11

THE LAST OF THE SOLDIER EMPERORS

After the destruction of Palmyra, 273 was quiet, and Aurelian used the rest of the year to make clear his intention of dealing with Tetricus in Gaul now that the east seemed pacified. In 274, he marched an army into Gaul, where Tetricus decided to surrender. He did so after his army had already drawn up for battle in modern Champagne, in the Catalaunian fields that are now more famous for a different battle that took place there almost two centuries later, in which Attila the Hun suffered a crushing defeat. This third-century battle of the Catalaunian fields was equally devastating, but for the Gallic legions, who were massacred by Aurelian’s troops. Given their long-standing habit of rebellion, this may have been a sensible move. With Gaul under control, Aurelian returned to Rome and celebrated a lavish triumph, which both Tetricus and Zenobia graced. Unlike the victi of the republican period, they were not executed at triumph’s end. Quite the reverse. Zenobia married into the Roman aristocracy, and was proudly claimed as an ancestor by a fourth-century scion of her house. Tetricus, for his part, was honoured with a minor governorship in the south of Italy that kept him safely out of the way.

Aurelian’s success really did look absolute and it had been decades since a single emperor had ruled unopposed. Aurelian had accomplished that with relative ease and comparatively limited bloodshed. More importantly, no new military challengers had risen up in the wake of the suppression of the regimes in Gaul and Palmyra. After his triumph, in a city that was being enclosed in a wall of great magnificence, another symbolic coup de théâtre would follow. The revolt of the imperial mint workers in Rome had been one of the first challenges of Aurelian’s reign and, though there is no reason to connect that fact to his next initiative, it is interesting that the men who produced Rome’s currency should have had so prominent a place for a second time in a relatively short reign: having welded the whole empire back together, Aurelian in 274 launched a new currency, replacing the old-fashioned gold aureus that had been in use for centuries with a new one. While the old aureus had circulated with varying weights and purities, it had always been consistently tariffed against the silver denarius at 1:25 and had maintained that ratio throughout fluctuations in weight and purity which might also cause some aurei, in some places, to be traded as bullion rather than coins. Aurelian’s new aureus was by contrast of extreme fineness and clearly intended to drive other, less fine, coins from the market. It was also no longer formally tariffed against silver at all, but was, in effect, stamped gold bullion. In and of itself, that might not have had any dramatic effect, but it was not the limit of Aurelian’s monetary activity. In place of the supposedly silver antoninianus, which at this point had virtually no silver content left in it, he minted a new silver coin at a standard 5 per cent fineness. The new Aurelianic antoniniani were marked with a XXI or XX, which presumably means they were tariffed at 1:21 or 1:20 against the copper token coinage. The emperor’s intent must have been to replace all the coins of his predecessors and their rivals then in circulation with a new coinage of his own, advertising to the whole world his achievement in restoring the unity of the empire.

Sadly, the impact of his move was disastrous, and came close to demonetising a Roman economy that was already suffering from the effects of epidemic disease. The debasement of the silver currency – from about 40 per cent fineness to less than 5 per cent in 270 – that had characterised the whole of the third century had not (amazingly, according to modern economic theory) had much of an impact on the economic stability of the empire. That is to say, there is no evidence of widespread inflation in response to the falling intrinsic value of the silver coinage year in and year out since the introduction of the antoninianus as a (theoretically) double sestertius under Caracalla. The imperial image on the coins, as well as their familiar tariffing, meant that they survived as a fiduciary currency – and as such were hoarded in troubled times, especially in the 250s and 260s, during which a hugely disproportionate number of known coin hoards were deposited – even when their intrinsic value had effectively disappeared. Aurelian’s reform – and it is clear that he did mean it as a reform to unify the minting of a newly reunited empire – changed all that, and very painfully. Where debasement had caused no reliably measurable inflation, Aurelian’s decoupling of gold from the fiduciary system of early imperial silver and base metal currencies did. Prices began to soar. The one place we can measure this is Egypt, where the papyri are able to track the almost instantaneous damage done by the decoupling of currencies. As Aurelian’s new coinage was exported from Rome and entered into circulation in the various provinces, it seems certain that the same process repeated itself, though that is harder to document.

Aurelian’s new coinage did drive many of the older currencies of various regions out of circulation, but it could not do so completely, if only because he could not supply demand fast enough. Worse, the floating of the relationship of gold and silver made nonsense of the link between silver and the intrinsically worthless copper currency. Without buyers and sellers sharing the faith that a handful of slugs of base metal bearing the imperial image and name were, somewhere in the universe, worth a set amount of gold, the fiduciary system collapsed. The sophisticated banking system of the early empire went with it. Coins were now worth what local markets thought they were worth, and that was always a great deal less than the tariffed price. It would be two decades before an effort was made to control the chaos that the Aurelianic reforms had unleashed, during which time payment in kind became a driving engine of the Roman economy for the first time in almost half a millennium – so much so that the next great reforming emperor, Diocletian (r. 284–305), was forced to build payment in kind into his administrative system. It is ironic that the empire began to face widespread economic problems only as the political situation seemed to have changed definitively for the better. But, as things turned out, Aurelian lived to see the impact of neither his military successes nor his economic reforms.

In 275, he ventured into Gaul, presumably to the Rhineland and the region of Colonia Agrippina, to reassert his presence in a region that had not seen an Italian emperor in decades. He then travelled with his army to the Balkans via Raetia, where he dealt with a minor Iuthungian raid. We do not know what Aurelian was doing in the Balkans, where there is no evidence of ‘Scythian’ or Gothic trouble; he may have been heading to the eastern provinces for a kind of ‘beating of the bounds’ along the lines of the Gallic trip. Regardless, in late summer or early autumn 275, at the imperial postal station at Caenophrurium, on the road between Perinthus and Byzantium, he fell victim to a putsch similar to the one that had unthroned Gallienus. It seems certain that the army as a whole was genuinely happy with their emperor’s leadership, and that the murder was an unpleasant accident: a court official, for reasons of his own, forged a letter suggesting that the emperor had ordered the execution of some junior officers, who proceeded to murder him – a Thracian guardsman named Mucapor is named as the killer. His action cannot have been popular, for neither he nor anyone else on the spot claimed Aurelian’s throne. That suggests frustration among an officer corps that had not, for a change, wanted to see their emperor dead. The man who did in the end succeed Aurelian – Claudius Tacitus – is an enigmatic figure, because the very few extant sources veil his succession in a bizarrely romantic fiction.

There are only two points of fact. Presumably because of internal disagreements, there was a six-week interregnum, with diplomatic twists and turns among rival factions that are now irrecoverable. At the end of it, Tacitus was invited to become emperor, left retirement in Campania in southern Italy and took up the purple in Rome. Because his name recalled that of the famous early imperial historian, the fourth-century author of the Historia Augusta concocted a purely imaginary mishmash: the army defers to the senate, which defers to the army, which again defers to the senate, so that a distant descendant of the historian takes the throne as the last senatorial emperor and includes among his edicts one that requires the copying and reproduction of his forefather’s works. (If only this fairy tale were true, the tenuous thread on which hangs the preservation of Tacitus’s extant writings – most of it from a single manuscript, more of it lost for ever – would be more robust.) The charms of this happy story are such that even Gibbon was deceived, and more recent commentators have not refrained from buying into one or another part of it. But the prosaic reality is that Tacitus was just another general from the Danube, older than most and now retired, hence a safe compromise candidate; he was quite possibly another marshal of Gallienus of whom we happen to lack a prior record. Clearly he was a known quantity, trusted by the Balkan soldiers, but so far away from the scene of Aurelian’s murder that he could not possibly have had anything to do with it. That explanation at least fits our tiny residue of reliable evidence, which suggests that after six weeks there was universal agreement that Tacitus should become emperor and could make his way out to the Balkan armies. The most remarkable thing about the entire episode is that no other frontier army ventured to set up an emperor of its own while Aurelian’s Balkan forces sat paralysed. That must be a measure of just how honoured – and also feared – Aurelian had been.

Tacitus crossed into Asia in command of the late emperor’s field army, dealing en route with a naval attack from north of the Black Sea by Scythians, perhaps Goths or perhaps the same Heruli attested in that role half a decade earlier. His explicit aim was to catch and chastise Aurelian’s murderers, some of whom had fled east. The Thracian Mucapor who had killed Aurelian, otherwise unknown, was captured and tortured to death. Tacitus then sent a relative named Maximinus to Antioch, intending to return to the west himself, but this Maximinus made himself so hated that the leading Antiochenes plotted with the surviving murderers of Aurelian to kill him. They then raised a force to pursue Tacitus and killed him as well, at Tyana, a city not long previously spared by Aurelian despite its support for Zenobia. It is possible, if speculative, that we are witnessing in this episode the vengeance of the Antiochene upper classes for the damage wrought on Daphne by Aurelian’s victory there in 272 – the villain was dead, but his self-declared avenger could still be killed. Aurelian’s army – having briefly been Tacitus’s army – now chose the latter’s half-brother, the praetorian prefect M. Annius Florianus, as his successor. But the Syrian army, presumably complicit in the Antiochene opposition to Tacitus, elected Marcus Aurelius Probus instead.

Probus was another of the old marshals of Gallienus, and matters appeared to be returning to the dreadfully familiar status quo of a decade previously – coup and counter-coup, putsch and push-back balancing each other out with no one the long-term winner. Florianus, about whom we know virtually nothing, began to lead his army towards Syria to confront his challenger, but had to call a halt at Tarsus in Cilicia when plague struck. Perhaps lacking the will to fight in their stricken condition, the army mutinied and declared for Probus rather than pursue the war. Florianus abdicated, but was promptly murdered, accused of plotting to retake the throne. With Florianus dead, Probus then executed the surviving murderers of Aurelian at a banquet.

Despite this promising start, Probus had nothing like Aurelian’s force of personality. His reign, which is the worst documented half decade in all of imperial Roman history, seems to have been a series of constant revolts from one end of the empire to the other. Indeed, the biggest surprise is that he managed to hold his throne for a full six years without being assassinated. Chronology is imprecise at best, but the reign opened with the Egyptian communities of the Thebaid joining with the pastoralist Blemmyes, who lived at the desert edge of the province, to attack the city of Coptos. This reflects the peculiar political and social world of southern Upper Egypt, hundreds of miles upriver from the main centres of the province and confined to a relatively narrow strip of land along the upper Nile, ending with the garrison town of Elephantine just below the river’s first cataract. Coptos was the southernmost of the really prosperous Nilotic towns and it was also the station at which the military road across steppe and desert from the Red Sea port of Berenike reached the Nile corridor. As such, it was an attractive target for periodic nomadic predation, but what is interesting here, and not fully explicable, is the participation of local provincials alongside the nomads in attacking a major town. Whether our evidence is a stray glimpse of a local feud that just happens to survive, or whether it represents something more systemic is unclear: perhaps it is a sign of the disruption and disturbance that Aurelian’s currency reform had generated even in far off corners of the empire.

There was also fighting on the Danube against Goths and Sarmatians, but this had become a chronic rather than an acute problem, and one that would take many years to deal with effectively. Indeed, the renewed threat to the Balkan provinces from the Sarmatians is almost certainly a consequence of Gothic expansion, inasmuch as they had proved relatively biddable neighbours over many previous decades. While Probus was campaigning on the Danube, the governor of Syria, Saturninus, revolted, claiming the purple for himself, before his own men turned on him and killed him at Apamea. Then there was an uprising at Cremna in Pisidia, led by a local worthy named Lydios, and the town withstood a major siege operation at the hands of the governor, Terentius Marcianus, before surrendering (the surviving remains of the siegeworks are some of the best evidence we have for Roman military engineering).

Perhaps at the same time, the Rhine armies acclaimed Proculus and Bonosus in turn as their emperors at Colonia Agrippina. The soldiers had good reason to be dissatisfied, for it seems that Gaul and Germany had been badly mauled by raiding since the collapse of the Gallic regime and had received little or no help from the imperial centre. On the other hand, some argue that Probus fought a series of campaigns in Germany, before returning to the Balkans, and that the usurpations followed his departure. The sources are too fragmentary to tell us. Finally, the governor of Britain, whose name is unknown, was also acclaimed by his troops – which was unusual in that it was a province that had generally been at peace throughout the century. In all these cases, and most unusually, there are no coins, which suggests the extent to which Aurelian’s monetary reforms had ruined the money economy.

Be that as it may, Aurelian had at least enforced unity and a certain peace on the empire. All that now looked as if it was disappearing. Probus could strike gold coins depicting the labours of Hercules, but he was not up to the Herculean challenges he faced. With the chronic frontier warfare, which fed a cycle of civil violence, the gap between the militarised parts of the empire and those that were generally at peace widened, and so too did the swathe of territory that was permanently exposed to the passage of imperial armies and the quartering of soldiers on them. And these soldiers, as the events of Probus’s reign just described show, were terribly prone to violent rebellion. Probus himself discovered as much in the end, despite having hung on to power for longer even than Aurelian. But, as with so many of his predecessors, the hostility of a cabal of senior officers would prove fatal. Probus celebrated a triumph at Rome in 281 and then returned to the east in the following year, reaching Sirmium in Pannonia before he was murdered, having just learned that his praetorian prefect M. Aurelius Carus had been proclaimed emperor.

Carus is interesting if only because he is the only emperor in the long sequence of third-century soldier claimants not to have a Danubian background – he was, instead, from Narbonensis in the south of Gaul, and the nature of his connection to the world of senior command is unclear. We lack any sense of his power base or how he was able to succeed in a highly competitive environment that seems to have excluded men of his regional background. If the reasons for and circumstances of his accession are obscure, Carus at least had a family with dynastic potential: there was an adult son, Carinus, and a younger one, Numerianus, as well as a grandson, Carinus’s son Nigrinianus. The family’s dynastic ambitions were clear from the start: while Carinus campaigned against the Quadi in Pannonia, celebrating a triumph at Mediolanum in 282, Carus sought to enhance his personal prestige by leading Probus’s campaign army into a Persian war. The timing was propitious, for the ruling Sasanian dynasty had collapsed into the sort of civil war that had afflicted the Roman state for decades. Shapur I had intensified his father Ardashir’s practice of setting up his various sons as kings of the various parts of his empire. The eldest son, Ohrmazd, had been imposed on Armenia, but other sons ruled Mesene, at the head of the Persian Gulf; Sind, Seistan and Turan in the deserts of present-day Baluchistan; Gilan in the mountainous southwest of the Caspian; and in Kerman and Adiabene. That left plenty of rivals who might hope to succeed Shapur when he died, as he finally did in 272. The fact that he had never capitalised on his defeat and capture of Valerian was the result of his taking much more serious account of his multiple frontiers than did Valerian’s Roman successors, all of whom would immediately privilege the eastern frontier over any other, and prioritise a usurper’s challenge over any possible damage their provincials might be suffering.

Shapur, like his father Ardashir and for reasons that the absence of Sasanian literary sources makes it impossible for us to analyse, seems to have sought prestige and ideological legitimation among his core supporters in Fars by appropriating the necropolis of the Achaemenids and advertising far and wide the way he had humbled three different Roman emperors. But in pragmatic strategic terms, he focused more time and military attention on Central Asia: Sogdiana, which the Sasanians were never fully able to control; Margiana and Bactria, which fell to Persia during his reign; and beyond the Hindu Kush, in Kabul, Gandhara and Swat, where the Sasanians displaced the last of the Kushans. Thereafter, the Kushanshahr became a client of the main Sasanian line, usually ruled by one of its cadet branches, occasionally acting as a launching pad for rebellions, and striking a currency we know as Kushano-Sasanian (because it was minted on the Kushan gold standard rather than the silver drachms of Iran proper). As fourth-century history will show us once again, it is the numismatic evidence that often gives us information that we lack from any other source. In terms of the Sasanian conquest of the Central Asian and north-west Indian hinterland, which used to be dated as much as a hundred years later, new coin finds and in particular the proper identification of mints, including travelling mints, have definitively shown the conquest to be the work of Shapur I: the royal silver coinage is increasingly struck in the Far East during the latter part of Shapur’s reign, some of it by engravers from the Iranian heartland, which must indicate travelling mints created in order to pay multiple armies on campaign.

After Shapur’s death in 272, his eldest son Ohrmazd I succeeded him, and is recorded fighting campaigns in Central Asia. Ohrmazd, however, reigned for just one year. His successors were all children, their courts dominated by the Mazdean priest Kardir, who had risen to prominence in the last years of Shapur and whose priority seems to have been to solidify the status of Zoroastrianism and the worship of Ohrmazd as the official cult of the Sasanian state. Although he was himself a convinced believer in the truth of his Zoroastrian religion, Shapur had been tolerant of other religions. There was room in the Zoroastrian worldview for non-believers, as indeed there had to be because its practice was so tightly connected to an Iranian ethnic identity. Christians and Jews in particular could contribute to the eternal fight of good against evil by cultivating their own, legitimate but inferior, faiths in their own way. They could not seek converts among the Iranian nobility, and the leaders of their communities were expected to play a similar role to the Magian priesthood in organising the economic affairs of their own communities and ensuring that their taxes were paid. Shapur’s tolerance had gone further; he even encouraged the new revelations of the prophet Mani, whose gnostic and dualist beliefs would go on to play a divisive role in Roman society. Some have suggested that Shapur saw in Mani’s revelation a belief system that would resonate, more than would his own Mazdaism, with the many non-Iranians of his empire, from Christians and Jews in Mesopotamia to Buddhists in the east and north-east.

Then, with Shapur’s death, the Zoroastrian priesthood seized its chance to stifle what it saw as a potent threat to its social and religious dominance. The minorities of Varahran I (273–6) and Varahran II (276–93) seem to have been dominated by the Mazdean priesthood on the one hand and the Persian nobility on the other. It is during these reigns that we find the otherwise unknown phenomenon of a non-royal figure (Kardir again) inscribing his exploits in a public and royal context. It was also now that Mani was arrested and left to die in prison, perhaps in 276. Then, in the 280s, there was a civil war between supporters of Varahran II and those of his cousin Ohrmazd. The latter relied upon nomadic Sakas and levies from the Kushanshahr for support, minted his own coinage and may have even taken the title of Kushanshah. The war between the cousins also divided the empire’s upper nobility, and the great families of both Persian and Parthian background now began to demonstrate an attitude they would maintain right until the collapse of the Sasanian dynasty: while the right of the Sasanian family to rule went unchallenged, the nobility reserved their right to choose among potential royal claimants, and depose one Sasanian in favour of another should that seem necessary. For the first, but not last, time in the history of the two empires, a combination of internal religious ferment and external distraction on Persia’s eastern frontiers made the 270s and 280s a period during which Rome had little to fear from the armies of its imperial rival.

Perhaps realising this – though with the caveat that Roman military intelligence on Persian affairs was never very comprehensive – Carus marched into Mesopotamia in the summer of 283, straight down the Euphrates to the capital at Ctesiphon. The campaign went so well that word was put about that Ctesiphon had actually fallen to the emperor, and one strand of sources preserves that story. But Seleucia-on-Tigris, by then a suburb of Ctesiphon, was indeed sacked, as attested by Ammianus Marcellinus, a good fourth-century source, who had inspected the ruins with his own eyes when serving on the emperor Julian’s Persian campaign.

Carus’s military success was not rewarded by his army’s loyalty: he was murdered, like so many of his predecessors, though there is also a story, surely legendary, that his tent was struck by lightning. His murder left the army stranded deep inside Persia, and the first order of business was to extricate it. Whoever was responsible for Carus’s death, no one claimed his title, which passed to his sons Carinus and Numerian, the latter a young boy who had accompanied his father’s campaign army into Persia. Carus’s brother-in-law Aper, who had succeeded him as praetorian prefect, probably took de facto control of affairs until the army was back in Syria. Disarray among the Persians had helped them to extract the army intact and largely unscathed – Aper and his officers were lucky that no new Shapur emerged to harass them on their retreat.

The Roman forces reached Emesa in Syria by March 284, and Cyzicus and Nicomedia in Bithynia later in the year. There, in November, Aper announced the death of Numerian. Although the boy had been ill, his death was almost certainly murder. There was now just one emperor, Carinus back in the west, but Carus’s campaign army was not likely to accept its subordination to a western rival. As Carus’s brother-in-law, Aper believed he should succeed to the throne, but the army did not concur. Instead, its choice landed on a relatively junior officer, C. Valerius Diocles, a man of about forty who was the comes domesticorum, commander of the main guard unit which travelled with the emperor. Diocles accepted his acclamation, and renamed himself Diocletianus, a more Latinate-sounding name than his obviously Greek (and low-born) original. At a meeting of the whole army, he personally executed Aper in full public view, claiming thereby to avenge the murder of Numerian. He then proclaimed himself consul prior, along with L. Caesonius Bassus, a member of the Roman senatorial aristocracy. The truth of what happened can never be known. Numerian may have died naturally, with both Aper and Diocletian seeking to profit by his death, or one of them may have killed him. (We do not need to believe the story that, for some time after his murder, his corpse had been carried alongside the army in a litter in order to disguise it, word being put about about that he was suffering from a disease of the eyes.) A child emperor with no protector was a victim awaiting the slaughter, his death a foregone conclusion, but Diocletian’s acclamation at Nicomedia necessarily meant civil war. Carus’s remaining son, the emperor Carinus, could not be expected to acquiesce in the eastern army’s presumption and, by taking the consular title for himself and appointing a colleague, Diocletian effectively declared war on Carinus.

Diocletian overwintered in Asia, before crossing into the Balkans, with an eye on attacking Carinus. The latter marched east from Rome, knowing that he would need to face Diocletian in the Balkans, but he had none of the personal authority his father Carus had commanded, and he had no dynastic prospects as his son Nigrinianus had died. The praetorian prefect Sabinus Iulianus, left behind by Carus to watch over his son, now revolted, presumably viewing him as unlikely to survive a war with the seasoned army of Diocletian. This revolt was put down at Verona, but then the corrector Venetiae, Marcus Aurelius Julianus, revolted in Pannonia, striking coins at Siscia before being defeated by Carinus early in 285 as he marched to the Balkans. There, at the Margus river, his army faced its more formidable rival, and Carinus’s newly appointed praetorian prefect, Tiberius Claudius Aurelius Aristobulus, promptly betrayed him as well. So too did M. Flavius Constantius, who was then serving in an extended military command over Dalmatia and the Balkan interior. Constantius had probably previously served as a tribune of the domestici under Diocletian; he would go on to be a major prop of the latter’s new regime. As we have seen many times, third-century soldiers were quick to desert a commander whose prospects looked dim, so Carinus was promptly assassinated by his own men. He and his wife Magnia Urbica suffered damnatio memoriae, as did Carus and Numerian, their names being chiselled from inscriptions.

The battle of the Margus ended in a relatively easy victory for Diocletian, but the past decades had shown that initial success counted for very little without persistent labour. Diocletian marched his army into northern Italy and took up residence at Mediolanum. There he made another general of no great pedigree his colleague as caesar, on 25 July 285. This general was Marcus Aurelius Maximianus, the son of a shopkeeper from very near Sirmium whose family had gained citizenship only with the edict of Caracalla. Diocletian and Maximian had come up through the ranks together and both had been present on Carus’s Persian campaign. Neither man troubled to go to Rome at this point – there were too many problems on the northern frontiers. By the autumn of 285, Maximian was campaigning in Gaul, where the death of Carinus had prompted a revolt by a man named Amandus, who proclaimed himself augustus. Carus was from Narbonensis and Amandus may have been a relative; later sources also name one Aelianus as a part of the revolt, but he remains a mystery, as no authentic coins in his name have survived, only modern fakes. Maximian seems to have suppressed the revolt fairly efficiently, and later tradition reinvented Amandus and Aelianus as rustic brigands rather than the provincial notables they were. Still, their uprising prompted the usual uncertainty on the frontier, and Maximian also fought against Franci or Alamanni from across the Rhine. Diocletian, at the same time, was fighting against Sarmatians on the Danube bend, now a perpetual sore point trapped between the empire and what we infer was Gothic power expanding up the Danube and in the former province of Dacia. We do not know where Diocletian overwintered in 285–6, but he was back in Asia Minor in March of 286. Gaul was far too unsettled for Maximian to leave it.

The expedient of having an augustus in one place and a caesar of similar age and experience in another was a novelty. No one, least of all his soldiers, could have expected Maximian to remain in a subordinate role for ever. On 1 April 286, he was duly proclaimed augustus, without Diocletian being present, but with his full, stated approval. The cynical interpretation would be that Diocletian gave his old comrade something he would obviously have wanted before he took it for himself and thereby sparked a civil war; the loftier and more noble reading would have Diocletian and Maximian embarking on a bold experiment in power-sharing, the better for each to hang on to the imperial thrones they had won. What is certain is that they, or Diocletian, now began to concoct an elaborate ideological system to explain what everyone could see was an unprecedented relationship. Diocletian began to style himself as the child or companion of that most Roman of gods, Jupiter, the chief of the Capitoline triad, and thus the greatest god in the Roman pantheon. Maximian, for his part, was to be Hercules, Jupiter’s son and loyal subordinate. Both emperors were equally augusti, both augusti were equally divine, but Diocletian was the senior, as Jupiter was senior to Hercules. This articulated a traditional Roman paternalism but injected the issue of divine election into the question of who could be emperor. Here we can see how far things had changed since the Antonine age: a hundred years before Diocletian, Commodus had been mercilessly ridiculed for his self-assimilation to Hercules; fifty years earlier, Elagabalus’s belief in his own divine incarnation led directly to his assassination. By contrast, when Aurelian advertised his personal relationship with the unconquered sun it was accepted as perfectly reasonable, as, so it seems, was Diocletian and Maximian’s identification with Jupiter and Hercules.

It remained to be seen whether this bold experiment in power-sharing between two men with no dynastic connection would work, especially because Maximian had a son on the threshold of adulthood, while Diocletian had only daughters. The presence of a presumptive heir to the junior augustus must have complicated expectations about both the present and the future balance of power. In the moment, though, both men had plenty of work to do.

In 286, a general named Carausius revolted in Britain, declared himself augustus and began minting coins. A Menapian by birth, from the territory between the Rhine and Scheldt, he had been the commander of the Channel fleet, protecting the coastline from Saxon and Frankish pirates. His revolt was a serious concern: not only the armies of Britain, but also many stationed in Gaul itself took his side, and he began striking coins at Rotomagus (modern Rouen) as well. Maximian was too busy fighting on the frontiers of eastern Gaul throughout the later 280s to do anything about this revolt, which is surprising given that usurpations almost axiomatically trumped barbarian incursions on the scale of imperial threats. The fact that Maximian divided his time among the main cities of the Rhineland – Trier, Moguntiacum and Colonia Agrippina – may suggest that he doubted his ability to challenge the usurper successfully, and it may be that a wide tranche of western Gaul recognised Carausius rather than the new regime of Diocletian and Maximian. Rather than face the usurpation outright, Maximian launched a police action on the frontier to shore up his authority, crossing the Rhine, sowing terror among the barbarians and installing a king named Gennobaudes among some of the Franci there. The victory emboldened him and, in 288, he led an army against Carausius, winning a battle at Rotomagus and regaining control over north-western Gaul. He then began to construct a fleet, taking the better part of a year to do so, only to see it destroyed in a North Sea gale before he could launch an invasion. Carausius promptly retook the Gallic towns he had recently lost to Maximian. It must have seemed like the fatal rhythm of the mid third century was returning.

While Maximian was in Gaul, Diocletian was fighting in the Balkans, with journeys to Syria in 286 and 287 to observe developments in Persia, which had been ignored for more than a decade apart from the brief campaign of Carus. Aurelian’s destruction of rebellious Palmyra had destabilised politics on the Syrian frontier, for in 289–90 we find Diocletian at the old caravan city fighting the desert tribes and also visiting the important Severan site of Emesa. Further north, Shapur’s weak successors had lost control of Armenia, and Diocletian was able to install as king the Arsacid Tiridates III, who had fled to the Romans as a child decades earlier.

On the whole, Diocletian had been having much greater success in his sphere of activity than Maximian in his. In 290, Diocletian inspected the field armies in the Balkans, en route to northern Italy, where he met Maximian in the winter of 290–91. Whatever else it may have been – and there was no love lost between the two augusti, however much they might need one another – the meeting at Mediolanum was a show of unity. The two received embassies from the cities of the west and from the senate at Rome. This reaffirmed the tradition by which earlier soldier emperors had coopted the support of the imperial capital’s aristocracy, although neither Diocletian nor Maximian showed the slightest inclination to go there. Despite the symbolic power that Rome and its senate still commanded, sometime in the quarter century between Gallienus and Diocletian, the traditional link between emperor and eternal city had been comprehensively broken. In fact, there was no longer really an imperial capital, just a series of residences of greater or lesser importance: Trier, Mediolanum, Nicomedia and Sirmium were the most important at this stage in Diocletian and Maximian’s joint reign, but there would be others. The outlines of the old, high imperial world were growing very blurred, and we – with the full benefit of hindsight – can see a new late imperial order beginning to take shape.

Shortly after the meeting in Milan, Maximian resigned the command against Carausius. It was taken over by a subordinate general, the same Flavius Constantius whose timely betrayal of Carinus eased Diocletian’s path to power in 284 – and who had already been rewarded by marriage to Maximian’s daughter Theodora in 289. The events of later 291 and all of 292 are obscure in the extreme, almost as opaque to the historian as the reign of Probus, but when the sources resume in 293, it is with a series of momentous, indeed unprecedented, reforms. Perhaps these were agreed upon at Mediolanum in 291; perhaps Diocletian spent the intervening years working out how he might shore up a regime that remained shaky even with two cooperative joint rulers at its head. What emerged at the start of 293 was nothing short of extraordinary – having discovered that even two emperors were not enough to secure a stable regime, and having quite unaccountably survived longer than any emperor since Gallienus, Diocletian and his colleague overhauled the shape of imperial government – from the currency to the army, to the administration of the provinces and even to the imperial office itself. On 1 March 293, two new caesars were appointed, and the shared emperorship of Diocletian and Maximian became an imperial college of four members, two senior augusti and two junior caesars. With the creation of this tetrarchy, a new stage in Roman history begins, one that reinvented the very nature of the empire.