The Roman Republic

The Roman Republic began by removing one man from a machine that had been built for him, and keeping every other part of it.

That is the first thing to understand about the constitution this course takes apart. It was not designed. It was a monarchy with the monarch subtracted, patched for four and a half centuries by people solving the argument in front of them, and no Roman ever read it as a document, because there was no document. This lesson assumes nothing about Rome. It sets out what the kings left behind, and, first, how much of the story can be believed at all, since that governs everything said about the next two hundred years.

Three centuries of silence

Rome dated its own founding to 753 BC and the expulsion of its last king to 509 BC, both dates fixed by scholars of the first century BC working backwards. The first Roman to write a history of Rome was Quintus Fabius Pictor, a senator who wrote in Greek around 200 BC. Between the fall of the monarchy and the first Roman account of it lie 309 years. At a conventional twenty-five years to a generation, that is a dozen generations of transmission before anyone wrote a line.

The two long narratives that survive are later still. Livy began his history in the 20s BC, 482 years after the events of his second book, and Dionysius of Halicarnassus was writing in Rome at the same time. Both had earlier annalists to draw on, and those annalists had the pontifical records, the priestly year-by-year notes of prodigies, eclipses, grain prices and magistrates. Livy himself reports that the Gauls burned the city in 390 BC and that the records went with it, which he offers as the reason the early centuries are thin.

So the working rule is not that early Roman history is fiction. It is that the further back a claim sits, the likelier it is to be a later reconstruction dressed as a fifth-century fact. Speeches are invented, as every ancient historian invented them; battles multiply; institutions get pushed back in time so as to look venerable. What survives this scepticism best is the spine: the list of annual magistrates, the fasti, kept because the Romans dated their years by it, and the physical city.

What the ground says

Archaeology is the check on the literary tradition, and on the question of kings it is unusually direct. Under the Comitium in the Forum sits the Lapis Niger, a black stone paving over an inscribed pillar of the mid sixth century BC. The Latin is archaic and broken, but the word recei, a dative of rex, is legible on it. There was a king in Rome, in the sixth century, and he was a public figure whose title appeared on a monument in the political centre of the city.

The other check is the temple of Jupiter Optimus Maximus on the Capitoline, dedicated by tradition in the Republic's first year but built by the last kings. Its podium, uncovered under the Palazzo Caffarelli, measures roughly 53 by 63 metres, a footprint of about 3,300 m². The Parthenon's stylobate is 30.9 by 69.5 metres, about 2,150 m². Sixth-century Rome, which the literary tradition treats as a modest place, laid foundations half again as large as classical Athens would later manage on its acropolis, and it did so with the labour and the organisation of a state rather than a village.

That is the shape of the evidence throughout: the tradition's stories are doubtful and its framework is often right.

Example. Livy names seven kings ruling from 753 to 509 BC. Test the list by arithmetic before testing it by argument.

Seven reigns across 244 years gives a mean of 34.9 years each. The English throne between 1066 and 1901 passed through thirty-five or thirty-six holders in 835 years, depending on how joint reigns are counted, which is 23 to 24 years apiece, and that is a hereditary succession where a son often follows a father in youth. Roman kingship was not hereditary: Livy has kings chosen, and two of them murdered. A non-hereditary throne should turn over faster than a hereditary one, not fifty per cent slower. Either the list has lost kings or the founding date is too early. The arithmetic cannot say which, and it does say the list as given is not a complete record.

Now you. Dionysius and Livy both put the first pair of consuls in 509 BC and name them. What does the existence of the consular list establish, and what does it not?

What a king was

The Roman king was not a monarch in the medieval sense, with a court and a bureaucracy. He was a commander, a judge and a priest, and his power was made of three things the Republic inherited intact.

The first was imperium: the right to command, to levy an army, to lead it, to compel obedience, and to punish. It was visible. A holder of imperium walked the city preceded by lictors, attendants carrying the fasces, a bundle of rods with an axe in it, the rods for beating and the axe for killing. Nobody had to explain what it meant.

The second was auspicium, the right to ask the gods whether a proposed action had their approval. Roman public business could not begin without it, and this was not decoration: it gave its holder the power to stop an assembly, a battle or an election by reporting an unfavourable sign, and the Republic would use that lever constantly.

The third was the machinery of consent around him. A council of elders, the senatus, advised. An assembly of citizens organised by their thirty curiae confirmed the king's power by a law, the lex curiata de imperio. Neither could initiate anything, and both would still exist under the Republic, with the same names.

Subtraction, not replacement

Now look at what the men of 509 actually did. They did not abolish imperium; they split it between two holders and made it expire after a year. They did not abolish the auspices; they handed them to the new magistrates. They did not abolish the Senate or the assembly; they kept both and enlarged the Senate's role, since a magistrate who lasts twelve months needs advisers who last for life.

The clearest proof that they understood themselves to have removed a person rather than a function is the priesthood invented to hold the leftovers. The king had religious duties only a king could perform, so the Republic created a rex sacrorum, a king for sacred purposes, who carried out the rites and was barred from every political office. The title was too dangerous to leave attached to power and too necessary to abolish, so they cut it in two.

The word for what they created was res publica, the public thing, the property or business of the people, as opposed to something belonging to one household. It is not a word for a set of institutions. It is a claim about ownership.

The fear that never went away

Because the Republic was defined against a person, its politics kept a permanent charge that no modern constitution has: aiming at kingship, adfectatio regni, lethal and needing no evidence beyond behaviour.

The tradition preserves three early cases and they follow one pattern. Spurius Cassius, three times consul, was executed in 485 BC after proposing to distribute land to the poor. Spurius Maelius, a rich grain dealer, was killed in 439 BC after selling grain cheaply in a famine. Marcus Manlius Capitolinus, credited with saving the Capitol from the Gauls, was thrown from the Tarpeian Rock in 384 BC after paying the debts of the poor. The details are exactly what annalists elaborate, but the pattern is the political vocabulary of the class that told the stories: a man who acquires a personal following among the poor is a king in waiting.

That charge was still live 465 years after the expulsion. When a crowd hailed Caesar as rex in 44 BC, he answered that he was not king but Caesar, and a few weeks later he was dead, stabbed by men who told themselves they were repeating 509.

Example. On the tradition, Spurius Maelius bought grain at his own expense in the famine of 439 BC and sold it below cost to the plebs; the Senate appointed a dictator, and Maelius was killed without trial by the dictator's master of horse. What is the episode good evidence for?

Not for Maelius's intentions, about which the tradition reports only what his enemies said, and not for the constitutional detail, which is the kind of back-formation later writers were fond of. It is good evidence for the political grammar of the aristocracy that preserved the story, and that grammar is unambiguous: private generosity to the poor, done outside the magistracies, was read as a bid for personal power, and killing the man who tried it was remembered as legitimate rather than as murder.

Now you. At the Lupercalia of 44 BC, Mark Antony twice offered Caesar a diadem, the Greek badge of kingship. Caesar twice refused it and had it carried to the temple of Jupiter, saying Jupiter alone was king of the Romans. What does the scene tell you, and what does it not?

How this course reads its sources

Two writers do most of the work in what follows, and they are unequal. Polybius, a Greek politician deported to Italy in 167 BC and taken into the household of the Scipios, wrote to explain to Greeks how Rome conquered the Mediterranean in fifty-three years. He interviewed participants, walked the ground and used Roman family archives, and for the third and second centuries BC he is a first-rate contemporary witness with a first-rate analytical mind. He is also the friend of one of the great houses, which shows.

Livy, writing under Augustus, is a different instrument. He works from earlier annalists, rarely marks where they differ, and writes to a moral scheme in which Rome rises through virtue and falls through wealth. For events before about 300 BC he preserves a tradition rather than reporting facts, and he sometimes says so. For the last century of the Republic we lose him almost entirely and rely on Appian, Plutarch, Sallust, and on the one archive nobody could tidy: the letters of Cicero, written week by week with no idea how the story ended.

The rule for the rest of this course is to ask of every claim who could have known it and how it reached us. A tax figure in Polybius, a land law quoted on bronze, a census total in the fasti and a deathbed speech in Livy are four different kinds of statement, and only the last one is free.

Example. Two statements about the fifth century BC: (a) the consular fasti record two magistrates for 509 BC, and (b) Brutus condemned his own sons to death for conspiring to restore the Tarquins, and watched the executions. Which is the stronger evidence?

The first, by a wide margin. The fasti are a working document, used to date contracts, priesthoods and family honours, kept and consulted by people with a motive to police it, and its earliest entries are constrained by the entries after them. The second is a scene: it has an audience, a moral, no possible witness who could have carried the detail through twelve generations, and a hero whose descendants were still trading on the family's tyrannicide in 44 BC. That does not make it invented; it makes it a story about what Romans valued rather than a record of what happened.

Now you. Polybius describes the constitution as he saw it working in the 150s BC, and projects the same description back onto the Rome of 216 BC. Where should you trust him, and where be careful?

Where this leaves the Republic

The men who expelled the king inherited a working state and a single problem: a king's powers had to be exercised by somebody, and nobody could be allowed to hold them the way a king had. Every institution in the next lesson is an answer to it. Two consuls rather than one, because a colleague can say no. One year rather than a lifetime, because a term ends. A Senate that could not command and could not be ignored, and assemblies that could vote yes or no but could not propose.

None of it was planned in 509, and much of it took another two centuries to appear. What was there from the start was the intention, and the next lesson takes the finished machine apart to see what the intention cost.

Having removed a king, the Romans had to give somebody his powers, and the entire constitution is a set of answers to the question of how to do that safely.

The previous lesson left the Republic with an inheritance: imperium, the auspices, a Senate and an assembly, all working, none designed for a state without a monarch. This lesson describes the machine that resulted. There was no constitutional text, no court to interpret one, and no way to amend it other than passing a law or, more often, doing something and finding that nobody stopped you.

Four constraints on one power

Roman magistrates held imperium in full. What the Republic did was surround it with four constraints, each simple and each aimed at the same target.

The first is collegiality. Every office had at least two holders, and each held the whole power rather than half of it. Two consuls did not divide imperium between them; they each had all of it, which meant either could block the other by the plain method of ordering the opposite. In the city they alternated the fasces month by month so that only one set of lictors walked in front of a consul at a time.

The second is annuality. A magistracy lasted twelve months, after which the holder was a private citizen again, liable to prosecution for what he had done in office. Nothing concentrated Roman political minds like the prospect of a year ending.

The third is provocatio, the citizen's right of appeal to the people against a magistrate's coercion. The tradition dates it to 509 and repeats it in 449 and 300, which suggests a rule that needed constant reassertion. It applied domi, at home, inside the sacred boundary of the city, the pomerium, and not militiae, in the field, where a commander could flog and execute at will. The boundary is what forced a general to lay down his imperium before entering Rome, and it is why a man who kept an army and crossed that line was not making a political error but committing a defined constitutional offence.

The fourth is the veto. Any magistrate could block an equal or lesser one, and the ten tribunes of the plebs could block almost anything, including each other. A veto needed no reason.

Add these up and the system's default is inaction. Anything that happens has to survive a colleague, ten tribunes, the calendar and the auspices, and Roman politics is mostly the story of what could be pushed through that filter.

Example. If two consuls each held the whole of imperium, why did the Republic also need a dictator?

Because the constraints that make an ordinary year safe make a bad year fatal. A blocked pair of consuls produces no decision, and when a hostile army is four days' march away, no decision is the worst available one. So the Romans built an escape hatch with its own limits: on the Senate's advice a consul named a single dictator, who held undivided imperium with twenty-four lictors, was not subject to a colleague's veto, and held office for a maximum of six months or until the task was done. The fear of one-man rule was met by time, not by division. That the office was used repeatedly in the fourth and third centuries BC and then abandoned after 202 BC is itself informative: the escape hatch was for emergencies of weeks, and the problems of the later Republic did not have that shape.

Now you. Provocatio protected a citizen inside the pomerium and not outside it. What follows for a commander who wanted to keep his army and stand for office at the same time?

The ladder

The offices were arranged in an order of increasing seniority, the cursus honorum, with minimum ages fixed by law. The lex Villia annalis of 180 BC set them, on the usual reconstruction, at 39 for the praetorship and 42 for the consulship; Sulla later put the quaestorship at 30 and made it the compulsory first step. Two years had to separate one office from the next and, in principle, ten years had to separate two consulships.

A quaestor handled money, in the treasury at Rome or as a governor's finance officer, and the quaestorship became the door into the Senate. An aedile ran the city, its markets, streets, water and games: no power and enormous visibility, since the games came partly out of his own pocket and were remembered at the next election. A praetor held imperium, presided over the courts and could command an army. A consul was head of state for twelve months and gave his name to the year. Two offices sit outside the ladder. The censors, two of them elected every five years for eighteen months, took the census, assigned every citizen to a class and a tribe, revised the roll of the Senate with power to strike a man off it for conduct, and let the great public contracts. The dictator was the emergency device described above.

Now the arithmetic, using Sulla's numbers. Twenty quaestors were elected each year, eight praetors, two consuls. Of a cohort of twenty entering the Senate, at most eight could reach the praetorship and two the consulship, so ten per cent of those who started the climb finished it, and the true figure is worse because some consulships went to men holding a second one. That is the mathematical fact behind everything else in this course. The Roman aristocracy was not a group of colleagues sharing power. It was a permanent tournament with two winners a year, in which the accepted way to distinguish yourself was to win a war, and in which a man with no prospect of the consulship might prefer to break the rules than to lose politely.

The Senate: authority without power

Read the Senate's formal position and it looks trivial. It was not elected. It passed no laws. It gave advice, and its resolution was a senatus consultum, a consultation, which a magistrate was free to ignore. It could not levy a tax, declare a war or convict a citizen.

Read what it actually did and it ran the state. It assigned provinces to magistrates, received foreign embassies, authorised spending from the treasury, decided how many legions to raise and where they went, appointed commissions, prorogued commands and set the agenda for the assemblies through the magistrates who were themselves its members. In the second century BC no important decision was taken without it.

The gap between those two paragraphs is the most instructive thing in the Roman constitution, and it has three causes.

The first is continuity. A consul had twelve months; a senator, enrolled at about thirty and serving until death, had perhaps thirty years, so the standing body outlasted every magistrate by a factor of around thirty to one. The men who knew what had been agreed with Massilia in 154 BC were sitting in the chamber, and the consul was not.

The second is money. The treasury sat in the temple of Saturn under quaestors, and the convention that only the Senate authorised withdrawals meant a magistrate with an idea and no money had an idea and nothing else.

The third is that the Senate was the pool the magistrates came from and returned to. A consul spent one year above the Senate and the rest of his life inside it, judged by the men he had overruled. Ignoring a senatus consultum was legal and expensive.

Example. After Sulla, twenty quaestors entered the Senate each year and the chamber held about six hundred members. What does that imply about how long a senator served, and does it hang together?

If the size is steady, entries must match departures, so 600 divided by 20 gives a mean tenure of 30 years. A man entering at the legal minimum of 30 would then be leaving at about 60, by death or by a censor's mark. That is consistent with ancient life expectancy for a man who has already survived to thirty, so the numbers hang together. Try it with the pre-Sullan figures and it does not: 300 senators fed by eight quaestors a year implies 37.5 years of service, ending at 67, which is too long. The mismatch is a real one, and it is the reason the pre-Sullan censors also enrolled ex-tribunes and ex-aediles, and why the chamber often sat below its notional strength.

Now you. A tribune proposes a law the Senate hates. The Senate passes a resolution against it. The tribune puts it to the assembly anyway and it passes. Is it law?

The assemblies, and what they could not do

Every law and every magistrate came from a vote of the citizens, which sounds like a democracy until you look at what a Roman assembly was allowed to do.

It could not debate. It could not amend. It could not meet on its own initiative. A magistrate summoned it, put a single question, and it answered yes or no, having heard whatever speakers that magistrate chose to allow at the informal meeting beforehand. There were three of them in practice: the comitia centuriata, organised by property and originally by military class, which elected consuls, praetors and censors and formally declared war; the comitia tributa, organised by the thirty-five tribes, which elected the lesser magistrates and passed laws; and the concilium plebis, the plebeians alone by tribe, which elected the tribunes and passed the plebiscites that after 287 BC were law.

None of them voted as individuals in a single pool, and the consequences of that are so large that the next lesson but one is devoted to them.

Polybius on the mixture

The best ancient analysis is Polybius's sixth book, written in the 150s BC for Greek readers who wanted to know why Rome kept winning. His answer is that Rome had accidentally achieved what Greek theory prescribed: a mixed constitution, monarchic in the consuls, aristocratic in the Senate, democratic in the people, each element able to obstruct the others and therefore obliged to co-operate.

He argues it concretely, which is what makes it good evidence. A consul commands in the field, but his supplies and his soldiers' pay are voted by the Senate, and after his year another man can replace him. The Senate controls the money, and the great censorial contracts for harbours, mines and taxes are held by companies of citizens who sit in the assemblies. The people elect and legislate, and depend on the Senate for the wars and settlements that shape their lives.

Two cautions. Polybius describes the system at the moment of its greatest success, and a theory of balance looks convincing when the state it describes has just beaten everyone. He is also a client of the Scipios, disposed to see aristocratic guidance where a hostile witness would see aristocratic control.

Example. Polybius says the people are sovereign because they elect magistrates, pass laws and try capital cases. Test the claim against the machinery just described.

The powers are real: no man held office without a vote, and no law passed without one. The limits are equally real. The people could not propose, amend or debate a bill, and could not meet unless a magistrate called them; the choice at an election was between candidates the presiding magistrate had accepted, drawn almost entirely from families already in the Senate; and the agenda was set by men whose careers depended on senatorial goodwill. Sovereignty of that shape is a veto over the aristocracy's proposals rather than a capacity to govern, which is a genuine power and not the one the word usually implies.

Now you. Polybius wrote that no element could act without the others, and inferred stability. What later event tests the inference hardest?

What the machine could not do

Notice what the design assumes. Every magistrate is in Rome or a few days from it. Every command is short. Every officeholder returns to the body he came from. Every dispute is settled by men who expect to see each other next year.

None of that would be true of the state Rome became, and the rest of this course is the story of the assumptions failing one by one. But the first pressure on the machine came from inside the city and much earlier, from citizens who noticed that all these carefully divided offices were held by the same fifty families, and who spent two centuries prising them open. That is the next lesson.

The offices described in the previous lesson were, at the start, the property of a few dozen families, and the next two hundred years of Roman history are about how that changed.

The two groups have names that sound like classes and are not. Patricians were a closed hereditary caste, perhaps fifty clans, who claimed a monopoly of the priesthoods and of the auspices and therefore, they argued, of the magistracies that required them. Plebeians were everybody else: the poor, but also rich landowners, traders and soldiers who were excluded from office by birth alone. The quarrel is not rich against poor. It is a hereditary caste against everyone outside it, and the people who led the fight from the plebeian side were mostly wealthy men who wanted the consulship.

Why the plebs had leverage

A group with no offices, no priesthoods and no legal standing had one asset: it was the army. Rome fought almost every year, its soldiers were its property-owning citizens, and the great majority of them were plebeians. A state that needs several thousand infantry every spring cannot afford to have them decline to muster.

That is the mechanism behind the tactic the tradition calls secessio, secession. On the account preserved by Livy, in 494 BC the plebs withdrew from the city to a hill outside it and refused to serve until their grievances were met. The device was used repeatedly across the next two centuries and it worked every time, which is what you would expect of a general strike by the only people who can fight.

The immediate outcome was the tribunate of the plebs. What matters is how it was protected. The tribune's person was made sacrosanct not by a statute of the Roman people but by an oath sworn among the plebs themselves: whoever harmed a tribune was forfeit to the gods, and any plebeian might kill him. This is not a power the state granted. It is a threat the state accommodated, and it was the foundation of an office that could stop the state's business. From it came two tools that shaped the rest of Roman history: auxilium, the right to interpose the tribune's body between a magistrate and a citizen, and intercessio, the veto over any act of any magistrate, including a colleague.

Example. Why did withdrawing from the city work when a petition would not have?

Because it attacked the one thing the patricians could not supply for themselves. Rome's wars were annual and its army was a levy of property-owning citizens, so a patrician consul who could not fill the ranks had no army, no campaign, no triumph and no defence against the neighbours. The plebs did not need to win an argument; they needed only to be absent at the moment their absence cost most. That also explains the shape of the concessions. Each secession bought an institution (a tribunate, a code, a law about plebiscites) rather than money or land, because institutions were what the withdrawal could be traded for and because the men leading the plebs wanted access rather than relief.

Now you. The tribune's sacrosanctity rested on a private oath among the plebs rather than on a law of the whole people. What does that tell you about the early Republic?

Putting the law in writing

The second demand was publication. The law was known to the priests and applied by patrician magistrates, and a plebeian entering a lawsuit could not read the rule he was being judged by.

In 451 and 450 BC, on the tradition, a board of ten men suspended the ordinary magistracies and produced the Twelve Tables, posted in the Forum. Livy called them the source of all public and private law. The surviving fragments, quoted piecemeal by later writers, are not a constitution and not a bill of rights. They are mostly procedure and property: how to summon a man to court, what a creditor may do, how much land a widow keeps, what happens if a beam is built into another man's house.

They are also harsh in exactly the places where the poor were vulnerable. A debtor had thirty days to pay, then was handed to his creditor, who might bind him in chains of not more than fifteen pounds' weight, about 4.9 kg, and after sixty days sell him abroad. One table forbade marriage between patricians and plebeians, which was repealed within five years by the lex Canuleia of 445 BC.

The gain was not that the law became fair. It was that it became fixed, public and citable, so an argument about what the rule said could be settled by looking. Every later Roman claim about legality, including the ones that mattered in 133 and 49 BC, depends on that.

Two centuries of laws

The rest of the struggle is a sequence of statutes, and the sequence is the most trustworthy part of the whole tradition, because the consular lists record when plebeian names begin to appear in each office.

The turning point is the leges Liciniae Sextiae of 367 BC, which required that one consul each year be a plebeian. After that the doors opened one by one: the dictatorship in 356, the censorship in 351, the praetorship in 337, and in 300 the lex Ogulnia opened the great priestly colleges of pontiffs and augurs, which mattered because the priesthoods carried the auspices the patricians had claimed as their reason for excluding everyone else. Debt bondage, the mechanism that had turned poor citizens into their creditors' property, was abolished by the lex Poetelia Papiria in 326 BC.

The last act was constitutional. After a final secession, the lex Hortensia of 287 BC made resolutions of the plebeian assembly binding on the whole Roman people, patricians included. From 494 to 287 is 207 years, and at the end of it a body originally created as a defensive club for the excluded could legislate for the state.

Example. The 367 law required one plebeian consul each year. When did Rome first elect two plebeian consuls, and what does the answer tell you?

Not until 172 BC, 195 years later. The rule that looks like a floor operated as a ceiling for nearly two centuries: one place was reserved for a plebeian, and one was effectively reserved for a patrician, and both sides found the arrangement comfortable. This is the standard shape of the settlement. Formal access was conceded, and the group that held it converted the concession into a share. It is also a warning about reading Roman legislation as a description of practice, since the statute and the fasti disagree for two hundred years and only the fasti record what happened.

Now you. The 367 law is also said to have capped holdings of public land at 500 iugera, the same limit Tiberius Gracchus proposed in 133 BC. How should that be weighed?

What the settlement actually produced

The natural expectation is that opening the offices to the majority produces a broader politics. It did not. It produced a new ruling group, the nobiles, the families of either order whose members had held the consulship, and that group was as narrow as the patriciate had been and considerably more capable, since it now contained the ablest plebeian families as well.

The evidence is in the consular lists. Across the second century BC the consulship circulated among a few dozen families, and a man from outside them, a novus homo, reached it rarely enough to be remarkable. Cicero, consul in 63 BC, was the first new man to reach the office since Gaius Coelius Caldus in 94, a gap of 31 years in which some sixty consulships went to men whose families already held one.

So the struggle of the orders ends in a compromise that is easy to misread. The plebs won legal equality, the abolition of debt slavery, a code, and the theoretical right of any citizen to hold any office. What they did not win was any change in who governed.

The tribunate absorbed

The clearest case of that absorption is the tribunate itself. An office invented to protect the plebs against magistrates became a normal step in a senator's career, held by young men from consular families on their way up, and its central weapon turned out to be more useful to the Senate than against it. Ten tribunes with an unconditional veto means that stopping a measure requires the Senate to persuade one man out of ten, while passing one against senatorial opposition requires persuading all ten not to interfere.

That asymmetry is why the tribunate is quiet for roughly a century and a half after 287 BC, and why the histories of that period mention tribunes mainly as obstructions. When the office wakes up in 133 BC it does so in the hands of a man whose father had been consul twice and whose brother-in-law was the conqueror of Carthage, using the plebs' oldest instrument against the class that had spent two hundred years domesticating it.

Example. The consular list for the first fifty years of the Republic contains names, among them Junius, Cassius and Genucius, that later belong to famous plebeian families. What might that mean?

There are two live readings. On the first, the patriciate was not yet closed in the fifth century, and it hardened later, so the early consuls were simply the leading men of the city and the exclusive caste was created afterwards and then read back into the record. On the second, those families did hold the consulship as patricians and later branches passed into the plebs, an idea the Romans themselves used to explain awkward names. The first reading is the more economical and is widely held, and either way the case is a good reminder that a document as solid as the fasti is not self-interpreting: it gives names, and every claim about what a name meant is an argument built on top of it.

Now you. Debt bondage was abolished in 326 BC. What does its abolition imply about where Rome's future labour would come from?

Where this leaves the Republic

By 287 BC the constitution described in the previous lesson had reached something like its final shape, and it now belonged to a wider class than before. It had also acquired the feature that decided its future: a career structure in which perhaps a dozen serious contenders each year competed for two consulships, in which the recognised route to the top was command in war, and in which the assemblies that awarded the offices were open to every citizen.

That last point is the one worth pressing. If every citizen votes, why did the same families keep winning? The answer is in the machinery of voting itself, and it is the subject of the next lesson.

Every Roman magistrate was elected by the citizens, and the same few dozen families held the consulship for centuries, so the interesting question is how both of those things were true at once.

The previous lesson ended with a puzzle. After 287 BC any citizen could in principle hold any office, and the assemblies that conferred those offices were open to every citizen. Yet the fasti show a narrow and hereditary consulship. Part of the answer is money and prestige, which are true of every aristocracy. The specifically Roman part is in the mechanics of voting, and the mechanics are unusual enough that they have to be set out before anything about Roman politics makes sense.

Nobody voted as an individual

A Roman never cast a vote that was counted alongside other citizens' votes. He voted inside a group, the group produced one verdict by an internal majority, and the groups were counted. Whether your vote mattered depended entirely on which group you were in and how large it was.

The comitia centuriata, which elected consuls, praetors and censors, sorted citizens into 193 centuries by wealth. The richest were counted first, and the count stopped as soon as one side had a majority, because the outcome was then settled and further voting was pointless.

Now do the arithmetic on the older scheme, which Livy describes. Eighteen centuries of equites, the wealthiest men of all, voted first, followed by the eighty centuries of the first property class. That is 98 centuries out of 193, and a majority is 97. If the rich agreed, the election was decided before the second class was called, and the fifth class, the smallest holders, whose 30 centuries contained by far the largest number of human beings, could vote only when the rich were split.

A reform in the third century BC tied the centuries to the tribes and cut the first class to 70 centuries. With the 18 equestrian centuries that gives 88, nine short of 97, so the top of the system now had to carry the second class as well. The details of the reformed scheme are disputed and the direction is not: the change made the richest citizens negotiate with the next group down, and it left the poorest exactly where they were.

Example. Under the older scheme, what did the arrangement mean for a citizen in the fifth property class?

That his vote existed and was almost never reached. He belonged to one of thirty centuries called after 98 others, in a count that stopped at 97, so he was polled only when the wealthy divided. The system was not a fraud, because when the rich did split, the lower classes decided the outcome, and Roman politicians behaved as though they might. But its normal operation was a weighted ballot in which about half of the voting units belonged to a small minority of citizens, and the weighting was explicitly by property, since the classes were originally the ranks in which men were expected to equip themselves for war.

Now you. After the reform the first class held 70 centuries and the equites 18. How many more centuries did they need for a majority of 193, and where did those have to come from?

The tribes, and where the city was put

Laws and the lesser magistracies went through the assemblies organised by tribe, of which there were 35: four urban and thirty-one rural. A tribe was not a place of residence but a registration, inherited from a father, and each tribe cast one vote decided by a majority of whichever of its members turned up.

Two consequences follow, and they run in opposite directions.

First, the city's poor were penned. Freedmen and the propertyless of Rome were registered in the four urban tribes, an arrangement fought over repeatedly in the fourth and third centuries BC and settled by 220 BC. However many of them there were, they controlled 4 votes out of 35, which is 11.4 per cent, against the 18 needed for a majority.

Second, and less often noticed, the rural tribes were cheap. Voting happened only in Rome, on days announced in advance, with no absentee ballot. A rural tribe's verdict was decided by those of its members who were standing in the enclosure that day, which might be a few dozen men, and a great house with clients registered in that tribe could deliver it at low cost. Both effects push the same way: the assembly was a machine that converted a small, organised, well-connected turnout into decisions.

Who was actually in the enclosure

The Roman citizen body was enormous and the space was not. The census of 70 BC registered about 910,000 citizens. Voting took place in the Forum or on the Campus Martius, and the enclosures used for it could hold, on the most generous estimates, some tens of thousands of men, so even a full and orderly assembly would have contained around 3 per cent of the electorate, and the ordinary attendance was far below that.

Distance did the rest. Roman citizenship by the first century BC extended the length of Italy, and a citizen registered near Mediolanum was about 500 km from the voting enclosure, which at a marching pace of 30 km a day is nearly seventeen days each way. A man walking for a month to cast one thirty-fifth of a vote is not a normal event, and it did not normally happen.

This has produced a genuine scholarly quarrel. Fergus Millar argued that the assemblies were real political theatre, where speeches changed outcomes and the crowd's verdict was sovereign in fact as well as in law. Henrik Mouritsen replied that the physical spaces were too small and the electorate too dispersed for anything but a tiny, largely urban, and largely mobilised sample to be present. The evidence supports a compromise that is uncomfortable for both: the votes were real and could go against the Senate, and the body casting them was a small fraction of the citizens, assembled by the people who had reasons to assemble it.

The canvass

What did a candidate actually do? A remarkable document survives to answer that: the Commentariolum Petitionis, a handbook on electioneering addressed to Cicero for his consular campaign of 64 BC and attributed to his brother Quintus. It is practical to the point of cynicism, and it is the best evidence we have for what winning required.

Its advice is to be seen constantly with a crowd, to learn and use names (a slave, the nomenclator, was kept to whisper them), to attach the men who control blocks of votes rather than the voters themselves, to remind everyone of favours done, and above all to promise. A refusal offends immediately and certainly; a promise may never have to be honoured, and if it is broken the injury falls later and on one man.

Behind the handbook lies the structure that made it work: clientela, the standing relation of obligation between a great man and the many lesser men who attended his morning reception, took his advice in lawsuits, and were expected to support him. Nothing about this was secret or shameful, and no Roman thought it corrupt.

Outright bribery was different, illegal and constant. The law was tightened again and again, in 67 and 63 BC among others, which is the usual sign of a rule that is not working.

Example. Cicero wrote to his brother in July 54 BC that the rate of interest at Rome had gone from 4 to 8 per cent because of the money being taken up for the consular elections. What does the remark establish?

That electoral bribery was large enough to move the price of credit in the largest financial market in the Mediterranean, and that it happened in a season, since Cicero reports a rate doubling in the weeks before a poll. It is also the right kind of evidence: an aside in a private letter to his brother, written for information rather than persuasion, with no reason to exaggerate and no audience to impress. A speech complaining about bribery would be worth much less, since complaint about bribery was a standard rhetorical move. Note what the figure does not tell us: how many voters were bought, or at what price, or whether the buyers got value.

Now you. The handbook advises Cicero to promise everything to everybody, on the grounds that a refusal is a certain injury and a broken promise a possible one. What does that advice tell you about the electorate?

What the voters could and could not decide

It would be wrong to end with the machinery alone, because the vote was not a formality and Roman politicians did not treat it as one.

Candidates lost. Assemblies threw out bills the Senate wanted and passed bills it hated, including the great commands of the 60s BC. A tribune with a popular measure could and did go over the Senate's head, and the fact that the Senate's usual response was to find another tribune to veto rather than to ignore the vote tells you the vote counted.

What the citizens could not do was set the menu. The choice at a consular election was among men of the same class, nearly all of them from families already in the Senate, presented by a presiding magistrate who could refuse a candidature. The bills came from magistrates. Debate happened at informal meetings, the contiones, where the presiding magistrate chose who spoke.

So the honest summary is that Rome had real elections without a real alternative, and the effect on the aristocracy was to make it competitive rather than to make it accountable. Since the reliable way to be remembered by voters was a war won, a triumph paraded, and games paid for out of the proceeds, the electoral system was a pump pushing ambitious men outward, towards commands, provinces and conquests.

Example. The four urban tribes held 4 votes out of 35 in the tribal assembly. Compute the share and say what it implies for a politician appealing to the city's poor.

Four of thirty-five is 11.4 per cent, against the 18 tribes needed to carry a vote. So a politician with the undivided support of the urban plebs, and of the freedmen registered with them, starts with roughly an eighth of what he needs and cannot get to a majority from that base alone. He must therefore either win rural tribes, which means the landowners whose clients fill them, or bring rural voters into the city, which means paying for their journey, or wait for a moment when the countryside is already in Rome. That is exactly the pattern of the popular politics of the late Republic, which peaks at festivals, in years of crisis, and around men rich enough to move people.

Now you. In 133 BC Tiberius Gracchus put a land bill to the tribal assembly, and the tradition says supporters came into Rome from the countryside to vote for it. Why is that detail worth noticing?

Where this leaves the Republic

Put the last three lessons together. Rome had a constitution that made action difficult, a career ladder with two winners a year, and an electorate that rewarded military distinction and was reached through networks of obligation. Every incentive in that system pointed the ambitious outward, towards a command against somebody.

Conveniently, Italy was full of somebodies. The next lesson follows the ambition out of the city and asks how Rome, having beaten its neighbours, did something no other ancient city did with the defeated: it turned them into more Romans, and thereby built the manpower that would fight Carthage.

Between 340 and 264 BC Rome beat every people in peninsular Italy, and what it did with them afterwards is the single most consequential decision in its history.

The previous lesson left a city whose political system pushed ambitious men towards war. This one follows them out of the gates. The military narrative matters less than the settlement that followed each victory, because the settlement is what turned conquest into strength rather than into a permanent policing problem. Almost every ancient state that won a war took land, tribute and slaves. Rome took soldiers.

Beating the neighbours

The bare sequence is quickly told. The Latin cities, Rome's own kin and long-standing allies, revolted and were defeated in the Latin War of 340 to 338 BC. Three wars against the Samnites, the tough highland confederation of the central Apennines, ran from 343 to 290 with a long middle phase that included Rome's worst humiliation, the surrender at the Caudine Forks in 321. A coalition of Samnites, Etruscans, Umbrians and Gauls was broken at Sentinum in 295. Finally Tarentum, a Greek city on the southern coast, called in King Pyrrhus of Epirus, who beat Roman armies at Heraclea in 280 and Ausculum in 279 at a cost that gave us the phrase for a victory not worth having, and lost at Beneventum in 275. Tarentum fell in 272.

Two things in that list are worth pausing on. Rome lost repeatedly, at the Caudine Forks, at Heraclea, at Ausculum, and kept fighting, which is a fact about its political system as much as its army. And Pyrrhus, a professional Hellenistic king with the best army in the western Mediterranean, beat the Romans twice and still could not finish them, because they came back the following year with fresh men.

The settlement of 338

The Latin War ended with a decision that had no obvious precedent. Rome dissolved the Latin League and refused to replace it with a new one. Instead it made a separate arrangement with every community, and forbade them to deal with each other. The arrangements came in grades.

Some towns received full Roman citizenship, with the vote and every legal right, and simply stopped being independent states. Others received citizenship without the vote, civitas sine suffragio: their people were Romans for private law, marriage, contract and inheritance, they owed military service, and they had no say in Roman elections. Latin status covered old Latin towns and the new Latin colonies, and carried the right to trade and marry with Romans and to become a Roman citizen by moving to Rome. Everyone else became an ally, a socius, bound to Rome by a treaty of its own, keeping local self-government and its own laws, owing troops and nothing else: no tribute in money, and no Roman governor.

Look at what that list does. It offers something at every level, from full membership down to a treaty that leaves a town running its own affairs, and it makes advancement possible, since Latin status leads to citizenship. It gives the local elites a reason to prefer Rome to their neighbours. And it converts each defeat into an obligation to send soldiers, so that every war produced the means to fight the next one.

Example. Why was citizenship without the vote attractive to Rome, rather than an obvious insult?

Because it solved both halves of the problem at once. Rome wanted manpower and did not want the assemblies filling with men it could not manage, and this status delivers the first without the second: full obligation to serve, full protection in private law, no political weight. For the receiving community it was not nothing, since Roman private law and the protection of Roman power had real value to a landowning class with local rivals. The design is legible for what it is, a graded membership with duties at the bottom and rights at the top, and it also contains the germ of the Social War, because a status that is tolerable in 338 for towns near Rome becomes intolerable in 91 BC for half of Italy after two centuries of fighting Rome's wars.

Now you. The allies owed troops and paid no tribute. What does the absence of tribute imply about how Rome's Italian system was financed?

Colonies and roads

The settlements were held in place by two physical instruments.

Colonies were new towns planted on confiscated land at chosen points, populated by Roman and Latin settlers who received plots. They were garrisons that fed themselves, and they were substantial: Livy and Dionysius give settler numbers such as 2,500 for Cales in 334 BC, 6,000 for Alba Fucens in 303, and 20,000 for Venusia in 291, planted in the middle of Samnite country. By 218 BC around thirty Latin colonies formed a net across the peninsula, each one a piece of Rome inside somebody else's territory, with its own land, walls and loyalty.

Roads connected them. The Via Appia, begun by the censor Appius Claudius in 312 BC, ran from Rome to Capua, 132 Roman miles, about 195 km, and was later pushed to Brundisium on the heel of Italy. Roman roads were not built for trade. They were built so that an army could move to a colony in days rather than weeks, in any weather, and their construction is dated by the wars they served.

Counting the manpower

The result can be measured, and Polybius measured it. Faced with a Gallic invasion in 225 BC, the Senate called for returns of men able to bear arms from Rome and every ally, and Polybius reproduces the lists: Romans and Campanians at about 250,000 foot and 23,000 horse, and a grand total for Rome and its Italian allies of over 700,000 foot and 70,000 horse.

Take the figure carefully. It is not an army; it is a register of men of military age who could in principle be called. Modern work, above all Peter Brunt's Italian Manpower, has argued over what the numbers counted and concluded that they are of roughly the right order while being neither a census nor a reliable total. Even discounted, the order of magnitude is the point. Athens at its height had something like 40,000 adult male citizens. Rome's citizen body alone was several times that, and the Italian pool behind it was of the order of eighteen times the Athenian figure.

That is the resource that the rest of this course keeps returning to. It is why a state can lose three armies to Hannibal and field a fourth, and it is why the Hellenistic kingdoms, whose greatest set-piece battles put perhaps 70,000 men on a side after months of preparation, lost to a city that could put comparable numbers into a single day's fighting and replace them.

Example. Polybius gives a legion as 4,200 foot and 300 horse, and says the allies supplied infantry equal to the Roman infantry and cavalry at three times the Roman number. Work out a standard consular army of two legions and its allied share.

Roman foot: $2\times 4200=8400$. Allied foot equal to it: 8,400. Roman horse: $2\times 300=600$. Allied horse at three times that: 1,800. The whole force is $8400+8400+600+1800=19\{,\}200$ men, of whom 10,200 are allies, which is 53 per cent. So more than half of a normal Roman army was not Roman, and three-quarters of its cavalry was not Roman. Hold on to that figure: it is the arithmetic that makes the Italian allies' demand for citizenship in 91 BC something other than a request.

Now you. At Cannae in 216 BC, Polybius says Rome fielded eight legions at strength, with allied contingents. Estimate the infantry, and say what the estimate assumes.

What the system cost the allies

It is easy to present all this as a story of Roman generosity, and Roman writers did. The allies' side of the ledger is worth stating plainly.

An ally supplied and paid for troops for as long as Rome wanted them, in wars it had no part in choosing, and after 200 BC those wars were increasingly overseas and long. It received a share of booty, but the general who divided it was Roman. It had no vote in the assemblies that declared the wars. Its magistrates could be humiliated by a passing Roman official, and the surviving complaints from the second century BC about the treatment of Italian notables by Roman magistrates are bitter and specific.

Meanwhile the rewards of the empire flowed disproportionately to Rome: land distributions to Roman citizens, the grain distributions later, the tax immunity of 167 BC. Allies who had fought in the same battle line watched the citizen half of the army receive what the allied half did not.

Example. Rome's system is often praised as an unprecedented generosity in extending citizenship. What is the strongest evidence against reading it that way?

The Social War of 91 to 88 BC, in which most of Italy took up arms specifically to obtain the citizenship, and which Rome ended by granting it. If the system had been generous in the way the compliment implies, the allies who had fought Hannibal, Philip and Antiochus alongside Roman legions for two centuries would not have needed to kill 100,000 people to be enrolled. The right formulation is that the settlement was generous by the standards of ancient imperialism, in that it offered a route in at all, and that it was designed for Roman advantage, and that it stopped expanding once Rome had all the manpower it required.

Now you. Consider two policies towards a defeated city: enslave the population and take the land, or leave it self-governing and require troops. On what does the choice depend, and which did Rome use?

Where this leaves the Republic

By 264 BC Rome controlled Italy south of the Po with an army it could rebuild after any single defeat, a road network to move it, and a class of ambitious men who needed wars to fight. It had no fleet, no overseas possessions and no institutions for governing anything outside Italy.

Directly across the strait from the toe of Italy lay Sicily, half of it under the influence of Carthage, the greatest naval power in the western Mediterranean and a state organised on the opposite principle: money, mercenaries and ships. The collision, and what it did to both of them, is the next lesson.

In 264 BC Rome, which had no navy, went to war with the greatest naval power in the western Mediterranean, and won by refusing to stop.

The previous lesson left Rome in control of Italy with an army half made of allies and a political system that rewarded commanders. This lesson is about what happened when that system met a state built on the opposite principle. Carthage, a Phoenician foundation on the coast of modern Tunisia, was governed by an oligarchy of merchant families, drew its wealth from trade, and fought with a professional navy and hired soldiers. Rome fought with its own farmers and had never built a warship.

Why the war began

The cause was small and the sources are frank about it. A band of Campanian mercenaries called the Mamertines had seized the Sicilian city of Messana, and when they came under pressure they appealed to both Carthage and Rome. Accepting meant protecting men who had done to a Greek city exactly what Rome had recently punished a garrison of its own for doing to Rhegium.

Polybius says the Senate could not bring itself to decide, and that the matter went to the people, who voted for war because the consuls held out the prospect of plunder. That is a striking claim from a writer sympathetic to the Roman aristocracy: the greatest war of the age was begun by an assembly vote against senatorial hesitation, on a straightforwardly acquisitive argument. Historians argue about whether Polybius is reporting or moralising. Either way, the mechanism he describes is the one the previous lessons built: a system in which commands are prizes and the people who award them expect a share.

Learning to build a fleet

Sicily is an island, and after a few years of marching about in it the Romans found that they could not win without controlling the sea. Polybius tells the story of what they did: a Carthaginian quinquereme ran aground, the Romans copied it, and while the shipwrights worked, the crews were trained on benches set up on dry land. He says a hundred quinqueremes and twenty triremes were built in sixty days from the felling of the timber.

Take the sixty days as a boast and the substance as true. The interesting number is not the building time but the crews. Polybius gives a quinquereme a complement of 300 rowers and 120 marines, so 420 men. A fleet of 100 such ships needs 42,000 men, which is more than two consular armies of the kind counted in the previous lesson, and they must be found, fed and paid every year that the fleet is at sea.

This is where Rome's Italian settlement paid for itself in a way nobody had planned. The rowers came from the citizen poor, the allies, and above all the Greek coastal cities of southern Italy, which had ships and seamen and were bound to Rome by treaty. A city of farmers could put a fleet to sea because it had spent the previous century acquiring maritime allies.

Example. Polybius says that at Ecnomus in 256 BC the Romans had 330 decked ships and the Carthaginians about 350, and that the total number of men engaged was close to 290,000. Check whether his figures agree with each other.

Using his own complement of 420 men to a quinquereme: the Roman fleet carries $330\times 420=138\{,\}600$ and the Carthaginian $350\times 420=147\{,\}000$, for a total of 285,600. Polybius's round figure of about 290,000 sits within two per cent of what his own ship numbers imply, so the account is at least internally consistent, which is more than can be said for many ancient battle numbers. It does not prove the ship counts are right, and it does show they were not invented independently of the manpower claim. If those figures are near the truth, Ecnomus involved more men than any sea battle before or since.

Now you. Polybius says the first fleet of 120 ships was built in sixty days from standing timber. Is the claim usable?

Winning by attrition

The war ran from 264 to 241 BC, twenty-three years, and Polybius calls it the longest and most severely contested war he knows of. Its shape is a Roman refusal to accept the verdict of any single event.

Rome won its first sea battle at Mylae in 260 with a boarding bridge, the corvus, that let its infantry fight on enemy decks and turned a naval action into the kind of fight it was good at. It invaded Africa in 256, and in 255 the invasion was destroyed at Tunis by a Spartan mercenary commander, Xanthippus, with the consul Regulus captured. The fleet that came to evacuate the survivors was caught in a storm off Camarina: Polybius says 364 ships sailed and 80 survived, so 284 were lost, and at 420 men a ship the drowned would number about 119,000 men. He calls it the greatest disaster at sea in history, and on his own numbers it is hard to argue.

Rome built another fleet, and lost most of it to a second storm in 253. It built another, and at Drepana in 249 lost 93 ships out of 123, the one Roman naval defeat in a fair fight. It then had no money.

Across the war Polybius reckons Rome lost about 700 ships and Carthage about 500. At full complement 700 ships is nearly 300,000 places, and while many crews were saved from wrecks, the losses in men were on a scale no Hellenistic state could have replaced. Rome replaced them repeatedly.

Paying for the last fleet

The end of the war is the most revealing episode in it. By 243 BC the treasury was empty and the Senate would not levy again on a population that had been paying for two decades. So the fleet that finally won was funded by private subscription: leading citizens, singly or in groups of two and three, paid for a quinquereme each on the understanding that they would be repaid if the war was won.

Two hundred ships were built to a better design, copied from a captured blockade runner, and manned by crews that had time to train. In March 241 they destroyed a Carthaginian relief fleet off the Aegates Islands. The Carthaginian army in Sicily, which could not now be supplied, was left where it stood, and Carthage asked for terms.

Example. Why did Rome win? Isolate the decisive factor rather than listing causes.

Not tactics: the corvus helped early and was abandoned, probably because it made ships unstable in weather. Not seamanship, which stayed inferior to the end. Not wealth, since Carthage was richer and Rome ended the war borrowing from its own citizens. The decisive factor is the capacity to absorb catastrophe: four fleets lost and a fifth built, an African invasion destroyed with the war continued anyway. That capacity has two components, the manpower pool counted in the previous lesson and a political system in which no single family's prestige depended on admitting defeat, since the consuls changed every year and the Senate that made policy was permanent. Carthage, whose commanders were held personally responsible and whose soldiers had to be paid in coin, could not match either.

Now you. The winning fleet of 242 BC was paid for by private citizens on the promise of repayment after victory. What does that tell you about the Roman state?

A province, and how to lose a peace

The treaty of 241 BC required Carthage to leave Sicily and to pay 3,200 talents, a thousand at once and the rest over ten years. In silver that is about 83 tonnes, and in Roman money about 77 million sesterces, a sum that fixed Carthage's finances as a problem for a decade.

Sicily then presented Rome with something it had never had: territory outside Italy, full of Greek cities that were not going to be turned into Italian allies. The solution was improvised and became permanent. Sicily was made a provincia, a sphere of duty assigned to a magistrate, and the Romans kept the tax system already operating there under Hiero of Syracuse, a tithe of the grain harvest. So the first province was taxed rather than levied for troops, and it was governed by a Roman magistrate sent out for a year. Both features are the opposite of the Italian settlement, and both would define the empire.

What followed is the part Polybius could not forgive. Carthage, unable to pay its returning mercenaries, fell into a savage revolt at home. In 238 BC, with Carthage fighting for its life, Rome annexed Sardinia and Corsica and demanded a further 1,200 talents to avoid a new war. Polybius, no friend of Carthage, calls this action contrary to every justice and names it as the greatest single cause of the war that followed.

The constitutional consequence arrived in 227 BC: the number of praetors was raised from two to four so that Sicily and Sardinia could have governors. That is the city-state machine of annual magistrates growing a new limb, and it is the first sign that a constitution built for one city would have to be stretched to fit an empire.

Example. Polybius says the seizure of Sardinia in 238 BC was unjust and was a cause of the next war. He is a Greek writing under Roman patronage. How much weight should the judgment carry?

More than usual, precisely because of who he is. Polybius wrote for a Roman readership, lived in a Roman house, and generally defends Roman conduct, so a passage condemning Rome runs against his interest and his habit. He also has the analytical apparatus to mean something specific: he distinguishes the causes of a war from its pretexts, and puts Sardinia in the first category, along with Carthaginian anger and the growth of Carthaginian power in Spain. The judgment is a considered one by a well-placed writer arguing against type, which is about as strong as an ancient verdict gets.

Now you. Rome levied troops from its Italian allies and taxed its Sicilian subjects. Why did the two systems not converge?

Where this leaves the Republic

Rome ended the war with Sicily, Sardinia, a navy it did not want to maintain, a governing class enriched by twenty-three years of contracts and plunder, and a defeated rival that had lost its islands but not its army, its trade or its best general's family.

Hamilcar Barca took that army to Spain and spent the next twenty years building a territorial empire with silver mines and Iberian recruits, out of reach of the Roman fleet. His son Hannibal was nine years old when he went with him. The next lesson is what happened when he came back.

Between 218 and 216 BC Hannibal destroyed three Roman armies in Italy, and Rome did not sue for peace, which is the fact the whole of the rest of this course depends on.

The previous lesson ended with Carthage rebuilding in Spain under the Barcid family, out of reach of the Roman fleet and paying for itself with Iberian silver and Iberian recruits. Hannibal, who took over that command in 221 BC, understood what the first war had taught: Rome could not be beaten at sea and could not be beaten by attrition, so it would have to be beaten politically, by detaching the Italians who supplied half its army. Everything he did in Italy for fifteen years follows from that plan, and the plan is the reason his victories did not win the war.

The invasion

The immediate trigger was Saguntum, a city in Spain under Roman protection, which Hannibal besieged and took in 219 BC. Rome declared war, expecting to fight in Spain and Africa. Hannibal marched to Italy instead, over the Alps, arriving in the autumn of 218.

The march cost him most of his army. Polybius, who says he used an inscription Hannibal set up in southern Italy recording his own numbers, gives 90,000 foot and 12,000 horse leaving New Carthage and 20,000 foot and 6,000 horse arriving in the Po valley. That is 102,000 down to 26,000, a loss of 76,000 men or 75 per cent, of whom many were detached as garrisons and some were sent home rather than lost.

With that remnant he beat a Roman army at the Trebia in December 218, ambushed and destroyed another at Lake Trasimene in June 217, killing a consul, and in August 216 met the largest army Rome had ever put in the field at Cannae.

Cannae

Polybius gives the Roman force as about 80,000 foot and 6,000 horse against Hannibal's 40,000 foot and 10,000 horse. Hannibal formed his infantry in a forward curve that gave way as the Romans pushed into it, held the flanks with his African veterans, and used his superior cavalry to clear the field and come in behind. The Roman army was surrounded and killed where it stood.

Example. Polybius puts the Roman dead at 70,000; Livy gives 48,200. Convert both to proportions and consider what the difference is worth.

Against a force of 86,000, Polybius's figure is 81 per cent killed and Livy's is 56 per cent. Both are catastrophic and they describe different events: 81 per cent is an annihilation with no possibility of flight, 56 per cent is a rout in which nearly half got away. Polybius wrote earlier, used family archives and had a particular interest in military mechanics; Livy names a source count that includes survivors regrouping at Canusium, which fits his lower number. The honest position is that the true figure is somewhere in that band and that the choice does not change the argument. Even the low figure is a death toll no ancient state was expected to survive politically.

Now you. Take Polybius's 70,000 dead and suppose the killing occupied about six hours. What rate does that imply, and what does the number tell you about the mechanics of the battle?

The refusal

What happened next is the reason Cannae is remembered as a Roman disaster and not as the end of Rome. Eighty senators had died in the battle. Capua, the second city of Italy, went over to Hannibal, and so did much of the south. Syracuse changed sides in 214, Philip V of Macedon allied with Carthage in 215.

Rome did not negotiate. It refused to ransom the prisoners Hannibal held, at a price he had set low precisely to make the offer attractive, and Polybius treats the refusal as the key to Roman character: the Senate calculated that buying men back would teach every future army that surrender was survivable. It enrolled 8,000 slaves as soldiers, bought from their owners at public expense, and armed them with weapons taken from temple walls. It appointed Fabius Maximus, who avoided battle for a year and was mocked for it, and then adopted his strategy anyway. It kept as many as twenty-five legions in the field in 212 and 211 BC, which at Polybius's establishment is about 105,000 citizen soldiers, close to 39 per cent of the 270,713 adult male citizens the census had recorded in 234 BC.

The census figures record the strain. The count for 209 BC gives 137,108 citizens against 270,713 twenty-five years earlier, a fall of 49 per cent. Part of that is dead men and part of it is a state too disrupted to register them, and no reading of it is comfortable.

Example. Why did the Senate refuse to ransom the prisoners after Cannae, when it was simultaneously arming slaves?

Because the two decisions answer different questions. Arming slaves solves a shortage of bodies at a moment when bodies are the constraint. Ransoming prisoners would solve the same shortage and destroy something the Republic could not replace: the expectation that a Roman army fights until it wins or dies. Polybius makes this explicit and treats it as the deliberate maintenance of an incentive, not as cruelty or pride. Read as policy it is coherent and cold: the state bought men from slave owners because that transaction changed nothing about what a citizen soldier could expect, and refused to buy men from Hannibal because that transaction would have changed everything.

Now you. In 209 BC, twelve of the thirty Latin colonies told the Senate they could supply no more men or money. Which half of that sentence is the striking one?

Why the alliance held

Hannibal's plan was to break Italy off, and it half worked. Capua, Tarentum and most of the Samnite and Lucanian country went over. The Latin colonies, the Etruscan and Umbrian cities, the Greek cities of the far south under Roman garrison, and central Italy did not.

Three reasons stand out. The first is that Rome had bound the local aristocracies to itself rather than the populations: defection required a city's ruling class to bet its property on a Carthaginian victory, and after Cannae many did, and after 211 the survivors of Capua's ruling class were executed and its territory confiscated, which was noticed everywhere else. The second is that the colonies were not subject towns but Roman and Latin populations living on land Rome had given them, whose neighbours would take it back the moment Rome fell. The third is that Hannibal could offer only revenge and a share of the fighting, since he never took Rome and never destroyed Rome's capacity to punish a defector the following season.

So the answer to the question the settlement posed in 338 BC arrived in 216 BC and was favourable. It is worth being precise about what it proves. It does not prove the allies were content; a third of Italy rose the moment a credible alternative appeared. It proves that Rome had made defection expensive and loyalty survivable for the men who made the decision.

Scipio, and a command that broke the rules

The war was won by a young man operating outside the constitutional rules the Republic had built to contain imperium.

Publius Cornelius Scipio was about twenty-five in 210 BC and had held only the aedileship when the assembly voted him proconsular imperium for the war in Spain, where his father and uncle had just been killed. He was a private citizen with the powers of a consul, appointed by the people to a theatre of war for an indefinite term. There was no precedent for it. He took New Carthage in 209, cleared Spain by 206, was elected consul for 205 despite being years under the legal age, crossed to Africa in 204 against strong senatorial opposition, and at Zama in 202 beat Hannibal, who had been recalled to defend Carthage.

Every element of that career reappears in the last century of the Republic: the command voted by the people rather than assigned by the Senate, the age rules waived, the theatre held for years, the army whose loyalty is to the man who leads it, and the popular general the Senate cannot control and eventually prosecutes. Scipio was investigated in the 180s over the accounting of the Antiochus campaign and left Rome for his estate at Liternum, where he died. The Republic had solved the Hannibalic emergency by suspending its own rules, and the solution proved habit-forming.

Example. In 210 BC the assembly gave imperium to a private citizen who had never held the praetorship or the consulship. Which of the Republic's constraints on imperium does that break, and which does it leave standing?

It breaks the cursus honorum, since imperium is being conferred outside the ladder of offices, and it breaks the Senate's practical monopoly on assigning commands, since the appointment came from a popular vote. It leaves annuality nominally standing, since the command was renewable rather than permanent, and it leaves collegiality untouched, since Scipio's imperium did not override the consuls'. That mixture is what makes the precedent dangerous rather than revolutionary: nothing in it declares the constitution void, and each element of it can be defended as an emergency measure, which is exactly what Pompey's supporters said in 67 BC.

Now you. The peace of 201 BC required Carthage to pay 10,000 talents over fifty years, against the 3,200 over ten imposed in 241. Compare the two and say what the difference is designed to do.

Where this leaves the Republic

Rome ended the war in 201 BC as the only great power in the western Mediterranean, holding Spain as well as the islands, with an aristocracy trained by seventeen years of continuous command, a citizen body that had lost a large fraction of its men, and southern Italy in ruins.

It also ended with something less visible and more important: armies that had served together for a decade under the same commanders, and a habit of granting long commands to men who could win with them. Within two years those armies were sent east, against the kingdoms that had inherited Alexander's empire, and Rome acquired in fifty years an empire it had never decided to want. How it governed what it took, and what that did to the constitution, is the next lesson.

In 200 BC Rome held Italy, Sicily, Sardinia and Spain; by 146 BC it had beaten Macedon, Syria and Carthage, destroyed two of the great cities of the Mediterranean, and acquired provinces it had no machinery to govern.

The previous lesson left Rome with veteran armies, experienced commanders and a taste for long commands. This lesson follows those armies east and then asks the question that matters more than any battle: once Rome had taken these places, what did it actually do with them? The answer is that it improvised, and the improvisations became the empire.

Fifty-three years

Polybius set out to explain how Rome subjected almost the whole inhabited world in less than fifty-three years, and the compression is real.

The Second Macedonian War ended at Cynoscephalae in 197 BC, where legions beat a Macedonian phalanx of the kind that had conquered Persia. In 196 the victorious commander Flamininus stood at the Isthmian Games and had a herald proclaim the Greek cities free, garrison-free and taxless, which Polybius says caused such an uproar that birds fell out of the sky. Roman armies then left Greece.

Antiochus III of Syria, the largest of the successor kingdoms, was beaten at Thermopylae in 191 and destroyed at Magnesia in 190. The peace of Apamea in 188 took Asia Minor west of the Taurus away from him and imposed 15,000 talents, about 360 million sesterces, the largest indemnity in ancient history. Rome annexed none of it, and gave the territory to Pergamum and Rhodes.

Then the pattern hardens. Macedon revolted and was crushed at Pydna in 168; this time it was broken into four republics and later, after another revolt, made a province in 146. In the same year Roman armies destroyed Carthage, which had complied with every clause of the treaty of 201, and Corinth, after an Achaean revolt. In 133 the last king of Pergamum left his kingdom to Rome in his will, and it became the province of Asia.

That sequence contains two different Romes: one that wins and withdraws, and one that annexes and destroys. Explaining the change from the first to the second is the central problem of Roman imperialism.

Why did Rome keep fighting?

Two answers are on offer and both have serious evidence behind them.

The defensive reading, which goes back to the ancient sources themselves, takes each war as a response to a threat or an appeal from an ally, and points to the withdrawals: Rome left Greece in 194 BC, left Asia in 188, and did not annex Macedonia for twenty years after Pydna. On this account the empire was acquired absent-mindedly by a state that kept being drawn back in.

The structural reading, argued most forcefully by William Harris, starts from the Roman aristocracy rather than from foreign policy. Rome fought a war in almost every year of the third and second centuries BC. Its highest honour, the triumph, required a victory with a minimum body count. Its elections rewarded military glory, as the earlier lesson on canvassing showed, and its temples, roads and games were built from booty by men whose careers depended on winning commands. A political class organised that way does not need a threat in order to find a war.

The evidence tends to support the structural reading of motive without discrediting every particular defensive claim, and the most useful way to hold the two together is to notice they answer different questions. Individual wars had occasions, often genuine ones. The system had a bias, and the bias explains why the occasions were so reliably found.

Example. In 196 BC Flamininus proclaimed the Greeks free, and Rome withdrew its armies. In 146 BC Rome destroyed Corinth and made Macedonia a province. What changed?

Not the intentions of a single policy-making mind, because there was no such mind: the Senate contained factions, and its members served short commands and wanted results within them. What changed was the accumulated experience of what freedom-granting produced. Every settlement left in place a Greek state capable of quarrelling with its neighbours and appealing to Rome, so Roman armies kept returning to the same places, and each return was more expensive than the last. Annexation is what a power does when it has decided that repeated intervention is dearer than permanent presence. The change is a learned habit rather than a decision, which is why it is so hard to date.

Now you. From 167 BC Roman citizens in Italy paid no direct tax on their property, the tributum having been suspended after the Macedonian booty came in. What does that fact do to the debate above?

Government by whoever was free

A provincia originally meant a sphere of duty assigned to a magistrate, not a piece of land, and the Roman empire was governed by that word long after it had come to mean territory.

The mechanism was simple and it broke on arithmetic. Rome created two extra praetors in 227 BC for Sicily and Sardinia, and two more in 197 for the Spains, giving six praetors and two consuls: eight holders of imperium a year. It then stopped creating them, while the provinces went on multiplying, so that by the 60s BC there were about fourteen provinces and still ten senior magistrates, of whom the consuls were usually needed elsewhere.

The gap was closed by prorogation: extending a magistrate's imperium past his year so that he governed pro consule, in place of a consul. Prorogation was a device of the Hannibalic emergency that became the ordinary way to staff an empire. Under Sulla's arrangements it was formalised: the year's magistrates served in Rome, then went out as promagistrates afterwards.

Notice what this does to the constraints the Republic had placed on imperium. Annuality is gone, because the commands last as long as the Senate finds it convenient. Collegiality is gone, because a governor has no colleague. Provocatio does not apply, because it protects citizens inside Italy and not the provincials. And there is no civil service: a governor arrived with a quaestor, a few legates, personal friends and his own slaves, and he was general, judge, tax supervisor and diplomat, answerable to nobody nearer than Rome and answerable there only after his term.

The tax farmers

Rome had no revenue department either, so it sold the job.

Provincial taxes were let out to companies of publicani, contractors who bid for the right to collect and kept the difference. In Sicily the arrangement was mild, because Rome inherited a working local system and the contracts were let in Sicily itself. In Asia, the richest province, the law of Gaius Gracchus in 123 BC moved the letting of the contract to Rome, where the censors sold it to companies of Roman equestrians.

The consequences are worth spelling out. A publican company bid a fixed sum for a province's taxes and profited from whatever it extracted above that, in a place where the only authority that could restrain it was a governor who needed its goodwill and its money. Cicero, defending a governor years later, described the provincials' hatred of the tax companies as one of the standing facts of Roman rule, and he was not an unfriendly witness to the equestrian order.

The Senate did respond. In 149 BC it created the first standing criminal court, the court de rebus repetundis, for the recovery of money extorted by magistrates. Its existence is the strongest evidence that Romans recognised the problem; its record is the strongest evidence that recognition was not enough. From 122 BC the juries of that court were staffed by equestrians, the same order as the tax companies, and control of those juries became one of the great political prizes of the late Republic, fought over for the next fifty years.

Example. Cicero, prosecuting the governor Verres in 70 BC, claimed he had extracted 40 million sesterces from Sicily in three years. Put that number in a scale a Roman would feel.

A legionary at this date was paid about 120 denarii a year, which is 480 sesterces. Forty million sesterces is therefore about 83,000 years of a soldier's pay, or the annual wage bill of roughly seventeen legions. One governor, one province, three years. Treat the figure as a prosecutor's maximum, since Cicero had every reason to inflate it and was addressing a court whose members might themselves govern provinces. Even halved, it says that provincial government was a route to wealth of an order that no career at Rome could match, and that is enough to explain everything that follows about the price of elections.

Now you. The extortion court was created in 149 BC and Verres was tried in 70 BC. What does the existence of the court prove, and what does it not?

What the money did at home

The wealth arriving in Italy after 200 BC was of an order the city had never handled. Aemilius Paullus's Macedonian triumph in 167 BC filled the treasury so completely that direct taxation of citizens was suspended. Pompey's eastern settlement a century later raised the state's annual revenues, on Plutarch's figures, from 50 million drachmas to 85 million, a rise of 70 per cent, which in Roman terms is roughly 200 million sesterces a year going to 340 million.

Alongside the public money came private money on a scale that unbalanced politics. A successful commander took a share of booty, distributed some to his soldiers, spent some on a temple or a road that carried his name, and kept the rest. The men who let and worked the contracts grew into a second wealthy order beside the Senate. And the whole system was self-reinforcing, because wealth won elections, elections won commands, and commands won wealth.

Roman writers noticed and moralised. Sallust argued that fear of Carthage had kept Rome honest and that its destruction in 146 BC removed the discipline, dating the decline of the Republic to that year. The theory is too neat, and something in it survives inspection: the generation after 146 is the one in which the political killings start.

Example. The Isthmian declaration of 196 BC promised the Greek cities freedom without garrison or tribute. Fifty years later Rome was destroying Corinth. Was the declaration a lie?

Probably not, and that is what makes it instructive. Flamininus needed a settlement that would let him take his army home within his command, and freedom for the Greek cities was a formula that cost Rome nothing and bought immediate goodwill. What the Romans meant by freedom was what a patron means: independence exercised within the expectations of a superior. What the Greeks heard was independence. Both sides used the word honestly and read it differently, and every subsequent quarrel, up to the war that ended with Corinth in ashes, grew in the gap between the two readings. Diplomatic language that both parties can accept for different reasons is not a lie, and it is a fault line.

Now you. Rome annexed no territory after Magnesia in 190 BC but imposed the largest indemnity in ancient history. What kind of empire does that describe?

Where this leaves the Republic

By 133 BC Rome governed the Mediterranean with eight or ten senior magistrates a year, no bureaucracy, no permanent revenue service, no professional army, and a constitution designed for a city that could see its own farmland from the walls.

The strains that broke it, though, did not appear first in the provinces. They appeared in Italy, in the countryside that supplied the soldiers, which fifty years of overseas war and imported wealth had changed beyond recognition. What actually happened to Italian land and Italian people is the next lesson, and it is the one place in this course where the ancient explanation and the modern evidence come apart.

By the 130s BC Roman politicians were saying that the men who conquered the world had no land in it, and the interesting question is how much of that was true.

The previous lesson followed the armies out and the money back. This one stays in Italy. It is the only lesson in this course where the ancient explanation and the modern evidence pull hard against each other, so it is worth being careful, because the reformers of the next lesson acted on the ancient explanation and the consequences followed from what they believed as much as from what was the case.

The diagnosis the Romans gave

Appian and Plutarch tell the same story, and it is coherent. The rich occupied more and more of the public land, the ager publicus taken from defeated Italians, beyond the legal limits. They worked it with slaves rather than free tenants, because slaves could not be conscripted and so were never taken away at harvest. The free smallholders, absent on campaign for years at a time, came home to debt and sold up, and drifted to Rome, where they had votes and no property. Since only a man with property could serve in the legions, the state was destroying its own recruiting base every time it won a war.

Plutarch adds the scene that made the case famous: Tiberius Gracchus travelling through Etruria on his way to Spain in 137 BC, seeing a countryside worked by imported slaves and almost empty of free inhabitants, and deciding that something had to be done.

Three parts of that account can be checked against independent evidence, and they hold up. Service abroad really was long and unpopular: the Spanish wars ran with little pause from 197 to 133 BC, and the levies of 151 and 138 produced riots in Rome, on one occasion with tribunes jailing a consul. Slaves really did arrive in numbers: 150,000 were enslaved in Epirus alone in 167 BC, and modern estimates put the slave population of Italy at the end of the Republic somewhere around a million, which on the usual estimates of Italy's total population is of the order of a fifth to a quarter of everyone alive there. And the property qualification for legionary service really did fall, from the 11,000 asses of the old fifth class to Polybius's 4,000 in the second century BC and to a figure of 1,500 by the time of Marius, a reduction of more than seven to one.

Cato's farm

The best single document is a handbook. Cato the Elder wrote De Agri Cultura around 160 BC as practical advice to a man with capital, and it describes without embarrassment the kind of estate the reformers complained about: slave-worked, market-oriented, and specialised in wine or oil rather than grain.

Cato staffs a 100-iugera vineyard with sixteen people: an overseer, his wife, ten labourers, a ploughman, a muleteer, a man for the willows and a swineherd. He also gives instructions on selling worn-out oxen and old or sickly slaves.

Example. A hundred iugera is about 25.2 hectares. Roman colonial allotments to settlers were often in the range of two to seven iugera. Work out what Cato's estate represents.

At the upper end of the allotment range, a hundred iugera is about fourteen family farms, and at the lower end it is fifty. The labour figure is 16 people on 100 iugera, or 0.16 workers per iugerum, which is thin compared with fourteen peasant households working the same ground, so the estate supports far fewer people on the same land. That is the mechanism the reformers were pointing at, stated in a form Cato would have recognised and approved: the villa is more profitable per unit of capital and it carries fewer men, and men are what the legions need. Notice also what Cato's list does not contain: any free tenant, and any product for local subsistence.

Now you. The property qualification for legionary service fell from 11,000 asses to about 1,500 over roughly a century. Two readings are possible. What are they?

What the census says

The censors counted citizens every five years, and the totals survive for much of this period. They are the closest thing the Republic has to a statistic.

The count stands at 337,022 in 164 BC and drifts down for thirty years, reaching 317,933 in 136 BC, a fall of 5.7 per cent at a time when nothing suggests a demographic collapse. Then, after the land commission set up in 133 BC had been working for several years, the census of 125 BC returns 394,736, an increase of 75,913 over the count of 131 BC, which is 23.8 per cent in six years.

Example. Interpret that jump of 75,913.

Populations do not grow by a quarter in six years, so the change is in registration rather than in births. The obvious candidate is the land commission: a man given a plot has property, property makes him liable to service and worth registering, and he now has a reason to present himself to the censors. On that reading the figure is the best evidence anywhere that the Gracchan programme actually distributed land to tens of thousands of people, which no narrative source states in numbers. The caution is that census procedure changed over time and that the surviving totals come to us through summaries of Livy, so a change in what was counted cannot be excluded. Even allowing for that, a jump of this size at exactly this moment is hard to explain by anything but the commission.

Now you. The census of 70 BC gives about 910,000 citizens and the census of 28 BC, reported by Augustus himself, gives 4,063,000. What is the problem, and what turns on it?

What the ground says

Since the 1950s, field survey has walked Italian landscapes collecting the pottery scatters that mark ancient settlement, and the results complicate the literary picture rather than confirming it.

Slave-run villas of Cato's type are real and are found where the sources say they were, in coastal Etruria, Latium and Campania, close to Rome and to ports. Elsewhere, surveys in inland Samnium, in the Biferno valley and in parts of the south find small sites persisting in numbers through the second century BC, which is not what wholesale displacement of the peasantry would look like.

So the honest summary is regional. The villa economy expanded in the best-connected parts of Italy, at the expense of small holders there, while much of the interior went on being farmed by families. The crisis the Gracchi described was real where they saw it, along the roads a Roman politician actually travelled, and it was not uniform across Italy.

That correction matters less than it might, for a specific reason. The recruiting crisis did not require the whole peninsula to be depopulated. It required only that the number of men with enough property to qualify for the legions stopped keeping pace with the number of legions Rome wanted to field, and that is consistent with the survey evidence and with the falling threshold together.

The city that filled up

The other half of the change is urban. Rome by the late second century BC was a city of several hundred thousand people, growing towards the near-million of Augustus's day, and it could not feed itself from its own hinterland. Grain came from Sicily, later from Africa and Egypt, and the price of it in the city became a political question with a constituency attached.

The men in that city were citizens with votes, concentrated in the four urban tribes, available on any given day when a tribal assembly met. Everything in the earlier lesson on how an election was won now acquires a new significance: for the first time there was a large, poor, permanently present electorate in the one place where votes were cast, and any politician willing to address its grievances had an instrument nobody had had before.

Example. Appian says the free peasantry was displaced and Italy filled with slaves. Field survey shows small farms persisting across much of the interior. How should the two be held together?

By asking what each source can see. Appian, writing much later from second-century sources, reports what a politically engaged Roman believed about the parts of Italy that Roman politics could see: the coastal plains, the estates of senators, the roads to the ports. Survey sees whole landscapes indifferently, including the ones nobody wrote about. Neither is lying, and the disagreement is mostly about scope. The correct statement is that a real transformation of the best land was generalised into a claim about Italy, that the generalisation was believed by the men who acted on it, and that beliefs about a crisis are themselves historical causes.

Now you. Suppose Lo Cascio is right and Italy's population was much larger than Brunt allowed. Does the Gracchan land programme still make sense?

Where this leaves the Republic

Put the pieces together. Italy's most valuable land was increasingly worked by slaves for absentee owners. The public land, which in law belonged to the Roman people, was in practice occupied by the men who sat in the Senate. The property qualification for the legions had fallen sevenfold. The city was full of citizens with votes and no land, and the countryside was full of veterans with claims and no compensation.

Every one of those is a problem the constitution had no procedure for solving, because solving any of them meant taking something from the men who ran the state. In 133 BC a tribune tried, using the assembly's legal power to legislate over the Senate's head. What happened to him is the next lesson, and it is the moment the Republic's political vocabulary changes.

In 133 BC a tribune of the plebs proposed to give public land to the poor, and by the end of the year he and about three hundred of his supporters had been beaten to death in the streets of Rome.

The previous lesson set out the problem the Gracchi thought they were solving: public land occupied by the rich, a shrinking pool of men who qualified for the legions, and a city filling with citizens who had votes and nothing else. This lesson is about the attempt and its consequences, and the consequences are not mainly about land. What changed in these twelve years is the answer to the question of how a Roman political dispute ends.

Tiberius and the land bill

Tiberius Sempronius Gracchus was not an outsider. His father had been consul twice and censor, his mother Cornelia was the daughter of the man who beat Hannibal, and his sister was married to Scipio Aemilianus, who destroyed Carthage. He was as close to the centre of the aristocracy as a Roman could be.

His bill, passed in 133 BC, revived the old limit on holdings of public land: 500 iugera per man, about 126 hectares, with an extra 250 for each of two sons. The excess reverted to the state and was distributed in small inalienable plots to landless citizens, administered by a commission of three with the power to decide what was public land and what was private, which in a country with no land registry was the real power in the bill.

Nothing in it was legally novel. Public land belonged to the Roman people, a limit on its occupation had a respectable pedigree, and the assembly could pass what it liked. What was novel was the procedure. Tiberius took the bill straight to the plebeian assembly without the customary prior discussion in the Senate, and when a fellow tribune, Marcus Octavius, vetoed it, he put a motion to the assembly to remove Octavius from office, and carried it.

Example. Was the deposition of Octavius legal? Set out the strongest case on each side.

For it: a tribune exists to represent the plebs, the plebs are sovereign in their own assembly, and an officer who blocks what the plebs demand has abandoned his function, so the body that elected him may unelect him. Tiberius argued exactly this, and no statute forbade it. Against it: the tribunate's whole value lay in an unconditional veto protected by sacrosanctity, and a veto that can be removed by the majority it obstructs is not a veto at all. The deposition therefore destroyed the office in order to use it. Both arguments are good, which is the point. The Roman constitution had no court and no text, so a genuinely new question had no answer beyond what one side could carry, and by the end of the year the answer was being decided with clubs.

Now you. Tiberius later stood for a second consecutive tribunate, and his enemies said this proved he wanted permanent power. What was he actually facing?

The killing

Two other acts made the collision certain. Attalus III of Pergamum died and left his kingdom to Rome, and Tiberius proposed that the assembly, not the Senate, should decide how the money was used, which touched the Senate's oldest practical monopolies, finance and foreign affairs. Then he stood for re-election.

On election day the Senate met, and the consul refused to act. Publius Cornelius Scipio Nasica, the chief priest, called on those who wished to save the Republic to follow him, and led senators, their clients and their slaves, armed with broken furniture, into the assembly. Tiberius and some three hundred others were killed and their bodies thrown into the Tiber.

Two features of that afternoon matter more than the deaths. The men who did it were not a mob but the governing class in person, and no court ever tried them. And the justification offered was not a law but a necessity: the Republic was in danger, and a citizen might act. Every political killing for the next hundred years appeals to that formula.

What the killing did not undo

Note what did not happen next. The land commission was not abolished. It kept its three members, kept its power to rule on what was public land, and went on distributing plots for years, which is why the census jumped by 75,913 between 131 and 125 BC. The Senate had killed the man and kept the law.

That is worth sitting with, because it defines the quarrel. If the land bill had been the intolerable thing, it would have been repealed the moment its author was dead, and repealing a law was easy. What had been intolerable was the method: a tribune who went round the Senate, deposed a colleague, disposed of a foreign bequest and looked like staying in office. The governing class could accept losing a policy and could not accept a rival centre of authority, and every crisis in the rest of this course has that shape.

Gaius, and a programme instead of a bill

Gaius Gracchus was elected tribune for 123 BC, ten years after his brother's death, and held the office two years running. Where Tiberius had one measure, Gaius had a system, and it is the most coherent legislative programme anyone attempted under the Republic.

He restored and extended the land commission and founded colonies, including one on the site of Carthage. He built roads with milestones. He passed a grain law under which the state sold wheat in Rome at a fixed price of 6 and one third asses a modius, in rations of five modii a month. He moved the collection of the Asian taxes to a contract sold at Rome, which handed the richest province in the empire to the equestrian companies. He took the extortion court away from senatorial juries and gave it to equestrian ones. He legislated on the levy, forbidding conscription under seventeen and requiring the state to supply clothing. And he proposed to extend Roman citizenship to the Latins and Latin rights to the other Italian allies.

Read as a whole, the programme is an attempt to build a coalition that the Senate could not break: the urban plebs held by grain, the rural poor by land, the equestrians by the tax contract and the courts, the soldiers by the levy law, and the Italians by citizenship. It is the first Roman politics that looks modern.

Example. A modius of wheat is about 6.55 kg. Work out what a monthly ration of five modii actually provided.

Five modii is about 32.8 kg a month, which is 1.09 kg a day. Wheat runs at roughly 3,340 kcal a kilogram, so the ration supplies about 3,650 kcal a day, against an adult man's requirement of something like 2,800. The ration therefore fed one adult comfortably and a family not at all, which is exactly right for its purpose: it was drawn by the male citizen, in person, in Rome, and it subsidised the household rather than supporting it. Note also what it was not. The grain was sold, not given, at a fixed and below-market price, and the free distributions that Roman moralists later denounced are a development of the 50s BC, not a Gracchan invention.

Now you. Gaius's citizenship proposal was defeated, and the sources say it cost him support among the very voters who had backed everything else. Why?

The final decree

Gaius failed to be elected for a third year, and in 121 BC his opponents moved to repeal his colony at Carthage. In the disorder that followed, a servant of the consul Opimius was killed, and the Senate passed a resolution asking the consuls to see that the state took no harm.

That formula, later called the senatus consultum ultimum, was new. It did not declare martial law, since the Senate had no power to do that, and it did not suspend any right, since it was formally only advice. What it did was announce in advance that the Senate would treat the killing of a citizen by a magistrate as legitimate. Opimius raised armed men, stormed the Aventine, and killed Gaius and some three thousand of his supporters, many of them after the fighting, without trial. He was later prosecuted and acquitted.

Everything about the device is characteristic of the Roman constitution. It is not a law. It creates no power. It works entirely by telling a magistrate what the men who will judge him afterwards think, and it worked because the same class supplied both the magistrate and the jury. It was used again against Saturninus in 100, against Catiline's associates in 63, and against Caesar in 49, and the argument about whether it could authorise the killing of citizens without trial ran until the Republic ended: Cicero was exiled in 58 for the executions of 63 and recalled the year after.

Example. In what sense did the events of 133 and 121 BC change the constitution, given that no law was passed either time?

They changed the set of moves available. Before 133 a political dispute at Rome ended in a vote, a veto, a delay or a prosecution, and the participants expected to keep meeting. After 121 the repertoire included killing, done by men who suffered nothing for it, justified by an appeal to necessity that the Senate had formally blessed. That is a constitutional change in every sense that matters, even though nothing was written down. It also worked in both directions: a reformer now had to expect violence, which made his own resort to it reasonable, and made a bodyguard a normal part of a political career. Within forty years both sides have private armies.

Now you. Roman politicians of the following century are often labelled populares or optimates. Given what the Gracchi did, why is it wrong to treat these as parties?

Where this leaves the Republic

The Gracchi are usually described as failures, and the description needs qualifying. The land commission distributed land on a scale the census records. The grain law survived and grew. The equestrian juries and the Asian tax contract lasted for decades and made the equestrian order a permanent player. The citizenship proposal was defeated in 122 BC and enacted in 89 after a war.

What failed was the attempt to do it inside the constitution. The Gracchi proved that a determined tribune with a popular measure could beat the Senate in a vote, and that the Senate would respond with force it could not be punished for. Both lessons were learned by everyone.

The next crisis came from the other direction. Rome's wars in Africa and against the Cimbri went badly under senatorial commanders, and the man who fixed them, Gaius Marius, solved the recruiting problem this lesson began with by simply abolishing the property qualification. What that did to the relationship between a soldier and his general is the next lesson.

When a Roman soldier owned a farm, discharge sent him home; when he owned nothing, discharge left him with whatever his commander could obtain for him, and that single change turned armies into political constituencies.

The previous lesson left the Republic with the Gracchan land programme half executed, the Senate willing to kill, and the recruiting problem unsolved. This lesson is about the man who solved the recruiting problem and, in doing so, created the instrument that would destroy the Republic. It is also the lesson where the sources most need handling with care, because the main narrative of Marius's rise was written sixty years later by a moralist with a case to make.

The war that discredited the Senate

Jugurtha, a prince of Numidia in North Africa, murdered his way to sole power over a kingdom allied to Rome and then fought off Roman intervention for seven years, from 112 to 105 BC. What made the war politically explosive was not its difficulty but its handling. Roman commanders made agreements with Jugurtha and had them repudiated. Bribery was alleged at every stage, and a special court was set up to investigate it. Jugurtha visited Rome in person under safe conduct and, on the story Sallust tells, left saying the city was for sale and would perish as soon as it found a buyer.

Sallust wrote his account in the 40s BC, after the Republic was effectively over, to argue that the nobility's greed had destroyed it. He is a first-class writer with a thesis, and the thesis shapes the selection of everything in his book. What survives independent of him is that the war went badly for years, that a court was created to try senators for corruption in it, and that the Roman people responded by taking the command away from the aristocracy and giving it to an outsider. The last of those is a fact of legislation and needs no source's goodwill.

That outsider was Gaius Marius, an equestrian from Arpinum with no consular ancestors, who had risen as a client of the Metelli, served with distinction, and then stood against his own patron's family. Elected consul for 107 BC, he had the African command transferred to him by a vote of the people, over the Senate's assignment. The precedent was Scipio's in 210 BC, and it was becoming a habit.

Enrolling men with nothing

For the new army Marius did something Sallust records in a sentence: he enrolled soldiers from the capite censi, the head count, the men below the property qualification, contrary to custom.

Be careful with what that means. The threshold had been falling for a century, from 11,000 asses to Polybius's 4,000 to about 1,500, and poor men had served before, in emergencies. There was no law abolishing the qualification, and no ancient writer says there was. What Marius did was to take volunteers without property as a matter of course, for a foreign war, when there was no emergency compelling it. Recent scholarship has been steadily reducing the "Marian reform" from a revolution to an acceleration, and the correction is right about the mechanism and can be pushed too far, because the consequences that follow do not depend on how sudden the change was.

Here are those consequences. A propertied conscript serves, wins, and goes back to a farm that is still his. A propertyless volunteer serves for years, and at discharge has his savings, his share of booty, and no land. If he is to have land, somebody must obtain a law giving it to him, and the only person with an interest in doing that is his general. So the soldier's retirement became a political favour, delivered by a patron, and the patron's ability to deliver depended on his standing in Rome.

Example. Work out the size of the problem. Suppose one legion of about 5,000 men is discharged and each veteran is to receive ten iugera, a modest plot. How much land is that, and what follows?

Fifty thousand iugera, which is about 12,600 hectares, for a single legion. An army of four legions needs something over 50,000 hectares, all of it in Italy, all of it currently belonging to somebody, and all of it obtainable only by a law passed in an assembly against the resistance of the occupiers. There is no standing fund, no reserve of land set aside, and no procedure. Every discharge is therefore a separate political campaign, and a general who cannot win it has failed the men who won his battles. That is the structural fact that makes armies into followings, and it does not require anybody to be disloyal or ambitious: it is what the absence of a pension system produces.

Now you. Why did the Senate not simply create a standing procedure for settling veterans and remove the problem?

Six consulships

Marius finished the African war in 105, or rather his quaestor Sulla did, by persuading Jugurtha's father-in-law to hand him over, an episode both men spent the next twenty years arguing about.

He then had to deal with something worse. Two migrating northern peoples, the Cimbri and the Teutones, had been destroying Roman armies since 113, and in 105 BC at Arausio, on the Rhone, they destroyed two of them at once, with losses reported at 80,000 soldiers and 40,000 camp followers, a figure of Cannae's order and this time inflicted on armies that were not surrounded but badly commanded.

Rome elected Marius consul for 104 and then for every year to 100, six consulships in all and five of them consecutive, in flat contradiction of the rule requiring a ten-year interval. He rebuilt the army, waited two years, and destroyed the Teutones at Aquae Sextiae in 102 and the Cimbri at Vercellae in 101.

The legion remade

The military changes credited to Marius belong here with a caution: our sources for them are late and fond of attributing systems to individuals. The legion was reorganised around the cohort of about 480 men rather than the older manipular lines, which gave a commander ten manoeuvre units instead of thirty and worked better with troops of uniform equipment. The eagle became the legion's single standard, an object soldiers would die to recover. The men carried their own kit to reduce the baggage train, which earned them the nickname of Marius's mules. Whether Marius invented these or ratified changes already under way, the direction is clear: a standing professional force with its own identity, in which the legion, not the property class, was what a soldier belonged to.

Example. Marius held six consulships, five of them in consecutive years, against a rule requiring a ten-year gap. What does that tell you about the constitution at this date?

That its rules were now negotiable under pressure, and that the pressure came from the people who voted rather than from a strongman. Marius did not seize these consulships; he was elected to them, repeatedly, by an electorate frightened of the Cimbri and unimpressed by the aristocracy's record. The rule was not repealed, and no court struck the elections down, because there was no court. So the constraint this course described at the start as fundamental, annuality with intervals, turned out to be a convention that a sufficiently frightened assembly would set aside for a sufficiently successful general. Once set aside for Marius it could be set aside for Pompey, and the argument in 52 BC over Caesar's candidacy is conducted with these precedents on the table.

Now you. Sallust says Marius enrolled the head count "contrary to custom". Modern historians argue that the property qualification had been eroding for a century and that Marius changed less than the sources imply. Does the correction matter for the argument of this lesson?

The first test, and the first betrayal

The consequences arrived immediately. To settle his African and Gallic veterans, Marius worked with the tribune Saturninus, who carried land bills in 103 and 100 BC by methods that had become standard: assembly votes, violence in the streets, and an oath imposed on senators to obey the law.

In 100 BC Saturninus went further, had a rival candidate killed, and lost control. The Senate passed its final decree, and Marius, consul for the sixth time, took arms against his own tribune, besieged him on the Capitol, accepted his surrender and stood by while he was lynched.

That episode is worth more attention than it usually gets, because it shows the position of a man like Marius exactly. He needed a popular tribune to obtain his veterans' land; he needed senatorial acceptance to have a political future; and when the two requirements collided he chose the Senate and destroyed his own instrument. He then retired into an unhappy semi-retirement, having got his soldiers their land and lost the alliance that produced it. The lesson his successors drew was not that the popular route was wrong but that it had to be pursued with a bigger army and fewer scruples.

Example. In 100 BC Marius put down Saturninus. In 88 BC Sulla marched on Rome. What changed in twelve years?

The willingness to use the army itself against the city. Marius in 100 was a serving consul acting under the Senate's decree with hastily armed citizens, which is a use of force within the political system as the Gracchan crisis had left it. Sulla in 88 brought legions from a foreign command against Rome to overturn a vote of the assembly, which is not a move inside the system but a demonstration that the system's decisions could be reversed by force from outside it. Between the two lies the Social War, which put whole armies of Roman soldiers under commanders who had learned that the state's decisions were negotiable. The precedent Marius set was that rules bend; the precedent Sulla set was that they can be broken by anyone with legions who is prepared to accept the consequences.

Now you. Why did the propertyless soldier's dependence on his general not simply make the general dependent on the state, which paid and discharged the army?

Where this leaves the Republic

By 100 BC Rome had a professional army recruited from men with nothing, generals who had to be politicians to look after them, a Senate that would use force to defend its position, and a set of constitutional rules that had proved bendable in an emergency.

One large group had watched all of this from the outside: the Italian allies, who supplied half the army, received none of the land, and had no vote in any of the assemblies where these fights were decided. In 91 BC a Roman tribune proposed to enfranchise them, and was murdered. What Italy did next, and what it let loose inside Rome, is the next lesson.

In 91 BC most of Italy revolted against Rome to obtain Roman citizenship, and three years later a Roman consul used the armies that war had created to capture his own capital.

The previous lesson left a professional army whose soldiers depended on their generals, and a governing class that had learned to bend its own rules. This lesson puts those two facts together with the one group that had been left out of every settlement since 338 BC. The Italian allies had supplied half of Rome's armies for two centuries and held no vote in the assemblies that sent them to war, and when they finally demanded a remedy the Republic discovered how thin the distance was between an army and a coup.

The allies revolt

The immediate cause was a murder. Marcus Livius Drusus, tribune in 91 BC, proposed a package that included citizenship for the Italians, and was stabbed in his own house one evening, his killer never found. With the constitutional route closed, the allies took the other one.

What they built is the most revealing thing about the war. The rebels formed a state of their own called Italia, with its capital at Corfinium, renamed Italica. It had a senate of five hundred, two consuls, twelve praetors and its own coinage, some of it showing an Italian bull goring the Roman wolf, with legends in Oscan rather than Latin.

Example. The Italian rebels copied Roman institutions almost exactly. What does that tell you about what they wanted?

That the war was about admission, not separation. A movement seeking independence from Rome does not organise itself as a duplicate of Rome, with a senate, paired consuls and a praetorian college; it revives whatever it had before. The rebels built a second Rome because two centuries of alliance had made them politically Roman in everything but the vote, and their leaders were exactly the local aristocrats whom the settlement of 338 BC had bound to Roman interests. The Oscan legends cut the other way and are the qualification worth keeping: local identity was real, and the alternative state was a serious construction rather than a bargaining posture. The best reading is that they wanted in, and were prepared to build something else if refused.

Now you. Rome conceded the citizenship in 90 and 89 BC, and enrolled the new citizens in a limited number of tribes, on some accounts eight or ten out of thirty-five. Why, and what does it echo?

Winning by conceding

The fighting was severe. Appian reports that some 300,000 men died in Italy, and though ancient totals of that kind are unreliable, the war put Roman legions against soldiers trained in exactly the same way, since the rebels were the men who had fought Rome's wars. Roman commanders included both Marius and Sulla.

Rome's decisive move was legislative. The lex Julia of 90 BC offered citizenship to allies who had not revolted, and to those who laid down their arms; the lex Plautia Papiria of 89 extended it further. Communities took the offer and left the war, and the revolt collapsed piecemeal during 89 and 88.

The result is one of the strangest outcomes in Roman history: the rebels lost the war and obtained the thing they had gone to war for. Italy south of the Po became Roman, which is the reason "Italy" thereafter means a country of citizens rather than a peninsula of subject allies, and it is the foundation of everything the empire later did with citizenship. It also meant that the men who fought each other in the civil wars of the next sixty years were, on both sides, Romans.

The march

In 88 BC Mithridates VI, king of Pontus, overran the Roman province of Asia and organised the massacre of the Romans and Italians living there, a figure our sources put at 80,000 in a single coordinated day. The command against him was the richest opportunity in a generation, and it was assigned to the consul Sulla.

The tribune Sulpicius then carried a law transferring it to Marius, by a vote of the assembly. The move was not new: commands had been transferred by popular vote to Scipio in 210 and to Marius himself in 107. What was new was the response. Sulla left Rome, went to his six legions at Nola, told them what had happened, and marched on the city.

The detail that matters most is that every one of his senior officers except one refused to go with him. Roman soldiers followed and Roman senators did not, which tells you both that the act was understood as monstrous and that the professional army of the previous lesson would do it anyway, because the men who marched expected their futures to come from Sulla rather than from the state.

Rome was taken. Sulpicius was hunted down and killed, Marius escaped to Africa, and Sulla left for the east. In his absence his enemies took the city back, killed his supporters, and held it for five years, during which Marius returned, held a seventh consulship and died in it. When Sulla came back in 83, a second civil war ran for over a year and ended at the Colline Gate outside Rome in November 82.

The proscriptions

Sulla's victory was followed by an innovation in political killing. He posted lists of names in the Forum. A proscribed man could be killed by anyone, his killer was paid, his property was confiscated and sold, and his sons and grandsons were barred from office.

The numbers vary by source, which is itself informative. Appian gives 40 senators and 1,600 equestrians on the first list; Plutarch describes 80 names, then 220, then 220 more; later writers give totals of 4,700 and 9,000. The lowest figures describe hundreds of the wealthiest men in Italy and the highest describe thousands.

What the confiscations did was as important as the killings. Property changed hands at auction on a scale that created fortunes, and Crassus, who reappears in the next lesson as the richest man in Rome, built much of his wealth this way. The proscription was not only terror but a transfer of assets to Sulla's supporters, which bound them to the survival of his settlement.

Example. Sulla had himself appointed dictator to write laws and reconstitute the Republic, an office nobody had held for 120 years and this time with no fixed term. How should that be classified?

As a monarchy operating through a republican form, and as an accurate reading of what the constitution allowed. The dictatorship was a real Roman office, the appointment went through a law, and the whole apparatus of magistrates and assemblies continued to function. What was missing was the six-month limit that had made the old office safe, and the emergency that had justified it: Sulla's commission was to rebuild the state, which has no natural end date. The Romans of 82 BC therefore watched a constitutional office being used to hold unlimited power for as long as its holder wanted, and this is the template Caesar takes up in 49, with one difference that decides everything. Sulla resigned.

Now you. The proscription totals in our sources range from a few hundred to nine thousand. What should be done with that range?

A constitution built to stop the next Sulla

Then, having taken power by an act without precedent, Sulla used it to legislate against every method he had used.

The Senate was doubled to about 600 by enrolling equestrians, and the number of quaestors raised to twenty, so that entry to the Senate became automatic and the censors' discretion over its membership largely disappeared. Praetors were raised to eight, and the ten senior magistrates each year went out to govern provinces afterwards, which regularised the promagistracy. The minimum ages and intervals of the cursus honorum were re-enacted and enforced. The permanent courts were reorganised and their juries returned to senators.

Above all, the tribunate was destroyed as an instrument. A tribune could not put a bill to the assembly without the Senate's prior approval, his veto was restricted, and any man who held the office was barred from holding any other, which removed it from the career path of anyone with prospects. And a law of treason made it an offence for a governor to leave his province or make war without authority, which describes precisely what Sulla had done in 88.

He resigned the dictatorship in 79 BC and died the next year. The settlement began to come apart at once: the ban on ex-tribunes holding further office was lifted in 75, and in 70 BC the consuls Pompey and Crassus restored the tribunate's powers in full, eleven years after Sulla's laws had taken them away.

Example. Why did the Sullan constitution fail so quickly?

Because it diagnosed the tribunate as the disease. Every measure in it assumes that the danger to the Republic is a popular politician using the assembly against the Senate, which is a fair description of 133 and 123 and no description at all of 88. Sulla did not take Rome with a tribune's bill; he took it with six legions whose men expected land from him, and nothing in his settlement touched the recruitment, the pay, the discharge or the length of commands that produced such legions. He also left behind the demonstration that a commander who was willing to march could have anything he wanted, which no law can undo. Within a decade Pompey was extracting a consulship at 36, having held no office at all, by standing outside the city with an army he declined to disband.

Now you. Sulla's treason law made it an offence for a governor to leave his province or start a war without authority. Was it useless?

Where this leaves the Republic

By 70 BC Italy was a country of Roman citizens, the Sullan restrictions on the tribunate were gone, and the two men who removed them were a general with no legal claim to his consulship and the richest man in Rome.

The empire had also gone on growing while Rome fought itself, and the problems it now presented, a war against Mithridates that had run on and off for twenty years, piracy that interrupted the grain supply of the city, a rebellion in Spain, could not be handled by an annual magistrate with a legion or two. What the Republic did about that, and what the men who received those commands became, is the next lesson.

Between 70 and 50 BC the Republic solved every one of its practical problems by giving one man powers no Roman had held before, and each solution left him harder to control than the problem had been.

The previous lesson ended with Sulla's settlement dismantled and Italy enfranchised. This one is about the twenty years in which the Republic worked, in the sense that pirates were cleared, Mithridates finished, Gaul conquered and the grain kept flowing, and in which the institutions this course began by describing stopped mattering. The tension between those two statements is the subject.

A career made of exceptions

Take Pompey, whose progress through the constitution is best read as a list of things that could not legally happen.

In 83 BC, aged twenty-three and holding no office, he raised three legions from his father's clients and estates and brought them to Sulla. In 81 or 80 he demanded and received a triumph as a private citizen. In 77 he was given proconsular imperium against Sertorius in Spain, again as a private citizen, on the Senate's own motion. In 71 he stood for the consulship of 70 and won it at the age of 36, six years under the legal minimum, having never been quaestor, aedile or praetor, and while standing outside the city with an army he had not disbanded.

Then came the two commands that mattered. The lex Gabinia of 67 BC gave him a three-year command against the pirates covering the whole Mediterranean and the coasts fifty miles inland, with 500 ships, 120,000 infantry, 5,000 cavalry, twenty-four legates of praetorian rank and 6,000 talents. In 66 the lex Manilia transferred the war against Mithridates to him with no limit of time or territory, and he spent five years reorganising the entire eastern Mediterranean, founding cities, making kings and settling boundaries, on his own authority.

Example. Set the pirate command against a normal magistrate's resources and say what the difference amounts to.

An ordinary consul commanded two legions, about 10,000 men, for one year, in one assigned province. Pompey received 120,000 infantry, equivalent to about twenty-four legions, for three years, across every province with a coastline, with twenty-four subordinates each of the rank a province would normally have as its governor. The money, 6,000 talents, is about 144 million sesterces, of the same order as the state's entire annual revenue at the time. It is not a magistracy with unusual scope; it is a parallel government of the Mediterranean, created by a vote of the assembly against the opposition of most of the Senate. The tribune who proposed it was following the precedents of 210, 107 and 88 BC, and the scale is what had changed.

Now you. Cicero says that grain prices in Rome fell on the day Pompey's appointment was announced. What does that tell you, and how far should it be trusted?

The money and the muscle

Marcus Licinius Crassus is the second figure, and he is the one usually underrated, because he won no famous battles and his fortune is the least glamorous kind of power.

Crassus made his money out of Sulla's proscriptions, buying confiscated estates cheaply, and enlarged it with urban property, mines and slaves. Plutarch puts his fortune at 7,100 talents, about 170 million sesterces, roughly half the state's annual revenue after Pompey's eastern settlement. What that bought was not soldiers but politicians: Crassus lent money to men on their way up, including Caesar, whose debts he guaranteed, and a creditor of that kind holds a permanent claim on the borrower's career.

He also has the one military achievement of the period that was purely internal. The rising of Spartacus from 73 to 71 BC destroyed several Roman armies before Crassus put it down, and he ended it with a gesture: 6,000 captives crucified along the Via Appia from Capua to Rome.

Example. The Via Appia from Capua to Rome is about 195 km. Work out what 6,000 crosses along it means physically, and what the gesture was for.

Six thousand crosses spread over 195,000 metres is one every 32.5 metres, so a traveller on the main road into Rome from the south passed a crucified man roughly every thirty seconds of walking, for a week. It is not a punishment, since the men were already prisoners and could have been killed anywhere. It is a message addressed to the slave population of Italy, the one the earlier lesson on Italy counted in the hundreds of thousands, and its scale is a measure of how frightened their owners were. Notice also who is speaking: not the Roman state, which passed no decree about it, but a commander disposing of prisoners as he saw fit, in Italy, in a manner designed to be seen. Power of that kind belonged to individuals now.

Now you. Suetonius says Caesar took so much gold from Gaul that he sold it in Italy at 3,000 sesterces the pound against a standard price of 4,000. What can be drawn from that?

A private arrangement

In 60 BC three men found that each of them had been blocked by the Senate. Pompey could not get his eastern settlement ratified or land for his veterans. Crassus could not get relief for the tax companies that had overbid for the Asian contract. Caesar, back from Spain, was being forced to choose between a triumph and standing for the consulship.

They agreed to help each other, and the agreement worked: Caesar was elected consul for 59, passed Pompey's settlement and his land bill and Crassus's rebate, and received Cisalpine Gaul and Illyricum for five years by a law of the tribune Vatinius, to which the Senate added Transalpine Gaul. In 56 the three renewed the arrangement, and in 55 Pompey and Crassus took a second consulship together and extended Caesar's command by another five years, giving themselves Spain and Syria.

This is what Roman writers called a conspiracy of three men and modern books call the First Triumvirate, and the name is misleading in a way worth being precise about. It was not an office, a magistracy or a legal body of any kind. Three private citizens agreed to combine their votes, their clients, their money and their soldiers, and the Republic's institutions did what they asked. We know about it because Cicero, who was outside it, describes its operation week by week in letters to Atticus that he never expected anyone else to read.

Example. What does the existence of an arrangement like this tell you about the constitution in 59 BC?

That the formal system had become a set of instruments rather than a source of authority. Everything the three men wanted was obtained through proper procedures: a consul proposed, an assembly voted, a law resulted. Nothing was seized. And the decisions were made in private by men whose power came from armies, fortunes and clients, with the assemblies functioning as a ratification stage. That is a hollowed constitution rather than a violated one, and it is much harder to defend against, because there is no moment at which anyone can point and say a law was broken. Cicero spent the decade trying to find such a moment and never did.

Now you. In 52 BC, after months of rioting in which elections could not be held and Clodius was killed on the Appian Way, Pompey was made consul without a colleague. What is odd about that title?

The city while this happened

Rome itself in these years is worth a paragraph, because it explains why so many people preferred a strong man to the alternative.

In 63 BC Cicero, as consul, uncovered a conspiracy by the bankrupt aristocrat Catiline and had five citizens executed without trial under the Senate's final decree, which was legally contested at the time and drove him into exile five years later. From 58 the streets were run by organised gangs: Clodius, a patrician turned tribune with a following among the urban poor, against Milo, backed by the Senate's hard men. Elections were postponed for months at a time, and 53 BC opened with no consuls at all because none could be elected. In January 52 Clodius was killed in a brawl on the Appian Way, his supporters burned the Senate house as his pyre, and it was in the ashes of that building that Pompey received his sole consulship.

A citizen watching this had reason to conclude that the Republic could no longer keep order in one city, let alone an empire.

Why the Republic needed the men it feared

It would be easy to read all this as ambition overwhelming institutions, and that reading is half of it. The other half is that the problems were real and the ordinary machinery could not touch them.

Piracy had made the sea unusable, interrupted the grain that fed Rome, and reached the point where two praetors were carried off with their attendants from the Italian coast. Mithridates had been fought inconclusively for twenty years by a succession of annual commanders, each of whom lost the campaign his predecessor had begun. Spain, Gaul and the east all needed multi-year commitments by forces larger than a consul commanded.

The constitution had exactly one way to supply that: suspend its own rules and give one man a long command with large forces. It did so repeatedly, each time for a good reason, and each time the beneficiary emerged with more soldiers, more money and more clients than any institution could balance. By 50 BC Caesar had spent nine campaigning seasons in Gaul with an army that had grown from four legions to about eleven, all of them recruited, paid and enriched under him.

The Senate's dilemma is genuine. Refusing the commands meant leaving the empire ungoverned; granting them meant creating men it could not control. Nothing in the constitution offered a third option, because it had been designed for a city whose commanders were a week's march away and back in the Forum by the following spring.

Where this leaves the Republic

By 50 BC the arrangement of 60 was dead. Crassus, who had held it together as much as anyone, took an army into Mesopotamia and was destroyed at Carrhae in 53, and Julia, Caesar's daughter and Pompey's wife, had died in childbirth the year before. What remained was two men with armies, no institution able to arbitrate between them, and a legal question that had no honest answer: whether Caesar could stand for a second consulship in his absence, or had to come home as a private citizen and face the prosecutions his enemies had been preparing for a decade.

That question is not really about law, as the next lesson shows. It is about what happens when the rules require a man to make himself defenceless and he has fifty thousand soldiers who would rather he did not.

In December 50 BC the Senate voted by 370 to 22 that Caesar and Pompey should both give up their commands, and six weeks later the war began anyway.

The previous lesson left two men with armies and no institution able to judge between them. This one is about how that situation turned into a war, what Caesar did with the power he won, and why the men who stabbed him in 44 BC found that they had removed a person and not a regime, which is the same discovery the Republic's founders had made in reverse five centuries earlier.

The question nobody could answer

The legal dispute of 51 to 49 BC is genuinely difficult, and that is the point of it.

Caesar's Gallic command was ending. He wanted to stand for a second consulship without first returning to Rome as a private citizen, and a law carried by all ten tribunes in 52 BC gave him permission to stand in absence. His enemies wanted him to come home unarmed, where prosecution waited for his acts as consul in 59 and afterwards, and they had Pompey's own legislation on magistracies to cite. Both sides could construct an argument that was legally respectable, and there was no court with jurisdiction to decide between them, because the Republic had never built one.

What breaks the deadlock is not the law but the incentives underneath it. The rules required Caesar to make himself defenceless in front of men who had spent a decade promising to destroy him, and the previous lessons have shown what a Roman politician without office and without protection could expect.

Example. In December 50 BC the tribune Curio moved that Caesar and Pompey should both lay down their commands, and the Senate carried it by 370 votes to 22. What does that division prove, and what does it not?

It proves that the war was not wanted by the political class. Of 392 senators voting, 94.4 per cent preferred mutual disarmament, which means the hard line came from a small group, on the usual reckoning around Cato and the Claudii Marcelli, who nonetheless controlled the presiding magistrates and could prevent the motion from taking effect. It does not prove the majority had a policy. A vote for both men to disarm is a vote for someone else to solve the problem, and it came with no mechanism to make either of them comply. Roman decision-making could register an overwhelming preference and be unable to execute it, which is a fair epitaph for the late Republic.

Now you. On 7 January 49 BC the Senate passed its final decree, and the tribunes Antony and Cassius left Rome for Caesar's camp. Why did that detail matter to Caesar's case?

The war

Caesar crossed the Rubicon, the stream marking the southern boundary of his province, with a single legion around 10 January 49 BC. Italy had no army in being: Pompey's veterans were in Spain, and the recruits he raised could not be concentrated in time. Sixty-six days later Pompey and the consuls sailed from Brundisium to Greece, and Caesar held Italy without a battle.

He then spent 49 BC destroying the Pompeian forces in Spain, and in 48 crossed to Greece, was beaten at Dyrrachium, and won the war at Pharsalus on 9 August. In his own account he had 22,000 infantry against Pompey's 45,000 and lost 200 men to Pompey's 15,000. Pompey fled to Egypt and was murdered on the beach as he landed. The remaining campaigns took three years: Alexandria, then Zela in Asia, then Thapsus in Africa in 46, then Munda in Spain in 45, where the dead included the last of Pompey's sons to matter.

Example. The figures for Pharsalus come from Caesar's own Commentarii, published by him during the war. How should they be used?

As the best evidence available and as advocacy. Caesar was writing for a Roman audience that contained many participants, so gross invention about the shape of the battle was risky, and his numbers for his own legions can be checked loosely against what is known of his recruitment. The ratios are where the interest lies, and they all flatter: he is always outnumbered, his losses are always trivial, and his enemies are always numerous and badly led. Asinius Pollio, who was present, is reported as giving 6,000 Pompeian dead against Caesar's 15,000. The sound method is to accept the framework, treat the casualty ratio as a claim rather than a count, and notice that a general who publishes his own dispatches has invented a political instrument as well as writing history.

Now you. Caesar took Italy in about two months without fighting a battle for it. What does that speed indicate about the state of the Republic?

What he did with it

Caesar held a series of dictatorships, briefly in 49, for a year in 48, for ten years in 46, and for life in February 44, alongside consulships in 48, 46, 45 and 44. He took the right to nominate magistrates and to command all the armies, raised the Senate to 900 by admitting Italians, provincials and some of his own centurions, and did all of it without abolishing a single institution.

The government he ran was effective in a way the Republic had not managed for decades. Veterans and the urban poor were settled in colonies overseas rather than by confiscations in Italy, which defused the quarrel that had run since the Gracchi. Debts were handled by writing down what had been paid in interest and valuing property at pre-war prices, which satisfied neither creditors nor the men who wanted debts cancelled outright, and is therefore probably about right. Cisalpine Gaul got the citizenship.

What he did not do is as striking. He wrote no new constitution, abolished no assembly, and left the consulship, the tribunate, the courts and the Senate standing, filling them with his own men rather than replacing them. The regime had no name and no rule of succession, which is why the question of what would happen when he died had no answer inside it, and why the men who killed him could believe that removing him was the whole of the task.

Reforms a republic could not pass

The most durable act is the least political. The Roman calendar had drifted so far from the seasons that the harvest festivals fell in the wrong months, and every serious Roman knew it and nobody had fixed it. Caesar, as chief priest, took Greek astronomical advice, inserted enough days to make 46 BC a year of 445 days, and started a new calendar of 365.25 days on 1 January 45.

Example. The Julian year is 365.25 days and the true tropical year is about 365.2422. Work out the accuracy of the reform and say what it demonstrates.

The error is 0.0078 days a year, which accumulates to one day in about 128 years, so the calendar was good to roughly one part in 47,000 and held with a single correction for over 1,600 years. What it demonstrates is not Caesar's genius, since the astronomy was Greek and available to anyone, but the difference a decision-maker makes. The reform required agreement across the priestly colleges, the magistrates and the assemblies, all of which had reasons to obstruct it, and under the Republic it had failed for a century. One man with unquestioned authority did it in a year. That is exactly the argument for autocracy, and it is worth stating in its strongest form, because the Romans who accepted the Principate were not fools and this is the kind of thing they were buying.

Now you. Suetonius says Caesar cut the number of people receiving subsidised grain in Rome from 320,000 to 150,000. Compute the reduction and consider how a Republican politician would have fared attempting it.

The Ides

None of that answers the question of what Caesar intended for the long run, and the honest position is that we do not know, because he was preparing to leave for a Parthian campaign of several years and had made arrangements for magistracies to be filled in his absence.

What he had done was make the permanence explicit. Dictator for life is not an emergency office, and the title said so in the one word Roman politics could not absorb. The rest, the statues, the gold chair, the coins with his portrait, the refusal to stand when the Senate approached, mattered because they pointed the same way.

On 15 March 44 BC some sixty senators, many of them men he had pardoned in the civil war, stabbed him in a meeting of the Senate. He was struck twenty-three times. The leading conspirator, Marcus Junius Brutus, was descended, or claimed to be, from the Brutus who expelled the last king in 509, and the whole act was staged as a repetition of that one, 465 years later.

The staging is also the clue to why the act failed. Removing a would-be king had worked in 509, on the tradition, because the powers he held reverted to institutions that could hold them, and in 44 BC there were no such institutions. The armies belonged to their commanders, the veterans expected land from Caesar's arrangements, the treasury and the state papers were in Antony's hands within a day, and Caesar's heir was an eighteen-year-old with a name worth legions. The conspirators had no plan for the day after, made no attempt to seize the treasury or the city, and expected the Senate to resume. Two days later that Senate voted an amnesty for the killers and confirmed all of Caesar's acts in the same session, which is a precise measure of the situation: it could not condemn the murder or undo the regime.

Nor did Caesar's clementia, his policy of pardoning defeated enemies rather than proscribing them, protect him, though it is usually contrasted favourably with Sulla's method. A pardon is an exercise of the power to kill, and every man who received one was reminded that his life was now a gift. Sulla's proscriptions removed his enemies; Caesar's clemency left them alive, in the Senate, holding office, and diminished, and several of the men who stabbed him had been pardoned at Pharsalus. Cicero saw the point exactly and wrote that the fault lay in a position where a Roman could be spared rather than acquitted. The comparison therefore cuts both ways: Sulla's method was monstrous and it worked, in that he died in his bed and his settlement outlived him by a decade, and Caesar's was humane and it killed him.

Where this leaves the Republic

The Republic on 15 March 44 BC had a dead dictator, an amnesty for his murderers, a confirmation of all his laws, a consul in possession of his papers, an heir nobody had heard of, and eleven legions in Gaul, thirteen in the east and a dozen more elsewhere, every one of them attached to a man rather than to the state.

Nobody in Rome was proposing monarchy and nobody could produce a working republic. The thirteen years that resolved that, and the settlement that replaced the res publica with something that used all its words, are the last lesson.

The Roman Republic was not abolished in 27 BC; it was declared restored, by a man who kept every legion, and almost nobody who mattered was in a position to disagree.

The previous lesson ended with Caesar dead, his acts confirmed, his murderers amnestied and every army in the empire attached to an individual. This lesson covers the thirteen years that resolved that, the settlement that followed, and the question the whole course has been building towards: what actually failed, and was it bound to?

The heir

Caesar's will adopted his great-nephew Gaius Octavius, aged eighteen, who took his name and, more usefully, his veterans' loyalty and his money. Cicero, leading the Senate against Mark Antony, thought the young man could be used and discarded, and said so in a phrase that got back to him.

In 43 BC Octavian fought for the Senate against Antony at Mutina, and when the Senate declined to reward him, he marched on Rome with eight legions and took the consulship by force. He was nineteen years old, twenty-three years under the legal minimum, and had held no office of any kind.

Then he changed sides. In November 43 he met Antony and Lepidus and they had a tribune carry the lex Titia, which created the three of them as tresviri rei publicae constituendae, a board of three for settling the Republic, for five years, with the power to make law, appoint magistrates and govern the provinces.

Example. Compare the arrangement of 43 BC with the private agreement of 60 BC and say which was more dangerous to the Republic.

The one that was legal. In 60 BC three citizens agreed privately to combine their resources and then obtained what they wanted through the ordinary machinery, which left the institutions intact and hollow. In 43 BC a law of the Roman people conferred on three named men the powers of the state for a term of years, including the right to legislate and to appoint the magistrates who were supposed to check them. The Republic did not fall to a coup here. It voted itself out of existence in a properly carried statute, which is why the second triumvirate, unlike the first, is a magistracy that historians can date and cite. A constitution with no procedure for amendment turns out to have a procedure for surrender.

Now you. The triumvirs opened with proscriptions on the Sullan model, and Cicero's name was on the first list. What was different this time?

The end of the republican cause

Brutus and Cassius had taken the eastern provinces and raised nineteen legions. At Philippi in Macedonia in October 42 BC they were beaten in two engagements and both killed themselves.

Philippi matters more than Actium for understanding what happened to the Republic, because it is the last time an army took the field for the Republic as a cause. Everything after it is a quarrel between Caesarian commanders over which of them would hold the state: Octavian against Antony's brother in the Perusine war, then against Sextus Pompeius over the grain supply until Naulochus in 36, then Lepidus discarded in the same year, then the long divorce between Octavian and Antony.

That is why the propaganda of the 30s BC is so revealing. Octavian could not present the final war as republicans against Caesarians, so he presented it as Rome against Egypt: Antony had been captured by a foreign queen, had given Roman provinces to her children at Alexandria in 34, and had left a will (which Octavian seized from the Vestals and read aloud, in itself an outrage) directing his burial in Egypt. War was declared in 32 BC on Cleopatra and not on Antony, so that a civil war could be fought as a foreign one.

They met at Actium off the west coast of Greece on 2 September 31 BC. Antony's fleet was blockaded, his army deserting; Cleopatra's squadron broke out, Antony followed, and the rest surrendered. Alexandria fell the following August, Antony and Cleopatra killed themselves, and Egypt became Roman as the personal possession of the victor.

The settlement

Octavian was now the only man with an army, and he had about sixty legions. He reduced them to twenty-eight, a cut of 53 per cent, and settled the discharged men in colonies, paying for the land in cash, which he was careful to record in his own account of his reign.

On 13 January 27 BC he came before the Senate and gave up his extraordinary powers. In the Res Gestae, the summary of his achievements published after his death, he wrote that he transferred the Republic from his own power to the discretion of the Senate and the Roman people. The Senate responded by voting him a command over Spain, Gaul, Syria, Cyprus and Egypt for ten years, and three days later by giving him the name Augustus.

Example. Test the claim of restoration against the distribution of the armies.

The provinces Augustus kept in 27 BC were precisely the ones with legions in them, and the Senate received the peaceful ones. By the end of his reign the empire had 28 legions, and exactly one, the third legion in Africa, stood in a province whose governor the Senate appointed: 96 per cent of the army answered to him. Every republican form was in place, consuls were elected annually, the Senate met, the assemblies voted, and the man who controlled 27 of 28 legions held them by a provincial command that a grateful Senate renewed whenever it expired. The claim of restoration is not a lie about the institutions, which were genuinely revived and functioning. It is a claim that omits the only variable that had decided every political question since 88 BC.

Now you. Augustus's own account says he had greater auctoritas than anyone else, but no more formal power than his colleagues in each magistracy. Is that true?

What failed

Now the question the course has been circling. The Republic ran for 482 years and broke down over the 106 years between 133 and 27 BC. What went wrong?

Five structural answers have been assembled along the way, and they compound.

Commands got too big. An empire needs multi-year operations by large forces, and the constitution could produce them only by suspending its own rules, so every solution created a man larger than the offices.

Armies stopped belonging to the state. A soldier with no property depended on his general for land at discharge, and no institution filled that role until Augustus created a military treasury in AD 6.

Disputes stopped having endings. From 133 BC the answer to a deadlock was violence, first by mobs, then by armies, and no court, veto or vote could produce a ruling that the losing side accepted.

The city could not govern the empire. Ten senior magistrates a year, no bureaucracy, tax farmed to contractors, and governors who could not be effectively supervised.

And the citizens could not physically be consulted. After the Social War the electorate ran the length of Italy while the voting enclosures stayed in one city, so the assemblies represented whoever was in Rome that day.

Structure or accident

Against all of this stands a serious objection, argued most fully by Erich Gruen: that the institutions were working reasonably well into the 50s BC, that most crises were resolved by ordinary politics, and that what destroyed the Republic was the civil war itself rather than any long decay. On that view the fall was contingent, the product of particular men and particular mistakes, and a different decision in 50 BC gives a different century.

The ancient writers had a third answer, and it should be reported rather than adopted. Sallust and Livy explained the collapse by moral decline: wealth from the conquests corrupted a people whose virtue had been sustained by poverty and by fear of Carthage. As history that is weak, since it explains a specific institutional failure by a general change in character and cannot say why the failure took the form it did. As evidence it is valuable, because it tells you how the governing class understood itself, and men who believe their state is dying of luxury make different decisions from men who believe it has a design fault.

Example. Ronald Syme argued that what happened was a revolution in the governing class, an old oligarchy replaced by a new one under a leader. Erich Gruen argued that the Republic was not visibly dying and that the war caused the fall. Can both be right?

Largely, yes, because they answer different questions. Gruen is describing the health of the institutions year by year, and his evidence is good: elections were contested, laws passed, courts sat, and most disputes ended without bloodshed even in the 50s BC. Syme is describing who ended up holding power and how, and his evidence is equally good: the consular families of the late Republic are substantially replaced within two generations by Italian and provincial newcomers attached to the ruler. A system can be functioning and fragile at once. The most defensible position is that the structural pressures made some breakdown likely and did not specify which one, and that the shape of what emerged, a monarchy in republican dress with a new elite, was decided by individuals.

Now you. What would have had to be different for the Republic to survive in a recognisable form?

What survived

It is easy to end at 27 BC with the Republic dead, and the ending is misleading in an interesting way.

The forms did not go. Consuls continued to be elected and to give their names to the year for nearly six centuries after Actium, the last of them in AD 541, which is 1,049 years after the first pair in the fasti. The Senate sat for as long. The vocabulary of the Republic, res publica, senatus populusque Romanus, magistracy, provocatio, went into the political language of Europe and has not come out: every modern republic, every constitution with a senate, a veto, a term limit or a rule against holding two offices at once, is quoting this one.

What was lost is precise. Not the institutions but the thing they existed to prevent: after 27 BC there was one man whose position no vote could end, and the question of who governed Rome was settled by his household, his adoption of an heir, and the army. That question had been answered by the citizens, badly and unequally and in a rigged assembly, for nearly five hundred years.

The last word belongs to the man who won. What Augustus claimed to have done was give the Republic back, and he chose that claim over any other because it was the only thing his Romans would accept. A regime that has to describe itself in the language of what it replaced is paying that thing a compliment, and the compliment is the most reliable evidence we have of what the Republic meant to the people who lost it.