Despite huge interest in its decipherment, the writing of the Harappans still cannot be read, for a number of reasons: the absence of bilingual inscriptions to provide a starting point, the stylized form of the signs, the very limited length and nature of the texts, ignorance of the language that the script was being used to record, and the fact that the script died out instead of giving rise to later scripts. In addition, the number of signs indicates that the script was probably logosyllabic; so the number of components to be deciphered and the complexities of their use and interrelations are much greater than they would be with a syllabic or alphabetic script.
The Development of the Indus Script
Scratched lines or other graffiti are known from earlier pottery in the greater Indus region, and graffiti continued to be used throughout the Mature Harappan period and into later times. There is no significant relationship between these and the signs of the Indus script, despite a few superficial resemblances; both include simple geometric marks that are common to all cultures and recur without any connection to the uses to which they are put. However, signs resembling distinctive Indus signs began to appear on pottery at Harappa during the late fourth millennium. In the following Early Harappan period, short sequences of two or three such signs began to be used there, often using the same sign order as that in the Indus script.
During the initial period of development, between 3200 and 2800 BCE, there was an Elamite presence at Shahr-i Sohta in neighboring Helmand, a town with which the cultures of the borderlands were in trading contact. The Elamites used a written accounting system with complex numerical notation and a script, Proto-Elamite, used mainly to write personal names and lists of commodities. It has not been deciphered, but the numbers and a few of the signs can be read, and there is considerable understanding of the structure of this script.
It is likely that the inhabitants of the Indus region and the borderlands were aware of the Proto-Elamite script and its uses; this knowledge may have encouraged them to develop their own script when the circumstances arose that made writing useful to them, even though they put it to different uses; similarly, although the Harappans developed completely different signs to those of Proto-Elamite, they may have drawn on the latter as a general model, borrowing some of its features. This is suggested by the fact that, of all the possible combinations of direction of writing and the use of space on their written media, the Harappans chose the same one as the Elamites: in lines from right to
The baffling though beautiful Indus script cannot be deciphered until it can be determined what language it was written in. This is as uncertain now as it was when the civilization was discovered more than eighty years ago. (Harappa. com)
Left and from top to bottom. (For comparison, at much the same time, the Sumerians were writing in boxes arranged in columns.) The result of this influence was probably to kick-start the development of the Harappan script, which grew into a complete writing system during the Transition period. In contrast, the Proto-Elamite script disappeared.
Lack of Successors. When urban life declined in the early centuries of the second millennium, the script went out of use. A few Harappan signs occurred among the graffiti of the Posturban period, but these were rare and, when several were used, their order was random. Some signs resembling those of the Indus script continued to appear occasionally in later contexts; it is probable that these signs, while no longer encrypting speech, held a symbolic meaning for those who drew them, for example as auspicious or religious symbols or royal insignia.
Nothing links the Indus script with the Brahmi script that came into use more than a thousand years later. Brahmi derives from a north Semitic script first used somewhere in West Asia, one of the descendants of the consonantal alphabet invented in the Levant during the second millennium BCE. Another Indian script, Kharoshti, used for a while in the northwest, had a similar ancestry. The letters of the parent alphabet were adapted to form the basic consonantal signs of Brahmi; to these, diacritics were added to mark their combination with the different vowels. The direction of writing was from left to right, in contrast to that of the Harappan script. In the early first millennium CE, a modified version was created in south India to write Tamil. From Brahmi are descended more than two hundred modern Indian scripts, whose ancestry can be clearly traced.
Written Media
The majority (more than 80 percent) of Harappan inscribed objects, as well as the greatest diversity of them, come from the cities of Mohenjo-daro and Harappa. The rest, mainly seals and sealings, have been found widely distributed in Harappan towns and cities and in a few foreign locations.
Seals and Sealings. Most surviving Harappan inscribed materials are square stamp seals that bear an image and a short text. These are usually of steatite, though a few are made in other stones; two from Mohenjo-daro are of silver, presumably deluxe versions. The majority of images are of animals, especially the unicorn, but also bulls, elephants, and a number of other, mostly wild, animals; a few show scenes or composite animals. No link has been found between the images and the inscriptions on the seals. The seal texts are generally short, averaging four to five signs, and use a large repertoire of signs. The inscription was written in reverse so that it could be read correctly when impressed on sealings or elsewhere, for example occasionally on pottery. Clay sealings, of which relatively few have survived, might bear the impression of more than one seal or the same one several times. A small number of larger seals are known, all of an unusually high standard of workmanship.
Other types of seals occur in small numbers. These include a cylinder seal from Kalibangan and a few seals with inscriptions on both faces and no boss. A seal from Gola Dhoro took the form of a slim box with an inscribed face and an opening at one end, and several others are known from Mohenjo-daro. A few round Persian Gulf seals made by the people of Dilmun (see Chapter 6) also bear Harappan inscriptions. There were also long rectangular seals bearing only an inscription.
Miniature Tablets. From the upper levels at Harappa come many miniature tablets. These bear a short inscription on one face, written to be read directly from the tablet, and often a number on the other, sometimes accompanied by one or two other signs. The incised steatite examples often repeated the same inscription, while molds were used to produce multiple copies of the same design in faience or terra-cotta. Some bear a design as well as an inscription. These are often narrative scenes such as a deity watching a contest between two people; others had a single image, such as a tree inside a railing, or one or more auspicious symbols.
There are also inscribed cylinders from the later levels at Harappa; one is also known from Mohenjo-daro. These were made in two-piece molds. One side generally bears an inscription and the other a scene or an image with a religious theme.
The majority of these tablets come from Harappa, where they were probably all manufactured. Many have also been found at Mohenjo-daro, and they are also known from Dholavira, Lothal, Kalibangan, Chanhu-daro, and Ropar. Kenoyer (1998) suggests they may have been accounting tokens; a collection of thirteen tablets, nine with one design, four with another, was found together with weights, seals, and pendants in one location at Harappa. Others consider it more likely that they were votive offerings or amulets, perhaps bearing an auspicious message or prayer. On some tablets a jar sign forms part of the inscription or accompanies the number, and occasionally they also bear an image of an individual making an offering to a tree or seated deity; Parpola (1994) convincingly interprets these as records of offerings.
Perhaps related to these were a few triangular prisms, found at Harappa and Mohenjo-daro in the Indus realms and in Bahrein; a rather dissimilar example is also known from Maysar in Oman. These were of faience or terracotta and may have been made in a mold. One face bears a line of script, and the other two bear unusual images such as a houseboat or a procession in which individuals bear sacred images on poles, as well as more familiar themes such as the horned figure seated on a low throne.
At Mohenjo-daro, but at no other site, more than one hundred small engraved copper tablets have been found. These bear an inscription on one face and a design on the other. Generally a particular design is paired with a particular inscription, and the repertoire of inscriptions and pictures is quite restricted. Sometimes the image is replaced by another short inscription. The designs include composite animals, the endless knot motif, and a wild man with the hindquarters of a bull, armed with a bow and wearing a horned head-dress—possibly a shaman or deity. These may have been amulets. A few quite different cast copper tablets with a raised design were found at both Harappa and Mohenjo-daro.
Inscribed Objects. Inscriptions were engraved or stamped on a number of other objects. These included ivory and bone rods that may have had some ritual significance and the stoneware bangles that may have been worn in a political or religious context. Inscriptions were impressed on many pots before firing or scratched on afterward, perhaps giving information on the designated contents, the capacity of the vessel, or its ownership. Large storage jars had an inscription in relief formed by incising the inscription into the mold used to shape their base section.
A number of small objects bear an inscription: bronze axes and chisels, gold jewelry from Mohenjo-daro, bone hairpins, painted terra-cotta bangles, and others. While in some cases these inscriptions may have denoted ownership, they are more likely in most cases to have been prophylactic charms to ensure the object's safe use or to protect the owner. Some objects, judged on their quality, seem more likely to have been votive than functional; on these the inscription may name the person or group dedicating the object or the deity to which it was being offered.
The Signboard. One other inscription is known that is at present unique. During their 1989-1990 season, the excavators at Dholavira uncovered a line of nine huge signs in the Indus script, made of a white crystalline material, lying in a side chamber of the North Gate to the citadel. The signs were about 37 centimeters high, and their widths ranged from 25 to 27 centimeters. They had probably originally been set on a board, probably of wood, displayed over the gate. Its location at the boundary between the citadel and the outside world suggests that the text probably had a religious significance, perhaps purifying or blessing those who passed under it. It is possible that comparable inscriptions existed in other cities and towns; given the fragility of the Dholavira lettering, the extant traces of this signboard clearly survived only by extreme good fortune.
Lost Media? The surviving inscribed objects served only some of the potential functions of writing, probably including personal or official identification (seals as badges), the marking of ownership (possibly names on artifacts), the prevention of tampering (sealings), the management of distribution (seals and sealings), and dedications, spells, prayers, or benedictions (miniature tablets, copper tablets, inscribed artifacts). Most literate societies also used their writing to record economic transactions and to glorify the deeds of their leaders. The Harappans seem to have had no interest in such propaganda, but it seems unlikely that the Harappan polity could have functioned as efficiently as it obviously did without using some kind of recording system. Viewed from another perspective, the literate Mesopotamians obsessively recorded everything and it is improbable that the Harappans, being intimately involved through trade with them, could have resisted the urge to use their own script in ways beyond the uses revealed by the surviving media. It therefore seems probable that the Harappans also used other media of which no trace remains.
In more recent times many perishable media have been extensively used in the subcontinent for writing: wood, cotton cloth, bark, palm leaves, and parchment, all of which, if used by the Harappans, would long since have decayed away. The absence from the Harappan repertoire of the clay tablets that were so much used in the Near East may well reflect the ready availability of these perishable media, which were more portable, easier to store, and less fragile than unfired clay. In Egypt, where a comparable writing medium was made from papyrus plants, it is only because the arid climate was favorable to its preservation that such texts have survived. Since the Harappans did not make monuments, there could be no monumental inscriptions in their realms.
In modern South Asia, children in school use wooden boards to write their lessons. These are covered in a white substance and can be wiped clean after use, ready for the next day's work. Two terra-cotta objects from Mohenjo-daro have exactly the same shape as these wooden writing boards, a rectangular surface with a short handle at one end; they may be models of wooden originals or may even have been used as writing boards themselves, covered with an appropriate writing surface. Two signs in the Indus script may represent such writing boards; these are composed of a square with a line rising from its top; one is blank and the other bears horizontal lines that may represent writing.
Arguing against the Harappan use of any lost media, Farmer and his collaborators (2004) point to the absence of any writing equipment such as ink and styli, though one may counter this by pointing to the absence also of the paints
The inscribed materials from a single house in Mound E at Harappa illustrate some of the range of written media: these include a seal with a unicorn design (top left), with a broken seal beneath it; a terra-cotta sealing (top right) bearing two impressions of the same seal; beside it, two incised steatite tablets: and below, four faience tablets bearing short inscriptions, which were mass-produced in molds. (Harappa Archaeological Research Project, Courtesy Department of Archaeology and Museums, Government of Pakistan)
And brushes that were certainly used to decorate pottery and that could also have been used to write.
An Economic Argument. The Harappan state had an efficient internal distribution network and organized flourishing overseas trade, supplying goods from far and wide to people in every corner of their realms. The sealings that occasionally survive through being accidentally fired bear witness to the management of this distribution; their backs show they were attached to goods in transit packed in cloth bales, sacks, or jars, probably both to prevent unauthorized opening and for identification. The seals used to impress these sealings bore a design and an inscription, and they were carefully made objects for long-term use. The message they conveyed was therefore one that needed to be repeated unchanged, over a long period, as the wear on some examples shows. I argued in Chapter 8 that sealings both supplied general information, allowing packages to be correctly handled, and identified the sender. They would not have been used to provide the ephemeral information on the packages' contents, which were likely to vary in detail with each shipment even if the same types of goods were regularly dispatched, or the similarly variable instructions on what was to be done with the goods. This information might have been conveyed by word of mouth, but it is more likely that it was written down in an invoice that either accompanied the goods or was sent separately. Records of goods dispatched, issued, or received may have been less essential, but comparison with other early states suggests that these may also have been kept. The absence of any evidence for invoices or records seems more likely to reflect the use of perishable media than a Harappan failure to use such records at all.
An almost contemporary example may put this in perspective. A fantastically detailed picture of such a system in operation survives from the Near East, where the torching of the town of Kanesh in Anatolia around 1830 BCE caused the baking and therefore the preservation of the clay tablets recording a trading network operated by Assyrian merchants from Assur. These tablets, sent from Assur, include much information about private life and the circumstances in which the merchants were operating, but the vast majority of their contents are details about consignments: lists of what was being sent and by whom, the names of the carriers, and instructions about disposal on arrival in Kanesh, along with requests for particular goods to be sent back and minutiae on the payment of various local taxes and customs dues. The sender's seal was often impressed on the invoice or on its clay envelope, as well as on the sealings on the goods. The invoices were carefully stored in the trading quarter at Kanesh but were taken back to the home city of Assur at the end of a merchant's stay in Anatolia. Almost nothing related to this trade survives from Assur itself, and there would have been no evidence at all of this trade (which dealt in textiles, tin, and silver) had the Assyrians not used clay as their writing medium.
The Importance of Context
A great deal of valuable information can be extracted from the analysis of the various contexts of the Harappan inscriptions: These include the type of object on which they were inscribed, the placement of the inscription, the associated images, and the objects' place of discovery and distribution.
There are various types of inscribed material. Because the function of each type dictated the content of its inscriptions, it is essential to study the inscriptions associated with each as a separate group. For example, personal seals usually include some of the following information: the owner's name, patronymic, home town or place of origin, and title, as well as a dedication to a personal god and the name of the authority to whom the owner is responsible. Sign sequences that are frequently repeated or frequently substituted for each other on seals might therefore represent an official title or a personal name. Many texts in other cultures were bureaucratic records; the nature of the surviving Harappan inscribed media and the form of their texts preclude their use for this purpose. One possible exception is the small series of tablets that have a text on the front and a number on the back; these may have been used as sanctified receipts for offerings made to the gods or, more prosaically, as ordinary receipts for goods or taxes or as ration vouchers issued for work undertaken.
In some cases the placement of an inscription may be informative. For example, on some inscribed axes the text is placed where the haft would originally have covered it: in such cases it is unlikely that the inscription named the owner and more likely that it was a prophylactic charm, whose efficacy did not depend on its visibility.
Many inscribed objects also bear one or more images. Often, particularly on seals, there seems to have been no link between the inscription and the animal or scene depicted. It is, however, worth looking for regularities within groups; for example, the texts bearing the zebu motif may have been owned by rulers and might therefore include royal titles. In other cases, however, particularly the copper tablets, there is a close relationship between particular images and certain sign sequences; here the inscription may relate in some way to the content of the image.
Where an inscribed object was found may also be significant, and both the exact find spot and the general location are potentially revealing. Certain types of inscribed objects have a limited distribution, such as the stoneware bangles, which are known almost exclusively from the great cities of Mohenjo-daro and Harappa; they may therefore be badges of high office. Parpola (1994) noted that the sign ||/ was found only at Chanhu-daro, where it occurred on eleven objects (around one-sixth of all Chanhu-daro's inscribed objects); he therefore suggested that the sign may represent the town's name. Exact findspots may provide little information if the inscribed object was accidentally lost. But in other cases, such as the burned warehouse at Lothal, there may be a great deal of useful information encapsulated in the associations between an inscribed object, other artifacts, and the structure in which it was found.
While contextual clues may not enable us actually to read the inscriptions, they may allow us to identify certain sequences as probably belonging to particular categories, such as names of individuals or deities or dedicatory formulas. These deductions should then feed into overall decipherment by pinning down the general meaning of certain signs or phrases that, in turn, can be used to check the credibility of suggested readings of these signs or sign sequences, perhaps also discouraging the adoption of absurd interpretations.
The Harappan Sign Repertoire
The Indus script uses a large number of signs. Some are unambiguously pictorial, such as the (headless) stick figure and the palm squirrel running headfirst down a branch. Many seem abstract, and many others appear to be pictorial but it is not clear what they show; for example, a sign that looks like a jar with handles might, alternatively, represent a cow's head. A few signs appear very frequently, with the most common twenty signs accounting for around 50 percent of the content of the inscriptions over the whole corpus. Some are known from only one occurrence (and are known as singletons), while many others occur only a few times.
An essential first step in studying the Indus writing system is to identify and count the repertoire of signs. This is far from straightforward.
Allographs. Many signs look similar. Some are likely to be allographs, that is, different accepted ways of writing the same sign. Others, however, must be different signs, with different semantic or phonetic values. The problem is to distinguish between these possibilities. If two similar signs behave in the same way, for example appearing interchangeably in sign sequences and forming ligatures with the same signs, they are likely to be allographs. In other cases, allographs of a sign form a cline, a series displaying a gradual shift in form. Even though the signs at either end of the cline look significantly different, each sign is close in form to its neighbor. Nevertheless, there are many ambiguities.
Ligatures. A related problem is that of ligatures. Ligatures are signs combined in various ways: by joining them with a line, by writing one inside another or one above another, or by superimposing parts of one sign on another (for example, the stick figure sign with the handled jar sign replacing the neck). Some ligatures are composed of more than two signs, sometimes as many as six. A ligature may also be formed of a sign and another element that does not appear as an independent sign. Establishing the boundaries of signs requires analysis of the sign sequences throughout the corpus, for example, to determine whether two parallel lines with a different sign between them represent three separate signs or a single ligature. Ligatures may be a space-saving device, allowing a complete word or phrase to be written as one, in which case they should be broken down into their separate components. Alternatively, the ligatured sign may express a meaning obliquely related to the sum of its parts or differing entirely from all of them, in which case such dissection is not helpful.
Sign Lists. A number of attempts have been made to establish a complete sign list, beginning with G. R. Hunter in 1934. Iravatham Mahadevan compiled one as part of his concordance published in 1977, and Asko Parpola included another in his 1994 volume. Most recently, a new sign list has been compiled by Bryan Wells (n. d.). These differ considerably in the number of signs they arrive at: Mahadevan distinguishes 419 signs, Parpola 386 plus twelve possibles, and
Wells 676. These numbers are the results of long considered decisions on the status of individual signs and potential allographs and ligatures. The sign lists form the basic tool when analyzing the texts to detect structure and eventually meaning, and many of the analyses use statistics; both lumping together signs that are different and separating out signs that are the same obviously affect the outcome of such analyses. It is an extremely difficult exercise, and new data necessitate constant revisions.
Corpora and Concordances. Detailed studies of the script depend on the availability of reliable published copies of all the inscribed objects (currently numbering around three thousand seven hundred) and their background information. Mahadevan (1977) compiled a concordance in which he listed all the texts then available, using a font generated to match his sign list. He accompanied this with background information on the settlement where each text was found and the design (or field symbol) on each seal, along with some analyses of the frequency and distribution of individual signs, within the texts and with respect to site and field symbol. Parpola and his collaborators produced several such concordances (Koskenniemi, Parpola, and Parpola 1973; Koskenniemi and Parpola 1979, 1982) but, dissatisfied with the quality and completeness of the data with which they were working, Parpola embarked on the mammoth task of producing a complete photographic record of all known inscribed Harappan material. Two volumes have now appeared (Joshi and Parpola 1987; Shah and Parpola 1991), and a third is expected soon. Wells has produced his own electronic corpus of the inscribed objects, along with much more detailed information on the find context of each; this is not yet publicly available.
All these corpora suffer from a major dilemma: The raw data from the inscribed objects have to be recorded in a manageable form before they can be analyzed, but this inevitably involves some loss of precision due to the decisions on sign identity taken by each scholar in compiling the sign list.
How Scripts Developed
A script is essentially a code whose meaning is accepted by those who use it, communicating a message between writer and reader. Most early scripts began with pictorial signs, but even for the most obvious pictograph the meaning represented was that agreed on by its users, which did not necessarily match its pictorial meaning. A pictograph might cover a range of related meanings (for example, a sign might mean both "legs" and "to walk"). Abstract signs might also be part of the initial sign repertoire (for example, the Sumerians used a circle quartered by lines to signify "sheep"), but pictographs predominated as new signs were added to the repertoire. Some new signs were made by modifying existing ones. When a new sign was introduced, it would be current among a limited group of people; some became more widely accepted and entered the standard repertoire, while others remained restricted or died out. Some signs also shifted their primary meaning through time. From an early stage, widely accepted conventions began to govern the use of signs, and a few signs were in frequent use over a wide area.
Initially the signs represented objects, but they soon came also to stand for sounds. The range of words that could be represented was now extended by punning (the rebus principle): using a pictograph to represent a homonym, a word that sounded the same (or almost the same) but had a different meaning (for example, "bee" for "be"). Further meanings could be achieved by ligatures (for example, "bee" plus "leaf" making "belief") or by modifying a sign (for instance, changing a jar sign in various ways to signify a jar of honey, beer, or milk).
Early scripts were devised to serve a restricted purpose or purposes, and it was adequate to use signs that just represented word stems or roots, with the appropriate form of the word being understood from the context. As the uses of the script diversified, the need to supply this information developed, and signs came increasingly to be used phonetically, often as single syllables, though logographic use also continued. Individual signs might by now have many possible semantic or phonetic values, so some scripts added extra, silent signs to aid identification of the value being used: semantic determinatives to indicate which of several alternative meanings was intended, and phonetic complements that reproduced part of the sound of the appropriate word. These developments took place in the Near East (including Egypt) from the late fourth through the third millennia, and later in China and Mesoamerica. In Mesopotamia, many modifications of the script, which had been developed to write the Sumerian language, took place as a result of the difficulties that arose when using it to write the quite differently structured Akkadian language.
At the same time the signs in the Sumerian script were simplified to make them easier to write, and the pictorial aspect was lost. The Egyptians maintained their pictorial hieroglyphic script for monumental inscriptions but developed a simplified hieratic (cursive) form for writing on papyrus.
While some scripts continued to be logosyllabic, by the late third millennium several scripts in the Near East used mainly syllabic signs. These allowed the sounds of speech to be reproduced more economically and efficiently. A further development took place during the second millennium BCE when Semitic speakers took the initial consonant of their own words for the pictorial meaning of a number of Egyptian hieroglyphs and assigned this consonantal value to a modified form of each sign, creating the first alphabet. From this developed all other alphabetic scripts of the world, including those of South Asia. (A roughly contemporary cuneiform alphabet, also devised in the Levant, was short-lived.) The Greeks developed a version more suited to their language in which vowels were also marked. In principle alphabetic scripts have a single sign for each phoneme (vowel or consonant), making it possible to write down any word so that its pronunciation is unambiguous; alphabetic scripts can therefore be adapted to render any language.
Analyzing the Indus Script
Understanding the Indus writing system depends on finding out as much about it as possible. Only then can meaningful attempts be made to decipher the script. Much has been revealed by detailed analyses of the text corpus, looking at features such as sign associations, frequency, and position.
Characterizing the Script. Alphabetic scripts can be recognized by the small number of different signs that they require, generally somewhere between twenty and thirty-five, and by the number of signs needed to make a word, which in English, for example, ranges from one to more than thirty. Syllabic scripts require a considerably larger number of different signs, generally in the range of a hundred to a hundred and fifty, but far fewer are needed to write a single word. Each sign represents a vowel (V) on its own or in combination with one or two consonants (C): V, CV, or CVC.
Logosyllabic scripts have a far larger sign repertoire, generally four hundred to seven hundred signs, though far fewer may be in regular use. The number of signs depends partly on the relative proportion of logographic and phonetic signs and on whether the latter are polysyllabic or monosyllabic. Words are typically around one to three signs long in logographic scripts.
The number of signs used in the Harappan script should reveal its nature. As well as being historically implausible, its identification as alphabetic or syllabic is ruled out by the number of its signs. While an exact number is not agreed, it lies somewhere in the range three hundred and fifty to seven hundred, of which about half were basic signs. This fits the number of signs expected in a logosyllabic script.
Text Length. The majority of Harappan texts are very short, around one to eight signs; fewer than one in a hundred has ten or more signs and quite a few have only one. The longest continuous text is seventeen signs; the longest text (on the faces of a triangular prism) is twenty-six signs but is probably composed of three shorter inscriptions. In part, this reflects the nature of the texts, most of which needed to convey very little information. Those on the seals probably gave brief information identifying the owner. Other objects might have borne the owner's name, a protective invocation or charm, or a dedication. The texts repeated on the mass-produced tablets seem likely to have been stock inscriptions, such as records of standard offerings or formula benedictions.
While this may provide an adequate explanation for the inscriptions' brevity, the latter could also be due to the use of many logographs and relatively few syllabic signs. This is probably consistent with the large number of rarely used signs in the script. Another possibility is that the use of ligatured signs, relatively common in the Indus script, reflects a tendency to agglutination, so that one ligatured sign may be composed of a number of morphemes, encoding a phrase or sentence.
Detecting Patterns. Attempts to understand the script's workings generally involve a detailed analysis of the full corpus of texts to reveal how the individual signs behave. Such laborious studies have been undertaken by a Russian team under Yuri Knorozov, the Finnish team under Asko Parpola, the Tamil scholar
Iravatham Mahadevan, and Bryan Wells, a Harvard scholar. One approach was to generate statistical information about the association of signs in pairs, with respect to the frequency of each sign in the inscriptions. This identified regularly associated pairs of signs that may reflect syntactical features of the script.
Some scripts mark word divisions, but the Indus script does not. It is therefore necessary to find other ways of determining where the divisions between words or phrases fall. Some texts have only one sign, which must represent a complete word. Many have two or three signs that must constitute a complete message, of one word or several. These sequences recur in some longer texts, and the remainder of such texts may then also represent self-contained segments. Where particular signs occur in a number of texts, they often occupy the same position: for example, the "handled jar" sign is placed at the end of many texts. This suggests that, when it appears within a text, the signs following it form a separate segment. The analysis of objects with more than one line of text shows that the line breaks almost always fall at points identified elsewhere as segment boundaries.
Various analyses of position and association allowed the identification of signs that behave in the same way and that may therefore be interchangeable within certain segments. For example, the "arrow" sign seems to have been interchangeable with the "handled jar" in the segment terminal position. For a while it was thought that these two signs might represent the dative and genitive markers, respectively. This theory is now discounted, but recently Mahadevan (2000) has suggested that they may instead mark person-number-gender (nominal-singular-male/nonmale, the gender distinction used in Dravidian).
Using a number of such clues and with immense labor, some characteristics of Harappan texts have been exposed. One is that a self-contained text segment is typically one to three signs long, matching the typical word length in logosyllabic scripts, particular those in which grammatical affixes are omitted. Another is that there is a definite pattern to many of the longer texts, which are made up of distinct sequences, each of which has a predictable composition. The behavior of many individual signs and sign sequences has been established, giving clues to morphology and syntax.
Farmer and his associates (2004), however, argue that the known third-millennium scripts include so many polyvalent signs and omit so much syntactic, semantic, and phonetic data that statistical studies of these sorts cannot identify any such features; this is probably an unduly pessimistic viewpoint.
Direction of Writing. Various lines of evidence have satisfactorily demonstrated that the Indus script was written from right to left, in rows from the top downward. (This is the direction of the text on tablets, objects, and sealings. The seals were cut the other way round so that they could be read in impressions on sealings and elsewhere; seal texts are therefore reversed before being studied.) Inscriptions that do not fill the whole upper line on a seal always begin on the right. Ones that run over on to another line generally start again on the right side, though a few run around onto the next line going left to right
(backwards but continuous). Sometimes the writer started on the right using normal-sized signs but ran out of space toward the end and cramped the last few signs on the left or progressively reduced their size. In some instances, one sign overlapped its neighbor, the left-hand sign partially obliterating the previously written right-hand sign. One unusual inscription ran round three sides of a seal: The first line was complete; the seal was then rotated through 90 degrees clockwise, and the next line was written in the space remaining to the left of the first line; this was repeated, the third line finishing before the left edge of the seal.
Mahadevan (1977) used internal evidence to prove conclusively that the direction of writing was right to left. His studies had shown that certain signs regularly appear in a particular position in the inscriptions. Specifically, the "handled jar" sign is frequently at the left-hand end of the text, while certain others, such as the "chakra" (an oval divided by six lines), frequently appear at the right-hand end, and several regular combinations of signs are written to the right of the "handled jar," to the left of the "chakra," and together elsewhere.
By looking at the signs that were written on the second line when text was broken into two lines, Mahadevan was able to demonstrate that the "handled jar" comes at the end of texts and that the "chakra" and others from the right side come at the beginning.
Studying the Signs. The work undertaken by various scholars in compiling sign lists has established that there was a corpus of basic signs, numbering around one hundred and fifty to two hundred, and an equal or greater number of other signs, many of which are compounds of the basic signs.
Some of the signs are unambiguously pictorial. There can be no doubt, for example, that the stick figure is a person, though this figure may be male or genderless, human or divine. In a few cases, Harappan art confirms what appears to be the pictorial meaning of the sign: for example, the fish sign is identical to a fish shown in a gharial's jaws.
For many more signs, it is harder to identify their pictorial meaning: One scholar's crab is another's tongs; one sees a cobra where another sees a seated man; a vertical stroke with horizontal lines from it could be a comb or a harrow or even a tree. The art or script of other societies may helpfully suggest possible interpretations of signs: For example, a triangular sign with one extended arm bears a strong resemblance to a hoe used by the ancient Egyptians. These comparisons are useful but cannot have the compelling weight of images identified from Harappan art or later Indian symbolism that may be derived from Harappan iconography.
In some cases, suggested meanings can be ruled out on cultural grounds: For example, the oval or circle divided into six looks just like a spoked wheel, but this interpretation is impossible because the spoked wheel was invented centuries after the sign came into use and thousands of miles away; much more plausible is its interpretation as a chakra, a symbol that was part of later Indian iconography, representing royalty and divinity.
Many scholars have taken the pictorial signs as a starting point in their attempts at decipherment. Ideally these could yield words clustered around their pictorial meaning and one or several homonyms. It is always possible, however, that a sign was not chosen for its pictorial value, that it had shifted its logographic meaning through time, or that it had become a sign with a purely phonetic value. In addition, in logographic scripts there is often polyphony and polysemy (a number of different sounds and meanings represented by a single sign), so that even in scripts that have been deciphered there is often uncertainty about the selection of the appropriate value.
The problem of interpretation becomes even greater with compounds. These are often composed of two or more signs from the basic repertoire. Ideally, these should be words or phrases put together from their various parts, but experience of the way scripts behave shows that things need not be so straightforward. Wells has demonstrated a number of cases in which the compound behaves like neither or none of its components, suggesting that the compound is not the sum of its parts. He also identifies some cases in which two signs may be written either side by side or as a compound; these may be alternative ways of writing the same thing or may have entirely different meanings.
Other scripts show that there is a considerable range of possible interpretations of elements added to a basic sign, including grammatical or semantic affixes, indicating inflexion or agglutination, diacritics, phonetic complements, and semantic determinatives, not to mention the practice characteristic of evolving scripts of modifying existing signs to create new ones.
The decipherment of alphabetic or syllabic scripts, such as Linear B, is relatively straightforward, in that the range of possible values is limited and the discovery of some of the values assists in the interpretation of others, the results interlocking to reinforce each other. In contrast, logographic scripts are open-ended, with the discovery of some sign values providing little or no assistance in interpreting others. The recent decipherment of the Maya script, a logographic script of fiendish complexity, whose users delighted in employing a wide variety of different "spellings" (ways of expressing the same word), provides a beacon of hope that progress can be made in deciphering the Harappan script too.
Numerals. Numbers form an important subset of signs in scripts and have often proved the easiest type of sign to identify, using clues from their form and behavior. These, therefore, were looked for in the Indus script, and it seems likely that at least some of them have been identified.
Obvious candidates for numerals are three series of parallel strokes that vary in number: short strokes arranged in a single line; short strokes arranged in several parallel lines, each of up to five strokes; and long strokes. The probability of these representing numbers is supported by their consistent position in front of certain signs, coupled with the variation in their number in this position, apparently signifying 1x, 2x, 3x, 3y, 3z, and so on. This seems extremely likely in the case of the short strokes, which number up to ten; the long strokes, on the other hand, do not behave quite as consistently, are less common, and only go up to five, so it is possible that they were not in fact numerals.
Not all such signs are likely to be numerals. Positional data show that twelve short strokes arranged in three rows of four, as well as one or two short strokes written in a raised position, seem not to behave like the numerals. In addition, in a number of contexts, numeral signs appear to have had a different significance, such as being read as phonetic signs. Wells argues that several other signs may have functioned as numerals, on the basis of their positions in texts and their frequent association with other numeral signs. Parpola identifies a different sign (semicircle) as a possible larger numeral, perhaps ten.
The Language of the Texts
Decipherment requires knowledge of the language that the script was used to write. A thorough understanding of the script's structure can assist in narrowing the field of possible candidates for the still-unknown Harappan language.
Global linguistic studies indicate that sentences may be constructed in a number of ways. The principal feature relates to the order of the basic elements, subject (S), verb (V), and object (O), particularly the latter pair; these elements, however, cannot yet be identified in the Harappan texts. However, many other features of word order are also significant. One is the respective order of nouns and qualifying numbers. The analysis of sign sequences and text segmentation in the Harappan script showed that the numeral signs always preceded the signs with which they were regularly associated; the word order was therefore number followed by noun. This tends to correlate with languages using the OV rather than the VO order, as do a number of other features: for example, in OV languages, titles follow names. This patterning offers clues to what may be looked for in the structure of Harappan texts, though, naturally, it is necessary to find independent evidence to demonstrate that these features have been correctly identified.
Evidence from the segmentation of the texts may also suggest that the Harappan language was agglutinative, using suffixes (like the Dravidian languages, including Proto-Dravidian), rather than inflected (like Latin and French), but this cannot be confirmed in the present state of knowledge. If the Harappan language should prove to have been of this type, this would strengthen the case for its identification as Proto-Dravidian, one of the main possibilities.
The analyses of the Harappan seal texts show that they follow a consistent structural pattern, regardless of where they were found in the Harappan realms. This indicates that a single language was used in writing throughout the state. Further confirmation comes from a number of Harappan seals found overseas, in Mesopotamia and the Gulf; these, in contrast, often bear sign sequences that are differently structured, suggesting that they were being used to write names, and perhaps other information, in a different language.
Experimentation. Another line of approach is to look at the signs that can be interpreted pictorially and assign plausible sound values to them from a candidate language. The words for the objects depicted should include some (probably many) that have homonyms in the chosen language. A sequence of signs that can be interpreted in this way should combine to form meaningful words, phrases, or sentences in the correct language, and gibberish in any other. Unfortunately this is a very hit-and-miss approach, given the great latitude available in the interpretation of the pictorial value of signs, and some would-be decipherers have used this to accord themselves so much freedom that they can interpret the script in any way they choose. In the early stages of decipherment—unfortunately still the point at which the Indus script is stuck—many plausible interpretations can be offered and justified. Through time, however, the correct path should become apparent, as further sign identifications using the correct language reveal themselves to be consistent with previously identified signs, and identifications in the wrong language prove inconsistent. At present, a number of convincing Dravidian-based identifications have been made that seem to provide consistent results, but they are too few to offer proof. A major drawback is that, apart from Proto-Dravidian, so little of any of the plausible candidate languages is known.
The Alternative:Not a Script at All?
Recently Steve Farmer, in collaboration with others (Farmer, Sproat, and Witzel 2004; Farmer 2006), proposed that the Harappan signs were not in fact a script but only a series of religious symbols. He argued that there were no lost manuscripts made of perishable materials and that what survives therefore represents the full range of uses of these signs. Various features led him to this conclusion. For a start, he points to the failure of the Harappan signs to evolve with frequent scribal usage as, for example, the Sumerian script developed from pictorial and geometric signs into highly abstract signs composed of a few wedges. However, Parpola (2005) notes that there was, in fact, some scribal simplification of Harappan signs. One may counter Farmer's argument by making a comparison with the Egyptian script: As with the formal carved or painted Egyptian hieroglyphic texts, with their highly conservative sign forms, the surviving Harappan inscriptions were mainly on finely made formal media where the signs were executed to a high standard; a cursive script—comparable to Egyptian hieratic, written at speed, and far less carefully rendered—could have existed for use on perishable media.
Farmer also draws attention to the absence of long Harappan inscriptions on potsherds. If the Harappan signs were a script, he contends, this absence would make it unique among the scripts of literate cultures, who all used potsherds, often like scrap paper. This need only imply, however, that the Harappans had other media that were easier to scribble on, such as cotton cloth or wooden boards, or that the writing medium was not well suited for use on sherds. Likewise, the absence of long monumental inscriptions seems significant to Farmer, but the Harappans did not create monumental art or architecture on which such inscriptions might have been written; the nearest they came to this is the Dholavira signboard, possibly the tip of an iceberg of now vanished public inscriptions. Again, Farmer points to the absence of representations of scribes in Harappan art or script, but in no branch of Harappan art are there portraits of occupational specialists, apart from a tiny number of figurines engaged in daily tasks, such as grinding grain or driving carts; the narrative
One of Farmer’s main arguments against the Harappan signs representing a script is the absence of inscribed sherds (ostraca). This rare find from Harappa, however, contradicts this contention: it is a sherd which has been used as scrap on which to scratch a short message or temporary record. (Harappa Archaeological Research Project, Courtesy Department of Archaeology and Museums, Government of Pakistan)
Scenes on a few seals are the closest the Harappans came to specifically themed art, and these are all concerned with religious subject matter.
The frequent singletons in the Harappan script strike Farmer as significant. These seem to increase in number with every new discovery of Harappan texts. This should not be the case with a true script, he argues; new discoveries should decrease the number by containing additional examples of previously unique signs. He also considers that the proportion of singletons and rare signs is unusually high; other scholars, such as Parpola (2005), demonstrate that this is not so, since in general logosyllabic scripts contain a small corpus of frequently used signs and a large number of much less common ones. Moreover, new signs are continuously added, even when the writing system is a fully developed one, something Farmer also denies. Statistically the Harappan script does not differ significantly in its sign proportions from other logographic scripts. A further point regarding the singletons is that Wells (n. d.) has demonstrated that many are variants or ligatures of basic signs, rather than completely different signs; again, this is something to be expected in a genuine script.
Conversely, Farmer points out that there is a high repetition rate of individual signs within texts in most scripts, something that is not apparent in Harappan texts. This, he suggests, indicates that the Harappan signs were not phonetic. However, the large number of signs suggests that the script used a considerable number of logographic signs, implying a lower proportion of phonetic sounds than in some logographic scripts. Perhaps more significantly, the brevity of the majority of Harappan texts (four to five signs, on average) makes it less likely that signs would repeat within them than it is in the longer texts with which Farmer compares them.
Farmer's arguments fail to account convincingly for the structural regularities that analyses have revealed in the use of the Harappan signs; these seem strongly to support the hypothesis that the Harappan signs represent a writing system. The theory put forward by Farmer and his collaborators has not been widely accepted, but it has been valuable in compelling scholars to look afresh at their assumptions about the script and in provoking a stimulating debate from which a deeper understanding of the script should emerge.
World History
