From the accession of Diocletian to the death of Theodosius I (284–395), the Roman state was ruled by warrior emperors, who strenuously campaigned against internal rivals and defended its frontiers against external enemies. Then, for more than two centuries until the accession of Heraclius in 610, emperors and their courts became sedentary. Emperors were based at Constantinople in the East and Ravenna (or for short periods at Rome) in the West, and evolved a style of ruling that was radically different from what had gone before. Thus the basic framework of political and military activity in the earlier part of late antiquity up to 395 continued the pattern of the third century. In political terms the main features of the fourth-century empire resembled those of the principate that had been founded by Augustus. After 395 the eastern emperor and his court were permanently resident at Constantinople. As emperors ceased to take part in campaigns themselves and became palace-based, the empire's ability to defend the northern frontier and retain the western provinces was critically reduced.
The beginning of the Byzantine Empire is sometimes placed around 395. That label cannot of course apply to the western provinces. From the late fourth century the borders of the western empire became porous, and north European tribal groups, speaking early forms of the German language, infiltrated the regions adjacent to the former frontiers. Eventually, by the end of the fifth century, all the former western provinces of the empire had been transformed into kingdoms of a hybrid nature. They retained Roman legal systems and other institutions, but were ruled by tribal Germanic kings. Here, therefore, was a political transformation much more radical than anything experienced in the East. Out of these conditions emerged the earliest structures of medieval Europe.
The history of the Roman state in late antiquity was forged in an environment of warfare, battle, and military activity. It has been estimated that more than two-thirds of the annual state budget, the income derived from taxation, tribute, and the lease of imperial lands, was expended on soldiers' wages and benefits, on supplying and equipping the armies, and maintaining the military infrastructure.9 This modern assessment is broadly confirmed by the remark of a sixth-century military manual that military wages took up most of annual imperial revenues.10 The Roman army was overwhelmingly the most important component of the organized Roman state.
Throughout the fourth century the aims and motives of Roman wars were essentially threefold. Firstly, they were designed to maintain the integrity of the empire within established frontiers against barbarian enemies. Secondly, they were fought on Roman territory by rivals for power; bloody civil wars are a leitmotif of late Roman history. Thirdly, and no less importantly, warfare provided the armies, especially their elite squadrons, with their chief occupation, fighting, and was the source of the spoils and booty that motivated the soldiers. Each of these themes requires elaboration, and is illustrated from the detailed record of Ammianus. His extensive, colorful, and often gripping description of the events of 353–78 presents the many faces of Roman warfare. A recent study suggests that his narrative omitted no significant engagements from this period.11 There were skirmishes and ambushes on either side of the northern river frontiers with Germanic barbarian groups. There was internecine warfare among these groups themselves (cf. Pan. Lat. XI , 16.1–2). Punitive Roman expeditions were sent beyond the Rhine and the Danube to assert Roman authority over their often elusive foes. There were pitched battles with the barbarians, successful against the Alemanni at Strasburg in 357, disastrous against the Goths at Adrianople in 378. Warfare in the east included large-scale sieges of the contested cities of Mesopotamia (Amida, Nisibis, Bezabde), as well as the cumbersome and draining logistic preparations for major invasions of Sassanian territory. No less important were the bloody campaigns fought on Roman soil, usually in the central Danubian region, between rival claimants to power.
Evaluation of the institutions of the Roman army and their economic implications belongs in a later chapter, but it is important to give military history pride of place during the first century of late imperial history, from the accession of Diocletian in 284 until that of Theodosius in 379. One of the critical, but little emphasized, aspects of this history is that throughout this period the rulers were expected to lead all major campaigns in person, and almost always did so. The youthful Constantine, serving as a cavalry officer, personally seized a Sarmatian warrior by his hair and threw him in triumph at the feet of his commander Galerius, and led his troops through the marshes against the enemy.12 In 302 Constantius I had to be hauled up by a rope when the gates of the Gallic town of Langres were hurriedly shut to keep out an Alemannic raid, while the Caesar was outside. His army managed to rescue him five hours later and 60,000 Alemanni are said to have paid with their lives (Eutropius 9.23; Orosius 7.25). In 355 it was only after a long discussion that Constantius II relinquished personal command of a minor sortie against the Alemannic Lentienses along the south side of Lake Constance to his general, the magister militum Arbitio (Ammianus 15.4). Julian was killed when he recklessly plunged into a minor rearguard engagement without putting on his body armor (Ammianus 25.3). The next emperor to be lost in battle was Valens, who simply vanished in the pandemonium of the battle of Adrianople against the Goths in 378 (Ammianus 31.13). Adrianople was the first time that a Roman army suffered a major defeat at the hands of barbarians, and almost the last occasion, it seems, that any Roman emperor led an army into battle in person for more than two hundred years. The defeat changed the role and character of Rome's monarchy.
The burdens of warfare were without question the main reason why Diocletian created an imperial college to aid him in his ruler's task. Maximianus had been made Augustus in 286, and the second surviving panegyric of the period, delivered in 291, proclaimed the concord and unity which existed between the two emperors. They fulfilled the twin roles of Jupiter and Hercules in ruling a subdued and pacified world (Pan. Lat. XI ).
What age ever saw such harmony in the highest power? Which full or twin brothers enjoy an undivided inheritance as equally as you share the Roman world?…The Rhine and the Danube and the Nile and the Tigris with its twin the Euphrates and the two Oceans where the sun sets and rises again, and whatever lands and rivers and shores are between them, are as common to you, with ready goodwill, as the daylight in which the eyes take pleasure is common to them. (Pan. Lat. XI  6.3, trans. Rees).
In 293 two Caesars were added to the team, Constantius, who had been a senior military commander under Maximianus in the west and was already married to his daughter Theodora,13 and Galerius, who married Diocletian's daughter Valeria (Lactantius, DMP 50. 2; Diagram 3.1).
Diagram 3.1 The first tetrarchy
Much modern speculation has been expended on the political origins and the motivation behind this new coalition, the first tetrarchy. Its main objective was to achieve a manageable division of imperial powers, and it led to an unofficial territorial division of the empire:
Since the burden of the wars, which we have already recorded, was pressing more fiercely, the empire was divided into four parts, and all the territory lying beyond the Gallic Alps was entrusted to Constantius, Africa and Italy was given to Herculius [Maximianus], the shores of Illyricum as far as the Pontic strait to Galerius, while Valerius [Diocletian] retained the remainder. (Aur. Victor, Caes. 39.30)