Roman Empire

The republic that did not end

formationDrafted from Millar, The Emperor in the Roman World (31 BC - AD 337) (1977)

In 27 BC Octavian stood up in the senate and gave the state back. He resigned his extraordinary powers, was voted a set of ordinary ones in return, received the name Augustus, and spent the next forty-one years governing an empire while holding no office that had not existed under the republic.

Musée de la Romanité, Nîmes · Millar, The Emperor in the Roman World (31 BC - AD 337) (1977) · Photograph: Krzysztof Golik · CC BY-SA 4.0

That is the constitutional arrangement the empire begins with, and it is worth taking seriously rather than treating as a fraud. Augustus held tribunician power, which gave him the tribune's veto and personal inviolability without the office; and a proconsular command over the provinces that contained the armies, which put every legion under a man with no formal superiority over any other magistrate. The senate met, the consuls were elected, the courts sat. There was no coronation, no regalia, no title meaning king, and no law that said the republic had ended.

The bronze in the Capitol

The best evidence that the fiction was maintained rather than merely asserted is a document from a century later.

In 69, after a year in which four men claimed the throne, the senate passed a law conferring powers on Vespasian, and a bronze copy of part of it survives. It grants him, clause by clause, the right to make treaties, to convene the senate, to extend the boundary of the city, to do whatever he judges to be in the interests of the state — and it cites precedent for each: as it was lawful for the deified Augustus, for Tiberius, for Claudius.

A general who had just won a civil war stopped to have his powers voted to him, itemised, with references. For three centuries the Roman answer to by what right do you rule was: the senate and people conferred these specific powers, and here is the law that did it.

What an emperor did

Millar's book arrives at the same conclusion from the opposite direction. Rather than asking what the constitution said the emperor was, it asks what emperors demonstrably spent their time doing, and the answer is: replying.

Hearing embassies from cities. Judging appeals. Answering petitions from individuals, in writing, at length, on questions of inheritance, taxation, citizenship and local privilege — the rescripts, thousands of which survive in the legal codes. Confirming or refusing what somebody had asked for. The emperor's power was enormous and it was overwhelmingly reactive, exercised on cases brought to him by subjects who expected an answer and usually got one.

An empire of fifty to sixty million people was governed by a few hundred senior officials and a man who answered letters, and the reason that was possible is the subject of a later chapter.

What it was not

It is worth saying plainly what the empire did not have, because most readers arrive with an image of it.

There was no Roman bureaucracy in the sense that China or the later Byzantine state had one. There is no standing police. Below the provincial governor and his small staff, the empire was administered by the cities — councils of local landowners who collected the taxes, maintained the roads, ran the courts of first instance and paid for the games out of their own pockets, in exchange for status and Roman citizenship. Rome ruled by co-opting the people who were already in charge locally and making it worth their while.

That arrangement worked for two and a half centuries. What broke it is a story told several chapters on, and it is closely connected to Caracalla's edict, which comes not long after this one.

The first dynasty

expansionDrafted from Woolf, Rome: An Empire's Story (2012)

Augustus had built a monarchy without naming it, and for the next fifty-four years the empire's central question was whether one family could keep holding an office that officially did not exist.

Tiberius, his stepson and adopted heir, succeeded in AD 14 with the senate's powers voted to him in the same form Augustus had received them, and spent the back half of his twenty-three-year reign governing from the island of Capri, communicating with Rome by letter and leaving the city to a prefect, Sejanus, who used the distance to make himself the second man in the empire until Tiberius had him arrested and killed. It was the first demonstration that the office could be run at a remove, by correspondence, which is exactly how Millar's later emperors mostly governed anyway.

Caligula, Tiberius's great-nephew, succeeded in 37 to an enthusiastic welcome and reigned not quite four years before his own Praetorian officers murdered him in a palace corridor, the first emperor killed by the men meant to guard him rather than by a rival army. The ancient accounts make him erratic to the point of madness; what can be said with more confidence is that he demanded worship as a living god in a way Augustus and Tiberius had both been careful to avoid, and that the demand did nothing to protect him.

An accession nobody planned

Claudius, found by the Praetorians hiding behind a curtain in the palace after his nephew's murder and reportedly made emperor because he was the last adult male of the family left standing rather than because anyone had chosen him, turned out to be the most capable administrator of the four. He completed the conquest of Britain in 43, extended citizenship into Gaul, and built the harbour at Ostia that fed Rome for centuries afterward. An emperor discovered behind a curtain governed longer, and more competently, than the two young men who had inherited the position by right.

Nero

Nero came to the throne at sixteen in 54, murdered his own mother five years into his reign after at least one failed attempt, and spent much of what followed on music, poetry and the stage rather than administration, while the empire around him was actually run by his advisers for as long as they kept his confidence. The great fire of Rome in 64 burned two-thirds of the city; Nero rebuilt it on a more regular plan and built himself an enormous palace on the cleared ground, and the rumour that he had started the fire himself, to clear space for it, could not be disproved and has never gone away.

Musei Capitolini, Rome · Woolf, Rome: An Empire's Story (2012) · Photograph: Commons contributor · CC BY 2.5

He killed himself in 68, the last man with a blood claim to Augustus, having alienated the senate, the Praetorians and finally the provincial armies whose commanders began declaring against him one after another. Tacitus's judgment, that a secret of empire had been revealed - that an emperor could be made somewhere other than Rome - describes what happened within the year, but the condition that made it possible was created here: a dynasty that had spent four reigns treating the office as a family possession left no institutional answer for what happened when the family ran out.

Rome · Woolf, Rome: An Empire's Story (2012) · Photograph: Tyler Bell · CC BY 2.0

What legitimacy had rested on

Augustus's own cult sat carefully on one side of a line: temples to Roma et Augustus were built and encouraged in the provinces, especially in the Greek east where ruler cult was an old and familiar idiom, but Augustus himself was titled a god only after his death, by senatorial vote, and no living emperor in Rome itself claimed outright divinity. Tiberius declined honours he thought excessive. Caligula did not, and asked to be worshipped as one while he still breathed - the clearest sign the ancient sources give that he had stopped reading the room his predecessors had read carefully. The pattern set under Augustus, of divine honours granted rather than demanded and mostly reserved for the dead, is what every later emperor except Caligula actually followed, Christian ones included: an emperor's authority was backed by claims about divine favour, not by a living claim to be a god outright.

The frontier

expansionDrafted from Woolf, Rome: An Empire's Story (2012)

In AD 9 three legions under Varus were destroyed in the Teutoburg forest by a Germanic coalition led by a Roman-trained auxiliary officer, Arminius, whom Varus trusted. Suetonius has Augustus beating his head against a door and calling for Varus to give him back his legions. The legion numbers XVII, XVIII and XIX were never used again.

Rome did not stop expanding because of it — Britain was invaded thirty-four years later and Dacia a century after that — but the Rhine became a frontier rather than a stage on the way to the Elbe, and it stayed one for four hundred years. That is why the Teutoburg disaster belongs in a chapter rather than among the hinges: it changed where a line was drawn and not how the state worked.

An empire that stops

The structural fact underneath the whole record is that a conquest state has a fiscal problem when it stops conquering.

Northumberland, England · Woolf, Rome: An Empire's Story (2012) · Photograph: Benjamin Smith · CC BY-SA 4.0

The republic's expansion had been paid for by the expansion: plunder, indemnities, slaves, and the provinces that each war produced. By the middle of the second century the profitable directions had run out. North of the Rhine and Danube there was nothing worth the cost of taking. East of the Euphrates there was Parthia, which could not be conquered — Trajan took Ctesiphon in 116 and Hadrian gave it back the following year, and every subsequent Roman attempt did the same thing with the same result.

So the empire lived on taxes — land, heads and trade — and not on plunder. For the republic, plunder plainly had been structural income: the conquests paid for themselves and for everything else. For the empire it was not, and the difference between those two states of affairs is the reason Hadrian's Wall exists.

What a frontier cost

The army was between three hundred and fifty and four hundred thousand men and it consumed the great majority of the state's revenue. It was stationed almost entirely on the frontiers, in Britain, on the Rhine, on the Danube, in Syria and in Egypt, and Italy — the notional centre — had no legions in it at all.

This produced the arrangement a later chapter is about. The soldiers were paid, long-service professionals whose loyalty was to their commander and their unit, concentrated in a few provinces far from Rome, and the man who commanded a frontier army was three weeks' march from an undefended capital. Almost everything in the empire's political history follows from where the troops were.

The peace

It is also worth stating what the arrangement bought, because it is not the kind of thing that shows up in a list of provinces.

For roughly two centuries the Mediterranean basin was a single tax area with no internal customs barriers worth the name, one currency in three metals, a road network measured in the tens of thousands of kilometres, and no piracy. Wine from Italy and Spain, oil from Baetica, grain from Egypt and Africa and marble from Asia Minor moved in quantities the western Mediterranean did not see again for more than a thousand years, and the archaeological evidence for it is the shipwrecks: the count of dated Mediterranean wrecks peaks in this period and falls to a fraction of it afterwards.

The empire covered about five million square kilometres, which does not put it in the first rank by area. The Umayyad and Abbasid caliphates were larger. So were the Yuan, the Ming, the Han, the Xiongnu confederation and even the Rashidun conquests. The Mongol empire was seven times the size.

That is worth stating rather than hiding, because Rome is generally imagined as the largest empire in history and it was not close. What was unusual about it was never the extent. It was the integration: one tax area, one currency, one law, a road network, and a sea with no piracy on it, held together for two centuries. Nobody else did that at all, at any size.

Five emperors chosen rather than born

peakDrafted from Woolf, Rome: An Empire's Story (2012)

Trajan reached Ctesiphon in 116, the deepest a Roman army ever marched into Mesopotamia, and died the following year with the empire at its widest extent: something over five million square kilometres, from Britain to the Persian Gulf. Hadrian, his successor, gave nearly all of it back within months - the new provinces of Mesopotamia and Assyria were abandoned almost as soon as they had been declared, on the judgment that they could not be held at a cost worth paying. The empire's territorial peak and the decision to stop expanding for good happened in the same eighteen months.

Rome · Woolf, Rome: An Empire's Story (2012) · Photograph: Alvesgaspar · CC BY-SA 4.0

What made that succession possible without a war is worth setting out on its own, because for most of the second century the empire solved its oldest problem by means that did not survive it.

Adoption instead of birth

Nerva, an elderly senator raised to the throne after the Praetorians murdered Domitian in 96, had no son. He adopted Trajan, a general commanding on the Rhine with no blood connection to him at all, and the empire passed to a capable outsider rather than to a disputed heir. Trajan, dying without a son of his own, adopted Hadrian - the ancient sources disagree about whether the adoption happened cleanly or was arranged after the fact by Trajan's wife once he was already dead, which is itself a measure of how much rode on the formality being observed. Hadrian in turn adopted Antoninus Pius, on the explicit condition that Antoninus adopt two further heirs of Hadrian's own choosing: Marcus Aurelius and Lucius Verus.

Four consecutive successions, over eighty-four years, without a disputed succession or a usurping general. Nerva, Trajan, Hadrian and Antoninus Pius each chose an adult of demonstrated competence rather than passing the office to a son, because none of the first three had a surviving son to pass it to.

It looked, for the better part of a century, like a solution: not a rule of descent but a habit of picking the ablest available man and calling him son by adoption instead of by birth.

Musei Capitolini, Rome · Woolf, Rome: An Empire's Story (2012) · Photograph: Nicholas Hartmann · CC BY-SA 4.0

What the habit was not

It was not a rule. Nothing compelled an emperor with an actual son to adopt someone else instead, and the test came the moment one did.

Marcus Aurelius, who reigned from 161, spent much of his eighteen years on the Danube fighting the Marcomanni and other confederations pressing across the frontier, while a plague - brought back, by the ancient accounts, from the Parthian campaigns of his co-emperor Lucius Verus - killed a significant fraction of the empire's population and, with it, a share of the tax base and the army's own strength. It is the point at which the integrated, lightly-taxed empire of the second century begins visibly straining, two decades before the third century made the strain a crisis.

Marcus had a son, Commodus, and made him co-emperor and then sole heir rather than adopting a Nerva or a Trajan of his own generation's choosing. Commodus proved unfit for the office by nearly every account that survives him, governed erratically for twelve years, and was strangled by a wrestling partner in 192 in a plot his own household organised. The four adoptive successions had not been a new constitutional settlement. They had been four decisions by four childless men, and the fifth emperor in the sequence, the first one who did not need to choose an heir from outside his own house, took the alternative the moment it was available to him. What his successor Pertinax got for it, and what the Praetorians did the moment Pertinax was dead, is a story of an auction, told in full a little further on.

What it taxed

peakDrafted from Saller, The Roman Empire: Economy, Society and Culture (2014)

The rates were low. A tax on land, assessed from periodic censuses, and a tax on heads, at levels that varied by province and were generally a small fraction of production; customs dues of two to five per cent at the internal frontiers of the empire and rather more at its external ones; and, until Diocletian, no direct taxation of Italy at all.

The yield was enormous, because the base was fifty to sixty million people and because the collection cost the state almost nothing.

Government by co-option

This is the part that is hard to hold in the head, and it is the reason to read the economic historians rather than the narrative ones.

An empire of that size was administered by something on the order of a hundred and fifty senior officials — governors, procurators, legionary commanders — plus their small staffs. There was no imperial revenue service. The taxes were assessed and collected by the cities: councils of local landowners, several thousand of them across the empire, who were personally liable for the sum their community owed and who competed for the honour of being liable.

What they got for it was status, citizenship, priesthoods, the right to wear particular clothes, statues with inscriptions, and access to the emperor. What Rome got was an empire it did not have to garrison internally, run by the people who would otherwise have led the resistance.

The classic statement of the deal is that Rome ruled the Mediterranean by making the local elite of every city into Romans. What happened when it finished doing that is the harder question.

The grain

One flow of goods was not a market and it is the clearest single illustration of what the empire was for.

The city of Rome was fed by a state grain supply — the annona — bringing something in the order of two hundred thousand tonnes a year from Egypt and North Africa, shipped in a fleet the state contracted, stored in warehouses the state built, and distributed free to a registered list of a few hundred thousand adult male citizens. Egypt was acquired by Augustus as an imperial possession governed by an equestrian prefect that senators were forbidden to enter without permission, and the grain is the reason.

Ostia Antica, Italy · Saller, The Roman Empire: Economy, Society and Culture (2014) · Photograph: Sailko · CC BY-SA 3.0

When the eastern half went its own way, the grain went with it. Egypt fed Constantinople instead, which got an annona of its own; Rome did not, and the city's population falls by something like ninety per cent over the following two centuries.

What the coins say

The empire ran on a silver denarius that Augustus struck at about ninety-eight per cent fine. By the reign of Septimius Severus it was around fifty per cent. By the 260s the coin of the same name was a copper piece with a silver wash that wore off.

Debasement is the state paying its army with money it does not have, and the army in the third century is being paid a great deal more often because there are a great many more emperors. The sequence runs: no more conquest, so no new revenue; more frontier pressure, so a larger army; more usurpations, so more donatives; less silver per coin, so prices rise; taxes collected in coin lose value, so taxes are collected in kind instead; and by Diocletian the state is requisitioning grain, uniforms and transport directly and assessing them in units of land and labour.

A later chapter is what that looks like from the top. The point of putting it here is that the fiscal machinery described above — light, cheap, delegated, dependent on a stable coin and a co-operative local elite — is what the third century destroyed, and the state that came out the other side taxed harder, counted more and trusted nobody.

A law written down twice

aside?What aside meansDrafted from Saller, The Roman Empire: Economy, Society and Culture (2014)

The claim made for this empire more often than any other is that its law is the ancestor of most legal systems still in use, and the claim is worth examining rather than simply repeating, because it rests on two very different moments three and a half centuries apart.

The Twelve Tables

Roman law's own origin story is a republican one, not an imperial one. Around 450 BC, after a plebeian demand that the law stop being an unwritten privilege of the patrician priesthood who alone knew and interpreted it, a commission produced a code of ten tables, later extended to twelve, and had it posted in the Forum. Livy calls it, writing centuries later, the source of all public and private law - and whatever the truth of the tradition around its drafting, the underlying grievance it answered is a familiar one: a community that could not be governed by a law nobody but its rulers were permitted to read.

Almost nothing of the Twelve Tables survives except as quotations in much later writers, and what does survive is procedural and specific rather than philosophical - rules about summoning a defendant to court, about debt bondage, about the burial of the dead outside the city walls. It was not a constitution and did not claim to be one. It was a fixed, public starting point that every later jurist could cite, amend by interpretation, and argue from, for the better part of a thousand years.

The jurists

The law that later ages actually inherited was not the Twelve Tables themselves but the enormous body of professional commentary built on top of them, above all in the second and early third centuries AD, when a small number of jurists - Gaius, Papinian, Ulpian, Paulus - produced systematic treatises addressing nearly every question private law could raise: property, inheritance, contract, the status of slaves and freedmen, the obligations owed between family members. Gaius's Institutes, written as a teaching text, organised the whole of private law into a scheme - persons, things, actions - that later codifications, including modern civil codes written in the nineteenth and twentieth centuries, still substantially follow.

Alongside the jurists' own writing sat a second source of law already mentioned in connection with how an emperor actually spent his days: the rescripts, the written answers to petitions that accumulated by the thousand across three centuries and were treated, cited and argued from exactly as a jurist's opinion or an old statute was. A legal system built this way - case answered by case, opinion layered on opinion, with no single moment of comprehensive drafting - is difficult to summarise and was, by the fourth century, difficult even for trained lawyers to navigate in full.

Oxford · Saller, The Roman Empire: Economy, Society and Culture (2014) · Papyrus Oxyrhynchus XVII 2103 · Public domain

What came after

The effort to make it navigable again came three and a half centuries after this empire's own conventional end, when the eastern emperor Justinian had the whole inherited mass of statute, juristic opinion and imperial rescript collected, edited down, and reissued as a single authoritative compilation in the 530s. That codification is why the classical jurists' work survives at all in anything like its original form - most of it would otherwise have been lost the way the Twelve Tables themselves mostly were - and it is the version of Roman law that medieval and early modern Europe actually rediscovered and taught, centuries after the western half of the empire that had produced the material was gone.

Two moments, then, sit behind the claim: a plebeian demand in the fifth century BC that law be public and written, and a scholarly and administrative achievement of the second and third centuries AD that gave later ages something organised enough to rediscover. Neither survives complete on its own terms. What was transmitted was the second moment's own systematic reading of the first.

Everyone is a Roman

peakDrafted from Saller, The Roman Empire: Economy, Society and Culture (2014)

In 212 Caracalla issued an edict granting Roman citizenship to very nearly every free inhabitant of the empire. A papyrus from Egypt preserves a damaged Greek copy of it. Between twenty and thirty million people became Romans in a sentence.

Citizenship had been the empire's central instrument for two centuries. It was what an auxiliary earned by twenty-five years' service, what a provincial city negotiated for over generations, what a freedman's son could acquire and a freedman could not, and what separated a man who could not be flogged without trial from a man who could. The system described earlier — government by co-opting local elites — worked because citizenship was scarce and desirable.

Caracalla gave it to everybody, and the near-contemporary explanation is that he did it to widen the base of the inheritance tax, which only citizens paid. Later writers add a religious motive: more citizens, more people obliged to sacrifice for the empire's welfare. The edict itself, in the surviving text, says something about giving thanks to the gods.

Universitätsbibliothek Gießen · Saller, The Roman Empire: Economy, Society and Culture (2014) · Papyrus Gissensis 40 · Public domain

What it did

It removed the distinction the empire had been built on, and the distinction that replaced it was wealth.

Within a generation the legal sources stop dividing people into citizens and non-citizens and start dividing them into honestiores and humiliores — the more honourable and the more lowly — with different penalties for the same offence. A humilior could be flogged, tortured, sent to the mines or thrown to the beasts; an honestior was fined, exiled or, at worst, beheaded. The old protections attached to citizenship survived for the people who were already powerful and were withdrawn from everybody else at exactly the moment everybody else acquired them.

That is not an accident of drafting. A society that has abolished a legal distinction it depended on will produce another one, and the one it produced was franker about what it was measuring.

What it altered

The empire's tax system was administered by cities whose elites paid for the privilege of administering it. The privilege was citizenship. After 212 a city councillor's status came from the office rather than from the citizenship, the office became compulsory and hereditary during the third century because nobody wanted the liability any more, and by the fourth the curiales were a legally bound class forbidden to leave their cities.

An empire that had ruled by making people want to be Roman finished making them Roman and then had to start compelling the work that the wanting had paid for.

The army decides

contractionDrafted from Woolf, Rome: An Empire's Story (2012)

Rome never solved the succession. There was no law of inheritance for the principate, because the principate officially did not exist — Augustus had constructed a monarchy out of republican offices and never named it, which left the question of who came next without a rule to answer it.

Musée du Louvre, Paris · Woolf, Rome: An Empire's Story (2012) · Photograph: Jamie Heath · CC BY-SA 2.0

For two centuries the gap was papered over by adoption, dynastic luck, and the Senate's willingness to ratify whatever had already happened. In 193 the Praetorian Guard auctioned the office. In the fifty years after 235 there were more than twenty emperors, almost all of them raised by their own troops and killed by other troops, while the currency debased and the frontiers gave way in several places at once.

It is the commonest failure in the history of states that hire their own soldiers. The Abbasids bought Turkish slave soldiers to escape provincial armies and were chosen and destroyed by them at Samarra within thirty years. The Ayyubids' Bahri regiment took the sultanate in 1250. The Safavids broke the Qizilbash with Georgian slave troops who became a court faction in turn. An army that can make a ruler will eventually make one it prefers, and no polity here found a durable answer.

Diocletian's answer was to divide the office among four holders and to make the division administrative rather than personal, which stabilised the empire and began the split that became permanent in 395.

The auction

contractionDrafted from Dio, Dio's Roman History (1906)

In March 193 the Praetorian Guard killed the emperor and sold the office to the highest bidder, and a serving senator who was in the city that day wrote down how the bidding went.

Cassius Dio owed his praetorship to Pertinax, the emperor who had just been murdered, and he says so in the middle of the account, which is one reason to trust the rest of it. When the news broke, he says, some people ran home and some ran to the soldiers' houses and everybody looked to his own safety. Sulpicianus, the city prefect, happened to be inside the Praetorian camp already — Pertinax had sent him there to settle things — and began working on his own appointment. And Didius Julianus, a senator Dio describes as an insatiable money-getter and a reckless spender, always wanting a change of government, heard the news, went straight to the camp and stood outside the gates making offers.

American Numismatic Society · Dio, Dio's Roman History (1906) · American Numismatic Society · CC0

Then the sentence. Just as is done in some market and auction-room, both the city and her whole empire were bid off. The sellers were the people who had killed their emperor, and the would-be buyers were Sulpicianus and Julianus, who vied to outbid each other, one from within, the other from without.

The guardsmen ran between them. Sulpicianus is willing to give so much; now what will you add? And back the other way: Julianus offers so much; how much more do you make it? They reached five thousand denarii a man. Sulpicianus, who was inside the walls and was prefect of the city and had said five thousand first, would have won it — except that Julianus stopped raising by small amounts and jumped twelve hundred and fifty denarii in one bid, shouting the figure and holding it up on his fingers. The guard took the higher offer, partly for the difference and partly because Julianus had put it into their heads that Sulpicianus might avenge Pertinax. They opened the gate and proclaimed him.

Towards evening he went down to the Forum and the senate house escorted by a great many Praetorians with standards, as if going into action, in order — Dio says plainly — to frighten the senate and the people into acquiescence. The soldiers shouted Commodus at him, which was a compliment.

And here is the part a modern account of the third-century crisis cannot supply. Dio describes what it was like to be a senator receiving this news at home. We were possessed with fear of Julianus and the soldiers, he writes — himself especially, since he owed his career to the dead emperor and had repeatedly proved Julianus in the wrong when appearing against him in court. And then the senate went in and did what was required of it.

Julianus lasted sixty-six days. Three provincial armies proclaimed their own commanders within weeks, and Septimius Severus reached Rome first.

Everything structural about the later empire is legible in that afternoon. The office was detachable from the man; the price was payable in cash to the troops who held the capital; and the body that had once conferred it stood in the Forum being shown standards. What Dio adds to the analysis is that everyone present understood all of this at the time, and that the shame of it was the thing he remembered — a most disgraceful affair, and one unworthy of Rome.

The third century

contractionDrafted from Woolf, Rome: An Empire's Story (2012)

In 260 Shapur I of Persia captured the emperor Valerian alive at Edessa. It is the opening event of Sasanian Persia's own imperial history, and the Persian version of it is carved on a cliff face at Naqsh-e Rostam: the king of kings on horseback, a Roman emperor kneeling, another standing with his wrists held. Nothing like it had happened before and nothing like it happened again.

Within a decade the empire was in three pieces. Postumus held Gaul, Britain and Spain as a separate Roman state with its own senate and consuls. Palmyra, an Eastern Mediterranean caravan city with imperial ambitions of its own, took Syria, Egypt and most of Anatolia, and its queen struck coins with her son as Augustus. The central government held Italy, the Balkans and Africa.

The count

Between 235 and 284 there were more than twenty men that later lists count as emperors and a good many more who claimed it. Almost all were raised by the troops they commanded and almost all were killed by soldiers. The average reign is under three years.

At the same time: the Rhine and Danube frontiers were crossed repeatedly and deeply, with raids reaching Spain, northern Italy and Athens; the silver coinage became copper with a wash; a plague ran through the empire in the 250s and 260s; and the cities of the west began building walls, in many cases out of their own demolished monuments, for the first time since the republic.

Rome · Woolf, Rome: An Empire's Story (2012) · Photograph: Karelj · Public domain

What did not happen

It did not end, which is the fact worth putting in front of the list.

Aurelian, in a reign of five years from 270, retook Palmyra and then Gaul and reunited the empire; he also built a wall around the city of Rome, which had not needed one for six hundred years and which says most of what there is to say about the period. Diocletian, from 284, then did the reorganising, and the chapter that follows is about it.

The recovery is the reason the empire's end is dated to 395 and a division rather than to 284 and a collapse. Whatever the third century was, it was survivable, and the thing that survived it was recognisably the same state with a different constitution.

The mechanism

The empire had no way of transferring the office that the army could not override, and once the army had learned that an emperor could be made outside Rome it kept making them.

Two things are worth adding to it here.

The first is that the frequency was not madness but arithmetic. The empire had several frontier commands, each with an army that could plausibly march on the centre, each facing a real emergency in this period, and each with soldiers whose pay was losing value. Any commander who did not take the purple when his troops offered it was likely to be killed by them; any who did was likely to be killed by somebody else's troops later. There was no stable choice available to an individual.

The second is that the rule and the practice came completely apart, and both were real. For fifty years the succession worked the way a mamluk sultanate's did: whoever the troops preferred, until the troops preferred somebody else. And yet every usurper still had himself formally acclaimed and sought a senatorial ratification he did not need. The form was performed by men who had just murdered their way past it, every time, for half a century, because they could not think of a way of being emperor that did not involve it.

Diocletian and Constantine

contractionDrafted from Woolf, Rome: An Empire's Story (2012)

Diocletian ruled twenty-one years, from 284, and then retired to grow cabbages in a fortified palace at Split, which nobody in the office had done before or did afterwards. Between those two events he rebuilt the state.

The office was divided: two senior emperors called Augusti and two junior called Caesars, each with a territory and an army, the Caesars expected to succeed. The provinces were roughly doubled in number and made smaller, grouped into dioceses under vicars, with the civil and military commands separated so that no man held both in one place. The tax system was rebuilt on a standard unit of land and labour with a periodic reassessment and collection largely in kind. Prices were fixed by edict across the empire, which did not work. The army was enlarged and split into frontier troops and a mobile field army.

St Mark's Basilica, Venice · Woolf, Rome: An Empire's Story (2012) · Photograph: Archaeodontosaurus · CC BY-SA 4.0

And the imperial person was removed from view: purple and jewels, prostration before the emperor, an elaborate court ceremonial, and a title — dominus, master — that Augustus and his successors had spent three centuries carefully avoiding.

What the tetrarchy was for

It was an attempt to solve the problem the earlier chapters on the army and the auction both describe, by making the succession administrative. If there are four emperors, a frontier army that acclaims its commander has not created a civil war, because its commander is already an emperor; and if the junior emperors are designated in advance, the question of who comes next has an answer before it is asked.

Split, Croatia · Woolf, Rome: An Empire's Story (2012) · Photograph: TimeTravelRome · CC BY 2.0

It worked once, at Diocletian's own abdication in 305, and then failed comprehensively. Within a year Constantine had been acclaimed by his father's troops in Britain, which is precisely the thing the system existed to prevent, and eighteen years of civil war followed.

Rome · Woolf, Rome: An Empire's Story (2012) · Photograph: Livioandronico2013 · CC BY-SA 4.0

Constantine

He won them, and made three decisions that outlasted him by more than a millennium.

He attributed his victory at the Milvian Bridge in 312 to the Christian God, and and what followed was not the conversion of the empire — that took a century and was never complete — but the conversion of imperial patronage. Bishops were given civil jurisdiction, churches were endowed on a scale nothing else could match, and the emperor summoned and paid for the council at Nicaea in 325 to settle a doctrinal question by vote. Everything Christianity afterwards became in Europe rests on that decision to fund it.

He struck a new gold coin, the solidus, at a weight that held for seven hundred years, remained the Byzantine currency for the whole of that time, and as the bezant was the money of Mediterranean trade generally.

And in 330 he founded a second capital at Byzantium and called it after himself.

The city

Constantinople is why the Roman empire and the Byzantine empire are usually treated as two things.

It was not intended as a division. It was a second imperial residence in a strategically superb position — controlling the straits, close to both the Danube and the eastern frontier, defensible on three sides — and emperors had been governing from wherever the war was for a century already. What it became was a permanent eastern capital with a senate, an annona from Egypt, and an administration that outlasted the western one by a thousand years.

The year the division is conventionally dated to is 395, and the date is more arbitrary than it looks.

Scarcely one-third escaped

contractionDrafted from Marcellinus, The Roman History of Ammianus Marcellinus (1902)

Ammianus Marcellinus had been a staff officer in the eastern armies and was writing within a few years of the battle, and his account of 9 August 378 is the last thing in the record of the undivided empire that reads like testimony rather than administration.

The Romans came up to the Gothic position in high summer, after a long march, and found the plain in front of them on fire: the Goths had set the standing crops alight with burning faggots on purpose, so that the army arrived parched and exhausted, with men and animals both short of food.

Then the embassy. Fritigern, who Ammianus says was good at guessing how things would go and did not want a doubtful battle, sent a herald asking for a few nobles as hostages for his own safety, after which he would come over with military help and supplies. The proposal was well received. The tribune Equitius, a relative of the emperor and steward of the palace, was chosen by general agreement to go — and refused, because he had been a prisoner of the Goths once already and had escaped, and was afraid of what they would do to him. Richomeres volunteered instead, thinking it a bold thing to do and becoming in a brave man, and set off carrying documents proving his rank.

He never arrived. The archers and Scutarii escorting him, under two officers Ammianus names, gave way to what he calls an indiscreet impetuosity, attacked without orders as they came near the Goths, and then ran — disgracing, he says, the beginning of the campaign. And while that was happening the Gothic cavalry came back from foraging with the Alans, and descended from the hills, in the phrase every account of this battle has borrowed since, like a thunderbolt.

The casualty list Ammianus gives is a roll of the eastern command: the two senior generals, thirty-five tribunes holding no particular command, many battalion commanders, the master of the horse and the steward of the palace, and a young tribune whose father had commanded the armies and whom everybody had liked. Scarcely one-third of the whole army escaped. And then the comparison, which is the sentence the empire remembered: except for Cannae, no such destructive slaughter is recorded in our annals.

The emperor died somewhere in it and no body was found. Ammianus then writes Valens's obituary in the manner of his age — a faithful friend, a severe punisher of ambition, a man who insisted that everything be decided according to law and would not allow any decision against his own wishes, who listened to every informer without troubling about whether the charge was true. And then, without a transition, the body: dilatory and sluggish, swarthy, a cast in one eye not visible at a distance, well-set limbs, neither tall nor short, knock-kneed and rather pot-bellied.

He also notes that Valens had been given an oracle in three lines, the last of which mentioned a plain called Mimas, and that being unlettered he had laughed at it and then, as things got worse, became terrified of the very name of Asia. A pillar with that name on it was found near where he was last seen.

Marcellinus, The Roman History of Ammianus Marcellinus (1902) · Nineteenth-century engraving · Public domain

The empire did not end in 378 and Ammianus does not say it did. What the chapter records is the last time a Roman field army of that size was destroyed on Roman soil with an emperor inside it, described by a man who had served in it, who knew the officers by name, and who thought the disgrace had begun before the fighting did.

395

endDrafted from Woolf, Rome: An Empire's Story (2012)

Theodosius I died at Milan in January 395 and the empire passed to his two sons, Arcadius in the east and Honorius in the west, aged seventeen and ten. That year is the conventional end of the Roman empire and the beginning of the Byzantine.

It is a choice about where to cut a continuous thing, and the honest position is that no date works.

The candidates

395 is the date used here. Nothing constitutional happened: the empire had been governed by two or more emperors for most of the previous century, the division was administrative, the law was still one law, and everyone involved expected the halves to be reunited as they had been before. What makes it usable is that they never were. It is the last division that stayed.

Real Academia de la Historia, Madrid · Woolf, Rome: An Empire's Story (2012) · Photograph: Commons contributor · Public domain

476 is the traditional answer and it is the worst of them. Odoacer deposed a twelve-year-old western emperor whose authority did not extend much beyond Italy, sent the imperial regalia to Constantinople with a message that the west did not need an emperor of its own, and ruled Italy as a patrician nominally under the eastern emperor. Contemporaries did not treat it as the end of anything. The date's prominence is a later western construction.

1453 is what the eastern half's own inhabitants would have said, since they called their state the empire of the Romans until the walls were breached.

Or no date at all, which is the position that the Roman state ended gradually and in different places at different times: in Britain around 410, in Gaul in stages across the fifth century, in Africa in 439, in Italy over the sixth century's wars, and in the east not until an Ottoman cannon.

What kind of relation this is

Whatever happened in 395, it was not a conquest, not a collapse and not a claim by one state on another's legitimacy. It was a partition.

And it is a partition unlike the other cases where a state disappears and reappears. Babylon, Assyria and Elam each have two periods separated by centuries of somebody else's rule, which is genuinely one thing seen twice across a dark age. Rome and Byzantium have no gap at all. The eastern empire begins in the year the division is dated to, with the same law, the same army, the same taxes, the same capital that Constantine had built, and the same emperor's brother on the throne.

Which is why every attempt to say precisely what ended in 395 comes back to the same answer: nothing did, and the division became permanent afterwards for reasons nobody in 395 could have known.

What actually ended

The western administration, over the following eighty years, and with it the integrated Mediterranean economy. The shipwreck count falls. The African grain stops reaching Italy after 439. Britain drops out of the coin economy entirely within a decade of 410. The city of Rome, at over half a million people when Constantine founded his rival, is under fifty thousand by the sixth century.

The eastern administration did not end. It ran for another ten and a half centuries, which makes the Roman state, counted honestly from Augustus to the last emperor, the longest-lived polity in European history by a very wide margin.

The name everyone wanted afterwards

afterlifeDrafted from Woolf, Rome: An Empire's Story (2012)

The most durable thing the Roman empire produced was its own name, and the history of the next fifteen centuries is partly a history of people claiming it.

Charlemagne's coronation in 800 was an attempt to have the western half again. Mehmed II took Constantinople in 1453 and the title Kayser-i Rûm with it. Those two claims span six hundred and fifty years and two religions, and neither claimant had any connection to the Roman state beyond standing on ground it had once held.

Imperial Treasury, Vienna · Woolf, Rome: An Empire's Story (2012) · Photograph: Commons contributor · CC BY-SA 3.0

The Byzantines are not on that list, and it is worth saying why. They did not claim to be the successors of Rome; they claimed, correctly, to be Rome — the eastern half continuing without a break in institutions, law or self-description, calling itself Basileia Rhomaion until the walls came down. A state does not claim itself.

The empire itself is worth stating plainly. At Trajan's death it held about five million square kilometres and perhaps a fifth of the world's population, ran on a professional standing army and a body of civil law that is the ancestor of most legal systems in use today, and maintained a Mediterranean-wide economy that did not recover its scale in the west for a thousand years after it stopped.

Dating its end to 395 is a choice about where to cut a continuous thing. The west was deposed in 476; the east continued to 1453; and neither of those is the end of the same object.