Bulletin on Sumerian Agriculture - 8 (1995): Domestic Animals of Mesopotamia, Part II

DOMESTIC ANIMALS OF MESOPOTAMIA PART I1

BULLETIN ON SUMERIAN AGRICULTURE Volume VIII

Cambridge 1995

General editors CONTENTS

J.N. Postgate Faculty of Oriental Studies University of Cambridge Sidgwick Avenue Cambridge CB3 9DA U.K.

M.A. Powell Department of History Northern Illinois University De Kalb Illinois 60115 U.S.A.

iv List of contributors and addresses ........................................................................................... Preface

........................................................................................................................... V-vi

Ethnographic evidence for water buffalo and the disposal of animal bones in southern Iraq: Ethnoarchaeology at Al-Hiba ....................................................................... 1-9 Bonnie Gustav & Edward Ochsenschlager Plough and Power: the economic and social significance of cultivation with the ox-drawn ard in the Mediterranean ................................................................................ 11-22 Paul Halstead How can modem food technology help to identify dairy products mentioned in Sumerian texts? ... 23-3 1 Michael Teuber Late Uruk Period cattle and dairy products: evidence from proto-cuneiform sources .....................33-48 Robert K. Englund Sheep and goat terminology in Ur I11 sources from Drehem ....................................................49-70 Piotr Steinkeller Plow animal inspection records from Ur 111Girsu and Umma .............................................. 7 1- 171 Wolfgang Heimpel

ISSN 0267-0658 Orders may be placed through booksellers or direct to the Sumerian Agriculture Group, Faculty of Oriental Studies, Sidgwick Avenue, Cambridge CB3 9DA, U.K.

Old Babylonian cattle ..................................................................................................173-213 Marten St01 215-240 Cattle in the Neo-Babylonian period ............................................................................... G. van Driel

O~umerianAgriculture Group 1995. All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted in any form or by any means without the prior permission of the publishers. Printed in England by Aris & Phillips Ltd., Warminster, Wiltshire

iii

Addresses of contributors Dr. G. van Driel

Dr. R.K. Englund

Department of Assyriology, University of Leiden, POB 9515, 2300 RA Leiden, The Netherlands Seminar fiir Vorderasiatische Altertumskunde, FreieUniversitiit Berlin, Bitterstr. 8-12, D 14195 Berlin, Germany

Bonnie Gustav

Department of Anthropology, Brooklyn College, Brooklyn, NY 11210, U.S.A.

Dr. P. Halstead

Department of Archaeology and Prehistory, University of Sheffield, Sheffield S10 2TN, U.K.

Prof. W. Heimpel

Department of the Near and Middle East, University of California, Berkeley, Ca. 94703, U.S.A.

PREFACE In our second volume on the domestic animals of Mesopotamia we resume the topic of sheep and goats, in the shape of the article of Piotr Steinkeller, and move into the topic of cattle discussed at the 1993meeting of the Sunerian Agriculture Group at Wassenaar, The Netherlands. This therefore gives us the opportunity to express our considerable gratitude for the hospitality of our Leiden hosts at that meeting which took place in the ideal surroundings of the NIAS Centre for Advanced Studies at Wassenaar: we are delighted to acknowledge the substantial contribution towards the costs of mounting the conference and towards the expenses of the participants received from the Royal Dutch Academy, from the CNWS at Leiden and from the Mesopotamian funds of the Department of Archaeology, Leiden University. We are most grateful to the NIAS and its staff for their kind and generous reception of the participants; and to our colleagues Marten St01 and Govert van Driel in particular, very many thanks from us all.

Prof. E.L. Ochsenschlager Department of Anthropology, Brooklyn College, Brooklyn, NY 11210, U.S.A. Prof. P. Steinkeller

Department of Near Eastern Languages and Civilizations, Harvard University, 6 Divinity Avenue, Cambridge, Ma. 02138, U.S.A.

Prof. M. St01

Heivlinder 27,2317 JS Leiden, The Netherlands

Prof. M. Teuber

Department of Food Science, ETEI Zentrum, CH-8092 Ziirich, Switzerland

While on the subject of Leiden, it is very satisfactory for us to be able to report that Miguel Civil's edition of the Fanner's Instructions or Sumerian Georgica, presented to our meeting there in 1987 (see BSA 5, p. iv) has now appeared, through the co-operation of the Oriental Institute at Barcelona: M. Civil, The Fanner's Instructions: A Sumerian agricultural manual (Aula Orientalis - Supplementa 5; 1994, Editorial AUSA Apartado de C o m s 101, 08280 Sabadell (Barcelona), Spain. ISBN 84-8881003-2). The volume includes the reconstructed text and translation, a general commentary, discussion of some specific points, and an edition of new texts from Chicago.

On a sadder note, we have to record the death of Evan Guest in 1992, at the age of 90. Anyone who has worked on the natural resources of Mesopotamia must be familiar with the superb volumes of the Flora of Iraq, a series sadly suspended by the political and financial disasters which have overtaken that country. The first, and enormously informative, introductory volume of the Flora was mainly from the pen of Evan Guest, and during his long association with the Royal Herbarium at Kew he was instrumental in pushing the series along, contributing notes on the ethnography of the plants. Evan Guest attended the first meeting of the Sumerian Agriculture Group in Cambridge in 1983, and those who were present were privileged to hear an absorbing account of his early years in the service of the Iraq government agricultural administration. An obituary notice and a sketch appeared in the Daily Telegraph for the 3rd April 1992.

This is almost certainly the penultimate volume of the Bulletin. No further meetings of the Group are planned, but a ninth volume of the Bulletin 'is expected to include some further papers on sheep and goat, cattle, and pigs. We also hope to prepare an index of all volumes of the Bulletin for inclusion in Vol. 9.

ETHNOGRAPHIC EVIDENCE FOR WATERBUFFALO AND THE DISPOSAL OF ANIMAL BONES IN SOUTHERN IRAQ: ETHNOARCHAEOLOGY AT AL-HIBA Edward Ochsenschlager and Bonnie Gustav

As in previous volumes, bold type is used for Sumerian, italics for Akkadian words. Neo-Babylonian logograms are left in capitals. Bibliographical conventions generally follow the usages of the individual authors, but Harvard system references are used for books and articles, whereas more traditional Assyriological abbreviations are often preferred for text editions. These can be tracked down in one of the following works: R. Borger, Handbuch der Keilschriftliteratur, W. von Soden, Akkadisches Handworterbuch, and the Chicago Assyrian Dictionary. For Ur 111 references see M. Sigrist & T. Gorni, The comprehensive catalogue of published Ur 111 tablets (CDL Press, Bethesda, Maryland).

This volume was prepared on Macintosh computers and uses the font "Sippar" developed by Dr. Wilfred van Soldt of Leiden University, which, with his characteristic generosity, he has permitted us to adopt, ~ h a n E are s due to Jonathan Blanchard Smith of Cockroft Adams Ltd., for help with processing the texts, and to Trinity College, Cambridge for continuing material support in the production process. Nicholas Postgate Marvin Po well

July 1995

(Brooklyn CoUege, City University of New York)

WATERBUFFALO The Waterbuffalo of Iraq are river type buffalo (Bubalus bubalis). Some think they are most closely related to the Waterbuffalo of India, but a white patch which is rather common on the foreheads indicates some influence of the Nile breed. This latter influence, of course, could be quite recent. It was estimated in 1970 by an official from the agricultural department that cattle outnumbered waterbuffalo in Iraq by about 6 to 1. A well-cared for animal can live for about 18 to 20 years, and several have lived to be 25 years old or older.

Use The Mi'dan near al-Hiba keep waterbuffalo primarily for milk, dung and hides. Of the three, dung is the most important. One of the primary tasks for little girls, at a very early age, is to follow the waterbuffalo, collect their dung, and bring it back home carried atop their heads in baskets water-proofed with bitumen. It is mixed with straw or crushed reed, patted into thin disks by the women of the family, and then allowed to dry in the sun. These disks are used for fuel, when one desires to maintain a fairly even temperature over a period of time, for cooking, baking food or pottery and providing heat in cold, rainy weather. During summer dung patty fires provide an acrid smoke which keeps mosquito and fierce biting flies at bay for both owners and their animals. At the height of the insect season when the Mi'dan build a fire the whole herd jostles for the best position on the downward side in order to push their heads and faces into the smoke. In spite of their rugged appearance, waterbuffalo are very sensitive to large numbers of insects and an unhappy buffalo soon loses condition. The dung is also used for repairing leaks in a reed structure and for both waterproofing and sealing storage containers in the courtyard. Applied fresh, it dries to a cement-like hardness in the sun. It is applied to the forehead for headaches and used as a healing agent for burns. For cuts and wounds it is used to stop the flow of blood as well as for healing. Buffalo milk is served at room temperature, as well as heated and sweetened with sugar, as a drink, and is used in making gravies. In the heating process the heavy cream rises to the top and is scraped off and used on bread. The cream which rises to the top of unheated whole milk is made into butter. Butter is sometimes made in a hollowed-out gourd which is shaken from Bull. Sumerian Agric. 8 (1995)

1

Ochsenschlager and Gustav

Waterbuffalo and bone disposal at Al-Hiba

side to side and sometimes in a sheepskin. The apertures of a tanned sheep skin are sewn shut except for the neck. The cream is put in this leather bag, air blown into it (as into a balloon) and the neck tied tightly closed. Butter is produced by rolling the skin from side to side, shaking it briskly, and in the final stages kneading it with both hands. The buffalo are milked by both men and women, into wooden pails with tapered, knobbed or pointed, bottoms or, into sabbat, baskets coated with bitumen on the interior. Most people milk them from their flanks, but one man was seen milking from behind the cow. When questioned, he claimed it was easier to milk this way if the cow was as quiet and reliable as the old one he was milking. Younger ones were too nervous and could cause serious injury to the milker if they idcked. Calves are restricted in the access to their mother's milk at an early age. They stay tethered in the courtyard while the cow forages for grasses during the day, and wear muzzles made out of cord when the adults return from the marshes in the evening until the cow's owner has first drawn part of her milk for his own use. Meat from the waterbuffalo can always be found for sale in the market towns. This usually comes from male animals sold by the Mi'dan for cash. It is very unusual for a Mi'dan to slaughter a waterbuffalo himself although they will occasionally do so if an animal is injured past saving or if the family is very hungry during a period of food shortage. Male buffalo, of course, are as productive as females when it comes to the most important product of these animals - manure. No waterbuffalo in the area is ever used as a draught or pack animal. Even in other parts of Iraq where cattle are often used for draft, buffalo are kept for milk. A good cow will produce between 18 and 19 pounds of milk while a good buffalo will give as much as 25 pounds.

Feeding Water Buffalo exist almost entirely on grasses, sedges and young reeds from within or on the edge of the marsh. For most of the year they forage for themselves during the day while their owners collect fodder to feed them when they return from the marshes. During the rainy season, however, the marshes are sometimes too deep for the buffalo to reach the grasses and their owners and the owners' families must collect enough to feed them during the day as well. If the water is very high, the forage must be carried to the banks or house for the buffalo to eat. If it is only moderately high, the buffalo may accompany a member of their owner's family into the marsh. The harvester will cut the young growth beneath the water's surface, and the buffalo will eat it as it floats to the surface. This is probably also the way the animal is trained to crop forage from beneath the water's surface, as grazing beneath the water is not natural to the waterbuffalo. If the water is altogether too deep or if there are too few people in the household to cut forage for them, an owner may feed their herd barley and chopped wheat or barley straw for a brief time. Waterbuffalo are more stubborn than cattle about the introduction of a new food should a dietary supplement be needed. They will resist trying something new for days, often until their condition has deteriorated. Cattle, on the other hand tend to experiment with the new food after a few hours of hunger.

Housing Some owners of waterbuffalo provide a roofed structure or sitra to protect their livestock against inclement weather (see BSA 6, 1992, pp. 59-60). Occasionally an owner will tether his

Ochsenschlager and Gustav

Waterbuffalo and bone disposal at Al-Hiba

buffalo to stakes at night in the lee of his dwelling. Most, however keep all their livestock in a courtyard fenced with reeds or mud. During the chilly weather of late winter and early spring, the closer the buffalo are kept to the family's dwelling the warmer the interior of the house will remain. If the animals are not tethered to stakes but allowed to roam free in the courtyard, ditches, two to three feet in width, or holes of the same width are dug every two feet alongside interior walls of mud or reed to protect them against the lumbering strength of the waterbuffalo. Such protection is particularly needed in the spring when the animals seek anything solid against which they can scratch off their itchy winter coats. Husbandry In the villages around al-Hiba the largest herds of waterbuffalo numbered near 20. The ordinary owner kept closer to eight to ten. In the morning the cows are milked, calves allowed to nurse and then the adult buffalo are released from their evening's confinement to make their way across the mound to the marshes. The cattle seem always to follow the same paths from the villages emerging eventually into long lines on three main paths which the cattle from the surrounding villages seem to share. Most of the time the cattle set off without urging when released from their compound. If the water is still fairly deep or if a cow has a calf at home, the cattle may require some urging. Village boys or girls will urge them along with a stick until the animals are caught up in the mass movement and move off on their own. The breeding of the waterbuffalo seems to be very haphazard. No control is apparently exercised by the owners. The animals can mate indiscriminately in the marshes with any of the males from the surrounding villages. In most cases, an owner would be hard put to tell you who was the father of a newborn calf. Holes are often drilled through the end of the horns and pieces of yarn or cloth tied through them. These are talisman to prevent injury and bring luck but also function in making each of his herd recognizable to their owner, even at some distance. Cows carry their young for eleven to twelve months. When they are about to give birth it is not unusual for the owners to bring them inside their houses for the event, and some calves left behind when their mothers go to the marsh pastures during the day seem to have free use of the house interior wandering in and out at will. A cow sells for about 75 dinars and a bull for 45 to 50 dinars. Sick animals are usually treated with a hot shish and wounds are dressed with a concoction of reed ashes and salt. Waterbuffalo appear stolid, peaceful and content and most of the time they are so. They are capable, however, of building up great resentment against an owner if mistreated or against another waterbuffalo of either sex whether in or out of their own herd. Once they attack each other, the battle is almost impossible to stop until one of the animals is dead or dying. Such a confrontation between two cows broke out on the mound at al-Hiba one day and the owners were joined by sixty of our workmen who did everything possible to distract the cows' attention and herd them away from each other. All was to no avail. Lumbering beasts turned into skilled fighters which charged and maneuvered with deadly speed and grace, unfazed by blows from sticks and clubs, and apparently totally incapable in their frenzy of seeing anyone but their enemy.

Ochsenschlager and Gustav

Waterbuffalo and bone disposal at Al-Hiba Ochsenschlager and Gustav

Waterbuffalo can acquire a taste for bitumen. This can be very hard on bitumen covered boats which the Mid'an depend on for transportation. (see BSA 6,1992, p. 52). Although the Mi'dan and waterbuffalo live together in a symbiotic relationship one should probably not think of the waterbuffalo in this area as fully domesticated. Man offers some protection and at times food in return for what the animals produce naturally. There has been no attempt to "improve" them by controlled breeding programs or to train them as draft animals.

ANIMAL BONES Refuse pits are associated with each family's housing complex. A single pit serves the entire family and is located outside the courtyard and within 10 to 25 meters of the same. They consist of rather shallow holes about one meter in depth and from two to three meters in diameter. Household refuse is disposed of in these pits. When they are nearly full, they are covered over and a new pit dug close-by. Among the items consigned to these pits are the leftover remains from meals which usually consist only of some of the bones, cartilage and sometimes pieces of skin. Butchering of animals for meat is always carried out near the currently used pit and the remains from this process are cast into them as well. One would imagine that these pits would become quite odorous, especially in the summer, but that is not the case. Dogs, who have attached themselves to the family, stake out their independent territories around this pit and pick it clean of anything remotely digestible. Domestic birds such as chickens or turkeys, and domestic cats, which die of natural causes are also assigned to these pits, but not the larger domestic animals such as dogs, sheep or cattle. The latter are dragged off as far as possible so that the odor of their decay will not affect the quality of life. A large carcass such as this immediately becomes the focus of all the dogs in the neighborhood. They descend on it, fight over it, and if the carcass is particularly large, such as a waterbuffalo, establish temporary territories as close to it as possible which they maintain until the animal is fully eaten. Weaker dogs take advantage of an opportunity to race towards the carcass, grab some edible morsel and retreat with it to their permanent homes, the environs of their family's garbage pit. When the dead animal is reduced to bones, dogs will continue to frequent the area selecting an appropriate bone to chew on and often carrying it off to their permanent territory. The same is sometimes true of the remains of the dog's regular quarry, the wowie (jackal) or fox, if it is successfully hunted by one or two dogs. If a whole pack is involved in the chase, however, the entire animal disappears in an eating frenzy and the ground where it is captured is spattered with blood, not always just that of the captured animal. As a result of all this, the refuse pit and its environs take on a most interesting pattern or configuration. Our study of two of these pits shows that by the time the pit is buried, it contains less than 10% of the surviving fragments of bones that were deposited there by humans. About 20% more of the surviving fragments can be found within a radius of 10 meters. The remaining 70% have either been entirely consumed or are scattered all over the village and its surroundings. Family dogs carry them off to shady places for leisurely gnawing, and dogs belonging elsewhere make successful forays and raids. Furthermore, there are bones represented, sometimes in the pit itself, which are the results of scavenging and hunting by the dogs and are not the product of human activity. It is clear to us that we must take into account the possibility of similar bone dispersal in the study of the refuse pits at al-Hiba.

Waterbuffalo and bone disposal at Al-Hiba

AFTERWORD - 20 YEARS LATER Returning to the site in 1990 we found many changes. Among these one of the most striking was the complete disappearance of the Mi'dan, who no longer lived in the area. In 1970 the Beni Hassan had looked down with contempt on the Mi'dan for keeping waterbuffalo. They considered the waterbuffalo as at least partially unclean, for many would not knowingly eat of its flesh, and as somewhat useless when compared with the productivity of their domesticated cattle. Imagine our surprise then to find that despite the absence of the Mi'dan in 1990, the number of waterbuffalo in the area had only decreased by about 20 to 25% (about the same amount of decrease as had occurred in the size of the marsh land in the area). Furthermore fhe waterbuffalo were now being kept by the Beni Hassan who had so thoroughly disparaged them in the past. The reason for their change in attitude was made clear by some rather embarrassed owners who remembered elaborating twenty years before on the disgusting nature of the waterbuffalo and the Mi'dan who kept them. Waterbuffalo exploit an ecological niche in the marshes in which no other domestic animal can function. Clearly the economic benefits derived with but moderate effort (the marshes rarely flood to the depth where families must forage for their waterbuffalo), is enough to justify a change in attitude. Waterbuffalo meat is eaten more frequently in the area than it used to be, but it is still against the law for butchers to sell it labeled as meat from domestic cattle. During our stay in 1990 two butchers in the area were closed down by the police for doing so, and one who was said to have done so knowingly was put in jail.

Ochsenschlager and Gustnv

m.

-- --

-

Waterbuffalo and bone disposal at Al-Hiba

PLATE I

Waterbuffalo and bone disposal at Al-Hiba

Ochsenschlager and Gustav

PLATE I1

.-.

1.Women heating waterbuffalo milk

1, Sitra, or c o d for waterbuffdo, built with the owner's home in the center of the back wall.

2. Building a new sitra with the help of neighbors.

2. Woman using a gourd butter churn

Ochsenschlager and Gustav

Waterbuffalo and bone disposal at Al-Hiba

Waterbuffalo and bone disposal at Al-Hiba

Ochsenschlager and Gustav

PLATE 111

PLATE IV

1. A sheepskin butter churn

1. Dried dung patties stacked to shed water during the rainy season

2. Using dung patties to heat a tabag which serves as a kind of frying pan 2. Making dung patties

PLOUGH AND POWER: THE ECONOMIC AND SOCIAL SIGNIFICANCE OF CULTIVATION WITH THE OX-DRAWN ARD IN THE MEDITERRANEAN Paul Halstead (University of Shefield)

A rich and diverse body of archaeological evidence indicates that an 'ard' or 'scratch plough' drawn by a pair of cattle was known, and in some cases used, in many parts of Europe from at least the 4th-3rd millennium BC (Sherratt 1981; Chapman 1982). In the Mediterranean region, 3rd millennium BC models of a ploughing scene from Vounous in Cyprus and of yoked cattle ~ the ~ ard and of from Nemea in Greece (Pullen 1992) provide early evidence for m w l e d of animal traction; apparent osteological evidence for improved survivorship of male cattle at 3rd millennium BC Pevkakia in Greece may indicate the raising of oxen (i-e. castrated male cattle) for traction (Halstead 1987a, 81 Table 2), while Linear B texts from the late 2nd millennium BC palaces of southern Greece unambiguously record the existence of pairs of working oxen (Ventris and Chadwick 1973,212; Palairna 1989,91). In addition to its economic significance, it has been suggested that the introduction of the ox-drawn ard played a major causal role in the development of stratified society in bronze age Europe. Two principal arguments have been advanced. Gilman (1981) has suggested that ploughing with oxen necessitates fixed fields, cleared of obstacles such as boulders and tree stumps, and thus represents a major capital intensification of arable farming. This in turn enforces more permanent settlement and so prevents farmers from, literally, escaping the clutches of emerging elites. Gilman's case rests partly on the assumption, now almost universally rejected (e.g. Rowley-Conwy 1981; Behre and Jacomet 1991; Jones 1992), that early neolithic farming in Europe was characterised by short-term, swidden cultivation of temporary clearances and consequently by impermanent settlement. Sherratt (1981) draws on Goody's survey of ethnographic literature (Goody 1976), which demonstrates a clear cross-cultural association between plough-based agriculture and stratified societies, on the one hand, and between hoe cultivation and more egalitarian societies, on the other hand. Goody draws attention to two possible causes which may underlie this association. First, compared to manual cultivation, the plough raises output and so makes it possible to support non-producers (e.g. a ruling elite plus retainers). Secondly, the plough involves more extensive land-use and so increases the likelihood that land will become a scarce resource, distributed unevenly between rich and poor. Both propositions are inherently plausible and, in each case, Goody cites sparse supporting evidence of diverse geographical origin. On the other hand, area yields (i.e. yields per unit area) of ard agriculture may be lower than those of manual cultivation (e-g. Gallant 1991, 51; Sallares 1991,477; Bogucki 1993,498), though low area yields are of course off-set by the capacity of oxen to cultivate a large area. Moreover, in Bull. Sumerian Agric. 8 (1995)

11

Halstead

Plough and power

assessing the productivity of ard agriculture in terms of yields per unit labour, the cost of maintaining work oxen must also be taken into account (Bayliss-Smith 1982,107-8). In an attempt to clarify the social significance of the ox-drawn ard in Mediterranean prehistory and early history, this paper reviews a variety of evidence for the costs and benefits of ard agriculture and manual cultivation. Because both costs and benefits are likely to differ greatly between contrasting environments, the evidence reviewed is mostly drawn from the circum-Mediterranean region.

The costs of maintaining work oxen Large work animals such as oxen have both capital and running costs. The capital costs of raising or buying young oxen obviously vary, depending on the number of female cattle kept for breeding and dairy purposes (e.g. Bogucki 1993). In 19th century Greece, farmers paid the equivalent of the purchase price of 1.2 ha of land or 40-45 sheep for a pair of oxen (Psikhogios 1987, 48), while some early 20th century farmers who bred their own oxen from the family cow(s) waited several years to raise a full ploughing team (Halstead field notes [Assiros, Central Macedonia]). For many farmers, therefore, neither buying nor raising oxen was easy. Large animals also need significant areas of pasture, the availability of which inevitably varies in accordance with differences in climate and soil, rural population density, land tenure and land use. The animals may need to be taken to pasture throughout the year (e.g. Aschenbrenner 1972, 52), though such day to day herding may be undertaken by children without other major work commitments (e.g. du Boulay 1974, 246-7). In addition, work animals must be fed supplementary, harvested fodder, especially during the working season when their energy expenditure is high and time available for grazing is reduced. One study in Greece (Halstead and Jones 1989) has documented some of the labour costs entailed in providing harvested fodder for work animals. If crops are grown specifically for fodder, additional human labour is required throughout the arable farming cycle from ploughing and sowing through harvest to threshing and winnowing (fodder crops are usually threshed and winnowed, because farmers feed the grain and straw to different animals in different proportions at different times). Elderly farmers in northern Greece report sowing between 0.5 and 2 ha of bitter vetch specifically for the plough oxen, with the upper limit on the area sown being set not by the nutritional demands of the oxen but by the availability of human labour for harvesting this low and scrambling crop (Halstead field notes [Assiros, Central Macedonia]). Much of the fodder for large work animals may be derived from the processing byproducts of crops grown for human consumption. Such by-products are often treated by farmers as involving no cost (cf. Langdon 1986, 158-9), but they greatly increase the burden on human labour at harvest and threshing time. Cereal crops are cut low to recover much of the straw, making reaping both slower and more back-breaking and greatly increasing the volume of crop to be transported from field to threshing floor. The crop to be threshed now consists largely of straw rather than just of grain and chaff and, although much of the labour of threshing is effected by the trampling of work animals' hooves, this remains a laborious task and has to be undertaken at the hottest time of day when the midday sun makes the crop brittle. Finally, because the light straw and chaff are needed as fodder, the threshed crop can only be winnowed in a light breeze, thus significantly slowing down the cleaning of the crop (Halstead and Jones 1989).

Halstead

Plough and power

In the Kingdom of Naples, capital and running costs together made a day's work by a pair of oxen as expensive as that of ten labourers in the 18th century and of six labourers in the late 19th century (Delille 1977,147). Even if young male cattle, pasture and child labour for herding are freely available, and even if additional grain crops are not sown, the use of food crop residues as fodder enforces a substantial increase in agricultural work during the harvest and threshing period of maximum labour stress. In making this substantial investment of labour, the farmer is effectively 'storing labour' (cf. Rowley-Conwy and Zvelebil 1989, 49) by buying 'energy slaves' (Bayliss-Smith 1982,52-53) capable of considerable amounts of work during the critical season of tillage and sowing in the autumn and winter. This point may be made clear by comparing the areas cultivable by hand and by ox-drawn ard.

The benefits of oxen: tillage by hand and by ox-drawn ard To ask how much land can be ploughed by a pair of oxen in a day is rather like asking 'How long is a piece of string?'. The answer depends, inter alia, on the size and condition of the oxen, the type of plough and plough share, the nature of the soil and terrain, the weather before and during ploughing, when and how well the field was last ploughed, the spacing between furrows, and the size, distance apart and distance from the farmer's base of individual fields. Nonetheless, many cultures measure land in terms of a rule-of-thumb figure for the area which can be ploughed in one day (e.g. Evans 1960, 40-41). The classical Roman unit of land measurement, the jugerum (cf. jugum = 'yoke') represents 0.25 ha; in 19th century Greece, the stremma ('turning' and hence area which oxen can plow in a day) varied locally between 0.09 and 0.16 ha (Psikhogios 1987, 24-25); and in various parts of the East Mediterranean, the Turkish doniim (also 'turning') covered a similar range. Recent estimates of the area which can be ploughed in a day by a pair of oxen range from 0.1 to 0.4 ha in 20th century Palestine and Jordan (Russell 1988, 123 Table 27; Palmer and Russell 1993). Estimates from 20th century Greece range from 0.1 to 0.3 ha, though the higher figures are sometimes linked to the use of two or three pairs of oxen in rotation with one plough (Aschenbrenner 1972, 57; du Boulay 1974, 242; Halstead and Jones 1989; Halstead field notes [Assiros, Central Macedonia]). The norm for ploughing in Mediterranean conditions under traditional technology, therefore, may be regarded as roughly 0.1-0.3 hatday. By contrast, estimates of the area which can be dug by hand with a variety of tools in different parts of the world range between 0.02 and 0.05 hdday (Russell 1988, 114 Table 19). In 19th century Greece an axinari, the area cultivable by hand in a day, was reckoned at ca. 0.05 ha (Psikhogios 1987, 34), while one estimate of the area cultivable manually by early 20th century market ,gardeners in northern Greece suggests a range of 0.02-0.05 hdday (Halstead field notes [Assiros, Central Macedonia]). In other words, tillage with an ox-drawn ard is something like 2-15 times faster than manual cultivation. In effect, the greater speed of the oxdrawn ard allows farmers to break one of the principal bottlenecks in the agricultural year - the autumn and winter period of tillage. The increased scale of cultivation thus made possible has radical implications for agricultural methods at subsequent stages in the crop growing cycle (Halstead 1987b). The implications of tillage by ox-drawn ard for subsequent stages of crop husbandry First, a major increase in the area sown requires more rapid methods of sowing and, as a consequence, ard-based agriculture tends to be associated with broadcast sowing (Sigaut 1975,

Halstead

Plough and power

220-21), while small-scale manual cultivation tends to involve more precise but more labourintensive methods such as dibbling or planting. The disadvantage of broadcast sowing is that it tends to use up twice as much seed corn as the more intensive methods (Halstead 1990). Secondly, small-scale manual cultivation is often associated with relatively intensive husbandry such as weeding by hand, whereas weeds tend to be controlled under large-scale ard-agriculture by regular, ploughed fallowing. Thirdly, small-scale cultivation tends to be characterised by annual cropping of land, often involving cereal-pulse rotation, while large-scale ard-agriculture tends to specialise in the less labour-intensive cereal crops. Thus two broad strategies may be distinguished (Halstead 1987b; 1992a): (1) Manual cultivation, with intensive husbandry methods, is characterised by relatively high area yields, but is small-scale and has high labour inputs, and so produces relatively little surplus over and above what is needed to feed the human workforce involved. (2) Ard cultivation, with extensive methods, is characterised by low area yields, but is large-scale and has low human labour inputs, and so has the potential to produce a large surplus. The greater capacity for surplus of ard agriculture is facilitated by the fact that a significant part of the fodder for the critical work animals is made up by the processing byproducts of human food crops.

The harvesting problem While more extensive methods of crop husbandry can be adopted to compensate for the larger scale of ard agriculture, and while work animals can contribute to the transporting and threshing of the crop after harvest, potential improvements in the speed of harvesting were limited before the relatively recent introduction of true scythes and then of mechanical reaping machines. A series of estimates by Greek farmers for reaping with sickles suggests normal cereal harvesting speeds of UD to 0.1 hdman day. Slower rates are reported for pulses and also for elderly workers, while some of the faster estimates exclude the labour of a second person (often the wife of the reaper) engaged in binding the crop into sheaves (Halstead and Jones 1989,47; Halstead field notes [Assiros, Central Macedonia]; Wagstaff and Augustson 1982, 131). Figures from a variety of recent European and Near Eastern contexts (Russell 1988, 116-17, Tables 20-21) suggest harvesting and binding speeds for cereals cut with a small- or mediumsized iron sickle of ca. 0.05-0.1 hdman day. This compares with rather higher speeds for ploughing with the ox-drawn ard of 0.1-0.3 hdday (above). Russell has also assembled figures for possible reaping speeds with more primitive technology. Reaping experiments with flint, obsidian, copper and bronze sickles suggest speeds of ca. 0.02-0.05 hatman day (Russell 1988, 116 Table 20), accentuating the contrast between ploughing and reaping rates. In sum, an ox-drawn ard can plough more land in a day than one man can reap, but some of the faster tillage rates may involve the labour of two persons (one to plough and one to sow) and of course farming families commonly mobilise their entire work force at harvest time. Where the farmer has a large family including several adult offspring, therefore, the area which can be ploughed and sown per day with the aid of an ox-drawn ard and the area which can be reaped per day by the family may be in balance. Where the farmer commands a small family labour force, however, the implicatibn is that the services of additional reapers must be secured or that the harvest season must be substantially longer than the ploughing season (perhaps involving underuse of the oxen).

Halstead

Plough and power

Up to a point, the length of the harvest season can be extended by growing a range of crops with different maturation times, by staggering the sowing of each crop type and by sowing crops in several fields differing in altitude, aspect and soil type (e.g. Forbes 1982). Once the crop approaches ripeness, however, it must be harvested with some urgency, before the grain is shed or robbed by birds or rodents (Halstead and Jones 1989). In Greece, before mechanisation, the barley and wheat harvest might typically be spread over up to one month, and this period might be extended to up to two months if the crops to be harvested included the early-ripening pulses and the late-ripening oat, rye and millet. The pulses are usually sown on a relatively small scale on large farms, however, and oat and rye are not yet attested as crops in prehistoric Greece and millet only rarely so (Halstead 1992a, 108 Table 1; 1994,204-5 Table 7.1). By contrast, the winter ploughing season in Greece might typically extend over one to three or even four months between October and February, depending on how early the autumn rains began in a given year and how much time was lost because the ground was too wet, heavy or frozen for cultivation (Aschenbrenner 1972,50-51; du Boulay 1974,242; Kostis 1993,41; Halstead and Jones 1989; Halstead field notes [Assiros, Central Macedonia]). Consideration of the relative lengths of the harvesting and ploughing seasons, therefore, again accentuates the contradiction between the ploughing potential of a pair of oxen and the harvesting constraints imposed by a family labour force. More meaningful than a comparison between daily ploughing and reaping rates is a comparison of the area which can be ploughed and harvested per year, though estimates of these figures are of course greatly complicated by variation in family size and structure and in the length of the ploughing and harvesting seasons. Nonetheless, some regularities again emerge.

The scale of tillage and harvesting possible on a family farm For 20th century northern Greece, recent estimates by several elderly farmers suggest that the maximum area cultivable per family per year by hand was ca 3 ha and the maximum with oxen ca 10 ha, of which about half would be in winter cereals and half fallow or in summer crops (Halstead field notes [Assiros, Central Macedonia]). Figures from Byzantine times are broadly comparable: average allocations per ox pair of 8 ha and 15 ha (including fallow or summer crops) are recorded (Harvey 1989,50-51,140), while farmers without oxen were allocated one quarter the area of land assigned to those with a pair of oxen (Harvey 1989,52). An average allocation of 10 ha per ploughman is also recorded for a 17th century estate in Macedonia (Moskof 1979,60-62). For southern Italy, two parliamentary enquiries into late 19th century and early 20th century agriculture offer further comparative data (Delille 1977, 123). In the rich plains of Campania, where a pasture-maize-wheat-oat rotation was practised, a farmer with oxen might take on 10 ha (presumably 5 ha in winter cereals), while a tenant farmer without oxen might take on a maximum of 2-3 ha (Delille 1977, 128). In the less fertile Appenine hills and in large parts of Apulia, under more extensive conditions of husbandry, the areas which might be farmed are ca 10 ha with oxen and up to 4-5 ha with manual cultivation (Delille 1977, 129). In the area of Crotone, a family without oxen could not rent more than 4 ha (Delille 1977, 136 n 89). Finally, in the Auvergne region of France, where oxen were used for ploughing until ca. 1950, farmers were able to cultivate up to 5-6 ha of winter cereals with one pair of oxen (Halstead field notes wireflews village]).

,

-

Halstead

Plough and power

Again comparison with the area which can be harvested by a family labour force is illuminating. In theory, at a rate of 0.1 hdday, a recent Greek reaper armed with a sickle could cut up to 3 ha of wheat and barley during a harvest season of one month; a family of two active parents and two grown sons could cut up to 12 ha. As noted above, however, the maximum figure of 0.1 hdday was probably only attainable by the most fit, industrious and skilled reapers, and perhaps ignores the labour involved in binding the crop. Farmers suggest that 1-2 ha of winter cereals per reaper is a more realistic figure for a harvest season (Halstead field notes [Assiros, Central Macedonia]). Moreover, the family will have to divide its labour force between harvesting, transport of the crop to the threshing floor, and normal maintenance activity with livestock, while time may also be lost if the crop is wet by summer rain. In practice, estimates by recent Greek farmers suggest that the maximum area of winter cereals which can be harvested by a reasonably active family is about 4 ha, perhaps 6 ha for a large family with adult offspring (Halstead field notes [Assiros, Central Macedonia] ; also du Boulay 1974, 242). The maximum harvesting capacity of a family labour force, therefore, at a favourable point in the domestic cycle, barely matches the normal cultivation potential of a pair of oxen. Further light may be shed on these figures by attempting an even more problematic estimate of the minimum areas in cultivation necessary to support a family and to justify the keeping of a pair of oxen.

The minimum size of a viable family farm Delille (1977, 132) estimates that, in southern Italy, 4-5 ha in a two-year cereal-fallow rotation would suffice to support a family of 4-5 persons (including some allowance for payment of taxes and other expenses, but not for the consequences of a bad harvest): the Mediterranean smallholder working the land by hand can only cultivate enough for bare survival (Delille 1977, 135). On the other hand, it may be inappropriate (see above) to extrapolate the extensive husbandry methods and low area yields of large scale cereal agriculture to a smallholding cultivated by hand. With intensive hoeing, weeding and perhaps watering and manuring, and with cereal-pulse rotation, a small holder might well produce average yields approaching 1 tonlha. In a prehistoric context, therefore, a farmer unencumbered by the need to pay taxes in money, might well be able to feed a family of 4-5 persons (needing, say, 1.5 tons of graidyear) from as little as 2-3 ha of intensively cultivated land. Hopkins (cited in White 1970, 336) similarly calculated the minimum area to support a Roman farming family at 2 ha without oxen. A family subsistence requirement of 2-3 ha of winter cereals is reported from 19th century and 20th century Greece (Psikhogios 1987, 39 n 13; Halstead field notes [Assiros, Central Macedonia]). In other words, the area needed to support the family of a 'subsistence farmer' (ca 2-3 ha) is comparable with the areas which a family labour force can cultivate by hand (ca 2-4 ha) and harvest (up to 4 ha). Conversely, a farmer making full use of the cultivation potential of a pair of oxen (5+ ha of winter cereals) is likely to be supporting a very large family or producing a substantial surplus. Finally, is it possible to estimate the area in cultivation necessary to justify the keeping of oxen, given the substantial if highly variable costs involved? Recent farmers in a part of northern Greece traditionally dominated by cereals, and whose own experience dates to the period in the earlier 20th century when large estates were dissolved and the land redistributed to large numbers of smallholders, suggested that ownership of a pair of oxen was only economic on holdings exceeding 3-4 ha (Halstead fieldnotes); below this level, farmers would variously

Halstead

Plough and power

share or rent oxen, use cows or equids or cultivate by hand (see below). The minimum holdings of the classical Athenian hoplite class, the zeugitai (i.e. ox-farmers), have similarly been estimated at ca 4-5 ha (Isager and Skydsgaard 1992, 79; also Foxhall 1992; Jarneson 1992; Osborne 1992, 24), while Hopkins (cited in White 1970, 336) has suggested a minimum requirement of 5 ha for a Roman farmer with oxen. Delille (1977, 135) emphasises the inability of the smallholder, in the Mediterranean as a whole, to raise large livestock. As a result, ownership of oxen has often been highly restricted. In early 19th century Greece, only one third of farmers possessed oxen and in 1860 there was an average of only one ox per farmer; many farmers cultivated by hand (Psikhogios 1987,47). Conversely, much of the cereal land in 19th century Greece was in the hands of large, oxpowered estates (Vergopoulos 1975) or 'chifliks' - from the Turkish ~ i for t 'pair' [of oxen] (cf. Asdrakhas 1978,41-43). A similar contrast between numerous small-holdings, too small to justify the keeping of oxen, and a few large estates with oxen is apparent for Byzantine Greece (Harvey 1989,63-64,71). Likewise, while many 19th century Italian farmers cultivated small holdings by hand, in Calabria in 1811 70% of the 13,750 oxen were in herds of 100-1200 head (Delille 1977, 136), that is attached to very substantial agricultural estates.

The economic implications of keeping plough oxen The implication of these various calculations is that, as regards the growing of grain crops, the 'traditional' Mediterranean farmer has been trapped between the minimum scale needed to justify the keeping of plough oxen and the maximum scale of harvest manageable with the labour force of a normal family. In other words, for all but the largest families at the most favourable point in the domestic cycle, farmers were faced with the choice of growing grain crops by hand on a scale sufficient for subsistence, or of investing in a pair of oxen, hiring outside labour at harvest time and producing a substantial surplus of grain for the market. In practice, of course, a number of alternative strategies are possible between these two extremes. First, the preceding discussion has avoided defining a 'typical' farming family. A nuclear family household may only temporarily muster the harvesting labour needed to match the ploughing potential of a good pair of oxen (while offspring are grown up but still attached to the parental household), but more extended 'stem' or 'compound' families, consisting of two or three generations under the same roof, were not uncommon in 19th-20th century Greece (Psikhogios 1987, 97-1 12). Such households can muster a relatively large and continuous labour force, but may also include enough consumers to neutralise the surplus producing potential of a pair of oxen. Secondly, in recent decades, Greek farmers with too little land to maintain oxen have usually ploughed with a pair of cows, a cow and a donkey, or a mule rather than resorting to manual cultivation. In this case, the capital and running costs of oxen are reduced by yoking a dual purpose animal, also used for milking or transport. Elderly farmers are in almost unanimous agreement, however, that the use of non-specialist plough animals greatly limits both the quantity and quality of ploughing, while also jeopardising the productivity of the cow in milk and calves, and making the donkey or mule unavailable for other work. Some smallholders have instead shared or hired oxen, but both solutions again limit the possible scale of cultivation and also mean that the farmer has only partial control over when he ploughs and sows and so incurs greater risk of crop failure or low yields (Halstead field notes [Assiros, Central Macedonia; Naxos]). As a result, grain production in much of the Mediterranean in

Halstead

Plough and power

early modern times has tended to be undertaken either by smallholders producing, at best, enough for subsistence or by large estates producing on a large scale, and with minimal input of human labour, for market (Delille 1977; Psikhoyios 1987; Vergopoulos 1975).

The social implications of keeping plough oxen The preceding discussion has documented a close relationship between the use of plough oxen and large-scale cereal agriculture with extensive methods of husbandry. It is now possible to reevaluate, in a Mediterranean context, Goody's suggestions as to the causal relationship between the plough and social stratification. As Goody noted, the ox-drawn plough permits large-scale surplus production and so enables the support of non-producers, such as an elite and its retainers; extensive plough agriculture is also 'wasteful7of land (compared to intensive manual cultivation) and so promotes land shortage, which in turn makes possible uneven access to land. In addition, at least under the scheduling constraints of Mediterranean climate, ox-based agriculture geared to surplus production of grain requires the existence of an external force of landless or otherwise dependent labourers at harvest time. The use by plough owners of fallow as a substitute for labour-intensive, manual control of weeds also presupposes preferential access to scarce land, while the costs of raising and maintaining oxen, in most parts of the Mediterranean, are prohibitively high for most farmers engaged in 'subsistence' cultivation. Social stratification thus emerges,pace Goody, as a precondition as well as consequence of oxbased agriculture in the Mediterranean (also Gudeman 1977). Finally, while both the anthropological and archaeological literature tends to be concerned with the relationship between the plough (or ard) and social complexity (Goody 1976; Sherratt 1981), the foregoing discussion suggests that the critical variable is ownership of specialised plough oxen and not mere use of the ard (also Braudel1975,77; Spurr 1986,171). Conclusion In the context of European or Near Eastern prehistory, therefore, it would not be appropriate to assign great social significance to archaeological evidence for the existence of the ard or yoked pairs of cattle or even for the practice of ploughing (see Rowley-Conwy 1987). On the other hand, references in the Linear B records of the late bronze age palaces of southern Greece to pairs of working oxen are more significant. First, working oxen were concentrated near major centres and sub-centres, as were recorded palace land holdings and cereal harvests, suggesting a connection between oxen and palatial agriculture. Similarly, some palatial land was allocated in units of ca 5 ha, that is of a size suitable for cultivation with a pair of oxen. Secondly, palatial records of grain harvest are restricted to one type of cereal, suggesting extensive agriculture, in sharp contrast with the archaeobotanical record, which suggests a balanced mixture of various cereals and pulses more characteristic of small-scale, intensive husbandry. Textual and bio-archaeological evidence in combination, therefore, suggest that, while much of the population practised intensive horticulture, the palaces (like the land-owning elites of the early modern Mediterranean) were in large measure supported by the surplus grain produced on large, extensively farmed estates, cultivated by plough oxen (Halstead 1992a; 1992b). While prehistorians tend to emphasise the role of economic intensification in supporting early states, it is tempting to speculate, on theoretical grounds, that extensive ard-agriculture played a major role in financing the surplus necessary to support many Near Eastern

Plough and power

Halstead

civilisations. In 3rd millennium BC Mesopotamia, the contrast between texts overwhelmingly concerned with one cereal and a diverse archaeobotanical record (Halstead 1990) again hints at specialised administrative involvement in extensive agriculture. In Sumerian literature, the association of the plough with gods and kings likewise suggests an elite context for extensive agriculture (Vanstiphout 1984,242-3; Hruska 1992,73), while the disputation poem between the hoe and the plough concludes that, for Sumerian society as a whole, the hoe was more important (Vanstiphout 1984,247).

Acknowledgements Attendance at the Wassenaar meeting of the Sumerian Agriculture Group was assisted by a grant from Sheffield University Foreign Travel Fund. The British Academy funded fieldwork in Greek Macedonia in 1993, during which interviews were conducted with elderly farmers who had worked on Turkish chiflik estates in the first two decades of this century. I am grateful to Yannis Hamilakis, Nicholas Postgate and Tom Williamson for useful bibliographic suggestions, and to Carol Palmer for drawing my attention to the meaning of the Turkish doniim.

I

BIBLIOGRAPHY Aschenbrenner, S. 1972 "A contemporary community", in W.A. McDonald and G.R. Rapp (eds.) The Minnesota Messenia Expedition (Mmneapolis: University of Minnesota Press), 47-63. Asdrakhas, S. 1978 Mikhanismoi tis Agrotikis Oikonomias stin Tourkokratia (Athens: Themelio). Bayliss-Smith, T. 1982 The Ecology of Agricultural Systems (Cambridge University Press). Behre, K.-E. and Jacomet, S. 1991 "The ecological interpretation of archaeobotanical data", in W. van Zeist, K. Wasylikowa and K.-E. Behre (eds.) Progress in Old World Palaeoethnobotany (Rotterdam: Balkema), 81-108. Bogucki, P. 1993 "Animal traction and household economies in neolithic Europe", Antiquity 67,492-503. du Boulay, J. 1974 Portrait of a Greek Mountain Village (Oxford University Press). Braudel, F. 1975 The Mediterranean and the Mediterranean World in the Age of Philip 11, volume 1 (London: Fontana). Chapman, J. 1982 "The secondary products revolution and the limitations of the Neolithic", Institute of Archaeology Bulletin 19, 107-22. Delille, G. 1977 Agricoltura e Demografm nel Regno di Napoli nei Secoli 18 e 19 (Naples: Guida Editori).

-

Halstead

Plough and power

Evans, G.E. 1960 The Horse in the Furrow (London: Faber and Faber). Forbes, H. 1982 Strategies and Soils: Technology, Production and Environment in the Peninsula of Methana, Greece (Ph.D. dissertation, University of Pennsylvania).

Halstead

Plough and power

Jones, G. 1992 "Weed phytosociology and crop husbandry: identifying a contrast between ancient and modern practice", Review of Palaeobotany and Palynology 73: 133-43. Kostis, K. 1993 Aforiu, Akriviu kai Pina (Athens: Ekdosis Alexandrias).

Foxhall, L. 1992 "The control of the Attic landscape", in B. Wells (ed.) Agriculture in Ancient Greece (Stockholm: Swedish Eastitute at Athens), 155-59.

Langdon, J. 1986 Horses, Oxen and Technological Innovation: the Use of Draught Animals in English Farming from 1066 to 1500 (Cambridge University Press).

Gallant, T.W 1991 Risk and Survival in Ancient Greece (Cambridge: Polity Press).

Moskof, K. 1979 Istoria tou Kinimatos tis Ergatikis Taxis (Thessaloniki).

Gilman, A. 1981 "The development of social stratification in bronze age Europe", Current Anthropology 22, 1-8.

Osborne, R. 1992 "'Is it a farm?' The definition of agricultural sites and settlements in ancient Greece", in B. Wells (ed.) Agriculture in Ancient Greece (Stockholm: Swedish Institute at Athens), 21-25.

Goody, J. 1976 Production and Reproduction (Cambridge University Press). Gudeman, S. 1977 "Morgan in Africa", Reviews in Anthropology, 575-80. Halstead, P. 1987a "Man and other animals in later Greek prehistory", Annual of British School at Athens 82, 71-83. 1987b "Traditional and ancient rural economy in Mediterranean Europe", Journal of Hellenic Studies 107, 77-87. 1990

"Quantifying Sumerian agriculture - some seeds of doubt and hope", Bulletin on Sumerian Agriculture 5, 187-95.

199%

"Agriculture in the bronze age Aegean", in B. Wells (ed.) Agriculture in Ancient Greece (Stockholm: Swedish Institute at Athens), 105-16.

1992b "The Mycenaean palatial economy: making the most of the gaps in the evidence", Proceedings of Cambridge Philological Society 38,5746. 1994

"The North-South divide: regional paths to complexity in prehistoric Greece", in C. Mathers and S. Stoddart (eds.) Development and Decline in the Mediterranean Bronze Age (Sheffield: J.R. Collis), 195-219.

Halstead, P. and Jones, G. 1989 "Agrarian ecology in the Greek islands", Journal of Hellenic Studies 109,41-55. Harvey, A. 1989 Economic expansion in the Byzantine empire, 900-1200 (Cambridge University Press). HruIka, B. 1992 "Hacke oder Pflug? Zur Bodenvorbereitung in den sumerischen Literaturtexten", in J. Prosecky (ed.) Ex Pede Pontis (Prague: Oriental Institute, Czechoslovak Academy of Sciences), 70-79. Isager, S. and Skydsgaard, J.E. 1992 Ancient Greek Agriculture (London: Routledge). Jameson, M. 1992 "Agricultural labor in ancient Greece", in B. Wells (ed.) Agriculture in Ancient Greece (Stockholm: Swedish Institute at Athens), 135-46.

Palaima T.G. 1989 "Perspectives on the Pylos oxen tablets", Studia Mycenaea 1988 (Skopje), 85-124. Palmer, C. and Russell, K. 1993 "Traditional ards of Jordan", Annual of the Department of Antiquities of Jordan 37, 37-53. Psikhogios, D.K. 1987 Proikes, Foroi, Stafida kai Psomi: Oikonomia kai Oikoyenia stin Agrotiki Ellada tou 19 Aiona (Athens: Ethniko Kentro Koinonikon Erevnon). Pullen. D.J. 1992 "Ox and Plough in the Early Bronze Age Aegean", American Journal of Archaeology 96,4554. Rowley-Conwy, P. 1981 "Slash and burn in the temperate European Neolithic", in R.J. Mercer (ed.) Fanning Practice in British Prehistory (Edinburgh University Press), 85-96. 1987

"The interpretation of ard marks", Antiquity 61, 263-6.

Rowley-Conwy, P. and Zvelebil, M. 1989 "'Saving it for latet", in P. Halstead and J. 0' Shea (eds.) Bad Year Economics (Cambridge University Press), 40-56. Russell, K.W. 1988 After Eden: the Behavioral Ecology of Early Food Production in the Near East and North Africa (BAR Int. Series 391; Oxford: British Archaeological Reports). Sallares, R. 1991 The Ecology of the Ancient Greek World (London: Duckworth). Sherratt, A.G. 1981 "Plough and pastoralism", in I. Hodder, G. Isaac and N. Hammond (eds.) Pattern of the Past: Studies in Honour of David Clarke (Cambridge University Press), 261-305. Sigaut, F. 1975 LJAgricultureet le Feu (Cahiers des ~ t u d e Rurales s 1; Paris). Spurr, M.S. 1986 "Agriculture and the Georgics", Greece and Rome 33, 164-87.

Halstead

Plough and power

Vanstiphout, H. 1984 "On the Sumerian disputation between the hoe and the plough, Aula Orientalis 2, 239-51. Ventris, M. and Chadwick, 3. reek;) ~ u(Cambridge ~ University Press). 1973 Documents in M ~ c ~ M Vergopoulos, K. 1975 To Agrotiko Zitima stin Ellada (Athens: Exandas). Wagstaff, M. and Augustson, S. 1982 "Traditional land use", in C. Renfrew and M. Wagstaff (eds.) An Island Polity (Cambridge University Press), 106-33.

HOW CAN MODERN FOOD TECHNOLOGY HELP TO IDENTIFY DAIRY PRODUCTS MENTIONED IN ANCIENT SUMERIAN TEXTS ? Michael Teuber (Laboratory of Food Microbiology, Institute of Food Science, Swks Federal Institute of Technology)

White, K.D. 1970 Roman Farming (London: Thames and Hudson).

In principle two routes are available to us: 1. We can analyse and study products made by nomads in that area and compare the properties of these products with the ones mentioned in the ancient texts (Anderegg, 1894; Mair-Waldburg, 1974). 2. We can try to prepare such products in the laboratory to find out whether the necessary conditions coincide with conditions presumably prevailing in Mesopotamia 3000 to 5000 years ago. The reason why we can be quite confident about this approach is the fact that the biochemical composition of milk (of different species) has not significantly changed since that time. In addition, the scientific progress during the last 100 years in dairy science has yielded so much experience and knowledge that we are able to predict (even retrospectively) what happens to milk and its components when specific actions are taken, e.g. spontaneous fermentation with lactic acid bacteria, heating or boiling, churning, draining of whey (milk serum), ripening or drying of curd (Teuber, 1993; Teuber et al. 1987). First I shall give a brief explanation of the properties of milk. The composition depends on the animal species and is given in Table 1. The main chemical fractions determining the biochemical and nutritional properties are accordingly:

1 Water (about 85 to 90%)

2 Protein (3 to 3.5%) The protein fraction is composed of 2 completely different kinds of molecules: 2.1 The caseins They are organized in micelles together with minerals (mainly calcium phosphate and potassium), and citric acid. The casein micelles (50 to 300 pm in size, about loi4 per ml) are in colloidal solution and stable at boiling temperatures. Caseins, however, are precipitated (coagulated and denaturated)

Bull. Sumerian Agric. 8 (1995)

Identification of dairy products in Sumerian texts

Identification of dairy products in Sumerian texts

Teuber

under acidic conditions (lactic acid fermentation or addition of acetic acid) or by the action of specific enzymes like those present in calf rennet or lamb stomachs.

Table 1 Composition of milk from different animal species (average values, undefined lactation phases)

2.2 The whey proteins They are globular proteins which are soluble under acidic conditions but denaturated (precipitated) at the surface of the casein micelles by boiling. For that reason, the coagulum obtained from boiled milk by lactic fermentation is much finer than the one obtained from non heated raw or butter milk. In addition, denatured whey proteins act as a kind of molecular glue at the surface of casein micelles to yield a very dense, almost glass like structure of the curd after drying. This is important for the structure and stability of dried curd (cheese). Curd can be separated from whey (milk serum) by draining (e.g. through cloth, sieves or riddles) to yield fresh cheese.

animal species

dry matter

%

%

beef goat sheep buffalo zebu

12,7 13.2 19,3 17,2 13,5 17,3 29,9 15,O 11,2 11,7 18,8 233 17.6 31,O

yak reindeer camel horse donkey

3 Fat Milk fat is organized in small fat droplets surrounded by a fat globule membrane composed of lecithin and proteins which render the fat globules water "soluble". When the fat globule membrane is destroyed by mechanical stress (e.g. by churning) the fat globules melt together into large lumps of butter which still contain about 15% of water. Boiling of butter separates the remaining water from the fat, yielding butter oil (which is called ghee in India). The boiled water-free butter oil is very stable and can be kept for many months, and even years.

4 Milk sugar (lactose, about 5%) Milk sugar is the component which is transformed into lactic acid by lactic acid bacteria. The concentration of lactic acid obtained in sour milk, sour cream or yoghurt (between 1 and 2%) is high enough to prevent the growth of pathogenic bacteria and of spoilage microbes. However, sour milk is still an excellent substrate for yeasts (Kefir!) and moulds (Roquefort and Camembert cheese). Yeasts and moulds can only be kept out of dairy products if the water content is reduced as in butter oil or dried cheese. Dried fresh cheese is stable for decades! The stability of cheese can also be increased by addition of high amounts of salt (sodium chloride). All classical and ancient techniques for the manufacture of dairy products use these few basic manipulations, however, in different combinations (see also Ryder 1993). The most important factor for the ancient dairy farmer was the desire and necessity to obtain a product which was stable enough to reach the market (consumer) and to have a preserve for the season of the year with low milk production. We can devise a technologydependent ladder of increasing stability against spoilage (at room temperature) as shown in Table 2. The only products which keep for much longer than about 2 weeks are hard cheeses, dried curd cheeses and butter oil. Whenever the ancient texts from Mesopotamia mention dairy products with presumable keeping times of more than two weeks we have to look for them and their identity within the short list given in that table. A strongly salted curd (at least 15% NaCl) could keep also for a few weeks. It seem questionable, however, whether nomads would be able to carry the necessary large amount of dry salt with them.

Teuber

pig dog cat dolphin

water

fat

casein

whey protein

lactose

%

%

%

%

%

87,3 86,8 80,7 82,3 86,5 82,7 70,l 85,O 88,8 88,3 81,2 76,5 82,4 69,O

3,7 4s 7,4 7,4 4,7 6,5 14,5 5,4

23 2,5 4,6 32 2,6

0,6 0,4 0,9 0,6 0,6 5,8* 1,7 1 ,o

43 4,l 4,8 4,8 4,9 4,6 32 5,1 62 7,4 5,5 3-1 4.8 0,6

19 1,4 6,8 12,9 43 18,O

8,l 2.9 1,3 1 ,o 2,8 53 3,7

-

12 1 ,o 2,o 2,l 3,3 9,4*

ashes

* : casein + whey protein

Table 2 Keeping times of dairy products

I

I

keeping time

1 milk and its protein fraction I

112 day

raw milk

1 day

boiled milk

1 week

!

sour milk

2 weeks

sour curdJfresh cheese

2-3weeks

soft cheese

1-2years

hard cheese

several years

dried cheese

fat fraction sweet cream

!

sour cream butter

butter oil

I

Identification of dairy products in Sumerian texts

Teuber

General procedures for the preparations of nomadic cheeses Very useful details on indigenous cheeses of the Middle East, their properties and production technologies can be be found in three recent publications by P.F. Fox (1993), Hamid A. Dirar (1993), and Abou-Donia (1978). Recent accounts of nomads still living in the Middle East are given by Baum (1989) and Janzen (1980). For the preparation of ripened soft and especially hard and semi-hard cheeses, we need precise timeltemperature regimes and management which probably were not available in ancient Mesopotamia due to climatic conditions. I believe that we can therefore strike such products with a complex production technology from our main considerations. However, fresh cheese curds stored in skin bags from goats and sheep may have ripened to Roquefort-like soft cheeses, e.g. in Anatolia, or in the Zagros mountains (i.e. under temperate climatic conditions with lower temperatures and air humidity than in the lower Euphrates and Tigris valleys). The substrates (raw materials) for nomadic cheeses can be either whole milk, skimmed or partially skimmed milk, whey or buttermilk. The milk can be mixed milk if different species of animals compose the flocks and herds.The following procedures are documented (see Fox, 1993, and Dirar, 1993): A.

1st step: 2nd step: 3rd step:

B.

1st step: 2nd step:

3rd step: 4th step: 5th step: C.

1st step: 2nd step: 3rd step:

Coagulation of milk with the aid of rennet (from young animal stomachs) Draining of whey through cloth with concomitant spontaneous lactic fermentation during coagulation and drainage. Resulting fresh cheese consumed directly or dried (with or without addition of salt and spices) Condensation of whole milk or butter milk by boiling. Coagulation of boiled, condensed whole milk or butter milk by thermophilic lactic fermentation (e.g. by using yoghurt culture), yielding sour rnilk or yoghurt. Removal of butter by churning if whole milk was used as raw material Draining of whey e.g. with cloth or through leaf mats Drying of formed cheeses Coagulation of whole boiled andlor condensed milk by fermentation as above under B Removal of whey by decantation or evaporation Addition of new milk until vessel is filled completely.

Hypothesis that the sign GA'AR in the ancient texts from Uruk I Warka stands for dried cheese This characteristic sign has a relation of the long side to the broad side of 2:l. The dot in the middle has been proposed by Nissen (as cited in an elaborate treatise by M. Stol, 1993) to

Identification of dairy products in Sumerian texts

'

Teuber

symbolize a string which has been used to hold small pieces of cheese. For reasons of mechanical stability these have to be dried cheeses. In the collection of our Institute made by the late Prof. Marc Bachmann we have a chain of small dried cheeses assembled on a string of 1 m length. The 18 rectangular small cheeses have a mean size of 15 x 18 x 35 rnm and a weight of about 5 g each. Under good manufacturing conditions, this amount of cheese represents the protein content of between 1 and 1.5 liters of cow's milk. This cheese-chain was given to Mr. Bachmann by nomads in Afghanistan about 25 years ago. The exact mode of manufacture is not known, in particular we do not know how and when during the production process the string was put through the cheeses. However, the cheese has most probably been produced from mixed milk (sheep, goats, cows and maybe even camels). The general appearance of the cheeses is close to Krut (or Qurilt) which is made by boiling buttermilk, draining of whey through cloth, kneading of the curd, drying in tablets, and cutting into the final size with a sharp knife giving sharp edges. I tried to reproduce such cheese in the laboratory. We are able to produce concave cheese surfaces, resembling the sign GA'AR as outlined in Fig 1. The 18,120 units of GA'AR mentioned by Nissen et al. (1990) when similar in size to the ones in the Afghanistan cheese chain would represent the yield from about 1,120 liters of rnilk coming from about 3 cows. This number could be supported by the amount of butter fat mentioned on the same tablet. It seems to be no problem to record the whole yield of a cheese season together in this form. In Fig. 2, I present photographs of the mentioned cheese chain from Afghanistan (2C), dried cheese curd from Afghanistan (2A) which probably was made as described by AbouDonia for Ekt cheese, and a cheese chain made in our laboratory from boiled buttermilk (2B) as outlined in Fig. 1. That the measures of GA'AR changed from number to volumes in later times (see Stol, 1993) could be an indication that production of cheese by sedentary farmers changed from production mode A or B (as practised by nomads) to production mode C. The mode C was still in use in Bavaria at the turn of this century to secure protein preserves for the wintertime ("Herbstmilch"). Production mode C was possibly the forerunner of the traditional ripened soft a d hard cheeses now so common in Europe and elsewhere.

Note added in proof In spring 1995, a PhD student of our University, a native from Bhutan, showed me a video he produced in the summer of 1994 in Bhutan when he observed a tribe of yak breeders and nomads in the Himalaya valleys at about 4000 m altitude. Dried cheese as shown in Fig. 2 C is a main product and source of income. About 1 to 2 litres of milk are taken from cows while their calves are kept in front of them (as in the famous El-Obed frieze). The fresh milk from several cows is put into a wooden barrel and beaten with a properly equipped stick from above. The butter is removed and placed into leaves until locally consumed. The skimmed milk is placed into a metal vat and slowly heated on an open fire to about 80" C. The milk is slowly stirred and several cups of whey from the previous day are added to aid acidification by lactic

Identification of dairy products in Sumerian texts

Teuber

fermentation. When the curd formation is completed, the coarsely coagulated milk protein is collected in a cheese cloth. Whey is removed by kneading. Surplus whey is given to the "yak dogs". The lump of curd obtained is formed into a square tablet and then pressed in the cloth over night between two milling stones. From the roughly square and flat piece of still moist cheese two kinds of products are made:

How to produce concave dried cheese pieces

1. By cutting the tablet into 4 square pieces, larger cheese loafs are obtained which are dried 4 on a rope by putting the rope through the middle of the loaf. A small wooden stick knotted into the string below each loaf keeps them in place and apart from each other. 2. By splicing a curd tablet along the flat side and cutting the obtained thinner tablets into smaller pieces like the ones seen in the cheese chain from Afghanistan (see Fig. 2 C), a similar product is obtained by putting 20 small cheese pieces with a needle onto one rope. Both cheese types are dried by these nomads in the air for several weeks by hanging them between the poles of their tents. Before sale, cheese chains are briefly dipped into boiling milk to provide for a white surface. Bhutanese nomads use the small cheese pieces mainly as a snack when climbing in the mountains. A dried piece of cheese is kept in the mouth like a candy and slowly dissolved by sucking. Cheese chains are also commonly used as a gift for visitors. Export of dried cheeses to India is of prime economic importance. During his summer term in Bhutan, our student will investigate and record the exact details of cheese making so that we will be able to calculate the exact yields of butter, cheese and whey in order to compare it with the numbers known from the ancient texts from Uruk.

Teuber

Identification of dairy products in Surnerian texts

1. cheese curd tablet 2.drying of the tablet for a limited time

-

to harden surface

3. cutting into small pieces

/p

4. pulling a thread through the middle along the

large axis 5. further drying leads to concave edges of the still soft curd

8

Fig. 1. Outline for the production of dried cheese curd pieces with concave edges

Identification of dairy products in Sumerian texts

Identification of dairy products in Sumerian texts

Teuber

BIBLIOGRAPHY Abou-Donia, S.A. 1978 'Microbial flora and chemical composition of Ekt cheese", Egypt. J. Dairy Sci. 6,49-52 Anderegg, F. 1894 Allgemeine Geschichte der Milchwirtscha# (Orell Fiissli, Ziirich) Baum, E. 1989 Nomaden und ihre Umwelt im Wandel (J. of Agriculture in the Tropics and Subtropics. Beiheft 38. Witzenhausen). Dirar, H.A. 1993 The indigenous fermented foods of the Sudan - A study in African food and nutrition (CAB International, WallingfordIOxon) (The chapter on dairy products contains descriptions of the composition and production technology of cheeses similar to nomadic cheeses of the middle east. The product kush-kush is obtained from fermenting sour milk by skimming the curd coming to the surface by strong gas formation. The curd is sun dried on palm leaf mats either as powder or after moulding into balls. Such balls are known from Akkadian seals as recently described by P. Steinkeller, "Early Semitic literature and Third Millenium Seals with mythological motifs", in Literature and Literary Language of Ebla (Quademi di Semistica 18; Florence 1992), p. 249). Fox, P.F. 1993 Cheese: chemistry, physics and microbiology. Vol. 2 Major cheese groups (2nd ed.; Elsevier, London). (with chapters and remarks on non-European varieties including examples from India, Afghanistan, Egypt, Syria, Saudi Arabia, North Yemen, Iraq, Israel, Turkey and Cyprus). Janzen, J. 1980 Die Nomaden DhofarsBultanat Oman - Traditionelle Lebensfonnen im Wandel (Bamberger Geographische Schriften Heft 3; Bamberg). Mair-Waldburg, H. 1974 Handbuch der Kase. Kase der Welt von A-2, eine Enzyklopedie (VollcswirtschaftlicherVerlag, Kempten). Nissen, H.J., Damerow, P., Englund, R.K. 1990 Friihe Schrift und Techniken der Wirtschaftsvenvaltung im alten Vorderen Orient - Informationsspeicherung und -verarbeitung vor 5000 Jahren (Verlag Franzbecker, Berlin). Ryder. M.L. 1993 "Sheep and goat husbandry with particular reference to textile fibre and milk production", BSA7,9-32. Stol, M. 1993 'Milk, butter, and cheese", BSA7, 99-1 13. Teuber, M., Geis, A., Krusch, U., Lembke, J., Moebus, 0. 1987 '%iotechnological Processes for the Manufacture of Foodstuffs and Fodders", in P.Prave et al. eds., Fundamentals of Biotechnology (Verlag Chemie, Weinheim) pp. 325-379. Teuber, M. 1993 "Lactic Acid Bacteria", in H. Sahm, ed., Biotechnology, Vol. I , Biological Fundamentals (2nd ed.; Verlag Chemie, Weinheim) pp. 325-366.

Fig. 2. Photographs of dried curd cheeses from Afghanistan (2A), cheese chain made in our laboratory from boiled condensed buttermilk as outlined in Fig. 1 (2B), and cheese chain from Afghanistan (2C) collected by the late Professor Marc Bachmann of the Institute of Food Science of the ETH Ziirich.

LATE URUK PERIOD CATTLE AND DAIRY PRODUCTS: EVIDENCE FROM PROTO-CUNEIFORM SOURCES Robert K. Englund (Free University of Berlin)

Introduction Of the two major uses of large cattle we may expect to find documented in proto-cuneiform sources, namely their exploitation as draft animals and as producers of meat and dairy fats, the former, in particular plowing, is not recognizable in its clear form known from administrative texts of the later third millennium. No texts known to me refer to numbers of oxen specifically assigned to plowing units, nor do the sources attest to the substantial costs-feed grainknown to arise with the use of oxen in plowing and seeding fields. Only several uncertain accounts register together the existence of both the plow represented by the sign APIN and oxen represented by the sign Gu4.1 Whether oxen played a large role in field work in the Late Uruk period is thus a matter of conjecture. It seems unlikely that large cattle were widely exploited as producers of meat any more in the Late Uruk period than in later times.2 As sources of dairy fats and cheese, however, cows were clearly prized and closely controlled. A relative abundance of documentation registers quantities of these products, from which herd sizes can be extrapolated that are comparable with those of the pre-Sargonic Lagash period, but that fall well short of the numbers of animals herded during the IIIrd Dynasty of Ur.3 Proto-cuneiform documentation of early cattle herding derives with few exceptions entirely from the large numbers of tablets and fragments unearthed during the German excavations of the southern settlement of Uruk.4 This Uruk documentation exhibits a clear dichotomy between texts dating to the Uruk IV (ca. 3200-3100 B.C.) and those dating to the Uruk IIIIJemdet Nasr (ca. 3100-3000 B.C.) periods. The earliest texts record numbers of cattle apparently assigned named officials or institutions, to the near exclusion of records of dairy produce, whereas among the texts dating to the Uruk I11 period, exceedingly few accounts of groups of cattle are found, but large numbers of records of dairy fats and cheeses, complemented with the existence of an involved metrological system seemingly developed to afford greater control of these products.

Identification of proto-cuneiform signs representing cattle Following the publication and analysis of a small number of Late Uruk documents between 1917 and 1927,s S. Langdon's work on the Uruk I11 period proto-cuneiform tablets from Jemdet NasI.6 resulted in 1928 in the first identification of large numbers of archaic signs with apparent later forms. It was readily obvious that in the case of large cattle pictograms representing the animals' heads, and, pars pro toto, the animals themselves, were the precursors Bull. Sumerian Agric. 8 (1995) 35-50

33

Englund

Late Uruk Cattle

of signs which in later periods represented animals differentiated according to sex and age using ideographic qualifications. Based on the primarily Uruk IV period texts excavated by German teams in Uruk in the years 1928-1931,7 A. Falkentstein noted in 1936 the paleographic development, beginning in the Uruk IV period, of the signs AB2, GU4 and A M A R . ~ Recent considerations about the possible semantic significance of a small number of so-called tokens found both alone and in association with sealed clay bullae, dating to the Late Uruk period prior to the development of writing,g may be considered highly speculative; one can only presume that there is no reason why such semiotic devices should not have been used together with tokens of obvious numerical significance, however the forms which were unearthed next to-none within-bullae in Uruk bear no more than a passing resemblance to the later signs GU4 or possibly AMAR. The Late Uruk sign AB2 depicted, seen from the front, the head of the domesticated female Bos with down-turned horns, the sign GU4 the head of the bull or ox with horns upturned,lO and the sign AMAR the hornless calf with ears held upright. This latter sign, which otherwise did not indicate the sex of the animal, could be qualified in the second writing stage Uruk I11 with the signs SAL and KUR, representing females and males, respectively.1 1 In the course of Uruk excavations following the publication of ATU 1 in 1936, a total of 24 Uruk I11 period fragments representing witnesses of an archaic lexical list dealing with large cattle were discovered and have been published in a volume containing all known archaic lexical material from Uruk and other sites.12 Although in a very poor state of preservation, these witnesses did allow a sufficient reconstruction of the text to confirm early suspicions that it represented the original version of a list which in a possibly redacted form was part of the scholastic curriculum in Early Dynastic Fara, Abu Salabikh and Ebla.13 The compendium contained in its archaic form three clear sections dealing in turn with AB2, GU4 and AMAR; each section consisted of a canonized sequence of sign combinations beginning with entries recording 'stall', hide colors ('reddish', 'white', 'black') and other both standardized and specific characteristics of the respective animals concerning above all their hides. A fourth section partially preserved in only one text consists of animals represented by the sign AM, (AMARgunfi), possibly corresponding to later am (GU4+KUR),"wild bull".14 A section of the well preserved archaic 'tribute' list-in all likelihood an early form of folkloristic literature-contains among standardized sequences of animals and agricultural products the notations INl4 AB2 I IN1 GU4, "10 cows, 1 bull". The same relation of 10:l is found in two following couplets of the text registering female and male sheep and goats, thus listing together the three major domestic animals in the early Babylonian economy.15

Records of heads of cattle In general, accounts of numbers of large cattle are well represented in texts dating to the period Uruk IV,l6 but are essentially unknown in texts from the following Uruk I11 period. The dairy products milk fat and cheese, on the other hand, are well attested in accounts from the latter period, whereas such accounts are seldom found in Uruk IV texts. Cattle as discrete objects were as a rule registered in proto-cuneiform texts in the sexagesimal system; the exceptional use of the sign N8 in the Uruk IV period to designate immature animals is discussed below.

Ikglund

Late Uruk Cattle

Small, characteristically pillow-shaped Uruk IV period tablets17 record the receipt by a named individual of one or as many as several head of cattle (see figure 118). Inscriptions in these accounts consist of numerical notations employing the sexagesimal system, one or more signs representing heads of cattle and one or more signs which seem to represent receiving individuals or officials and possibly, for example, in the case of the signs ERIM, and A L , ' ~the function of the animals.20 Numerical notations representing numbers of both male and female animals were explicitly qualified by the sign combination GU4 AB2 as in later periods both for large cattle and, using a related sign combination, for sheep and goats.21Reverse faces of the 'receipts' remained uninscribed. Simple records of single transactions were booked into larger accounts in a format represented by the two tablets W 9656,ev and W 7227,a in figure 2. Up to five columns on the obverse face of these texts contained from four to seven individual entries, each of which corresponded to one of the simple records discussed above. The numerical total of the cattle recorded in these entries was entered on the reverse face of the account (rotating the tablet around its 'horizontal' axis).22 This total was qualified with the ideograms GU4 or GU4 AB2, respectively, representing cattle and, in the case of the former text depicted, with a further sign combination possibly recording the distributing organization (? NUN,) and the purpose of the distribution (? 'GI' [ I). A third account in figure 2, W 9656,ex, demonstrates the use in the Uruk IV period of a bookkeeping mechanism which recorded an apparently complete herd of adult and young cattle probably separated according to the function of the individual animals. Employing the principle of listing most to least valuable, the text registers numbers of oxen, followed by counted cows and calves (AMAR), in some cases assigned named individuals. In accounts from the Uruk IV period, the calves could, just as is true of lambs and the children of dependent laborers who were probably too young to be put to work, be qualified using the sign N8 (o,that is, the basic sign N1 turned 90" clockwise) which in sexagesimal notations generally designated "112" of a discrete unit.23 Thus the fourth case of the text's second obverse column contains a notation 'Nll N8 representing one cow and one calf; the latter animal was included on the text reverse, column 1, case 3, among a group of four animals qualified as AMAR. Uruk I11 period accounts of herds of large cattle are very rare and register only modest numbers of animals. The preserved sections of text W 14275 in figure 2 contain notations representing just 8 head; it is just the same an interesting example of the correspondence in the archaic period between designations attested in the lexical cattle list and those attested in administrative documents. The sign combinations AB2 NE, (obv. i I), AB2 U4 (i 2), AB2 SU (ii I), AB2 GI6 (ii 2) and GU4 'NE,' (ii 4) in this account, all known from the canonized lexical text ATU 3, 89-93,24are found neither in accounts from the preceding Uruk IV period, nor in the administrative corpus of the following Early Dynastic period, including the age represented in the SIS 4-8 texts from Ur,25 during which the lexical list in its archaic form was part of scribal curriculum. Recorded categorization of animals according to age seems to have been more developed in the Uruk I11 period as well; the text W 14361 (figure 2, bottom right) registers in three cases of its second column notations representing oxen in their fifth, fourth and second years, respectively (sign combinations U4+5NS7,U4+4N57and U4+2N57G U ~ ) . ~ ~ These oxen were apparently in the charge of an official called GAL, UTUL, (obv. iii 2), "head cowherder".

Late Uruk Cattle

Englund

Examples of receipts (?) for one cow (sign

Englund

Late Uruk Cattle

@)

Examples of receipts (?) for one or at most several bulls/oxen (sign& fl

Examples of receipts (?) for one calf ( s i g n p ) and for mixed cattle (?; sign combination

W 9579,da

W %56,in

Figure I: Examples of receipts (?)for one or at most several cattle (all copies from R. Englund, ATU 5).

@)

W 9335,~

Figure 2: Examples of complex accounts of cattle. The upper three texts (copiesfrom R. Englund, ATU 5 ) date to the Uruk IV period and register numbers of animals totaled on the texts' reversefaces; W 7227,a books a total of 54 cows and bulls. The lower two Uruk 111period texts contain accounts of small numbers of cattle qualified with signs known from the lexical list ATU 3, pp. 89-93 (W 14275), and with sign combinations known to represent the animals' ages (W 14361; Jirst published in ATU 2, pl. 60).

Englund

Late Uruk Cattle summatfion

(sign DUGb) per two (W 20274,12) or four (W 20274,63) milk cows. We may compare these relationships with the amount of the dairy fat required as a yearly delivery of herders from the pre-Sargonic Lagash (ca. 2400-2300 B.C.) and the Ur I11 (ca. 2100-2000 B.C.) periods. Numerous accounts from the earlier period set this quota at 10 Old Sumerian sila or ca. 15 liters, accounts from the later period on the other hand assume a yearly delivery of just 5 neoSumerian sila or ca. 5 liters. If we assume a yearly milk production of reasonably well fed cows kept in semi-arid climates of between 700 and 800 liters, of which about half will have been given to their calves, the remaining milk with a fat content of ca. 4% would result in ca.

2 cows (reconstructed)

I heifer calf

b~rthof l calffw cows)

D~D

BQ ry S

designation of calf as "yearling"

Late Uruk Cattle

1:nglund

responsible herdsman

summation

1 heifer calf

4 cows

1 bull calf

"dairy fat"

responsible herdsman W 20274,63

Figure 3: Two Uruk III accounts of small herds of milk cows. The sign combinations in the lower left case of each account represent the herders responsible for the cows.

Records of archaic dairy products Related herding accounts from the Uruk I11 period, of which only two are preserved well enough to permit a reconstruction of their c0ntents,~7record small numbers of cows together with their offspring, qualified SAL+AMAR and KURa+AMAR ('heifer calf' and 'bull calf'), from the accounting year of the text. Both texts record a ratio of two adult cows per recorded calf.28 The two accounts depicted in figure 3 further serve as examples of the greater accounting effort invested in the dairy production of cows in the Uruk I11 period. In a fashion parallel to that employed in recording the amount of dairy fat to be delivered by herders of sheep or g0ats,~9the accounts book in the totals on their reverse faces one jar of dairy fat AB, 'cow'

6@@

GU, 'hull', 'ox'

AMAR 'calf

SALAMAR 'heifer calf

KURAMAR 'bull calf 2~7,200+3~1,200+1~120 = 18,120 units of

D

Figure 4: Proto-cuneiform signs representing cattle.

Figure 5: Simple accounts of dairy productsfrom the Uruk Nperiod (above; copiesfrom R. Englund, ATU 5 ) and a large account from the Uruk Illperiod (below; see ATU 2, pl. 55, and Archaic Bookkeeping, p. 94) of products from animal husbandry, including the signs for dairy fat (DUG,,) and cheese (GA'AR).

Englund

Late Uruk Cattle

Englund

15 liters of dairy fat ('clarified butter', 'ghee'). This suggests that the pre-Sargonic Lagash herders delivered all the produce of their cows, the Ur I11 herders in contrast perhaps a third. Since, moreover, the archaic sign DUGb seems to have represented a ceramic jar no larger than ca. 10 liters (see below), it may be proposed that the archaic herders delivered dairy produce at a rate comparable to that of the Ur I11 period and were thus akin to contract herders who kept a percentage of the produce of their herds in lieu of other compensation for their labor. The first eight lines of the archaic lexical list 'Vessels,' one of several compendia for which precursors from the Uruk IV period are known, in fact consist of entries with the signs DUGb, KISIM* and other signs which represent containers of fats used in the administration of ar, ~ ~often found inchaic dairies.30 These signs, including NI,, DUG, and U K K I N ~ + N I ,are scribed together in administrative documents beginning in the Uruk IV period (see figures 58). In particular, the association of the sign NI, with DUGb in such documents as W 9206,c, W 9579,ad and W 9579,ah, and of NI, in the same case with AB2 and with DUG, in the account W 9656,eq (all figure 5) demonstrate that this sign should represent a container of dairy fat from its first use in the Uruk IV period. The sign, the real referent of which is unknown, is in later cuneiform documents the general designation of fats of all kinds. Only indirectly associated with the sign representing dairy fat, DUGb, is on the other hand the sign GA'AR in such texts as W 20274,97 (figure 5). This sign, found as a general object designation in a section of the archaic vessels list following a long section on containers of fats and other products,32 is, as a clear precursor of the Fara and pre-Sargonic Lagash sign LAK 490-itself replaced in Ur I11 documents by the sign combination ga.kJARIUDgunli-, posited to represent a unit of cheese (Sumerian reading possibly lga'aral or Igamurl). Whereas oil vessels were counted with the sexagesimal system, GA'AR was reckoned in discrete units using the so-called bisexagesimal (120-base) system and so may be associated with the products GAR. (dry grain products) and KU6 (fresh? fish). We consider such objects to have formed a part of the archaic rationing system.33 Unfortunately, no archaic texts known to me document the arithmetical relationship between dairy fat and cheese well attested in preSargonic Lagash and Ur I11 accounts.34 Containers of dairy fat and other liquids were not only as discrete objects counted using the sexagesimal system, but were also as members of a liquid capacity metrological system divided into smaller units using one of three numerical conventions. In the first place, the sign N8 (T), discussed above as a designation of immature cattle in the sexagesimal system, as a rule qualified 'I2 of some discrete unit, above all the contents of vessels and baskets, including those represented by the signs DUG, and KAS, (beer), KISIM*, DUGb, DUG, and DUG+AS, (fats) and GA2,+kJI (a fish product).35 Notations in a number of Uruk IV period texts suggest that the sign N8 in the sexagesimal system could also represent a smaller fraction than 'I2 of an object, possibly lIlo; the objects so qualified in these notations are, unfortunately, not always clear, although DUG, seems attested in at least two of the accounts.36 A second, Uruk I11 period convention used explicitly and alone for dairy fats seems on its surface substantially more complex, yet shares the basic structure of 'I2 and 'Ilo of the unit 'container.' A large number of accounts, of which nearly all derive from the same locus W 20274 as the herding texts described in figure 3 above, contain notations in this metroloFigure 6: Accounts concerning dairy fat stored in the jar DUG,.

Late Uruk Cattle

Late Uruk Cattle

Englund

gical system which exhibits the structure lx N1 vessel = 2x Nl+KU3a= 1 0 N2 ~ (corresponding to the basic unit N1 crossed by a horizontal stroke: B; see figure 9). W 20274,6 in figure 6 offers a simple summation of three entries with numbers of a container of fats represented by the sign DUG:^: 7N1 DUG, IN1 KU3, + 4N1 (DUG,) 1 KU3, + 5N1 (DUG,) = INl4 7N1 DUG,, that is, 7 112 + 4 112 + 5 = 17. The only known duplicate administrative texts from the archaic text corpus, W 20274,33 and W 20274,89 (see figure 6 and the transliterations below), contain a somewhat more involved account, yet the reckoning steps exhibited by both are easily recognizable as simple additions of whole numbers and fractions from the same metrological system.38

Obv. i l a lblal lbla2 lblb lb2 2 3 4

5 ii 1

2 3 4 5 6 Rev. i 1 2 3 4

5

6 ii 1

'5N11 ; 'DUG,' DUB, lN1 ;BA AN ZABALAM, lN1 ; SAL BA PINGbl ZATU75 1, DUR2 3N5, 'BU,+DUQ 3N1 ;BA 'KI; IN1 I lN1 KU, ;ZATU649 lN1 'KU3: ;'ZABALAM; AB2 lN1 ; SI,, NE, SE, GI GA, MUD NUN, 2N1 ;ZATU648 ABb P1NGbl+3N5, SU, 3N1 ; SI U4 'SIC? 2N1 ;TUR,, A GA, BU, 'Ijr BA GI+GI ZATU648

5N1 ;DUG, DUB, lN1 ; BA AN ZABALAM, lN1 ; SAL BA PINGbl ZATU75 1, DUR2 3N5, BUa+DU6 3N1 ; BA KI, lN1 I lN1 KU, ; ZATU649 lN1 KU,, ; ZABALAM, AB2 lN1 ; SI,, NE, SE, 'Gr 'GA; [MUD NUN,] 2N1 ;ZATU648 PIRIGb1+3NS7ABb SU, 3N1 ; SI U, SIG 2N1 ;TUR3, A GA, BU, HI [BA GI+GI ZATU6481

2N1 ; ZATU648 '3N11; 'SI U4 SIC? '2N11;'TUR,, A' lN1 I lN, KU, ;ZATU649 IN1 KU3, ; ZABALAM, AB2 lN1 ; SI,, NE, 'Gr SE, 'INl4 3N11;DUG, GI+GI BA

[2N1 ;ZATU6481 '3N11; SI 'U4 SIG' 2N1 ; TUR,, A lN1 I lN1 KU3, ; ZATU649 lN1 KU,, ; ZABALAM, 'AB,' lN1 ;NE, SI,, GI SE, INl4 3Nl ;'ZATU648' DUG, GI+GI BA

An example of an account containing the full structure of the second conventional system used to record fractions of containers of fat is found in figure 7. This and a small number of other accounts clearly demonstrate the relationships N1 KU, = If2, and N2 ( B ) = lIl0 N1 DUG, in this system39 and thus make plausible the assumption that it may represent a development from the Uruk IV system with, dependent on context, N8 equal both to N1 KU3, and to N2. The meaning of KU3, in this connection is, aside from the fact that it indicated a half measure, not obvious.40 The third means of designating fractions of oil jars is found in just one text, presented in figure 8, but here fully documented. W 21682 contains on its obverse face two columns with 5 entries, each of which consists of the numerical sign N1 together with the sign combinations S1LA3,+GARA2, or SILA3$GAa-the former explicitly written in the first four cases of the first column, the latter probably only in the lost first case of the second column-representing

Late Uruk Cattle

linglund reconstructionof the summation:

obverse

reverse I

memlogical relations:

ID$,

=

.D(

Figure 7: Reconstructed account of dairy far stored in jars demonstrating the metrological relations in the system DUG,.

units of a dairy product, the sign SI (meaning unknown) and further ideograms probably representing receiving individuals. The meaning of both signs GARA2, and GA, is not certain. Clearly, they represent vessels and are both invariably in context with dairy products. The former sign, a gunified variant of the sign DUG,, is attested in the archaic Ur (ED 1-11) version of the lexical list L6 A, 1. 20, as a

,,, >s

standardized vessel SILA, (pro@ly a Blumentopf wlth ca. 0.8 liter capacity)

Bd> E

relations between the standardized vessels:

,>

product

5 units

>s

+,

=I

grand total of the dairy products:

1 unit

S;?dgGy

Figure 8: Metrological relationship between SILA, and DUG,.

disbursement

>P

Englund

Late Uruk Cattle

variant of GA, in the combination GALa GARA2,, "head of GARA2,", and representing a product among notations for domestic animals and other agricultural products in the list ' T r i b ~ t e ' The . ~ ~ Uruk IV period form of the latter sign (see figure 10 below) is apparently the representation of a flat basket, the inner surface of which was probably coated with bitumen so that the basket could be used in the milking of dairy animals. The reverse face of the tablet contains in the first column (to the right) subtotals of each of the obverse columns, numerical notations representing five units qualified by the sign combinations SILA3,+GARAZaand SILAJa+GAa,in the second column the final total N1 DUGb qualified with SI and the sign GU7 (SAG+GAR = 'human' + 'rationing bowl'; the specific administrative meanings in the use in archaic texts of the two signs BA and GU7, both apparently qualifying distributions, are poorly ~ n d e r s t o o d ~representing ~) "disbursement". This equation proves that the sign SILA3,, to be identified as a pictographic representation of the mass-produced Blumentopf which followed and for some time in Late Uruk levels coexisted with use of the beveled-rim bowl GAR, assumed the same metrological function in the three systems discussed as Uruk IV period N8 and Uruk I11 N2, all representing a measure equal to 'Ilo of the amount of liquids or semi-liquids contained in some larger vessel.43 According to data derived from excavations, above all measurements conducted of the masses of beveledrim bowls found in Late Uruk settlements, and in accordance with textual analysis, the most plausible current working hypothesis of the absolute capacities of these various units is the following:44 GAR = IN8 (?; Uruk IV) = SILA,, = 1N2 1N8 = IN, KU,, IN, DUGb,, etc.

= ca. 4/5 liter

The sign DUGb, representing a ceramic jar without a spout, was consistently distinguished from the sign DUG, L, including the representation of a spout. N,, N , DUG, N, SILA,~ This fact and the contextual usage of -lo_ L both signs suggest that the former jar will most likely have contained semiliquids, the latter liquids, above all beers. A large number of signs were Figure 9: Metrological systems employed in dairy notations (cp. impressed in DUGh in archaic lexical Englund, Iraq 53 [I9911 101-4). The application o f the .. " upper system with dairy products is not proven, the lowtexts, to a lesser extent attested in ader two systems are only known from the Uruk Illperiod. -

N , DUG~NKKIN,+NI~

N, K U , ~

lhuk IV

sign name

~0 0

meantng

DUG*

beer

K A ~

beer

Uruk l V

Uruk Ill

srgn nome

00 00

meaning

DUGb

dairy fat

DUG.

dairy fat

5 ~ ~ 1 %dai, fat

p G

KISIMd

butter fat from sheep's milk

P E P

KISIM,

butter fat from goat's milk

0,&p

Uruk I V

Umk I l l

mme

0 0

dairy fat ?

b 8 $3

meanlng

milk ?

@@

dairy fat mixed with crushed barley?

BB>

GAR*,

irem ?

Figure 10: Probable archaic designations of liquid /semiAiquid products.

lninistrative texts, to specify the product contained in the jar represented by the sign, including among others SE ('barley'), NAGA ('an alkaline plant' ?), TI (?), MAS (male goat), KUR (a plant related to the grapevine ?), GIS ('wood'), KU6 ('fish') and S A ('pig').45 ~ ~ The sign K A S is ~ also one of these composita, consisting of dots impressed within the sign DUGb in its Uruk IV form, abstracted to vertical lines in the Uruk I11 writing phase. These impressed dots are in my opinion the same as those added to numerical signs from the grain capacity system (forming the system S*)46 to indicate amounts of barley groats used in the production of dry and liquid grain products.47 This K A S ~is also consistently distinguished from the related sign KAS,; whereas the latter sign is generally attested together with notations representing both rough-ground barley and malt, thus representing beer, the former is invariably found in a dairy context.48 The same applies for the sign KAS, (see figure lo).@

NOTES TO TEXT 1

2

N,

D>

Uruk I l l

= ca. 4 liters = ca. 8 liters

These measures would, moreover, suggest a yearly delivery in the Uruk I11 period of from two to four liters of dairy fat per milk cow (see figure 3 above) and be in general accordance with figures known from the Ur I11 period. With an expected yearly delivery of between 'I4 and 112 DUGb per cow-year, the notation of '26 DUGb' in W 20274,97 obv. i 1 (figure N8 5 ) , understood as part of an annual acNI4 N, 2 count, would correspond to ca. 50- 100 a - lo D -4,. N~ milk cows, or total herds of perhaps U 100-200 head. N,,

Late Uruk Cattle

I:nglund

3

See, for example, the two texts R. Englund, Archaic Administrative Texts from Uruk: The Early Campaigns (= ATU 5; Berlin 1994) pl. 86, W 9656,f, and pl. 100, W 9656,dr, with counted APIN GU, apparently assigned to temple households. The inscription of the latter text is duplicated in the second column of the obverse of the former. The very meager bone remains from Uruk of Bos taurus identified by J . Boessneck, A. von den Driesch and U. Steger, BagM 15 (1984) 170-172, were almost entirely of adult animals. Although the authors believed the crushed remains indicated the exploitation of cattle for meat, the numbers of bones-only 30 of the 73 specimens were from Late Uruk levels-permit no more than speculation as to whether the animals were selected for meat or were simply butchered in old age or after having died from some other cause. A general idea of the sizes of state-controlled herds in the pre-Sargonic Lagash period is offered by the text DP 93, which registers a total of 119 adult and juvenile animals in the care of dairy herders (cf. K. Maekawa, Zinbun 15 [I9791 122-123) in the year Enentarzi 5. The numbers in DP 93 are confirmed by further documents recording the delivery quotas imposed on the herders of the dairy fat i fib ssb.gald6.a and the cheese LAK 490 (DP 273-276; Nik. 1,257,259; RTC 64; TSA 37; VS 14,89) on the one hand, the amount of feed distributed to the herders (Nik. 1, 68, HSS 3, 37, etc.) on the other. This suggests total holdings, including draft oxen, of ca. 250 head in the economic unit represented by the archive from Girsu. State-owned herds in the Ur I11 period were substantially larger; the account MVN 15, 108, books for a single dairy herd supervisor from Umma over 300 milk cows, from which may be extrapolated a total of close to 1000 cattle. The supervisor Atu was, moreover, one of possibly several such managers active in Umma (see also the numbers in the text SET 130). The archaic texts from Uruk discussed in the

Englund

4

5

6

7 8

9

10

11 12

13 14 15

16 17 18 19

20

Late Uruk Cattle

present study document cattle herd sizes of between 50 in the Uruk IV period and possibly 100-200 in the Uruk I11 period. While it is true that with a current count of 5000 texts and fragments the corpus from Uruk can be expected to contain the majority of all early accounting genres, the ca. 450 archaic texts now registered from excavations of other sites, above all Jemdet Nasr (together with some few texts from Kish numbering 244 tablets and fragments), and from the antiquities market, do exhibit a broad spectrum of features known from archaic bookkeeping, in some cases containing the best documentation of particular economic activities. Still, the certainly uneven coverage of household activities in the proto-cuneiform documents, resulting from the vagaries of excavations, is the more to be underscored given the relatively modest number of texts and fragments from the Late Uruk period. V. Scheil, "Quelques signes originaux de 1'Ccriture cunCiforme," RA 14 (1917) 93-94; A. Deimel, LAK (1922) p. 73, no. 2; F. Thureau-Dangin, "Tablettes ? signes i picturaux," RA 24 (1927) 26-29. The inscriptions on the archaic stone tablets republished by I. J. Gelb, P. Steinkeller and R. Whiting, OIP 104 (1989) as nos. 1-11 and 13, of which the Blau pieces (10-11) were first published in 1885, and that on the tablet published by S. Langdon in Excavations at Kish I (Paris 1924) 99-101 and pl. XXXI, are in my opinion to be dated to Early Dynastic 1-11. The Herbert Weld Collection in the Ashmulean Museum: Pictographic Inscriptions from Jemdet Nasr Excavated by the Oxford and Field Museum Expedition (= Oxford Editions of Cuneiform Texts 7; London etc. 1928). See now the edition by R. Englund and J.-P. GrCgoire, The Proto-Cuneiform Texts from Jemdet Nasr (= Materialien zu den friihen Schrifzeugnissen des Vorderen Orients [MSVO] 1; Berlin 1991). See the introduction to ATU 5, pp. 11-23. ATU 1, p. 52+5-53+1.Falkenstein drew particular attention to the different graphic forms of the two signs AB2 and SIG (first noted by S. Langdon in OECT 7 s.v.),the latter a tend-form of the sign U4 which seems to designate "evening." R. Englund, review of D. Schmandt-Besserat,Before Writing, in: Science, from 11 June 1993, p. 1671; P. Damerow and H.-P. Meinzer, "Computertomografische Untersuchung ungeijffneter archaischer Tonkugeln aus Uruk: W 20987,9, W 20987,ll und W 20987,12," BagM 26 (1995) 9. I am not aware of a graphic differentiation in archaic documents between breeding bulls and castrated oxen, both apparently = GU4 (the few bulls kept for breeding in pre-Sargonic Girsu were called simply gu4 hb, "bull of the cow"). See figure 4 for a table of the graphic forms which represented large cattle in the Late Uruk texts. R. Englund and H. Nissen, Die lexikalischen Listen der archaischen Texte aus Uruk (= ATU 3; Berlin 1993), esp. pp. 22 and 89-93. SF 81; OIP 99, nos. 25-27; MEE 3, nos. 12-17, pp. 47-50 ("Lista di animali A"); s. Th. Krispijn, JEOL 27 (1981-82) 47-53, and J. Krecher, OrAnt. 22 (1983) 179-189, to the syllabic version MEE 3, no. 62. See ATU 3,2241. See ATU 3, 25-29. The numerical relationship probably mirrors the number of males kept in herds for purposes of breeding. Incidentally, these accounts were without exception uncovered during the first Uruk campaigns, published as photos in A. Falkenstein's ATU 1 and subsequently as copies and indexed in my ATU 5. For an overview of the text formats employed in such accounts see M. Green, "The Construction and Implementation of the Cuneiform Writing System," Visible Language 15 (1981) 345-372, esp. 349-356. All tablet copies found in this and the following figures are represented at 75% of original size. See the texts included in figure 1, W 9579,cz (upper row) and W 9579,dd (second row). ERIM, may mean in this context "for the yoke," and since both W 9579,dd and W 9656,ak depicted in the second row of this figure contain the sign combination NESAG, GAR3, it may be reasonable to assume that the sign AL in the former text represented a function of the ox (GU,) recorded there (see also ATU 3, 91, to line 5 1: U4 AL X GU,, with no apparent correspondence in the ED texts, assuming MEE 3, p. 53,ll. 51-52 = ATU 3 , 91,ll. 50 and 52). While this is not the place to discuss the methods available for the isolation and identification of object designations, in the cases involved here of cattle, as against the other ideographic representations found in

1;nglund

Late Uruk Cattle

syntax-free archaic accounts, it may be stated that as a general rule signs representing counted objects are situated closest to the numerical notation, inscribed, insofar as this is discernible due to the existence of sign distortions caused by subsequent inscription, immediately after the numerical notation and before the impression of the accompanying ideograms. This is true both for receipts, for example, W 9335,c in figure 1, lower right, and in such large accounts as W 7227,a in figure 2. The largest attested total of adult animals is '54' contained on the reverse of W 7227,a in figure 2. Although the bundling step '60' was thus not crossed, there can be little doubt that this notation is from the sexagesimal system, since the only other candidate known from the early Near East used to count animals, the decimal system from proto-Elamite Susiana, was not used in Babylonia (see P. Damerow and R. Englund, The Proto-Elamite Texts from Tepe Yahya [= American School of Prehistoric Research Bulletin 39; Cambridge, MA, 19891 24 and 53-55 with footnotes) and since other animals and humans were demonstrably counted sexagesimally in the archaic period. See A, Vaiman, "Die Bezeichnungen von Sklaven und Sklavinnen in der protosumerischen Schrift," BagM 20 (1989) 121-133, and the comments of P. Damerow and R. Englund, BagM 20, 137-138. See above, notes 12-14. The sign combination AM, KUR, of obv. ii 3 may bear some relationship to the presumed AMb section of the archaic cattle list cited above. E. Burrows, Archaic Texts (= UET 2; London 1935). For a description of archaic designations of years and animal ages see ATU 2, 146, and R. Englund, "Administrative Timekeeping in Ancient Mesopotamia," JESHO 31 (1988) 140-148 and 156-162. The standard age sequence for Ur I11 bullsloxen attested, for example, in the theoretical account TCL 2, 5499 , --), mu.2, mu.3, gu4 gal, "milk (I. J. Gelb, JCS 21 [I9671 64-69), was gu4 amar ga, gu4 m u . l ( ~ S sign bull-calf, one-year bull, two-year bull, three-year bull, large (full-grown) bull." First published by M. Green, "Animal Husbandry at Uruk in the Archaic Period," JNES 39 (1980) 32, nos. 35-36; see now H. Nissen, P. Damerow and R. Englund, Archaic Bookkeeping (ChicagoLondon 1993) 89-93 with figure 7 1. The same ratio is attested in the Ur I11 account cited in note 26 above; this reference should be understood as hard evidence for an administrative rule, since as a theoretical exercise (see Archaic Bookkeeping, 97102) the text mirrored traditional quotas and since those quotas and value equivalencies-yearly deliveries per milk cow of dairy fat and cheese, and silver values of these products, respectively-also underlying numerical entries in the text are found to be exactly the same in contemporaneous administrative documentation. Insofar as an adult cow should bear one calf per year, the ratio of 1:2 may result from a set number of calves either culled from the herd for meat, sacrifice or some other purpose, or a number accorded hired herders as their income. However, it is impossible to derive a similar rule for the archaic period based on just two small accounts, in particular since the production of dairy fat per cow registered in the two texts seems to be inconsistent. M. Green, op.cit., pp. 12-13 (the last column of this table records the fraction of the oil contained in the jars KISIM& delivered per adult female animals; for example, the first text W 20274,55 books obv. i 1 70 ewes and rev. i 2 a fat delivery of 3Nl IN8 = 3 11, KISIM, for a quota of 'I2,, or, with the author, 0.050 KISIM, fat per ewe). See ATU 3, p. 29, figure 12, and pp. 30-32. Of the proto-cuneiform signs representing ceramic vessels, only NI, may have been a two-dimensional depiction of clay objects found in the pre-literate accounting devices known as clay bullae; see now the summary of D. Schmandt-Besserat's important work on these symbols i i her Before Writing, vol. 1 (Austin 1992) 108-128, and my review of her treatise in Science, 11 June 1993, p. 1671. The 'oil tokens,' believed themselves to have represented concrete containers, have been found in bullae from Uruk and from Habuba Kabira in Syria. ATU 3, pp. 29-32. See P. Darnerow and R. Englund, ATU 2, 132-134. Alone the use of different numerical systems with the two products in archaic texts would obscure a possible connection between DUG, and GA>AR.Cheese was in later periods quantified not as discrete units but in capacity measures. The specific contents of dairy milk (87% water, and 13% solids, of which ca.

Englund

35 36

37 38 39 40

41

42

43

44 45

46 47

48

49

Late Uruk Cattle

3 3 % was fat, 33% casein, and 5% lactose) may have played a role in determining that herders from the pre-Sargonic Lagash period were required to deliver 18 sila, from the neo-Sumerian period 7 'I2 sila of cheese per year, i.e., ca. 1.5 x the norm set for dairy fat. This roughly corresponds to the volume of kaSk cheese which may be derived from a measure of milk relative to its fat content. I assume that the norm of 7 '1, sila derived from a nice relationship to the norm of 5 sila of i nun fat from Ur I11 times, and that 18 was also a "nice number" in pre-Sargonic Girsu bookkeeping: 3 bhn, or 11, of the barig consisting of 36 sila. ATU 2, 128 c. Cf. the texts W 19466,a (unpubl.) obv. i 1 ('5N1 5 N i ; DUG,) and W 20652 (unpubl.) obv. i 1 (4N1 [ ] '6N; ; DUG, [ I). The notation 3N1 9N8 in ATU 5, pl. 111, W 9656,gl (cited ATU 2, 129 d, as ATU 1, no. 490) refers to an object not preserved in the second case of the tablet, and this and the preceding two notations could in principle derive from a number of other numerical systems. Clearly sexagesimal, however, is the notation r3N342N11[ 1 '4N; in ATU 5, pl. 64, W 9 5 7 9 , ~rev: 1 (cited ATU 2, 129 d, as ATU 1, no. 352); the apparent object represented by the sign combination SUHUR KAS,, literally 'jar of dried fish flour oil,' must at least be admitted as a weak reference for the use of N8 < 11, in a sexagesimal notation of oil jars. Note further the possible value N8 = 1Il0 'iku' in the area system attested in the two Jemdet Nasr texts R. Englund and J.-P. Grkgoire, MSVO 1, nos. 2 obv. i 2d, and 4 obv. i 5d. See R. Englund, "Archaic Dairy Metrology," Iraq 53 (1991) 101-104. Including only the 3 units qualified as BA KI, in the second subcase of the first case of each text's obverse face, the addition is: 3 + 1 '1, + 112+ 1 + 2 + 3 + 2 = 13 DUG,. See ATU 2, 131 c. No dairy accounts known to me contain a notation 5-9N2, in compliance with the expected replacement of 5N2 with IN1 KU,,. I might draw attention to the fact that tokens often related to this sign have been found in clear association with sealed bullae in Uruk (see D. Schmandt-Besserat, Before Writing, vol. 1, p. 119, to W 20987,27; unmarked crescent) and possibly within still complete bullae &om Susa (identified tomographically; this information was given me by J. Friberg). See ATU 3, pp. 73 and 114-116, respectively; in 'Tribute' followed by a notation of '10 cows'. See preliminarily my "Grain Accounting Practices in Archaic Mesopotamia," in J. HOyrup, ed., Changing Views on Ancient Near Eastern Mathematics (Berlin forthcoming). The text W 20274,72 (unpublished) seems to contain an addition '2N11[ ] + '2N1 SILA,,+GARA,,' + IN8 = IN, DUG,, implying that, as might be expected, N8 also served in this system to represent 'I2 of a basic unit and 5x N, SILA,,. The use of N2 and N, SILA,, together in W 20274,92 (unpubl.) with an apparent summation on the reverse is not otherwise attested among dairy accounts and may so represent a scribal error. See ATU 2, 15360. See 11. 21-61 of the archaic lexical list 'Vessels', ATU 3, 125-129. Far fewer examples are known of signs impressed in the beer sign DUG,, including KASKAL, LAM,, NAGA, and U2, (types of herbs ?), all of which are attested only in administrative texts. ATU 2, 140-141. This interpretation should be noted to the suggestion of R. Michel, P. McGovern and V. Badler, "Chemical Evidence for Ancient Beer," Nature 360 (5 November 1992) 24, that the grooves found on the inside surface of vessel shards excavated in Late Uruk levels of Godin Tepe were represented by the internal markings in the DUG signs. Cf., for example, the unpublished texts W 15774,b obv. i 1 (KAS, together with DUG, representing a container of dairy fat) and W 2051 1,2 obv. v 3a, vi lb3, 2b3 (registering counted KAS, SILA% BA in a large account of dairy fats). Cf., for example, the texts'W 19408,5 (KAS, together with DUG,), 12, 17, W 20044,40 (all unpublished), and M. Green, Visible Language 15, 355, fig. 4a, W 21049 (together with DUG,), MSVO 1, 179 obv. i 3 (KAS, following DUG,) (in no instance together with DUG,; to be noted to ATU 2, 131).

SHEEP AND GOAT TERMINOLOGY IN UR I11 SOURCES FROM DREHEM Piotr Steinkeller (Harvard University)

1 Introductory remarks First, for the benefit of readers not versed in Assyriology, a few words must be said about Drehem and its written sources. Drehem, ancient Puzrish-Dagan, was founded by king Shulgi in the middle of the 21st century BC as a center for the collection and distribution of domestic and wild animals that were delivered to the state as tax revenues or were obtained as booty during the course of military campaigns.l In fact, Drehem was not the only such center in Babylonia; in several other cities (Ur, Uruk, and possibly Eshnuna) there existed similar but smaller animal centers, which functioned as local branches of Drehem. This enterprise operated for some thirty years. Among the animals that passed through Drehem and its subsidiaries one finds primarily cattle, sheep, and goats, both domestic and wild, but also gazelles, deer, bears, pigs, marsh boars, and various types of birds. As far as it can be determined, the Drehem enterprise was a completely unique occurrence in Mesopotamian history. Unlike other livestock operations known from Mesopotamia, Drehem was not involved in animal production, its sole purpose having been the collection and distribution or allocation of animals. Furthermore, the overwhelming majority of animals handled by Drehem and its branches originated not in Babylonia itself but on its northern and eastern peripheries: the Assyrian piedmont, the Zagros mountains, and the western portion of the Iranian plateau. The special character of the Drehem enterprise gave rise to the development of an exceptionally rich animal terminology. The importance of this terminology to the modern scholar lies in the fact that it was used by the same group of scribes during a comparatively brief period of time. This assures that a single classificatory system is involved, which in turn allows highly reliable identifications of species and breeds. The present study deals specifically with Drehem terminology pertaining to domestic and wild caprinae. The types of terms considered are only those which, however broadly, relate to biological characteristics of animals. Not treated here are the numerous terms specifying the intended use of animals, such as kin-@&a,which identifies the lamb suitable for extispicy? or the various descriptions of meat preparation and processing, such as mu-du-lum, "sated",3 uzuhhd, "dried": urn-a-bala, "meat broth"? ka-izi, " r ~ a s t e d "and ~ ne-mur-ta ba-leg6, "roasted in hot embers."7 Because of limitations of space, I will not discuss the historical evolution of the Drehem terminology and the distribution of terms depending on text types. Nor shall I offer a Bull. Sumerian Agric. 8 (1995)

49

Steinkeller

Sheep and goat terminology in Ur I11 sources from Drehem

systematic comparison of the Drehem usage with that documented in other Ur I11 centers; comparisons will be made, but only selectively.

2 Species Beginning with the question of species, four species of caprinae can be confidently identified in Drehem sources: the domestic sheep (generic udu), the domestic goat (generic maS), the wild sheep or the Asiatic mouflon (udu-hur-sag),8and the wild goat or the bezoar (dard~ldara~).~ The identifications of udu-hur-sag and dar* as the mouflon (Ovis orientalis) and the bezoar (Capra hircus aegagrus), respectively, are assured by the existence of hybrids of these animals with their domestic counterparts.10 At the same time, it is striking that the Drehem terminology and, for that matter, Mesopotamian animal taxonomies in general, lack specific terms for wild caprinae such as the urial (Ovis vignei), the ibex (Capra ibex), and the markhor (Capra falconeri), which, as we can judge from surviving pictorial representations,ll were well known to ancient Mesopotamians. It is possible, therefore, that, in its broad sense, the term udu-hur-sag includes also the urial, whereas the term dar* designates also the ibex and the markhor. Other species that need to be discussed in this connection are mBS ~ a - ~ a n ("Makkan ~~), goat", and maSda (A&. ~abitu),"gazelle." Since the "Makkan goat" occurs infrequently in Drehem sources,l2 and since this animal came from (or was at least associated with) the land of Makkan, which can positively be identified as the Oman peninsula,l3 it almost certainly was not simply a breed of domestic goat, but rather a distinctive exotic species, whose physical appearance was somewhat goat-like. In my opinion, the best candidate for that species is the Arabian white oryx (Oryx leucoryx), a type of antelope, which is by far the most characteristic animal native to Oman.14 That the oryx was known in Mesopotamia, is proved by ancient iconography.15 An alternative, but, in my view, less likely identification,l6 would be the Arabian tahr (Hemitragusjayakari Thomas 1894), a type of small mountain goat.17 Turning to the mdda, there is no doubt that we find here the "gazelle." In all probability, maSda does not designate any specific type of gazelle, but rather is an inclusive label for all gazelles native to the region, e.g., the goitered gazelle, the mountain gazelle, and the Dorkas gazelle. Finally, we should consider the question of Segg-bar,18Akk. sapparu. The only reason why it is at all necessary to address this issue here is the fact that this animal is thought to be either a wild goat or a wild sheep.19 However, it is quite certain that both these identifications are wrong. That the legg-bar was neither a sheep nor a goat is demonstrated quite conclusively by the fact that, in Drehem summary accounts, the legg-bar are grouped together with deer (lulim), equids, and bovines,20 and not with sheep and goats, domestic or wild. Equally telling is the fact that the age of Segg-bar is occasionally identified,al and that its young are described as amarJ2 both features never being attested in connection with sheep and goats. Since Segg-bar are commonly associated with deer and gazelles,23 and since this species very likely had antlers,24 it appears highly probable that the Segg-bar was a type of deer. Assuming that lulim is the red deer (Cewus elaphus), the Segg-bar should in all probability be identified as the Mesoptamian fallow deer (Duma mesopotamica),25 which has characteristic palmated antlers,26 and whose coat is spotted with white.27 In this connection, it may be significant that, according to the Kesh Temple Hymn, lines 48-49,28 the Segg-bar was "spotted" or "dappled" (an-a).

Steinkeller

Sheep and goat terminology in Ur I11 sources from Drehem

3 Breeds 3.1 Sheep 3.1.1 Native breed of sheep It is safe to assume that the sheep identified simply as udu, without additional breed designations, represents the standard native breed. A specific name of this breed is udu Ki-engir,(GI), the "native" or "Sumerian" sheep, which appears only in three Drehem documents.29 It is noteworthy that in two of those sources (AUCT 1 405; BIN 3 350), both concerning the Suaid of the city URU.SAG.RIG7, udu Ki-en-gir, are listed in apposition to udu Lu-lu-bu, the "Lullubean" sheep. In the Ur 111 sources from Lagash, Urnrna, and Ur, the native breed of sheep is described by the appellative udu-gir, or udu-em-gir,, "native sheep."30 3.1.2 Fat-tailed sheep Ilrehem sources mention at least two types of fat-tailed sheep: gukkal and udu kun-gid. The gukkal, which is first documented in Babylonia in the Uruk I11 period,31 is very common in Ilrehem documentation. There, the term g ~ k k a letymologically ,~~ "big tail" (*kun-gal),appears to describe specifically a sheep with a short, broad fat-tail.33 The udu kun-gid, a "long-tail", which is considerably rarer than the gukkal at D ~ - e h e mis, ~in~all likelihood the variety with a long fat-tail.35 This type of sheep, otherwise unknown in Mesopotamian documentation, is probably identical with the udu-gukkal-ib-18= zib-ba-nu of tlh. XI11 26,36 which, according to the Akkadian translation, was a gukkal with an extra long tail. An alternative, but less likely, identification might be udu-gukkal-TU = gu-uk-kal-la-nu, listed in Hh. XI11 27,37 which probably denotes the gukkal with an extra big or fat tail. It appears that yet another term for the fat-tailed sheep is udu gal-tab-bu-um. In contrast to the gukkal and udu kun-gid, this type of sheep is extremely rare, as it is found in only five published documents.38 The possibility that udu gal-tab-bu-urn was a type of fat-tailed sheep is suggested by the text FAOS 16 1052, where, in three separate sections, udu gal-tab-bu-um are grouped together with the udu kun-gid and the gukka1.39 It is also significant that, in two other documents, udu gal-tab-bu-urn are associated with the udu kun-gid and gukkal, respectively.40 Given the great rarity of this type of sheep, it is possibly the "fat-tail" with an extra big tail. For this possibility, note further that Hh. XI11 names two specific varieties of the gukkal, called zibbanu and gukkallanu (see above). If, as I suggested earlier, udu kun-gid is the zibbanu, then udu gal-tab-bu-um is perhaps to be connected with gukkallanu = udu-gukkalT U . ~ ~ The term gal-tab-bu-urn, which appears to be a Sumerian loanword, may conceivably be identical with kaltappu, a variant form of kilzappu, "footstool", also "threshing board."42 The problem with this solution, however, is that the form kaltappu is not documented before NeoAssyrian times.43 No less important, it is difficult to think of a reason as to why a fat-tailed sheep should be associated with a footstool or a threshing board.44

Steinkeller

Sheep and goat terminology in Ur I11 sources from Drehem

3.1.3 udu aslum,(A.LUM),"long-fleeced sheep" The breed called udu A.LUM is very common in Drehem sources. Outside Drehem, this breed is extremely rare, as it is attested in only two other Ur 111texts: one from Umma (TLB 3 44: 1,3) and one from Lagash (Barton, H a v e ~ o r d2, pl. 75 HLC 57 ii 5 , l l , iii 3, 12, iv 1, 4).45 Although udu A.LUM does not appear in economic tablets before Ur III times, the term is attested already in an Early Dynastic list of animals, preserved in manuscripts from Abu Salabikh and Ebla.46 Following the Ur I11 period, udu A.LUM is documented in the early OB texts from Isin$7 at Old Babylonian Mari, and in the inventories from Qatna, as well as in Sumerian literary texts.48 There are no clear attestations of this term following OBMB times.49 The term udu A L U M is listed in the forerunners to Hh. XIII, where it appears with the qualifications niga, niga sigg, and gukkal (MSL 811, p. 83 lines 8-11). In the standard recension of Hh. XIII, udu A.LUM is replaced by udu asq(LAGABxA)-lum,with a translation pasillu: udu L A G A B X A ~ - ~ ~ L=pa-sil-lum UM (Hh. XIII 12; see also lines 13-15).50 In the Hg. commentary to Hh. XIII, pasillu is explained as immeru, "sheep", and udu L A G ~ X A . L U M is further glossed sulul~u/ sulum~ii,"long-fleeced sheep": udu aSLAGABxA.[LUM] = pa-si-i[l-lum] = [im-me-ru], udu S ~ - ~ ~ L A G A B X A ~ = ~ .su-lum-ku-ri ~UM] = MIN (MSL 811, p. 54 A II 235-36). Based on the writing asq-lum of Hh. XIII, and the fact that this writing replaces the earlier A.LUM, one must conclude that ALUM is to be read aslum,. aslum,, in turn, must be identical with Akkadian a ~ l ua,well-known ~ ~ term for "sheep."52 Since ALUM is first attested in an ED lexical text of undoubtedly northern Babylonian ("Kishite") background (see above), this term is very likely an abbreviated / defective Akkadogram, similar examples of which are welldocumented in the ED script of northern Babylonia.53 As can be concluded from the fact that it was a rare word already during OB times, aslu had ceased to be a viable breed-designation sometime in the Isin-Larsa period, that function having been taken over by pasillu (see Hh. XIII 12-15).54 Turning now to the question of the identity of this breed, it appears virtually certain that aslum, (and likewise the Akkadian aslu) is a word for the "long-fleeced" or "hairy" sheep, usually identified by the term SU~@U(S~G.G~D), Akk.sulumhii.55 Among the evidence pointing to such a conclusion, we must mention, first, the aforementioned Hg. commentary to Hh. XIII, which provides AS4.LUM with the gloss lsulu~ul. Second, in the leather texts from Isin skins of udu ALUM are consistently distinguished from those of ordinary sheep.56This shows that there was something distinctive about the skin of the aslum,, and, since that characteristic was not its color,57 it must have been long hair that set it apart from ordinary sheepskin.58 And, in fact, skins of aslum, sheep are often described as "hairy" (siki-mu) in the very same sources.59 Third, in the composition Curse of Akkade, line 23, the aslum, sheep is described as "longfleeced": ur-girls ur-nim darab kur-ra udu A.LUM suluhu(~fG.Gf~) si, "dogs, lions, mountain bezoars, the a l u m sheep richly provided (si = malii) with long fleece."60 And last but not least, the identification of aslum, as the "long-fleeced" sheep is strongly indicated by the very absence of that breed in Drehem sources. Considering that these employ the most extensive taxonomy of sheep breeds known from ancient Mesopotamia, it is precisely here that some mention of this breed is expected to be found.

Steinkeller

Sheep and goat terminology in Ur I11 sources from Drehem

3.1.4 Shimashkian sheep The sheep known as udu S i m a ~ k i ( L u . ~ ~ ( . A which ~ ) ) , is ~ ~exceedingly common at Drehem, represents a domestic breed that was indigenous to the lands of Shimashki. As we now know,62 those lands comprised the whole western portion of the Persian plateau, extending from the Caspian Sea in the north to the borders of Fars in the south. The term udu Simagki is found exclusively in the Ur 111documentation from Drehem. The only exception is the occurrence of this term in the Lagash text Barton, Haver$ord 2 pl. 75 HLC 57.iii.17, iv. 1, 8. As noted ear lie^-,63 the use of Drehem terminology in that document is due to its recording deliveries of livestock from Drehem and Ur. 3.1.5 Lullubean sheep Another certain example of a foreign breed is the "Lullubean" sheep. This breed is frequent at Drehem, although not as common as the Shimashkian sheep. It is usually identified by the term udu L ~ I I ~ ~ , ( L u . u L uwhich, - ~ ~ ) ,with one exception (Barton, Haverford 2 pl. 75 HLC 57.iii.14, iv. l , 5 - L a g a ~ h )is, ~attested ~ exclusively at Ur 111Drehem. In two Drehem sources, the same breed is identified as udu Lu-lu-buythe "Lullubean" sheep (AUCT 1 405; BIN 3 350).65 As indicated by its name, this breed was native to the country of Lullubum, situated in the Zagros ranges, probably in the area of present-day Kurdistan.66 I assume that L U . U L U - U-~to be read L U I I U ~ , ( L U . U L Uor - ULu-ulu-um ~) - is a fancy writing of Lullum,67 a byform of Lullubum, which is documented in the Old Babylonian texts from Shemshara, Mari, and Tall Rimah, as well as in later sources.68 There is independent evidence that this variant was known and used already in Ur 111times.@ 3.1.6 Sharurnijum sheep This breed, which is identified by the term udu Sa-ru-rni-~m,~~ is documented exclusively at Drehem. With the exception of the udu gal-tab-bu-um, it is the rarest of the breeds documented there.71 Since Sharurnijum is a gentilic formation, we apparently find here yet another foreign breed. It is troubling, however, that neither Ur 111 nor any other third millennium sources record any toponym from which the form Sharumijum could plausibly be derived. To my knowledge, the only toponym that would permit such a derivation is Sa-ru(?)-maki, which is uniquely attested in an Old Babylonian geographical list from Tell Harmal (MSL 11, p. 58 iv 122). 3.1.7 To conclude this part of my discussion, the following summary may be offered. If the suggested identifications are correct, eight specific breeds of sheep can be identified in Drehem sources: a) the native breed (generic udu / udu Ki-en-gir,); b) three breeds of fat-tailed sheep: the short broad tail - gukkal, the long tail - kun-gid, and the extra big tail - gal-tab-bu-um; c) the long-fleeced sheep (aslum,); d) three foreign breeds: the Shimashkian or Persian sheep, the Lullubean or Kurdistani sheep, and the Sharurnijum sheep.

Steinkeller

Sheep and goat terminology in Ur I11 sources from Drehem

The breed terminology used outside Drehem is much simpler. Thus, at Lagash, sources differentiate only between "native" (udu or udu-gir, or udu-eme-gir,) and "fat-tailed" (udugukkal) ~ h e e ~At . 7Ur, ~ the distinction is between "native" sheep (udu-eme-gir,), "fat-tails" (udugukkal), and "fat-tails of the Upper Country" (udu-gukkal-igi-nim-ma).73 And at Umrna, sheep the latter are classified as either "native" (udu-eme-gir,) or "foreign/mountain" (udu-k~r-ra)?~ i ~ indicated ) . ~ ~ by further subclassified as either "fat-tailed7'(udu-gukkal) or "black" ( ~ d u - ~ As these data, outside Drehem one distinguished basically only between native and fat-tailed sheep, with "fat-tail" being a virtual synonym for "foreign" breeds. This contrasts sharply with Drehem practice, where fat-tails (gukkd, kun-gid, and gal-tab-bu-um) are always distinguished from foreign sheep (Shimashkian, Lullubean, and Sharumijum), with no instances of the former included or subsumed among the latter.

3.2 Goats In regard to breeds of domestic goat, the picture is considerably simpler. Only two such breeds are documented at Drehem: the "native" (generic mag) and the "Shimashkian" (mag Sima~ki),76 the latter being very common. Outside Drehem, only the native breed is attested.77

3.3 Hybrids As noted above (see under $2), Drehem sources mention two types of caprine hybrids: a hybrid and a hybrid of the domestic goat of the domestic sheep with the mouflon (udu-a-udu-hur-~a~),~~ with the bezoar ( m i l - a - d a r ~ d a r a ~Both ) . ~ ~hybrids are very common at Drehem. The terms audu-hur-sag, "mouflon seed", and a-dara?l/dara4,"bezoar seed", are limited to Drehem sources. We otherwise find no references to either hybrid in cuneiform documentation. Yet another example of a caprine hybrid, uniquely attested in CMET 7 72:7, is possibly udu-a-gukkal, "sheep (with) the seed of the fat-tailed sheep."

4 Sex and Age 4.1 The basic terminological distinction is between adult and young animals. The former are identified as udu (nita), "ram", ug, "ewe", ma6 (nita) or mag-gal (nita), "buck, and UZ,~O "nanny goat." Unless qualified by the designation gig-dii, "breeder" (see under $4.2), udu (nita) and mi6 (nita) 1mag-gal (nita) invariably describe castrated a n i m a l ~ .The ~ l terms for the young are sua4, "male lamb", kirll, "female lamb", mi& "male kid", and "iri~-~hr, "female kid." The young are further subclassified as ga, "suckling",82 gaba, and gub, the latter two describing weaned or semi-weaned animals. The only terms warranting discussion here are gaba and gub. In his contribution to this volume, Heimpel considers gaba and gub to be two separate terms. As understood by him, gaba, "Brust", describes young animals that were small enough to be clutched against one's breast,83 whereas gub, "Bestand", signifies animals that were brought for their first shearing and "werden gleichzeitig zum ersten Mal im Bestand gezZhlt."84 However, since gaba never occurs together with gub,85and since both terms describe animals older than ga, "suckling", but younger than fully mature, and thus refer to the same agecategory, I am inclined to assume, following Landsberger's original suggestion,86 that these are variant spellings of the same word, with gub being a syllabic realization of gab(a).

S Ieinkeller

Sheep and goat terminology in Ur I11 sources from Drehem

However, regardless of whether gub is the same as gaba, the interpretation of gaba presents certain difficulties. This term, lit. "breast", is almost certainly an abbreviation of the descriptions dla4 gaba ug and m h gaba iiz, which are recorded in an ED lexical source.87 Accordingly, gaba means "(animal at the) breast (of its mother)", a further corroboration of which is provided by Hh. XIII 84, where udu-gaba is translated im-mer ir-ti, "sheep of the breast." However, such a translation obviously clashes with the commonly-held view that gaba describes fully weaned a n i m a l ~ . ~On 8 the other hand, the possibility that gaba could be an alternative term for "suckling" (as in well-known (dumu-)gaba"suckling infant") is precluded by examples in which this term appears together with As a possible solution to this dilemma, I propose that gaba (and similarly gub) serves to describe semi-weaned animals. As is practiced by some pastoralists, the young over onemonth-old are separated from the ewes and grazed in a separate flock, though they are still allowed to suck the ewes once or twice a day, till the weaning is completed.90 In support of this interpretation, note that the animals classified as gaba or gub are never qualified as niga, "barleyfed", or u, "grass-fed", which suggests that milk was still the primary source of their nutrition.

4.2 In addition, Drehem sources use the following ageisex related terms: a) h-tu-da, "newborn"; b) gig-dii, "breeder";91 c) en-zi, "lead animal";92 d) us sila4 nir/nu-a1hz miS nWnu-a, "pregnant ewe / nanny goat";93 e) ug silaq dii-a I iiz ma6 dii-a, "pregnant ewe / nanny Rejecting the long-accepted interpretation of nsnu-a sila4/m6S as "pregnant (with) a lamb/kidm,95Heimpel now proposes the translation "infertile", taking nulnu-a to mean "without."96 In my view, this interpretation is extremely unlikely, primarily because this description is very common at Drehem. By contrast, ug silaq dii-a / iiz m6S dii-a, the other known term for "pregnant", is attested in only two Drehem texts (see above). If the former expression were to mean "infertile",97 we would be forced to conclude that, at Drehem, infertile animals were regularly identified (by what means and for what purpose?), whereas pregnant ones were not. This I find impossible to accept. As for Heimpel's objection that it is difficult to think of any rationale for "slaughtering" pregnant animals, it should be pointed out that, at Drehem, the term ba-liS means not only "slaughtered", but also "dead (of n.atural causes)." As for silaq/maiS dh-a, this term is probably to be analyzed as "planted (dii-a = zaqpu) (with) a lambkid." Accordingly, its specific meaning seems to be "impregnated."

5 Various physical characteristics 5.1 Color Five color designations are attested: babbar?8 "whiteV,99816, "black",100 si4, "brown, red",lol sig17(GI), "yellow, tanm,102and giin-a, "spotted, mottled."l03 With the exception of "brown, red", which, as far as I know, is attested only for goats,lo4 these descriptions are documented for sheep and goats alike.

Steinkeller

Sheep and goat terminology in Ur I11 sources from Drehem

Among these terms, only sigl7(GI),"yellow, tan", requires further comment. This term has uniformly been read as gi by scholars, with the translation "(fed with) reeds."l05 However, that interpretation is obviously erroneous, as shown by the examples where GI is combined with such qualifications as niga, "barley-fed",l06 and gay "suckling",lO7 which make it absolutely clear that GI cannot describe a type of feed. On the other hand, yellow or tan color is given as a description of both sheep and goats in Hh. XI11 (see above), and, since sig17 is otherwise attested as a graph of /sig/ "yellow, greenV,l08 there can be no doubt that this is how GI should be read in this context. 5.2 Condition of fleece I goat-hair Drehem texts use four terms to describe the condition of the sheep's fleece: bar-gal, "with fleece",l09 bar-su-ga, "without fleeceWy1 lo sila4/kirll AS(=dili?)-ur4,"lamb plucked for the first time9',lll and si-2-16, "plucked twice."l12 The first three are frequent in contemporaneous sources from other places. The fourth term, si-2-laYwhich is also attested for goats, occurs only at Drehem.

5.3 Physical deformities Finally, we find a number of terms describing various physical deformities: a) ba-za, "cripp1ed;l l3 b) gu-gu-turn;l14 c ) umbin TAB .TAB ;I d) G U R ~G. U R ~ . Of these, only the meaning of ba-za, "crippled", is beyond question. Assuming, as suggested by CAD G, p. 123a, that gu-gu-turn, which is attested exclusively of lambs, is a loanword from Sumerian kud-kud-du, Akk. /jurnrnuru,l l7 its meaning would be "crippled" or "deformed." More likely, however, gu-gu-turn is a type of meal or a ritual designation. This is suggested by the text MVN 2 1595, which records 1 udu niga mu udu mu-du-lum uzu-had ii silaq gu-gu-turn-ma&, "1 barley-fed sheep (earmarked) for salted (and) dried sheep and the gugu-turn lambs." The meaning of the term umbin TAB.TAB, which is found only in connection with goats, is , ~ ~ of ~ somewhat obscure. If translated literally, umbin TAB.TAB would mean " f ~ u r - h o o f e d " but, course, such a description would be meaningless in reference to quadrupeds. One possibility is that umbin is to be understood here not as "hoof' but as "digit" or "toe." In that case the term could describe a goat with the overdeveloped remnants of its back toes. A more plausible solution, however, is that TAB.TAB is to be analyzed as an early graph of adq, i.e., ZA-tenli or EREN-tend,Akk. kubbulu, 'lame."l l9 Accordingly, umbin TAB.TAB would mean "(goat) with a lame hoof." The term GUR8.GUR8, which is documented also of cattle,120 is even more difficult. Possibly, it is to be connected with [udu-gur4-gur4]= kur-kur-ra-nu (Hh. XIII 48); udu-gur4gurq and [uldu-gur-gur in Forerunners (MSL 811, p. 83 line 33, p. 91 line 3'). Assuming that kurkurrgnu derives from kurkurru, a type of bulky jarY121this term could describe an animal with a swollen or enlarged belly, and should be translated "pot-bellied" or the like.

Stcinkeller

Sheep and goat terminology in Ur I11 sources from Drehem

Grades of quality 'I'he following terms grade sheep and goats according to type of feed and feeding rates: a) niga sigg, "barley-fed, top quality (first grade);122 b) niga sigg-us, "barley-fed, top quality, next grade"; c) niga 3-kam-us, "barley-fed, third grade"; d) niga 4-kam-us, "barley-fed, fourth grade"; e) niga, "barle~-fed";l~~ f) niga gud-e-us-sa,"barley-fed, 'following the oxen"'; g) u, "grass-fed";124

6

The basic distinction is between barley-fed and grass-fed animals. The former are further subdivided into six categories, which can be grouped as follows: four superior categories (a, b, c, and d), the standard category (e), and the inferior category (f). The sequence of these categories is assured by texts such as BIN 3 239 and 240, where all seven categories appear, always in the same order. Except for gud-e-us-saythe meaning of these terms is clear. Contrary to the common belief that gud-e-us-sa designates sheep and goats of the highest quality, whose value approximated or was closest to that of cattle,l25 a systematic examination of this problem shows that, in fact, it was the least valued category of barley-fed animal~.l2~ If, as seems likely, gud-e-us-sa is to be taken literally as "(an animal) that follows oxen", the term probably describes sheep and goats that were fed with barley left over from cattle feeding. It is certain that this six-tier classification of barley-fed animals is based primarily (if not exclusively) on the daily rates of feeding. Such feeding rates are, of course, well-documented in sources from Umma, Lagash, and other Ur I11 centers. For this conclusion, see, especially, MVN 15 64 i 1-3 (Drehedummal), which records 111udu niga sigg at the rate of 1112 sila of barley per sheep, followed by 196 udu niga at 1 sila, and 90 udu at 112 sila. Cf. also UET 3 1197 (Ur), recording udu niga sig5 and udu niga <sig5->usat 1112 sila of barley, and udu niga at 1 sila.

7 Wool and goat-hair The following are designations of wool and goat-hair found in Drehem sources: a) siki gir, (= siki udu-eme-gir,), "wool of the native sheepW;127 b) siki aslum,(A.LUM) (= siki udu aslum,), "wool of the long-fleeced sheepn;128 c) siki udu S i m a g k i ( ~ O . ~"wool ~ ) ~ ~of, the Shimashkian sheep"; 129 d) siki kur(-ra) (= siki udu-kur-ra), "wool of the foreign/mountain sheepm;130 e) siki gi6 (= siki U~U-gi6), "wool of black sheepW;l31 f) siki gir-gul, "wool (plucked with) the aritu-knife" or, less likely, "wool (of the sheep slaughtered with) the arttu-knife7';132 g) siki mug, "wool of poorlinferior quality7';133 h) siki i i ~"goat-hair."l34 ,

Steinkeller

Sheep and goat terminology in UI I11 sources from Drehem

The Drehem terminology for wool and goat-hair comes from the texts that originated in a separate office of the Drehem establishment, whose operations appear to have been largely independent of the animal office. It is characteristic that this terminology is essentially the same as the one used at Umma, although it is also influenced by the breed designations employed at the Drehem animal office. The dependence of Drehem wool terminology on Umma is reflected in its adoption of the terms gir, (for udu-erne-girls), kur(-ra) (for udu-kur-ra), and gi6 (for udugi6), which are basic classificatory terms for sheep in Umma sources (see above under §3.1.7).135 On the other hand, influence of the Drehem animal terminology is evidenced in the occasional use of the terms siki aslum, and siM udu Gima~ki,two breeds that are documented only in Drehem sources.

Slcinkeller

Sheep and goat terminology in Ur I11 sources from Drehem

TABLES OF TERMINOLOGY

A. SPECIES a. udu - domestic sheep (Ovis aries). b. udu-hur-sag - (Western) Asiatic.mouflon (Ovis orientalis); secondarily may also describe the urial (Ovis vignei). c. mi6 - domestic goat (Capra hircus). d. darawdara4 - bezoar (Capra aegagrus); secondarily may also describe the ibex (Capra ibex) and the markhor (Capra falconeri). ( ~ ~ ) goat; probably to be identified as the Arabian white e. m G ~ i - ~ a n- "Makkan" oryx (Oryx leucoryx).

13. BREEDS

a. Breeds of domestic sheep 1. udu I udu-Ki-en-girx(GI) - native breed. 2. gukkal - fat-tailed sheep (short broad tail). 3. udu kun-gid - fat-tailed sheep (long tail). 4. udu gal-tab-bu-um - fat-tailed sheep (extra big tail). 5. udu aslumx(A.LUM) - long-fleeced sheep. 6. udu S i m a ~ k i ( ~ u . ~ -~Shimashkian ( . ~ ~ ~ ) ) sheep. 7. udu ~ u l l u m ~ ( ~ ~ .I ~Lu-lu-bu ~ - u m - Lullubean ) sheep. 8. udu Sa-ru-mi-um - Sharumijum sheep. b. Breeds of domestic goat 1. m6S - native breed. 2. mi5 S i m a ~ k i ( I , o . ~ ~ (- .Shimashkian ~~~)) goat.

C. HYBRIDS a. udu-a-udu-bur-sag - hybrid of the domestic sheep with the Asiatic mouflon. b. m8-a-daraWdara4 - hybrid of the domestic goat with the bezoar. c. udu-a-gukkal - hybrid of the domestic sheep with the fat-tailed sheep.

Steinkeller

Sheep and goat terminology in Ur I11 sources from Drehem

D. SEX AND AGE Adult sheep goat

Male udu(-nita)

Female u8

sila4

miis(-nita) m6H-gal(-nita)

uz

m6H

Young Male Female kin1 ga gabalgub mi66-ghr ga gabalgub

ga gabdgub ga gabdgub

a. a-tu-da - newborn. b. gig-dii - breeder. c. en-zi - lead animal. d. ug sila4 nunu-a 1 hz m6g nunu-a - pregnant ewe 1 nanny goat. e. ug sila4 du-a 1 iiz m6E du-a - pregnantlimpregnated ewe 1nanny goat.

E. VARIOUS PHYSICAL CHARACTERISTICS a. Color 1. babbar 1 barg-bar6 - white. 2. gi6 - black. 3. siq - brown, red. 4. sigl7(GI) - yellow, tan. 5. gun-a - spotted, mottled. b. Condition of fleece / goat-hair 1. bar-gail - with fleece. 2. bar-su-ga - without fleece. 3. sila4 / kirll ~ S - u- rlamb ~ plucked for the first time. 4. si-2-16 - plucked twice. c. Physical deformities 1. ba-za - crippled. 2. gu-gu-turn - ... 3. umbin TABTAB - lame(?). 4. GUR8.GUR8 - potbellied(?).

suckling (semi-)weaned suckling (semi-)weaned

1

Sheep and goat terminology in Ur I11 sources from Drehem

St cinkeller

I:. GRADES OF QUALITY a. niga sig5 - barley-fed, top quality (first grade). b. niga sigg-us - barley-fed, top quality, next grade. c. niga 3-kam-us - barley-fed, third grade. d. niga 4-kam-us - barley-fed, fourth grade. e. niga - barley-fed. f. niga gud-e-us-sa - barley-fed, "following the oxen." g. - grass-fed.

G . WOOL AND GOAT-HAIR a. siki &,(GI) - wool of the native sheep. b. siki aslumx(A.LUM) - wool of the long-fleeced sheep. ~ ) ~of~ the Shimashkian sheep. c. s k i udu S i m a H k i ( ~ f i .-~wool d. siki kur(-ra) - wool of the foreignlmountain sheep. e. siki gi6 - wool of the black sheep. f. siki gir-gul - wool (plucked with) the aritu-knife. g. siki mug - wool of poor/inferior quality. h. siki iiz - goat-hair.

NOTES Abbreviations used are those of the Chicago Assyrian Distionary and the Philadelphia Sumerian Dictionary with the following additions: BAOM

Bulletin of the Ancient Orient Museum (Tokyo)

CMET

Catalogo del Museo Egizio di Torino (TurinJMilan)

Gomi-Sato, NSAT

T. Gorni and S. Sato, Selected Neo-Sumerian Administrative Texts from the British Museum (Tokyo, 1990)

Hirose Coll.

T. Gomi, Y. Hirose, and K. Hirose, Neo-Sumerian Administrative Texts of the Hirose Collection (Potomac, MD, 1990)

Lafont-Yildiz. TCT

B. Lafont and F. Yildiz, Tablettes cune'ifonnes de Tello au Muse'e d'lstanbul (Leiden, 1989)

Lau, OBTC

R. J. Lau, Old Babylonian Temple Records (New York, 1906) H. Limet, Textes sume'riens de la Ille Dynastie d'Ur (Brussels, 1976) I . Spar, Cuneiform Texts in the Metropolitan Museum of Art, vol. 1 (New York,

Limet, TSDU Metropolitan Museum

1988) SACT

Sumerian and Akkadian Cuneiform Texts in the Collection of the World Heritage Museum of the University of Illinois (Urbana)

S teinkeller

Sheep and goat terminology in Ur I11 sources from Drehem

See, in general, T. B. Jones and J. W. Snyder, Sumerian Economic Texts from the Third Ur Dynasty (Minneapolis, 1961), pp. 212-38; G. Buccellati, The Amorites of the Ur 111 Period (Naples, 1966), pp. 274-302; Steinkeller in The Organization of Power: Aspects of Bureaucracy in the Ancient Near East, ed. by McG. Gibson and R. D. Biggs (Chicago, 1987), pp. 28-40; M. Sigrist, Drehem (Bethesda, 1992).

Slcinkeller 15

For excellent representations of the oryx, see Boehmer, Entwicklung, pl. 15 fig. 165; D. Collon, First Impressions: Cylinder Seals in the Ancient Near East (Chicago, 1987), p. 189 fig. 912.

16

As suggested by W. I-Ieimpel, 2.4 77 (1987) 64.

17

See P. N. Munton, ''The Ecology of the Arabian Tahr (Hemitragus jayakari Thomas 1894) and a Strategy for the Conservation of the Species", Journal of Oman Studies 8 (1985) 11-48; Uerpmann, Ungulate Mammals, pp. 111-12. In my opinion, this animal is too ordinary to have been imported to Babylonia. Uerpmam, op. cit., pp. 126-27 n. 9, cites a certain H Kauch, animal dealer at Dubai, who, though familiar with various types of wild caprinae and gazelles, "never received any thars, either dead or alive, nor has seen them in captivity. He thinks that the thar is considered as too ordinary and too similar to a domestic goat to receive much attention.''

18

For practical reasons, I retain the traditional reading Beg9-bar. However, it should be pointed out that the correct reading is Begbarx(S~~9)bar. I hope to discuss S E G ~ (= LAK-263) and related signs in a separate study. AHw., p. 1027a: 'Wildbock"; CAD S, p. 166: "wild sheep" in the occurreuces from Etana Epic and in bilingual examples, but, inexplicably, "a bovid", in the heading.

See Heimpel, BSA 7, 131-33. E.g., [x] 'udul niga mu-du-lum-Sb, "[XI grain-fed sheep for salting" (AUCT 2 338:4). See CAD Ml2, p. 162b. E.g., 1 silaq uzu-h6d-Sb, "1 lamb for dried meat" (AUCT 1 95:l). E.g., 1sila4 ga uzu-a-bala, "1 suckling lamb (for) meat broth" (AUCT 1 441:14). uzu-a-bala = ummar me^Jiri (CAD M12, p. 155). See [tu7]-a-uzu = rum-mar7me-e k-e-ri (MSL 11, p. 69 frg. c 1); tu7-uzubala, tu7-a-uzu (MSL 11 p. 152 lines 29-30); tu7-uzu-bala = SU-u [...I, tu7-im-me-uzu-bala = SU-u [...I (MSL 11, p. 89 Hg. B, VI 103-04). E.g., 9 sila4 ga ka-izi-R, "9 suckling lambs for a roast" (BIN 3 369: 1-2). See K A ~ ~ - ~ = - ~MIN ~ N(=E krrziifu) Zci NE (Antagal VIII 112 = MSL 17, p. 173); [uzul-ka-izi = Ji-ir Su-me-e, [uzu-kla-izi = pu-ut-tu-u' (Hh. XV 260-61 = MSL 9, pp. 13-14); uzu-ka-izi = :[urn-mu-u], uzu-ka-izi-tab = [MINI (Nabnitu XXIII 172, 175 = MSL 16, pp. 216-17).

I9 20

Thus we find the Beg9-bar listed: following bovids and deer and before equids (RA 63 [I9691 102 A 0 19548, 104 A 0 19550 = FAOS 16 1359); following bovids and before equids (FAOS 16 958); following bovids and deer, and before sheep (C'I' 32 pl. 14); followed by equids and sheep (CT 32 pls. 28-29 viii).

21

E.g., 1 silaq ga ne-mur-ta ba-Segg, "roasted in hot embers" (BIN 3 74:l-2). neldb-mur = tumru, p2mtu; Seg6 = ba3iilu. Cf. 45 sheep and goats gird-ta ba-Seg6, "roasted in an oven" (SACT 1 171:l-4). See H.-P. Uerpmann, The Ancient Distribution of Ungulate Mammals in the Middle East (Wiesbaden, 1987), pp. 126-27; J. Clutton-Brock, A Natural History of Domesticated Mammals (Austin, 1987), p. 53. For a good representation of the mufflon, see E. Porada in Investigating Artistic Environments in the Ancient Near Eust, ed. by A. C. Gunter (Washington, DC, 1990), p. 76 figs. 7 and 8.

Sheep and goat terminology in Ur I11 sources from Drehem

See Steinkeller, SEL 6 (1989) 3-4 and n. 5.

22

See, e.g., 1 amar Beg9-bar nita mu 1 (PDT 526 ii 4); 1 amar Begybar mi mu 1 (ibid iii 9); 2 Beg9bar nita mu 1 (BIN 3 234:4); 2 Segg-bar nita mu 2 , 3 Segg-bar nita mu 1 (ASJ 12 [I9901 41 no. lO:5-6). See, e.g., AUCT 2 294:2-3; PDT 526 passim.

See under 53.3.

23

With lulim: PDT 40; with lulim and maBda: BIN 3 234,260; with rnaSda: BIN 3 218; AUCT 2 294.

For the ibex, see Porada in Artistic Environments, p. 78 fig. 9; R. M. Boehmer, Die Entwicklung der Glyptik wahrend der Akkad-Zeit (Berlin, 1965), pl. 6 fig. 54; for the markhor, see Boehmer, op. cit., pl. 14 fig. 153. Due to its great similarity to the moufflon, the urial is more difficult to identify; a possible representation is found in Boehmer, op.cit., pl. 16 fig. 175.

24

This is suggested by the following literary passage: Begg-bar-resi-mui-bi nam-ta-an-i?,"the S. raiszs i sticks out its ... " (TCL 16 89:14), where si-mug, lit. "serpentine/coiled horns", is very likely a term for "antlers". For the sense "serpentinc, coiled" of mug, cf. (id-)a-muB, "serpentine water (course)", which describes small meandering canals in Ur 111 documentation from Urnma (H. Sauren, Topographie, pp. 44-45). In another source, the horns of the Segg-bar are described by the term a: 2 6 Begbarx(S~G9)(DP 490 v 2 - Pre-Sargonic). While si is a general word for animal horns, 8, or more commonly, a-darah (Akk. adrii), denotes the elaborate horns of wild goats and deer antlers. See ii-darahq6-ar-nu-umrfa(7)' a-[a-lim] (Proto-Izi I1 20; cited in CAD Q, p. 134b); 53 6-darah, followed by 96 si am and 90 si maBda (NATN 855:1-3 - Ur 111).

25

For this species, see Uerpmann, Ungulate Mammals, pp. 58-63.

26

See ibid., p. 58: "The Mesopotamian fallow deer shows a different form of the basal part of mature antlers, which only have a small brow tine immediately followed by a larger second tine, flattened at its base and sometimes palmated itself. It extends into a palmation along the whole beam, which gives the stags of this species a completely different appearance."

27

For ancient representations of the fallow deer, see Collon, First Impressions, p. 67 fig. 277, p. 189 fig. 915.

28

C an-Sk Beg9-bar ki-Bi?darah-mag, 6 an-Sb Seg9-bar-gim gun-a ki-Si?darah-mag-gim sig7-ga, "temple,

The attestations known to me are: msS M. (AUCT 1 786:3; PDT 130:1, CMET 7 361: 1; Hirose Coll. 28:1), m6L-gal M . (Hirose Coll. 75:2; Jones-Snyder 295). uz M. (SACT 1 94:2; FAOS 16 1347:4; Hirose 266:1), m6S ga M. (SACT 1 91:1), mi6S-g8r M. (AUCT 1 786:4), mi6S-ghr ga M. (Nesbit, Drehem IX:3). As demonstrated by examples in which the animals are qualified as ga, "suckling", this species could be raised in captivity. See, most recently, W. Heimpel, ZA 77 (1987) 30-31; D. T. Potts, The Arabian Gulf in Antiquity, vol. 1 (Oxford, 1990). pp. 133-50. A very important datum for this issue is provided by a recently-published Ur I11 tablet (Metropolitan Museum 17:84 - AS 4lvii/l0), which records the name of a ruler of Makkan: WA-du-um Ili-kin-gi4-a Na-du-be-li Cnsi ~ 6 - ~ aThe n ~fact. that Na-du-be-li I nadiib-'el-i 1 is an unmistakable Amorite name (see G. Buccellati, Amorites, p. 175), assures - in my opinion - that Makkan is to be sought on the southern coast of the Arabian Gulf. Here it may be highly significant that Mani'um, the name of the ruler of Makkan who was captured by Naram-Sin (Ma-ni-u[m] EN ~ c i - ~ a n ~ [ ~ ] - I. J . Gelb and B. Kienast, FAOS 7, p. 89 ii 4-6). is likewise documented as an Amorite name (Gelb, AS 21, p. 223, p. 621 no. 4525). See H. Jungius, "The Arabian Oryx: Its Distribution and Former Habitat in Oman and its Reintroduction", Journal of Oman Studies 8 (1985) 49-64; Uerpmann, Ungulate Mammals, pp. 80-83.

at (its) top a S., at (its) bottom a deer; temple, at (its) top spotted like a S., at (its) bottom yellow like a deer" (G. B. Gragg, TCS 3, p. 170).

Steinkeller 29

30

Sheep and goat terminology in Ur I11 sources from Drehem

Steinkeller

-

2586 udu Lu-lu-bu, 1051 udu Ki-en-girx (include 42 goats) (AUCT 1 405 S 46liv); 70 udu Lu-lubu, 902 udu Ki-en-gir, (include 234 goats) (BIN 3 350 - S 47lix) - both texts involve the Su-gid of URU.SAG.RIG7 and name the same officials; 15 sila4 ga Ki-en-girx, 1 6 kirll ga Ki-en-gir, (Buccellati, Arnorites no. 24:3'-4').

In the early Ur I11 sources from Lagash, the term is udu-girx; in the beginning of Amar-Sin's reign, it was replaced by udu-erne-girx. See the text Barton, Haverford 3 pl. 122 HLC 260, which tallies the native sheep that were available in the Lagash province from 3 32 to 3 47, or perhaps even later: 10,853 udu-girx @-a (i), 11,245 udu-gir, @-a (ii), and 22,653 udu-girx @-a (iv). A similar text, Reisner, Telloh 27, dated to AS 2, lists over 22,300 udu-erne-girx (@-a) (iv 10, v 5). See also sipad udu-girx (MVN 6 48: 1 [before Shulgi], 496:7 [before Shulgi]; ITT 4 7048:l [before Shulgi]; CT 1 6.iv.8 [S 431; Chiera, STA 17 i 8 [S 461; Lafont-Yildiz, TCT 741 edge line 7 [AS 271) and udu-ern[e-g]irx (LafontYildiz, TCT 620 viii 8 [SS 81); udu-erne-gir,-ra (I'IT 3 49152 [IS 21). The only exception to this rule is possibly found in Lambert, TEL 96:6 (transliteration only), which dates to IS 2, and appears to have sipad udu-gir,. At Umma and Ur, the term for "native breed of sheep" is consistently udu-erne-girx. See sipad uduerne-girx-ra-ke4-ne (Jones-Snyder 130:72; MVN 13 618 x 10; UET 3 1504 ii 26,1505 ii 39), udu-ernegir, (YOS 4 237:30,205), x GAN kud-r8 udu-erne-girx-ra (AnOr 1 303:4 - following x GAN kud-ra udu-gi6). For the meaning "native, domestic, indigenous", of girxlgirls (as in durnu-gir15/girx, "native sonldaughter", Ki-en-girx, "native land, erne-girls, "native tongue", and ur-girls, "domestic dog"), secondarily expressed also by erne-girls (as in sila erne-girls, "locallindigenous sila measure"), see my discussion in "Early Political Development in Mesopotamia and the Origins of the Sargonic Empire", in Akkad - The First World Empire: Structure, Ideology, and Traditions, ed. by M. Liverani (Padua, 1993), p. 110-11 n. 9.

In spite of the gloss EMEU-~~GI = [...I in Nabnitu XXIII 231 (MSL 16, p. 218), which led Th. Jacobsen, Studia Pedersen, p. 174 n. 10, to introduce a reading uli-gi-(r) for erne-girx, there is no doubt that the first sign is to be read erne. See sila me-girls in BIN 8 338: 1, which replaces sila erne-girls of other Sargonic sources (e.g., BIN 8 303:4, 306:3; 312:7, 320:1, 341:l). While the gloss u-li of Nabnitu XXIII 231 remains a mystery, it is quite certain that the lost Akkadian equivalent was Suweru or Sumeru. This follows from the fact that the respective section of Nabnitu lists such words as ier:erli, s'urrfi, ZZru, and iarierru. Cf. erne-girls = Su-we-rum in Sag iv 30 (MSL SS 1, p. 24). 31

See M. W. Green and H. J. Nissen, Zeichenliste der archaischen Texte aus Uruk (Berlin, 1987), p. 219 nos. 240-41. For other pre-Ur I11 attestations, see CAD G, p. 126a. Add the Sargonic examples CT 50 1251; OIP 14 143:4, 9, 10, rev. 7'; UET 2 Suppl. 10 i 4, iii 1. In lexical sources: gukkal (MEE 3, p. 67 nos. 18-25 line 52); u ~ u - ~ u - u ~ - ~ ~ ~ U =Dgu-uk-kal-lurn U.@L (Hh. XI11 23); the forerunners have UDU+@L (as in third mill. sources) (MSL 811, p. 8 n. to lines 23-25).

32

, rarely udu gukkal (as in TLB 3 The term for the male sheep is simply gukkal = U D U + ~ very 133:2). The ewe is usually identified as ug gukkal, sometimes as u ~ + (e.g., ~ LAnOr 7 17:9; AUCT 1 499:2, 3; MVN 13 60:6, 7), where U8 replaces the UDU of UDU+@L. The terms for male and female lamb are regularly sila4 gukkal and kirll gukkal, respectively.

33

For this breed, see M. L. Ryder, Sheep and Man (London: Duckworth, 1983), pp. 120, 148, 172, 246, 256, and passim.

34

Attested in, e.g., BIN 3 138:7, 209:1, 2; SACT 1 159:2; AUCT 1 530:4, 6; MVN 13 447:l; FAOS 16 1052 and passim in this source; BAOM 5 (1983) 31 no. 2:3. That udu kun-gid and gukkal are two distinct breeds is demonstrated by texts that list both types together: 1 sila4 gukkal, 1 ug kun-gid

Sheep and goat terminology in Ur 111 sources from Drehem

(BIN 3 138:6-7); 1 gukkal niga sig5, 1 gukkal, 1 kirll kun-gid (BAOM 5, p. 31 no. 2:l-3); and passim in FAOS 16 1052. 35 36

See Ryder, op. cit, pp. 120, 172, 263, and passim. udu-gukkal-ib-18-e in Nippur Forerunner (MSL 811, p. 83 line 15). udu-TU in Nippur Forerunner (MSL 811, p. 83 line 16). PDT 21:l (1 sila4 gal-tab-bu-urn gig-du), 644:9 (1 sila4 ga gal-tab-bu-urn); CMET 7 291:l (1 ug galtab-bu-urn niga), 305:6 (1 ug gal-tab-bu-um); FAOS 16 1052 iii 7 (36 sila4 gal-tab-bu-urn niga) and passim in this text, 1086:3 (1 udu gig-du gal-tab-bu-urn).

5 udu kun-gid gig-du, 1 udu gal-tab-bu-urn giS-du, 2 gukkal gig-dh (iii 19-21), 2 sila4 kun-gid, 6 sila4 gal-tab-bu-urn gun-a, 5 sila4 gukkal (iv 5-7), 3 kirll kun-gid, 3 kirll gal-tab-bu-urn gun-a, 5 kirll gukkal (iv 13-15). 1 US kun-gid niga, 1kirll gal-tab-bu-urn (FAOS 16 1067:2-3); 1 gukkal niga, 1u s gal-tab-bu-urn (CMET 7 3055-6). However, if ib-18 in udu-gukkal-ib-18 = zibbiinu is to be taken as "belt", and connected, following A. L. Oppenheim, JNES 4 (1945) 157, with the belts used to support the tails of fat-tailed sheep (see below n. 44), then a better match for udu gal-tab-bu-urn would probably be zibbiinu. See CAD K, pp. 361-63, under kilzappu (with variants kilizappu, kiltappu, kaltappu, galtappu, gis'tappu, kissappu, and gissappu). To be added is the form girtappu, appearing in ARM 19 104:1-2: 8 KUS UDU.GAL is' gir-tab-ba-tim,(DIN). For kilzappu as a threshing device, see most recently Steinkeller, Iraq 52 (1990) 20. kaltappu, a gold ornament of NB sources, and kaltappu, a writing for GI$.G~R.GUBin a NA royal inscription (CAD K, p. 362 c) and d)). Note, however, that AHw., p. 427a, considers the former item to be a separate lexeme - kaltappu, "ein Ggst. aus Gold." The only way such a connection could be made - and I offer this suggestion with due reservation - is to relate gal-tab-bu-urn to the "carts" that, according to the famous account of Herodotus, were used to support the tails of fat-tailed sheep in ancient Arabia: "One other thing is remarkable enough to deserve a mention - the sheep. There are two kinds, such as are found nowhere else: one kind has such longs tails not less than 4 112 feet - that if they were allowed to trail on the ground, they would develop sores from the constant friction; so to obviate this, the shepherds - who, fortunately, have sufficient knowledge of carpentry - make little carts and fix one of them under the tail of each sheep, to keep it clear of the ground. The other kind have flat tails, eighteen inches broad (Histories 3.115; translation after A. de Selincourt, Herodotus, The Histories [Harmondsworth: Penguin Books, 19541, p. 221). This account was repeated by Aristotle in Historia animaliurn 8.27. Although we lack evidence that would corroborate the existence of such "carts" in ancient times, the use of various devices, such as belts, nets, wooden rods, and even trolleys, to support the fat-tail is well-documented in the Middle East. For example, in his account of the Kazak fat-tails, C. D. Forde, Habitat, Economy and Society: A Geographical Introduction to Ethnology (London: Methuen, 1934), p. 340, writes: "The majority of the Kazak sheep belong to the fat-rumped breed, a large, rather goat-like variety derived from the wild Urial sheep of Asia ... [Its] most remarkable characteristic is the large double cushion of fat which covers and hangs from the buttocks, obscuring the short tail; in a grown sheep after good feeding this fat hangs down in two great ropes dragging on the ground, and may weigh over thirty pounds .... It is greatest in autumn when little wooden trolleys are often fitted to prevent the hanging cushions from dragging on the ground." Cf. also M. Hilzheimer, Antiquity 10 (1936) 202; A. L. Oppenheim, JNES 4 (1945) 157 n. 26. For representations, see Ryder, Sheep and Man, p. 230 fig. 5.15b, reproduced after An Illustrated History of the Sheep and Wool Industry, South African Wool Board (South Africa, 1970) [not accessible to me].

Stei~nkeller

45

Sheep and goat terminology in Ur 111 sources from Drehem

Stcinkeller

Sheep and goat terminology in Ur I11 sources from Drehem

Conceivably, then, gal-tab-bu-um, understood as either a "footstool" or, perhaps more likely, a "threshing board", might denote such a tail-supporting contraption.

Cf. J. S. Cooper's translation, The Curse of Agade (Baltimore, 1983), p. 51, "alu-sheep with long wool."

The same source also mentions Shimashkian (udu SimaSki) and Lullubean (udu L U . D L U - U ~sheep, ) which, like the u d u A.LUM, never appear in Lagash texts. The use of the distinctive Drehem terminology in this document is explained by its recording the deliveries of livestock from the distributive centers of Drehem (Nibruki-ta - ii 2, 12, iv 12, iv 14, vi 4) and Ur @rimki-ta - iii 5). Cf. Steinkeller, JAOS 108 (1988) 199 n. 15.

For the reading of LU.SU(.A) as SimaSki, see Steinkeller, JAOS 108 (1988) 197-202. The conclusive proof will be offered by M. Civil in a forthcoming note (to appear in NABU). See Steinkeller, NABU 1990113. See above n. 45. For this document, which records also Shimashkian sheep, see above n. 45.

46

A.LLJM, following gukkal and preceding udu-kur (MEE 3, p. 67 nos. 18-25 line 53).

47

kul udu A.LUM plus various designations of color and the ways of processing, passim in BIN 9 and 10. See also below nn. 56 and 59.

See H Klengel, RlA 713-4 (1988) 164-68.

48

CAD All, p. 374.

49

The alleged occurrences of *alu in Middle Assyrian sources - written e-lu and a-lu (ibid., p. 374b), almost certainly involve ajalu /jalu, "stag."

This writing is obviously based on the Akkadian lullfi, "man" (CAD L, p. 242a), itself a loanword from the Sumerian Iu-ulu (c*lu-lu), "men."

50

But note that the recension from Emar retains the writing udu A.LUM (Emar VU4, p. 107 lines 9'-12').

51

CAD A12, p. 336.

52

Once udu A.LUM is reclassified as aslu, then the lemma alu of CAD All, pp. 374-75, simply disappears, for its existence is not supported by any other evidence. Note that the examples of "udu-a-luhur-sag", cited ibid., p. 374b, represent actually udu-a-udu-hur-sag, for which see below under 93.3. Further, 1 GAL A.LU ZABAR of an OB Mari text, cited ibid., p. 374b, is a type of container, probably the same as alu D, ibid., p. 378b. For the alleged occurrences in MA sources, see above n. 49.

53

See, e.g., SU.ME for S U . ~ R . M Ecited , by W. G. Lambert, JCS 41 (1989) 27, and HA.ZI.URUDU (e.g., BIN 8 39 i 10) for @A.ZI.IN.URUDU.

54

Nora hene, in the modified spelling as4-lum, the rare value as4 = LAGABxA is very likely a (graphic) reminder of the archaic spelling with A. The use of asq-lum as a writing of aslu, a type of cubit measure (CAD Al2, pp. 336-37), is apparently a secondary development.

55

56

See Klengel, RlA 713-4, pp. 165-66. Add the following occurrences from Shemshara letters: LO Lu-ul-liirn (JCS 42 [I9901 138 SH. 913:10, 149 SII. 859+881:28) and LU LU-ul-li-imki (ibid.. p. 157 SH. 888:?7). See Lu-lu replacing Lu-lu-bi-im, discussed by Steinkeller, FAOS 17, p. 309 note to 1. 14. The alleged spelling Sar-ru-mi-um, occurring in Jones-Snyder 32:1, is almost certainly a misprint. The occurrences I can furnish are: 1 udu S. niga (PDT 1 596:3); 15 udu S. niga (BIN 3 539:ll); 1ug S. (FAOS 16 1168:5); 1 sila4 S. (Jones-Snyder 32:l); 1 sila4 S. (MVN 13 540:l); 1udu S., 1 udu S. gig-du, 4 US S. (AUCT 1 301:4-6 = StOr 9 no. 26); 2 udu 1 sila4 1ug S. (AUCT 2 366 i 7); 2 udu S. gig-du (FAOS 16 1052 iii 25), 5 sila4 S. (ibid. iv lo), 13 kirll S. (ibid. iv 18); 1 udu g. gig-du, 4 ug S. (Fish, Catalogue 295:7-8). See Reisner, Telloh 27; Lau, OBTC 161; ITT 4 7048:l-2 (a-Sag4 dabs-ba sipad udu-gir,, dabs-ba sipad udu-gukkal).

a-Sagq

See UET 3 1504 ii 23, 26, iii 7, 1505 ii 35, 39, iii 17.

Attested already in ED texts: AN-ur-Sb udu-sululp~,"A., (one of) the suluhu-sheep" (WVDOG 45 18 xvi); 1 udu-suluhu (TuM 5 88 i 1). This breed is not mentioned in Ur I11 texts, except for the Lagash PN Lugal-suluhu (H. Limet, Anthroponymie, p. 472; MVN 6, p. 356; 7, p. 180; Gomi-Sato, SNAT 154:3; etc.). However, it does occur in OB texts from Ur: UDU.NITA S ~ G . G (UET ~ 5 111:17, 817:7 courtesy M. Van de Mieroop). In lexical sources: u d u - s f ~ . ~ U(MSL D 811, p. 84 line 57 - Nippur S U D(Hh. XI11 16). Cf. Forerunner to Hh. XIII); U ~ U - S ~ G ~ U ' ~ U ' ~=USU-u = SU-u (= sikidu or the hke) (Hh. XI11 17), which indicates that yet another term for the "long-fleeced sheep" was udu-siki-du. However, this term is not otherwise documented in Mesopotamian sources.

See Jones-Snyder 130; MVM 13 618. The term udu-kur-ra appears already in an ED lexical text (MEE 3, p. 67 nos. 18-25 line 54 - written udu-kur). Cf. udu-kur-ra = im-me-ri 56-di-i in Hh. XI11 35. In economic sources, this term is documented from Sargonic times onwards (OIP 14 127:3, 8, 134:l [u8kur]; CT 50 62:l; HSS 10 171:2, 172 rev. 6'; RTC 244, 245, 246; etc.).

See BIN 9 281, 293, and 440, listing both kuS udu A.LUM babbar and kuS udu babbar, BIN 9 334 and 430, listing both kuS udu A.LUM gi6 and kul udu gi6, and BIN 10 128, listing both kuS udu A.LUM u-haib and kuS udu u-hab.

For the reading of L~.sU(.A)as SimaSki, see above n. 61. This breed is possibly identical with magkur, which appears in the Sargonic texts CT 50 125:3 and HSS 10 176:4.

57

See the preceding note.

58

Here note that the fleece of the long-haired sheep was treated as a distinctive type of fleece, which, like the sheep itself, was called suluhu in Sumerian and sulumhii in Akkadian. See S ~ G . = G su-lum-ku-li, ~ it-qum, "fleece" (Proto-Din 414-15), zu-lum-@ TUG.S~G.SUD= SU-u, it-qu (Din V 131-32) (both cited "5, shekels of in CAD IIJ, p. 299b); 5 GfN ga-bi-im a-na sa-ra-up 5 zu-lu-bi a: T ~ G . N ~ G . B A R A G 'alum' [black-dying agent - same as im-KUG.GI] to dye 5 sulumhii-fleeces for the u~siigarments" (ARM 23 142:l-3).

59

See above n. 29.

E.g., BIN 9 233:10, 334:3, 8.

See YOS 4 237, where the total of 4378 sheep is broken into 1939 udu-gukkal (1. 199), 2150 udueme-gir, (1. 205), and 289 udu-gi6 (1. 21 1). For udu-gi6, see also AnOr 1 303:3-4, which lists the areas of (pasture?) land assigned to the udu-gi6 and udu-eme-gir,. For the conclusion that, at Urnma, udukur-ra are virtually identical with udu-gukkal, see also IIeimpel, BSA 7, 137-8.

In this regard note that mag ~ u l l u m ~ ( ~ U . ~ ~ solely ~ - u r attested n ) , in the Lagash tablet Barton, Haverford 2 pl. 75 HLC 57.iii.16, iv.1, 6, is assuredly an error for m i 5 SimaSki. Since the goats in question were part of the delivery from Drehem (Nibruki-[ta], iv 14), which also included Shimashkian and Lullubean sheep, the Lagash scribe, unfamiliar with the Drehem terminology, apparently confused the two geographical designations. For this document, see also above n. 45. See Steinkeller, SEL 6 (1989) 4-5. See ibid., pp. 4-5. More correctly, the word for "nanny goat" is uzudx@z) (see Steinkeller, MC 4, p. 42), but I use the traditional reading uz for convenience.

Sheep and goat terminology in Ur I11 sources from Drehem

Steinkeller

Cf. uzu-umbin-TAB.TAB = MIN (= e-pi-r[i-turn]) (Hh. XV 235 = MSL 9, p. 13), where the meaning of epiritu (if restored correctly) is unfortunately unknown. See udu-ad4 = kub-bu-lu (FIh. XI11 28); at-t?l ERIN-ten6 = 3 te-nu-u = k~i-bu-lu,ad-*da GRIN-tenii = 4 te-nu-u = KI.MIN (Ea I1 230-31 = MSL 14, p. 257). Cf. Iu-ad4 = hu-um-mu-ru-um (MSE 12, p. 201 line 10).

PLOW ANIMAL INSPECTION RECORDS FROM UR I11 GIRSU AND UMMA

E.g., 1 6b GUR8.GUR8 (MVN 2 168:3); 2 gud GUR8.GUR8 (MVN 13 816:33); 1 gud GUR8.GUR8 babbar niga, 2 gud GURg.GUR8 (Legrain, TRU 80:3, 8).

Wolfgang Heimpel

See CAD K, pp. 563-64; Steinkeller, MC 4, pp. 53-54. In favour of such a connection, note that both gur8-gurg and gur4-gur4 are attested as the spellings of kurkurru (see MC 4, pp. 53-54).

(University of California, Berkeley)

See udu niga sig5 = MIN MIN (= im-me-ru ma-ru-u) darn-qa (Hh. XI11 3). See udu

N - ~ ~ =S MIN E (= im-me-ru) ma-ru-u' (Hh. XI11 2).

Contents

See udu u = im-me-ri :am-mu (tb. XI11 10). So Jacobsen, Copenhagen, p. 8 n. 2; also this author, ASJ 3 (1981) 88. For this conclusion, see already A. Archi and F. Pomponio, CMET 7, p. 89. E.g., AUCT 3 191:1, 1921, 212:l. For girx(GI) as an abbreviation of udu-erne-girx, see H. Waetzoldt, Textilindustrie, p. 6. E.g., PDT 634:15; AUCT 1 492:3. PDT 627:2; Nies, UDT 44:2. P D r 429: 1. E.g., PDT 429:3, 634:7. E.g., AUCT 1 910:l; 2 111:1, 151:l; 3 361 : l . I assume that -gul is syllabic for -gu-lalgal. See udu gir-gu-la = MIN (= im-me-ru) ar-ri (Hh. XI11 4); gir-gal, gir-gu-la = a-ri-tu (CAD ,412, p. 270b). Cf. also Waetzoldt, Textilindustrie, pp. 53-55. E.g., PDT 634:9. For siki-mug, Akk. mukku, see H. Waetzoldt, Textilindustrie, pp. 56-57; CAD M12, pp. 187-88. Note SIG.MUG =: mu-gu in VE 973 (MEE 4, p. 307).

Introductory remarks I Introduction to the Girsu texts 1 Agricultural units 2 Households 3 The partition of households into two divisions and the top administrators of the households 4 Distribution of units among households and diachronic fluctuation of numbers of units 5 Cultivators, bull lieutenants, ten-bull-scribes, and plow bull scribes

11 Types of guq-apin-gub-ba texts 6 The three text types and texts of type B 7 Texts of type A 8 Texts of type C 9 Remarks to individual texts 10 Sample text 12

E.g., PDT 381:2, 383:4, 634:lO. Quite likely, the Umma classification is based not so much on types of breed but rather on types of wool.

111 Oxen and donkeys 11 Animal designations (A), sequence of listing (B), and quantities (C) 12 Excursus on Maekawa's identification of ade-kunga with the Persian Onager 13 Oxen and donkey groups 14 Classification of animals in totals, grand totals, and summaries 15 Working and not working animals 16 Norm of working animals in a group 17 The ideal composition of a group of working animals 18 Not working animals in plow groups 19 Barley of the calves

IV Loss and replacement, withdrawal and supplementation 20 Deaths 21 Death rates 22 Replacement, including a note on the nu gloss 23 Causes of replacements 24 Deficit replacements 25 Substitution Bull. Sumerian Agric. 8 (1995)

Heimpel

Plow animal inspection records 26 Withdrawals 27 Supplementation

V Other administrative procedures of plow animal management 28 Intramural redistribution 29 Redistribution and borrowing between households 30 Redistribution within the household of Ningirsu 3 1 Dissolution of groups and change of managers 32 Inspections 33 gu4-apin-dib-batexts 34 The term dib-ba in gu4-apin-gub-batexts

VI Umma 35 Umma texts 36 Inspections of administrative units and field inspections 37 The districts 38 The inspection sites 39 The bull of Gula 40 guq-Suhub 41 Month dates 42 Text formats 43 The persons of text 8 44 The norm of working animals in Umma 45 Supplements 46 Withdrawals and a deficit replacement 47 Deaths and death rate 48 Replacement 49 Registers of outstanding replacements

VII Conclusions

Plow animal inspection records

l lcilnpel

from Girsu have been noted sporadically but not sytematically treated before. They were called irlter their subscript guq-apin-gub-ba texts by I. Gelb.7 I have found 48 such texts, a number which may seem impressive; but considering that about 100 such texts must have been written I I I a single year, they are really a pitiful lot. Many are also badly preserved, and the chronological distribution of the well preserved texts is not very happy as there are no exemplars of one and the same administrative unit from consecutive years; one text records the animals of all administrative units of the province, but it is a fragment; there is no overlap t~etweentexts of different types; there are few links with the well-researched seed-and-fodder tcxts. The texts from Urnrna are few and much shorter. If there is one wish that emerged from trying to extract knowledge about the management of plow animals in the 21st century B.C. in [he lower alluvium of Tigris and Euphrates, it is that more guq-apin-gub-ba texts will be identified and published. Numbers written with the number stylus are followed by an asterisk * in transliterations. Abbreviations follow M. Sigrist and T. Gomi, The Comprehensive Catalogue of Published Ur I11 Tablets, pp. 81-95, right hand column, without year dates. Abbreviations not found there are: ATL

Agricultural Texts of Ur I11 Lagash published by Maekawa in a series of articles. Texts 1-10 = ASJ 3 (1981) 50-61; 11-23 = ASJ 4 (1982) 108-127; 24-36 = ASJ 8 (1986) 104-120; 37-45 = Zinbun 21 (1986) plates I-XI1 following p.157; 46-58 = ASJ 9 (1987) 113-129; 59-68 = ASJ 11 (1989) 129-141; 69-78 = ASJ 13 (1991) 222-235; 79-92 = ASJ 14 (1992) 228-243; 93-99 = ASJ 15 (1993) 122-129.

Buffalo Bulletin of the Buffalo Society of Natural Sciences 11-2 (1915). SLTNi S.N. Kramer, Sumerian Literary Texts from Nippur (1944). SNAT T. Gomi and S. Sato, Selected Neo-Sumerian Administrative Texts from the British Museum (1990). UTI

F. Ylldlz and T. Gomi, Die Umma Texte aus den archaologischen Museen zu Istanbul, Band (ID) (1993).

In transliterations the sign GANA2-tenfi is abbreviated as Gt. Where deemed helpful the origin and date of a text is indicated behind the quotation. G = Girsu, U = Umma, D = Drehem.

Introductory remarks

A survey of Ur 111 texts shows that oxen and donkeys were used for draft in deep- and seedplowing, harrowing, and the unidentified task designated as "wet place" (ki-durs),l and donkeys and mules for pulling the vehicles of gods, nobility, and royal messengers.2 There exist representations of oxen pulling a threshing sledge from around 3000 B.C.,3 and textual corroboration from the Old Babylonian Farmer's Almanach? but the information on threshing in Ur 111 texts is silent on the use of animal labor.5 References to oxcarts are rare;6 grain and reed were moved by porters and boat. Evidence for the use of any other animal draft, as for example in lifting water, is completely lacking. The aim of this contribution is not so much to amplify on these uses of animal draft but to present information on draft animals in grain cultivation that is contained in records of inspections of such animals by the provincial governments of Umma and Lagash. The texts

Allen Estes checked the references and made many useful suggestions to improve this study. All remaining errors are mine.

Heimpel

Plow animal inspection records

I Introduction to the Girsu texts 1

inspection of the personnel of the household. The personnel was divided into sections with the li)llowing subscripts.

Agricultural units

Ur I11 administration used a basic agricultural unit of approximately 130 hectar (20 bur) of which one half lay fallow during a given year and the other half was cultivated by one cultivator with his three oxdrivers and a group of animals which was mainly used to pull the plow.8 Accordingly, the unit was called "plow"9 or, more often, "bull" in the extended meaning of the word as "plow animal group". The agricultural land of the province of Lagash that fell within the framework of the provincial administration, which alone is the subject matter of Girsu texts, consisted of 600 units: according to ATL 53 (undated) 11,652 bur, or 755 k m 2 , land was designated as ghn-guq kurg-dabs-ba gir-sb-ga guq 600-kam, that is "bull land", or land that was worked with plow animal teams, including the allotments of the personnel of 600 "bulls". In the itemization of assets of land of the province in ATL 52 of IS 5, the 600 unit group was called the "old 600 bulls" and augmented with a few other tracts. A distinct subgroup was formed by 480 units. It was designated in ATL 53 as "(the land worked by) the personnel of the 480 bulls" (gir-s&-gagu4 480), and represented an area of 9424 bur. It appears also in TCT I 629, a balanced account of income and expense of barley of the province in AS 7. The fifth expense item of that text amounted to 8640 gur, or 25380 hl, of "seed, fodder, and wages for the 480 bulls" (Se-numun mur-guq a-hug-gh guq 480-kam).1° In the text fragment TUT 27 (date broken) the 480 unit group was connected with plow animals: 3679 guq-ade-@-aapin-bi 480-iun "3679 various oxen and donkeys representing 480 plows". The high numbers of oxen, sheep, and goats that are recorded in the text suggest that it contained a count of the total livestock under management of the province. If this is correct, the oxen and donkeys of the 480 plows represented the total of plow animals that fell within the frame of the Girsu archlve. Maekawa concluded that the 480 unit group was "managed by the main public institutions headed by the sanga- and the ~abra-officials".llThe units of these "public institutions" were itemized in TUT 5, a record of the expenses for cultivation for the province in S 47. They add up to 419.12 It is not known to which administrative unit or units the remaining 60 or so units belonged. 2

Plow animal inspection records

l lcirnpel

Households

The "main public institutions headed by the sanga- and the Sabra-officials" were administrative entities that were designated with the word 6 "house" and the name of a divinity, a city, or a person. They are usually called temple households,l3 but the term may not fit the household of Namhani, whose name was never written with the divine determinative. The household of Lugal-URUxGt was often called 6 URUxGt, or simply URUXG~M. The households cannot simply be identified with one temple or residence and its possessions, personnel, and land holdings; they were composite administrative entities. Some consisted of a main temple and a number of small temples, or at least architectonic units called 6 "house", that were dispersed throughout the territory of the province. An instructive example is the household of Nanshe as itemized in UDT 58 43 (cf. Fig. 1). The text is the record of an

Figure 1 kS-didli 6 dNanSe

C dNanSe 6-a-ga-su-lim-ka 6 dHendur-sag 6-Sh-ge-ph-da 6 dHendur-sag g ~ n a~! a - d a - s a k - l a ~ ~ 6 d[ ]-ab 6 dNin-gig-zi-da u 6 dAma-gulO-g6 6~ M ~ S . Z A . A N - ~ ~ C dNanSe Pas-enku E-ning-e-gar-ra 6 dNin-x[ E-A-dam-dun Tir-ba-bil-la E-d~an~e-SeS-e-gar-ra 6 Ki-sa-la

ks-didli Sh Ninaki

Sh Gir-suki

was, according to its position at the beginning of the The "house of Nanshe" (6 d~an?ie) list and the fact that the section relating to it is by far the longest, the Sirara, the main temple of Nanshe in Nina.14 Other Nanshe temples were located in other cities and settlements, despite the misleading subscript "various shrines in Nina" (G-didli Sa Ninaki): one in Pas-enku (location unknown), the Eshagepada in Lagash,ls and the Enanshesheshegaral6 and the "house Kisala" in Girsu. In addition to Nanshe temples the list included temples of divinities that were closely associated with Nanshe: according to the Nanshe Hymn Hendursag was her "herald" (nimgir) and the sag-kn-tar (-keeper) of her house17 and according to the Hendursag Hymn he brought butter and cheese for the wedding rite of Nanshe and Nindara.18 The household of Hendursag is attested by itself in IT'I' 211, 865 (cf. section 5 for this text). The Eninegara was the temple of Nanshe's "brother" Ningirsu in Nina.19 The guq-apin-gub-ba texts contain several references for similar household structures. Maekawa noted that one of the top administrators of the household of Ningirsu, a certain Utumu, was titled bishop of Bagara, Bagara being Ningirsu's temple in the city of Lagash.20 The subscript of text 1 listed this and other sub-households: guq-apin gub-ba standing plow bulls of the house of Ningirsu 6 d~in-gir-su-ka house of the commander 6 Sabra

Heimpel

Plow animal inspection records

Plow animal inspection records

1 leimpel

C Ba-g6ra

house Bagara Bbaibbar White House ii 6 d~a-bil-sajj and house of Pabilsag ugula ~ r - ~ ~sanga a - u supervisor bishop Urbau The "house of Ningirsu" was presumably the Eninnu in Girsu; house Bagara Ningirsu's temple in the city of Lagash; White House was also a temple of Ningirsu, its location is unknown;21 Pabilsag was an alter ego of Ningirsu according to the Old Babylonian hymn SLTNi 61 which consists of characterizations that refer equally to Ninurta, Pabilsag, and Ningirsu. An additional sub-household was the house of Ninhursag. It consisted of one unit.22 Maekawa noted also that one of the administrators of the household of Igalima, a certain Urlamma, was titled "bishop of the house of Ninshubur" (sanga ~ . ~ ~ i n - i u b uItr ) . demonstrates that this household, too, was composite. The other administrator, a certain Adda or Addamu, was bishop of Inana.23 The section of text 4 for which this Adda was supervisor included 1 or 2 units that was, or that were, subscribed as guq d~nana"bull(s) of Inana". It demonstrates that not all units under supervision of the bishop of Inana belonged to the subhousehold of Inana.24 The same may have been true for the bishop of Ninshubur. It is strange bat no administrator of the household of Igalima was a bishop of Igalima. I cannot place the following sub-households: "house of Ninsun" with 4 agricultural units according to text 37, "house of Gishbare" with two or more units according to text 34, and "house of Shulpae" also with two or more units according to the same text. In addition to the households listed in Table 4 a household of Shusin is mentioned a few times.25 It is not known whether one of the established households was renamed after Shusin as happened with the household of Namhani which became the household of Arnarsin in the years from AS 3 to 3s 1,26 or whether the household of Shusin was a new creation. An alltogether different type of household may have been that of the governor. It was not divided into two divisions (see next section); the only administrator so far known was a certain Baddari whose title is unknown.27

3

The partition of households into two divisions and the top administrators of the households

For the purpose of administering the agricultural units each household was divided into two parts;28 they will be called "divisions" in this study. Each division was led by a bishop (sanga) or a commander (Sabra). Each administered one half of the number of agricultural units and their plow animal groups (cf. Fig. 2).29 The texts contain several examples that demonstrate strict halving between the divisional heads not only of plow animal groups but also of individual plow animals that came from outside into a household: according to text 2 the household of Gatumdu held 2 donkeys which had been found the year before. One had been assigned to each division (cf. the description of

the case in section 23). Of a transfer of 8 donkeys from the household of Ningishzida to that of Lugal-URUxGt 4 each went to the two divisions (see section 29). Figure 2 text

divisional administrator Urshagamu Urbau Urlamma Adda Dugazida Urigalirn Urbau Lugaluruda Urnumushda Nammah

units

The documentation of the late years of Shulgi allows an almost complete roster of the divisional heads of households. It is assembled in Table 4 together with the uncomplete data from the the guq-apin-gub-ba texts of AS 2. The distribution of bishops and commanders deserves a study of its own. I will only mention the curious case of Utumu. The household of Ningirsu was led in 3 41 by two bishops. Utumu was called expressly bishop of Bagara in ATL 10 of 41 ;31 yet in text 5 of 3 46 a transfer of donkeys to the the division of Urbau of the household of Ningirsu came "from the plows of commander Utumu" (apin d~tu-gu10iabrata).32 The format of texts 6 and 9 indicates the possibility that the two divisional heads of a household were sometimes temporarily subjected to the supervision of a third administrator (cf. section 7).

Distribution of units among households and diachronic fluctuation of numbers of units

4

In terms of agricultural units, the households were of different size, ranging from a high of 98 units of the household of Ningirsu in AS 2 according to HSS 4, 4 to a low of 14 of that of Nindara in 3 47 according to TUT 5. The information is assembled in Table 3. The number of units of a household fluctuated over the years. The numbers for the household of Ningirsu seem to indicate a peak in AS 2 that flattened out one year later. HSS 4, 4 of AS 2 listed 98 cultivators, corresponding to that many agricultural units, under the supervision of a single commander. In light of the numbers of the other texts, it is impossible that he administered only one division of the household. It would mean that the household had a total of 194 units. I assume that the commander of the text served as supervisor of the two divisional heads of the

Heimpel

Plow animal inspection records

household and that the 98 units were the total of the household. The numbers for Nanshe show a similar increase, albeit less extreme, from 50 in 3 47 to 70 in AS 2. By AS 5 they had fallen to 60 (only one division of the household is documented in text 31) at which level they stayed until [SS 11.33 The household of Ninmar experienced a steep increase in units from S 47 to AS 2 if the number of HSS 4 , 4 is not misleading. But already one year later they were back down at 63, and the number 30 in text 32 of an unknown year of AS, or less likely SS, fits perfectly with the 60 units of 3 47 if it referred to one division. Toward the end of the reign of AS, in his 7th year, the number had fallen to 50. The development in the household of Shulgi was similar but the peak in AS 2 was flatter and so was the reduction by [SS 11. The numbers in the households of Dumuzi, NamhanilAmarsin, and Igalima did not fall off after AS 2, but kept increasing, modestly in the case of Dumuzi, substantially in that of Amarsin, and spectacularly in that of Igalima. Those of the households of Gatumdu and Ningishzida did not change. On the whole the numbers indicate an expansion of agricultural land that peaked in AS 2 and declined subsequently. The individual households were affected quite differently by the changes. 5

Cultivators, bull lieutenants, ten-bull-scribes, and plow bull scribes

The great majority of persons appearing in the plow animal inspection texts were cultivators (engar) and bull lieutenants (nu-banda-gu4). Gelb understood them as two "classes of temple personnel", noting that they formed a unit as expressed in the summarizing formula engar nubanda-gu4-me "they are cultivators and bull lieutenants" in TUT 111 and elsewhere, and that the level of their earnings was not dependent on the size of the household as it was for the higher ranks. He also noted the different ratios between their numbers: one bull lieutenant supervised 5 cultivators according to HSS 4, 4 and also 4 or 2 according to the 9 personnel lists that he tabulated in chart 11.34 The evidence concerning these ratios in the guq-apin-gub-ba texts is summarized in Table 2. The ratio of lieutenants to cultivators was specific to households: in those of Ningirsu, Nanshe, Shulgi, Namhani, Lugal-URUxGt, and Dumuzi it was 1 5 , in those of Igalima and Ninmar 1:4, and in those of Gatumdu, Nindara, and the governor 1:2. In the household of Ningishzida it was 1:2 in an unknown year of Shulgi according to text 10 , but 1:5 in AS 2 according to texts 45 and 46. There exist minor deviations from these ratios. For example: in text 5 the first bull lieutenant supervised 4 cultivators, the next 3 bull lieutenants supervised 5 cultivators each, and the last bull lieutenant 6 cultivators; among the 14 bull lieutenants supervising 28 cultivators in text 10 was 1 bull lieutenant who supervised 3 cultivators and another bull lieutenant who supervised only 1 cultivator. The preponderance of the 1:5 ratio agrees with the fact that the number of units in many households was a factor of 5. The evidence of the texts used by Gelb deviates in many points. HSS 4, 4 of AS 2 recorded a ratio of 1 5 , or nearly 1 5 , for all households. For that of Ninmar it was 19 bull lieutenants and 95 cultivators, which is closer to a ratio 1:5 than 1:4 while the ratio 1:4 is attested in no less than 6 plow animal inspection texts from the same year; and in the 4 late personnel lists (AS 8 and SS 1) on the left hand of chart I1 the ratio was uniformly 1:2. It compares with the evidence of the plow animal inspection texts as shown in Fig. 3:

Plow animal inspection records

l lcimpel

Figure 3 household Hendursag Ijagara URUxGt Igalirna

ratio in Gelb and here 1:2 and 1:5 1:2 and ? 1:2 and 1:5 1:2 and 1:4

remarks as part of household of Nanshe Bagara is included in text 1

For the time being it looks as if the ratios were changed to 1:2 after AS 2 and before AS 8.35

Cultivators usually managed one unit and its plow animal group, but occasionally also two. Luurasari and Lugalsipa managed one group in the division of Urnumushda in the household of d ~ u ~ a l -in~AS~ 2 ~andx took ~ tcharge of a second in the other division of the household at the same time according to text 12 (cf. Table 5). The second cultivator in text 5 and the first in text 29 had a double group (6-bi 2 - h ) , and so did the second of text 43 (Ur-la6-ga engar 2) and Lugalshala in text 33 V 1'-5' (gu4-apin-bi 2-am). Cultivators could become supervisors and function like bull lieutenants occasionally. The cases known to me are quoted in section 49. It is not known whether they could become bull lieutenants. "Plow bull scribes" (dub-sar-gu4-apin)and "lo-bull-scribes" (dub-sar-gu4-10)took an intermediary position between bull lieutenants and divisional heads in the hierarchy of the field administration. The former was the higher rank. According to BM 12232 he belonged to the rank of the "elders" (ab-ba-ab-ba)while the 10-bull-scribedid not.36 The latter is well attested in the so-called Round Tablets that refer to the household of Ninrna~-37and in Umma. The texts from Umrna allowed Maekawa to demonstrate that a 10-bull-scribe supervised two bull lieutenants which agrees with the title since the two bull lieutenants in turn supervised 10 cultivators who managed 10 plow animal groups that were called "bullsn.38 In the gyq-apin-gubba texts 10-bull-scribes and plow bull scribes were mentioned occasionally as persons who had to replace plow animals. They may have written the texts that are discussed here, the plow bull scribe the large texts of type A, the 10-bull-scribes the small texts of type B. Bull lieutenants wrote occasionally themselves, as is indicated by the seal legend of Lugalkuzu which identifies him as scribe (see section 49). The involvement of plow bull scribes in recording the receipt of skins of fallen animals is described in section 20. Gelb noted that the ranks of household personnel above the bull lieutenants and cultivators were called "elders". The elders that occur in the gu4-apin-gub-ba texts are the bishops, commanders, and an occasional plow bull scribe. They will be referred to as "adnlinistrators", while the 10-bull-scribes, bull lieutenants, and cultivators will be called "managers".

Heimpel

Plow animal inspection records

I1 Types of gu4-apin-gub-batexts 6

The three text types and texts of type B

The corpus (Table 1) consists of published texts that include records of plow animal groups in the form of lists of oxen and donkeys each of which was associated with the name of a cultivator. They are easily recognized by the round and rounded impressions of the number stylus. Most included the phrase "standing plow bulls (guq-apin gub-ba) of household x" in their subscript.39 They were classified under that phrase on the tags of tablet boxes that contained such texts. Cf. e.g. CT 3, 10 BM 14608 SS 4: pisan-dub-ba tablet box guq-apin gub-ba standing plow bull (records) 6 d~in-marki (of) house Ninmar i-gail are in it The context of the tablet box label inscription shows that the phrase guq-apin-gub-ba could refer to the tablets on which the "standing plow bulls" were registered. This extended meaning may also apply to the subscript of the texts. The scope and format of the texts exhibit variations that divide them into 3 different types. They will be described below separately. Short texts, classified here as type B, recorded groups of plow animals and their cultivators that were managed by a single bull lieutenant. Most of these texts were dated AS 2; several were undated, and one, text 8, was dated S 47. Originally, the texts from AS 2 would have been stored together in their tablet boxes, arranged according to households. The boxes disintegrated over time, but the tablets kept their physical proximity in the ground, loosing it step by step and more or less as they were transferred to the saddlebags of their illicit excavators, the new boxes of antiquities dealers, the shelves of museums, and the publications of Assyriologists. It so happens that most found their way into museums in England. The subscript identified the bull lieutenant of the cultivators that managed the plow animal groups that were registered in the text, the household in which he was serving, and occasionally also the divisional head of that household. It did so in various forms. They are tabulated in Fig. 4. Column 2 refers to the identification of the numbers of plow animal groups that were registered in the text. In text 8 it constituted the first element of the subscript; it was phrased 5 guq, and the numeral was written with the letter stylus in vertical notation. It does not occur in the texts dated AS 2. In texts 41 (not tabulated here because the rest of subscript is not preserved), 45, and 46 it was written on the left edge as last entry, again with the letter stylus, but in horizontal notation. Column 3 refers to the formula guq PN which represented the first statement of the subscript in a few texts. The persons were Ab-ba-gi-na (22), Ab-ba-Sag-ga (24), and ~ r - d u t u dumu ur-d~amma(28). The last person cannot be the bull lieutenant or the head of the division of that name as both were mentioned further on in the subscript. Abbagina and Abbashaga should be the bull lieutenants as these would otherwise not have been recorded.

Plow animal inspection records

l lcimpel

Column 4 refers to the supervising bull lieutenant. He was identified as supervisor (ugula) and by name; his title was occasionally added (the cases are marked as "nbg"). Column 5 refers to the presence or absence of the formula guq-apin gub-ba. In text 23 p4-apin-dib-ba is found instead. The term is discussed below in section 33f. Columns 6-8 refer to the identification of the household in one of 3 forms: by recording in text 14; or by placing the word e the name of its divinity or person, for exainple d~in-marK or gu4 in front of it, for example 6 d~anEein text 8 and guq d~in-marW in text 13. Columns 9 and 10 are complementary, recording the name of the supervising divisional head with the formula ugula "supervisor" or i-dabs "took in charge". Figure 4 text

#

guq PN

ugula 1

gag

B

gu4-

N

i-dabs ugula 2

year

Heimpel

Plow animal inspection records

Plow animal inspection records

l leimpel

[guq-apin gulb-ba [(@ Nam-b]-ni 1-x (not -la, i.e. [A-kall-la,or -sum, i.e. wa-an]-sum) [ [~r-~~almma [enlsi / date

The copies of copyists who appreciated the physical aspects of tablets indicate many erasures of lines, and Reisner observed that one line of text 35 was scratched into the dry clay. It appears that the process of recording these texts was accompanied by much change of information. The last lines were occasionally written onto the left edge. The scribe, presumably the plow bull scribe (dub-sar-gu4-apin),of the household of Ninmar of AS 2 used the left edge for inscribing the name of the supervising bull lieutenant. It may have helped to identify a tablet after it was stacked among others in a tablet box. The sample of texts is too small to determine whether individual plow bull scribes can be recognized by a specific use of the various elements of the subscript.

7

Texts of type A

Texts of type A are long; they recorded the groups of an entire household or, less often, of one of its divisions. Text 12, which is perfectly preserved, has 340 lines in 10 columns; text 31, almost perfectly preserved, has about the same number of lines in 14 columns. Texts 5,9, and 10 are also long and of decent preservation. The rest are fragments large and small. They span the 13 years from 41 to AS 5 rather spottily. One of the few texts from the reign of Urnammu, a small fragment of a large tablet, labeled here text a, belongs to this type. Text formats are tabulated in Fig. 5. Texts registering the plow animal groups of two divisions have a threefold structure. First the groups were itemized according to the divisions (A1 and A2), then their numbers were totaled (B), and finally the text was characterized in a subscript (C). The itemization consisted of a definition of the composition of the plow animal groups in the division and the status of its animals (a sample is presented in section 10). The number of animals was then totaled and the number of groups, separated into oxen and donkey groups, formed by these totals was given. Finally the name of the divisional head was recorded, usually with the formula ugula PN, in texts 6 and 9 (column 4) with the formula guq PN "bulls of PN". Part B consisted of various totals. Texts 6 and 9 adhered to a rigorous separation of itemization and totaling and thus grouped the totals with the grand totals and summaries in part B, while the other texts kept the totals close to the itemization and thus left them in part 1. Totals, grand totals, and summaries were three categories of totaling. They are described in section 14. The subscript contained the key phrase of the texts "standing plow bulls of household X", the name and title of the ruling governor and the year date. In the texts of column 1 and 2 the names of the divisional heads were repeated; the scribe of text 12 (column 3) thought that was unnecessary. Texts 6 and 9 (column 4) pose a problem. Text 9 concluded section A1 with the entry guq Ma-an-sum in V 11, and section A2 with the entry guq A-kan-la] in rev. I11 2'. We expect the two names to be repeated and connected by u "and" in the subscript as in the texts of groups 1 and 2 of Fig. 5. Yet the traces are nor compatible with this expectation. The relevant section (rev. VI) reads as follows:

Figure 5

2 divisions A1

A2

B

c

contents itemization totals of animals # of plow animal groups ugula PN1 bulls of 1st division itemization totals of animals # of plow animal groups ugula PN2 bulls of 2nd division totals grand totals summaries total of groups total of 164 guq-apin gub-ba household PNl u PN2 PN governor Year

1

2

3

+ + + +

+ +

+ +

4

1 division 6 5

+

+

+ + + +

+ + + + - + + + + + + + + + -

-

-

+ - +

+

+ + +

+ + +

+ + +

+ + +

-

-

-

+ + +

+ + +

+ + +

+ +

+ +

-

-

-

+ +

+ +

+ + + + +

+ + + + + + +

+ +

+ + +

+ + +

-

-

-

-

-

-

-

-

-

-

-

-

-

column 1 = texts 2, 4, 11; 2 = 10; 3 = 12; 4 = 6, 9; 5 = 5; 6 = 31, 32

In text 6, the end of part A2 is preserved, and we find, as expected, one of the divisional heads of the household of Shulgi mentioned in the phrase guq ~ u - ~ i Iabra n a ~"bulls ~ of the commander Lunina". The supervisor of the first division should have been Urshugalama who was still divisional head in AS 2. In the subscript we expect again Urshugalama and Lunina,

Heimpel

Plow animal inspection records

but find Lugalsigsu instead. 3 47, the year of text 6, was the year in which Lugalsigsu is first attested as a top administrator in the household. In this text he appears to have been placed above the divisional heads. It must have been a temporary solution because by AS 2 he had succeeded Lunina as colleague of Urshugalama. The tripartite administrative top of the household would explain the make-up of text 9 where the person "( 1-x" was placed above Akala and Mansum, and of the evidence of HSS 4 , 4 where one administrator was in charge of all agricultural units of the household of Ningirsu (cf. section 4). Texts that included only the groups of one division differ from the rest by indicating 16-i amounts (cf. section 16 for their meaning). The texts of type A are not just long, they were also skillfully written and make an aesthetically pleasing impression which is enhanced not little by the use of the number stylus. Their importance was stressed by the choice of the unabbreviated version of the year dates that included the full titles of the reigning king.

8

Texts of typeC

Texts 30 and 37 differ in outward appearance as their scribes did not use the number stylus. In contents, they differ in not including itemizations of plow animal groups and mention of their managers and administrators. Recorded were the totals of animals of a household in sections of assets (gub-ba-h), withdrawals (zi-ga), deaths (ri-ri-ga), and supplements (libir-h), the count of oxen and donkey groups in a household, and the name of household. The designations of the totals correspond to those of the grand totals in texts of type A; yet they differ in including the supplements which were subsumed among the assets in texts of type A. It follows that a text such as 37 could not have been produced by mechanically copying the sections of grand totals of the tablets of the households. Text 37 included some unique features: it used the term i-bi-za "loss" (VII 29-30) and it mentioned several mules besides oxen and donkeys. The text was inscribed on a tablet of 16 columns that probably registered the plow animals of all households, or even the entire 480 unit group. Column I, XV, and XVI are lost, and of many columns only little remains. The date is broken, but there fortunately exists very good circumstantial evidence for it: column VIII 23 listed a dead female mule in the household of Nindara. It is very likely that it is the female mule of the household of Nindara whose skin should have been, but was not, received by the administration in SS 1 according to MVN 5, 180 (cf. section 20 II). Text 30 used the same terminology and a similar format. It recorded the plow animals of the households of Ningirsu and Ninmar in AS 3. In the first section assets and withdrawals were distinguished, in the second there is no distinction. Presumably the recorded animals were all assets. Texts of type C were probably used to establish the number of all plow animals in the province. That this was done is evidenced by TUT 27 (cf. section 1).

l lcirnpel

9

Plow animal inspection records

Remarks to individual texts

'the quality of publication of the texts is very uneven. Text 2 is almost completely preserved and thus a very important source, but it was published with an unreadable photo, no copy, and a problematic transliteration which includes impossible restorations. I have attempted to reconstruct the numbers by comparing itemizations, totals, grand totals and summaries with the following results: obv. I11 9 should be 1hb sag guq-gig-g& because 12 hb-ma + 8 8b-[mhh] of the sccond total = 20 6b-mhh of the grand total. Line 16 of the grand total must be 1 hb-[3]; line 18 6 guq-[I]; line 19 1hb-[I]. The milk calves pose unsolvable problems. 7 + 2 nu mar-ga su-su of the lirst and second total = 9 mar-ga [so-su] of line 20 of the grand total. But there is no equivalent for the 2 6b-amar-ga of the second total in the grand total. I can find only half of the adult donkeys of the grand total in the itemizations and totals. The subscript of text 2 in rev. V 7-10 should be read as follows: gu4-apin gub!-ba!d[gh-tiirn]-du10ugula Ur-Sag-ga[-gulo]ii! ~ r - ~ B a - [ u ] . 'The restoration of the name Gatumdu can be derived from the fact that ~ r - ~ ~dumu a - uUN-il and 1~r-gag-ga-gu10 were the administrators of the household of Gatumdu in S 43 according to ATL 27 (see Table 4). Text 6 was published by Th. Pinches in the Amherst Tablets and again by H. Limet in pigr rap hie I, as # 86. Limet quoted the Amherst text (p.78) but did not realize its identity with the text he was copying. Pinches' copy is better for details, Limet's for a view of the large fragment in its entirety. Text 32 is available only in Lafont's transliteration which contains some problems. oh]-i su-ga LLreplaced deficit" in I 21 and I1 4 cannot be correct. It makes no sense in the context and was never used in wapin-gub-ba texts. Read [duq instead. Column IX must contain the sukilib section, i.e. SU~LAGABXLAGAB (see section 14). To judge by the often hazy distinction between S U ~ L A G Aand B SUXLAGABXLAGAB in copies, the ligatures are not easily distinguished. Column X contained the end of the kilib-ba section; lines 10'-11' must be [guqapinl-bi 22 / [anSe-apinl-bi 8; line 12' should contain the total of deficits as in text 31 in the same position, i.e. @h-i]71 guq 06-i] 11ange. Line 13' is of course [guq-apinl-gub-ba.Line 14'-15' must contain the identification of the head of the division. Text 46: The first sign in line 21 was copied as two verticals written with the letter stylus. The format requires the name of the cultivator for the line. Read therefore A!. The cultivator, A-ab-ba, was dead, ba-TIL. This is the only such record in the corpus. Restore the lines on the left edge: [(guq-16.) d~in-gi]~-zi-da I [ugula U]r-6-ninnu. The numeral on the edge was copied as 2 by Radau. It should have been 5.

10

Sample text 12

This is an almost perfectly preserved text from AS 2 with a record of all plow animal groups of the household of Lugal-URUxGt. Its structure is very even. It is shown in Table 5. The household was divided into two equal divisions that were supervised by Urnumushda and Narnrnah. It will be seen from the material collected in Table 4 that Nammah was already divisional head in S 41. Urnumushda must have suceeded Niurum after S 47. Nammah or Urnumushda was bishop, or both were bishops, because text 12 registered animals that had

Heimpel

Plow animal inspection records

been replaced by "the bishopM.40If the unnamed bishop, or bishops, had been from a different household we would expect that to have been mentioned. The itemization within each division consisted of a list of decriptions of plow animal groups. Each was connected with its managing cultivator whose name and title was recorded. After a block of five groups the name and title of the supervising bull lieutenant was given. 10 cultivators and 2 bull lieutenants represented one division. In four cases, the formula PN engar, which identified the manager of a plow animal groups, was replaced by the formula g w a n i e PN1(-kam) PN2 i-dabs "(they are the) bulls/donkeys of PN1, PN2 took charge of them;" in one case (group 8 in the second division) it was replaced by the formula PN1 i-dabs. The formulas designate change of manager of a group. It is discussed in section 31. The itemization in text 12 follows the basic format that was used for all yearly livestock censuses: assets were recorded first, withdrawals, deaths, and supplements followed in that order. In Table 6 three samples of group descriptions are transliterated and translated. It will be seen that the numbers in the assets and supplements sections were written with the number stylus and those of the withdrawal and deaths sections with the letter stylus. The term "assets" is here used for gub-ba-Am, literally "it is standing", rather than the more common translations "on hand", "available", "present". None of these translations is completely satisfactory because the sections included also animals that were "yet to be replaced" as the "two year" cow in group 9. The designation implies that the animal was not "present", "on hand" etc. at the time of recording. Since the administration was not liable for its replacement the cow counted among its assets. The special status of absent assets was expressed in most texts of type A by placing the sign nu "not" close to the number sign, and in texts of type B by writing the number with the letter stylus. A full treatment of this category of animals is given in connection with the topic of replacement in section 22 below.

I11 Oxen,and donkeys 11

Animal designations (A), sequence of listing (B), and quantities (C)

A The gender and age designations of oxen are well known. I. Gelb demonstrated in his article "Growth of a Herd of Cattle"41 the basic fact that amar-a designated the full first year of life regardless of the time of weaning, mu-1 the second year, etc. Beginning with the fifth year of life the age of oxen (and donkeys) was not expressed anymore; females were Bb-ma and males, in the Drehem text which he discussed, guq-gal. In the Ab-gub-ba texts from Girsu, the males were guq-aib or gyq-giS.42 guq-aib designated the breed bull, the latter a male adult ox which was not used for breeding. In the plow animal inspection texts, guq-gig was the exclusive full designation of the adult male ox. It was often, but not consistently, abbreviated to gu4, preferably when other qualifications followed. Cf. e.g. the following sequences in text 37: hbm a L'cow"..db-mah-su-ga "replaced cow".. guq-gig "yoke bull" .. guq-gig su-ga "replaced yoke bull" .. gyq su-su "bull to be replaced" (III 12-16), or i b - m a .. Bb-ma su-ga .. I b - m a su-su .. g q -

l lcirnpel

Plow animal inspection records

fils .. guq su-ga .. guq su-su (IV 16-21). In text 4 supplemented bulls were called guq 6-tiir na-gabtum-ta "bulls of the pen house, from the NaGABtum" in the itemization while the very same

irnimals were designated gu4-gii na-gab-turn-ta "yoke bulls from the NaGABtum" in the total. While gu4-gii designated always individual animals, guq was also used for general and extended meanings. In the phrase Bb-guq-hi-a it designated any male ox, in guq-apin any plow animal including donkeys, in guq d~nanaplow animal groups. The element gig cannot have been an abbreviation of gig-du which qualified uncastrated male adult animals including oxen in the texts I'rom Drehem.43 It is contradicted by the opposition of guq-gii and guq-6b in the hb-gub-ba texts; we also must assume that the normal plow ox in Sumer was castrated. We might consider that fils in guq-gig meant "penis" and was abbreviated from some phrase which expressed castration. Yet, the term ade-gig contradicts this because ade-giS designated a collective of working plow donkeys including females. It is necessary to stress this point because B. Landsberger and others were, and are, convinced that castration must have been expressed in the terminology of domestic animals, spending much effort to find the expected designations and proposing to have found them. I do not know a single designation whose function was the identification of castrated animals as opposed to noncastrated ones. The distinction, in German, between "Stier" (uncastrated)and "Ochs" (castrated) seems to be restricted to a variety of Hochdeutsch which is somewhat removed from the reality of dialectal German as spoken by those who handle these animals. My attempts to understand the terminology of castration from farmers in the Schwarzwald (southwest Germany) always elicited head-shaking. They divided their oxen not into anatomical, but functional, categories, namely, among the males, draft, meat, and sires; and while these functional categories overlap the distinction of castrated and noncastrated they do so only more or less, but are not identical with them. I suspect that the situation is not much different in the case of "bull" (uncastrated) and "steer" (castrated) in Webster's English as opposed to the terminology of the countryside. In this study I use "bull" as translation of guq and "ox" as designation for a bos taurus. giS in guq-gig is the word for yoke. It is attested in lexical tradition in the equation gii = niru and in Ur 111terminology in the expression gig-i&"(fit) for the yoke". In the hb-gub-ba texts guq-gig can hardly have designated yoke bulls. It was presumably the male adult ox which was usually used as plow ox and kept, for reasons unknown to me, among the dairy cows. "3 year bulls" and "3 year stallions" do not occur in Girsu texts. As far as working is concerned they were functional adults so that the designation was really superfluous. "3 year" females are mentioned, yet their number creates a dip when plotted on a line between "2 year" and adult females, indicating that some of the "3 year" females were designated as adults. In the Umma texts "3 year" males and females occur routinely and in numbers that indicate consistent differentiation from adults. The term iu-giq poses a problem. There exists unambiguous lexical evidence equating guq-iu-giq with Sibu and Bb-iu-giq with Sibtu "old"$4 and this meaning fits the evidence very well. But text 10 VIII 31 recorded a "1 year" iu-gi4 stallion; a "2 year" iu-gi4 stallion is attested in text 1 I1 11-12, and a "2 year" Su-giq mare in the same text XI1 34.45 SU-giq must have designated a quality that applies typically, but not exclusively, to old age, such as "weak". iu-gi4 animals were always listed last, which indicates that they were the least valuable; they are typically found in withdrawal sections; and they did not work except for a single mare which

1 lcirnpel

Heimpel

Plow animal inspection records

was declared fit for the yoke (4 I1 8). The term Su-gi4 was not used in the Urnrna texts; instead, Su was placed in front of the animal designation.46 The term siKKurn stallion designated, in my opinion$7 a mule or donkey that was used to pull the coaches of royal messengers. Entries of siKKum stallions were, with one exception, marked with the "nu gloss" which indicates that they were not present before and at the time of recording (cf. section 22). In two Umma texts the designation za nu-Su "not branded" was added, once to a milk class male donkey (text 7 I1 2: 1dur-ga d nu-Su) and once to a replaced yoke bull (4 111 17'). These cases indicate that plow animals were usually branded. D. Foxvog, in a forthcoming study on branding in Sumerian terminology, quotes texts from Girsu (ITT V 6796; SNAT 25 1; 146) that mention the branding 'coppers' (urudu si-im-da) of individual households. Plow animals were occasionally transferred from one household to another (see section 29). SNAT 146 SS 8 seems to record a case where the question of branded animals arose in such a transfer. [n] anSe [nl donkey(s) guq-apin Ba-g8ra-ka ba-an-si were entered among the plow bulls of the Bagara ki LU-d~tu-ta from Luutu bull lieutenant Urbagara received them Ur-Ba-g8ra nu-banda-gu4 '1-dabs d u ~ ~Ba-ghra supervisor U., bishop of Bagara ugula ~ r - ~ g h - t i i m - sanga si-im-da u ~ u x ~ t - i-ak ba~~ the (branding) mark of48 URUxGt had been applied nu-bhnda-guq-keq igi-na i-lag the bull lieutenant ... Surprising is the absence of the term guq-hb-ur-ra which is well attested in texts recording the purchase of plow oxen (see section 27 below). It is known from a paragraph of the laws of Lipit-Ishtar where it was contrasted with guq-8b-sag-muru4.49For the guq-8b-ur-ra one had to pay 2400 liter barley rent for 2 years while 1800 liter were the rent for guq-ib-sagmuruq. CHs242f. treats the same topic. The first term was then written guq-8-ur-ra, and this is the form that entered lexical tradition where it was translated into the ungrammatical Akkadian form a-lap ar-ku (one expects ar-ku-ri) which served as the basis for the conventional English translation "rear ox".50 The passages quoted in section 27 confirm that the guq-hb-ur-ra was the most valuable plow animal and fetched the highest prices. Civil's interpretation of a plow ox, bull or cow (because of the element gu4-Ab),which was not found in the front (sag) or middle (muruq) position in a team of 3 pairs and thus in the rear as confirmed by the Akkadian translation and the corresponding synonym guq-eger-ra,is very difficult not to accept. But there are problems: the interpretation of the element guq-ab as bull or cow fits the legal context but not the references in section 27 which speak of 1 particular guq-8b-ur-ra;the forms guq-8-ur-ra and guq-ur-ra in post Ur 111tradition indicate that the form guq-8b-ur-rawas not anymore understood at the time and re-interpreted; the form a-lap ar-ku or ar-ki is, as noted, ungrammatical; the silent assumption that the pair closest to the pulled plow should require the best draft animal is not convincing; the OB tradition which seems to include the kind of information that would

Plow animal inspection records

help clarify the nature of the guq-8-ur-ra,as can be seen from the data assembled in M. Stol's contribution in this volume, has not been subjected to contextual interpretation. Note, for example, the letter CT 29,28f. 80186, a reference given by CAD S.V. arkci. The writer says that hc has 3 oxen but that there is no guq-ur-ra. He asks his father to look among his oxen, find a good guq-ur-ra, set its price, and send it. The implication seems to be that one good gu4-ur-ra will make it possible for the writer to use his three oxen, among which there is no gu4-ur-ra, effectively. The hypothesis of guq-iir-ra as one of a pair of rear oxen is not confirmed. #

11 The sequence of listing followed the principle of decreasing value and is mirrored in the

arrangement of Table 7: adult animals were listed before the young, and among the young the older animals came first. The Su-gi4 animals came in last place after the milk class. Within each age group, female preceded male as in the pasturing environment and not as in the fattening cnvironment.51 Female draft oxen may be expected to have ranked below males because of their inferior strength. Yet their potentially higher value is confirmed by SNAT 135 according to which a cow fetched the highest price (8 shekel silver) while 3 bulls fetched 7 each, and 2 more 6 each. According to SNAT 136 a cow and a bull fetched 7 shekel each. The term for oxen of various ages and gender in Girsu texts was 8b-guq-I$-a and thus conformed to the convention of listing the females first; in Umma it was guq-8b-@-aeven if there, too, females preceded males in itemizations.

C For the numbers in column 2 of Table 7 I have used all recoverable numbers from the Umma texts and the numbers from the assets and supplements sections of texts 3,4, 10-12,31, 32, and the assets sections of 5 and 9, from Girsu. Representative data for distribution of species in Girsu are discussed in section 13. In the Umma documentation there were 426 oxen and 157 donkeys, or 73.07% oxen. The ratio of bulls (guq-gig) to cows (bb-mih) in Girsu was 641 to 171, or 78.84% bulls; in Umma it was very similar: 169 to 47, or 78.24% bulls. The ratio of stallions to mares was 379 to 254, or 59.87% stallions; in Umma it was again close: 47 to 43, or 52.22%. 12

Excursus on Maekawa's identification of anSe-klinga with the Persian Onager

The translation "mule" for age-kunga refers to the offspring of a donkey and an onager, as was proposed by J.N. Postgate and is widely accepted now. Maekawa believes that it designated the Persian onager. He discussed his identification of the kunga with the subspecies equus hemionus onager, as opposed to the subspecies hemippus, the Syrian onager, repeatedly, last in ASJ 13 (1991) 206-210. His first argument refers to the anle-kunga that occur as plow animals in the texts Fortsch 195 and 66 of Lugalanda 4 and 5 (tabulated by Maekawa in Zinbun 15 (1978) 100f.) where "about half ... were either completely blind (igi-2-na-bi nu-dug) or wounded in one eye each (igi-I)". He observed: "Injury to the eyes of the AN€E.BARXAN equids should not be regarded as fortuitous, since almost all the male donkeys (AN€E.DUN.GI)and oxen (gu4) which were used as plow animals by the same household had healthy eyes in Lugalanda 4 and 5". And he concludes: "A possible interpretation is that (they) were wounded in their eyes when captured or during being trained as draught animals afterwards".

Heimpel

Plow animal inspection records

The status of the eyes of ade-kunga was also occasionally specified in Ur I11 records. BJRL 64,63 D S 44 IIm'n: 93 a g e kunga-nita 93 male mules 2 ange kunga-nita igi-1 (AS) 2 one-eyed male mules 124 a d e kunga munus 124 female mules 1 anSe kunga-munus igi 2 (DIS.DIS)g6-gar-ra 1 two-eyed female mule mu-DU lugal income of the king In the badly preserved Girsu text f i98 ~is found the entry 1a g e igi-2 (I1 4). While these references indicate that there existed a problem with the eyes of these equids, I cannot follow Maekawa's explanation. It seems to be derived from the underlying assumption that the adult "captured onagers", as he calls them, injured their eyes in the process of being captured or trained. Yet is there any knowledge about whether and how.onagers of any subspecies were caught and trained for the yoke? I find a statement in Brehm's Tierleben (where much information about practices now extinct survives) about the subspecies Kulan which may be of relevance: "Versuche, den Kulan zu ziihmen, sind in seinem Vaterland (Innerasien) selten und nie mit vollstandigem Erfolg angestellt worden. ...Sie Cjung eingefangene und mit Stuten als Stiefmiittern aufgezogene Tiere) ans Einspannen zu gewohnen war unmoglich".52 If it was hopeless to try to train onagers for draft that were caught as foals before weaning it must have been even more hopeless to try it after capturing adult animals as Maekawa assumes. Eye troubles may not have anything to do with capturing and training and may not have been limited to the ade-kunga. M. Greenberg referred me to a passage from the Mishna which implies that eye troubles with donkeys were anticipated: "He who hired a donkey and it went blind (hibriqa) or it was seized for state service. To him he (the owner) says: it is your problem".s3 Greenberg remarks: hibriqa is explained by "(dazzling) light" as though from the root brq "lightning". This has been traditionally taken - I think questionably - as euphemism for blind. I think that it may simply mean "dazzled". Maekawa's three remaining arguments are negative: (1) there exists no positive evidence that kunga were born by donkeys; (2) or that ange-eden-na was the father; (3) and references to kunga foals are almost nil. All three points are correct, but there is circumstantial evidence which supports the existence of mules whose parents were onager stallions and donkey mares. All three animals are attested as being found together in groups as has been mentioned by Postgate, J. Zarins, and Maekawa.54 I find most compelling the evidence of Buffalo 3 U [AS 81, a count of large cattle of the 6-Su-sum-ma55 of the governor of Umma. Among the animals is the following group of which Urshara, a well attested donkey herder (sipa-an~e),~6 was in charge @dab5) : 2 anSe-nita eden-na 2 male onagers 1 a g e kunga-munus mu-2 1 female 2 year mule 2 a g e koinga-nita mu-1 2 female 1 year mules 1 a g e kunga-nita-ga 1 male milk mule foal 3 em-ma 6 donkey mares

Plow animal inspection records

Heimpel

2 erne6 Su-giq 1diir Su-giq 1diir amar-ga

2 old donkey (mares) 1 old donkey stallion 1 male milk donkey foal

The text does not indicate that the mules were the offspring of the onager stallions and donkey mares. In fact, we can trace the male milk mule foal as it was mentioned also in UTI 2135 U AS 8:57 1 male mill< mule 1 a d e kunga-nita-ga ii-tu-da gig.apin 'plow birth' from cultivator Uana ki U2-6-na engar-ta uS-gu10i-dabs uS-.guloreceived it (now) in the E. Sa 6 Su-sum-ma A-a-kal-la 6nsi-ka of governor Aakala @rSa-ta-ku-zu intermediary Shatakuzu sukkal LkdNanna sukkal (sic) vizier Lunanna vizier and scribe Lunanna ii ~ u - ~ ~ a dub-sar nna The fact that the animal was received "from" the cultivator Uana according to UTI 2135 and was a charge of the donkey herder Urshara according to Buffalo 3 clearly indicates the direction of its movement from the cultivator to the donkey herder. Thus it was not born from parents in Urshara's herd. Unfortunately, the phrase 'plow birth' is not clear. Still, I regard the combination of animals in Buffalo 3 as good circumstantial evidence: male onagers, female donkeys, and mules were kept together. Why would they have not crossbred them in a time and in an environment where also donkeys and horses began to be crossbred? For the time being, there is little information on the availability of male onagers, and no information on the techniques of obtaining these animals. A glimpse into the world of wild animal trade, or tribute, is furnished by PDT 560 D &3 9 XI 12: 80 a d e eden-na nita 197 mag-dh nita Ab2-ru-um-ma-ba-an.The same person occurs repeatedly as originator of wild animals. His name was written in almost as many ways as there are references. 20 female wild goats (dara4)58and 20 shapparu animalss9 were received from "Ab2-ru-ma-KU-ba" in AS 5; a young of a gazelle from "Ab2-ru-um-ta-KU-ba"in AS 7;60and another already in 5 47 from king (lugal) '~a-bu-ru-um-rna-~u-ba".~~ If he could deliver 80 male onagers, he must have ruled over people who were adept in capturing, rearing, and working with onagers.

13 Oxen and donkey groups As a rule, plow animal groups consisted of oxen or donkeys, but mixed groups resulted occasionally from substitutions of donkeys for oxen and viceversa. Cf. e.g. the assets of the second group in text 42:

Heimpel

Plow animal inspection records

3 yoke bulls 1* anSe su-ga engar sag! gu4-gS-Sk 1 mare, replaced (by) the cultivator, as substitute for a yoke bull l * dur su-ga sag guq-gig-82 engar 1 stallion replaced, as substitute for a yoke bull, (by) the cultivator 1 guq-giS su-su nu-bhnda-guq 1 yoke bull to be replaced (by) the bull lieutenant gub-ba-iun assets Such replacements are attested in 7 additional cases, all but one62 from texts of type B.63 Presumably the animals of the "wrong" species were soon moved to an appropriate group in the process of intramural redistribution (cf. section 28). The equids among the plow animzls were called aGe "donkeys". The word was used as a general term for equids and as a specific term for donkeys (Equus asinus) so that it is not always clear whether "donkeys" were donkeys. The only other equid that can be expected is the mule. A few mules (an3e-kunga) were listed in text 37 (cf. Fig. 6): 3* guq-gig

Figure 6

household Nanshe Nindara Dumuzi

male

.female

There is no proof that the mules were used for plowing even if they appear among plow animals. They may have been identical with the "donkeys" of the siKKum which pulled chariots. The appearance of mules in text 37 assures us that the equids that were not called mules in that text were donkeys. Since the mules are attested in such small numbers, we can be sure that the "donkeys" of the other texts were donkeys. Overall, there were more oxen than donkey groups. In the extant parts of text 37 the species of plow animal of 311 groups can be identified. 176 consisted of oxen and 135 of donkeys for a ratio of roughly 4:3. In the individual households the ratio was widely different. In that of Ninmar (according to text 32) and Gatumdu it was 6:l or 7:l. In the household of Nanshe in the early reign of Urnammu it must have still been higher in favor of oxen. In the preserved summary of individual animals 125 (and a half) oxen contrast with 14 (and a half plus possibly a few more in the broken section) donkeys. According to text 37 the ratio was approximately 5:2 in [SS 11. In the household of Ningishzida on the other hand it was about 1:2. The case of this household is interesting because the unusually high number of donkey groups is attested in texts from 3 34 to SS 1, demonstrating that it was a characteristic of the household. The text from S 34 is CT 10,34f. 15322, a seed-and-fodder text which recorded 11 calves and 28 foals as recipients of "barley of the calves" (Se-amar-ra). The household of Arnarsin had a ratio of about 1:l with slightly more donkey groups. "House Amarsin" was, as

Heimpel

Plow animal inspection records

Maekawa demonstrated,64 the name of the household of Narnhani between AS 3 and Shusin 1. Is it coincidence that the households that were presumably founded in the period of the 2nd dynasty of Lagash, i.e. Ningishzida by Gudea and Namhani by the last independent governor, had so many donkeys? There exists a noteworthy correlation with plot categories in the household of Ningishzida. According to the numbers of TUT 5 47 (cf. Maekawa's tabulation in BSA 1, 90f.) this household cultivated an unusually high percentage of category A land, that is land where 10 furrows per nindan (6m) were seeded with plow animals. It means that donkey groups must have been used to cultivate category A land. If we ask whether donkeys also plowed category B land, that is land where 12 furrows per nindan were seeded with plow animals, the numbers that are available today fail to give an answer. The oxen groups could have easily plowed all category B land in all households. In the household of Dumuzi they would have been quite busy but, judging on the basis of the numbers in TLJT 5 and text 37, they could have done the work by themselves: 56.3% of the land was of category A and 43.7% of category B; 44.74% of the plow groups were oxen and 55.26% donkeys. If we assume that a team of oxen plowed as much as a team of donkeys we can see that the oxen plows could have plowed all category B land. The guq-apin-gub-ba texts contain no information on the particular type of work done by these animals. It is known from a variety of documents that plow animals were used for plowing, harrowing, and seeding. In texts from Girsu that record labor expenses for tilling and seeding,65 work teams were called invariably guq, that is guq-giS.ur "harrow bull", guqTUG2.KIN "deep-plow bull", guq-ki-durg, "wet land bull", and guq-numun "seed bull". The designation "bull" designated draft animal teams of both species. If we ask whether donkeys were preferred for the presumably lighter task of seeding, the answer must be negative because of the lack of evidence for a division of labor between tilling and seeding within agricultural units. The administration and management of cereal cultivation was based on the one to one connection between cultivator and plow animal group throughout the season of soil preparation and seeding. The lack of mixed plow groups indicates that the donkeys performed the same tasks as the oxen. Donkeys are not as strong as oxen and essentially disappear as plow animals after the Ur I11 period from Babylonian documentation. Why were so many donkeys used as plow animals in the Ur III period? They seem to have reproduced more successfully. According to text 12 the 'donkey household' of Ningishzida had generated a surplus of donkeys so that 8 of them could be transferred to the household of Lugal-URUxGt (see below under redistribution among households) and, judging by the numbers of working animals in text 4 (cf. Table 9) and a transfer to another household, the household of Igalima had developed a surplus of donkeys in S 45.

14

Classification of animals in totals, grand totals, and summaries

Text 12, and most other texts of type A, used a threefold system of-totaling and summarizing animals by which they were reclassified into successively fewer categories. Table 8 shows how it was done. The totals of the category that is encountered first in a text will be called "totals"

Heimpel

Plow animaI inspection records

here. They were written with a ligature of the signs Su and LAGAB. It is commonly read Su@in, but since this transliteration does not account for the ligature I use the transliteration Sunigin instead. Most designations found in the itemizations are also found in the totals. Left out were the often rather lengthy designations for causes of replacement, the identification of the liable persons, and the time elapsed since commencement of liability. For example the "2 year cow yet to be replaced since last year (by) the cultivator (and) bull lieutenant" of the assets section of group 9 became a "2 year cow to be replaced" (Bb-2 su-su) in the total. Also, the notation "nu" was not repeated. Thus the "non(present) siKKum stallion" of the assets section of g o u p 6 became simply a '%iKKumstallion" in the total. The two total (Sunigin) sections which concluded the descriptions of the divisions were combined to a grand total. The sign for grand total is a ligature of the signs Su and N I ~ I N The . sign N I ~ I Nitself is a ligature of two LAGAB signs. The whole complex is commonly transliterated Su-nigin. It is conceivable, but unlikely, that the two kinds of totals which had clearly defined separate functions were expressed with the same word in two different writings. I therefore assume that the ligature of LAGABLAGAB was pronounced kilib and transliterate the whole ligature complex as Sukilib. In grand totals the designations of substitutions were left out, the "siKKum donkeys" became simply "donkeys", and the category of supplements was integrated into the category of assets. In a third step, the grand total was summarized. Summaries were called NI~IN-bawhich is read commonly kilib-ba. At this stage only 3 categories per species were distinguished. They will be discussed next. 15

Plow animal inspection records

2ade

5 diir 1 dur-1

1 ade-mar-ga 1 diir Su-gi4 giS-56 1 dur Su-giq gub-ba-8m Kag-a-tur engar

2 mares 5 stallions 1 one year stallion 1 milk mare foal 1 Su-gi4 stallion (fit) for the yoke 1 Su-giq stallion is standing Katur cultivator

The division of working and not working animals in plow animal groups is tabulated in Fig. 7: Figure 7

working

-

in summaries

in grand totals

Bb-gu4-giS1anSe-giS

Bb(-m6h)lade ,guq-giSldur

I

BblanSe/guq/dur-Su-giq gig-&

not working

mar-guq-@-a

BbIanSe-2, guddiir-2

Working and not working animals

The categories of the summaries were (1) rib-guq-gig "yoke cows (and) bulls" or anSe-giS "yoke donkeys9,66 (2) amar-guq-M-a various calves" or amar-anSe-hi-a "various foals", and (3) guq/Bb/an8e/dur Su-giq "Su-gi4 cows/bullslmares/stallions". By matching the numbers of these categories with the corresponding numbers of the animals of the more numerous categories in the grand totals we can fix their semantic range: any h-giq animal in the grand totals remained a Su-giq animal in the summary. "Milk calflfoals" and "1 year" animals were classified as calveslfoals. The 8 amar-guq-@-aof the summary of text 12 correspond to 1 Bb-1 + 4 guq-1 + 3 Bb-amar-ga of the grand total. The "two year" animals and adults of the grand total were summarized as working oxen, i.e. Bb-guq-gig It is surprising that all "two year" animals were classified as working animals because of the existence of "two year" animals which were qualified as gig-Sk "(fit) for the yoke". This qualification makes only sense if the "two year" animals that were not so qualified were not fit for the yoke and did not work. Comparison between grand total and summary of texts 5, 10, and 31 (cf. Table 9 for the first two texts) shows that, contrary to text 12, the unqualified "2 year" animals counted as expected among the calveslfoals (amar) and those qualified as gi8-~& "for the yoke" among the working animals. Su-gi4 animals were also, but rarely, qualified as gig-R "for the yoke". There are two references in the corpus, namely text 37 VID 14 and the description of group 4 in text 4: 66

Only text 12 deviated from this scheme by counting "two year" animals as working animals.

16

Norm of working animals in a group

The summary of the number of plow animal groups in text 1 reads as follows: the plow bull (groups) are 15 guq-apin-bi 15-im the plow donkey (groups) are 17 anSe-apin-bi 17-8m 7 bulls each, 8 donkeys each. guq 7-ta a 6 e 8-ta In the individual group descriptions of the text the number of working animals was in most cases less than 7 oxen or 8 donkeys. The summary statement can therefore not be a characterization of the plow animal groups as they appeared in the text, but should be a norm. Its existence can be demonstrated with the help of text 31. In this text, the notation 16-i plus number was occasionally inserted at the end of a group description, immediately before the name and title of the cultivator. It did not, as in the other texts, designate deaths. The sum of the assets and supplements of working animals and the 16-i number equal the norm of 7 oxen or 8 donkeys. Cf. e.g. the description of the 9th group (LU 3-10):

Heimpel

Plow animal inspection records

2 mares 3* dur 3 stallions l*dur su-ga engar 1 stallion replaced (by) the cultivator gub-ba-hm assets 1 dur ba-TIL 1 stallion dead libir-hm is old 16-i 2 deficit 2 d~tu-gu10 engar Utumu cultivator The six working donkeys of the assets section plus the 2 unspecified animals of the deficit add up to 8, the norm for donkeys. The numbers of working animals in assets and supplements sections and the deficit number of fully preserved group descriptions are given in Table 10. They add up to the norm without exception. Judged on the basis of closeness to the norm, the guq-apin-gub-ba texts form a rather diverse group. There are some texts in which the majority of groups is below norm. Such a text is # 1. The numbers of working animals per fully preserved group description in the sequence of the text are (asterisks identify donkey groups): 4-6-8*-7-5*-6-4-5*-3-5*-6-7-5*-2*-2-6*-3. Three groups conform to the norm, the others are below norm, several considerably so. Text 2 has groups that correspond to the norm or are slightly below or above it: 8-6-7-7-6-5-7*-6-8. Text 4 has the least consistent numbers. There were 6 groups with numbers above the norm, 3 at norrn, and 9 below norm. 2* erne6

17

The ideal composition of a group of working animals

There probably existed an ideal of working animals in a plow animal group. Such an ideal would have been realized when a group was formed from scratch and in one stroke under conditions of unlimited supply. According to text 4, two newly formed groups consisted of "6 bulls of the house of pens from the NaGABtum" (6* gu4 6-thr na-gib-tum-ta); according to text 12 another pair of newly formed groups consisted of 8 stallions each, 4 from the NaGABtum and 4 from another household. The two donkey groups may well represent an ideal: an all male group of 8. The two oxen groups of 6 males were 1 animal short of the norm and could thus have been only partial realizations of an ideal. Among the great variety of compositions of oxen groups one can perceive a tendency to place one cow into oxen groups. The best data come from text 12. 7 of its 13 oxen groups consisted of six bulls and one cow; the details of supplementation and substitution are as follows (I and I1 refer to the divisions of the household, arabic numbers to the groups within these divisions): I 1: the one cow was a supplement I 8: there was one cow, the supplements were bulls 19:sameasI8 115: sameas1 1 II 7: the one cow was substitute for a bull

Plow animal inspection records

Heimpel

II8:sameasIl 119: same as 117 Other groups show that there was no strict rule. All male oxen groups existed, as I 5 and 114, and these groups were formed partially by supplementation which would have allowed addition of a cow. The seeming preference of groups of 6 bulls and 1 cow is intriguing insofar as it may be connected to the curious fact that the norm of working oxen in a plow animal group was an odd number.

18

Not working animals in the plow groups

The numbers in Fig. 8 for animals that were part of plow animal groups but too young for work come from the summary sections of well preserved texts of type A. The numbers in ( ) refer to the working animals.67 Figure 8

text

donkeys #

%

1 (14) 16 (70) 16 (70) 66 (124)

7.1 22.9 22.9 53.2

4 (15) 2 (63) 3 (78)

26.7 3.1 3.9

On the average, about one fifth of the plow oxen in a household were too young for work. But the numbers are quite disparate and suggest that the population of young animals in the groups was conditioned by a complex set of circumstances. The numbers for donkeys are even more disparate; more than half of the donkeys of the 'donkey household' of Ningishzida (text 10) were too young to work; according to text 12, on the other hand, hardly any young donkeys were kept. In the grand totals of the same texts the animals were distinguished by age and gender. The distribution is tabulated in Fig. 9. 2, 1, ga, designate "two year", "one year", and milk class animals; 37a the section of the household of Nanshe, 37b that of Shulgi).68 The numbers suggest a pronounced attrition rate. Among the female oxen and the donkeys, there were more milk class than "one year" animals, and more "one year" than "two year animals". Especially dramatic was the decline from "one year" to "two year" donkeys of the household of Ningishzida in [ ] according to text 10. Most "two year olds" seem to have entered the work force early. The high numbers for "one year" male oxen is remarkable.

,

-

Heimpel

Plow animal inspection records

calves was recorded. Correlation of the two sets of numbers yield an approximate, but not a fixed, proportion. Cf. the 5 sets of numbers in Fig. 10:

Figure 9

female oxen

male oxen

text 2

2 1

1 3

ga 9

2

5 10 11 31 37a 37b

3 2

3 4

3 2

-

-

4 2

2 8 8

2 4 4 3 12 7

sum

16

32

47

female donkeys

Figure 10

male donkeys

1 7

ga

2

1

ga

2

1

ga

-

-

-

-

-

-

-

3 7 8 1 8

1 1

1 16

4 18

2 1

2 11

6 19

-

-

-

-

-

2

2 1 3

1

4

2 1 4

6

4 5 4 8 10 11

10

[

23

58

44

8

-

2 8

1 2

I 21

30

8

22

39

Bu-gi4 oxen and donkeys occur steadily and in small numbers, many in withdrawal

sections, none in deaths sections. The lack in deaths sections is noteworthy considering the fact that most h-gi4 animals were probably old animals. It indicates that they were not kept in a plow animal group but withdrawn shortly after exhibiting the negative quality described as Bugi4-

19

Plow animal inspection records

l lcimpel

Barley of the calves

It is plausible that "2 year" animals were kept in the plow animal groups: they stood in the 3rd year of their life and would soon join the workforce. But why were milk class and h-gi4 animals kept there? As indicated in the previous section, ru-& animals were quickly withdrawn and so need not have caused an additional expense to the cultivator. Concerning the "one year" and milk class animals, we might speculate that their stay in the plow animal group, their growing familiarity with conspecifics, oxdrivers, and the rhythm of agricultural life generally primed them for later work.69 Yet it is hard to imagine that the benefit of such priming outweighed the considerable expense of feeding and caring for the not inconsiderable numbers of young animals in the group. Of course, they had to be fed and cared for by somebody. It could have been done within the confines and responsibilty of the NaGABtum, and it is likely that many of the adult animals that came from that facility to the plows were indeed fed and cared for there for some time. Yet others, begining from their first year of life, were turned over to cultivators and became part of their plow animal groups. The cultivators presumably fed them and cared for them. They also received barley, called "barley of the calves" (Be amar-ra), probably in partial compensation for the expenses. In the seed-and-fodder texts the barley of the calves was routinely listed after barley expenses for seed, plow animal fodder, and wages of human labor in the formula "n calves of n liter each". In TUT 5 S 47, the number of plow groups per household and the barley of the

household Ningirsu commander Nanshe Ninmar Shulgi

units 62 40 50 60 50

calves 21 10 30 10 18

On the average, about 1 calf corresponded to 3 agricultural units. Since most groups had non working young animals, and many of them several, only a fraction can have been eligible for the "barley of the calves". Text 6 in our corpus is dated to the same year as TUT 5 : it recorded the groups of the household of Shulgi. The text is a large fragment of a tablet with 16 columns but the sections of the totals and grand totals are so fortunately preserved that all subadult animals are accounted for. Among them we can look for the 18 calves of TUT 5. The subadults in the totals are: XI1 2'-8': 3 two year bulls 3 ~ 4 2 1 i b 2 sag guq-glH-lb* 1 two year cow, substitute for a yoke bull

1 guq-gi~sag gu4-2-Sb* 1 yoke bull, substitute for a 2 year bull 2 ab 1

2 one year cows

6 ~ 4 1 9 Bb amar-ga*

6 one year bulls 9 cow milk calves 6 bull milk calves

6 guq amar-ga*

XIV 6'-13': 1a d e 2 4 ade 1 11.5 dur su-su* 1 dur 2 gig-Sb* 1 a d e 2 Su-su* 2 dur 1 4 anSe amar-ga* 2 diir amar-ga*

1two year mare 4 one year mares 11.5 stallions, to be replaced 1 two year stallion for the yoke 1 two year mare, to be replaced 2 one year stallions 4 mare milk foals 2 stallion milk foals

Heimpel

Plow animal inspection records

The list contains 42 subadults. The only obvious group of 18 among them are the unqualified "1" and "2 year" animals (indicated by asterisk). The amount of barley per calf was 120 liters in seed-and-fodder texts of the years 3 337O, 3 3671, and S 4472, and 90 in the texts of 4173, 3 4374, S 4575, 3 4776, and AS 177. Some texts itemized the different rates for calves and foals. Figure 11 text

date

L.965478 SNAT 75

[

1

AS 2

Plow animal inspection records

l lcimpel

deficit 1 mare deficit 1 two year stallion la-i 1 dur-2 Texts 12 and 15 marked deficit constituting deaths by using the designations ba-TIL for the singular and ri-ri-ga for the plural.80 Cf. the following examples from text 12. 17: 16-i 1 guq-giS ba-TIL deficit 1 yoke bull dead 16-i 1 a d e

1114-5: calves

foals

120 120

90 60

There is no evidence that calf and foal fodder was given to the cultivators above and beyond the barley of the calves, so that we can assume that it represented the annual total. I do not have any clear criteria by which to measure the degree to which 120 liter for oxen, and 90 or 6 0 liter for donkeys, covered the expenses. The young, as well as all other, oxen and donkeys had probably suitable pasture in fallow fields and the nearby marsh, and stabling and herding may have been a minimal expense, so that one third or one fourth liter fodder daily throughout the year may have been a fair compensation.

IV Loss and replacement, withdrawal and supplementation 20 Deaths The term "deaths" is used here in a technical sense for animals that died of causes that exempted their caretakers from the liability to replace them. Three aspects are treated in this section: I the notation of deaths, I1 the requirement of restoring the skin of the dead animal to the administration, and III the cause of death that constituted deaths.

I As in other animal inspection records, deaths were recorded in a separate section that was placed after the withdrawal section. The numbers of dead animals were written with the letter stylus. The texts used several designations and notations. Most introduced the deaths section with the formula 16-i "deficit" and ended it with 164-gm.79 Cf. e.g. 1 I 9-13: la-i 1 6b-ma deficit 1 cow 1 guq-gig 1 yoke bull 1 Bb-3 1 three year cow 1 three year bull 1 gu4-3 18-i-gm is deficit If the section consisted of a single entry, the concluding 16-i-gmwas often left out. In text 5 16-i-h was left out but the initial 1B-i was repeated in every entry. Cf. e.g. V 14-15:

I 19-22:

18-i 2 anSe-2

deficit 2 two year mares

ri-ri-ga

dead

18-i 1 hb-me

deficit 1 COW 2 yoke bulls 1 one-year cow for the yoke

2 guq-gig

1 rib-1

dead In texts 31, 32, 34, 37,38, and 47, la-i was not used as notation for deaths. As before, death of one animal was indicated with ba-TIL, death of 2 or more with ri-ri-ga. The three types of notation follow each other chronologically. The first type was used from 3 41 to AS 2, the second in AS 2, the third in AS 5. The use of the formula 1B-i "deficit" in deaths sections betrays their original function: they constituted the place in the form where animals were registered that counted as deficit for the administration. Normally the deficit consisted exclusively of deaths, and accordingly the sections were, in effect, deaths sections. For an example of a combination of deaths and other deficit cf. section 24. 18-i ri-ri-ga

I1 The skin of a dead plow animal that belonged in the category of deaths went to the administration; that of a dead animal that was to be replaced did not. The principle is clear and logical and needs no proof. Yet since the proof is included in the information of two Urnrna texts it shall not be suppressed. A plow animal group of a cultivator named Aakala was registered in (Urnma) text 7.*l It includes the following two entries: dead: 1 yoke stallion, to be replaced (by) the cultivator TIL 1 dur-giS su-su engar dead: 1 two year mare TIL 1 em% mu-2 The second animal counted in the category of deaths, and its skin had to be delivered to the administration. In the comprehensive deaths register of the same year Aakala was indeed registered as the source of 1 donkey hide.82 ~L 86 and MVN 5, 180 from Girsu illustrate that the expense of plow animals by the administration to cultivators "for cause of having died" required the receipt of the hides of the fallen animal, and that the plow bull scribe (dub-sar-guqapin) was responsible for producing the written receipts. The texts seem to be audit reports. &L 86: 1 bull 1 donkey 1 guq 1 a d e AS 8 mu en ~ r i d u ~ 2 bulls 2 donkeys 2 guq 2 a d e Ss 1 mu dSu-d~uenlugal

Heimpel

Plow animal inspection records

16 guq 6 anSe mu mai dara-abzu

16 bulls 6 donkeys Ss 2

ba-ab-dug mu ri-ri-ga-S& engar-ra ba-na-zi kul-bi la-gid-da-S& guq anie-bi lul-la dub-sar guq-apin su-su-dam 6d~mar-d~uen

for cause of having died were expended for cultivator(s) because their skins were not received the plow bull scribe has to replace these bulls and donkeys ... house Amarsin

Plow animal inspection records

l leimpel

1 male goat for the "oath taking of bull lieutenant(s) and cultivators" (nam4rim kug-da nu-bandaguq u engar-e-ne).84The oath may be interpreted as an annual ceremony when bull lieutenants and cultivators swore that they did not misrepresent the death of an animal in their charge.

21 Death rates In order to get an idea about death rates among plow animals, I have added the numbers of assets, withdrawals, and deaths from grand totals and summaries of oxen and donkeys of suitable texts and computed the percentage of death from deaths sections and replacement entries in Fig. 12. The death rate appears to be high. Yet the data are thin and may not be representative.

The same type of irregularity was uncovered in the household of Nindara. Again the plow bull scribe was found liable.83

111The principal criterion for deaths was defined as 8-sig = asakku. Cf. MVN 7 , 2 0 3 S 30 (or IS 5 ) : 1 guq ba-TIL 1 bull, dead guq 8-sig-ga ba-TIL-a M a , the son of the gardener, nam-6rim-bi swore 6 d~in-marld-ka in the house of Ninmar Al-la dumu Sandana to the effect in-kug that the bull died of Asakku zi-ga expense Da-da of Dada Asakku was not a specific cause of death or a specific disease (or a specific demon) but a general term for the kind of death that could not be blamed on the caretaker of the dead animal. For example, the deaths among the goats of the palace and the temple households of Girsu in the uzud-gub-ba text DAS 50 IS 3 was characterized in toto as caused by Asakku: kilib-ba 225 uzud-hi-a ri-ri-ga 8-sig "summary 225 various goats, deaths, Asakku". The same animals were qualified as ri-ri-ga or ri-ri-ga 8-sig in the itemization. The same inconsistency in designation is found in the Bb-gub-ba text TCT I 740. Compare e.g. ri-ri-ga 8-sig in the itemization in column VI 19 and ri-ri-ga in the corresponding subtotal in line 30. There was obviously no difference between deaths and deaths caused by Asakku. MVN 7 , 203 demonstrates that there was room for fraud in the determination of the cause of death of livestock, as might be expected. The cultivator or his oxdrivers were normally present when the animal died, and given the choice between having to replace it - which could cost as much as 8 shekel of silver or 24 hektoliter of barley - or returning its skin he must have preferred the latter. Whatever really happened to the animal could easily be obscured by skinning the animal, and the oath remained as only recourse for the administration. A passage from an Umrna text suggests that the administration attempted to shield itself in some measure against fraud by involving the bull lieutenant. SET 130:313-3 15 AS 4 recorded the expense of

Figure 12

22

text

oxen

donkeys

2 4 5 6 12 27

14.2 12.9 14.6

15 19.8 5.9 23.8 14.3 7.9

13.3 18.2

Replacement, including a note on the nu gloss

Replacement was a serious issue which took up much room in the texts. It could impact severely on the fortunes of managers and administrators from cultivator to bishop. We have seen above that an audit revealed irregularities in the documentation of the receipt of skins of fallen plow animals and resulted in plow bull scribes having to replace dozens of animals (section 20 11). The impact on cultivators is possibly best captured in the subscript of the Umma text SNAT 509 SS 6 which listed amounts of barley that were taken out of the allotments of cultivators and withdrawn from the wages for work done by their wives and children: wives and children of cultivators dam dumu engar 6-hul-a-keq-ne of ruined houses for cause of having to replace bulls mu guq su-su-da-S& The designations of replacement specified a time relative to the incidence which triggered the obligation to replace, the liable person, and sometimes the cause of replacement. If the obligation to replace an animal was satisfied, the animal was called "replaced" su-ga. If the obligation to replace was established and the animal was not replaced by the time of recording, it was called su-su "to be replaced". After one year it became su-su nu-su im-ma, literally "to be replaced, not replaced, last year", in effect "one year overdue". This status is frequently attested and seems to have been commonplace. In text a we find su nu-su instead of su-su nu-su and a

Heimpel

Plow animal inspection records

contrast beween su nu-su libir and su nu-su im-ma. There is too little of the text preserved for a meaningful context, so that we are left with the literal meaning which is problematic ("oldAast year's not replaced replacing"?). Animals that were 2 and 3 years overdue are attested rarely.85 Text 26 rev. 5 may even have recorded an animal that was 4 years overdue.86 The managers and administrators of a household, from the cultivator to the divisional head, and occasionally also persons from outside the household (the NaGABtum, repsectively its scribe, a man of the siKKum, and a shepherd) could be found liable for replacement. Their rank and the frequency of incidence of their responsibility is summarized in Table 11. Frequency of incidence correlated with rank in the following way: as a group, cultivators replaced most often, bull lieutenants less, bishops and commanders even less, and plow bull scribes, 10-bull-scribes, and others least. If we adjust for the relative number of individuals in these ranks, and leaving aside cases of shared responsibilty, the situation changes. The corpus includes 107 replacements of cultivators and 35 of bull lieutenants. Assuming sin average of 4 cultivators per bull lieutenant, a conservative estimate, the individual bull lieutenant was more likely to become liable for replacement than the individual cultivator. There were on average 3.8 bull lieutenants to 1 divisional head in our corpus. Divisional heads replaced 18 animals, less than one half the number of bull lieutenants. The individual divisional head was clearly more likely to become liable for replacement than the individual bull lieutenant. If the numbers are not grossly unrepresentative we can conclude: the higher the rank the costlier the responsibility of replacement. Often, cultivators and bull lieutenants shared responsibility, once a "lo-bull-scribe" was a third liable party. It is likely that such arrangements were the cause for the half animals that are recorded occasionally.87 Animals to be replaced were listed in the assets section even if they were not physically present when the text was recorded. They also counted toward fulfillment of the norm of working animals. Cf. e.g. the 2nd group in text 3 1: I* Bb-ma 1 COW I* guq-gig 1 yoke bull 1 yoke bull, replaced (by) l * guq-giS su-ga Sabra the commander 1 yoke bull, replaced (by) I* guq-giS su-ga nu-bhnda-guq the bull lieutenant 1 yoke bull - non(present) I* nu guq-gig su-su engar to be replaced by the cultivator gub-ba-8m is standing

...

...

Ub[ir-&n]

is old 2 yoke bulls from the count of the NaGABtum Ur[ cultivator]

2 guq-[gig ni-kas71

na-lg6b-hlm-taI

Ur-[

engar]

l Ieimpel

Plow animal inspection records

Only if we count the yoke bull that was to be replaced by the cultivator do we obtain 7 working animals. According to the format of the text, 16-i 1 "deficit 1" would have been placed before the name of the cultivator if the bull to be replaced had not counted toward the norm. The physical absence of animals to be replaced - which is inherent in their status - was indicated in this example, as in most texts of type A, but not in text 12, by a small nu "not" that was placed after or below the number sign. Lafont observed that it was written with a smaller character and "aprhs In the texts of type B the nu gloss is not found; instead, the number was written with the letter stylus and thereby associated with dead and withdrawn animals whose numbers were also written with the letter stylus, the common denominator being absence. I assume that the nu gloss designated also absence in the case of the "donkeys of the siKKum" and the other references, which are tabulated in Fig. 13: Figure 13 reference

animal gu4 Su-gi4 guq-ku-ta sale-a guq-su-ga Igi-LUM.LUMg9 guq-gig guq-1 su-ga engar sai guq-2-S& guq-fiS engar su-ga nu-bhnda-gyq Bb-1 su-gal engar?

Replacements were closely monitored as is demonstrated by TUT 93 AS 3, the fragment of a VIII column tablet with the subscript guq a d e su-su C d~in-gir-su"bulls and donkeys, to be replaced, house Ningirsu", and the difficult texts STA 27 AS 1 and ATL 99 AS 3. The latter two texts attest that replacement in kind was accepted. For that case equivalences were established. According to STA 127 I11 3-5, adult oxen and donkeys were equivalent to 6 gin silver or 6 gur barley and calves and foals (amar) to 2 gin silver or 2 gur barley.

23

Causes of replacements

In most cases the cause of replacement was not mentioned, which is unfortunate because it deprives us of the information that would allow determination of the rules of liability. The cases where the cause of liability was spelled out are arranged below according to their nature: deaths, loss, theft, predation, distress, miscellaneous. Deaths: 31 VI 19: 1 guq su-ga Ba-zi sipa mu dumu-ni guq i-gaz-S&"1 bull, replaced (by) the shepherd Bazi because his son killed a bull". CH $255 reckons with the possibility that somebody kills a bull by hitting it. It is not easy to do that without using much strength and a mace or similar weapon.

l lci~t~pcl

Plow animal inspection records

9 1 21': [I] gu4 su-su mu p N ] nu-bhnda-guq-keq [x (x)] i-na-16 ba-TIL-a-s&"1 bull, to be replaced because the bull lieutenant hitched it for him [ ] (and) it died". TUT 93 I 3'-6': 1 dur [sul-su engar mu mar-[x] IbP-I6 ba-TIL [x (x)] nu-gu7-a-@ "I stallion, to be replaced (by) the cultivator because he hitched it [to] a cart(?), it died, (and) ... did not eat". Or: "because he hitched it [to] a cart (and) it died, and not (because) [a beast (ur-re)] ate it". This is one of the rare references from Ur 111known to me which may attest the use of a cart. 12 I 2-3: 1 su-ga engar mu az-a TIL-a45 "1 bull for cause of a bear having killed it7'. This is a bizarre case. Syria and Persia contained until fairly recently a few bear habitats. But it is hard to imagine that these forest animals ever roamed the countryside of Lagash. It may have been the bear of an entertainer ~q-da-tuSthat got away. -a after nz is perhaps a phonetic rendering of the agentive postposition. Loss: Disappearance as cause of replacement (u-gu d6-a "lost7') was fairly common; it is mentioned for 22 animals in our corpus, understandably without any indication of the circumstances. An illustration for a case of lost and found is contained in text 2. In each division of the household of Gatumdu the withdrawal of 1 mare was recorded and described as follows: 1 anSe ph-da im-ma

1 mare found last year @r ~a-tar-d~a-li intermediary Katarbau mu ~ r - ~ ~ dumuu e n because Ursin, son of the king, lugal-keq Su-na brought it back into ba-a-gi4-a-55 his possession I assume that 2 mares had been found and given to the household of Gatumdu in 42, the past year (pl-da im-ma) from the perspective of the text, and the heads of the divisions split them among themselves. It turned out that they belonged to Ursin, a prince, and they were handed over to an intermediary to be returned to their rightful owner. The last, apocopated and badly phrased, law in YOS 1,28 reads: tukum-bi guq-N .g'~n-na guq u-gu ba-an-dC guq guq-gin7 "If a bull that is wandering about, (and that) bull was lost, bull like bull (it shall be replaced)". Theft: 4 14'-5' seems to be the only example for theft as cause of replacement: [I 1 sag a d e 1 [ 1, substitute for a mare (that was) [S& I]u-la-ga Ah-a [ dlur su-su

stolen by a rustler ?

I la-ga x ? A l a p person was a type of out-law, perhaps specifically a cattle rustler.90 According to the messenger text MVN 7,98 a certain Gurruru traveled Iu-la-ga dabs-dabs-dl "in order to seize a rustler". SNAT 210 = E. Sollberger, AOAT 25,447 records the following sworn promise of a certain Urd~bshen:~l [

uq-Hakar iti munuq-gu7-S& Iu-la-ga udu-zm-a LU-d~anna-ka

I will have brought by the beginning of month IV the rustler of the stolen sheep of Lunanna.

Plow animal inspection records

Heimpel

If not, I will be Iu-la-ga-bii-me-a that rustler. NG 169 contains the case of a bridegroom who engaged i11laga lifestyle which seems to have been reason for annulling the betrothal (lines 21-23): nam-lu-la-gaba-a-hul-a-82nam-dam-nita Nin-hi-li-sii-keqtug fb-da-an-ur "because he was mined by living as a rustler, Ninhilisu was released from being his wife". Predation: Loss to wild animals is attested 13 times in the corpus and twice in TLJT 93, The formula for it was ur-re/e gu7-a "eaten by a beast". Most of the plow animals that were declared "eaten by a beast" were adult. The only predator of adult oxen and donkccys in Babylonia was the lion. Among the Urruna texts replacement as result of predation of a Lion was mentioned once (text 6, group 6). In that case the word ur-m& was used.92 Tt is remarkable how common predation by lions was.93 It provides a telling background for Shulgi's praise as lion hunter in hymn R, and it demonstrates that large tracts of the southe1-nBabylonian countryside were wild and uninhabited country, as they are today. Cultivators94, cultivators md bull lieutenants,95 bishops,g6 and thc administration could be responsible for replacement caused by predation. The last case is attested in 35 IV 4-5: a mare and stallion that had been "eaten by a beast" were recorded as deficit (la-i). The Sumerian law AS 16,7 III 9'-13', stated the case of a lion - here called as expected ur-m& - having killed "a bull 'by' its yoke" (gu4 BiSgudun-b6).The preservation of the negative prefix of the verb of the apodosis indicates that the caretaker was not responsible for replacing it. YOS 1, 28 $8, a Sumerian law in questionable Sumerian, seems to say that a bull that was killed by a lion while wandering about (nigin-na) was to be replaced by its keeper. CH $244 on the other hand stated that the renter of a bull that was killed by a lion ina ,Grim ("in open country"?) was not Ijable. Distress: 4 YJII 6: 3* nu guj su-su saraga nu-lbanda-yu.pke4 mu ge-kurg-ra engar-ra nu-nastam-ma46 "3 non (plcsent) bulls, to be replaced (by) the bishop because the bull lieutenant did not give the cultivator the allotmerit barley". The cultivator withheld 3 bulls as distress in lieu of the barley that the administration was obliged to give him as allotment. The bull lieutenant was supposed to hand it over to the cultivator. The fact that the bishop, not he, was responsible for replacing indicates that the allotment barley was issued by the bishop and distributed by the bull lieutenant. Maekawa demonstrated that the allotment barley was l/lOth of the harvest of a domain unit that was worked by a cultivator who held 12 iku as allotment plot, or 1/20th of the harvest if he held a 6 iku plot.g7 According to the standard yield ratio of 30 gur per bur, a cultivator with 6 iku could expect an allotment of 10 gur. Accoriling to the replacement value of 6 shekel silver, or 6 gur barley, for 1 bull that was used in Lagash, 10 gus barley corresponded to 1 and two thirds bulls.98 With three bulls the cultivator held a value of 18 gur which was presumably more than his allotment barley. CH $241 refers to a bull as distress: "If a person takes a bull as distress t ~ eshall pay one third pound silver", if 11: had no valid elairn. In our case the cultivator had a valid claim, he kept 3 bulls, and the bishop the barley. thus becoming liable for the replacement of the three bulls. Miscellaneous: 9 IV 13 and V 6: I* nu dur su-su US-e ttim-ma "1 non (present) stallion, brought to US". It is very tempting to understand this as a case where a stallion was removed from the plow group because it was needed for breeding: gig-e "for the penis", i.e. for stud mu-tum nu-mu-ttim

Heimpel

Plow animal inspection records

service. The interpretation would mean that plow stallions were not necessarily castrated. I have long sought for, but not been able to obtain, reliable information on the existence or nonexistence of uncastrated males as draft animals from written material. Convictions pro and contra are occasionally expressed. The best I can offer is the statement of an old Black Forest farmer concerning oxen: "wenn er gutartig ist, spannt man einen Muni schon einmal an" [a docile bull (i.e. uncastrated adult male bos taurus)99 is occasionally hitched]. 5 I11 19-21 (cf. collation Maeda): l* anSe su-ga engar sag anSe-Sk mu su su-a-S& 6 V 7-8: I* nu gu4 na-grib-tum nu-sum-ma mu-3 su-su ~ i i - d ~ i n - S u b u"1r non (present) bull, not given (by) the NaGABtum since 3 years, to be replaced by Luninshubura". This is not the only time that an animal was "not given (by) the NaGABtum.lo0 17, 22: l * anSe-1 sag anSe-85 mu in-su sag &rum-ma ba-TIL-R 27, 5: l*gu4 su-su engar mu hug-gri ~ r - d x - x - ~

24

Deficit replacements

We have seen that animals that had to be replaced were recorded in the assets sections. Occasionally, however, animals that had to be replaced were recorded as deficits. Cf. e.g. the deaths and deficit section 4 IX 8-10: 16-i 4 anSe su-su nu-su deficit 4 mares yet to be replaced engar ri-gu dC-a (by) the cultivator, lost 1 dur su-su nu-su 1 stallion yet to be replaced engar u-gu dC-a (by) the cultivator, lost 2 dur 2 stallions 16-i-iun is deficit The administration had apparently no expectation that the cultivator would fulfill his obligation to replace 5 lost donkeys and reckoned them as deficit. The few other references101 for deficit replacements contain no clue about the particular circumstances. In the type C text 30 deficit replacements were totaled as 16-1su-su after the total of deaths.102 It is not likely that the administration simply waived the obligation of replacement. Indeed, a case of deficit replacement from Umma shows that a person liable for replacement did not replace in kind but had his family work off his debt (cf. section 46). The debt could also be paid in silver, barley, or wool as is attested in STA 27 and ATL 99.

25

Substitution

Substitutions were expressed with the fomula sag animal-& "substitute for animal". Their notation took up even more space than that of replacements. The data are collected in Table 14. Substitutions were often connected with replacements. Cf. e.g. 1 rib-mribsu-ga engar sag guq-giS-SG "1 cow, replaced (by) the cultivator, as substitute for a yoke bull". This should mean that the cultivator had to replace a yoke bull and offered a cow as replacement, and that the administration accepted the cow as substitute for the yoke bull. The same process was also

Heimpel

Plow animal inspection records

expressed in a different way: 1 dur sag gu4-gig-G su-ga engar "1 stallion as substitute for a yoke bull, replaced (by) the cultivator" (25 rev. 3-4). Another occasion for substitutions was the process of intramural redistribution for which see section 28. Cf. the following case from text 12: a yoke bull was transferred from cultivator Lugishbare to cultivator Lugalsipa; in the assets section of Lugishbare is found the entry 1 6 b - m a sag guq-gig-3 "1 COW as substitute for a yoke bull".lo3 The substituted yoke bull could be the very animal that was transferred to Lugalsipa. There were surely other occasions for substitutions, but the current documentation is too thin to uncover them. Compared to the animals which they substituted, the substitutes were: replaced 3 67 of same age and different gender 29 one year older1o4 7 two years older none three years older 29 one year younger 14 two years younger none three years younger 1 a Su-giq class animal 2 substitute for a Su-gi4 1 a siKKum donkey 14 of the other species

26

Withdrawals

Withdrawals were registered after assets and designated with the label zi-ga "withdrawal". The numbers of withdrawn animals were written with the letter stylus. In the majority of cases no animals were withdrawn from a plow animal group during a given year. When a withdrawal was made it consisted typically of 1 animal. According to text 37,26 animals were withdrawn from all groups of the households of Shulgi, Nindara, Dumuzi, Amarsin, Ninsun, and Gatumdu. According to the numbers collected in Table 15, withdrawals averaged 3% of assets. Oxen and donkeys of both sexes and all ages except the first year were withdrawn. Most were adult, less than 9% subadult. One "two year" and many adult animals were characterized as Su-giq. Most withdrawn animals went to the NaGABtum. This was expressed as "to the NaGABtum" (na-grib-turn-%), or abbreviated "NaGABtum". There, they were received by the "fattener" (kiiruSda) Si-KAK. Transfers to other households were registered as withdrawals in the household from which they were transferred (cf. section 29).

Heimpel

27

Plow animal inspection records

Supplementation

Sections consisting of records of supplernented animals were introduced with the formula libiram "is old". It identified the preceding entries as the state of affairs before supplementation. Supplement sections occur regularly in texts of type A. They are lacking in texts of type B and C ; the former rnay thus have been written before supplements reached the groups, the latter after supplements were integrated among assets. The goai of supplementation was the preservation of efficient plow animal groups, specifically the replacement of deaths, and the new formation of groups in times of expansion of arable land (cf. section 4). According to TUT 27 of AS 2, the 480 unit group employed 3679 oxen and donkeys that were organized into 480 plow animal groups in AS 2. The average rate of deaths was 10.5% according to the numbers collected in Table 15. It meant that the administration had to replace an average of 386 animals per year. The flow of these animals to the households is documented in the supplement sections. The information of these sections is assembled in Table 16. It will be treated according to the different sources of the supplemented animals. A special form of supplementation was redistribution of animals within a household and between households. It will be treated separately in sections 28-30. Table 16 (1): Most supplemented animals came from the NaGABtum. The designations vary: "from the NaGABtum" (4,7) "from the NaGABtum, from Urningishzida" (9) "NaGABtum, from the kurugda Si-KAK, intermediary Urningishzida" (9) "from the house of pens,l05 intermediary bishop of Ningirsu" (10) "from the count (of) NaGAB tum" (12,3 1,32,37) "from the house of pens, from the NaGABtum" (6 and 7) "from the house of pens" (2,6,39) "house of pens" (9) "from Si-KAK (9) "from the fattener (kurugda) Si-KAK" (5) Oxen came from the house of pens and the NaGABtum, donkeys came from the NaGABtum but not the house of pens. It indicates that the house of pens belonged to the NaGABtum but was a separate entity devoted to oxen. It was probably the central administrative unit for oxen where the decisions about the future of each animal were made: wether it was fattened for slaughter, kept for breeding, used for dairying, or for agricultural work. The fragment HLC 260 table 122 contains a passage that recorded the number of oxen of the house of pens of the year 9 47, 1329 cows and 600 bulls, and the animals among - them that were destined for agricultural work: ih-ba 111 guq apin-na si-ga i-g6l "1 11 bulls among - them are incorporated among the plows". The number represents probably the yearly total of oxen that were supplemented in the 480 unit group. 62 bulls were supplemented from the "account of the NaGABtum" in [SS 11 to households representing 244 units according to text 37 (cf. note 1 to Table 16).

Heimpel

Plow animal inspection records

The institution NaGABtum deserves a study of its own. For the time being cf. CAD. S.V.nagabtum, where many references are collected, principally from Drehem texts. M. Sigrist mentions the "royal NaGABtum" (na-gsib-tum lugal) in Uruk which is attested in BIN 3, 281 and, generalizing the reference, says: "Le roi avait dans le diffkrentes villes du royaume ses propres parcs h bCtail desquels il pouvait tirer les animaux qu'il voulait distribuer aux autoritks des temples du lieuV.l06The NaGABtum of the texts discussed here obviously had a much broader function. An "account of the NaGABtum" of the KA-la-ti type is preserved in MVN 6, 294. Table 16 (4): A considerable number of animals was "bought with silver" (ku-ta sale-a), especially in the year 3 47 according to texts 6 and 7. Text 3 of 43 identified the source of the purchase, a merchant by the name of Iluma. He also sold a "rear ox" in 3 42 to the household of URUxGt according to NATN 408: 1 guq-6b-ur 1 rear ox from Iluma merchant ki Ilu-ma dam-gar!-ta apin d ~ u g a l - ~ ~ ~ ! x ~ t k i - g & for the plows of Lugal-URUxGt received by Lugahpa cultivator Lugal-sipa engar i-dabs supervisor Abbakala commander ugula A[b?-b]a-kal-laga[bra] gir LU-d~in-gubur intermediary Luninshubur There is a small group of similar texts which augments the infomiation from the gu4apin-gub-ba texts and demonstrates that purchase of plow animals was always an important aspect of supplementation. 57 donkeys that had been bought for silver and brought from Uruk, with a royal messenger serving as intermediary, were received by Adda, the bishop of Inana, for the household of Igalima at the end of the year AS 3 according to TUT 49; Berens 36 AS 4 I11 10 recorded the receipt of 35 donkeys "from merchants for the commanders (mu gabra-nenesic-g&) for the "plows of Ningirsu;" and 68 oxen and 13 donkeys were bought for wool (sig-ta sale-a) "(for) the yokes of the house of Amarsin" (gig C d ~ m a r - d ~ u eaccording n) to HLC 164 table 103 (without year date). With these animals each of the 50 plow animal groups of the household (cf. Table 3) could have received 1, and more than half of them a second, animal as supplement. According to MVN 7, 111 IS 1 IX the sum of almost 18 pound silver was used for the purchase of the oxen and donkeys for the district Kinunir-Nina. The price for one ox averaged 7 shekel, for a donkey 5 shekel; that is 18 pound silver bought about 180 animals. It translates into about 1 and one third animals per plow animal group if we localize in the district the households of Nanshe, Dumuzi, Nindara, Gatumdu, and Lugal-URUxGt, representing 125 agricultural units in S 47 (cf. Table 3).1M A small group of texts of the year mu Shashru ba-bul (5 42 or AS 6) recorded the purchase of "rear oxen" gu4-6b-hr-ra.log Cf. e.g. SNAT 16 (envelope): 1 g. for the plow 1 guq-aib-hr apin-a from the merchant Luningirsu ki LU-d~in-gir-su-ta dam-ghr-ta tablet of Urnumushda dub ur-d~u-mu8-da bishop of URUxGt sanga U R U X G ~ ~ ~

Heimpel

Plow animal inspection records Heimpel

Table 16 (5): It escapes me why animals of the category "for cause of fallen cattle" mu guq ri-ri-ga-gbshould have been supplements. Their description resembles entries of replaced and to be replaced animals in asset sections. They were replaced by the commander Luzagesil09, a bishop and a bull lieutenant,llo and unidentified persons.111 The main purpose of supplementation must have been maintenance of groups with full norms of working animals, and thus the great majority of supplemented animals were adult and fit for work. Only rarely do we find other kinds of animals among them. Text 1 recorded supplements from gifts (a-ru-a) that had presumably been given to Ningirsu andlor other divinities of his household. Since one may not have been fussy about accepting votive gifts of questionable value,l12and since there may have been a scarcety of sources of supplementable animals at the time, we find as an exception a "2 year" stallion which was not yet fit for the plow (dir-2) and even a stallion and mare of the Gu-giq class. A hapax is the milk cow calf (ibamar-pa) which came from the house of pens and was supplemented in a group of the household of Shulgi (6 I1 6). Among the adult animals that were fit for plowing, the males outweighed the females by a larger margin than that of the average in assets sections. According to the preserved sections of text 37 for example, 63 bulls and no cow, 14 stallions and 3 mares, were supplemented.

V Other administrative procedures of plow animal management 28

Intramural redistribution

Many animals in supplement sections were qualified with the phrase "from PN cultivator". Cf. e.g. text 12 18-12: libir-im is old 2* guq-gig 2 yoke bulls ki ur-d~umu-ziengar-ta

I* ib-m& ni-kas7 na-gib-tum-ta

from Urdumuzi cultivator 1 COW

from the count of the NaGABtum Urdumuzi was the cultivator who managed group 8 (I11 14-31). In fact, every cultivator who was mentioned as source of animals in supplement sections managed one of the groups that were registered in the text. The movement of animals constituted apparently intramural redistribution. The pattern is shown in Table 12. It will be seen that redistributions involved the groups of one and the same bull lieutenant, or those of the two bull lieutenants of one division, or two bull lieutenants of different divisions. In one exceptional case a cultivator, the aforementioned Urdumuzi of group 8, received a supplement from himself via a bull lieutenant zi "1 yoke bull, from the other division: 1 guq-j@ s u - y G u - p nu-bhnda-guq W ~ r - d ~ u m u -engar-ta replaced (by) bull lieutenant Gugu, from cultivator Urdumuzi". The redistributed animals were not totaled as supplements even if they were registered in the supplement sections of the itemization, but disappear instead in the total of assets.

29

Plow animal inspection records

Redistribution and borrowing between households

Group 10 in text 12 was managed by the cultivator Lugalmud*; it was described as follows: deficit 2 yoke bulls 18-i 2 guq-gig ri-ri-ga libir-hm 4 dir ni-kas7 na-gib-tum-ta 4 dir apin d~in-gig-zi-da-ta

deaths is old 4 stallions from the count of the NaGABtum 4 stallions from the plows of Ningishzida

As can be seen from Table 12, three oxen had been withdrawn from the group of Lugalrnudah to serve as supplements for as many cultivators. The three oxen had constituted the assets of his original group. Two dead yoke bulls were left on his account. With all oxen gone, a new group, this time of donkeys, was formed from supplements and assigned to him. Four stallions came from the household of Ningishzida. Another batch of 4 stallions from the household of Ningishzida was used for the constitution of group 16. Group 10 belonged to the division of Urnumushda, group 16 to that of Narnrnah: the transfer from Ningishzida was evenly split between the two divisions. The transfer of animals between households was registered as a supplement in the receiving household, as here, and as a withdrawal in the household where the animals originated. The latter case is attested in text 4 which registered the plow animals of the household of Igalima (IX 4-7): 4 diir 4 stallions for the plows of Lugal-URUxGt apin d ~ u g a l - ~ ~ ~ x ~ t - g & Nam-mab i-dabs zi-ga

Narnmah received it withdrawal.

One household borrowed occasionally plow animals from another. Cf. the memorandum Gpigraphie 1,56 of 3s 2 IV 4: 4 harrow bulls, to harrow 1.62 h each 4 g ~ q - i i 4.5 r ~ iku ~ ~ gig-ir-ta 4 bulls to harrow 1.44 h each 4 guq 4 iku gig-iir-ta the bulls are to be handed back gu4 gu giq-giq-dam bulls of Urbau, commander of the guq ~ r - ~ ~ agabra - l i 6 d~inmarki house(ho1d) of Ninmar place of the bishop of Nanshe ki sanga d ~ a n ~ e plow land (managed by) bull lieutenant Luutu gina-gu4 LU-dutu nu-bhnda-guq field Gigara a-gi Gi-gira

30

Redistribution within the household of Ningirsu

The household of Ningirsu was the largest of its kind and it was probably also the most complex. In the context of redistribution of plow animals the divisions could behave like two

Heimpel

Plow animal inspection records

separate households. Texts 1 and 5 recorded the plow animal groups in the division of bishop Urbau. The administrator of the other division at the time was Utumu, bishop of Bagara and commander. In text I we find the following entry (IX 23'-24'): 1 dur 1 stallion ki ~ 6 - ~ ~ i n - g i r - s u - t a from Luningirsu apin d~tu-gulO-ta from the plows of Utumu

In text 5 VII 23-29: 1 stallion, replaced [(by) ] ki ~ f i - ~ ~ i n - g i rengar-ta -su from the cultivator Luningirsu 1 dur-2 gig-R sag guq-Si? 1 two year stallion for the yoke as substitute for a bull ki Ur-sukkal engar-ta frorn the cultivator Ursukkal 1 dur su-ga sag guq-62 1 replaced stallion as substitute for a bull ki Lugal-ni-zu engar-ta from the cultivator Lugalnizu apin d~tu-gu10 gab[ra-ta] [from] the plows of the commander Utumu As in intramural redistribution the cultivators from whom the supplements originated were mentioned by name and title, and as in redistribution between households the animals were designated as belonging to the "plows" of another administrative unit, and listed among the totals of supplements. 1 dur su[-ga

31

]

Dissolution of groups and change of managers

Plow animal groups were subject to constant change as animals died, were lost, withdrawn, redistributed, replaced, substituted, and supplemented. Normally these changes were balanced so that the group composition stayed close to the ideal. Not infrequently, however, group composition deteriorated so far that drastic action was needed. Unfortunately, the text corpus does not yet provide the density of documentation that would allow the reconstruction of the development that led to the dissolution of groups, or for that matter any developments over the years. So far, only short episodes can be grasped. Some are quoted below: (1) According to text 4, a group of donkeys managed by Henati, cultivator in the household of Igali~na,had ceased being viable. Its assets were two donkeys that had not yet been replaced by the bishop, 4 donkeys were withdrawn and transferred to another household, 5 lost donkeys were reckoned as unrecoverable and counted as deficit, and the rest constituted deaths. As a non functioning group it had no supervisor anymore and was placed as appendix at the end of the itemizations in the division of Adda (IX 1-12). According to VII 16-19 a cultivator by the name of Shagube "took charge of 6 bulls of the house of pens from the NaGABtum; they are the bulls of Henati" (6* gu4 6-tur na-gab-tum-tagu4 IjC-na-ti-kam Sh-gti-b6 idabs). It appears that Henati, after dissolution of his donkey group, had received of a group of oxen which had been newly formed from animals of the house of pens. But by the time the text was recorded, this group had been transferred from Henati to Shagube.

Heimpel

Plow animal inspection records

(2) The story of Urzikuma of text 4 was similar. He too had originally a donkey group. It had 4 adult donkeys - of which 1 was one year overdue from the NaGABtum - and 1 gu-gij animal as assets and deaths of 2 animals, and was attached in this state as a supplement to the group of Lugirnun who managed group 6 of the second division of the household (VII 6-14). Just as Henati, Urzikuma received a newly formed group of 6 bulls of the house of pens frorn the NaGABtum, but they were slated to be transferred to another cultivator. The latter had not been determined by the time the text was written so that the place where his name would have appeared was left uninscribed (VI 21). In the end, Henati and Urzikuma did not manage any group in the household of Igalima. Urzikuma appears as cultivator in the household of Ningishzida according to text 10. He must have been transferred from one household to another; the date of text 10 is unfortunatley broken, and it remains unclear in which direction the transfer was made. The cases of Henati and Urzikuma constitute the longest episodes in the corpus. They include three developments: dissolution of groups, assignment of newly formed groups, and change of managers. Dissolution and assignment of a newly formed group is attested in the case of the 10th group of the first division and the 6th group of the 2nd division in text 12 (see above in section 29). A nonfunctional group which had not yet been completely dissolved and was not reassigned occurs as first group in text 47: 1 two year mare 1 emeg-2 1 one year mare 1 emeg-1 deaths ri-ri-ga libir-hm is old 1 non (present) stallion bought for silver I* nu dur ku-ta-sale-a Ur-Sag-ga engar cultivator Urshaga The group consisted of deaths and a nonpresent supplement. Urshaga's was a ghost group which can only have had an administrative function. Change of cultivators of a group was often recorded.l14 It was designated with the formula guqlange PN1-kam PNz (engar) i-dabs "the oxerddonkeys of PN1 were taken in charge by (the cultivator) PN2". Once, in text 16, the formula i-dabs was left out. In text 19, the change was expressed in a curious way by designating the group as supplement coming from the former cultivator to join, as it were, a zero group of the current cultivator: libir-hm is old 6 yoke bulls 6 guq-gig gub-ba-hm are assets they were the bulls of Ursheila guq Ur-Le-ll-la-kam Kulikala took them in charge Ku-li-kal-la i-dab5 The reason for change of cultivators was never recorded. It may have been death or financial ruin of the original cultivator, or transfer between households. The only information we can wrest from the texts is to determine whether the cultivator who gave up a plow animal group was kept on the roster of the household. Even that is possible only in the case texts 4, 10,

H

Heimpel

Plow animal inspection records

and 12, which are well enough preserved for the complete roster of cultivators to be known. Henati and Unikuma of text 4 were not on it, neither was Urashnan whose animals were taken over. by A N . ( ? I R ~ - ~according [~ to text 10, and neither were Lugaldalla and Lupabilsag according to text 12 (cf. Table 5). On the other hand, Urmes and Urdumuzi of this text were on the roster. The donkeys of a cultivator Urmes changed to Luurasari, and a cultivator Urmes took the oxen of Lupabilsag; the oxen of a cultivator Urdumuzi changed to Lukunun, and a cultivator Urdumuzi managed the 8th plow animal group in the first division. Even if Urmes and Urdumuzi are very common names, we can be reasonably sure that these were, infact, the same persons because the scribes distinguished two persons of the same rank with the same name appearing in one and the same text by adding the father's name of one or both. Change of bull lieutenants in the supervision of a group is attested in four texts of type B. It was designated by the same formula and formed part of the subscript in texts 22 and 24: guq PN1-kam PN2 i-dabs. In text 28 ugula PN2 is found instead of PN2 i-dabs; and in text 47 ugula P N l nu-bhnda-guq instead of guq PN1-kam. Another way to designate the previous supervisor of a group or a block of groups seems to have been the placement of a ki PN-ta formula at the end of the itemization and before the subscript and, in cases where the subscript was separated from the itemization by a blank, before that blank. The placement of the formula suggests at first sight that it qualified the last group of the itemization, but the fact that it is not found in any other position indicates that it referred to all groups of the itemization. An exception is text 27 where the formula is found after each itemized group. The names of the persons in the formula are listed in Fig. 14. One of them was titled bull lieutenant. Ld-IB and Igishasha occur in two texts each, Ld-IB in two texts of the same date. If the transfer of the groups took place in the same year, Ld-IB would have been in charge of at least 10 groups at one time. He could not have been a bull lieutenant in that case, and may have been a 10-bull-scribe. Figure 14 text

group

1 19 1 all 120

1 all

name o f source

I LB-IB

I LB-IB Lugal-x-x

1

1

5

1

~ r - ~ nu-bhnda-guq ~ t u ~ r - ~ udumu t u ~r-d~amma Igi-Sag-Sag Igi-Sag-Sag

Heimpel

Plow animal inspection records

It is unfortunate that the realities behind these dry pieces of information cannot yet be perceived. I suspect that they will only become clear when the system of land distribution for cultivation is explained.

32

Inspections

The Umma texts characterized the process which resulted in their recording as "inspections" (giirum); since these texts are very close in subject matter and format to the guqapin-gub-ba texts from Girsu, it is hardly plausible that the latter were products of a different process. The term gururn appears in text 1 XI1 20-21 where a mare was qualified as egers drumta "from after the inspection". There was apparently enough time beween inspection and recording to fit in late-comers. HLC 346 table 133, a tag of a tablet box, mentions an inspection in connection with plow animals: tablet box pisan-dub-ba d r u m aka done inspections plow bulls having passed guq-apin dib-ba balanced accounts of wages ni-kas7 aka ai C d~umu-zi house Dumuzi C d~in-dar-ra house Nindara C d~in-mar~j house Ninmar house Shulgi cdSu~-~i C d~mar-d~uen house Amarsin and house Gatumdu a C dgsi-tam-du10 i-gril are in it. The terse document did not express the syntactic relationships between the phrases and thus leaves us guessing whether 1,2, or 3 types of texts were in the box, "done inspections of plow bulls having passed (and of) balanced accounts of wages" or "done inspections of plow bulls having passed" and "balanced accounts of wages", or "done inspections", "plow bulls having passed", and "balanced accounts of wages (of agricultural workers)". A check of the texts of tablet box tags that were collected by R.C.Nelsonlls indicates that the assumption of 2 text types is the most likely because the object of an inspection was nearly always mentioned after the phrase "done inspection",l16 and inspections of accounts, balanced or not, are not attested, while "balanced accounts of wages" are attested as contents of tablet boxes.117 The "inspections of bulls having passed" were presumably texts with the subscript guq-apin-dib-ba "bulls having passed". 5 such texts are known to me. One of them is # 23 of our corpus. It differs only in this phrase from the guq-apin-dib-ba texts of type B. The other four texts differ substantially in format.

Heimpel

33

Plow animal inspection records

Plow animal inspection records

gu4-apin-dib-batexts

Lunindara supervisor bull lieutenant Uludj

~u-d~in-dar-a

CT 7,32 BM 18394 AS 2 consists of short lists of oxen and donkeys followed by the name of a cultivator. The first group looks very much like a plow animal group from a guq-apin-gob-ba text: 3 ab-m& 3 COWS 4 guq-gig 4 yoke bulls

ugula Ulu3-df nu-blnda-gud

The subscript is: guq su-ga engar nu-bhnda-guq

2 guq-1

2 one year bulls 1 COW milk calf 1 nu guq-gig ki-gig-i 1 non (present) oilplant land yoke bull 1 anSe su-ga engar 1 mare replaced (by) the cultivator LU-d~in-dar-a Lunindara There is a working group of 7 adult oxen, 3 young oxen, a non present bull connected with oilplant cultivation, and a replaced mare. The remaining 6 groups were much smaller. They included a fair number of animals connected with oilplant cultivation, none of them non present. There was no division into sections recording assets, withdrawals, deaths, and supplements. A supervisor for the first 5 groups and another for the remaining two groups was mentioned. Then follows the subscript:

guq dib-ba SB ~ a ~ a g ~ ~

1 aib-amar-ga

ugula sanga d~in-dar-a

bulls, replaced (by) cultivators (and) bull lieutenants bulls having passed, of/in Lagash supervisor bishop of Nindara

The three texts demonstrate that the guq-dib-ba was connected with a place, and that it was an occasion when assets and deaths of animals of plow animal groups were established and replacements were executed and registered. According to CT 7,29 18383, the occasion fell late in the year as the text is dated to the 1lth month. The fourth text, HLC 251 table 126 AS 5 is the fragment of a 4 column tablet. 'The subscript does not include a geographical name. It consists of a list of plow animal groups without division into sections of assets, deaths etc., includes several unique features, and is essentially unintelligible thus reminding us that the mechanisms of monitoring plow animals remain essentially ununderstood.

guq-apin dib-ba

plow bulls having passed house of Nindara ugula SeS-kal-la supervisor Sheshkala SB a-SB La-za-PI oflon the field La-za-PI The second text is CT 7,29 BM 18383 AS 3. It recorded 4 donkey groups. There are sections of assets and deaths, but not of withdrawals and supplements. The numbers in the assets sections decline from 10 working animals in the first group, to 8, 6, and 4. The fourth group includes "1 yoke stallion, lost, to be replaced (by) the cultivator" which was listed ----- between the assets and deaths sections. Blocks of two groups each were subscribed by anb nlgurll PN "donkeys, ... of PN". The first PN was Urbau, the divisional head and bishop in the household of Ningirsu.l18 The second was his colleague Dudu. Then follows the subscript: C d~in-dar-a

ugula Ur-gar nu-blnda-gu4 guq dib-ba dumu Cnsi-ke4-ne

31Ur-sag-pa-&

supervisor Urgar bull lieutenant bulls having passed (of the) children of the governor oflin Ursagpae

@rUr-mes Cnsi

intermediary Urmes governor h g"lr Ur-nigin sag-dug lugal and intermediary U. land registrar of the king iti Se-gurlo-kug month 11 The third text, HSS 2,32 AS 2, records distribution of oxen that had been returned by cultivators and bull lieutenants and were distributed to cultivators under their bull lieutenants. A typical entry is obv. 7-9: 1 guq su-ga nu-bBnda-gu4

1 bull, returned (by) thela bull lieutenant

34

The term dib-ba in gucapin-gub-ba texts

In guq-apin-gub-ba texts, the phrases 6 1 dib-ba, egerg dib-ba, and nu-dib-ba occur as characterizations of animals. The totals of text 1 include 1 yoke bull, 1 "two year" cow, and 1 "two year" mare, all characterized as egers dib-ba-ta "from after the passing" (XI 37; 40; XI1 15). Next to it (XI1 20-21) we find 1 mare egerg gurumta "from after the inspection". An entry from a supplements section of the itemization (1 IX 20') reads: l* dur a-ru-a-taegers dib-ba-ta "1 stallion from gifts from after the passi1lg".~~9 An example for Sh dib-ba "of the passing" is 33 "IV" 17-20: 5 yoke bulls of the passing 5 guq-gig Sa dib-ba na-grib-tum-S2 gir Si-KAK kurugda zi-ga

to the NaGABtum intermediary fattener Si-KAK withdrawal

The withdrawal left just 2 fallen animals in the group. It seems that the passing was the occasion on which it was determined to move the 5 yoke bulls to the NaGABtum.120 According to 2 I 3 a bull "of the passing, cultivator Urshaga" was part of the assets of the group of Urenki. The passing was here apparently the occasion on which a bull was transferred from the group of Urshaga to that of Urenki. A similar entry is found in 40,9: I* dOr Sh dib-ba ~b-ba-gu~~.'~~

The phrase nu-dib-ba occurs in 43, 3: 1 guq-2 su-su engar nu-dib-ba im-ma "1 two year bull, to be replaced (by) the cultivator, not having passed last year". Here, nu-dib-ba fills in for nu-su and designates the occasion when its status changed from "to be replaced" to "one year overdue".

Heimpel

Plow animal inspection records

Combining the clues of this and the last section indicates that dib-ba designated an inspection: it took place in a specific location, it served as occasion for transfer of animals from one group to another and back to the administration, it was the time when the replacement status changed and when the composition of plow animal groups was determined. In section 46 a text from Umma is quoted which demonstrates that withdrawals were made at the occasion of a field inspection. The verb dib was written with the sign L A G A B X P A ~which ~ ~ was fused with the sign LAGABXASin post Ur 111 script. In lexical tradition dib = etequ = to pass. But it was often identified (i.e. confused) with dabs = jabatu = to grasp. Many scholars follow this identification (i.e. confusion) and assume that already in Ur I11 texts dib was simply a scribal variant of dabs. This cannot be so because the contexts of dib and dabs fall into distinct groups and because dabs occurs consistently with the prefix i- while dib occurs consistently with the prefix ba-. Not a single dib is found in thousands of references for the formula i-dabs "helshe received". dib-ba occurs as a noun for which a translation with the gerund "passing" with infinitive function fits best, thus egers dib-ba "after the passing" and Sh dib-ba "of the passing". It also occurs as a participle, as in nu-dib-ba. Being the hamtu form, it should mean "not having passed". The literal meaning of the term indicates probably the particular form of the inspection so designated. In Cylinder B of Gudea, the inauguration of the newly built Eninnu is described. Its central part was a procession in which the servant gods passed before their master god Ningirsu with their "divine powers" (me-ni-damu-na-da-dib-e).Described was apparently a form of inspection in which the head of a household surveyed his servants as they passed before him. According to this parallel I suggest that in dib-ba inspections plow animals were lead in a file before the owner or his representative and the recording scribe.123 The results were the gu4apin-dib-ba texts. They were close to the gu4-apin-gub-batexts of type B, and the fact that text 23 was characterized as guq-apin-dib-ba in the subscript while being a guq-apin-gub-ba text in format may be understood as confirmation of that closeness. Yet they were not identical. The gu4-apin-gub-batexts of type B were not linked to a particular place, and they regularly recorded assets, withdrawals, and deaths. Unfortunately, there does not seem to exist for the time being guq-apin-dib-baand guq-apin-gub-ba texts registering the plow animal groups of one and the same bull lieutenant from one and the same year.

Heimpel

1

1I I

I

Plow animal inspection records

comparison of the two groups, but it is often difficult to decide whether differences resulted from different processes of field administration in the two provinces - which looked back on a long history of independent development by Ur 111 times - or from different scribal conventions. The subscripts are a case in point. In Girsu, they consisted of the identification of the household as administrative unit and the phrase gu4-apin-gub-ba "standing plow bulls" which expressed the verification of the number and kind of assets of plow animals as essential feature of the process that led to the recording of the texts. In Umrna, that phrase is not found. Instead, the process that led to the recording of the texts was designated as "done inspection7'(g"rum aka), and the inspected groups were not identified by household but by region. Was the process by which the gu4-apin-gub-ba texts were produced really essentially the same as that which produced the aka of plow animals? And was the field administration in Lagash not divided into regions and that of Umrna not into households?

rum

I

36

Inspections of administrative units and field inspections

The numbers of cultivators and bull lieutenants in texts 7, 9, and 11, correspond to the hierarchical relationship of 5 to 1 between these ranks that was standard in Lagash and Umma agricultural administration.124 In texts 1, 2, 4, and 5, the number of cultivators and bull lieutenants was equal. This ratio of 1:l reflected the distribution of cultivators on the field. Contrary to what we might expect, the administrative organization was invertly proportional to distribution on the field: cultivators who worked next to each other were supervised by different Figure 15

cultivator Dug d

~

r

3

-

~

~

~

+

~

~

~

VI Umma 35 Umma texts The Umma plow animal inspection texts are listed in Table 17. They are fewer and shorter; their scribes did not use the number stylus, and the texts differ in substance and details of notation and terminology from the Girsu texts and among each other. Yet their essential feature is the same, namely a description of the assets, withdrawals, deaths, replacements, and supplements of groups of plow oxen or donkeys that were managed by cultivators. The texts from Urnrna and Girsu complement each other in several aspects so that we can gain information from

13-sum Lugal Zi600

Gi-apin-kug-r6 Zi600

bull lieutenants, and these bull lieutenants were in turn supervised by different superiors. The principle can be demonstrated with the help of BM 110116.125The text listed the yield of the

Heimpel

Plow animal inspection records

district Da U m a , detailing the yield of the individual domain units and their cultivators. The sequence of units and cultivators followed the administrative organization: after groups of five cultivators the supervising bull lieutenant was mentioned, and after two groups of five cultivators and their bull lieutenants the untitled supervisor (uguls), elsewhere called the 10-bullscribe (dub-sar-gu4-lo).Cf. the first group of 5 cultivators in Fig. 15. There is very little overlap; the 5 cultivators worked plots in 9 different field tracts. If we choose a field and observe who is working there, we obtain the following result (the field Uku3nu-ti is chosen in Fig. 16).

Mushbiana. Accordingly the public agricultural land of the province increased slightly between AS 2 and IS 3. Figure 17

district

bull lieutenant

!h-kii-ge

22

area in h 2197.02

#of units 68

-

100

Anegara Mushbiana

3

-

-

all

-

4849.6"

7,

I-Sar-ru-um Ab-ba-gi-na Seg-kal-la

# offiedrs

1 Da Umma I

Figure 16 cultivator

Plow animal inspection records

Heimpel

I

-

I

-

1

105

I

text BM 110116

99

date

IIS 3 77

BM 105334

1 AAS 83

AS 2 )[

1

I

* This is the total of the text, 5184.4, minus oil-plant land and garden plots. The unit numbers show that even the comparatively large plow animal inspection texts recorded only small fractions of the plow animal groups that worked in the districts.

lrll-dSrira

38

7 cultivators worked on the field Ukunuti; all had different bull lieutenants, and all bull lieutenants had different superjors.l26 It follows that inspection texts with the same number of cultivators and bull lieutenants were field inspections and that texts with 5 cultivators per bull lieutenant were not. The ratios between the two ranks in texts 6 'and 8 remain unexplained. 37

The districts

The subscripts include geographical nanles. Some were placed after Sa "in", others not. Text 5 combined both: "done inspection, gu4-LCb, Guedena (and) Mushbima, in Edur Shulpae". The wording suggests that plow animals working in Guedena and Mushbiana were inspected in Edur Shulpae. Guedena and Mushbiana were districts of the province, Edur Shulpae presumably a village jn the vicinity of the Shulpae field. Text 6 relates the "donkeys" - if Gomi's restoration is correct - to Guedcna, and text 8 its 13 plow animal groups to Da Umma, another district. The districts of Urnma have not yet been treated systematically, and this cannot be done here. Instead I tabulate the data that are helpful as background for understanding the plow animal texts in Fig. 17. The references were assernbled and partly treated already by Maekawa in Zinbun 22. Fish 539 did not record the numbers of units of Akasal and Guedena. The ones given here are based on the assumption that the relationship between area and units was equal to that of BM 110116.127Both texts recorded the yield of cereals harvested by so many cultivators from so much land. The total number of units in IS 3 was 102 plus a few units from

The inspection sites

Dukusig "Gold Hill", presumably a tell where an Ummaite struck it rich, was also the name of a field. The latter was located in the district of Akasal according to RA 79, 3 1 # 23; according to the same text the field Shulpae was located there too, and this in turn localizes Edur Shulpae "Shulpae village". It is unclear why plow animals that were connected with the districts of Guedena and Mushbiana should have been inspected in the district of Akasal. Ukunuti was a field in the district Da Umma. Enkar, or Menkar,l28 was treated as field or as district. In the phrase "bulls (of) Men[kar]" in the comprehensive deaths list (cf. Table 18) it was placed after the deaths of the combined districts of Guedena and Mushbiana and was thus seemingly equaled with districts. Yet there exist other examples for uniting fields and districts in one descriptive category.129 In AnOr 7, 313 Menkar was called a field among other fields which were summarized in the subscript as a-SB Gu-dBna u Amar-GIR2-gun3[ ] x-na. Accordingly, Menkar was a field and located in or near the district Guedena. According to AAS 83 Menkar was the location of gu4-iuhub land (see section 40); DCS 271 AS 7 V listed five cultivators who were responsible for harrowing 6 iku in the "field Menkar". The actual locality of the inspections was the threshing floor according to MVN 4,75 AS 6 IV 6 which recorded the results of an inspection of plow animals and agricultural workers. Its subscript reads: glirum aka 4-6-kam ki-su, a-SBd~in-ur4-ra"done inspection of the 6th day (on) the threshing floor of the Ninura field". Closer examination of the text shows that it recorded all plow animals of the district of Akasal where the field was located (cf. Fig. 18):

Heimpel

Plow animal inspection records

Figure 18

a-Sh A-ki-lhh

tu-ra

Lu-kal-la engar

ugula I Da-DU-MU Lugal-6-ma Lugal-ra-ni Lugal-kii-zu Lugal-nisag-e Ur-ma ~r-~~n-lfl-ld 1.1-pa-&

If we divide the number of oxen and donkeys by the average group strength 8.4,13O we arrive at 28 or 29 groups. The result is close to the 26 units which were postulated above for the district in IS 3 according to Fish 539. 39

The bull of Gula

Text 3 recorded the inspection of 1 unit which was designated as "bull of Gula". The term is reminiscent of the small households which were part of large households in Lagash, especially that of Ninhursag which also had just one unit (cf section 2). The "bull of Gula" appears also in BM 105334, a balanced account of the total surveyed public agricultural land (ni-kas7 aka a-Sh @d-da)of the province in AS 2. After the description of the domain land of 100 units, including the allotments of its managers and the fallow, the following passage is found: 6 (bar) ghn-guq

38.88 h domain 1 engar 1 Me 1 cultivator 2.16 h 3 BB-guq 3 iku-ta 3 oxdrivers 1.08 h each @r-sG-gaguq d~u-la-me they are personnel of the bull of Gula Accordingly, the "bull of Gula" described an administrative unit of the public land consisting of 1 agricultural unit including the allotment plots of its cultivator and his 3 oxdrivers. The cultivator of the unit was a certain Lukala in SS 5 acc. to the comprehensive deaths list IV 20-22 (cf. Table 18): 1 kul guq

1 ox hide Lukala cultivator guq d ~ u - l a bull of Gula Lukala managed the unit as cultivator still in IS 3 according to BM 110116 XVII 11-16 where he and his unit were recorded as 68th, and last, unit of the district Da Umrna: Lu-[kall-la engar

3 (bur) 7.5 (iku) ghn 10-ta

le-bi 90 gur

Plow animal inspection records

Heimpel

22.14 h land of 10 (furrows per 6 m) each the barley (yield) 27000 liter

nu-bhnda-guq Lugal-6-zi-da

field Akishah Lukala cultivator bull lieutenant Lugalazida

blank bull of Gula The plow animal inspection of the bull of Gula, our text 3, is from AS 5. The cultivator at that time was a certain Sipaanshe.132 There existed two Gula figures in the province, Gula of KI.AN and Gula of Urnrna. It is not clear which Gula was associated with the agricultural unit. Gula of Umma employed 2 shepherds; one had a flock of Sumerian sheep, another one of goats according to YOS 4,237 SS 7 11, a count of the small cattle of the p r 0 ~ i n c e . l ~ ~ guq d ~ u - l a

Three of the 12 plow animal inspection texts from Umma concern guq-Suhub. 17 of the 19 groups consisted of oxen, the remaining 2 of donkeys. The group compositions varied from group to group, but they do not stand out in any way from the others or from those of the groups in the texts from Girsu. They were apparently unaffected by the special attributes of that which is designated with the term guq-Suhub. Maekawa has repeatedly addressed the problem of the g ~ ~ - S u h u and b l ~developed ~ a hypothesis based on several lines of evidence, namely (1) its occurrence in Ur I11 texts of agricultural context from Girsu and Umma and (2) a passage in the OB Georgica, and (3) its alleged literal meaning. He showed that it designated land consisting mainly of C plots and expressed his conviction that they were specifically C1 plots. His plot classification is as follows: C refers to plots for which 360 sila barley were expended for seeding 1biir of land. If plow animals were used, 240 sila were used for seeding 1 b b of land in 8 rows per nindan and the remaining 120 sila for plow animal fodder. Such plots are called specifically C2. If plow animals were not used for seeding, the whole amount of 360 sila was expended for seeding 12 rows per nindan. Such plots are called C1. Normally, the context of the references for C plots did not express whether C1 or C2 plots were meant. Maekawa, BSA 5, 127, states with respect to the large tracts of guq-iuhub land of C plots in TUT 5: "I assume that C1 plots were much more extensive than C2 plots in Girsu in Shulgi 47". Thus guq-luhub land would not be seeded with plow animals but instead with the implement called @;-gaba-tab which was operated by workers. He shows further that this land was newly converted from waste land, especially reed thickets, producing relatively low yields. Being new land, it was not yet integrated into the "temple households" managed by bishops and commanders but stood under the direct control of the governor. Lines 6-7 of the Georgica indicated to Maekawa that the Babylonians introduced luhub oxen into fields that they had flooded early in the year. Following a suggestion of Civil, he connected the practice with "puddling" in which oxen or buffaloes are let into fields that are to be planted with rice in South East Asia.

Heimpel

Plow animal inspection records

The word Saub may mean "boot" and is, in that meaning, written with the determinative kuS "leather". The Umma field name a-Sg guq-MUL was indeed spelled once, in BIN 5,266, with the determinative, suggesting that guhub in gu4-guhub represented the word "boot". Literally, gu4-guhub would then mean "boot bull". These boot bulls "were probably brought into the fields so that they could break up clods and make the bottom ground smooth. This is also an effective method for stamping out weeds and plants, as is described in Georgica 8. Indeed, oxen seem to have had their hooves and legs covered primarily for protection from the thorny plants and roots of reeds cut out of the field. Another benefit of this method was that salts, which would otherwise remain on the ground surface, were scattered deeply and thoroughly into the ground (BSA 5, 129)". In ASJ 14, Maekawa made some modifications: (1) He voiced doubts about the relevance of the spelling in BIN 5,266 and quoted the equation MUL = l u b b = suguNdnu = member of a herd which Civil had suggested. Indeed, since iuhub represents two words - sugulldnu and suhuppatu - the scribe of BIN 5,266 may have used the determinative to merely indicate the pronounciation Suhub as opposed to mul, without intending the word "boot". (2) He embraced the fact that the verbal form in line 7 of the Georgica has prohibitive function, and now understands it correspondingly as "an advice to avoid using gu4-suhub oxen's tramplinglpuddling in a field of average soil conditions after the artificial flooding in early summer". (3) He quoted evidence for small areas of gu4-suhub land managed by temple households. Maekawa did not consider the occurence of the term in the plow animal inspection texts from Umma. The evidence of seemingly normal plow oxen and donkey groups that were designated as gu4-suhiib does not seem compatible with essential elements of his hypothesis. If trampling the flooded fields that were to be planted in C l plots was the sole objective, any herd of oxen would have sufficed, and the more the better. Why would groups be needed that were managed by cultivators? Would donkeys be suitable for puddling? In order to resolve this difficulty while adhering to Maekawa's conclusions that gu4suhub was a term for land that had been recently converted to agricultural use and was cultivated in a special way, and adopting Civil's identification of gulplb with sugulldnu, 1suggest to use two meanings for guq-suh6b: (1) herd oxen and (2) (former) herd oxen (land). In Georgica the first meaning applies, in the Ur III texts the second. Georgica 5-8 has been translated often. I add my own version: uq a-ta im-mu-e-8d8-a

a-gii ki-durug-bi en-nu-un ak-ab ki-L-er a-ra-ab-tuk guq-luhub @r na-ra-ab-zukum-e

u KIN-bi u-bi-zi5

...

"When it (the ground) emerges from the water watch the field and its wet spots. When it develops kiSer, do not let herd oxen trample it.

Heimpel

Plow animal inspection records

When you have weeded it ..." Presumably this refers to the fields that had been lying fallow the previous year had been used as oxen pasture. In preparation for cultivation the fields were flooded early in the year. When the soil dried up it was time for the oxen to move on. Otherwise the ridges and depressions produced by its tracks would prevent even drying. In the Ur I11 sources discussed by Maekawa, the term gu4-suhub clearly characterized a type of land. Originally, it may have been "herd oxen" land, especially reed thickets (gig-pi) that may well have been used by oxen herds as pasture. Oxen, next to buffaloes, are kept in the marshes still today. In converting such land to fields, the water had to be drained and reeds and other grasses and sedges had to be eradicated. Enki converted marsh (sup) to fields for Enrnerkar - in Old Babylonian view - as follows: d ~ n - klugal i ~ r i d u ~ * - ~ agi- kTIL-bi e ~ lp-ma-an-zb a-bi ha-maan-ti] "Enki, the king of Eridu, eradicated its ... reed for me, brought its water to an end for me".l35 Most work-intensive must have been the removal of the troublesome rhizomes of grasses, including reeds, and sedges.136 It was probably a lengthy progress that took several years137 during which the cultivation methods may have changed and the yields improved in steps. Important are the observations of Maekawa on the fields of the household of Ninmar in the district of Guabba (BSA 5, 132-134) which document slow improvement of wide areas with very low initial productivity. The fact that this was newly converted land is - in addition to Maekawa's arguments - indicated by the nature of field names.l38 Returning to the inspection of plow animal groups on gu4-suhub land, it still remains unclear what the plow animals did on that land. If Maekawa is correct in his assumption that gu4-suhub land consisted mainly of C1 plots and thus was seeded without the use of plow animals and if that was also the case in Umrna, it follows that the plow animal groups were used exclusively for the preparation of the ground prior to seeding. The early date of text 5 may confirm this conclusion, but it is counterweighted by text 8 which is dated late in the planting season. This text may prove to be important for the question of the nature of gu4-suhub. As is argued in section 43 below, the gu4-suhhb groups were formed from normal, that is non gu4suhub groups, after they had finished their seeding work on other land. I believe that it is premature to attempt a complete interpretation because of several intangible factors: (1) We do not yet know what kind of work was meant with the designation gu4 ki-durg. It is mentioned in MVN 6, 305 beside harrowing and seeding; it required many plow teams and a remarkable number of 8 workers pro animal team; the literal meaning "wet spot bulls" indicates that it had something to do with wet conditions on the field. Perhaps the translation of Georgica 7-8 as given above is wrong after all, and the gu4-suhub did have something to do with ki-durg. (2) The process by which newly developed gu4-suhub land was integrated with old farmland is not clear, and neither is the point of time when gu4-suhub land was fully converted and stopped being so designated. The designation may still have been used in Umma during a period when original gu4-suhub land was already so far improved that cultivation fits into the seeding could be done with plow animals. (3) It is not clear how oil~~lant picture of plow animal use. A connection of gu4-suhhb land and oilplant cultivation is attested in

Heimpel

Inana field @r Ha-ba-an-zi-zi intermediary Habanzizi There are 7 group descriptions. All end in the formula PN engar i-dabs "cultivator PN took in charge". Apparently, the groups came without the cultivators who had managed them under Nigarkidu and were received by cultivators under Abbamu. The year date is 5 33 (so Owen), 5 45 or IS 4. The fact that the persons do not seem to be attested elsewhere argues against S 45. The assignment of the text to Umma seems inevitable because of the month name RI and is confirmed by the personal names Ur-Ell-e and Nig-ki-dulo. The terminology points to Lagash (anle instead of emeg, amar-ga instead of riblguq-ga, Su-giq instead of Su before the animal designation, and 164"deficit" for deaths). a-Si d~nana

a fragmentary text that is now destroyed: ITT IV 7331 (cf. MVN 6, page 208). Clearly, more source material is needed to advance the understanding of this important topic.

41

Month dates

One obstacle to interpretation of the texts from Girsu is the almost complete lack of month dates. The only exception is the guq-apin-dib-batext CT 7, 29 18383 which is dated to the 1lth month. In the Urnma texts five month dates are preserved. Three, the 5th, 6th, and 7th, fall squarely into the season of seed plowing. Using Maekawa's Table 1 in BSA 5, we can see that in Lagash most expenses in the 5th month went for deep plowing and harrowing, and those for seed plowing commenced. In the 6th and 7th month the expenses for seed plowing dominated and those for deep plowing and harrowing trailed. The inspection record MVN 4, 75 AS 6 (quoted above in section 38) was dated to the early 4th month when expenses in Lagash went only for soil preparation. Texts 5 and 8, both referring to gu4-suhub land, date to the 3rd and the 9th. Maekawa found a single text from a 9th month with expenses for seeding, and none with expenses for soil preparation or seeding for the 3rd month. The wide spread of month dates indicates the lack of a seasonally fixed date for plow animal inspections in Urnma. This finding agrees with the fact that the gu4-apin-gub-ba texts of type A from Girsu show much difference in the degree to which the norm of plow animal groups was reached. If the existence of a norm indicates that there was a season during which all animals of the norm were needed it follows that the degree of divergence from the norm was proportional to the closeness to that season.

42

Text formats

The only features that are shared by all Umma plow animal inspection texts are lists of assets and deaths of plow animals and the record of their managing cultivator and his supervising bull lieutenant. All other features are strewn among the texts seemingly at random so that there is no convincing division into types. Using the criterion of the ratio of cultivators to bull lieutenants we have separated field inspections from "administrative" inspections. Looking at the criterion of the place of inspection, we expect field inspections to identify the place of inspection. Indeed texts 1,4, and 5 do so, yet 2 does not, and 7 does without being a field inspection. Looking at the criterion of absence or presence of sections, the small scope of most texts obscures possible typological differences. The absence of withdrawals and supplements among the 5 groups text 9 for example may be due to the absence of such sections from the format of the text, or to the fact that in the recorded year no animals were withdrawn from, and supplemented to, the recorded groups. Only in the case of text 4 with its 9 groups does the latter possibility become unlikely. Text 12 shares with the other texts descriptions of plow animal groups, but deviates in almost every other aspect. The full subscript is: guq anie si-la engar oxen (and) donkeys, entrusted (to) cultivators ki Nig-ki-dulo-ta from Ni