It’s important to note at the outset that the Romans had no written constitution and indeed most of the rules for how the Roman Republic functioned were, well, customary. The Roman term for this was the mos maiorum, the “custom of the ancestors”, but Roman practice here isn’t that different from how common law and precedent guide the functioning of something like the British government (which also lacks a written constitution). Later Roman writers, particularly Cicero, occasionally offer theoretical commentary on the “rules” of the Republic (as a retrojected, ideal version), but just as often their observations do not actually conform to the practice we can observe from earlier periods. In practice, the idea here was that the “constitution” of the Republic consisting in doing things as they had always been done, or at least as they were understood to have always been done.

Consequently, as historians, we adopt the formulation that the Republic is what the Republic does – that is that one determines the rules of offices and laws based on how they are implemented, not through a hard-and-fast firm legal framework. Thus “how does the dictatorship work?” is less a question of formal rules and more a question of, “how did the eighty-odd Roman dictatorships work?”

The basic idea behind the office was that the dictator was a special official, appointed only in times of crisis (typically a military crisis), who could direct the immediate solution to that crisis. Rome’s government was in many ways unlike a modern government; in most modern governments the activities of the government are carried out by a large professional bureaucracy which typically reports to a single executive, be that a Prime Minister or a President or what have you. By contrast, the Roman Republic divided the various major tasks between a bunch of different magistrates, each of whom was directly elected and notionally had full authority to carry out their duties within that sphere, independent of any of the other magistrates. In crude analogy, it would be as if every member of the United States cabinet was directly elected and none of them reported to any of the rest of them but instead all of them were advised by Congress (but in a non-binding manner). Notionally, the more senior magistrates (particularly the consuls) could command more junior magistrates, but this wasn’t a “direct-report” sort of relationship, but rather an unusual imposition of a more senior magistrate on a less senior one, governed as much by the informal auctoritas of the consul as by law.

In that context, you can see the value, when rapid action was required, of consolidating the direction of a given crisis into a single individual. This is, after all, why we have single executive magistrates or officials in most countries. So, assuming you have a crisis, how does this process work?

The typical first step is that the Senate would issue its non-binding advice, a senatus consultum, suggesting that one or both of the consuls appoint a dictator. The consuls could ignore this direct, but almost never did (save once in 431, Liv. 4.26.5-7). The consuls would then have to nominate someone; they might agree on the choice (which would make things simple) or one of them might be indisposed (out of the city, etc.), which would leave the choice to the one that remained. If both consuls were present and did not agree, they’d draw lots to determine who got to pick (which happens in the aforementioned instance in 431 after the tribunes got the consuls of that year to relent and pick someone, Liv. 4.26.11).

The nominating consul could pick anyone except himself; if you, as consul, wanted to be dictator, you would need your co-consul to so nominate you. There were no formal requirements; of course nominations tended to go to experienced commanders, which tended to mean former consuls, but this was not a requirement. Publius Claudius Pulcher (cos. 249), enraged when the Senate directed him to appoint a dictator (because of his own bungled military command) infamously nominated his own freedman, Claudius Glicia, as dictator (Liv. Per. 19.2; Seut. Tib. 2.2), which was apparently a bridge too far; Glicia was forced to abdicate but his name was duly entered onto the Fasti because the appointment was valid, if ill-advised (despite the fact that, as a freedman, Glicia would have been ineligible to run for any [office]). Nevertheless, dictators were usually former consuls.

Once the name was picked, in at some cases the appointment may have been confirmed by a vote of the Comitia Curiata, Rome’s oldest voting assembly, which was responsible for conferring imperium (the power to command armies and organize law courts; essentially “the power of the kings”) on magistrates; not all magistrates had imperium (consuls, praetors, proconsuls, propraetors, dictators and their magistri equitum did; quaestors, aediles, tribunes, both plebeian and military, and censors did not). We do not know of any instance where the Comitia Curiata put the kibosh on the appointment of a dictator, so this step was little more than a rubber-stamp, and may have been entirely optional (Lintott, op. cit., 110, n. 75), but it may have also reflected the notion that all imperium had to be conferred by the people through a voting assembly. It is often hard to know with clarity about pro forma elements of Roman politics because the sources rarely report such things.

The dictator was appointed to respond to a specific issue or causa, the formula for which are occasionally recorded in our sources. The most common was rei gerundae causa, “for the business to be done” which in practice meant a military campaign or crisis. In cases where the consuls were absent (out on campaign), a dictator might also be nominated comitiorum habendorum causa, “for having an assembly”, that is, to preside over elections for the next year’s consuls, so that neither of the current consuls had to rush back to the city to do it. Dictators might also be appointed to do a few religious tasks which required someone with imperium. Less commonly but still significantly, a dictator might be appointed seditionis sedenae causa, “to quell sedition”; only one instance clearly under this causa is known, P. Manlius Capitolinus in 368, but several other instances, e.g. L. Quinctius Cincinnatus in 439, also dealt with internal matters. Finally, once in 216, Marcus Fabius Buteo held the office of dictator senatus legendi causa, “to enroll the Senate”, as the Battle of Cannae, earlier that year, had killed so many Senators that new inductions were needed (Liv. 23.23).

The dictator then named a subordinate, the magister equitum (“master of the horse”). The magister equitum was a lieutenant, not a colleague, but interestingly once selected by a dictator could not be unselected or removed, though his office ended when the dictator laid down his powers. We should note Marcus Minucius, magister equitum for Q. Fabius Maximus in 217 as an exception; his selection was forced by the people via a law and his powers were later made equal to Fabius’ powers. This turned out to be a substantial mistake, with Fabius having to bail the less prudent Minucius out at Geronium – the undermining of Fabius generally during 217 was, in retrospect viewed as a disaster, since the abandonment of his strategy led directly to the crushing defeat at Cannae in 216.

One of the ways that legal power was visually communicated in Rome was through lictors, attendants to the magistrates who carried the fasces, a bundle of rods (with an axe inserted when outside the sacred bounds of the city, called the pomerium). More lictors generally indicated a greater power of imperium (consuls, for instance, could in theory give orders to the praetors). Praetors were accompanied by six lictors; consuls by 12. The dictator had 24 lictors when outside of the pomerium to indicate his absolute power in that sphere (that is, in war), but only 12 inside the city. The magister equitum, as the dictator’s subordinate, got only six, like the praetors.

It also seems fairly clear that while dictators had almost complete power within their causa, those powers didn’t necessarily extend beyond it (e.g. Liv. 2.31.9-11, the dictator Manius Valerius, having been made dictator to resolve a military problem, insists to the Senate that he cannot resolve internal strife through his dictatorial powers and instead lays down his office early). The appointment of a dictator did not abolish the other offices (Cicero thinks they do, but he is clearly mistaken on the matter, Cic. De Leg. 3.9, see Lintott, 111). In essence then, the dictator was both a supreme military commander and also expected to coordinate the other magistrates with his greater degree of imperium, though of course in practice the ability to do that is going to substantially depend on the individual dictator’s ability to get cooperation from the other magistrates (but then, on the flip side, the dictator has just been designated as the leader in the crisis, so the social pressure to conform to his vision must have been intense). Notably, dictators could not make a law (a lex) on their own power or legislate by fiat outside of their causa; they could and did call assemblies which could by vote approve laws proposed by a dictator, however.

The dictator served for six months or until the task for which he was appointed was resolved, whichever came first. There is a tendency in teaching Roman history to represent a figure like Cincinnatus, who laid down his dictatorship after just fifteen days in 458, as exceptional but while the extreme shortness of term was exceptional, laying down power was not. Indeed, Cincinnatus (or perhaps a relative) served as dictator again in 439 and again laid down his power, this time in merely 21 days. In practice, the time-limited nature of the dictatorship meant there were few incentives to “run out the clock” on the office since it was so short anyway – better, politically, to solve the crisis quickly and lay down power ostentatiously early and “bank” the political capital than try to run out the period of power, accomplish relatively little and squander a reputation for being public-spirited.

Bret Devereaux, “Collections: The Roman Dictatorship: How Did It Work? Did It Work?”, A Collection of Unmitigated Pedantry, 2022-03-18.