The empire that came out of the third century was rebuilt from the frame outwards by a soldier from Dalmatia who held power for twenty-one years and then retired, and almost every feature of the later Roman state is his.
The previous lesson left three conclusions from the crisis. One frontier could be held by the old army and two could not. A coinage could not be debased indefinitely without ceasing to be money. And an emperor with a two-year horizon cannot govern. Diocletian, proclaimed in 284, addressed all three, and the solutions cost more than the problems had.
Four rulers
The first move answered the political problem directly. In 285 Diocletian made a fellow officer, Maximian, his colleague, and in 293 he added two junior emperors, Constantius and Galerius, each with a territory, a court, an army and a capital: Trier, Milan, Sirmium, Nicomedia. Rome was the capital of none of them.
The logic was the one the previous lesson identified. If a frontier army's incentive to proclaim its own commander comes from the emperor's absence, then have an emperor present on every frontier. If a capable general is a threat, make him a colleague. Diocletian and Maximian took the divine surnames Jovius and Herculius, presenting themselves not as two men who had agreed to share but as a single sacred authority in two persons, with the Caesars beneath them.
Whether this counts as dividing the empire is worth settling now, because it recurs to the end of the course. It did not. There was one empire, one body of law, one citizenship and one coinage, and legislation issued by any emperor ran in all the territories. What was divided was the work. That distinction holds through 395 and beyond, and much of the confusion about the fall of the west comes from people at the time and afterwards treating a division of labour as a division of states.
More provinces, more officials
The administrative reform was equally direct. Provinces were subdivided from about fifty to close to a hundred, grouped into twelve dioceses under officials called vicars, and above them regional prefectures. A governor now ran a much smaller territory, and, crucially, ran no troops: civil and military commands were separated, so that a man with soldiers had no tax revenue and a man with tax revenue had no soldiers.
The cost was a bureaucracy of an order the earlier empire would not have recognised. An earlier lesson counted around 350 senior officials for sixty million people; the fourth-century establishment ran into tens of thousands of salaried posts, and the change is visible in the sources as a new kind of complaint, about fees, delays, and men who must be paid to move a document. Contemporaries thought it enormous. Measured against the Chinese comparison from that earlier lesson, it was still small.
An army twice the size, and how to feed it
Example. John the Lydian, using official material, gives Diocletian's army as 435,266 soldiers with 45,562 in the fleets. Set that against the quarter of a million of the early empire and against a tax base that two plagues and fifty years of disorder had probably shrunk.
The total is 480,828, or roughly twice the force of the first century. An earlier lesson found the old army absorbing over half of an imperial revenue near 800 million sesterces, when the economy was at its largest and the coinage was sound. Now the army is doubled, the population is smaller, and the silver coin has become a copper token. Lactantius, a hostile witness, says the number of men receiving was becoming greater than the number paying, which is rhetoric, but the direction is right and no arrangement built on cash wages could survive the arithmetic.
So the state stopped relying on cash. The annona militaris, formerly an emergency requisition, became the ordinary way of supplying the army: grain, meat, oil, wine, uniforms, animals and labour, assessed on the provinces and delivered in kind. Soldiers were fed and clothed from what was collected rather than paid to buy it. Once you see that the coinage had failed as a store of value, this is not a step backwards into barbarism but the only workable response: a state whose money is unreliable moves to goods, and it has to build the machinery to assess and move them, which is where the new bureaucracy comes from.
Now you. Why not simply raise cash taxes and pay the army properly?
Answer
Because the tax would be collected in a currency the state itself had ruined, and the interval between assessment and spending would eat the proceeds. When prices are rising fast, any fixed money tax falls in real value between the day it is set and the day it is spent, and the state is the largest holder of the depreciating asset. Requisitioning wheat avoids the problem entirely: a modius is a modius whatever the coin is doing.
There is a second reason, which is that the earlier system relied on cities to assess and deliver a cash total, and city councils were failing. Their members were fleeing an office that had become a liability, exactly as an earlier lesson predicted. A state that cannot rely on the intermediaries has to do the assessing itself, which means its own registers, its own officials and its own units. The move to tax in kind and the growth of the bureaucracy are the same event seen from two sides, and both follow from the collapse of the money.
Counting everybody
Requisition in kind needs a much more detailed register than a cash levy, because the state must know what each district can actually produce. The answer was a new assessment built on two units: the iugum, a notional unit of land adjusted for quality and crop, so that a given area of vineyard, olive grove and arable counted differently, and the caput, a unit of labour. Liability was set per unit, and the assessment was revised on a fixed cycle, which became the fifteen-year indiction and remained a way of dating documents for centuries.
Example. In 297 the prefect of Egypt, Aristius Optatus, published an edict announcing the new census, and it survives on papyrus. It says that the emperors, having learned that the public levies were being made capriciously, so that some were let off lightly and others overburdened, have decided to fix a rate for each aroura of land according to its quality and for each head of the rural population, and it instructs everyone to pay promptly and warns officials against exaction. Read it as evidence.
Three things are visible at once. First, the reform is presented as a remedy for unfairness, and the complaint it names is real: the old system delegated assessment to local men who protected their own. Second, the units are land measured by quality and people counted by head, exactly the iugum and caput pairing, so a document from an Egyptian village confirms the structure that late Roman law codes describe from the top down. Third, and most revealing, the prefect finds it necessary to tell taxpayers to pay before they are compelled and to warn collectors off extortion, which tells you what everyone expected to happen.
The caution is that Egypt is the province whose documents survive and the province with the longest tradition of detailed registration, so it is both the best evidence and the least typical. What this edict establishes securely is that the reform was real, dated and implemented on the ground, which is more than a law code can prove on its own. What it cannot establish is that Gaul or Africa was assessed the same way or as thoroughly.
Now you. The new system taxes a piece of land together with the people who work it. What follows for those people?
Answer
They stop being able to leave. If a village's liability is fixed as so many iuga and so many capita, and the residents are jointly answerable for the total, then a departing tenant transfers his share of the burden to his neighbours, and the landlord loses the labour that makes his land assessable. Both have a direct interest in stopping him, and so does the state.
Fourth-century law duly obliges. Tenants become tied to the estates on which they are registered, in the condition later called the colonate; sons of soldiers must serve; sons of bakers must bake; and, in the sharpest reversal in this course, membership of a city council becomes hereditary and compulsory. An earlier lesson described local notables competing and paying for the honour of governing their towns. Within a century the state is legislating to stop them running away from it.
It is worth being careful about how far this went. The laws survive and were repeated, which shows both intent and non-compliance, since a rule enacted five times is a rule being broken. The late empire was not a caste society in practice, and people did move. But the direction is unmistakable: a state that needs a guaranteed physical yield from a named place has to hold people in place, and that requirement, not any theory of social order, produced the immobility of the later empire.
The price edict, and why it failed
In 301 Diocletian issued maximum prices for over a thousand goods and services throughout the empire, with death as the penalty for exceeding them. The preamble blames profiteers, and it is worth noticing what it does not blame.
The text survives in dozens of fragmentary copies, all from the Greek-speaking provinces, and it is the best price list from the ancient world. A farm labourer may be paid 25 denarii a day plus food; a modius of wheat is capped at 100 denarii; a pound of purple-dyed silk, the most expensive item listed, at 150,000.
Example. Explain the failure, using the wheat price.
At the capped prices, four days of a labourer's wage buys a modius of wheat, and a pound of purple silk costs six thousand days of it. The ratios are informative and the failure is not about ratios. A maximum price binds only where it sits below what a seller would otherwise accept, and if it does, the seller's options are to sell at a loss, sell elsewhere, sell secretly, or not sell. Wheat does not appear on the market at 100 denarii because a farmer told to accept less than the grain is worth stops bringing it in. Lactantius says exactly this: much blood was shed over trifles, goods disappeared from sale, and the law was eventually allowed to lapse.
The deeper point is in the preamble. Diocletian attributes rising prices to greed, and there is no reason to think he was being cynical, because nobody in antiquity had a theory connecting the quantity of coin to the level of prices. He had spent his reign issuing increasingly nominal money and was seeing the consequence, and he identified the visible party, the merchant raising his price, rather than the invisible one, the mint. Blaming the intermediary for a monetary phenomenon is a mistake with a very long history, and it is not fair to call it a Roman one.
Now you. Diocletian also launched the most systematic persecution of Christians the empire ever attempted, from 303. Given everything else in this lesson, why then?
Answer
Because the whole programme was an attempt to restore correctness, and the previous lesson's account of Roman religion makes that a religious project as much as a fiscal one. Diocletian was reorganising the currency, the tax, the provinces and the army on the principle that things had gone wrong because they had become disordered, and he presented himself and his colleagues as under the protection of Jupiter and Hercules. In that frame, a large and growing body of people who refused the sacrifices is not a private matter; it is a hole in the restoration.
The scale matched the ambition: churches demolished, scriptures surrendered and burned, clergy imprisoned, and in 304 a general order to sacrifice on the Decian model. It also failed, and failed visibly. Enforcement varied enormously by region, Constantius in the west did little, and in 311 Galerius, dying, issued an edict admitting that the policy had not returned Christians to the old rites and permitting them to exist and to pray for the emperor's health. That admission, from the man who had pushed hardest for the persecution, is the single clearest piece of evidence that the growth curve of the previous lesson had run too far to be reversed by force.
The one part designed to work
Diocletian's most original act was to retire. In May 305 he abdicated, compelled Maximian to do the same, and the two Caesars became Augusti with two new Caesars beneath them. The point was to convert the succession from a lethal accident into a procedure: merit, seniority, promotion and a living predecessor to supervise the handover. Nothing like it had been attempted since Augustus, and it addressed the deepest problem this course has, which is that the position was never an office.
It lasted a little over a year. Constantius died at York on 25 July 306, fourteen and a half months after the abdication, and his troops immediately proclaimed his son Constantine, who had no place in the scheme. In October, Maximian's son Maxentius was proclaimed at Rome, seventeen months and three weeks after the abdication. By 310 there were four or five men claiming imperial rank and the wars between them ran for eighteen years.
The reason for the failure is not obscure, and it is the same one the whole course keeps returning to. The tetrarchy allocated the succession by merit as judged by the senior emperor, and the armies allocated it by inheritance and presence. When Constantius died in front of his army with an adult son beside him, the army applied its own rule, and there was no institution, no law and no ceremony capable of saying that the army was wrong. Diocletian, living in his fortified retirement palace on the Dalmatian coast, was asked to come back and refused, reportedly telling Maximian that if he could see the cabbages he had planted with his own hands he would not talk to him about power.
What he had built survived the collapse of the way he had arranged to pass it on. The provinces, the dioceses, the tax in kind, the separated commands and the enlarged army were all still standing a century later. The next lesson follows the man who won the wars, kept every one of those structures, and added the one thing Diocletian would have found unimaginable: an emperor who backed the church he had tried to destroy.