in the preceding chapters provide a rich sample of the diversity of technologies for recording and communicating in early cities. Before we can compare these ancient cities, however, it is important to establish exactly what the object of study is. How we define both writing and record-keeping more generally reflects our understanding not only of the material forms and symbolic meanings involved, but what those forms were used for. Written records, any kind that uses a tangible medium to register information, have two important qualities to bear in mind in our comparison of these technologies across the ancient world. Writing is not the only human solution to the need to store and communicate information. Humans innately have a finite, if large, capacity to store information and are wired for supple language to communicate that store of information to others.

Writing and other forms of durable, external communication are not, however, independent of individual memory nor spoken language. Portraying writing as language in visible form does not allow us to discuss the differences between diverse technologies, from Maya hieroglyphs, Inkan khipus, Late Uruk cuneiform writing, and earlier Mesopotamian seals. Recording technologies rely on the memory and shared understanding of both author and recipient, since in order to be effective, a record must make use of appropriate forms, and the recipient must know the proper interpret­ation of those forms. Ethnography and common sense tell us that the life of objects and the life of people are tightly entangled: a wedding ring, a stuffed animal, a picture of a family reunion... Each object is a memory aid, embodying a piece of memory to somebody. They become a sign, a representation of something else. But in many cases, these “aides-memoires” are idiosyncratic. They only mean something to their maker because of that person's unique lived experience. While an object or sign might help that individual recall something that would otherwise be difficult to retrieve, anyone else confronted with these idiosyncratic signs would be unable to recover the original meaning assigned to them by their maker. In order for a sign to be communicative, that is, to express information not already known by the interpreter, it needs to have a shared, socially circulated value or meaning, one that anybody initiated into the use of that particular sign would recognize.

But the utility of a single sign is limited. The more signs with socially circulated values, the greater flexibility in communication and the less the need for contextual cues, as strings of signs provide context for one another. A set of signs that follows the same rules of use, modality, and social function produces a communicative system, one that can be used, among other things, to encode historical, logistical, administrative, or ritual mes­sages in such a way that the information can be retrieved at a later point by anyone initiated into the rules of the system. The message does not need to be known in advance, only the system of representation. Spoken language is the universal example of this. But all of the ancient record-keeping and communicative technologies discussed here are also examples, though they differ substantially in their relationship to spoken language. Recording systems such as early cuneiform (Chapter 6) and khipu (Chapter 9) use conventionalized marks, or what we could term “markers” (in the case of the khipu), but are not necessarily bound to a particular language. The oracle bone inscriptions in Shang China (Chapter 7) and Maya hieroglyphs (Chapter 8) are capable of communicating exact graphical transcriptions of spoken language. A system that represents spoken language has the advantage of allowing people to communicate with that system anything that they are capable of talking about. However, that flexibility comes at a cost, as messages will often need to be quite lengthy and they will be unavailable to individuals who do not speak the language in question. Other forms of representing information, by restricting the semantic range of the system, are able to communicate certain types of information much more efficiently. Thus the color of a cord or a particular type of knot in the khipu system can communicate information to the initiated in a “semasiographic” modality in a manner that would likely require many words, if not para­graphs, to adequately describe in language.

It is difficult to compare the communicative technologies that emerged in the context of ancient cities in different parts of the world. The comparison is hampered, in the first place, by the clear lacunae in the surviving corpus, particularly from China and Mesoamerica. These ancient writing systems seem, from the surviving record, to stride onto the scene fully formed with the purpose of recording the complexity of spoken language. In the chapter on Maya writing, Law suggests that the precursor to Maya writing, or at least a crucial element in the context of its initial innovation, was a rich and highly conventionalized iconographic tradition in which cosmological meanings were central.

In the Andean case, a wide range of technologies had for millennia been based on the construction of various configurations of spun and plied cords; therefore the ancient Andeans seem to have naturally turned to the possi­bilities offered by the cord medium (for example, construction variation, thickness, color, various topological configurations such as knots and attach­ments). Inkan khipus are not obviously connected to spoken language, and would only have been interpretable within certain carefully defined param­eters, such as the management of tribute and corvee labor forces. Thus, we may say that when facing the need to devise a record-keeping system, a society will first turn to media that are readily available and familiar from existing technologies; later systems may innovate on the original medium and if existing media do not offer sufficient flexibility and variability to achieve the level of signing required, a new medium may be chosen.

A good example of the invention of a new medium for records is the invention of paper in early imperial China, which gradually replaced the use of wood and bamboo slips that had been the dominant writing surface during the Bronze Age. Like clay in Mesopotamia and cord in the Andes, wood and bamboo were cheap and readily available materials that had been used for various purposes (building, carpentry, basket-weaving, etc.) millen­nia before the rise of early cities; some early and middle Neolithic fragments have survived to this day under waterlogged conditions. The use of a brush in Neolithic times can likewise be ascertained by examining the exquisitely painted pottery; the calligraphic lines on bronze decorations suggest that the skillful wielding of the brush was continued into the Bronze Age. It is not surprising, therefore, that wood and bamboo were chosen as the media for record-keeping, not just used as a surface to convey brushed marks, but also made into calculi, comparable to the use of clay tokens and cotton or camelid fibers to count things in Mesopotamia and the Andes, respectively. Although the earliest extant bamboo slips are only dated to the fifth century bce, their use at Anyang since the twelfth century is confirmed both by a character depicting an actual book (made of slips bundled together by cords), and by some bronze inscriptions, the peculiar columnar format of which strongly suggests that they were modeled on individual slips. One scholar¹ suggests that the characters for numerals in the oracle bone inscrip­tions originated from the different arrangement of bamboo calculi on a flat surface, echoing Schmandt-BesseratV theory on the origin of cuneiform numerals in the shape of token impressions.

In Mesopotamia, on the other hand, surviving materials show that strategies of record-keeping run the range from counting tokens and owner seals with a very limited range of meaning to cuneiform texts clearly based on spoken language. The counters and stamp seals were only able to denote simple numbers of goods or a sealing individual. It isn't until the middle of the Late Uruk period (Uruk Level VI) that we begin to find artifacts that were able to store both pieces of information in the same mode. It has not been possible to identify the language behind the earliest texts. Yet even those clearly linguistically based writings were ^{[137] [138] not used to directly capture speech, but instead employed a truncated style of expression, serving as aides-memoires.}

Each of these types of communication technology has strengths and weaknesses that have consequences for the kinds of functions to which they could be put in early cities. In the next section, we will turn to a comparison of these diverse functions, and how those functions appear to have changed over the centuries and millennia.

Functions

The functions to which information technologies were put in early cities run the gamut from economic administration to the performance and commem­oration of ritual. A functional description of these different systems must necessarily attend to the trajectory of development: for what purpose was the technology originally developed, and to what purposes was the technol­ogy later applied? In truth, of the cases surveyed here, only the Mesopota­mian context has direct data to support a reasonably complete sequence of development. To generate hypotheses about the purpose of the initial invention of writing or other information technologies in China, Mesoamer­ica, or the Andes, inferences must be made from other documented cases (Mesopotamia), or from the functional domains of use apparent in the surviving material record. Here we will focus on three salient functions, economic administration, accounting, and ritual, and what these case studies suggest about how these relate to the development of writing around the world.

Economic administration

By far the most widely proposed functional reason for developing writing or other information technologies was to facilitate the economic administra­tion of large communities. When Uruk first grew to the size of a city, people could avail themselves only of seals to leave personal marks on the surface of clay fasteners, and clay tokens of different shapes used to denote numbers. These technologies allowed for the recording of quantity, type, and ownership of goods, but the limited number of types in the sign repertoire and the inefficiency of this method for dealing with large numbers of objects, as well as the growing demands of an ever increasing economic administration, led to attempts to increase the range of storeable infor­mation types. In Uruk, incremental advances in these technologies of accounting and administration are attested archaeologically: sealed clay bullae with tokens inside; sealed clay tablets with numerical indentations. Only after a considerable time did the first system of writing emerge as a tool. Cuneiform offered a relatively economical method to store as many pieces of information as were desired. However, as an extension of the former methods, it would be hard to make the case that writing was intended to represent language from the onset. That it did so to a degree was secondary to the quest for greater expressive range in records.

Because of a relatively complete sequence of development found at Uruk, and its clear administrative and accounting function, from the beginning, scholars often assume that economic administration, control and monitor­ing of flows of goods and services throughout a city would be the impetus for all of these advanced record-keeping technologies. Nevertheless, it is important to note that aside from Mesopotamia, there is precious little direct evidence for the actual trajectory of development. Early cuneiform at Uruk, as with Inkan khipus, was first and foremost a way to aid in accounting and economic administration, both of goods and of labor. They were records of the flow of goods and people involved in transactions. Writing is used in the beginning to denote only those items that are deemed necessary to reconstruct a certain transaction. The origin of the goods is never mentioned, presumably because it would have been obvious back­ground knowledge. At the time, these inscriptions recorded and monitored goods entering or leaving the central store.

At Uruk some major administrative tasks that were covered by documents include land survey, calculating the amount of seed and distributing it to each field, and feeding the laborers. Animal husbandry was another import­ant set of activities that needed to be controlled. The tablets gave a superior and clear sense of the amount and whereabouts of goods and manpower. This knowledge would allow the decision-makers to allocate them accord­ingly. In other words, the tablets served budgeting purposes. The contents of the tablets leave no doubt that redistribution played an important role in the city's economy - the provision of city dwellers has always been a real and day-to-day concern and it goes without saying that redistribution must have involved a certain degree of advance planning. At Uruk the building of monumental architecture and the production of elite objects would have been inconceivable without budgeting the city's resources.

The central institutions at both Uruk (in the Eanna precinct) and Cuzco seem to have had extensive control over the economic transactions within the city and its hinterland. They allocated large tracts of land to various officials and organized labor forces, presumably to work in the fields and for public projects such as digging canals and building roads. The khipu accounts generated at Cuzco must have served similar budgeting functions, enabling the Inka and his royal counsel to move goods and corvee labor along their famed highway system, which in itself was a superb example of central planning. The provincial governor (Tocricoc) gathered the summa­tion of figures from the provincial level, which in turn were summations of data from the local level. The knowledge needed for the state budget had to be very abstract and drastically simplified, consisting mainly of numbers in various categories.

In China and the Maya Lowlands, however, there is virtually no direct evidence of surviving early texts that were explicitly used for accounting or economic administration. Palace scenes painted on polychrome ceramic vessels and wall murals from the Maya Late Classic period (600-900 ce) show bundles of goods with hieroglyphic labels of quantity and type (8,000 cacao beans, for example), suggesting that writing was used among the Classic Maya for economic administration of some degree, but no direct evidence of such uses in earlier periods exists. In the absence of direct data, we can only infer their existence, since the need in these early cities would have existed, and the technology would have certainly allowed for the production of such records; therefore, it is very likely that writing was used from early on in both China and the Maya Lowlands for economic administration.

Accountability

One important function of the khipu at Cuzco and other Inka cities and administrative centers is that the technology also served as proof of the official's fulfilment of his duty to execute and document the administrative task so ordered by his superior. Both actual khipu samples and early Colonial documents suggest that the khipu system was itself one of checks and balances. Under the principle of dual organization, the cord-keeper of one moiety could point to a khipu and declare it recorded the truth about his moiety, but his counterpart in the other moiety kept his own copy of statistics regarding the first moiety, and vice versa. If the two copies did not match, then something was wrong. It was a strategy of group surveil­lance, but it also guarded against individual administrators' mistakes or malfeasance.

Administration is a marked form of control. It creates accountability, which invites explicit accounting practices that record the administrative tasks generated by the system. Most of the early cuneiform administrative

Writing and record-keeping in early cities tablets and the khipus list the exact number of a certain material or labor, leaving no doubt that the liability for materials or labor involved in the task was of great concern to the central institutions. But whose liability was incurred: the institution's, a specific office's, or an individual's? In the Near East a time-honored way to record individual liability was with glyphic seals whose impressions on clay identify the individuals or the responsible parties involved in economic transactions. Some early cuneiform tablets were sealed, but they seem to have been exceptional. Some officials' names can be identified, but again they seem to have been rare. However, considering that the proto-cuneiform texts span a time of 200 to 300 years, and that some of these give evidence of daily account-making, the overall number of close to 6,000 is only a fraction of what once must have been written. It is very likely that we are missing entire thematic groups.

A similar concern with accountability can be seen in Shang China on some narrative inscriptions on the interior of ritual bronzes, inscriptions that would have been much easier for the ancestral spirits to access than for a living audience, especially when the vessels were filled with food or wine. These inscriptions usually state that a court official was rewarded for some deeds so he cast a bronze for his ancestor so-and-so: “On the day guisi the king awarded the Xiao Chen Yi ten strings of cowries, which he used to make this sacred vessel for Mu Gui. It was in the king's sixth si, during the yong cycle, in the fourth month.”^[139] This is a written report. It has a list consisting of numbers and names, reminiscent of the administrative lists at Uruk and the information recorded in khipus at Cuzco. But the narrative content of this report would be hard to derive from the highly circumscribed spreadsheet-like systems for describing quantities and categories used in those places. Oral narrations during the ancestral ritual would be one way to make this report, but these would be ephemeral. Full writing that was attached to language made possible communication across time and space.

The mate to a written report from the subordinate to the superior is a written authorization from above. The oracle bone inscriptions sometimes specify the source of authorization for certain actions (for example, plowing and harvesting) with a complete sentence: “The king orders/commands so- and-so to carry out such-and-such a task.” The king might worry about the accuracy of oral transmissions, but his officials are perhaps more anxious to

have a written instruction in order to protect themselves in the future. In short, the pair of authorization and reporting in writing constituted another aspect of accountability. Unlike proto- and early cuneiform and khipu records, these texts from Anyang could be classified as letters of a sort, written to communicate without relying on the unreliability of oral com­mentaries. Because administrative documents were almost certainly written on perishable materials like wood and bamboo, we will probably never find them. Only after writing came to be used for display does archaeology begin to find traces of it.

Ritual activity

The straightforward logistical accounting that dominates khipus and early cuneiform inscriptions is essentially absent from both Chinese and Mesoa­merican texts. Some of the divination texts at Anyang show characteristics of both display and administration. Likewise, while the subject matter of many Maya hieroglyphic texts is commemorating ritual activities, in another sense they are a statement of fulfilment of ritual obligations - a type of ritual accounting.^[140] As John Baines has pointed out, display and administration are not mutually exclusive functions.^[141] The manner in which a text is displayed is related to its anticipated and intended audience. Magistrates in many soci­eties have been accountable to both human and divine interlocutors, and the need to provide an account, in these cases, through writing, unifies logistical administration and ritual activity.

Like the Preclassic Maya inscriptions, the size of the Chinese characters meant for display on the oracle bones is diminutive, visible only at close range. The audience for actual physical inspection of this display therefore must have been small, consisting of the inner elite around the royalty and the diviners. Though not public, the texts still constitute a report of activ­ities, suggesting close monitoring and control, whether by man or deity. Although writing was not part of the divination process at Anyang, and hence not essential in communicating with the royal ancestors, royal ances­tors were nevertheless part of the intended audience of the divination texts, as can be deduced from the fact that the bones were almost exclusively buried in the royal precinct, close to the ancestral temples. After a period of accumulation above ground these bones would be buried in large pits, reminiscent of the disposition of sacrificial animals and humans outside the ancestral temple. At Anyang one royal duty was to consult divine intentions on matters concerning the dynasty's rule. The display of the medium of divination, sometimes inscribed with the divined questions and outcomes, was possibly intended as a proof of the discharge of this royal duty. This suggests an administrative concern, namely accountability, espe­cially toward the ancestors.

Most of the early Mesoamerican writings, including the dozen or so Preclassic Maya texts, seem to have used grammar and syntax of an actual language to express narrative contents, to judge from their length, linearity, and iconographic context. Some Preclassic texts perhaps include a mixture of logographs, syllables, and dates (including several concurrent calendar cycles, which in turn were composed of numbers and logographs for days/ months). The carriers of these texts were of various types: wall murals, stone stelae, and portable objects usually made of precious stone. The latter two types have lost their original contexts, while wall murals seem to have existed mainly in temple buildings or pyramids, often as part of larger ritual architectural complexes constructed in accordance with Maya ritual and cosmology. These complexes arguably marked the creation of the city as a moral community, under the leadership of the king, whose duty was to perpetuate cosmological structures and operations.^[142] Although the early texts on these murals remain mostly undeciphered, they do have a recognizable glyph for “king” ajaw, associated with depictions of ancient royalty. This, as well as the sacred setting of these murals, seems to suggest that their contents and purpose were closely related to Maya kingship and cosmology.

Since the setting for these texts was explicitly private, secret, and elite, public display does not seem to have been a major function of writing at this stage, in contrast with later, Classic period texts. The tiny scale of the Preclassic Maya texts, even on monumental murals, suggests an expectation of close and careful study of the text by a small and privileged audience. The absence of monumentalism in these Preclassic texts, in contrast with the large figural scenes that they accompany, indicates that the placement of glyphs on monuments was intended to caption pictures, a function continued into the Classic period, by which time the scale of the glyphs had been enlarged to be proportional to the image. The glyphs could be names of gods and people, just as they are on inscribed objects of the Classic period. Some early texts on portable objects appear to be essentially lists of nouns, in a sense similar to the lists found in Uruk. Yet the nature of the recorded nouns is different. In early cuneiform it is the names of commodities that dominate, with only a few attested personal names. The reverse seems to be true in Preclassic Maya texts. Lists of deities occur on small portable objects with intrinsic value, perhaps as a means to identify their ownership, or to solicit godly invocations, or to provide incantatory cues for those reading the names. Names of real persons on small objects probably served as name­tags to label the owners or makers.^[143]

The preoccupation with names of deities and powerful people in Maya writing, and their function as labels of images, betray an elite obsession with cosmological symbols and a fetishization of images and written names. This was the Maya's response to a human drive to represent the world around us and at the same time to create that world through representation. The act of naming is one manifestation of that impulse. Names have an intellectual or psychological importance that is well summed up by the Egyptologist Barry Kemp: “[T]o the ancients knowing the name of a thing made it familiar, gave it a place in one's mind, reduced it to something that was manageable and could be fitted into one's mental universe.”^[144] Graphical recording systems were devised to store and retrieve information across space and time very early on in human history. But it was not simply a desire to overcome bodily limitations of memory, time, and distance that prompted people to devise these systems. Through devising and using these systems people sought to capture the world. The compilation of lexical lists at Uruk, the veneration of a written king list at Anyang and the Maya Lowlands (as well as in Eygpt), and the census data at Cuzco were all attempts to represent elements of the universe. One main difference between them and the Maya case is that in the former the emphasis of representation seems to have been to attain intellectual control over man's immediate environment, the more mundane aspects of the universe, including both material and human resources, while in the latter the Maya seem to have been chiefly interested in the more spiritual aspect, seeking to connect the human world with cosmological power. Perhaps it is telling to recall that at Uruk there was a list of professions in hierarchical order but no list of gods, in contrast to the situation in Maya cities.

Another difference between writing in Mesopotamia and Mesoamerica is that writing was not used to caption divine and royal figures on valuable objects from Uruk (though this did happen after the Uruk period), but for the Maya to write on a surface was to create objects of value in their own right. The content and the artistic quality of Maya writing, together with the expensive materials and the religious contexts, all made the inscribed objects into valued and powerful goods. In this respect inscribed Shang bronzes and jades from Anyang are comparable examples. But if an object's value is increased by the addition of an inscription, there has to be a commonly accepted recognition that writing, especially beautiful writing, is valuable. The only social mechanism to bring out this recognition is to create a script community, the members of which would accord great value to the thing that binds them together. Such communities must have existed in all cities employing graphical recording systems for communication, because com­munication depends on signs with shared meanings. Thinking in concrete terms, there must have been some sort of supporting infrastructure to perpetuate, mediate, and legitimize these systems of communication. This supporting infrastructure - schools, institutional supports, and the like - is the subject of our next section.

Supporting infrastructure

Training is indispensable if a record-keeping system is to be kept alive and functional. In Mesopotamia, pre-writing systems such as different kinds of seals, tokens, and their combinations were already complex enough to require established methods to transfer them, along with other skills, like measuring fields or performing mathematics. The literate civilizations in Mesopotamia, Egypt, China, and Mesoamerica all have a long history of script use - in the first three cases more than 3,000 years - so they clearly had effective means for teaching the scribal art. The Mesoamerican and Andean civilizations established schools to teach young students other ways of storing and communicating information. The training of scribes involves several interlocking key factors: a teaching place - a school in the physical sense; a curriculum - procedural and conceptual knowledge to be imparted; institutional or private sponsorship; sources of teachers and students; the logistics of running the school; and the occupations for which the students are being prepared. In many cases, the details of these important institutions and practices are not present in the surviving material record, though in all cases some inferences can be made.

Of particular interest for understanding scribal training at Uruk is the presence of so-called lexical lists: lists of words and phrases, arranged according to semantic groups. In certain erudite circles, attempts to control their universe had led to lists of names, which during the invention of writing served as guidelines for setting up the system. These lists were faithfully copied in large quantities over a period of more than a thousand years and held in high esteem because of their role in the initial process. Lists account for about io percent of the total extant archaic cuneiform tablets; the rest are administrative texts. There are no letters, legal docu­ments, or literary works. This distribution of contents suggests that in the earliest phase of script development in Mesopotamia, the simple word lists used for scribal training were indispensable, in addition to learning the system of administrative control, and not other genres of writing featuring connected discourse.^[145]

Another important point is that the curriculum of early schools would have involved training in a variety of other skills to be mastered by the scribes: the making of tablets, the layouts of the different tablet formats, book-keeping procedures, mathematics, history, mythology, and in some cases ritual practice. In early texts from Uruk, information about the relationships between entries and groups of entries in an administrative tablet is coded in a bewildering array of sub-cases, sub-columns, and varying column widths. Numbers make up a large part of the content in the archaic tablets. Unlike later Mesopotamian arithmetical practice, which principally employed the sexagesimal system regardless of the objects that were to be qualified, archaic book-keeping has several numerical systems that were used for different objects: the bisexagesimal system, the grain capacity system, the area system, and other systems that are still poorly understood. There were also derived systems for time-keeping and measurement.^[146] The choice of a specific numerical system roughly corresponded to the bureau­cratic division (land surveyor, tax collector, etc.). It is possible that individual scribes needed only to learn one system specific to their office. These non- grammatical and non-syntactical devices for encoding information were

Writing and record-keeping in early cities developed centuries before grammatical and syntactical elements appeared.^[CXLVII] Connected discourse was not the stimulus for the invention of early cuneiform; ledgers then as now did not need complete sentences.

Several oracle bone inscriptions at Anyang contain a character that has been transcribed as xue, which in later classical Chinese has three basic meanings: (ι) school; (2) to teach; (3) to learn. One fragment of an inscription mentions the approval of a proposal to build a xue, possibly within the Shang royal residence. Besides court schools there seem to have existed schools located outside the royal household. One inscription reads, “Crack-making on the day bingzi, divining: ‘[Should] the Many Children go to school? Will it not rain on their way home?'” A similar divination was made on the next day. Still another inscription seems to imply that noblemen or their children from other polities were “taught and admonished” in the Shang capital. Could these children be political hostages who nevertheless received educa­tion in court schools, like the ones in the Inka Empire (see below)?

The oracle bone inscriptions tell us little about what was taught in the schools. Some inscriptions mention learning ritual dance and music, but not necessarily in schools. No compelling evidence for literacy schooling at Anyang exists - apart from the conclusive evidence of literacy itself. A character written with brush and ink on a disused potsherd is at present one of the few direct witnesses of writing practice, and it is far too skillfully executed to have been done by a beginner. That Mesopotamian pupils spent much time practicing simple wedges and their combinations reminds us that Chinese beginners likewise have always had to start from basic strokes and then proceed to simple characters, adhering to a strict stroke order. Unfor­tunately no student exercise containing only basic strokes is extant from Shang times.

Ethnohistorical records provide more detail about education and training in the Inka Empire. Inka rulers certainly gave serious attention to the education of both the male and female members of the nobility. Lower- ranking women were trained to weave, cook, and brew for the state; higher- ranking women additionally were instructed in religious matters. Important provincial nobles were required to send their sons and close relatives to the court in Cuzco at the age of fourteen or fifteen years. Together with the

sons of Inka nobles they attended yachawasi, special schools run by learned men (amautakuna) who were also noblemen. Among the school children in the court schools were the eldest sons of the most important provincial nobles. They were hostages for their fathers' loyalty and second-generation nobles-in-training at the same time.

A later writer, Martin de Munaa,^[148] informs us that the length of study at the court schools was four years. He gives us an outline of the curriculum organized by year. It included not only the chief subject - the court version of the Inka language - but also Inka rituals and calendrics, khipu record­keeping, Inka history, law, statecraft, military tactics, and behavior appro­priate to the students' social class. Although Murιua's account is suspiciously Europeanized, the subjects taught do not conflict with those listed by earlier writers. However, it is not clear what kinds of khipu record-keeping were taught to what kinds of student at school. Other accounts seem to suggest that there was a specialist group of “khipu makers” (khipukamayuq). How they transmitted the knowledge of reading a khipu is not recorded by the Colonial chroniclers. A seventeenth-century friar, Antonio de la Calancha, left us the only general account about studying khipus. But we are not even sure whether the “khipu makers” actually made the khipus or simply read them.

whether because of the privileges with which they honored the office, or because if they did not give a good accounting concerning that on which they were questioned they would be severely castigated, they [the khipu­kamayuq] continually studied the signs, ciphers, and relations, teaching them to those who would succeed them in office, and there were many of these Secretaries, each of whom was assigned his particular class of material, having to suit [or fit] the story, tale, or song to the knots of which they served as indices, and points of “site memory.”^[149]

In contrast to the sparse accounts of the learning of record-keeping, there are numerous reports that the Inka rulers ordered all their subordinates to learn the Inka lingua franca - Quechua. The order recognized the problem of heteroglossia in the vast empire and solved it in a way that was simple and consistent with the state ideology, making one language official and requiring its use. To make such an order effective the Inka rulers must have required the establishment of local schools to educate the sub-elite. But we are ignorant of details, such as who had the right or obligation to go to the local schools. It is clear that the recruitment policy for the court schools recognized the authority of the local nobles. In the meantime by educating (and indoctrinating) the sons of the local and provincial nobility, the court sought to secure the loyalty of local and regional administrators, because it was the sons who attended the Inka court schools who would succeed to their fathers' offices, not their siblings who were not educated in the capital. Inka control of the provinces was further strengthened by the authority of governors appointed directly by the Inka rulers.

Little is known about scribal training in the ancient Maya tradition. There is a term “houses of writing” (ts’ibal na:h) in Classical Mayan, which, like the Sumerian “tablet house” or Egyptian “room of teaching” or “house of life,” might refer to scribal schools. Several architectural structures, at Copan and elsewhere, have been identified as houses of writing, both because of texts and iconography on the structures themselves, and, more unusually, because of the discovery of remains of scribal paraphernalia (inkwells, mortar and pestle for grinding pigments), as identified by Takeshi Inomata at the Late Classic site of Aguateca.^[150] There are only a few texts that can be compared to the pedagogical lists of Mesopotamia and Egypt. A glyph band carved on stone blocks on a “house of writing” at Chichen Itza pairs glyphs that share a vowel but differ in their consonants. This is similar in some ways to the tu-ta-ti syllabic list from Mesopotamia, and, as Stephen Houston has suggested, may well represent a Maya syllabic primer.^[151] A more recent discovery of astronomical tables painted on the walls of a structure at Xultun, Guatemala, may have been used for pedagogical pur­poses, as well as reference for experienced scribes responsible for performing complex astronomical calculations regarding the movement of the Moon, Venus, and Mars. Unlike the Mesopotamian list, which comes to us through student exercises on clay tablets, the Maya examples are permanent fixtures carved on stone or painted in a room. They perhaps functioned as a permanent and canonical model for students to copy on perishable bark and palm leaves. Alternatively it might just be a sort of display announcing the purpose of the building. The likely loss of writings on perishable materials, painful to acknowledge, unfortunately leaves much about the mechanisms for standardizing and transmitting the script to the vagaries of our imagination.

Scribal training in early cities seems to have had a mixed nature from the very beginning. City institutions (palaces, temples, and administrative offices) and private homes (teaching the teachers' own children or students from other families) complemented each other. In Mesopotamia and Egypt there seems to have been a standard pool of school texts for individual schools or teachers to choose from. This made it possible to achieve a measure of uniformity, especially in the sphere of administration. The mutual comprehensibility of scribes from different city-states in Mesopota­mia or from different nomes in Egypt suggests that a more or less universal curriculum regulated at least at first by the state and by itinerant teachers helped to set standards. Training in numerical notation and mathematics was universal because of administrative requirements. Calendrical calcula­tion, indispensable for ritual activities, was another significant subject.

Conclusions

The technologies for writing and record-keeping discussed here are diverse, both formally and functionally. Cuneiform, Chinese, and Maya hieroglyphs all developed a direct relationship with language. Inkan khipus are less tightly - or quite possibly not at all - related to language, but because of a more restricted communicative range were able to encode complex organ­izational information very efficiently. The shape of these technologies, in each instance, also reflects the history from which they emerged. The knotted-cords of khipus developed in the context of a highly valued and elaborated textile tradition in the Andes. Maya hieroglyphs developed in a rich Mesoamerican tradition of cosmological iconography. Cuneiform tablets were preceded and shaped by clay tokens and seals, and Chinese script was shaped by the bamboo strips on which it was written.

If we extend our comparison to the functions of these technologies, and how knowledge of them was maintained and transmitted to new gener­ations, we are seriously hampered by the large gaps in the surviving historical record. The richness of data in the case of Uruk and Mesopotamia suggests that administrative and economic book-keeping was the central motivation for, and function of, writing. It is perhaps plausible to extend inferences about writing in Mesopotamia to its innovation and purpose in other parts of the world, but in the absence of direct evidence, we cannot be certain. Indeed, in the case of Maya writing, an early and abiding function was ritual and cosmological, not economic and administrative.

Regardless of the functions performed by writing and other record­keeping technologies, it is clear that in order to be effective tools, these systems needed to be conventionalized and rigidly maintained. In this sense, they are a very direct physical expression of control. Writing and other forms of record-keeping depended for their existence and perpetu­ation on the high degree of uniformity and control that cities made available. As with the earliest functional motivations for the emergence of writing, we are forced to make inferences about the institutional support for these technologies on the basis of incomplete data. Insti­tutions for training scribes are apparent in the material remains for Uruk and are mentioned in ethnohistorical records for the Andes. The earliest records in China and Mesoamerica, however, offer very little direct evidence for schools and curricula. The uniformity of the systems, across time and space, makes it clear that institutions did exist, but we can only speculate about their exact nature.

Administering a city largely consists of managing taxonomies, and keeping records is an exercise in taxonomy. The aim of this chapter has been to investigate what early recording systems did and how they were perpetuated rather than just what they were. These technologies differed substantially in terms of their communicative capacities, their primary social or institutional functions, and their relationship to language. How­ever, whether representing language or not, their lexicons and numerical systems gave these early recording systems a remarkable ability to sort and quantify. In an illuminating study of modern states, James Scott has stressed the state's need for what he calls “legibility,” that is, for a clear and detailed knowledge of its population and resources. As Yoffee noted, ancient states, in the form of cities, had similar needs. With or without writing, legibility was achieved by simplification and classification. Indeed, the very act of writing and drawing is a human attempt “to reduce a complex and often chaotic reality to a comprehensible order.”^[152] Whether it is an aid in the process of urbanization, or a consequence of it, record-keeping goes hand in hand with striving for a rational ordering of the city.