Is There Evidence for Ethnicity in

Since the discovery of the first cuneiform texts, there has been controversy about the people who spoke the various languages written therein. In 1874, Joseph Halévy began his challenge against the then-prevalent view that Sumerian and Akkadian were two separate languages spoken by separate peoples (though at that time the languages were referred to as Akkadian and Assyrian, respectively). He argued that there was nothing in art, geographic names, or the literary sources known at the time that would point to two different peoples living in Babylonia. Rather, he suggested, the language was merely an “ideographic system invented by the Assyrians themselves in addition to phonetic writing.”[1] But by the turn of the last century, a great deal of new evidence had come to light, which effectively discredited Halévy and proved the existence of Sumerian as a distinct language.

However, the fact that Sumerian existed apart from Akkadian as a language does not mean that the two existed as separate peoples. While language is in general an important ethnic signifier, people who speak different languages do not necessarily identify with different ethnic groups. Indian emigrants to the United States, for instance, often identify themselves as a single cohesive ethnic group, at least in relation to non-Indians, despite of the vast number of mother tongues they may speak as their first languages. There is, of course, the fact that these emigrants speak English to each other, if a different Indian language is spoken at home. This is very significant when discussing their unified ethnic identity, because the English language may be a uniting factor for groups that would otherwise see themselves as separate. Unfortunately it is not within the scope of this paper to more fully address such linguistic issues, so let us assume for now that different languages do not necessarily correspond to different ethnic groups, and hence that the question remains as to whether or not Sumerians and Akkadians can be understood as ethnically distinct.

If ethnicity is to be further discussed, some time must be taken to establish a working definition of the term. This definition should include both how people define their own ethnicity and how they define that of others. As a social construct, ethnicity can be said to exist wherever a significant number of people believe it exists, whether they believe it is a quality that they themselves have or one only possessed by others. We must also incorporate the characteristics that separate ethnicity from other forms of social identity. Its constructed nature does not make ethnicity at all unique, but it does tend to be the only form of social identity that puts quite so much emphasis on kinship relations, real or imagined. Volumes could be written about the nuances of ethnic identity, but for the purposes of this paper, “ethnic group” will be defined as a group of people whose members perceive themselves as kindred, or whose members are perceived by others as kindred, and which exists in contrast to one or more other such groups. “Ethnicity”, then, is the quality of perceived kinship that sets one ethnic group apart from another.

Ethnic identity can involve an imaginary kinship connection, and often does, but such a connection cannot exist even in the imagination if people are unable to convince themselves that it could exist in reality. It is unlikely that two people will perceive any sort of kin relationship to each other if they have very different attitudes about family, because human beings, as social animals, transfer familial attitudes from one generation to the next. Significantly different attitudes therefore suggest different ancestry, regardless of how closely related two families are in reality. A sudden change in these attitudes, with respect either to location or to time, may correspond either to distinct ethnic groups living at the same time or to a sudden influx of members of one ethnic group to an area previously dominated by another. Of course, the fact that two groups of people (as categorized by attitudes about family) are unlikely to have a single ethnic identity does not necessarily mean that they each have their own ethnic identity. It is possible that ethnicity plays no important role in the identities of members of either group. Personal identity, like the linguistic issues mentioned above, will have to be left to others to address, as it is also outside the scope of a paper dealing with familial systems.

Family units and households frequently coincide, but the difference between family and household is very important. “Family” refers to people whose relationship to each other is largely determined by society itself, while “household” refers to people whose relationship is physically determined by the building in which they live. J. N. Postgate describes a family as “composed of members related by blood or marriage, with limits which are inevitably vague,” and a household as “a ‘co-resident domestic group’, whose limits at any time should be more or less precise.”[2] Their more precise limits make households much easier to analyze, both for archaeologists studying ancient societies and for the people administrating a society. To determine the characteristics of a particular household, one need only look at the situation within the walls of a particular house (whether its residents are still alive or have been dead for millennia). To determine the characteristics of a particular family, on the other hand, one needs to look at all the complex blood and marital relationships within a group of people.

In addition to family structure, a number of other factors contribute to the arrangement and functioning of a household. One of these factors is socioeconomic status, with poverty possibly forcing more people to live within the same set of walls, and wealth possibly allowing smaller groups of people to inhabit separate houses. Another contributing factor is overall population density. Over time, increased crowding in a city tends to cause larger numbers of people to cohabit smaller living spaces, regardless of their attitudes about the value of an extended family structure. These non-familial factors affecting households make it very difficult to analyze family structures by looking only at households. Hence it is impossible to use trends in household arrangement to definitively answer questions of ethnicity, which pertains to many familial characteristics in addition to basic structure. To address these questions, we need to search more closely for trends pertaining directly to families themselves, rather than to the buildings in which they live.

It is difficult to determine the characteristics of a family by studying the objects left behind by those who lived in ancient Mesopotamia, with the exception of texts. While a collection of pottery cannot tell us all that much about a family, material written about family groups can provide valuable insight into how they were arranged and how people thought about family. Unfortunately, the majority of Sumerian and Akkadian textual sources deal more with households than with families, most likely for the reasons mentioned above. On the other hand, the vast number of texts that have been found means that the comparatively small proportion dealing with family units still provides us with a fair sample of families to analyze.

One of the most basic characteristics of a family unit is its size and extent. Many different levels of kinship grouping exist, with the two most fundamental being the nuclear and extended families. The nuclear family unit (in a patrilineal society) consists of a man, his wife (or wives), and their unmarried children. Variations of the nuclear family include stem families, where a married son continues living with his aged parent(s) and unmarried siblings, and fraternal families, where brothers live together with their spouses and unmarried children. The extended family unit includes the head of the family, his wife (or wives) and unmarried siblings, his unmarried children, and possibly his married children with their own spouses and children as well.[3] As the most fundamental family structures, these are the most likely to be in an individual’s consciousness as he or she participates in society, so they are also the most likely to play a part in administrative and legal documents.

The documents that most frequently show evidence of an extended family structure in ancient Mesopotamia are those dealing with the buying and selling of various kinds of property. In a nuclear family, it is likely that only a man and perhaps one or two of his adult sons would be named as sellers in the text of a contract. In an extended family, on the other hand, more people would likely have some right to whatever property was being sold, and thus more sellers would be named in the text. I. J. Gelb analyzed a synopsis of all the available sale contracts from the Fara to Ur III periods and discovered significant trends in the number of sellers listed in each document.[4] Most of the texts from the Fara and Pre-Sargonic (Early Dynastic) periods list five or more sellers, and there are no field sales and only one house sale with a single seller. This suggests that extended family groupings held joint rights to most types of property. In the Sargonic period, we see a shift toward a smaller number of sellers listed in each document. “The number of sellers of fields, houses, and orchards…varies from one to four, and, in contrast to the two preceding periods, over one half of the cases is represented by a single seller.”[5]

This shift might suggest a shift in familial attitudes, from a great deal of emphasis put on members of one’s extended family to a focus on the members of one’s immediate family. One might argue from here that there was indeed an ethnic difference between the Sargonic period and those that came before it. After all, the Sargonic period was linguistically Akkadian, whereas the preceding periods were linguistically Sumerian. The relatively sudden change in both the language and family structure gives stronger evidence for an ethnic change than either would by itself. The Sumerian population living under Akkadian rule could still have valued the extended family as much as it had previously, with the change only involving how many of the people with rights to a piece of property had to be named in sale documents. Perhaps the Akkadians, with their stronger emphasis on nuclear families, saw no reason to require more than the members of one nuclear family to be involved with the sale. A possible scenario might involve a Sumerian extended family choosing one of its members to be listed as the primary seller of a piece of property, even though the entire family held de facto rights to the property up to the time when it was officially sold.

Upon further investigation, however, this theory falls apart. The Ur III period, which the Gutian interregnum separated by a few decades from the Sargonic period, was again linguistically dominated by Sumerian. If the apparent change in family structure was due to an Akkadian ethnic group, one would expect the Sumerians to return the administrative system to its previous state and once again start listing a large number of sellers in sale contracts. This did not happen. On the contrary, the prevalence of single sellers continued to grow into the Ur III period. “The number of sellers of houses, orchards, slaves, and animals varies from one to four, with the overwhelming majority of cases represented by one seller only.”[6]

There are a few possible explanations for the continuation of this trend into a period that was again linguistically dominated by Sumerian. To support the claim that an ethnic change was responsible, one would have to argue that, despite the resurgence of Sumerian as the dominant language, the administrative population remained mostly Akkadian, and thus continued to focus more on the nuclear family than on the extended family when listing sellers of property. Another explanation is that some non-ethnic factor caused a widespread change in family structure that continued through the Sargonic period and into the Ur III period. Without further investigation, it is hard to say what such a factor might have been. A third possibility is that no widespread change in family structure occurred at all. Instead, economic or social factors may have simply made it more feasible for individuals to own property without the involvement of their extended families. Or perhaps there was simply a change in administrative practices, from requiring every seller to be listed in sale documents to only requiring the chief seller or sellers to be named. In either case, we would expect to see a corresponding decrease in the average number of sellers whose names appear on contracts. In light of the existence of these alternate possibilities and evidence to be further explained below, it seems most likely that, while the living arrangements of family groups may have changed, this change was accompanied neither by a significant change in how people thought about the family nor by an ethnic shift.

A change in family living arrangements might be revealed through documents regarding the sale of houses. There is no reason to assume that a large number of sellers for a field or orchard corresponded to a common residence, but at the same time, “there is no reason to assume that a large number of related individuals had the rights to a house unless they lived together.”[7] From the fact that the Fara and Pre-Sargonic periods were characterized by high numbers of sellers, between seven and nine on average, we can conclude that extended families tended to live in the same houses or at least in a localized cluster of houses. In the Sargonic and Ur III periods, the usual number of sellers dropped significantly to between one and four. However, as was shown above, there are explanations for such a change that do not involve an actual shift in family structure. It is doubtful that family living arrangements would change appreciably without a corresponding difference in the number of sellers, but it is entirely possible that the number of sellers would change without a corresponding difference in family living arrangements. Even if we do accept a change in living arrangements, this is not sufficient to conclude that there was a change in the general role of the extended family:

Whether or not a family remains co-resident has much to do with external conditions: within the city the finite availability of space combined with existing property rights forces an expanding family apart…. Nevertheless, it is critical to appreciate that the dissolution of co-residence and the consequential division of rights in urban property does not necessarily undermine the theoretical or practical role of the extended family in the areas of marriage and land tenure.[8]

So while this doesn’t constitute evidence against a change in the importance of the extended family, it does show that none of the evidence so far examined conclusively points to such a change as having taken place between the Fara and Ur III periods.

There is also a great deal of textual evidence that can reveal the importance of various aspects of the nuclear family unit. Unfortunately, when relying chiefly on impersonal and administrative texts, it is much more difficult to analyze the attitudes people had about how families ought to function than it is to analyze how they actually did function. While the actual functioning of families is certainly related to people’s attitudes about families, it is these attitudes themselves, more than the situations to which they pertain, that are most likely to be involved in any ideas of ethnicity, which is still the basic focus of this paper. The following look at examples of the structure and function of nuclear families is an attempt to glean some information about attitudes people had about families, rather than an analysis of the functioning for its own sake. We must rely on concrete indicators, such as the number of wives and children in a family, when seeking to discover whether there was any considerable change in attitudes concerning marriage and family size. Texts dealing with broken and/or destitute families can also be used to detect a general change in other aspects of family’s role in society, and hence to ascertain whether another kind of attitude change took place.

Throughout the second half of the third millennium, Mesopotamia was overwhelmingly monogamistic. “A man could not, with rare exceptions, have more than one formally recognized wife at a time. This is not usually stated explicitly, but both the law codes and accounts of court proceedings confirm it.”[9] In a series of texts analyzed by Gelb,[10] which range from Pre-Sargonic to Old Babylonian, only one out of the 112 married men listed had two wives. One group of texts does not specifically enumerate men and their wives, but rather deals with gifts to the wives of men in various temple households. It cannot be proven conclusively that these men each had one wife, but the fact that each of the nearly 50 named men appears only once as the husband of a gift recipient suggests this. The only other explanation would be that, by mere coincidence, only one of each polygamist’s wives happened to receive gifts. The rest of the documents list individual men, women, and children, grouped into family units. In only one of them is there a man listed as having more than one wife. It is again possible that only one wife of each of the other polygamists happened to be involved in whatever events were dealt with by each text, but it is extremely unlikely that this would have been the case for more than a few of the men listed as having one wife.

At first glance, it would appear that royal families throughout the period in question were as consistently monogamous as the families listed in the administrative texts discussed above. Ur-Nanshe, a Pre-Sargonic ruler of Lagash, is depicted as having a single wife and seven or eight children.[11] W. W. Hallo lists a single wife for Sargon and Shar-Kali-Sharri.[12] In the same chart, Manishtushu and Naram-Sin do not have wives listed, but the number of children (two and eight, respectively) is within a reasonable range for a single mother. For the Ur III dynasty, he lists one wife each for Ur-Nammu, Shulgi, Amar-Sin, Shu-Sin, and Ibbi-Sin. [13] However, he lists 19 names of children of Shulgi, and describes a possible unnamed twentieth as “wife of the ruler of Anshan”. One explanation for this is that the terms “dumu lugal” and “dumu-sal lugal” in the texts do not literally mean “son of the king” and “daughter of the king”, respectively. Rather, as Thorkild Jacobson assumed, they may be translated as referring to children of a king, that is, “prince” and “princess”.[14] The other obvious possibility is that Shulgi had, in addition to a queen, several concubines. This is plausible given that his wife is listed as the daughter of Apil-Kin, king of Mari. She, being of royal birth herself, would have been his reigning wife and thus the one named in royal genealogies. Any concubines may have been slaves or other women of lower birth, and so would not appear in these texts. This apparent change in the latter part of the Ur III dynasty[15] does not necessarily correspond to an actual change in marital practices among members of the royal family. It is possible that previous kings also had concubines, but that the children of these couplings only started being listed during Shulgi’s time.

Whichever explanation is correct for the unusually large number of children listed for Shulgi, it is impossible to know for sure how many children he fathered per wife (or concubine). If the Sumerian words refer to children of kings, rather than to children of a particular king, we have no way of knowing which of them were actually his (though we can be fairly certain that he fathered the next two kings). If he had several concubines in addition to a primary queen, we have no way of knowing how many he had, and thus no way to determine how many children each woman bore. Earlier kings, though, are listed as having up to children. This number is consistent with those in two votive inscriptions discussed by Gelb, one from the Pre-Sargonic period and one from the Sargonic.[16] These both name a votant, his wife, and eight children.

One gets a very different picture when looking at two of the administrative texts mentioned above.[17] The Pre-Sargonic text, which deals with 55 total people composing 12 families, lists 26 children. This gives an average of just under 2.2 children per family. The other text is from the Old Babylonian period. While this is later than any of the other texts discussed so far, I think it bears mentioning here as evidence against an overall change in nuclear family size. In its fully preserved descriptions of ten families, there are 22 children named. This gives an average of exactly 2.2 children per family, which is remarkably close to that found in the Pre-Sargonic text. The reason this number is so much lower than that found on votive inscriptions or in texts dealing with the royal family likely has to do with economic status. Clearly the royal family would have been wealthier than the average citizen, and could therefore afford to support more children per couple. More prosperous individuals also tended to be the ones making votive offerings, so it makes sense that the votive inscriptions would likewise refer to families that could afford to support a higher-than-average number of children.

This economic factor, though, suggests that attitudes about families and how they should be organized may not have had that great an impact on the number of children per average couple. Regardless of how large a family was seen as the ideal, if a big family was unaffordable, a couple was unlikely to have one. However, we can safely assume that the royal families, and perhaps those named in votive inscriptions, were able to afford as many children as they desired. This being the case, we can conclude from the consistency in the number of children that there was no considerable change in ideal family size from the Pre-Sargonic to Ur III periods.

A society’s attitudes about the family include how that society views the families on its margins. These attitudes are also related closely to its treatment of widows and orphans. Of the twelve families listed in the Pre-Sargonic document mentioned above, five appear to be headed by widows, and four by single men. As the colophon of that text states, the families made up a “total of 55 persons, small and great, who went to Guabba; property of Bau.”[18] This colophon and the skewed structure of the families led Gelb to the conclusion that the text “deals with destitute or impoverished families which placed themselves as clients at the disposal of the temple household of Bau in Girshu, whence they were sent to Guabba.”[19] It was common at that time for the temple households to provide a sort of charity for those who could not support themselves. This practice made sense, because temples could always use laborers, and they were well provided with food and drink given as offerings, which more often went to the people staffing the temple than to the gods. In addition to those offerings, temples made money from the commercial activities in which they took part. While many temple-employed craftsmen undoubtedly worked to provide for the needs of the temple itself, “textile production in particular was a highly organized commercial undertaking.”[20] Throughout Mesopotamian history, the weavers have almost invariably been women, whose children may have somehow assisted in the process. Rita P. Wright gives some evidence that the women working as weavers were in fact widows or other unmarried mothers:

Basic work units consisted of women and their sons and daughters. In the accounts, women were not referred to as the “wife of”, as was customary in naming high ranking women. Children of weavers were referred to by their matronym…. At death, a woman’s children were referred to as “orphans”.[21]

There is no evidence that this practice changed during the second half of the third millennium. By the Ur III period, the temple workshop at Guabba “employed 6000 workers, mostly women and children.”[22] This is a far larger number than earlier documents indicate, but that is more likely the result of economic changes than of a change in attitudes about the class of people in question. Due to the obvious benefits of having a cheap, dependant group of laborers (weavers did not get any land allotments, unlike most other professions, so they could not sustain themselves by their own means), it is unlikely that the temples were completely magnanimous in “allowing” women and children to work there. However, our concern here is not a society’s motivation for doing what it does with marginalized people, but rather whether or not what the society does changes enough to suggest an overall change in its views on those people.

In none of the sources examined thus far is there conclusive evidence for a shift in the role of the extended family, attitudes about marriage and family size, or the treatment of marginal family groups. In the absence of such evidence, it would be foolish for anyone to conclude that the existence of ethnicity is indicated by ancient Mesopotamian family structures, and thus the question asked by the title must be answered with an emphatic “no”. However, there is also no conclusive evidence against a change in attitudes about the family. It is possible that the sources here are too limited to definitively indicate a change that actually occurred in the role of the extended family, numbers of wives and children, or the treatment of marginalized individuals. Even if an actual change did not occur, attitudes could have changed without there being a corresponding shift in the sort of documented activities we’ve looked at. This is especially true of attitudes such as marriage rites or property inheritance, which were nowhere addressed in this paper. More comprehensive research is therefore required to discover what changes, if any, occurred in third-millennium attitudes about the structure and roll of the family.

Similarly, there is no conclusive evidence herein against the existence of ethnicity as an important part of the social identity of those who spoke Sumerian and Akkadian. We must not prematurely conclude from the continuity of attitudes about the family, assuming there really was such continuity, that the Sumerians and Akkadians definitely did not see each other as distinct ethnic groups. There is no real evidence in favor of this, but there was not in 1874 a huge amount of evidence for the distinct existence of the Sumerian and Akkadian languages. (Apart from written differences, which were not well enough understood at the time for anyone to be sure of the existence of spoken differences.) Yet Halévy’s argument from the continuity of art, geographic names, and the literary sources known at the time ultimately failed to produce an accurate account of the two primary languages written and spoken in ancient Mesopotamia.

[1] J. Halévy, quoted in The Sumerian Problem, ed. Tom B. Jones (1969), p. 26.

[2] J. N. Postgate, Early Mesopotamia: Society and Economy at the Dawn of History (1992), p. 88.

[3] Paraphrased from I. J. Gelb, “Household and Family in Early Mesopotamia” (1979), in State and Temple Economy in the Ancient Near East, ed. E. Lipinski, pp. 56-8.

[12] W. W. Hallo, “Women of Sumer” (1976), in The Legacy of Sumer (Bibliotheca Mesopotamica 4), ed. D. Schmandt-Besserat, p. 38.

[15] Gelb mentions that 18 children have elsewhere been listed for Amar-Sin (p. 66).

[17] Ibid,. pp. 60-65. The two texts discussed here are the only ones that specifically mention children per family.

[21] Rita P. Wright, “Crafting Social Identity in Ur III Southern Mesopotamia”, in Craft and Social Identity, ed. Cathy Lynne Costin and Rita P. Wright (1998), p. 65.