BMCR 1990.01.12

The Limits of Empire: The Roman Army in the East

, The Limits of Empire: The Roman Army in the East. Oxford: Oxford University Press, 1990. xiv, 492 pages. ISBN 0198148917. $75.00 (hb.).

Was there a coherent “grand strategy” for the defense of the Roman empire, a strategy dictated from Rome? Did this strategy aim at defending and enhancing the security of the empire, and did it evolve through a series of coherent responses by emperors who took thoughtful account of the wide range of internal and external pressures upon their military establishment in the centuries between Augustus and Constantine? These questions were answered affirmatively by Edward Luttwack in his deservedly influential book, The Grand Strategy of the Roman Empire from the First Century A.D. to the Third (Baltimore, 1976). This view has not, however, commanded universal assent among Roman historians. One reason has been that Luttwack is not a classicist by training—he is an authority (among the most important in the United States) on defense systems. Another is that such positive answers are not compatible with the style of government described by Fergus Millar in The Emperor and the Roman World (31 BC-AD 337) (London, 1977). Millar’s picture is of a government that rarely initiated policy, but rather made policy in response to appeals (and complaints) from Rome’s subjects. Furthermore, when critics objected that Millar had not taken adequate account of the military interests of the emperors in drawing his picture, he responded with an eloquent defense of his overall view by stressing the practical difficulties of central control over the frontiers given the primitive conditions of ancient communication. 1 Ben Isaac’s The Limits of Empire offers a new and formidable challenge to Luttwack, and is at the same time an excellent study of the implications of Millar’s view for our overall understanding of Roman military policy.

Isaac bases his case upon the evidence for the Roman army’s activities in Palestine, and he moves on from this material (of which he has a superb command) to generalize about the activity on the eastern frontier, and, thus, by implication, the empire as a whole from the first century A.D. to (at times) the sixth. It is an impressive accomplishment, and throughout 426 large, heavily annotated pages Isaac never loses sight of his argument. The length of the book, and the density of its argument, may well prove to be daunting to the nonspecialist—but these readers will still want to read chapters 1 and 9, where the main points of Isaac’s case are laid out. This argument is that there was no “grand strategy” of empire. Questions of war and peace were decided by the emperor, most often to enhance his own glory and to satisfy his soldiers, who would profit from foreign adventures. This was the most important factor because, in Isaac’s view, “there was no powerful officer class in Rome, no central army command” (p. 383). Furthermore, he maintains that, “it is unlikely that most Roman frontier lines were determined by choice and by a conscious decision to halt indefinitely all further advance” (pp. 387-88). In his view, the Roman limites were not thought of as lines to cut off movements by outsiders, but rather to facilitate communication among Roman forces. When the Romans thought about expansion, they did not do so with the intention to acquire territory, but rather to control peoples (an important point indeed, pp. 394-95), and they really knew very little about lands beyond their borders. The grand strategy of the Roman army, insofar as it existed at all, was simply to control internal disorder and to be ready to conquer other peoples.

Isaac’s propositions are challenging, and I think that there is much merit to what he says. But I do not think that the evidence supports him at every point, and it is dangerous to base such an important a set of propositions on the evidence from Judaea as heavily as he does. Judaea was a very unusual province insofar as the local ruling class was incapable of exercising enough control over its people to ensure internal security. 2 After the suppression of the great revolt in the 60s, this task became the primary mission of the large number of Roman troops who were stationed in the province. Troops were used to chase bandits in other provinces (there is even evidence for an auxiliary cohort on Cyprus that cannot be connected with the Jewish revolt of 116/7), 3 but not on the scale that they were used in Judaea. Above and beyond this, I think that there was a good deal more central direction to Roman policy than Isaac does, and that there were some major changes of direction in Roman frontier policy that were thought out in a coherent way at Rome. When Tacitus noted that Augustus had written in his will that the empire should be contained within fixed termini, 4 he was noting an important change of policy, and, as I will argue below, it was not unparalleled.

Roman frontier policy plainly differed from place to place as the Romans responded to local conditions, but this does not mean that they had no coherent idea about what they were doing. There is a parallel here with the way that the empire handled its taxes. An examination of the tax structure of the empire, at first glance, suggests utter chaos. The sorts of taxes collected, the sorts of collectors employed, and the rates of collection plainly differed widely from place to place as the Romans adopted local schemes to their own ends. But this does not mean that they did not have a clear idea of just how much they expected to get from each province in a given year. When he died, Augustus left a full accounting of the annual revenues and fiscal obligations of the state, written out in his own hand, and the elaborate system of provincial censuses was designed to update this information periodically. 5 When new revenues came in that insured the satisfying of existing obligations, an emperor such as Tiberius could reduce unpopular taxes because he knew what his total revenues would be. Similarly, members of the senate were well enough informed to realize that when Nero proposed the abolition of vectigalia, it would ruin the state. 6 This concern with an accurate accounting is perhaps best illustrated in the elaborate calculations, evident in Ulpian’s life table, of the return from the inheritance tax in certain circumstances. 7 That this system did not always work out, either through political pressures or imperial insanity, does not mean that most emperors did not recognize the need for a coherent tax policy tied to their expenditures, or that they lacked the ability to implement it. It is fair to say that, despite its ultimate failure in the third century, the Roman system served the empire better than did the systems of taxation in many early modern European states. The same can be said for the emperors’ handling of their foreign affairs. A veneer of inconsistency does not mean that the emperors (or most of them) did not deploy the army in terms of some more coherent policies than Isaac suggests.

The most important test case for Isaac is the reign of Hadrian. Did this period see a clear shift to a different type of behavior towards the outside world than was evident in the reign of Trajan? If Trajan’s Parthian war is viewed as a failure, as it is by Isaac (p. 30), then the answer can be no. Hadrian’s withdrawal was forced upon him, and the quiescent policy that he favored did not imply a new direction, but that Rome was licking its wounds until another opportunity for aggression offered itself, or, as Isaac puts it, “it follows that Hadrian gave up Mesopotamia because he could not afford to reconquer it” (p. 25). A recently published inscription from Iraq should make us rethink this, for it shows that Mesene remained independent of Ctesiphon between 117 and 151. 8 Furthermore, the complex coinage of Parthia in these years shows that there were no fewer than three pretenders to the throne in the second quarter of the second century, and the admittedly scant evidence from the Historia Augusta (drawing here upon a good source) suggests that Rome was able to dominate Parthia until the aggressive king Vologeses IV took the throne in 147/8. I think that it is fair to say that Hadrian did what he did not because he had been beaten, but rather because he thought that his policies conformed better with the resources of the state.

Looking beyond Hadrian, the other evidence from the second and third centuries does not suggest that, “Rome had long-standing ambitions to acquire parts of the Persian empire and frequently made attempts to realize them” (p. 52). While it is clear that this was Trajan’s initial aim, it is also clear that the great war that broke out in the 160s was started by Vologeses IV, and that it began with invasions of Armenia and Syria. Severus did launch two invasions of Parthia in the 190s, primarily to enhance his prestige after his seizure of the throne, and the subsequent civil wars—a point of particular importance to him as he seems to have been unpopular at Rome. Caracalla appears to have identified rather too strongly with Alexander the Great, but he died before anything came of his designs in the east. Matters changed greatly thereafter. Between 225 and 260 the new Sassanid dynasty in Persia was ever the aggressor. Alexander Severus’ invasion came as a response to Sassanid attacks on the eastern provinces, attacks that may have been motivated by Ardashir’s desire to root out survivors of the Arsacid dynasty who had taken refuge in the border regions. 9 In 238 the Sassanids came again; this invasion is the reason for Gordian III’s campaign in 242-44. In 251 Sapor I appears to have exploited a civil war in Armenia to seize power there, and in 252 he followed this up with a massive invasion of the eastern provinces, which resulted in the defeat of the eastern army at Barbalissos and the capture of Antioch. In 260, after several years of campaigning, that saw the Persian capture of Dura (257?) and Nisibis (259?), Sapor fell upon the weakened army of Valerian outside of Edessa, captured the emperor, and proceeded to ravage Cilicia and Cappadocia before returning to his own lands. 10 The picture of these years is one of very serious Persian aggression indeed, and even if the Sassanids had no intention of annexing Roman territory west of the Euphrates, these attacks certainly serve to explain why Aurelian planned (in 275) and Carus carried out (in 283) an invasion of Mesopotamia. Revenge is a very well attested motivation for Roman military action.

In looking at the broad picture of Roman relations with Persia, Isaac denies the conventional view that there was a real change in Persian attitudes after the accession of Ardashir in 225, and he argues that this only shows that the Persians could not tolerate a Roman presence east of the Euphrates (p. 32). The evidence of early Sassanid activity suggests to me that they were actively trying to establish a presence in northern Mesopotamia. They occupied Carrhae and Nisibis in 238, and Nisibis again in 260; they destroyed Hatra in 240, and they occupied Nisibis after Julian’s failure in 363. This was a major departure from Arsacid policy, which appears to have been based upon alliances with states on their western frontier, and which was impossible for the Sassanids to follow for internal political reasons (these places also had dynastic links with the Arsacids that Ardashir could not tolerate). Thus, Diocletian’s acquisition of the five trans-Tigritine provinces, through the treaty of 298/9, can very easily be seen not as the realization of an old Roman dream (it seems that no significant Roman presence was established in them), but rather as an effort to strengthen defenses for northern Mesopotamia, which had been the main route for Persian offensives in the third century. The powerful fortress cities of northern Mesopotamia were thereafter serve more often as defensive positions than as staging grounds for invasions of Parthia.

With the exception of Septimius Severus’ campaigns, the evidence speaks, I think, against Isaac’s basic interpretation. Furthermore, Isaac rules out the evidence of Antonine authors who speak of the empire as a fortress (pp. 26-27), arguing that the views of Appian, Aristides, and Pausanias are contradicted by Herodian’s positive view of expansionist policies and thus cannot have been important. 11 Isaac is clearly correct in seeing that there was a serious debate about the merits of expansion, and, as others have also argued, the debate was particularly relevant in the wake of Hadrian’s eastern settlement. But I differ from Isaac in that I think the panegyric statements of Appian or Aristides (and pseudo-Aristides) are a more significant reflection of opinion than occasional praise of expansion by Herodian (who also uses the image of the empire as a fortress when it suits him) or Fronto’s critique of Hadrian (whom he loathed) in his own panegyric on Lucius Verus during his Parthian war (one does not praise the virtues of a policy that was looking none too successful under such circumstances). What is more, the image of the empire as a fortress clearly gained wide currency in the second and third centuries outside of the exalted circles that Isaac discusses, so much so that the borders of the empire appear as “the walls of Rome” in the Sibylline Oracles (13.105; 14.165, 247), which provide many valuable reflections of imperial ideology. When it comes to Cassius Dio, I think that Dio’s views were more complex than Isaac does. Dio clearly approved of expansion in past Roman history, and, as a way of criticizing Commodus, he approved of Marcus’ alleged plans to transform the lands of the Quadi and Marcomanni into a province. But he disapproved of Severus’ annexation in northern Mesopotamia, and it is fairer to say that his attitude was generally ambivalent: he was no believer in the idea that in his own time Rome should attempt to conquer the world.

To what extent did the emperor dictate day-to-day policy on the frontiers? The letters of Pliny show us that the emperor could be called upon to make decisions or give advice upon a wide range of civil cases, as well as to decide disputes concerning the command of even a very few soldiers. 12 It is difficult to believe that he would have been less involved in provinces where there were large garrisons, and Tacitus does indeed reveal a pattern of imperial interference with such governors.2 He also shows a governor faced with serious problems as he tries to anticipate an emperor’s response. 13 This is not to say that governors could or did not act on their own when necessary, and that they did not drag their feet when they thought that the orders they received were foolish—but the parameters of their activity did tend to be laid out at Rome, and their actions could be reversed, as happened in the case of Agricola. An emperor’s duty to pay attention to the actions of his governors was perhaps brought out most clearly by Vespasian when he criticized Nero for failing to give Tiberius Plautius Silvanus Aelianus the honors that he deserved for his successes as governor of Moesia. 14 If nothing else, Augustus’ actions with respect to Crassus and Primus in 30 and 23 B.C.(?) show that he was at pains to point out that, under his regime, decisions with respect to the army were to be made by him.

There is reason, good reason I think, to doubt Isaac’s overall interpretation. But the great strength of this book will remain Isaac’s treatment of the evidence on the ground in Palestine. The best part of this analysis, aside from a few superb pages on milestones as propaganda (pp. 304-9), is Isaac’s study of Roman military installations in Judaea—and it is in this part of the book that Isaac lays the foundation for his case against Luttwack (p. 188). He makes a very convincing argument that the post-Diocletianic fortifications that can be securely dated do not add up to a systematic “defense in depth” against desert raiders, as Luttwack (and others) had once believed. He shows that they tend to be badly sited for defense and that their most obvious function is to serve as observation posts on supply lines for the maintenance of internal security. As Isaac points out, this is also the best explanation for earlier Roman posts in the region for and the development of the Roman road system (p. 113). I do not think that his argument in these cases can be faulted, and I think that the comparisons that he draws with the fortifications in the Egyptian desert between Coptos and Leukos limen and similar installations in Jordan are to the point. So too is his observation that many of the buildings that have been identified by archaeologists as military, on the basis of their shape, served other purposes (e.g. hospices for travelers along the road). His point that “it is significant that buildings may look like heavily defended forts while, at the same time they are sited without any regards for defensibility” (p. 204) raises some very real methodological questions along the lines of those once raised by Dilleman, in his critique of the use of aerial photography for establishing Roman frontier lines in Mesopotamia, 15 or those that J.P. Sodini’s excavations at Déhès have raised about the validity of G. Tchlenko’s methods in his survey of central Syria. 16 Before any firm conclusions can be drawn about the function of a building, it needs to be carefully excavated. What once seemed to be a line of forts may very well turn out to be a chain of hotels.

But if the Arabian limites are not a defense in depth, does this mean that there was no coherent defense policy? I do not think that this conclusion follows. As our knowledge of pre-Islamic Arabia increases, it is becoming clear that Rome relied much upon native confederations to keep raiders away from the more settled areas of its domains in the east, and that it had done so ever since the occupation of the Nabataean kingdom. 17 Despite the ultimate failure of this policy in the seventh century, it seems to me that it worked well during the intervening 500 years (not a bad record). 18 When this system is examined in a broader context, it appears to be in line with Roman practices in the west: the Romans adjusted their policies to match the people that they were dealing with—just as they did when they assessed their taxes. The variation in Rome’s response to various outside peoples suggests a more sophisticated approach to frontier defense than Isaac suggests, and, as I have said above, I think that the literature of the second century suggests that defense was what most often what the emperors had in mind.

There was a limit to what any emperor could hope to accomplish, and to the amount of direction that he could give commanders in the field. It is also true that the amount of interest that individual emperors took in the administration of their domain varied enormously. Most also appear to have taken the view that there was no need to change a system that was working perfectly well—but others did choose to initiate important changes in conduct and attitude. Each emperor also had his own priorities, be it the marital status of the divine meteorite Elagabal from Emesa or the defense of the Rhine frontier. The study of Roman military policy has to allow for a wide range of variables both at Rome and at the fringes of the empire. In the areas that concerned them most, individual emperors could and did intervene to give important direction that could have long lasting consequences. But they were also motivated by respect for precedent. It is this point that makes it possible to speak of “grand strategies.” The actions of Hadrian could, and I think did, give a direction that his immediate successors followed for several decades, as did the actions of rulers such as Severus, Diocletian, Constantine, or Augustus. The impact of their decisions could be immense in some parts of their domain and slight in others: but that the impact of a decision could be limited does not mean that the overall posture of government was passive or sluggish.

Isaac has raised very real questions about how that the empire operated, by working outward from the evidence of one province. I do not agree that this is the way to arrive at a model for the way that the empire worked. This is less important than that Isaac’s book is an immensely impressive accomplishment that tackles issues needing to be confronted, and that it does so very well. It is a book that students of Roman history will be reading with profit for a long time to come.


1. “Emperors, Frontiers and Foreign Relations,” Britannia 13 (1982), 1-23.

2. M. Goodman, The Ruling Class of Judaea: The Origins of the Jewish Revolt against Rome, A.D. 66-70 (Cambridge, 1987). Even though his arguments concerning the outbreak of the revolt are open to question, his analysis of the inadequacy of the ruling class is to the point.

3. CIL 3.215; ILS 1398 for Cyprus; for the police elsewhere see the useful summary by K. Hopwood, “Bandits, Elites and Rural Order,” in A. Wallace-Hadrill (ed.), Patronage in Ancient Society (London, 1989), 171-87.

4. Tac. Ann. 1.11.4: quae cuncta sua manu perscripserat Augustus addideratque consilium coercendi intra terminos imperii.

5. The basic surveys are L. Nessen, Untersuchungen zu den direkten Staatsabgaben der römischen Kaiserzeit (27 v. Chr.- 284 n. Chr.) (Bonn, 1980) and P.A. Brunt’s updated review of this book in P.A. Brunt, Roman Imperial Themes (Oxford, 1990), 324-46, 531-40 along with Brunt’s new study of the publicani under the principate in the same volume, 354-432. The recent publication of the Neronian tax law for Asia in EA 14 (1989) also adds much to our knowledge. For a thorough discussion of Augustus’ breviarium totius mundi see now Cl. Nicolet, Space, Geography, and Politics in the Roman Empire (Ann Arbor, 1990), esp. 180, pointing out that Tacitus’ quae perscripserat sua manu should mean that Augustus wrote out his numbers in words so that there would be no possibility of errors, and thus that Augustus was not giving rough estimates.

6. Tac. Ann. 2.42.4: regnum [Cappadocia] in provinciam redactum est, fructibusque eius levari posse centesimae vectigal professus Caesar ducentesimam in posterum statuit. Tac. Ann. 13.50: sed impetum eius, multum prius laudata magnitudine animi, attinuere senatores dissolutionem imperii docendo, si fructus, quibus res publica sustineretur, deminuentur. The reading senatores is in M, though not printed in the majority of modern editions; for discussion of this problem see M. Griffin, Nero. The End of a Dynasty (London, 1984), 92. Such knowledge on the part of members of the senate had been expected in the Republic: at de legibus 3.41, Cicero wrote: est senatori necessarium nosse rem publicam (idque late patet: quid habeat militum, quid valet aerario, quos socios res publica habeat, quos amicos, quos stipendarios, qua quisque sit lege, condicione, foedere). See also Nicolet, Space, Geography and Politics, 181-3.

7. B.W. Frier, “Ulpian’s Life-Table,” HSCP (1982), 216-23.

8. F.A. Pennachietti, “L’Iscrizione bilingue greco-parthica dell’ Eracle di Seleucia,” Mesopotamia 22 (1987), 169-85.

9. D.S. Potter, “Alexander Severus and Ardashir,” Mesopotamia 22 (1987), 147-57, reprinted with corrections in my Prophecy and History in the Crisis of the Roman Empire. A Historical Commentary on the Thirteenth Sibylline Oracle (Oxford, 1990), 370-80.

10. For the chronology and events of these campaigns see now Prophecy and History, 189-96, 290-7, 329-37.

11. App. Praef. 28: tên te archên en kyklô perikathêntai megalois stratopedois kai phylassousi tên tosênde gên kai thalassan ôsper chôrion; Arist. 26, 29, 82-84; [id.] 35, 36. Herod. II, 11, 5: phrouria de kai stratopeda tês archês proubaleto, misthophorous epi rhêtois sitêresiois stratiôtas katastêsamenos anti teichoustês Rhômaiôn archês [the person involved here is Severus], Paus. 1.9.3; 8. 43. Note also the interesting description of Hadrian’s activity (possibly from Marius Maximus), which implies that his frontier policy was seen as drawing a line between the Roman and barbarian worlds in HA V. Hadr. 12.6: per ea tempora et alias frequenter in plurimis locis, in quibus non barbari non fluminibus sed limitibus dividuntur, stipitibus magnis in modum muralis saepis funditus iactis et conexis barbaros seperavit. See also J. Palm, Rom, Römertum und Imperium in der griechischen Literatur der Kaiserzeit (Lund, 1959), 56-62, 76, 83

12. For the assignment of soldiers see esp. Plin. Epp. 10.21-2; 27-8, and 10.29-30 on recruits.

13. Thus, for example, Tac. Ann. 11.19-20, 13.9; 14. 38-9. 3 Tac. Ann. 12.48.3: ne tamen adnuisse facinori viderentur et diversa Caesar iuberet, missi ad Pharasmanen nuntii, ut abscederet a finibus Armeniis filiumque abstraheret.

14. ILS 986: Moesiae ita praefuit, ut non debuerit in me differri honor triumphalium eius ornamentorum. Note also the criticism of Commodus in the Historia Augusta (probably derived from Marius Maximus), HA V. Comm. 13.7: ipse Commodus in subscribendo tardus et neglegens, ita ut libellis una forma multis subscriberet, in epistolis autem plurimis vale ‘tantum’ scriberet.

15. L. Dilleman, Haute Mésopotamie orientale et pays adjacents: Contribution à la géographie historique du ve s. avant l’ère chrétienne au vie s. de cett ère (Paris, 1962).

16. J.P. Sodini et al., “Déhès (Syrie du Nord). Recherches sur l’habitat rural,” Syria 57 (1980), 1-301.

17. G.W. Bowersock, Roman Arabia (Cambridge, 1983), 97. Isaac discusses this evidence on p. 68-77 but does not draw the same conclusion that I do.

18. Note also F. Donner, The Early Islamic Conquests (Princeton, 1981), 100 on the importance of the Sassanian conquests for disrupting the Byzantine system of alliances just before the rise of Islam. It is also clear that in the early years of Islam these confederations were able to resist the armies of the Prophet, and inflicted a serious defeat on them at Mu’ta in 629 (see Donner, p. 105, for details).