Cuneiform Digital Library Journal
2009:5
ISSN 1540-8779
© Cuneiform Digital Library Initiative  

CDLI Home
CDLI Publications
Editorial Notes
Abbreviations
Bibliography

PDF Version of this Article
Get Acrobat Reader
Download Cuneiform Font

The Army of the Kings of Ur: The Textual Evidence^*

§1. Introduction

§1.1. The situation is paradoxical. During the last century of the 3rd millennium BC, several accounts testify that rulers of the Third Dynasty of Ur (Ur III) conducted, from the heart of Sumer, several wars of conquest beyond the frontiers of their kingdom, probably influenced by the imperial model initiated roughly two centuries earlier by the kings of Akkad. And yet very few documents, with little detail in them, are available on the armies that led these military conquests. Moreover, we know little about how these kings organized, from a military point of view, the defenses either at the center or at the outskirts of their kingdom. Although the military organization of this period is extremely important to the history of ancient Mesopotamia, the total evidence that we have to date is so sketchy and incomplete that it allows only minimal insight.^[1]

§1.2. Starting in the middle of his 48-year reign, the second ruler of the dynasty, Šulgi (Š, 2094-2047 BC in traditional chronology), launched the process of territorial expansion and conquest. The year names used to date the documents of that period, an essential source for any enquiry into military history,^[2] show that he led campaigns against the cities or territories of:

• Dêr (year 21, zone A of the map below, figure 1)
• Karahar (year 24, then 31; zone B)
• Simurum (years 26, 32 and 44; zone B)^[3]
• Harši (year 27, then 48; zone B)
• Anšan (year 34; zone D)
• Šašrum (year 42; zone C; Šušarra of the OB period)
• Urbilum (year 45; zone C, present Erbil)
• Kimaš (year 46, then 48; zone B)
• Hu’urti (year 48; zone B)

§1.3. Military expeditions continued under Šulgi’s three successors, even if they were less frequent toward the end of the period:

§1.4. When we examine these year names in detail, we observe that through almost half a century, once every two to four years and sometimes every year, major military campaigns were undertaken beyond the frontiers of the kingdom. Together, the campaigns constituted a policy of expansion that also had a preventive and defensive objective, with an eye to the potential threat represented by populations beyond the borders of the realm. Some of these campaigns were of sufficient importance to be remembered in the literary oracular tradition as late as the Seleucid period.^[4] All of them were carried out in two main directions (figure 1):

• The Zagros area (with the Diyala and the two Zab regions) and the Upper Tigris area (Kurdistan),^[5]
• Susiana (Khuzistan), Elam and southern/southeastern Iran (Fars and toward the Kerman).^[6]

Figure 1: Map of the Ur III Empire.

§1.5. At the height of the Ur III expansion, therefore, a considerable territory was encompassed, stretching from southeastern Anatolia to the Iranian shore of the Persian Gulf. But in contrast to the situation during the Akkadian expansion, the Middle Euphrates and Khabur regions, along with the central and western parts of Syria, remained almost unaffected by these campaigns.^[7] Relations with these western and northwestern areas seem to have been peaceful,^[8] based mainly on a close alliance with Mari.^[9] Danger (and especially the “Amorite” danger) never came from the west for the kings of Ur.

§1.6. These military expeditions by these Neo-Sumerian kings, often conducted in difficult mountainous areas and remote regions, sometimes over a thousand kilometers from the city of Ur as the crow flies, must have constituted major undertakings, requiring a strong military organization with complex logistical capacities. It is naturally frustrating, therefore, to find that the massive documentation for this period^[10] contains virtually no evidence about either these armies or the kingdom’s military organization in general. Even with such a quantity of texts, we cannot clearly fathom how soldiers were recruited, supplied, maintained, equipped, armed, organized, disciplined and launched into battle. Investigations based on the remains of the Neo-Sumerian bureaucracy yield few results, though these can be interesting and informative. We are left with the feeling that no proper military administration ever existed, solely because we lack archives relevant to the subject (but see fn. 1 above).

§1.7. Without such archives, the only possible method for research on the Ur III armies is to assemble scattered evidence from the multitude of available sources (administrative archives, royal inscriptions, literary texts, etc.), trying wherever possible to cross-check them with available archaeological and iconographic evidence. In the end, this process constitutes a considerable investigation that, if sufficiently exhaustive, would lead far beyond what is feasible within the scope of this contribution. We will thus attempt here only to explore selected lines of research.

§2. Some characteristic military vocabulary in the administrative archives

§2.1. To conduct such an enquiry, we can of course begin by gathering all the Ur III administrative texts that mention “soldiers” (aga₃-us₂, erin₂). But with the resulting amalgam of records, many of them not relevant, it becomes difficult to create a meaningful classification. Nevertheless, we will naturally return to these different categories of soldiers and what can be learned from this mass of references (below §4).

§2.2. Another characteristic term to look for is nam-ra-ak, Akk. šallatum, “booty.” The hundred or so texts that mention this word show in clear chronological terms the opportunities seized by the Ur III armies to pile up and bring back loot in the course of their victorious campaigns. These often corroborate the evidence from year names or royal inscriptions mentioning the relevant military expeditions:^[11]

• nam-ra-ak an-ša-an^ki (Š 33)
• nam-ra-ak kur mar-du₂ (Š 40, Š 44, Š 46, Š 47, Š 48, AS 1, AS 4)^[12]
• nam-ra-ak lu-lu-bu^ki (Š 44, cf. Maeda 1992: 157)
• nam-ra-ak šu-ru-ud-hu-um^ki (Š 44, AS 4)
• nam-ra-ak ur-bil₃-lum^ki-ma (Š 45, Š 48)
• nam-ra-ak LU₂.SU(.A) (Š 47, Š 48)^[13]
• nam-ra-ak ha-ar-ši^ki (Š 48)
• nam-ra-ak hu-ur₅-ti^ki (Š 48)
• nam-ra-ak ki-maš^ki (Š 48)
• nam-ra-ak ša-aš-ru^ki u₃ šu-ru-ud-hu-um^ki (AS 4)
• nam-ra-ak iri nergal^ki / iri mes-lam-ta-e₃-a (AS 5)

§2.3. We can see from this list that the Šulgi campaigns were by far the most profitable in terms of booty taken from the enemy, and that the kings of Ur ceased to earn war treasure by the middle of Amar-Suen’s reign. After the 5^th year of this king, there is no more evidence of booty. Further detailed analysis is needed, but a quick review of all the texts and documents gathered thus shows that the “loot” mainly consisted of men (lu₂ nam-ra-ak),^[14] women (geme₂ nam-ra-ak),^[15] and animals (gu₄ nam-ra-ak / udu nam-ra-ak / maš₂ nam-ra-ak).^[16] All of these had to be kept and fed (ša₃-gal nam-ra-ak / še-ba nam-ra-ak / i₃-ba nam-ra-ak).^[17] The booty also consisted of materials, including wool (siki nam-ra-ak, BPOA 6, 796^[18]), tarred baskets with unspecified contents (TCL 5, 6036), valued metal objects (TSU 39), animal skins (Princeton 1, 130; OIP 115, 355), etc. These texts also illuminate how the army, after returning home, could offer booty recovered from war to the king (Ontario 1, 50, etc.) and to the gods.^[19]

§2.4. All this evidence for “booty” is significant in that it shows the extent to which these armies, in their campaigns, were predatory conquerors, not just sitting virtually and feeding themselves on the backs of the vanquished,^[20] but also bringing back home (at least until the mid-reign of Amar-Suen) a surplus (slaves, livestock, treasures, various objects and materials), of which the king and the temples were the main beneficiaries.

§2.5. Another similar search can be conducted concerning the Sumerian word ugnim, Akk. ummānum. Does this word simply mean “army”/“troops,”^[21] or does it also represent a “military camp” and even a proper name for the place where one of these camps was located, especially when written with the {ki} determinative, characteristic of place names?^[22] R. Englund, who addressed this question some years ago, believed that ugnim was “The military troop gathered on the occasion of the implementation of an expedition.”^[23] Beyond this definition, it is worth noting that the fifty or so references to this word (not so many, in fact), which is subject to frequent variations in spelling, mainly come from Girsu.^[24] Would this mean that only the province of Girsu brought together, in one particular place, the whole royal army before departing on a foreign campaign? Whatever the answer, it is worth noting the following categories and professions that are frequently associated with the word ugnim:

We see thus that the ugnim would appear—at least at Girsu—to be a kind of military establishment, clearly organized in the form of a “household” (e₂, Akkadian bītum). In fact, we find inside it:

• On one side (a, b, c), all the soldiers (erin₂), officers (ugula / nu-banda₃) and a specific category of recruited men (lu₂-nig₂-dab₅)^[25]: in short, the troop and its officers;
• And on the other side (d, e, f, g, h), the administration and the supply corps in charge of the functioning and provision of the whole group: chief-manager (šabra), scribes (dub-sar), servants (giri₃-se₃-ga), millers (geme₂ kikken₂), potters (bahar₂), etc.

§2.6. The relevance of the ugnim is highlighted by another text that warrants citation in detail. Like all the others, this text comes from Girsu, and it dates to the year IS 3:

[A] Owen 1973: 135, 3 (Fs. Gordon 1)
(1) 532;0,3. še gur (2) 158 ziz₂ gur (3) erin₂ ugnim^ki(=KI.SU.LU.ŠE₃.GAR.KI)-ma-ke₄-ne (4) u₄ kaskal mar-du₂ -še₃ i₃-re-ša-a (5) šu ba-ab-ti (6) kišib₃ lu₂-^dnanna (7) u₃ iš-ku-un-e₂-a (8) ki gu-za-na-ta (9) u₃-ma-ni šu ba-ti (Date: -/vii/IS 3)

“159,600 liters of barley (and) 47,400 liters of flour, received by the troops of the military camp when they went on an expedition against the Amorites. Sealed tablets of Lu-Nanna and Iškun-Ea from Guzana did Umani receive.”

Above all, it is the quantity of food distributed to the soldiers of the ugnim (here with the {ki} determinative) that calls for attention: it is considerable (200,000 liters) and therefore shows that a huge army was involved.^[26] Furthermore, the two individuals who signed by rolling their seals on the tablets are well known to be generals (šagina) in the royal army. And Guzana, who delivered the grain, is known as an important supplier of the troops with various commodities.^[27] Unfortunately, this campaign against the Amorites in the second year of Ibbi-Suen left few other traces in the documentation; at that time, the situation had become difficult for the last king of Ur, precisely because of the Amorite threat.^[28]

§2.7. Another administrative tablet is also instructive concerning the ugnim:

[B] MVN 10, 149 ii 6-9^[29]
(6) 70 guruš u₄ 1-še₃ (7) ugnim(=SU.KU.ŠE₃.KI.GAR.RA) ma₂-gan^ki-še₃ bala-a (8) 30 guruš u₄ 1-še₂ (9) ugnim(=SU.KU.ŠE₃.KI.GAR.RA) an-ša-an^ki-ta bala-a

“70 workmen days having ensured the transfer of the army as far as Magan; 30 workmen days having ensured the transfer (back) of the army from Anšan”.

In this exceptional piece of text, belonging to a small dossier concerning the military campaign led by Šulgi against Anšan,^[30] we see the army mustered and dispatched on expedition by the king, thus taken in charge and transported by some men—likely mariners maneuvering the ships—as far as Magan, far away at the entrance of the Persian Gulf. Then on returning, the same army was taken in charge again for its transportation, this time from Anšan in Iran and by a smaller number of men, perhaps because of casualties in battle or because the return trip was easier. This text, which is dated precisely to the “year when Anšan was destroyed.” (Š 34) is also interesting for its evidence regarding the maritime route used for the journey to fight against Anšan.

§2.8. Finally, based on the contexts and origin of the documentation, the ugnim seems to be both the meeting place or departure camp for the royal army, flanked by stewardship and general supply corps, and the army itself once “on the march” (cf. Latin agmen) or ready to go after being mustered.

§2.9. One last relevant word in the available records is the term kaskal, Akk. harrānum, gerrum, “expedition.” If we set aside the numerous uses of this word in the messenger texts, some interesting documents still remain. Thus, for example, this account tablet from Drehem dated 25/vii/AS 6:

[C] SAT 2, 913 = MVN 5, 115
(1) 12 udu (2) 83 u₈ (3) 25 maš₂ (4) 35 ud₅ (5) mu aga₃-us₂ kaskal-ta er-ra-ne-še₃

“12 sheep, 83 ewes, 25 male and 35 female goats for the soldiers who have returned from expedition.”

In total, 155 head of small livestock were thus kept for feeding soldiers in the regular army (for aga₃-us₂ see below, §4), returning from a campaign. If we consider that all this meat was consumed at one time (though this is not proven), knowing that one sheep can feed approximately sixty people,^[31] roughly 9,000 men would have benefited from this expense (see below, §8, for the numbers of the royal army).

§2.10. We find similar feeding distributions in a similar context with a batch of three tablets from Umma, all dated to the year ŠS 1, but without specification of a day:

[D] UCP 9-2-2, 7, TCNU 507, Nik 2, 337
4 kaš DU gur / aga₃-us₂ kaskal-ta / gen-na

“1,200 liters of ordinary beer for the soldiers who have returned from expedition” (in the third text, the quantity is 4,4.1. kaš gur = 1,450 liters).

This time, it is beer that has been delivered to the soldiers after a military expedition, and on three successive(?) occasions. This could be related to the well attested custom of the banquet offered by the king to his troops after return from a victorious campaign. Such a custom is documented, for example, through a series of Umma tablets concerning the naptanum-banquets.^[32] Here, if we estimate two liters per person and per occasion, it might well be that between 600 to 700 soldiers received beer at Umma under these circumstances.

§2.11. And several of these military expeditions (kaskal) are also mentioned in some Umma tablets, on the occasion of propitiatory offerings made to the “standard-weapon going on the campaign” (šu-nir ^geštukul kaskal-še₃ gen-na) and around which the army was probably ready to be on the move (OrSP 47-49, 344; MVN 4, 263; MVN 2, 52; MVN 5, 46; UTI 5, 3424; MVN 16, 655; CHEU 19; the dates of these expeditions are given, respectively: xii/AS 5, xii/AS 5, iii/AS 6, -/AS 6, -/ŠS 2, iii/ŠS 3, iii/ŠS 4; they probably refer to the campaigns that were undertaken against Šašrum for Amar-Suen and against Simanum for Šu-Suen).

§2.12. Finally, this kind of enquiry into “some characteristic military vocabulary in the administrative archives” could be pursued for several other words, such as the names of weapons (see below §6). We must reemphasize here that such research is very substantially facilitated by the electronic databases BDTNS and CDLI, allowing for exhaustive search of the data, which are then possible to sort according to multiple criteria (date, place of origin, etc.).

§3. The contribution of literary texts and sources other than administrative

§3.1. We have seen how the year names that date the administrative documents of the five kings of Ur are important for reconstructing a military history of this period (above §1). Several successful investigations have linked these year names to administrative archival documents or royal inscriptions, which give some evidence for how these campaigns were conducted.^[33] But as for the year names themselves, how should we understand one such as the year Šulgi 20?

[E] Year name Šulgi 20
mu dumu uri₅^ki-ma lu₂-geš-gid₂-še₃ ka ba-ab-keš₂

“Year the citizens of Ur were drafted as spearmen.”

In 1987, P. Steinkeller proposed that in this particular year, Šulgi established a new professional army in the framework of a series of vigorous reforms undertaken during the last two-thirds of his long reign.^[34] This view has since been debated,^[35] and the event reported here may be a simple and small one, certainly interesting in itself, but limited in its effect and perhaps linked only, in this particular year, to the need for an additional corps of spearmen in the army.^[36]

§3.2. The royal Correspondence of Ur also offers useful evidence for military organization. Without returning to the ongoing debate over the “authenticity” of the relevant documents,^[37] the letter RCU 1 shows the royal grand vizier Aradmu reporting to king Šulgi that he met with Apillaša, a military chief in the Zagros region. Aradmu says the following about Apillaša:

[F] RCU 1
(19) aga₃-us₂ sag-ga₂-na 5 li-mu-um-ta-am₃ zi-da gub₃-bu-na ib₂-ta-an-gub-bu-uš

“To his right and left he had his elite(?) soldiers stationed, five thousand at each side.”

Here, we have a powerful military commander, appointed by the king of Ur, and about whom we learn that the troops available to him consisted of at least 2 × 5,000 = 10,000 soldiers. These are defined as aga₃-us₂ sag-ga₂-na, literally “soldiers of his head.” sometimes translated as “elite soldiers.”^[38] No other text from this period refers explicitly to so many soldiers at one time. Apillaša is a well known figure in the Ur III documentation:^[39] he was a very important military chief, and we will encounter further records concerning him (below text [Q]). But what is most important here is to see the stage set for the homecoming of a royal envoy, Aradmu, whom Apillaša received at the head of his elite force composed of ten thousand soldiers.^[40]

§3.3. On the crucial question of conscription, the Gudea Cylinder A offers important and unique evidence. This text helps understand how troops could be raised at the heart of the Lagaš kingdom, just before this land was incorporated into the domain of the kings at Ur. We find the following:^[41]

[G] Gudea Cylinder A, col. xiv
(7) u₄-ba ensi₂-ke₄ kalam-ma-na zi-ga ba-ni-gar
(…)
(11) iri du₃-a a₂-dam-gar-ra-na
(12) gu₂ geš-ba-ra ^dnanše-ka
(13) zi-ga ba-ni-gar
(14) gu₄ huš zi-ga gaba-gi₄ nu-tuku
(15) ^gešeren babbar₂-ra lugal-bi-ir dab₆-ba
(16) im-ru-a ^dnin-gir₂-su-ka-ka
(17) zi-ga mu-na-gal₂
(18) šu-nir-mah-bi lugal kur dub₂ sag-bi-a mu-gub

“In those days the ruler (= Gudea) imposed a levy on his land.^[42] (…) On his built-up cities, his rural settlements, on the edge of the desert^[43] of Nanše, he imposed a levy. There was a levy for him on Ningirsu’s clans (having the emblem) ‘Rampant fierce bull that has no one to oppose it’ (and) ‘White cedar passing along for its master,’ and he placed their magnificent standard, ‘King who makes the mountain tremble,’ in front of them.”

In this passage, which in fact continues to l. 27, we can see how Gudea undertook a systematic draft of all the population liable to corvée throughout his land, in order to complete the building of Ningirsu’s temple. This is one of the rare documents from this period that alludes to the conscription process. Furthermore, it shows how this conscription was carried out by clan or tribe (im-ru-a), each one with its emblem (šu-nir) or totem, and that no one was exempt. In this case, the conscription involves enrollment for public works, of course, but we can imagine in all likelihood a similar procedure for enrollment into the army when there was a need to go to war.

§3.4. Another literary text from this period, the Šulgi Hymn D, provides several details regarding the weapons used in battle (see below §6). In the section of this composition that mentions the destruction of his enemies, Šulgi affirms at first:^[44] “I set up my emblem at the border of the foreign lands” (^geššu-nir-gu₁₀ kur-ra zag-ba ga-am₃-du₃, l. 177). Then he describes the range of weapons that he would use and specifies what effect would each one would have on his opponents: the spear (geš-gid₂-da, l. 177), the battle axe (^geštukul-ha-zi-in, l. 191), the complex bow(?) (GEŠ.ŠUB=illuru)^[45] and other weapons for which the translation is uncertain. This section also mentions the regular bow (^gešban, l. 180), arrows (^gešti, l. 181) and the quiver (e₂-mar-uru₅, l. 179). Main of these are offensive weapons, which would ensure him victory.

§3.5. These literary passages on weapons and their use are echoed in the last literary text that we will mention here: the Lamentation over the destruction of Sumer and Ur, which likewise alludes to the weapons used to fight:^[46]

[H] LDSU
(382) uri₅^ki-ma ^urudaha-zi-in gal-gal-e igi-bi-še₃ u₃-sar i₃-ak-e
(383) geš-gid₂-da a₂ me₃-ke₄ si ib₂-sa₂-sa₂-e-ne
(384) ^gešban gal-gal ^gešilluru ^kuše-ib₂-ur₃-ra teš₂ im-da-gu₇-e
(385) ^gešti-zu₂-ke₄ muru₉ šeg₃-ga₂-gin₇ bar-ba mi-ni-in-si
(386) na₄ gal-gal-e ni₂-bi-a pu-ud-pa-ad im-mi-ni-ib-za

“In front of the city of Ur, formidable battle axes were sharpened. Spears, the arms of battle, were made ready. The leather shields were being devoured by the terrible bows and the complex bows, all together. Barbed arrows covered their outer surfaces as if from a storm cloud. Formidable (sling-)stones fell together with great thuds.”

We can observe through these examples that the army’s offensive weaponry seems to include: battle axes and spears for the hand-to-hand combat, and bows and slings for striking at a distance.

§3.6. From a methodological point of view, these few examples show how we may proceed to collect further evidence on our subject. By pursuing such enquiries, and by creating a synthesis of the results, we can address several important questions relating to armies in antiquity: military organization, hierarchy, weaponry, garrisons, and force. These five points are treated directly in the remainder of this article.

§4. The troops, the regular army and the different categories of soldiers

§4.1. To begin in brief, we must keep in mind an essential principle: the army, when mobilized, probably essentially consisted of the male population of the kingdom. Every able-bodied man (called guruš in the texts) could be called up for duty when troops were mustered throughout the land, as seen in the Gudea text (§3.3 text [G]). Throughout the year, every man (guruš) was thus obliged to devote a fixed time in service as a “member of the troop” (erin₂, Akk. şābum). He could then, depending on situation and need, be assigned either to work on civil projects such as construction, drainage, agriculture, etc., or to military service.

§4.2. Thus, as evident in the archival texts, the administration classified each individual according to whether he was on active duty, in the category of erin₂ bala gub-ba; or he was free from current service and was placed under the category of erin₂ bala tuš-a. K. Maekawa, to whom we are indebted for this discovery, has shown in several articles^[47] how the alternating status of the erin₂ was managed on a monthly basis in the framework of corvée. At Girsu, they passed from one month to the other, from the erin₂ bala-gub-ba category to the erin₂ bala-tuš-a category. Logically, this monthly rhythm seems impractical for military conscription, but evidence is lacking on this matter and, as stated by K. Maekawa, “the yearly length of days of the corvée to be assigned to a single person still remains to be studied.”^[48] Nevertheless, based on both the Girsu and the Umma archives, it is conceivable that the erin₂ would have devoted at least half of their time to state service.^[49]

§4.3. The extensive Ur III administrative documentation shows what kind of severe control was exercised on the population, so that no one could escape this form of conscription, whether civil or military. In some cases, men could be “seized by weapons” (^geštukul-e dab₅-ba). Texts also mention several categories of “seized” or “dragooned” individuals (lu₂-dab₅-ba, dumu-dab₅-ba, gan-dab₅-ba, etc.),^[50] as well as individuals who tried to escape, and as a result had to face punishment and imprisonment, once they surrendered or were caught.^[51] We are quite well informed about these people during their time in service, thanks to the many texts recording the rations, food or otherwise, that were delivered to them at that time, and to the tablets relating to their work assignments.

§4.4. Finally, even without definitive proof, it is reasonable to think that the assignments of the erin₂, when they were “on duty” (erin₂ bala gub-ba), were organized so as to cater to the needs of the state, not only for civil works (corvée), but also for all the military obligations: garrison service, internal defense and security for the kingdom, expeditions abroad, etc. This explains the many references to erin₂ in military contexts (see for example the erin₂ kaskal-še₃, troops ready to go in an expedition: BIN 5, 135; MVN 10, 216; etc.). Still, we do not know in detail either how the male population of the kingdom was called into the army or how it was actually gathered and organized for these military duties (but see above §2.7-8 the function of the ugnim).

§4.5. Besides the erin₂ troops, for which a mandatory but temporary service was imposed, the texts clearly show that there was also a permanent army, composed of professional privates. These are the aga₃-us₂, Akk. rēdû, “professional soldiers.” This Sumerian word appears for the first time in the Presargonic documentation from Lagaš, but it also is known at Ebla.^[52] An abundant literature on the term is already available, and there is no point in reviewing it all here. The prevailing idea, however, is that the aga₃-us₂ would not appear to be a real “soldier” as such, but more a kind of “guard.” considering the numerous civil security activities in which he seems to have been involved.^[53]

§4.6. Against the prevailing view, some arguments assert that, during the Ur III period, the aga₃-us₂ was first and foremost a professional soldier of the standing army.^[54] Many records clearly show the aga₃-us₂ in specifically military activities (some examples above §2.9-10 texts [C] and [D]), particularly in the entourage of the king and of the army’s leadership (below text [O]). His life was that of a soldier (below §7 about garrisons); he was provided with weapons, for the use of which a regular regime of training was necessary (below §6 text [Q] and TIM 6, 36), and he clearly served under a military chain of command (below §5).

§4.7. Nevertheless, as war was obviously not a permanent situation, and the armed forces under royal power could easily be deployed in many sectors of civilian life and administration, especially to ensure internal order and security, the aga₃-us₂ could also be employed:

• as personal bodyguards, not only for the king’s guard of course (aga₃-us₂ lugal), but also for the highest officials and department heads of the administration (aga₃-us₂ sukkal-mah / aga₃-us₂ ensi₂ / aga₃-us₂ sanga / aga₃-us₂ šabra / aga₃-us₂ šar₂-ra-ab-du / aga₃-us₂ nu-banda₃-gu₄ / aga₃-us₂ zabar-dab₅ / aga₃-us₂ PISAN-dub-ba / aga₃-us₂ PN);^[55]
• as part of teams in charge of maintaining order and security in the many administrative and official households, establishments and production units (e₂ and various giri₃-se₃-ga) of the kingdom.^[56] See also, in this category, some specialized aga₃-us₂, such as the aga₃-us₂ i₃-du₈, “main door guard / janitor” (SET 270 ii 41) for example, or the aga₃-us₂ ka (for ka₂) e₂-gal-ka, “soldiers of the palace gate”;^[57]
• as messengers or couriers (see below § 4.8-9), acting sometimes as representatives with diplomatic or political capacity (aga₃-us₂ kin-gi₄-a);^[58]
• as constables in charge of communications, circulation and transportation. They could even, in the latter case, be used for towing boats, providing security for convoys and supervising transportation teams, on the canal network that enabled circulation through the land of Sumer. This is notably evidenced by the many references connecting aga₃-us₂ to boats and involving them in inland water transport activities (see for example below, §4.16, or the many texts showing the activity of aga₃-us₂ ma₂-gid₂, “boat towing soldiers”). In many cases we also find aga₃-us₂ messengers acting as guides, or escorting individuals or groups. See also, along the same lines, the numerous records that mention aga₃-us₂ acting as official “intermediaries” or “conveyors” (giri₃) in administrative operations;
• in specified and one-time tasks requiring an abundant but temporary labor force, such as shearing sheep (SACT 2, 291), or especially during the harvest or for special agricultural tasks. The regular army could then be requisitioned by the authorities in addition to other categories of the population; see, for example:
  — TCTI 1, 621 i 9 (with 732 aga₃-us₂)
  — TÉL 239a i 13 (with 761 aga₃-us₂)
  — TCTI 1, 742 i 10 (with 796 aga₃-us₂)
  — ASJ 13, 227 72 ii 9 (with 900 guruš u₄ 1-še₃ = aga₃-us₂ lugal-me)
  — ASJ 19, 141 126, 23 // TCTI 2, 3817 (with 9,600 aga₃-us₂)
  — ASJ 8, 118 33, 9 (with 10,800 aga₃-us₂)
  These last two texts provide quite exceptional numbers—approximately 10,000 men. They represent perhaps the entire population of aga₃-us₂ available at a given time in the Girsu province. All this offers useful insight regarding the numbers in the professional army (see below §8).

§4.8. Two observations can be made regarding the many references to aga₃-us₂ in the specific tablets called messenger texts, besides the fact that the missions entrusted to them were extremely varied.^[59] First, in the messenger texts of Girsu, there is a category labeled as aga₃-us₂ gal and aga₃-us₂ gal-gal that occurs only in this kind of tablet and almost exclusively at Girsu.^[60] This category must refer to a distinctive group of special aga₃-us₂, appointed by the royal power to a particular communications service in the Girsu province. Accordingly, the epithet gal and gal-gal here are specific and have no hierarchical value in connection to the other categories of aga₃-us₂ soldiers serving in the army (see below §5 on the chain of command).

§4.9. Second, in the messenger texts of Umma,^[61] one could consider it odd that no aga₃-us₂ ever appears explicitly, unlike in the Girsu messenger texts. This leads then to ask whether the aga₃-us₂ may not in reality be hidden behind the professional category denoted ka-us₂-sa₂, which is peculiar to the Umma messenger texts. Thus, ka-us₂-sa₂ would only be a local phonetic variant of the aga₃-us₂. The contexts in which these ka-us₂-sa₂ appear at Umma fit well with such an idea.^[62]

§4.10. In conclusion, all these examples suggest that even if the aga₃-us₂ were military professionals of the standing army, they were also responsible within the kingdom for internal security, including the protection of state officials and public institutions. They had to oversee various types of movement within the land, ensuring the proper functioning of transportation and communications systems for the entire territory. In ordinary times, the aga₃-us₂ role could finally be compared in a way, mutatis mutandis and through the millennia, to the role played today in France by “gendarmes.” In this country, the gendarmes depend specifically on the Ministry of Defense and not on the Ministry of the Interior, as opposed to the police force; they are therefore soldiers.

§4.11. Classically, and like other categories of permanent dependants of the royal administration, remuneration of the aga₃-us₂ was organized through the allocation of cultivable land on one hand,^[63] and by the well known system of “rations” distribution on the other hand (see above, fn. 32). Many text files record such allocations to the aga₃-us₂ for various types of rations: barley (še-ba aga₃-us₂), oil (i₃-ba aga₃-us₂), fish (ku₆-ba aga₃-us₂), wool (siki-ba aga₃-us₂), clothing (tug₂-ba aga₃-us₂).^[64] We will also quote below (§4.20 texts [J] and [K]) two administrative letters that evoke the way in which the aga₃-us₂ were remunerated.

§4.12. We may also understand that these professional aga₃-us₂ were those who could hold cylinder seals in their name, proving their status as state officials.^[65] We currently know a hundred or so different cylinder seal impressions of aga₃-us₂ for the Ur III period. The great majority of these seals are from the archives of Umma. And it is worth noting that most of their holders are aga₃-us₂ lugal or aga₃-us₂ ensi₂, therefore in the service of the king or of the governor of Umma. In other respects, we can see that these professional aga₃-us₂ had full legal capacity. They acted sometimes as witnesses in legal transactions (igi+PN, lu₂ inim-ma-bi-me, etc.) or they possessed slaves of their own (for example: ITT 2, 3516 = NSGU 166, ll. 15-17).

§4.13. If we thus accept that the Ur III aga₃-us₂ were professional soldiers of the royal army, also employed for all kinds of occupations in logistics, communication and security duties throughout the kingdom, it is even perhaps possible to go further and to consider that this category of soldiers could have managed and trained the inexperienced conscript troops, after these were mustered. One Umma text, at least, would seem to show it: OrSP 47-49, 466, where a series of breakdowns is made in the form:

N aga₃-us₂ / N egir-erin₂ / nu-banda₃ PN

“So many professional soldiers / so many following conscripts / (their) captain (being) PN.”

This document shows in any case a clear separation between the two categories of soldiers and we see in this case that there was on average one aga₃-us₂ for every four erin₂ “following” him.^[66]

§4.14. To conclude on this point, we can summarize the military organization of the Ur III kings according to the pattern described in figure 2.^[67]

Figure 2: Men in Ur III service.

§4.15. At least two archival tablets may contradict the clear distinction that is proposed here between these two components of the royal army, with the conscripts (erin₂) on one hand and the professional soldiers (aga₃-us₂) on the other. These are the only texts where the two terms are joined in the form erin₂ aga₃-us₂, as if a single category:

Two possibilities, not necessarily mutually exclusive, could explain such a situation. Either we have a simple case of juxtaposition, with the pooling of a joint force including both conscripts and professional soldiers (“conscripts troop and royal soldiers of Garšana” in the first example, “sustenance of conscripts and regular soldiers” in the second example); or all these men were conscripts (erin₂), who were serving as soldiers (aga₃-us₂) during their term of service. This second explanation deserves consideration; in particular, it would help us understand a text as TÉL 182, which records “barley rations for regular soldiers from Nigin^[68] during their time of service” (še-ba aga₃-us₂ dumu nigin₆^ki bala gub-ba-še₃). According to the description proposed above, the expression bala gub-ba (“on duty”) should apply only to the conscript troops of the erin₂— who were alternating between time on and off duty—not to the professional aga₃-us₂ soldiers whose service was permanent. We must therefore acknowledge that conscripts (erin₂) could fulfill their duty by serving as aga₃-us₂ soldiers.^[69]

§4.16. The fifteen references where guruš and aga₃-us₂ are mentioned together in the single expression “guruš aga₃-us₂” pose no particular problem: they simply refer to men (guruš) hired or recruited as regular soldiers (aga₃-us₂).

§4.17. It should be noted that the texts display further parallels and alternations between the situation of the erin₂ and that of the aga₃-us₂. For example, in Girsu administrative documentation, the category of the erin₂ ma₂-u₄-zal-la (TCTI 2, 2747 or 3896) occurs in close parallel with that of the aga₃-us₂ ma₂-u₄-zal-la (TCTI 2, 3205 or 4074). Here, the soldiers in question, whether conscripts or professionals, were probably those assigned to go and sail an official boat that was scheduled to leave Girsu every evening by the canals network of the land of Sumer.^[70]

§4.18. Finally, two other important issues deserve further attention, as raised by allusions in a few texts: the recruitment of the aga₃-us₂, and the duration of their service. As will be seen, the aga₃-us₂ were most often recruited on an individual basis.

§4.19. Two Girsu tablets (ITT 4, 7131 = MVN 6, 130, and CT 1, 4-5 94-10-15 4: ii 13) refer to the situation of some “royal soldiers seized among the shepherds” (aga₃-us₂ lugal sipa-ta dab₅-ba-me). To this must be added a reference to some “soldiers (who are) fifteen young seized men” (aga₃-us₂ dumu-dab₅-ba 15-bi, TSU 9 = TCS 1, 234). The impression that dominates in these three examples is that of recruitment of soldiers “by force”. The context would thus be conscription for a term of service (bala). This interpretation, however, is hindered by one Umma text, dated Š 47, which evokes the payment of bonuses to an individual, the day he joined the regular army as a soldier:

[I] YBC 15411
(1) 6 udu ^dba-ba₆-mu (2) u₄ nam-aga₃-us₂ i₃-ni-in-ku₄-ra

“6 sheep for Babamu, the day he entered the status of regular soldier.”

This text could refer to the voluntary enlistment of an individual in the standing army,^[71] and the clear opposition observed in such situations would confirm (although the latter is, curiously, unique) the scheme proposed above, which distinguishes conscripts and volunteers.

§4.20. Two pieces of the Ur III administrative correspondence (letter orders) deal with the recruitment of soldiers and its consequences for the management of rations accounts. The first one, from Girsu, refers to the recruitment of former vintners:

[J] TCS 1, 86
(1) ka₅-a-mu (2) u₃-na-a-du₁₁ (3) lu₂-geštin-a (4) aga₃-us₂ ba-si-ga (5) siki-ba-bi ha-ba-ab-sum-mu (6) še-ba-a za₃ ha-ab-us₂-e

“Tell Ka’amu to give the(ir) wool rations to the vintners who have been made soldiers. (Tell) him to set aside the(ir) barley rations.”

A second similar letter, also from Girsu, uses the same Sumerian verb (sig) to evoke the transfer to the army of individuals and the payment of their rations:

[K] ITT 3, 5558 = TCS 1, 110
(1) lu₂-kal-la-ra (2) u₃-na-a-du₁₁ (3) lu₂ še-ba aga₃-us₂-lugal-ka ba-an-si-ga (4) mu didli-bi-še₃ (5) siki-ba-bi (6) ha-ba-ab-sum-mu (7) ˹im˺ siki-ba ba-[ga₂]-˹ga₂˺ en₃ ha-ab-tar-re

“Tell Lukalla to give the(ir) wool rations individually to those who get barley rations and have been made royal soldiers. (Tell) him to make certain that this is entered on the wool ration (account) tablet(s).”

It is hard to say here whether we are dealing with conscription for a term of obligatory service or with the recruitment of new professional soldiers, just hired.

§4.21. We must also mention one more Girsu text, which specifies that an individual recruited as a soldier must see his name registered on the original list tablets (im-ama = Akk. țuppāt ummātim):

[L] ITT 5, 6712
(1') [ku?]-li (2') al-la ra₂-gaba i₃-dab₅ (3') nam-aga₃-us₂-[še₃] (4') im-ama dah-he-dam

“Kuli, whom Alla the courier has taken in charge for (his integration with) the status of regular soldier, is to be added to the original record tablets.”

Thus, complete rolls listing the names of active soldiers and of the rations that were distributed to them were kept up to date very precisely, though unfortunately not one seems to have been discovered. In this context it is interesting to find in the tablets the title dub-sar aga₃-us₂, “scribe of the soldiers” (ITT 4, 7467 = MVN 6, 443).

§4.22. Another Girsu text sheds light on the bookkeeping of such lists, with evidence for how a child had to be removed from the workforce list of a weaving workshop, when he was removed from his mother to become a soldier:

[M] ITT 5, 6795
(1) 1 ur-^deš₃-he-nun-ka (2) im-e tak₄-a (3) dumu geme₂-ki-gu-la uš-bar (3) aga₃-us₂ tur-še₃ (4) inim lu₂-inim-nig₂-sa₆-ga (5) ša₃ gu₂-ab-ba^ki-ka (6) Date (IS 1)

“One (individual named) Ur-Ešhenunka, son of the weaver Geme-kigula, erased from the register, for (becoming) a child-soldier on the orders of Lu-inimnigšaga in the Gu’abba (province).”

Was this Lu-inimnigšaga the same one as the “captain” (nu-banda₃) known by ITT 2, 651 (also “ša₃ gu₂-ab-ba^ki”)?

§4.23. The individual recruitment of a man is also recorded in two remarkably similar tablets from Umma. Each of them starts in the same way:

[N] SACT 2, 134, and YOS 4, 155
(1) 1 lu₂-^dsuen (2) aga₃-us₂ ensi₂-ka-še₃ (3) …

“One (individual named) Lu-Suen, (to serve) as soldier of the Governor.”

The first of these two texts records the recruitment of Lu-Suen in the 11th month of the year ŠS 1; the second in the 1st month of the year ŠS 6. The key issue here is whether or not this is the same individual in both tablets. If so—and it is an attractive hypothesis, given the parallelism of these two texts, which are also the only ones of this kind in all our documentation—Lu-Suen would have been hired for the first time in the year ŠS 1, then renewed in his duties a few years later. This could allow us to estimate the average duration for the initial term of service to be performed by a committed aga₃-us₂: four years, renewable?

§4.24. Finally, this assumption regarding the limited duration for a term of service could also help clarify the meaning of the Girsu tablet ITT 5, 6902, which mentions some aga₃-us₂-lugal šu-bar-ra-me, “released royal soldiers.” Quite a number of Ur III tablets refer to people who are “released” (šu bar-ra) from their obligations or from their deprivation of liberty, but this text is the only one that relates specifically to the aga₃-us₂. Was it after such a period of four years that the aga₃-us₂ were usually “released”? Or were the specific aga₃-us₂ of this example simply liberated after having spent time in jail? An answer to this question and new and interesting information are given by an unpublished tablet of Irisagrig: this text, to be published by D. I. Owen in his forthcoming volume Nisaba 15, shows that land allotments (GAN₂) and rations (šuku) were attributed to these “soldiers after they have been released (from service)” (šu-bar-ra aga₃-us₂-me).

§4.25. Without further elaboration, let us finish this chapter by considering the categories of soldiers encountered in our archival texts that may be categorized according to their geographical origin. Two main groups are involved: Amorites (mar-du₂) and Elamites (elam). For each of these groups we have clear and frequent references linking these “ethnic” categories with the aga₃-us₂ function (aga₃-us₂ mar-du₂, aga₃-us₂ elam, passim).^[72] This identification is often on an individual basis, however, particularly in the messenger texts. Moreover, P. Michalowski has argued recently that “in the language of the Ur III administrative texts the word elam designates highlander bodyguards who were essentially the counterparts to ‘native’ aga₃-us₂ guardians.” These Elam aga₃-us₂ were “not directly employed by the Ur III state.” and “there is absolutely no evidence to support the notion that they were part of the Ur III military establishment.”^[73] A similar or parallel phenomenon has been observed by W. Sallaberger concerning the mar-du₂: he proposes not to identify them “by their homeland or language” and declares instead that “Mardu/Amorite came to mean also ‘nomad’ in Babylonia.”^[74] In the case of the mar-du₂, however, it must be kept in mind that the personal military guard nearest to the king himself was precisely composed of a dozen of these “Amorite” soldiers.^[75]

§5. The chain of command

§5.1. It is the king who leads the army to war and is usually in command, at least virtually and in propaganda. This is notably suggested by the year names, the iconography and the royal inscriptions and hymns. Close to him, and at least during two decades (between AS 3 and IS 3), it is clear that the grand vizier (sukkal-mah) Arad-Nanna^[76] played a particularly important role in the military life of the kingdom.

§5.2. The hierarchical organization within the army can be reconstructed quite clearly from numerous sources. There are three main ranks of officers above the basic soldier (aga₃-us₂):^[77]

§5.3. All these officers, as seen in many texts, can command both aga₃-us₂ and erin₂ units, of course. Here are three examples with aga₃-us₂:

§5.4. Among the numerous tablets from Drehem that record withdrawals of animals “delivered to the kitchen for (feeding) the soldiers” (šu-gid₂ e₂-muhaldim mu aga₃-us₂-e-ne-še₃),^[78] a recently published tablet is particularly interesting. It describes for the first time in detail the full chain of command for the army, with the ultimate precision that all these soldiers had just returned from an expedition (text dated 22/vii/ŠS 1):

[O] Dahl and Hebenstreit 2007: 35-37 no. 1
(1) 3 gu₄ 255 udu (2) u₄ 21-kam (3) 90 udu (4) šu-gid₂ e₂-muhaldim-še₃ (5) mu šagina nu-banda₃ (6) u₃ ugula-geš₂-da kaskal-ta (7) er-ra-ne-še₃ (…)

“3 oxen, 255 sheep on the 21st day, (and) 90 sheep, šugid-delivery to the kitchen, for the generals, the captains, and the ‘commanders-of-sixty’, who have returned from campaign.”

This tablet must be added to texts [C] and [D], seen above (§2.9-10). Here, only officers (šagina / nu-banda₃ / ugula geš₂-da) were the focus of the distribution; but the aga₃-us₂ soldiers, usually the main beneficiaries in such texts, were doubtless invited to the feast as well, if we judge by the quantity of meat obtained from the slaughter of three cattle and 345 sheep! If all that meat was consumed at once, it would have benefited as many as 22,500 men (or 11,250 men over two days),^[79] which is considerable!

§5.5. Finally, it is worth noting that this chain of command (šagina / nu-banda₃ / ugula) is exactly the one that is found in the Gilgameš and Akka tale. In a closing tribute to his vanquished enemy, king Gilgameš declares to Akka king of Kiš:^[80]

[P] Gilgameš and Akka
(102) ak-ka₃ ugula-gu₁₀ ak-ka₃ nu-banda₃-gu₁₀ (103) ak-ka₃ šagina-erin₂-na-gu₁₀

“Akka my lieutenant, Akka my captain, Akka my general-in-command of the troops…”

§6. The weapons and equipment of war

§6.1. How was an ordinary soldier armed at that time? Along with the question of numbers, it is without doubt one of the most difficult issues we need to address. It warrants a systematic search of the archival texts in order to record all the names of weapons and to build up appropriate dossiers: ^gešKAK, ^gešban, geš-gid₂, ^geštukul, ^gešilluru, ^geššukur, gir₂, ha-zi-in, etc. We have already examined (above §3.4-5) some evidence on this subject that can be gleaned from the literary texts.

§6.2. Three categories of offensive weapons, however, appear with greater frequency in Ur III administrative archives and elsewhere and must therefore represent the basic weaponry of a soldier:

• The mace (^geštukul),
• The spear (geš-gid₂),
• And the bow (^gešban).^[81]

§6.3. It appears that the spear was the basic weapon for simple conscripts (erin₂);^[82] the bow would have been more readily used by professional soldiers (aga₃-us₂), who were in all likelihood trained regularly for its use (text [Q]). We saw above (§3.1. text [E]) how in the 20th year of his reign, Šulgi drafted a troop of spearmen at his capital of Ur. Such a military organization, based on these three categories of weapons, may be illustrated in the letter of Lipit-Eštar mentioned below (text [X]), which refers to the sending of troops to fight as reinforcements, composed in equal parts of spearmen (lu₂ ^geššukur), bowmen (lu₂ ^gešban), and soldiers armed with battle-axes (lu₂ dur₁₀-tab-ba). If this three-party distribution was confirmed as general, it would provide at least a tenuous indication of how to think about the organization of the fighting army, as well as of how military strategy could have been considered and conducted.

Figure 3: Bow, mace and spear in fighting scenes of the Akkadian period (stela fragment from Girsu. Louvre Museum, AO 2876; from Abrahami and Battini 2008: p. 104).

§6.4. Who supplied the weapons? With regard to conscripts (erin₂), we cannot be sure that it was always the central power. Professional soldiers (aga₃-us₂), in contrast, were certainly armed by the royal administration, as evidenced for example by this interesting text from Drehem, dated 2/xi/AS 2:

[Q] TIM 6, 34
(1) 1200 ^gešban (2) 1200 ^kušsag-e₃ / ^gešban e₂-ba-an (3) mu aga₃-us₂-e-ne-še₃ (4) hu-ba-a (5) u₃ a-pil-la-ša-ar (6) 1 za₃-mi-ri₂-tum zabar (7) geš-bi ku₃-babbar šub-ba (8) ah-ba-bu mar-du₂ (9) lu₂-DUN-a a-bu-ni-ra (10) ugnim(=KI.SU.LU.UB₂.GAR.RA)-še₃ (11) bur-ma-ma nu-banda₃ lu₂ zimbir^ki-ke₄ (12) u₃ su₂-ku-ku-um (13) lu₂-DUN-a lugal-ku₃-zu-ke₄ (14) in-ne-de₆-eš₂ (15) ki di-ku₅-mi-šar-ta / ba-zi (16) ša₃ PU₃.ŠA-iš-^dda-gan (17-18) Date.

“1,200 bows and 1,200 leather quiver (for) pairs of bows, for the regular soldiers of Huba’a and Apillaša. A bronze javelin, the wooden (pole) of which plated with silver, for Ahbabu the Amorite, subordinated to Abuni; (all this) for the ready-to-go army. Burmama, the captain of the man of Sippar, and Sukkukum, subordinated to Lugal-kuzu, have provided (these weapons) to them; (they are) issued by Dayyanum-mišar.”^[83]

What is described here is therefore the royal administration’s provision of a contingent of 1,200 regular soldiers employed as bowmen, with bows and quivers. Their officer, an Amorite,^[84] received a ceremonial weapon, a javelin plated with silver. Otherwise, Huba’a, Apillaša and Abuni are three well-known generals (šagina) of the royal army in the archives from this period. We can even consider that this Apillaša is the same as the main protagonist in the letter CRU 1 mentioned above (§3.2, text [F]), where he is at the head of 10,000 aga₃-us₂ soldiers.

§6.5. How were these soldiers dressed? And how were they protected? The texts are not very loquacious on this topic, nor for all matters relating to defensive weapons. We must insist, to finish this short chapter, on the fact that the Ur III army was first and foremost an army of marching soldiers, even if the use of war-carts is not completely ruled out, despite its virtual absence in our archival texts.^[85]

§7. Occupying the land: the garrisons

§7.1. In his pioneering article published some twenty years ago on the political and administrative organization of the Ur III Empire,^[86] P. Steinkeller sketched a bipartite picture, distinguishing the “core” of the kingdom, composed of about twenty provinces that contributed to the system of centralized taxation called the bala,^[87] from the “periphery.” with more or less controlled regions, held by garrisons under the authority of military governors (šagina) and subject to tribute (gun₂ ma-da, see map, figure 1).

§7.2. As for these “marches” of the empire, P. Steinkeller highlighted a series of administrative texts from Drehem that show how garrisons and military personnel at the eastern and northeastern periphery of Sumer and Akkad had to generate annually, on behalf of the Sumerian central power, a fee counted in cattle and representing the “tribute (imposed on) the land” (gun₂ ma-da) where they were stationed. These texts show clearly the military chain of command that we just described, now applied to garrisons: namely, the sequence “general”-šagina / “captain”-nu-banda₃ / “lieutenant”-ugula / “troop soldier”-erin₂. Each of these categories had to produce, according to rank, a greater or lesser number of livestock. P. Steinkeller counted nearly ninety of these Sumerian garrisons abroad, but we know nothing about their size, which would obviously have varied.^[88]

§7.3. Within the empire, the twenty or so provinces were organized under two distinct hierarchies: one civilian, under the authority of a civil governor (ensik), and the other one military, under the authority of a general (šagina) who acted as military governor.^[89] P. Steinkeller makes a crucial observation regarding the origin of these officials: while the civilian governors (ensik) of each province came mainly from local dynasties, deep in the heart of Sumer, it appears, however, that military governors (šagina) often bear Akkadian, Hurrian, or Elamite names, betraying their “foreign” origin. Moreover, the military governors were more “mobile” throughout their careers and closer, from many points of view, to the royal power.

§7.4. Each province of the kingdom therefore had appointed to it one or more military governor or general, according to the locations of garrisons and of troop mobilization. Completely independent from the ensik, these šagina reported directly to the central authorities, either the grand vizier (sukkal-mah) or the king himself. Their closeness and their loyalty to the king are highlighted by an important and interesting event (only known by a unique and, unfortunately, elusive administrative text of Drehem published recently) that took place at Ur sometime during the tenth month of Amar-Suen's seventh regnal year: A number, or perhaps even all, of the generals (šagina) of the realm were gathered in the city to swear a loyalty oath to the king (Steinkeller 2008). Responsibility to recruit and maintain military forces in each province was probably entrusted to these šagina. Beyond this, however, it remains to know where these garrisons were located within the kingdom; this is truly a considerable problem, as they do not appear clearly in the archival texts.

§7.5. Thanks to the dossier of 800 or so Drehem texts related to the “kitchen” (e₂-muhaldim), which register livestock intended to supply, among others, some number of aga₃-us₂ with meat,^[90] we get a first impression that there were garrisons of aga₃-us₂, as one would expect, at least in Ur, Nippur and Uruk, that is to say in the three capitals (political, religious, and historical) of the kingdom. Some of the available texts actually specify the place where the soldiers were provided with meat: ša₃ uri₅^ki-ma (Ur), ša₃ nibru^ki (Nippur), and ša₃ unu^ki (Uruk). It may nevertheless be better to conclude that these regular soldiers (aga₃-us₂), who were fed at Ur, Nippur, or Uruk via the e₂-muhaldim of Drehem, actually represented the same royal guard of aga₃-us₂, moving regularly with the king to escort him during his journeys to all three capitals.^[91] The same may apply to journeys to sanctuary-towns in the kingdom such as Gaeš, Idlurugu, and Tummal, which the king had to visit frequently, especially for regular attendance at the rituals (Akiti, etc.) that are mentioned in these texts.^[92] One example of such ceremonies is the a-tu₅-a-“lustration” ceremony, for which the king was actually accompanied by a particular group of his soldiers, the aga₃-us₂ a-tu₅-a-me. Finally, this particular unit of aga₃-us₂ provided through the Drehem e₂-muhaldim can be understood as having represented a kind of Pretorian Guard at the king’s disposal, like the one that would be constituted by Roman Emperors. Following the calculations made by L. Allred from the quantities of delivered livestock,^[93] this contingent of royal soldiers perhaps consisted of a few hundred men.^[94]

§7.6. According to the Ur III documentation as a whole, however, beyond the three major cities, the provincial capitals, and the king’s royal guard, there were probably at least two other main garrisons of aga₃-us₂ soldiers in the core of the kingdom:^[95]

§7.7. Apart from this important new archive, a small group of tablets is particularly interesting concerning Garšana: mainly from the administration of Girsu, they show that one of the principal garrisons of soldiers in the core of the kingdom was located precisely at Garšana. The “Lagashite” origin of these texts is interesting, as Garšana generally depended on the province and governor of Umma, as did Nagsu, Apisal, and Zabalam, for example.^[101] The tell of Garšana, which has not yet been identified, may have been located at the crossroads of the central provinces of Umma, Girsu and Uruk, and therefore south of Umma, perhaps in the area of Gu’edena.^[102] This file comprises the following six texts (in chronological order):

[R] AfO 18, 105, MAH 16285 (=CUSAS 3, 1440). Date: AS 9 (Drehem)
(1) 570 ^kušsuhub₂ e₂-ba-an (2) ki ensi₂ gir₂-su^ki-ta (3) erin₂ aga₃-us₂ lugal gar-ša-na^ki-ka-ke₄ (rev. 4) šu ba-ab-ti (5) nu-banda₂ e-lu₂-da-an (6) ša₃ unu^ki (7) giri₃ lugal-am-gal (8) Date.

“The troop of royal soldiers of Garšana received 570 pairs of leather boots issued by the governor of Girsu. The(ir) captain (is) Elu-dan. At Uruk, through Lugal-amgal.”

[S] TÉL 171 (=CUSAS 3, 1447). Date: ix/ŠS 6 (Girsu)
(1) 438 še (2) gur lugal (3) nig₂-ba lugal (4) aga₃-us₂ ša₃ gar-ša-na (5) ki al-la-mu-ta (rev. 6) šu-i₃-li₂ (7) šu ba-ti (8) i₂-dub ^geštir-ma-nu-ta (9) še ur-šu-ga-lam-ma (10-11) Date.

“131,400 liters of barley according to the royal measure, royal gift from the king (to) the soldiers at Garšana. šu-ilī has received (it) from Allamu, from the Tirmanu warehouse. Barley (run by) Ur-šugalama.”

[T] ITT 3, 6174 (=CUSAS 3, 1453). Date: xii/ŠS 9 (Girsu)
(1) 3000 kuš udu a-gar gu₇-a (2) 600 kuš udu a-gar nu gu₇-a (rev. 3) ki nig₂-u₂-rum-ta (4) mu aga₃-us₂ gar-ša-an-na^ki-še₃ (5) i-pa₂-li₂-is-e (6) šu ba-ti (7) Date.

“Ippalis received 3,000 tanned sheep skins (and) 600 non-tanned sheep skins, delivered by Nigurum for the soldiers of Garšana.”

[U] ITT 3, 5405 (=CUSAS 3, 1491). Date: IS 2 (Girsu). Collated by the author [*] at the Archaeological Istanbul Museum^[103]
(1) 1897 še gur (2) aga₃*-us₂* gar-ša-na-ke₄ (3) šu ba-ab-ti (4) ki ensi₂ gir₂*-su*ki*-ta (rev. 5) ba-zi (6) kišib₃ sukkal-mah (7) Date. “569,100 liters of barley, received by the soldiers of Garšana. Expense by the governor of Girsu. Seal of the grand vizier.”

[V] MVN 6, 280 (=CUSAS 3, 1459) (Girsu)
… (14) 20 gu₂ si^ki (15) aga₃-us₂ dumu gar-ša-an-na^ki-me (16) kišib₃ sukkal-mah ra-ra-dam (17) …

“20 talents of wool for the soldiers originating from Garšana. The seal of the grand vizier has to be rolled (on the tablet).”

[W] TCTI 2, 3543 (=CUSAS 3, 1458). undated (Girsu)
(1) 360 erin₂ <nu-banda₃> lu₂-ša-lim (2) 500 nu-banda₃ la-num₂ (3) 360 nu-banda₃ šu-i₃-li₂ (4) 147 nu-banda₃ PU₃.ŠA-ha-mi (5) erin₂ gar-ša-an-naki(6) …

“360 men of the troop: the(ir) captain (is) Lu-šalim. 500: the(ir) captain (is) Lanum. 300: the(ir) captain (is) Šu-ili. 147: the(ir) captain (is) Puzur-Hammi. (These are) men of the troop of Garšana.”

We must add to this list texts that mention a “general” (šagina) of Garšana, on an Umma tablet (SAT 3, 2073) and in the Garšana archive (CUSAS 3, 1424, and seal impression of Šu-Kabta, CUSAS 3, p. 436).

§7.8. Despite its disparate character, we must emphasize the exceptional nature of this small dossier.^[104] It has no equivalent elsewhere in the Ur III documentation, and it shows how these soldiers, who do truly appear to belong to a garrison at Garšana, were regularly supplied: 570 pairs of boots (text [R]), 3,600 sheep skins (text [T]), 600 kg wool (text [V]), 130,000 and then 570,000 liters of barley (texts [S] and [U]). These texts on the garrison of Garšana come from Drehem, Girsu, and Umma; they refer to Uruk, Girsu, and to the grand vizier (sukkal-mah), whose importance in the military affairs of the kingdom is well known and whose official titles show his ties to Garšana (RIME 3/2.1.4.13, p. 324). All this suggests that, in a central position within the kingdom, there was a key garrison of regular soldiers maintained by the Neo-Sumerian state (text [R] asserts that they were “soldiers of the king”). Further, it is not impossible to imagine that the general who commanded this garrison was precisely, at least for some time, Šu-Kabta, the man whose private archives were found, along with those of his wife.^[105]

§7.9. The last tablet of this small dossier, text [W], provides useful evidence for the composition of units stationed in the garrison of Garšana, the whole dossier having shown that it consisted of both conscript troops (erin₂) and professional soldiers (aga₃-us₂). Each of these units was commanded by a captain (nu-banda₃) and included between 150 and 500 men. Also, we learn that the total number of troops identified in the Garšana garrison could reach 1,367 men (erin₂).

§7.10. The rest of text [W], however interesting and unusual, is more difficult: it seems to break down the assignment of troops actually identified in several other places in the kingdom, as opposed to the personnel who had been originally anticipated.^[106] The scribe eventually recorded a total deficit of 340 men from the 3,000 he had hoped to muster at an initial meeting, or approximately 10% missing the call.

§8. The issue of numbers

§8.1. Text [W] offers a perfect transition to address one final point in this study. Albeit important, it is difficult to analyze, due to lack of adequate evidence: this is the question of numbers or force. If one accepts that the garrison of Garšana constituted a major barracks for the whole kingdom, the 1,300 troops (erin₂) stationed there (text [W]) did not, in the final analysis, represent a very considerable number! We must nevertheless acknowledge that when units of soldiers, whether aga₃-us₂ or erin₂, appear in our archives, they rarely consist of more than a few hundred men. It is rare to encounter groups of more than 1,000 soldiers in a military context. This contrasts with the large and exceptional mobilizations of labor during harvests, when up to 10,000 aga₃-us₂ may be gathered (see above §4.7, in ASJ 8, 118 33).

§8.2. In the absence of any explicit evidence, one possible method to reconstruct a more significant number of men gathered for some military occasion, such as departure to or return from campaign, is to take into account the texts recording distribution of food or drink to the soldiers, and to collect the quantities distributed and consumed. The problem in this case is that the duration of consumption is rarely mentioned: is the food or drink for one day, ten days or a hundred days? In documents relating to the army, distributions made in the framework of “banquets” to celebrate victory must take place at one time and therefore allow the following results:

• Text [C], Drehem: 155 head of cattle to feed the soldiers returning = 9,000 men (or 4,500 on two days, 3,000 on three days, etc.)?
• Text [D], Umma: 1,200 liters of beer for quenching the thirst of soldiers returning from expedition = 600 men?
• Text [O], Drehem: 3 cattle and 345 sheep to feed the soldiers returning from expedition = 22,500 men (or 11,250 on two days, 7,500 on three days, etc.)?
• Texts in the dossier concerning the naptanum-banquet at Umma:^[107] between 760 and 1,200 soldiers, beneficiaries of beer?
• UET 3, 1114, Ur: 1,450 liters of regular beer and 1,500 liters of bread and flour, shared (ha-la-a) between soldiers (aga₃-us₂) = 750 men?
• TUT 120, Girsu: 44,500 liters of beer on each day of a full month (for this text, and this is exceptional, the duration of consumption is therefore known), shared by soldiers (aga₃-us₂-e ha-la-a) = 750 men.
• STA 3 iii 7-8, Umma: 37,710 liters of beer delivered to the soldiers in the frame of the monthly bala duties = 628 men?
• Dossier of the Drehem e₂-muhaldim (above §7.5): one contingent of several hundred soldiers acting as royal guard.
• Three typical Girsu texts (among others) record a še-ba delivery to the aga₃-us₂ for the 11th month of Š 46 (MVN 12, 118 and CM 26, 64) and the 11^th month of Š 47 (MVN 12, 249). If we calculate 2 liters per day and per soldier during one month, we get: 1,950 soldiers in the first text (with 117,035 liters of barley), 2,023 soldiers in the second text (121,390 liters of barley) and 6,095 soldiers in the third text (365,710 liters of barley). But is this way of calculating reliable? And who are these Girsu aga₃-us₂?
• Text [Q], Drehem: 1,200 bows distributed to arm 1,200 soldiers.
• Text [R], Drehem: 570 pairs of boots to put shoes on 570 soldiers.
• Text [W], Girsu: 1,300 men in the garrison of Garšana.
• TCTI 2, 3543 (remainder of the text [W]): breakdown of units between 360 and 820 men.

§8.3. In most cases and if we put aside the (exceptional) texts [C] and [O], we find that the sources indicate a range of unit size from 300 to 2,000 soldiers, with an average of around 600.^[108] It may be objected that nothing is said here about the army composed of erin₂-conscripts, when these were mustered en masse for military operations: adding these conscripts to the professional soldiers, the numbers would probably be much larger. But the problem is that they are hardly ever seen in our archives (see nevertheless below §8.6)!

§8.4. One more interesting account related to this issue is provided by an extract of the “historical” letter (its status is therefore literary) from Lipit-Eštar to Nanna-kiag, referring to the constitution and sending of a reinforcement army composed of three equal units of spearmen, bowmen, and soldiers armed with axes; the total reaches 6,000 men:

[X] Letter from Lipit-Eštar to Nanna-kiag (ETCSL 3.2.4)
(8) a₂-še₃ 2 li-mu-um erin₂ lu₂ ^geššukur (9) 2 li-mu-um erin₂ lu₂ ^gešban (10) 2 li-mu-um erin₂ lu₂ dur₁₀-tab-ba im-mu-e-ši-sar

“Now, I have sent to you in haste 2,000 spearmen troops, 2,000 bowmen troops, 2,000 double-axe wielding troops.”

The most interesting information here is the three-party distribution of the whole group (see above §6.2-3). For the rest, the total number of 3 × 2,000 = 6,000 is unfortunately not reliable, as demonstrated by the alternative numbers given by some copies of this text: for each unit, they alternate numbers between 2,000, 3,000 or 4,000 men.

§8.5. Regarding the reliability of numbers, the situation is probably similar for the one text that gives the largest number of soldiers explicitly attested in a military context, for all the documents we have seen for this period: the 10,000 aga₃-us₂ (2 × 5,000) commanded by Apillaša (text [F] above §3.2). Again, it must be kept in mind that this document from the royal correspondence of Ur is not an administrative tablet but a literary text. And even if an original letter was actually written by Aradmu, we cannot know how much reality resides in this breakdown of 10,000 regular soldiers: we have to consider the possibility that the author of the letter sought to impress his interlocutor, or that the scribes who “canonized” or “re-created” this letter may have rounded off the total number of soldiers, either to exaggerate or to minimize it.

§8.6. Finally, armies of several tens of thousands, as seen in documentation from the time of the Mari kings or from that of the neo-Assyrian empire, cannot in any case be seen explicitly in our Ur III texts. Therefore, we do not know ultimately with what means the kings of Ur were engaged in all these faraway fights that they undertook, as mentioned at the beginning of this work. Nevertheless, P. Michalowski recently pointed out a great number of erin₂ troops stationed in Susiana, which can perhaps be glimpsed from administrative tablets dating to Šulgi’s final years. At that time, several tens of thousands troops (erin₂) seem to have been stationed in such cities as Susa or AdamDUN. These cities could have thus been staging areas, virtually dominated by military personnel ready to go to war. And, according to Michalowski, “massive armies” were gathered there during the wars of Šulgi’s final years, this influx of soldiers having perhaps doubled the population of Susiana at that time: “some of these troops come from Sumer, some were local, and some came from other vassals, allies and provinces.” Ultimately and according to him, “the impact of a large military presence in the border areas should not be underestimated.”^[109]

§8.7. Nothing is available for the army on campaign, nor on siege warfare (unlike the situation for the time of the Mari kings for example), or on military strategy. These are still questions for which the silence of our sources, because of their nature, is unfortunately almost absolute.^[110]

§9. Conclusion

In reinforcing the institution of the aga₃-us₂ soldiers, already observable since the Pre-Sargonic period, the kings of Ur wished to have at their disposal, first, a force for administering and securing the territory of the kingdom, and second, a military intervention force ready for foreign conquest. Closest to the king, a specific contingent of royal aga₃-us₂ soldiers was used as his own household troops and elite infantry unit, as a kind of Pretorian Guard. For this Ur III period, the texts from the Girsu archives generally provide the greatest amount of evidence regarding the army and military affairs. If we are not misled by the random distribution of our sources and in particular by the fact that we do not have any central archive, it appears that this province of Girsu played a military role in the kingdom greater than that of its neighbors. And we must note in conclusion that the elements we have tried to gather and to describe briefly approximate a framework inherited from the Old-Akkadian period, one that would continue to be used in Mesopotamia until the end of the Old-Babylonian period, at least with regard to the constitution and hierarchical organization of the royal army.^[111]

Version: 21 October 2009