Knowledge brings power. Hence writing brings power to modern societies, by making it possible to transmit knowledge with far greater accuracy and in far greater quantity and detail, from more distant lands and more remote times. Of course, some peoples (notably the Incas) managed to administer empires without writing, and “civilized” peoples don’t always defeat “barbarians,” as Roman armies facing the Huns learned. But the European conquests of the Americas, Siberia, and Australia illustrate the typical recent outcome.

Writing marched together with weapons, microbes, and centralized political organization as a modern agent of conquest. The commands of the monarchs and merchants who organized colonizing fleets were conveyed in writing. The fleets set their courses by maps and written sailing directions prepared by previous expeditions. Written accounts of earlier expeditions motivated later ones, by describing the wealth and fertile lands awaiting the conquerors. The accounts taught subsequent explorers what conditions to expect, and helped them prepare themselves. The resulting empires were administered with the aid of writing. While all those types of information were also transmitted by other means in preliterate societies, writing made the transmission easier, more detailed, more accurate, and more persuasive.

Why, then, did only some peoples and not others develop writing, given its overwhelming value? For example, why did no traditional hunters-gatherers evolve or adopt writing? Among island empires, why did writing arise in Minoan Crete but not in Polynesian Tonga? How many separate times did writing evolve in human history, under what circumstances, and for what uses? Of those peoples who did develop it, why did some do so much earlier than others? For instance, today almost all Japanese and Scandinavians are literate but most Iraqis are not: why did writing nevertheless arise nearly four thousand years earlier in Iraq?

The diffusion of writing from its sites of origin also raises important questions. Why, for instance, did it spread to Ethiopia and Arabia from the Fertile Crescent, but not to the Andes from Mexico? Did writing systems spread by being copied, or did existing systems merely inspire neighboring peoples to invent their own systems? Given a writing system that works well for one language, how do you devise a system for a different language? Similar questions arise whenever one tries to understand the origins and spread of many other aspects of human culture—such as technology, religion, and food production. The historian interested in such questions about writing has the advantage that they can often be answered in unique detail by means of the written record itself. We shall therefore trace writing’s development not only because of its inherent importance, but also for the general insights into cultural history that it provides.

The second strategy uses so-called logograms, meaning that one written sign stands for a whole word. That’s the function of many signs of Chinese writing and of the predominant Japanese writing system (termed kanji). Before the spread of alphabetic writing, systems making much use of logograms were more common and included Egyptian hieroglyphs, Maya glyphs, and Sumerian cuneiform.

The third strategy, least familiar to most readers of this book, uses a sign for each syllable. In practice, most such writing systems (termed syllabaries) provide distinct signs just for syllables of one consonant followed by one vowel (like the syllables of the word “fa-mi-ly”), and resort to various tricks in order to write other types of syllables by means of those signs. Syllabaries were common in ancient times, as exemplified by the Linear B writing of Mycenaean Greece. Some syllabaries persist today, the most important being the kana syllabary that the Japanese use for telegrams, bank statements, and texts for blind readers.

I’ve intentionally termed these three approaches strategies rather than writing systems. No actual writing system employs one strategy exclusively. Chinese writing is not purely logographic, nor is English writing purely alphabetic. Like all alphabetic writing systems, English uses many logograms, such as numerals, $, %, and + : that is, arbitrary signs, not made up of phonetic elements, representing whole words. “Syllabic” Linear B had many logograms, and “logographic” Egyptian hieroglyphs included many syllabic signs as well as a virtual alphabet of individual letters for each consonant.

Somehow, the first scribes solved all those problems, without having in front of them any example of the final result to guide their efforts. That task was evidently so difficult that there have been only a few occasions in history when people invented writing entirely on their own. The two indisputably independent inventions of writing were achieved by the Sumerians of Mesopotamia somewhat before 3000 B.C. and by Mexican Indians before 600 B.C. (Figure 12.1); Egyptian writing of 3000 B.C. and Chinese writing (by 1300 B.C.) may also have arisen independently. Probably all other peoples who have developed writing since then have borrowed, adapted, or at least been inspired by existing systems.

The independent invention that we can trace in greatest detail is history’s oldest writing system, Sumerian cuneiform (Figure 12.1). For thousands of years before it jelled, people in some farming villages of the Fertile Crescent had been using clay tokens of various simple shapes for accounting purposes, such as recording numbers of sheep and amounts of grain. In the last centuries before 3000 B.C., developments in accounting technology, format, and signs rapidly led to the first system of writing. One such technological innovation was the use of flat clay tablets as a convenient writing surface. Initially, the clay was scratched with pointed tools, which gradually yielded to reed styluses for neatly pressing a mark into the tablet. Developments in format included the gradual adoption of conventions whose necessity is now universally accepted: that writing should be organized into ruled rows or columns (horizontal rows for the Sumerians, as for modern Europeans); that the lines should be read in a constant direction (left to right for Sumerians, as for modern Europeans); and that the lines should be read from top to bottom of the tablet rather than vice versa.

But the crucial change involved the solution of the problem basic to virtually all writing systems: how to devise agreed-on visible marks that represent actual spoken sounds, rather than only ideas or else words independent of their pronunciation. Early stages in the development of the solution have been detected especially in thousands of clay tablets excavated from the ruins of the former Sumerian city of Uruk, on the Euphrates River about 200 miles southeast of modern Baghdad. The first Sumerian writing signs were recognizable pictures of the object referred to (for instance, a picture of a fish or a bird). Naturally, those pictorial signs consisted mainly of numerals plus nouns for visible objects; the resulting texts were merely accounting reports in a telegraphic shorthand devoid of grammatical elements. Gradually, the forms of the signs became more abstract, especially when the pointed writing tools were replaced by reed styluses. New signs were created by combining old signs to produce new meanings: for example, the sign for head was combined with the sign for bread in order to produce a sign signifying eat.

The earliest Sumerian writing consisted of nonphonetic logograms. That’s to say, it was not based on the specific sounds of the Sumerian language, and it could have been pronounced with entirely different sounds to yield the same meaning in any other language—just as the numeral sign 4 is variously pronounced four, chetwíre, neljä, and empat by speakers of English, Russian, Finnish, and Indonesian, respectively. Perhaps the most important single step in the whole history of writing was the Sumerians’ introduction of phonetic representation, initially by writing an abstract noun (which could not be readily drawn as a picture) by means of the sign for a depictable noun that had the same phonetic pronunciation. For instance, it’s easy to draw a recognizable picture of arrow, hard to draw a recognizable picture of life, but both are pronounced ti in Sumerian, so a picture of an arrow came to mean either arrow or life. The resulting ambiguity was resolved by the addition of a silent sign called a determinative, to indicate the category of nouns to which the intended object belonged. Linguists term this decisive innovation, which also underlies puns today, the rebus principle.

Once Sumerians had hit upon this phonetic principle, they began to use it for much more than just writing abstract nouns. They employed it to write syllables or letters constituting grammatical endings. For instance, in English it’s not obvious how to draw a picture of the common syllable-tion, but we could instead draw a picture illustrating the verb shun, which has the same pronunciation. Phonetically interpreted signs were also used to “spell out” longer words, as a series of pictures each depicting the sound of one syllable. That’s as if an English speaker were to write the word believe as a picture of a bee followed by a picture of a leaf. Phonetic signs also permitted scribes to use the same pictorial sign for a set of related words (such as tooth, speech, and speaker), but to resolve the ambiguity with an additional phonetically interpreted sign (such as selecting the sign for two, each, or peak).

Thus, Sumerian writing came to consist of a complex mixture of three types of signs: logograms, referring to a whole word or name; phonetic signs, used in effect for spelling syllables, letters, grammatical elements, or parts of words; and determinatives, which were not pronounced but were used to resolve ambiguities. Nevertheless, the phonetic signs in Sumerian writing fell far short of a complete syllabary or alphabet. Some Sumerian syllables lacked any written signs; the same sign could be pronounced in different ways; and the same sign could variously be read as a word, a syllable, or a letter.

Besides Sumerian cuneiform, the other certain instance of independent origins of writing in human history comes from Native American societies of Mesoamerica, probably southern Mexico. Mesoamerican writing is believed to have arisen independently of Old World writing, because there is no convincing evidence for pre-Norse contact of New World societies with Old World societies possessing writing. In addition, the forms of Mesoamerican writing signs were entirely different from those of any Old World script. About a dozen Mesoamerican scripts are known, all or most of them apparently related to each other (for example, in their numerical and calendrical systems), and most of them still only partially deciphered. At the moment, the earliest preserved Mesoamerican script is from the Zapotec area of southern Mexico around 600 B.C., but by far the best-understood one is of the Lowland Maya region, where the oldest known written date corresponds to A.D. 292.

Despite its independent origins and distinctive sign forms, Maya writing is organized on principles basically similar to those of Sumerian writing and other western Eurasian writing systems that Sumerian inspired. Like Sumerian, Maya writing used both logograms and phonetic signs. Logograms for abstract words were often derived by the rebus principle. That is, an abstract word was written with the sign for another word pronounced similarly but with a different meaning that could be readily depicted. Like the signs of Japan’s kana and Mycenaean Greece’s Linear B syllabaries, Maya phonetic signs were mostly signs for syllables of one consonant plus one vowel (such as ta, te, ti, to, tu). Like letters of the early Semitic alphabet, Maya syllabic signs were derived from pictures of the object whose pronunciation began with that syllable (for example, the Maya syllabic sign “ne” resembles a tail, for which the Maya word is neh).

All of these parallels between Mesoamerican and ancient western Eurasian writing testify to the underlying universality of human creativity. While Sumerian and Mesoamerican languages bear no special relation to each other among the world’s languages, both raised similar basic issues in reducing them to writing. The solutions that Sumerians invented before 3000 B.C. were reinvented, halfway around the world, by early Mesoamerican Indians before 600 B.C.

We know that the development of Sumerian writing took at least hundreds, possibly thousands, of years. As we shall see, the prerequisites for those developments consisted of several features of human society that determined whether a society would find writing useful, and whether the society could support the necessary specialist scribes. Many other human societies besides those of the Sumerians and early Mexicans—such as those of ancient India, Crete, and Ethiopia—evolved these prerequisites. However, the Sumerians and early Mexicans happened to have been the first to evolve them in the Old World and the New World, respectively. Once the Sumerians and early Mexicans had invented writing, the details or principles of their writing spread rapidly to other societies, before they could go through the necessary centuries or millennia of independent experimentation with writing themselves. Thus, that potential for other, independent experiments was preempted or aborted.

The spread of writing has occurred by either of two contrasting methods, which find parallels throughout the history of technology and ideas. Someone invents something and puts it to use. How do you, another would-be user, then design something similar for your own use, knowing that other people have already got their own model built and working?

Such transmission of inventions assumes a whole spectrum of forms. At the one end lies “blueprint copying,” when you copy or modify an available detailed blueprint. At the opposite end lies “idea diffusion,” when you receive little more than the basic idea and have to reinvent the details. Knowing that it can be done stimulates you to try to do it yourself, but your eventual specific solution may or may not resemble that of the first inventor.

To take a recent example, historians are still debating whether blueprint copying or idea diffusion contributed more to Russia’s building of an atomic bomb. Did Russia’s bomb-building efforts depend critically on blueprints of the already constructed American bomb, stolen and transmitted to Russia by spies? Or was it merely that the revelation of America’s A-bomb at Hiroshima at last convinced Stalin of the feasibility of building such a bomb, and that Russian scientists then reinvented the principles in an independent crash program, with little detailed guidance from the earlier American effort? Similar questions arise for the history of the development of wheels, pyramids, and gunpowder. Let’s now examine how blueprint copying and idea diffusion contributed to the spread of writing systems.

In a few cases, we also know something about the individuals who designed writing systems by blueprint copying in the remote past. For instance, the Cyrillic alphabet itself (the one still used today in Russia) is descended from an adaptation of Greek and Hebrew letters devised by Saint Cyril, a Greek missionary to the Slavs in the ninth century A.D. The first preserved texts for any Germanic language (the language family that includes English) are in the Gothic alphabet created by Bishop Ulfilas, a missionary living with the Visigoths in what is now Bulgaria in the fourth century A.D. Like Saint Cyril’s invention, Ulfilas’s alphabet was a mishmash of letters borrowed from different sources: about 20 Greek letters, about five Roman letters, and two letters either taken from the runic alphabet or invented by Ulfilas himself. Much more often, we know nothing about the individuals responsible for devising famous alphabets of the past. But it’s still possible to compare newly emerged alphabets of the past with previously existing ones, and to deduce from letter forms which existing ones served as models. For the same reason, we can be sure that the Linear B syllabary of Mycenaean Greece had been adapted by around 1400 B.C. from the Linear A syllabary of Minoan Crete.

At all of the hundreds of times when an existing writing system of one language has been used as a blueprint to adapt to a different language, some problems have arisen, because no two languages have exactly the same sets of sounds. Some inherited letters or signs may simply be dropped, when the sounds that those letters represent in the lending language do not exist in the borrowing language. For example, Finnish lacks the sounds that many other European languages express by the letters b, c, f, g, w, x, and z, so the Finns dropped these letters from their version of the Roman alphabet. There has also been a frequent reverse problem, of devising letters to represent “new” sounds present in the borrowing language but absent in the lending language. That problem has been solved in several different ways: such as using an arbitrary combination of two or more letters (like the Englishth to represent a sound for which the Greek and runic alphabets used a single letter); adding a small distinguishing mark to an existing letter (like the Spanish tilde ñ, the German umlaut ö, and the proliferation of marks dancing around Polish and Turkish letters); co-opting existing letters for which the borrowing language had no use (such as modern Czechs recycling the letter c of the Roman alphabet to express the Czech sound ts); or just inventing a new letter (as our medieval ancestors did when they created the new letters j, u, and w).

The Roman alphabet itself was the end product of a long sequence of blueprint copying. Alphabets apparently arose only once in human history: among speakers of Semitic languages, in the area from modern Syria to the Sinai, during the second millenniumB.C. All of the hundreds of historical and now existing alphabets were ultimately derived from that ancestral Semitic alphabet, in a few cases (such as the Irish ogham alphabet) by idea diffusion, but in most by actual copying and modification of letter forms.

That evolution of the alphabet can be traced back to Egyptian hieroglyphs, which included a complete set of 24 signs for the 24 Egyptian consonants. The Egyptians never took the logical (to us) next step of discarding all their logograms, determinatives, and signs for pairs and trios of consonants, and using just their consonantal alphabet. Starting around 1700 B.C., though, Semites familiar with Egyptian hieroglyphs did begin to experiment with that logical step.

Restricting signs to those for single consonants was only the first of three crucial innovations that distinguished alphabets from other writing systems. The second was to help users memorize the alphabet by placing the letters in a fixed sequence and giving them easy-to-remember names. Our English names are mostly meaningless monosyllables (“a,” “bee,” “cee,” “dee,” and so on). But the Semitic names did possess meaning in Semitic languages: they were the words for familiar objects (’aleph = ox, beth = house, gimel = camel, daleth = door, and so on). These Semitic words were related “acrophonically” to the Semitic consonants to which they refer: that is, the first letter of the word for the object was also the letter named for the object (’a, b, g, d, and so on). In addition, the earliest forms of the Semitic letters appear in many cases to have been pictures of those same objects. All these features made the forms, names, and sequence of Semitic alphabet letters easy to remember. Many modern alphabets, including ours, retain with minor modifications that original sequence (and, in the case of Greek, even the letters’ original names: alpha, beta, gamma, delta, and so on) over 3,000 years later. One minor modification that readers will already have noticed is that the Semitic and Greek g became the Roman and English c, while the Romans invented a new g in its present position.

The third and last innovation leading to modern alphabets was to provide for vowels. Already in the early days of the Semitic alphabet, experiments began with methods for writing vowels by adding small extra letters to indicate selected vowels, or else by dots, lines, or hooks sprinkled over the consonantal letters. In the eighth century B.C. the Greeks became the first people to indicate all vowels systematically by the same types of letters used for consonants. Greeks derived the forms of their vowel letters (ed. Image missing) by “co-opting” five letters used in the Phoenician alphabet for consonantal sounds lacking in Greek.

From those earliest Semitic alphabets, one line of blueprint copying and evolutionary modification led via early Arabian alphabets to the modern Ethiopian alphabet. A far more important line evolved by way of the Aramaic alphabet, used for official documents of the Persian Empire, into the modern Arabic, Hebrew, Indian, and Southeast Asian alphabets. But the line most familiar to European and American readers is the one that led via the Phoenicians to the Greeks by the early eighth century B.C., thence to the Etruscans in the same century, and in the next century to the Romans, whose alphabet with slight modifications is the one used to print this book. Thanks to their potential advantage of combining precision with simplicity, alphabets have now been adopted in most areas of the modern world.

A striking example from the history of writing is the origin of the syllabary devised in Arkansas around 1820 by a Cherokee Indian named Sequoyah, for writing the Cherokee language. Sequoyah observed that white people made marks on paper, and that they derived great advantage by using those marks to record and repeat lengthy speeches. However, the detailed operations of those marks remained a mystery to him, since (like most Cherokees before 1820) Sequoyah was illiterate and could neither speak nor read English. Because he was a blacksmith, Sequoyah began by devising an accounting system to help him keep track of his customers’ debts. He drew a picture of each customer; then he drew circles and lines of various sizes to represent the amount of money owed.

Around 1810, Sequoyah decided to go on to design a system for writing the Cherokee language. He again began by drawing pictures, but gave them up as too complicated and too artistically demanding. He next started to invent separate signs for each word, and again became dissatisfied when he had coined thousands of signs and still needed more.

Finally, Sequoyah realized that words were made up of modest numbers of different sound bites that recurred in many different words—what we would call syllables. He initially devised 200 syllabic signs and gradually reduced them to 85, most of them for combinations of one consonant and one vowel.

As one source of the signs themselves, Sequoyah practiced copying the letters from an English spelling book given to him by a schoolteacher. About two dozen of his Cherokee syllabic signs were taken directly from those letters, though of course with completely changed meanings, since Sequoyah did not know the English meanings. For example, he chose the shapes D, R, b, h to represent the Cherokee syllables a, e, si, and ni, respectively, while the shape of the numeral 4 was borrowed for the syllable se. He coined other signs by modifying English letters, such as designing the signs , , and  to represent the syllables yu, sa, and na, respectively.

Still other signs were entirely of his creation, such as , , and  for ho, li, and nu, respectively. Sequoyah’s syllabary is widely admired by professional linguists for its good fit to Cherokee sounds, and for the ease with which it can be learned. Within a short time, the Cherokees achieved almost 100 percent literacy in the syllabary, bought a printing press, had Sequoyah’s signs cast as type, and began printing books and newspapers.

Cherokee writing remains one of the best-attested examples of a script that arose through idea diffusion. We know that Sequoyah received paper and other writing materials, the idea of a writing system, the idea of using separate marks, and the forms of several dozen marks. Since, however, he could neither read nor write English, he acquired no details or even principles from the existing scripts around him. Surrounded by alphabets he could not understand, he instead independently reinvented a syllabary, unaware that the Minoans of Crete had already invented another syllabary 3,500 years previously.

We can confidently attribute the han’gul and ogham alphabets to idea diffusion rather than to independent invention in isolation, because we know that both societies were in close contact with societies possessing writing and because it is clear which foreign scripts furnished the inspiration. In contrast, we can confidently attribute Sumerian cuneiform and the earliest Mesoamerican writing to independent invention, because at the times of their first appearances there existed no other script in their respective hemispheres that could have inspired them. Still debatable are the origins of writing on Easter Island, in China, and in Egypt.

The Polynesians living on Easter Island, in the Pacific Ocean, had a unique script of which the earliest preserved examples date back only to about A.D. 1851, long after Europeans reached Easter in 1722. Perhaps writing arose independently on Easter before the arrival of Europeans, although no examples have survived. But the most straightforward interpretation is to take the facts at face value, and to assume that Easter Islanders were stimulated to devise a script after seeing the written proclamation of annexation that a Spanish expedition handed to them in the year 1770.

As for Chinese writing, first attested around 1300 B.C. but with possible earlier precursors, it too has unique local signs and some unique principles, and most scholars assume that it evolved independently. Writing had developed before 3000 B.C. in Sumer, 4,000 miles west of early Chinese urban centers, and appeared by 2200 B.C. in the Indus Valley, 2,600 miles west, but no early writing systems are known from the whole area between the Indus Valley and China. Thus, there is no evidence that the earliest Chinese scribes could have had knowledge of any other writing system to inspire them.

Egyptian hieroglyphics, the most famous of all ancient writing systems, are also usually assumed to be the product of independent invention, but the alternative interpretation of idea diffusion is more feasible than in the case of Chinese writing. Hieroglyphic writing appeared rather suddenly, in nearly full-blown form, around 3000 B.C. Egypt lay only 800 miles west of Sumer, with which Egypt had trade contacts. I find it suspicious that no evidence of a gradual development of hieroglyphs has come down to us, even though Egypt’s dry climate would have been favorable for preserving earlier experiments in writing, and though the similarly dry climate of Sumer has yielded abundant evidence of the development of Sumerian cuneiform for at least several centuries before 3000 B.C. Equally suspicious is the appearance of several other, apparently independently designed, writing systems in Iran, Crete, and Turkey (so-called proto-Elamite writing, Cretan pictographs, and Hieroglyphic Hittite, respectively), after the rise of Sumerian and Egyptian writing. Although each of those systems used distinctive sets of signs not borrowed from Egypt or Sumer, the peoples involved could hardly have been unaware of the writing of their neighboring trade partners.

It would be a remarkable coincidence if, after millions of years of human existence without writing, all those Mediterranean and Near Eastern societies had just happened to hit independently on the idea of writing within a few centuries of each other. Hence a possible interpretation seems to me idea diffusion, as in the case of Sequoyah’s syllabary. That is, Egyptians and other peoples may have learned from Sumerians about the idea of writing and possibly about some of the principles, and then devised other principles and all the specific forms of the letters for themselves.

Early scripts were incomplete, ambiguous, or complex, or all three. For example, the oldest Sumerian cuneiform writing could not render normal prose but was a mere telegraphic shorthand, whose vocabulary was restricted to names, numerals, units of measure, words for objects counted, and a few adjectives. That’s as if a modern American court clerk were forced to write “John 27 fat sheep,” because English writing lacked the necessary words and grammar to write “We order John to deliver the 27 fat sheep that he owes to the government.” Later Sumerian cuneiform did become capable of rendering prose, but it did so by the messy system that I’ve already described, with mixtures of logograms, phonetic signs, and unpronounced determinatives totaling hundreds of separate signs. Linear B, the writing of Mycenaean Greece, was at least simpler, being based on a syllabary of about 90 signs plus logograms. Offsetting that virtue, Linear B was quite ambiguous. It omitted any consonant at the end of a word, and it used the same sign for several related consonants (for instance, one sign for both l and r, another for p and b and ph, and still another for g and k and kh). We know how confusing we find it when native-born Japanese people speak English without distinguishing l and r: imagine the confusion if our alphabet did the same while similarly homogenizing the other consonants that I mentioned! It’s as if we were to spell the words “rap,” “lap,” “lab,” and “laugh” identically.

A related limitation is that few people ever learned to write these early scripts. Knowledge of writing was confined to professional scribes in the employ of the king or temple. For instance, there is no hint that Linear B was used or understood by any Mycenaean Greek beyond small cadres of palace bureaucrats. Since individual Linear B scribes can be distinguished by their handwriting on preserved documents, we can say that all preserved Linear B documents from the palaces of Knossos and Pylos are the work of a mere 75 and 40 scribes, respectively.

The uses of these telegraphic, clumsy, ambiguous early scripts were as restricted as the number of their users. Anyone hoping to discover how Sumerians of 3000 B.C. thought and felt is in for a disappointment. Instead, the first Sumerian texts are emotionless accounts of palace and temple bureaucrats. About 90 percent of the tablets in the earliest known Sumerian archives, from the city of Uruk, are clerical records of goods paid in, workers given rations, and agricultural products distributed. Only later, as Sumerians progressed beyond logograms to phonetic writing, did they begin to write prose narratives, such as propaganda and myths.

Mycenaean Greeks never even reached that propaganda-and-myths stage. One-third of all Linear B tablets from the palace of Knossos are accountants’ records of sheep and wool, while an inordinate proportion of writing at the palace of Pylos consists of records of flax. Linear B was inherently so ambiguous that it remained restricted to palace accounts, whose context and limited word choices made the interpretation clear. Not a trace of its use for literature has survived. The Iliad and Odyssey were composed and transmitted by nonliterate bards for nonliterate listeners, and not committed to writing until the development of the Greek alphabet hundreds of years later.

Similarly restricted uses characterize early Egyptian, Mesoamerican, and Chinese writing. Early Egyptian hieroglyphs recorded religious and state propaganda and bureaucratic accounts. Preserved Maya writing was similarly devoted to propaganda, births and accessions and victories of kings, and astronomical observations of priests. The oldest preserved Chinese writing of the late Shang Dynasty consists of religious divination about dynastic affairs, incised into so-called oracle bones. A sample Shang text: “The king, reading the meaning of the crack [in a bone cracked by heating], said: ‘If the child is born on a keng day, it will be extremely auspicious.’”

To us today, it is tempting to ask why societies with early writing systems accepted the ambiguities that restricted writing to a few functions and a few scribes. But even to pose that question is to illustrate the gap between ancient perspectives and our own expectations of mass literacy. The intended restricted uses of early writing provided a positive disincentive for devising less ambiguous writing systems. The kings and priests of ancient Sumer wanted writing to be used by professional scribes to record numbers of sheep owed in taxes, not by the masses to write poetry and hatch plots. As the anthropologist Claude Lévi-Strauss put it, ancient writing’s main function was “to facilitate the enslavement of other human beings.” Personal uses of writing by nonprofessionals came only much later, as writing systems grew simpler and more expressive.

For instance, with the fall of Mycenaean Greek civilization, around 1200 B.C., Linear B disappeared, and Greece returned to an age of preliteracy. When writing finally returned to Greece, in the eighth century B.C., the new Greek writing, its users, and its uses were very different. The writing was no longer an ambiguous syllabary mixed with logograms but an alphabet borrowed from the Phoenician consonantal alphabet and improved by the Greek invention of vowels. In place of lists of sheep, legible only to scribes and read only in palaces, Greek alphabetic writing from the moment of its appearance was a vehicle of poetry and humor, to be read in private homes. For instance, the first preserved example of Greek alphabetic writing, scratched onto an Athenian wine jug of about 740 B.C., is a line of poetry announcing a dancing contest: “Whoever of all dancers performs most nimbly will win this vase as a prize.” The next example is three lines of dactylic hexameter scratched onto a drinking cup: “I am Nestor’s delicious drinking cup. Whoever drinks from this cup swiftly will the desire of fair-crowned Aphrodite seize him.” The earliest preserved examples of the Etruscan and Roman alphabets are also inscriptions on drinking cups and wine containers. Only later did the alphabet’s easily learned vehicle of private communication become co-opted for public or bureaucratic purposes. Thus, the developmental sequence of uses for alphabetic writing was the reverse of that for the earlier systems of logograms and syllabaries.

Thus, food production and thousands of years of societal evolution following its adoption were as essential for the evolution of writing as for the evolution of microbes causing human epidemic diseases. Writing arose independently only in the Fertile Crescent, Mexico, and probably China precisely because those were the first areas where food production emerged in their respective hemispheres. Once writing had been invented by those few societies, it then spread, by trade and conquest and religion, to other societies with similar economies and political organizations.

While food production was thus a necessary condition for the evolution or early adoption of writing, it was not a sufficient condition. At the beginning of this chapter, I mentioned the failure of some food-producing societies with complex political organization to develop or adopt writing before modern times. Those cases, initially so puzzling to us moderns accustomed to viewing writing as indispensable to a complex society, included one of the world’s largest empires as of A.D. 1520, the Inca Empire of South America. They also included Tonga’s maritime proto-empire, the Hawaiian state emerging in the late 18th century, all of the states and chiefdoms of subequatorial Africa and sub-Saharan West Africa before the arrival of Islam, and the largest native North American societies, those of the Mississippi Valley and its tributaries. Why did all those societies fail to acquire writing, despite their sharing prerequisites with societies that did do so?

Here we have to remind ourselves that the vast majority of societies with writing acquired it by borrowing it from neighbors or by being inspired by them to develop it, rather than by independently inventing it themselves. The societies without writing that I just mentioned are ones that got a later start on food production than did Sumer, Mexico, and China. (The only uncertainty in this statement concerns the relative dates for the onset of food production in Mexico and in the Andes, the eventual Inca realm.) Given enough time, the societies lacking writing might also have eventually developed it on their own. Had they been located nearer to Sumer, Mexico, and China, they might instead have acquired writing or the idea of writing from those centers, just as did India, the Maya, and most other societies with writing. But they were too far from the first centers of writing to have acquired it before modern times.

The importance of isolation is most obvious for Hawaii and Tonga, both of which were separated by at least 4,000 miles of ocean from the nearest societies with writing. The other societies illustrate the important point that distance as the crow flies is not an appropriate measure of isolation for humans. The Andes, West Africa’s kingdoms, and the mouth of the Mississippi River lay only about 1,200, 1,500, and 700 miles, respectively, from societies with writing in Mexico, North Africa, and Mexico, respectively. These distances are considerably less than the distances the alphabet had to travel from its homeland on the eastern shores of the Mediterranean to reach Ireland, Ethiopia, and Southeast Asia within 2,000 years of its invention. But humans are slowed by ecological and water barriers that crows can fly over. The states of North Africa (with writing) and West Africa (without writing) were separated from each other by Saharan desert unsuitable for agriculture and cities. The deserts of northern Mexico similarly separated the urban centers of southern Mexico from the chiefdoms of the Mississippi Valley. Communication between southern Mexico and the Andes required either a sea voyage or else a long chain of overland contacts via the narrow, forested, never urbanized Isthmus of Darien. Hence the Andes, West Africa, and the Mississippi Valley were effectively rather isolated from societies with writing.

That’s not to say that those societies without writing were totally isolated. West Africa eventually did receive Fertile Crescent domestic animals across the Sahara, and later accepted Islamic influence, including Arabic writing. Corn diffused from Mexico to the Andes and, more slowly, from Mexico to the Mississippi Valley. But we already saw in Chapter 10 that the north-south axes and ecological barriers within Africa and the Americas retarded the diffusion of crops and domestic animals. The history of writing illustrates strikingly the similar ways in which geography and ecology influenced the spread of human inventions.