The fall of the Roman Empire in the West (please, right now, just mentally add the phrase “in the west” next to every “the fall of Rome” and similar phrase here and elsewhere) is complicated. I don’t mean it is complicated in its causes or effects (though it is that too), I mean it is complicated in its raw events: the who, what, where and when of it. Most students are taught a fairly simple version of this because most of what they need to actually learn is the cause and the effects and so the actual “fall” part is a sort of black box where Huns, Vandals, Goths, plague, climate and economic decline go in and political fragmentation, more economic decline and the European Middle Ages come out. The fall itself ends up feeling like an event rather than a process because it is compressed down to a single point, the black box where all of the causes become all of the effects. That is, frankly, a defensible way to teach the topic at a survey level (where it might get at most a lecture or two either at the end of a Roman History survey or the beginning of a Medieval History survey) and it is honestly more or less how I teach it.
But if you want to actually try to say something intelligent about the whole thing, you need to grapple with what actually happened, rather than the classroom black-box model designed for teaching efficiency rather than detail. We are … not going to do that today … though I will have some bibliography here for those who want to. The key thing here is that the “Fall of Rome” (in the West) is not an event, but a century long process from 376 to 476. Roman power (in the West) contracts for a lot of that, but it expands in periods as well, particularly under the leadership of Aetius (433-454) and Majorian (457-461); there are points where it would have really looked like the Romans might actually be able to recover. Even in 476 it was not obvious to anyone that Roman rule had actually ended; Odoacer, who had just deposed what was to be the last Roman emperor in the west promptly offered the crown to Zeno, the Roman emperor in the East (there is argument about his sincerity but James O’Donnell argues – very well, though I disagree on some key points – that this represented a real opportunity for Rome to rise from defeat in a new form yet again).
Glancing even further back historically, this wasn’t even the first time the Roman Empire had been on the brink of collapse. Beginning in 238, the Roman Empire had suffered a long series of crippling civil wars and succession crises collectively known as the Crisis of the Third Century (238-284). At one point, the empire was de facto split into three, with one emperor in Britain and Gaul, another in Italy, and the client kingdom of Palmyra essentially running the Eastern half of the empire under their queen Zenobia. Empires do not usually survive those kinds of catastrophes, but the Roman Empire survived the Crisis, recovered all of its territory (save Dacia) and even enjoyed a period of relative peace afterwards, before trouble started up again.
The reason that empires do not generally survive those kinds of catastrophes is that generally when empires weaken, they find that they contain all sorts of people who have been waiting, sometimes patiently, sometimes less so, for any opportunity to break away. The rather sudden collapse of the (Neo-)Assyrian Empire (911-609 BC) is a good case study. After having conquered much of the Near East, the Assyrians fell into a series of succession wars beginning in 627; their Mesopotamian subjects smelled blood and revolted in 625. That was almost under control by 620 when the Medes and Persians, external vassals of the Assyrians, smelled blood too and invaded, allying with the rebelling Babylonians in 616. Assyria was effectively gone by 612 with the loss and destruction of Ninevah; they had gone from the largest empire in the world at that time or at any point prior to non-existent in 15 years. While the Assyrian collapse is remarkable for its speed and finality, the overall process is much the same in most cases; once imperial power begins to wane, revolt suddenly looks more possible and so the downward slope of collapse can be very steep indeed (one might equally use the case study of decolonization after WWII as an example: each newly independent country increased the pressure on all of the rest).
Yet there is no great rush to the doors for Rome. Instead, as Guy Halsall puts it in Barbarian Migrations and the Roman West (2007), “The West did not drift hopelessly towards its inevitable fate. It went down kicking, gouging and screaming”. Among the kicked and gouged of course were Attila and his Huns. Fought to a draw at the Battle of the Catalaunian Plains, his empire disintegrated after his death two years later under pressure from both Germanic tribes and the Eastern Roman Empire (and the standard tendency for Steppe empires to fragment); of his three sons, Ellac was killed by revolting Germanic peoples who had been subject to the Huns, Dengizich by the (Eastern) Romans (we’re told his head was put on display in Constantinople) and the last, Ernak just disappears in our narrative after the death of Dengizich. The Romans, it turns out, did eventually get down to business to defeat the Huns. But the Romans doing all of that kicking, gouging and screaming were not the handful of old families from the early days of the Repulic; most of those hard-fighting Romans were people who in 14 AD would have been provincials. And indeed, the Roman Empire would survive, in the East, where Rome wasn’t, making for a Roman Empire that by 476 consisted effectively entirely of “provincial” Romans.
Instead what we see are essentially three sets of actions by provincial elites who in any other empire would have been leading the charge for the exits. There were the kickers, gougers and screamers, as Halsall notes. There were also, as Ralph Mathisen, Roman Aristocrats in Barbarian Gaul (1993) has noted, elites who – seeing the writing on the wall – made no effort to hasten the collapse of the empire but instead retreated into their estates, their books and their letters; these fellows often end up married into and advising the new “barbarian” kings who set up in the old Roman provinces (which in turn contributes quite a bit to the preservation and continued influence of Roman law and culture in the various fragmented successor states of the early Middle Ages). Finally, there were elites so confident that the empire would survive – because it always had! – that they mostly focused on improving their position within the empire, even at the cost of weakening it, not because they wanted out, but because “out” was inconceivable to them; both Halsall and also James O’Donnell, The Ruin of the Roman Empire (2009) document many of these. If I may continue my analogy, when the exit door was yawning wide open, almost no one walked through; some tried to put out the burning building they were in, others were content to be at the center of the ruins. But no one actually left.
During the Crisis of the Third Century, that set of responses had been crucial for the empire’s survival and for brief moments in the 400s, it looked like they might even have saved it again. For all of the things that brought the Roman Empire down, it is striking that “internal revolts” of long-ruled peoples weren’t one of them. And that speaks to the power of Rome’s effective (if, again, largely unintentional) management of diversity. The Roman willingness to incorporate conquered peoples into the core citizen body and into “Roman-ness” meant that even by 238 to the extent that the residents of the Empire could even imagine its collapse, they saw that potentiality as a disaster, rather than as a liberation. That gave the empire tremendous resiliency in the face of disaster, such that it took a century of unremitting bad luck to bring it down and even then, it only managed to take down half of it.
(As an aside, those provincial Romans were correct in the judgement that the collapse of the empire would mean disaster. The running argument about the fall of the Roman Empire is generally between the “decline and fall” perspective, which presents the collapse of the Roman Empire as a Bad Thing and the “change and continuity” perspective, which both stresses continuity after the collapse but also tends to try minimize the negative impacts of it, even to the point of suggesting that the average Roman peasant might have been better off in the absence of heavy Roman taxes. That latter view is particularly common among many medievalists, who are understandably quite tired of the unfairly poor reputation their period gets. This is an argument that for some time lived in the airy space of narrative and perspective where both sides could put an argument out. Unfortunately for some of the change-and-continuity arguments about living standards, archaeology has a tendency to give us data that is somewhat less malleable. That archaeological data shows, with a high degree of consistency, that while there is certainly some continuity between the Late Antique and the early Middle Ages the fall of Rome (in the West) killed lots of people (precipitous declines in population in societies without reliable birth control; probably this is mostly food scarcity, not direct warfare) and that living standards also declined to a degree that the results are archaeologically visible. As Brian Ward-Perkins notes in The Fall of Rome and the End of Civilization (2005), the collapse causes cows to shrink, speaking to sudden scarcity of winter fodder (which in turn likely speaks to a general reduction in available nutrition). Some areas were worse hit than others; Robin Flemming, Britain After Rome (2010) notes, for instance, that in post-Roman Britain, pot-making technology was lost (because ceramic production had been focused in cities which had been largely depopulated out of existence). The fall of Rome might have been good for some people, but the evidence is, I think, at this point inescapable that it was quite bad for most people. Especially, one assumes, all of the people who got depopulated.)
Bret Devereaux, “Collections: The Queen’s Latin or Who Were the Romans, Part V: Saving and Losing and Empire”, A Collection of Unmitigated Pedantry, 2021-07-30.
September 10, 2025
QotD: “The [western Roman Empire] did not drift hopelessly towards its inevitable fate. It went down kicking, gouging and screaming”
September 4, 2025
QotD: The development of the “halftrack” during the interwar period
What drove the differences in tank development between countries were differences between how each of those countries imagined using their tanks, that is differences in tank doctrine. Now we should be clear here that there were some fundamental commonalities between the major schools of tank thinking: in just about all cases tanks were supposed to support infantry in the offensive by providing armor and direct fire support, including knocking out enemy tanks. Where doctrine differed is exactly how that would be accomplished: France’s doctrine of “Methodical Battle” generally envisaged tanks moving at the speed of mostly foot infantry and being distributed fairly evenly throughout primarily infantry formations. That led to tanks that were fairly slow with limited range but heavily armored, often with just a one-man turret (which was a terrible idea, but the doctrine reasoned you wouldn’t need more in a slow-moving combat environment). Of course this worked poorly in the event.
More successful maneuver warfare doctrines recognized that the tank needed infantry to perform its intended function (it has to have infantry to support) but that tanks could now move fast enough and coordinate well enough (with radios) that any supporting arms like infantry or artillery needed to move a lot faster than walking speed to keep up. Both German “maneuver warfare” (Bewegungskrieg) and Soviet “Deep Operations” (or “Deep Battle”) doctrine saw the value in concentrating their tanks into powerful striking formations that could punch hard and move fast. But tanks alone are very vulnerable and in any event to attack effectively they need things like artillery support or anti-air protection. So it was necessary to find ways to allow those arms to keep up with the tanks (and indeed, a “Panzer divsion” is not only or even mostly made up of tanks!).
At the most basic level, one could simply put the infantry on trucks or other converted unarmored civilian vehicles, making “motorized” infantry, but […] part of the design of tanks is to allow them to go places that conventional civilian vehicles designed for roads cannot and in any event an unarmored truck is a large, vulnerable tempting target on the battlefield.
The result is the steady emergence of what are sometimes jokingly called “battle taxis” – specialized armored vehicles designed to allow the infantry to keep up with the tanks so that they can continue to be mutually supporting, while being more off-road capable and less vulnerable than a truck. In WWII, these sorts of vehicles were often “half-tracks” – semi-armored, open-topped vehicles with tires on the front wheels and tracks for the back wheels, though the British “Universal Carrier” was fully tracked. Crucially, while these half-tracks might mount a heavy machine gun for defense, providing fire support was not their job; being open-topped made them particularly vulnerable to air-bursting shells and while they were less vulnerable to fire than a truck, they weren’t invulnerable by any means. The intended use was to deposit infantry at the edge of the combat area, which they’d then move through on foot, not to drive straight through the fight.
The particular vulnerability of the open-top design led to the emergence of fully-enclosed armored personnel carriers almost immediately after WWII in the form of vehicles like the M75 Armored Infantry Vehicle (though the later M113 APC was eventually to be far more common) and the Soviet BTRs (“Bronetransporter” or “armored transport”), beginning with the BTR-40; Soviet BTRs tended to be wheeled whereas American APCs tend to be tracked, something that also goes for their IFVs (discussed below). These vehicles often look to a journalist or the lay observer like a tank, but they do not function like tanks. The M113 APC, for instance, has just about 1.7 inches of aluminum-alloy armor, compared to the almost four inches of much heavier steel armor on the contemporary M60 “Patton” tank. So while these vehicles are armored, they are not intended to stick in the fight and are vulnerable to much lighter munitions than contemporary tank would be.
At the same time, it wasn’t just the infantry that needed to be able to keep up: these powerful striking units (German Panzer divisions, Soviet mechanized corps or US armored divisions, etc.) needed to be able to also bring their heavy weaponry with them. At the start of WWII, artillery, anti-tank guns and anti-air artillery remained almost entirely “towed” artillery – that is, it was pulled into position by a truck (or frequently in this period still by horses) and emplaced (“unlimbered”) to be fired. Such systems couldn’t really keep up with the tanks they needed to support and so we see those weapons also get mechanized into self-propelled artillery and anti-air (and for some armies, tank destroyers, although the tank eventually usurps this role entirely).
Self-propelled platforms proved to have another advantage that became a lot more important over time: they could fire and then immediately reposition. Whereas a conventional howitzer has to be towed into position, unlimbered, set up, loaded, fired, then limbered again before it can move, something like the M7 Priest can drive itself into position, fire almost immediately and then immediately move. This maneuver, called “shoot-and-scoot” (or, more boringly, “fire-and-displace”) enables artillery to avoid counter-battery fire (when an army tries to shut down enemy artillery by returning fire with its own artillery). As artillery got more accurate and especially with the advent of anti-artillery radars, being able to shoot-and-scoot became essential.
Now while self-propelled platforms were tracked (indeed, often using the same chassis as the tanks they supported), they’re not tanks. They’re designed primarily for indirect fire (there is, of course, a sidebar to be written here on German “assault guns” – Sturmgeschütz – and their awkward place in this typology, but let’s keep it simple), that is firing at a high arc from long range where the shell practically falls on the target and thus are expected to be operating well behind the lines. Consequently, their armor is generally much thinner because they’re not designed to be tanks, but to play the same role that towed artillery (or anti-air, or rocket artillery, etc.) would have, only with more mobility.
So by the end of WWII, we have both tanks of various weight-classes, along with a number of tank-like objects (APCs, self-propelled artillery and anti-air) which are not tanks but are instead meant to allow their various arms to keep up with the tanks as part of a combined arms package.
Bret Devereaux, “Collections: When is a ‘Tank’ Not a Tank?”, A Collection of Unmitigated Pedantry, 2022-05-06.
August 29, 2025
QotD: The early “Motte and Bailey” castles
But as noted, the personal manor home of a significant noble (the rank in this case is often a “castellan”, literally the keeper of a castle, so entrusted by one of the more powerful nobles who holds sway over a larger territory; the castellan has the job of holding the castle and administering the countryside around it) is also an administrative center, managing the extraction of agricultural surplus from the countryside and also a military base, housing the physical infrastructure for that noble’s retinue, which again is the fundamental building block of larger armies. Which means that it is going to need more structures to house those functions: stables for horses, storehouses for food, possibly food processing facilities (bakeries, mills) and living space both for retainers (be they administrators or military retainers) and for the small army of servants such a household expects. Those structures (to the degree they can’t exist in the keep) are put in the bailey, a wider enclosed part of the settlement constructed at the base of the motte. As with the motte, the bailey is typically enclosed only by a wooden palisade; naturally that means the most valuable things (the physical treasury, the lord’s family) go in the keep on the motte, while the more space-demanding but less valuable things go in the bailey. There is a lot of room for variation in this basic type, but for now the simple version will serve.
The resulting fortification seems almost paradoxically vulnerable. The bailey, after all, is protected only by a ditch and a wooden palisade which a determined work-party could breach with just iron axes and an afternoon to kill. The core defensive motte with its keep adds perhaps only one more palisade and a steep climb. But in fact, these relatively modest defenses have greatly increased the cost of attacking this settlement. The motte and bailey castle, at least in its early wooden form, won’t stand up to a determined assault by a large and well-coordinated enemy, but that isn’t its purpose. Instead, the purpose of the motte and bailey castle is to raise the cost of an assault such that a potential opponent must bring a significant force and make a careful, well-planned assault; this the motte and bailey accomplishes quite well, which explains the long durability of the basic design, with stone versions of the motte and bailey persisting into the 15th century.
The quick mounted raid is now impossible; precisely because it will take a solid afternoon to breach the defenses, there is little hope of surprising the defenders. At the same time, the ditches will make any such work party vulnerable to missile fire (arrows, yes, but also javelins or just large rocks) from the palisade. And most of all, taking the place now demands you coordinate a work party, with some of your attackers splitting up to suppress the defenders, some making sure to block the exits so the defenders don’t rush out and attack your work party directly, and still more of your attackers in the work party itself. These very basic defenses have suddenly taken you from a position where a bit of surprise and rough numerical parity was enough to contemplate an assault to a position where you need several times as many attackers (for each of those divisions needs to be large enough to confidently win against the defenders if assailed).
Perhaps most importantly, the basic structure of this defense demands that you do this multiple times in sequence. We’ve already discussed the value of defense-in-depth, but in brief, every attack is at its strongest in the moment after it jumps off: everyone is alive, in the right positions, at the right time, coordinated and at least in theory clear on their objectives. Every movement and action beyond this point diminishes the power of the effort as coordination breaks down, attackers are killed and things break; this is what Clausewitz terms (drink!) friction – the unpredictable interaction of probabilities takes their toll on any plan, no matter how carefully designed. This is, by the by, more true in real warfare, where coordination is limited by communications technology, than it is in film or video games, where armies appear to mostly communicate by some form of instantaneous telepathy (it is amazing just how many clever sounding movie or game assault plans fall apart once you imagine trying to coordinate them with nothing more than shouting, or even a radio). As more and more things turn out unexpectedly or have to be improvised, the plan slowly shakes apart until eventually all of the momentum is lost.
The basic structure of a motte and bailey castle exploits this feature of warfare, forcing an attacker to overcome a series of obstacles in sequence, all while in contact with the enemy. Recall that this is a defense which really doesn’t envisage enemy artillery (because armies with lots of effective siege artillery were not common in the often small-scale warfare of the period; that’s not to say they didn’t exist, but if your motte and bailey castle forces the enemy to only attack with a big, expensive army that can build catapults, it has done its job, not the least because most possible enemies won’t have that capability at all), so an attacker is going to have to breach each layer in sequence while in contact with the defense and to pierce them all more or less “in one go”. Consequently, taking the castle by storm means crossing (and probably filling in) at least one deep ditch, breaching a palisade under fire, then moving up a steep hill under fire, then breaching another palisade, at the end of all of which, the attacker must arrive at the keep with enough force and cohesion to take it. All of that is going to take a substantial attack and a lot of coordination and most potential attackers, the defender may hope, will lack either the resources or the determination to go through so much effort, especially as they are likely to have to do it multiple times: being entirely wooden, motte and bailey castles were fairly cheap and so a large territory could have quite a lot of them (note on the Bayeux Tapestry how William has to take several such castles in order to capture Conan II of Britanny). Each motte and bailey castle thus raises the cost of trying to seize control of the territory; collectively they make that cost prohibitive.
Of course our principle of “antagonistic co-evolution” is not done and the vulnerabilities of a wooden motte and bailey castle are fairly clear and easy to exploit. For one, the wooden palisade is mostly a blocking element, rather than a fighting position; attackers that reach the wall can actually use it as cover while tearing it down or setting it on fire. The entire setup, being made of wood, is vulnerable to fire but also to any kind of even-quite-modest catapult. And quite naturally, any military leader (which is to say, the military aristocracy which was emerging at the very same time as these castles) is going to want to build the kind of capabilities which will allow for successful castle assaults because, as we’ve already noted, castles function more or less as the “nails” on the map which hold down the canvas of revenue extraction and military power.
Which in turn means evolving castle design to resist the methods by which a motte and bailey castle might fall. The most immediate change is in building material: wooden walls can only be so high, so thick and so resistant to fire. Stone, though far more expensive, offers advantages on all three fronts. And so, already in the late 10th century, we start to see stone keeps and gatehouses (supporting still wooden palisades); full stone castles would soon follow.
As an aside, one solution to this problem which doesn’t much appear in the Middle Ages but was very well-used in Iron Age Europe was what the Romans called the murus Gallicus, a hybrid wood-and-stone wall system. Gallic hillforts (called oppida) were built on hills, as the name suggests; their outer walls could be built by using earth fill to construct what was essentially a retaining wall, faced in stone, with transverse reinforcing wood beams every few feet. That created, in turn, a vertical stone surface, supported by the hillside itself, on which could be additionally built a wooden palisade for added height. The result was a very formidable fortification, assuming one had the hill to work with initially. You couldn’t knock it over or really undermine it effectively and the stone face was nearly vertical; the height of the hill meant that effective escalade meant coming up with a mole, tower or ladder taller than the hill (a thing, naturally, that the Romans ended up doing). That this style of fortification didn’t really reemerge in the Middle Ages speaks to the degree of path dependence in fortification design. Because fortification design tends to be evolutionary, it is possible in similar conditions to get very different responses as different designers try to meet the same threats by modifying different preexisting systems of fortification.
Bret Devereaux, “Collections: Fortification, Part III: Castling”, A Collection of Unmitigated Pedantry, 2021-12-10.
August 23, 2025
QotD: The background of Tiberius Gracchus
Meanwhile, as D.J. Gargola has noted, Appian is also bending his account of Tiberius Gracchus’ reforms, in particular by presenting the Lex Sempronia Agraria as an entirely traditional, conventional response to a pressing crisis.1 But in fact, the provisions of the Lex Sempronia Agraria were not traditional: no similar law (save for a re-enactment by Gaius Gracchus) – had ever or would ever be passed in Rome and the legal precedent that Appian presents as providing the foundation for Tiberius’ law appears to be at least substantially an anachronistic invention. Meanwhile, the crisis Appian thinks Tiberius Gracchus thought he was addressing probably didn’t exist in the form he understood it.
But that’s what we have, so it is what we must work with. And we should note that both Plutarch and Appian are quite favorable to the Gracchi, even though both men were clearly very controversial in their day. So in a sense this is a reverse of the situation we had with Cleopatra, where we had to contend with relentlessly negative sources: here the sources are broadly positive.
So, on with what we know.
Tiberius Gracchus was elected tribune in 133. His election was already unusual in that he seems to have run on something like a program (land reform, which we’ll get to); Romans generally ran on character and background rather than promising specific political actions if elected, so this was unusual. Part of the reason for it was doubtless that Tiberius Gracchus’ political fortunes were in difficulties. Now we should note here that while Tiberius Gracchus was a plebian (that is, not a patrician) that doesn’t make him a political outsider: Tiberius Gracchus was not remotely a political outsider or poor man or lacking in influence. His father (also Ti. Sempronius Gracchus) had been consul in 177 and 163 and censor in 169; his father (or grandfather) was consul in 215 and 213. Our Tiberius Gracchus’ mother, Cornelia, was the daughter of P. Cornelius Scipio Africanus, the man who defeated Hannibal. Tiberius Gracchus was born into substantial wealth and influence, the sort of man whose eventual political ascent was almost guaranteed.
(Indeed, it was so guaranteed that he gets to bend the rules and hold many of his offices early. He’s quaestor at just 26, which implies that he started his military service at 15 or 16 instead of the normal 17, doing so as a military tribune, not a common soldier. I do think this is relevant to understanding Tiberius Gracchus: this was a man born with a silver spoon and a carefully paved, flat-and-easy road to power and influence laid out for him by his family and his political backers, the most notable among whom was his key supporter Scipio Aemilianus (destroyer of Carthage and shortly Numantia).)
Except. Except he got wrapped up in something of a nasty foreign policy scandal during his year as quaestor, when he was assigned to the amazingly named but less amazingly capable C. Hostilius Mancinus who as consul in 137 was supposed to deal with Numantia in Spain. Mancinus blew it and got his army effectively trapped and sent Tiberius – his quaestor and the next highest ranking Roman present – to negotiate to get his army out. Tiberius did this, but the whole thing caused a great stink and a scandal at Rome (Roman armies are supposed to go down fighting, not negotiate shameful retreats!). Indeed, the Senate was so enraged they rejected the treaty and instead sent Mancinus, bound in chains, to the Numantines as part of a ritual process by which his treaty was disowned. Tiberius doesn’t get packed off to Numantia, but some of the political stink does rub off on him, so while he’s connected enough to get elected as a plebeian tribune in 133, he must know he needs a big second act to get his political career back on track, or he may never reach the consulship. That context – a political insider who had a golden ticket but must now win it back, rather than an outsider without connections – is important for understanding the reaction he is going to get.
Bret Devereaux, “Collections: On the Gracchi, Part I: Tiberius Gracchus”, A Collection of Unmitigated Pedantry, 2025-01-17.
- 1. D.J. Gargola, “The Gracchan Reform and Appian’s Representation of an Agrarian Crisis” in People, Land and Politics, eds. L. De Ligt and S.J. Northwood (2008).
August 17, 2025
QotD: The benefits of using auxilia units to the Roman Empire
The most obvious of these is manpower. Especially with a long-service professional army, capable and qualified recruits are in limited supply. The size of the Roman army during the imperial period ranged from around 300,000 to around 500,000, but in 14 AD (the year of Augustus‘ death) there were only 4,937,000 Roman citizens (Res Gestae 8.11), a figure which probably (a word I am using to gloss over one of the most technical and complex arguments in the field) includes women and children. Needless to say, keeping something close to a fifth of the adult male citizen population under arms continually, forever was simply never going to be feasible. After his victory in 31 BC at Actium, Octavian (soon to be Augustus) had acted quickly to pare down the legions, disbanding some, merging others, until he reached a strength of just 28 (25 after the three legions lost in 9 AD were not replaced). It was a necessary move, as the massive armies that had been raised during the fever-pitch climax of the civil wars simply could not be kept under arms indefinitely, nor could a short-term service conscript army be expected to garrison the hundreds of miles of Roman limes (“frontier, border”) in perpetuity.
Harnessing the manpower of the provinces was simply the necessary solution – so necessary that almost every empire does it. By their very nature, empires consist of a core which rules over a much larger subject region, typically with far greater population; securing all of that territory almost always requires larger forces than the core’s population is able or willing to provide, leading to the recruitment of auxiliaries of all kinds. But whereas many imperial auxiliaries, as noted above, turn out to be potential dangers or weaknesses, Rome’s auxilia seem to have been fairly robustly “bought in” on the system, allowing Rome to access motivated, loyal, cohesive and highly effective manpower, quite literally doubling the amount of military force at their disposal. Which in turn mattered a great deal because the combat role of the auxilia was significant, in stark contrast to many other imperial armies which might use auxiliaries only in subsidiary roles.
The auxilia also served to supply many of the combat arms the Romans themselves weren’t particularly good at. The Romans had always performed very well as heavy infantry and combat engineers, but only passably as light infantry and truly poorly as shock cavalry; they generally hadn’t deployed meaningful numbers of their own missile cavalry or archers at all. We’ve already talked a lot about how social institutions and civilian culture can be important foundational elements for certain kinds of warfare, and this is no less true with the Romans. But by recruiting from subject peoples whose societies did value and practice the kinds of warfare the Romans were, frankly, bad at, the Roman skill-set could be diversified. And early on, this is exactly what we see the auxilia being used for (along with also providing supplemental heavy infantry), with sagitarii (archers), funditores (slingers), exploratores (scouts) and cavalry (light, heavy and missile), giving the Romans access to a combined arms fighting force with considerable flexibility. And the system clearly works – even accounting for exaggerated victories, it is clear that Roman armies, stretched over so long a frontier, were both routinely outnumbered but also routinely victorious anyway.
As Ian Haynes notes, the ethnic distinctiveness of various auxilia units does not seem to have lasted forever, though in some cases distinctive dress, equipment and fighting styles lasted longer. Most auxilia were posted far from their regions of origin and their units couldn’t rely on access to recruits from their “homeland” to sustain their numbers over the long haul (although some number of recruits would almost certainly come from the military families of veterans settled near the forts). But that didn’t mean the loss of the expertise and distinctive fighting styles of the auxilia. Rather skills, weapons and systems which worked tended to get diffused through the Roman army (particularly in the auxilia, but it is hard not to notice that eventually the spatha replaces the gladius as the sword of the legions). As Ovid quips, Fas est et ab hoste doceri, “It is right to learn, even from the enemy” (Met. 4.428); the Romans do that a lot. The long-service professional nature of these units presumably made a lot of this possible, with individual cohortes and alae becoming their own pockets of living tradition in the practice of various kinds of fighting and acclimating new recruits to it. Consequently, not only did the Roman army get access to these fighting-styles, because the auxilia were actually integrated into the military system rather than merely attached to it, they also got the opportunity to adopt or imitate the elements of the fighting styles that worked.
Finally, the auxilia system also minted new Romans. We’ve already mentioned that auxilia veterans received Roman citizenship on retirement, but that wasn’t the extent of it. We can see in inscriptions that the degree of cultural fluency that soldiers in the auxilia gained with Roman culture was high; they often adopted Roman or Romanized names and seem to have basically always learned Latin (presumably because their Roman officers wouldn’t have spoken their language). While some units of the auxilia kept distinctive national dress as a sort of uniform, most of the auxilia seem to have adopted a style of dress that, while distinct from the legions, was generally in keeping with the Roman tradition of military dress (which was not quite the same as Roman civilian dress). They also partook of the Roman military diet (Roman soldiers kept a similar diet all over the empire, even if that meant shipping thousands of amphora of olive-oil and sour wine to northern England) which would have given them a diet in common with many work-a-day Romans too. Once retired, auxilia soldiers tended to settle where they served (rather than returning to their “home” provinces), which meant settling in frontier provinces where their citizenship set them apart as distinctively Roman, wherever they may have come from.
Exactly how many auxilia would have retired like this requires a degree of number crunching. Given a 20-year tour of service and zero mortality, we might expect around 7,500 men to pass through the auxilia each year. But of course, mortality wasn’t zero and so we have to expect that of our c. 20-year-old recruits, some number are going to die before retirement. Using some model life tables (following B. Frier, “Demography” in CAH^2 XI (2000)), we should figure that very roughly one third of our recruits will have died before reaching discharge. We then we need to adjust our recruitment figures to retain the same total strength and we get something like 9,000 new recruits each year to keep a strength of c. 150,000 with mortality counted for and 20 year tours. That gives us roughly 6,000 auxilia living to retirement each year. That may seem a small number, but that gradual accretion matters when it runs for decades and centuries and the newly enfranchised family units (recall that the citizenship grant covers children and sort-of-kind-of his spouse1) tend to settle on the frontiers, which is a really handy place to have communities of citizens. If we assume that these new citizen families mostly reproduced themselves (or more correctly that they went extinct or split with multiple children at roughly the same rate with no natural population growth), then we’d expect this process to produce perhaps something like 1.5 million new citizen households up until the Constitutio Antoniniana. Being very back of the envelope then, we might – once we account for women and children descendants of those soldiers – assume that on the eve of the general grant of citizenship in 212, there were perhaps 4 million Romans whose citizenship status was a product of service in the auxilia somewhere in their history; perhaps representing something like 7% of the entire population (including non-free persons). Were we to assume larger households (which seems wise, given that retired auxiliaries are probably more likely than average to be in an economic position to have a larger family), that figure would be even higher.
That is a very meaningful number of new Romans. And those figures don’t account for some of the other ways Roman citizenship tended to expand through communities both through manumission but also the political networks citizenship created (your Latin-speaking former-auxiliary citizen neighbors are a lot more likely to be able to help intercede to get you citizenship or get your community recognized as a municipia with that attendant citizenship grant). And not only are those new Romans by legal status, but new Romans who have, by dint of military training and discipline, absorbed quite a lot of Roman culture. As best we can tell, they tended to view the Roman Empire as their polity, rather than as a foreign or oppressive entity. They were “bought in” as it were. Again, this does not seem to have been the Roman intent, but rather an opportunistic, self-serving response to the need to maintain the loyalty of these troops; citizenship was, after all, a free benefit the emperor might bestow at no cost to the treasury (since citizens who lived outside of Italy still owed taxes) or himself.
Of course that fits the auxilia in to a later pattern in the provinces which becomes perhaps most apparent as the Roman Empire begins to collapse …
Bret Devereaux, “Collections: The Queen’s Latin or Who Were the Romans, Part V: Saving and Losing and Empire”, A Collection of Unmitigated Pedantry, 2021-07-30.
- Note on the coverage of the spouse. The grant of citizenship covered any biological children of the discharged auxiliary but did not extend citizenship to his wife. It did however, give an auxiliary the right to contract a lawful marriage with effectively any free woman, including non-citizens and the children resulting from such a union would be citizens themselves. Consequently, it extended one of the core privileges of citizenship to the non-citizen wife of a discharged auxiliary: the right to bear citizen children. Since the wife would be part of the retired auxiliary’s household (and then later, if he predeceased her, potentially in the household of her male citizen children) she’d be legally covered in many cases because a legal action against her would generally be an action against her husband/child. Given that a number of the rights of citizens simply didn’t apply to women in the Roman world (e.g. office holding), this system left the wife of a retired auxiliary with many, but not all, of the privileges of citizenship, so long as her husband and her marriage survived. That said, the legal status remained vested in her husband or her children, which made it more than a little precarious. One of these days, we can talk more about the structure of the Roman familia.
August 11, 2025
QotD: The job of the fuller
The cleaning agents for fulling wool varied by time and place. Roman fulleries generally used urine allowed to sit for a time (becoming “stale” – such urine is known as “wash”) because that concentrated the ammonium in the urine which acted as the cleansing agent. By the Middle Ages, we see the use of “fuller’s earth” (ammonia-rich clay), although urine continued to be used as well, presumably for its greater availability. As J.S. Lee notes (op. cit., 53), from the late twelfth century, we begin to see the use of water-power to replace the fullery worker as the treading agent, with the use of heavy wooden hammers driven by a water wheel to pummel the fabric.
Once this process was done the clothes or fabric were removed from the basin, scrubbed and wrung out fully, before being rinsed. In the Roman context – Roman fulleries (fullonicae) are fairly well archaeologically preserved and so give clues to the process at that time – the rinsing basins are set up to allow workers to walk in and out of them (some have working benches) which suggests that rinsing may have included additional scrubbing and wringing to make sure to remove both all of the impurities as well as all of the cleaning agents (Flohr, The World of the Fullo, 179-81). Fabrics would then have to be hung to be dried. In the Roman context, artwork tends to show clothes hung over high beams in the fullonica to dry; in the medieval context they were often hung to dry outdoors on long wooden frames called “tenters”.
Finally, the cloth would be “napped” (also called “raising the nap”, “rowing”, “teasing”, or polishing), which may have actually been the most labor intensive part of the process. Cloth would be brushed first, to raise the nap (the fuzzy, rough raised surface on woolen cloth), which would then be sheared to leave the cloth smooth. This stage also provided an opportunity for burling (and now you know why the coat factory is in Burlington), the inspection of the cloth and the manual removal of burrs, knots and other defects. Flohr (op. cit.) argues that this stage in the process consumed the bulk of the time and labor of fulling (a point on which J.S. Lee concurs for the Middle Ages). It is to a significant degree unfortunate that the sensational “they washed clothes in urine!” element of fulling has tended to eclipse the rest of the process in not only the popular imagination but occasionally in the scholarly discourse (the already cited Flohr, The World of the Fullo is a good antidote to this).
The position of fulling in the production chain of textiles seems to have varied a bit over time. In the medieval and early modern periods, fulling was generally done only once, as a final finishing stage in cloth production. By contrast, as Miko Flohr argues (op. cit., 57ff), the primary job of the Roman fuller was effectively as a laundry (though they may have treated freshly woven wool as well). Part of this probably has to do with differences in Roman clothing; Roman clothes were generally fairly simple in shape which must have made them easier to put through a fullery as a completed garment. Myself, I wonder if the changing role of fulling has to do with the introduction of soap during the later Roman Empire, which would have made it more possible for clothes to be laundered domestically (the Romans cleaned their bodies with oil, scraping it off with a strigil, which while perfectly good for cleaning skin would obviously not do for clothes, but soap and scubbing will work for both).
Fulling was generally a commercial (that is, not household) operation, done by professional fullers and we’ll talk about them (along with dyers and cloth merchants) in just a moment in terms of their place in society.
Bret Devereaux, “Collections: Clothing, How Did They Make It? Part IVa: Dyed in the Wool”, A Collection of Unmitigated Pedantry, 2021-04-02.
August 5, 2025
QotD: Fighting a Middle Republic Legion
As you approach the velites, they scatter back to their lines and now the first real trial comes: when you are about twenty meters out from the enemy line, a storm of those heavy pila come in, thick and all at once. Each one masses around 1.3kg (just short of three pounds) so even if the tip doesn’t bite, one of these things clanging off of armor or a shield is going to hurt, the impacts stagger men near you as you struggle to keep formation (and for a Hellenistic army, get and keep those sarisa-points down). The impact of the massed volley, especially against close-order infantry with tight fighting-width, is going to be chaotic as some men are killed, others disabled, still more suddenly staggered. The volley is followed almost immediately by the on-rush of the hastati. These guys don’t have a long spear for keeping you at a distance, they’re all brandishing swords and aim to get in close, using their large body-shields to absorb any blows you might throw while they get right up in your face, where their swords can stab and slash viciously over or under your shield. These hastati are aggressive and they’re probably better armored than you are.
And of course an engagement in contact like this is unpredictable. Perhaps in some areas, your lines push forward, whereas in other places it bends back. For large maneuver units (like taxeis!) this can be a real problem, but Roman maniples are small, so one maniple can advance if it finds the opportunity while others hold position or are even forced back (we actually see a general give, essentially, an “advance at your own discretion” order at Pydna, Plut. Aem. 20.8).
After a short and terrifying experience – these moments of shock combat probably didn’t last all that long, perhaps as little as just a few minutes – the hastati fall back. The front of your line is already physically and mentally exhausted. Many men are wounded and certainly some have been killed or disabled. I don’t want to oversell the casualties aspect of this: armies don’t annihilate each other in stand-up engagements (instead more casualties happen in pursuit), but wounds and exhaustion matter. Latin has this phrase, of being confectus vulneribus, “exhausted by wounds” or perhaps “worn down by wounds” (Liv. 24.26.14, 31.17.11, Caes. BGall. 5.45) to describe how the accumulation of a lot of little wounds can sap soldiers of their ability to resist effectively, even if no one wound is lethal. And just as important, all of the emotional impetus of your initial attack is spent. And there’s a decent chance that, as you try to breath, you still have these light velites‘ javelins (the hasta velitaris) thudding into your line every few seconds, because – again – they carry seven of them. They’re not out.
And then, as you are getting your bearings, trying desperately to catch your breath, the principes come up. They’re not physically tired or emotionally exhausted, but eager (like you were a half an hour ago when you advanced), they’ve been waiting all this time. Worse yet, these are probably the most combat-effective soldiers the Romans have, in the prime of their life, with years of combat experience. Now the second volley of pila comes in, creating yet more chaos. And then more angry, heavily armored Romans, behind their big shields, stabbing and cutting with their deadly gladii.
Now the men at the back of your single line may be relatively fresh, but you have no real way to get them to the front, so the wrath of the principes falls on men who are already exhausted, already wounded, already tired and already out of fight. Your line isn’t advancing so stridently; the men in the back, if the formation is deep, don’t know why the advance is slowed, why the line seems to be wavering, only that it seems to be wavering. And meanwhile, everyone is hoping that, at any moment, the victorious cavalry on the flanks is going to show up and win the battle, but they can’t see it anywhere in the confusion. Maybe your cavalry has won and is moments away – or perhaps Antiochus III charged it off the field again and no help is coming. Or perhaps the enemy cavalry has tied it up or worse yet, the Romans’ highly skilled Numidian allies might have mastered the flanks. You have no idea, you only know that help isn’t here, you are tired and more Romans are upon you. And somewhere, the thin thread of human courage snaps, either from the exhausted men in front or the confused men behind and the formation begins to collapse.
As the collective defense of lapped shields or serried pikes gives way, the Romans are now truly in their element: their large shields function just fine in individual combat and their versatile swords do as well. Lead by their centurions, the principes, with practiced and experienced skill, are finding the gaps, cutting as they go. As the formation crumbles, the velites can pursue – lightly armored, but well enough armed, backed up by the equites if there are any left.
You can see thus how this is a formation designed to wear down an enemy’s main battle line. It isn’t that the Romans are set massively deeper than a Hellenistic army, either. Assuming a base-3 set of file-depths (which seems to me the most likely), the Roman ranks are probably six, six and three men deep (hastati, principes, triarii), for a total depth over each file of 15, one less than the normal Macedonian formation. And with the wider fighting intervals the Romans use, they won’t normally have much of a problem matching the fighting width of the enemy army, unless substantially outnumbered (as, for instance, at Magnesia).
It’s not the Roman formation is deeper, it’s that its successive battle lines avoid exposing the entire army to exhaustion, attrition and friction all at once. In effect, it uses the same principles as defense-in-depth, exploiting the effect of friction on the enemy line to wear it down, but does so on the offensive. I don’t think it is an accident that when the Romans do lose, it tends to be because this model battle was spoiled in some way, either because the army was ambushed, enveloped, something disrupted the triplex acies or because the enemy was able to carry the field with just the momentum of the first charge – the Roman lines essentially failing like a building undergoing controlled demolition, as each floor pancakes the next without slowing.
But an army that isn’t able to decisively win the battle either at the first onset or somewhere else on the line is going to find itself in quite a lot of trouble as the Romans almost inevitably sandpaper away the morale and stamina of the main line of resistance until it collapses.
Now many of you may already be realizing that this kind of force is going to present a Hellenistic army with a lot of problems, both because it is set up for a different kind of fight than they are, but also because it may end up matching much heavier troops against the lighter parts of a Hellenistic army. But before we jump into battles, we need to zoom up to the upper levels of military analysis – operations and strategy – and talk about the advantages the Romans have there.
Because if all the Romans had was an edge in their tactical system, we might expect them to win battles but sometimes lose wars. Instead, while the Romans sometimes lose battles, they seemingly always win the war.
Bret Devereaux, “Collections: Phalanx’s Twilight, Legion’s Triumph, Part IIa: How a Legion Fights”, A Collection of Unmitigated Pedantry, 2024-02-09.
July 30, 2025
QotD: Meetings of the Roman Republican Senate
We may begin with place: the Senate had no single fixed meeting spot, though the curia in the Forum was the most common location, however the place the Senate met had to be religiously prepared via inauguration (the taking of the auspices by the augurs) and by sacrifices in order to make sure the gods approved of the proceedings and its results. Consequently, the Senate always met in a templum in the sense of a consecrated space, but also it tended to meet literally in temples, with meetings in the temples of Jupiter Optimus Maximus, the temple of Fides, the temple of Concord, and so on. Notably, two locations, the temple of Bellona and the temple of Apollo were also used and these sat outside the pomerium, enabling the Senate to meet with magistrates who, because of their active command of an army, could not cross the pomerium; they were also sometimes used to meet with foreign dignitaries the Senate did not wish to let into the city. Later added to this number of sites outside the pomerium was Pompey’s theater, which included a temple of Venus Victrix and a curia as part of the overall complex.
In order to meet, the Senate had to be called or more correctly “driven together” (cogere, often translated adequately as “summoned”, but as Lintott notes, it has an element of compulsion to it) by a magistrate. There were a few standard dates on which this would effectively always happen, particularly the first day of the consular year, but beyond that it was expected that magistrates in Rome could call the Senate at any time to discuss any issue on relatively short notice. There was initially no requirement that Senators live in the city of Rome, but it was clearly assumed. Early on in the second century, we get regulations requiring Senators to stay close to Rome unless they had an official reason to be elsewhere, though Senators might be permitted to leave if they needed to fulfill a vow. In the Late Republic it seems to have been common also for Senators to leave the city during the spring res prolatae, a sort of recess from public business (literally “the deferring of business”), but these informal breaks did not mean the Senate was truly “out of session” and it could still be summoned by a magistrate.
Generally, meetings of the Senate began at dawn, though they could begin later, and they proceeded either until the business was concluded or to dusk. Because of the ritual preparations required, no meeting of the Senate could last more than a day, much like the assemblies, so if the business was not finished, a new meeting would need to be called and the process begun from scratch. While it seems that magistrates generally tried to avoid calling the Senate during festival days, dies nefandi (days unsuited for public business) and meetings of the popular assemblies, there was no requirement to do so and the Senate might be called for any day for most of the Republic, with laws restricting the Senate’s meeting days only coming midway through the first century.
Beyond this, Senators were expected to show up and we hear of threats of fines or other censure for failure to show up, but it also seems like no meeting of the senate was ever very close to the full body and quorums for the Senate were fairly low, 100 or 150. For the Sullan Senate, notionally of 600 members, the highest attendances we know of, as noted by Lintott, are 415, 417 and 392. Of course some significant number of Senators will, at any time, have been active magistrates overseas, or serving as military tribunes, or as senatorial legati, but it seems clear that even beyond this attendance was not universal even if it was in theory supposed to be.
Bret Devereaux, “Collections: How to Roman Republic 101, Part IV: The Senate”, A Collection of Unmitigated Pedantry, 2023-09-22.
July 23, 2025
QotD: The legion of the Middle Republic
With that said, we reach the first and arguably most important thing to know about the legion: the Roman legion (and socii ala) of the Middle Republic is an integrated combined arms unit. That is to say, unlike a Hellenistic army, where different “arms” (light infantry, heavy infantry, cavalry, etc.) are split into different, largely homogeneous units, these are “organic” to the legion, that is to say they are part of its internal structure (we might say they are “brigaded together” into the legion as well). Consequently, whereas the Hellenistic army aims to have different arms on the battlefield in different places doing different things to produce victory, the Roman legion instead understands these different arms to be functioning in a fairly tightly integrated fashion with a single theory of victory all operating on the same “space” in the enemy’s line.
And you may well ask, before we get to organization, “What is that theory of victory?” As we saw, the Hellenistic army aims to fix the enemy with its heavy infantry center, hold the flanks with lighter, more mobile infantry (to protect that formation) and win the battle with a decisive cavalry-led hammer-blow on a flank. By contrast, the Romans seem to have decided that the quickest way to an enemy’s vulnerable rear was through their front. The legion is thus not built for flanking, its cavalry component – while ample in numbers – is distinctly secondary. Instead, the legion is built to sandpaper away the enemy’s main battle line in the center through attrition, in order to produce a rupture and thus victory.
To do that, you need to create a lot of attrition and this is what the manipular legion is built to do.
The legion of the Middle Republic is built out of five components: three lines of heavy infantry (hastati, principes and triariivelites), and a cavalry contingent (the equites). Specifically, a normal legion has 1200 each of velites, hastati and principes, 600 triarii and 300 equites, making a total combined unit of 4,500. Organizationally, the light infantry velites were packaged in with the heavy infantry (Polyb. 6.24.2-5) for things like marching and duties in camp, but in battle they typically function separately as a screening force thrown forward of the legion.
So to take the legion as an enemy would experience them, the first force were the velites. These seem to have been deployed in open order in front of the legion to screen its advance. These fellows had lighter javelins, the hasta velitaris (Livy notes they carried seven, Livy 39.21.13), no body armor and a “simple headcovering” (λιτός περικεφάλαιος, Polyb. 6.22.3), possibly hide or textile; they also carried a smaller round shield, the parma, and the gladius Hispaniensis for close-in defense (Livy 38.21.13). These are, all things considered, fairly typical ancient javelin troops, aiming to use the mobility their light equipment offers them to stay out of close-combat.
Behind the velites was the first line of the heavy infantry, the hastati. These fellows were organized into units called maniples (lit: “a handful”) of 120, which in turn are divided into centuries of 60 each. The maniples are their own semi-independent maneuvering units (note how much smaller they are than the equivalent taxeis in the phalanx, this is a more flexible fighting system), each with its own small standard (Polyb. 6.24.6) to enable it to maintain coherence as it maneuvers. That said, they normally form up in a quincunx (5/12ths, after a Roman coin with the symbol of five punches, like on dice) formation with the rear ranks, as you can see above.
The hastati (and the principes, who are equipped the same way) have the large Roman shield, the scutum, two heavy javelins (pila), the gladius Hispaniensis sword, a helmet (almost always a Montefortino-type in bronze in this period) and body armor. Poorer soldiers, we’re told, wore a pectoral, wealthier soldiers (probably post-225, though we cannot be certain) wore mail. That is, by the standards of antiquity, quite a lot of armor, actually – probably more armor per-man than any other infantry formation on their contemporary battlefield. That relatively higher degree of protection – big shield, stout helmet (Montefortino’s in this period range from 1.5-2.5kg, making them unusually robust), and lots of body armor – makes sense because these fellows are going to aim to grind the enemy down.
Note that a lot of popular treatments of this assume that the hastati were worse equipped than the principes; there’s no reason to assume this is actually true. The principes are older than the hastati, but the way to understand this formation is that the velites are young or poor, whereas for the upper-classes of the infantry (probably pedites I-IV) after maybe the first year or so, they serve in the heavy infantry (hastati, principes, triarii) based on age, not on wealth (and then the equites are the truly rich, regardless of what age they are; the relevant passage here is Polyb. 6.21.7-9, which is, admittedly, not entirely clear on what is an age distinction and what is a wealth distinction).
We’ve discussed the combat width these guys fight with already – somewhat wider spacing than most, so that each man covers the other’s flanks but they all have room to maneuver. It seems like the standard depth in the Middle Republic was either base-3 (so 3 deep on close order, 6 deep for “fighting” open order) or base-4 (so 4 and 8). Even in open-order with the maniples stretched wide (possibly by having rear centuries move forward), there would have been open intervals (10-20m) between maniples, which reinforces the role of a maniple as a potentially independent maneuvering unit – it has the space to move.2
Behind the hastati are the principes, with the same equipment and organization, slightly off-set to cover the intervals between the hastati, with a gap between the two lines (we do not know how large a gap). These men are slightly older, though not “old”. The whole field army generally consists of iuniores (men under 46) and given how the Romans seem to like to conscript, the vast majority of men will be in their late teens and 20s. So we might imagine the velites to be poorer men, or men in their late teens (17 being the age when one become liable for conscription) or so, while the hastati are early twenties, the principes mid-twenties and the handful of triarii being men in their late twenties or perhaps early 30s. The positioning of the principes isn’t to spare older men the rigors of combat, but rather to put more experienced veterans in a position where they can steady the less experienced hastati.3
Finally, behind them are the triarii, who trade the pila for a thrusting spear, the hasta, the Roman version of the Mediterranean omni-spear. These men are, as noted, the oldest and so likely the calmest under pressure and thus form a reserve in the rear. The three-line system here is what the Romans call a triplex acies (“three battle lines”). This wasn’t the only way these armies engaged and they could sometimes be formed up into a single solid line, but the triplex acies seems to have been the standard. We don’t know exactly how deep such a formation would run, but we have fairly good evidence that a legion might occupy a space around 400m wide (with some variation), meaning a whole Roman army’s core heavy infantry component (the two legions and two alae) might be some 1.6km (about a mile) across.
The equites, while organic to the legion organizationally, will be tactically grouped in battle to form cavalry screens on the edges of the army, not as a grand flanking cavalry “hammer”, but as flank-protection for the advancing infantry body (as a result, they tend to fight more cautiously). The equites in this period are heavy cavalry, with armored riders (after c. 225, that would be mail), using a shield and a hasta, along with a gladius as a backup weapon and thus serving as “shock” cavalry. Roman cavalry, if we look at their deployments, is generally ample in numbers, but the Romans seem to have been well aware it wasn’t very good, and sought allied cavalry (especially non-Italian allied cavalry) whenever they could get it. But the cavalry, Roman or not, was almost never the decisive part of the army.
Polybius tells us that the socii supplies more cavalry than the Romans and implies that there was a standard rule of three socii cavalrymen to every Roman equites, while socii infantry matched Roman infantry numbers (Polyb. 6.26.7). Looking at actual deployments though, we see that the socii tend to outnumber the Romans modestly, on about a 2:3 ratio, with socii cavalry only modestly outnumbering Roman cavalry.4 Consequently a normal Roman consular field army (of which the Romans generally had at least two every year) was 8,400 Roman infantry, around 12,600 socii infantry, 600 Roman cavalry and perhaps a thousand or so socii cavalry, for a combined force of 21,000 infantry (c. 5,000 light 16,000 heavy, so that’s a lot of heavy infantry) and 1,600 cavalry. That somewhat undersells the cavalry force the Romans might bring, as Roman armies also often move with auxilia externa (allied forces not part of the socii), which are very frequently cavalry-heavy (especially, after 203, that really good Numidian cavalry).5 By and large, it’s not that the Romans bring a lot less cavalry (as a percentage of army size), but that Italian cavalry tends to perform poorly and the as a result the Romans do not built their battle plans around their weakest combat arm.
Perhaps ironically, the Romans used their cavalry like Alexander and Hellenistic armies used their light infantry: holding forces designed to keep the flanks of the battlefield busy while the decisive action happened somewhere else.
Bret Devereaux, “Collections: Phalanx’s Twilight, Legion’s Triumph, Part IIa: How a Legion Fights”, A Collection of Unmitigated Pedantry, 2024-02-09.
1. On this, see Burns, M. T. “The Homogenisation of Military Equipment under the Roman Republic”. In Romanization? Digressus Supplement I. London: Institute of Archaeology, University College London, 2003.
2. On this, M.J. Taylor, “Roman Infantry Tactics in the Mid-Republic: A Reassessment”, Historia 63.3 (2014): 301-322.
3. To expound at some length on my own thoughts on how I think the wealth/age issue was probably managed, Dionysius (4.19.2) claims that the Romans recruited by centuries in the comitia centuriata such that the wealthy, divided into fewer voting blocks, served more often, and we know from Polybius that the maximum period of service for the infantry was sixteen years and from some math done by N. Rosenstein in Rome at War (2004) that the average service must have been around seven years. My suspicion, which I cannot prove is that the very poorest Roman assidui (men liable for conscription) might have only been serving fewer years on average and so it wasn’t a problem having them do all of their service as velites (the only role they can afford), whereas wealthier Romans (my guess is pedites IV and up) are the ones who age into the heavy infantry, with pedites I, whose members probably serve more than the seven-year average (perhaps around 10?) might make up close to 40% of the actual heavy infantry body (which is their balance in the comitia centuriata). The velites thus serves two important functions: a place to “blood” wealthier young Roman men to prepare them to stand firm in the heavy infantry line, as well as a place for poorer Romans to contribute militarily in a way they could afford. But I think that, once in the heavy infantry, the division between hastati, principes and triarii was – as Polybius says (6.21.7-9 and 6.23.1) – an age division, not a wealth division. Instead, the next wealth line is for the equites.
4. The data on this is compiled by Taylor, Soldiers & Silver (2020), 26-28.
5. Taylor, op. cit., 54-7 compiles examples.
July 17, 2025
QotD: War elephants in India
This may seem a strange approach to use to end a discussion of war elephants – after all, these are war elephants – but as will soon become apparent, war elephants are almost impossible to fully understand outside of the social and political context in which they are most useful.
First, we are going to look at how elephants fit into the ancient and medieval political systems which used them as weapons of war. I want to stress very strongly here that what I am presenting is essentially the main argument of Trautmann’s Elephants and Kings (2015), not something I dreamed up. For the sake of brevity, I am leaving out a lot of detail here – but you know where to go to find the argument in full.
Last time, we introduced a problem: while awesome, war elephants were very expensive and relatively easy to counter on the battlefield. This answered the question of why the Romans and Chinese mostly ignored the elephant as a weapon-system despite having access to it, but it raised a second question: if the elephant was at best a limited weapon, why did its use persist in India? After all, if the Romans could figure out how to beat these things, surely the Indians could too!
Part of the answer, of course, is that some of the logistical problems that existed for states located at the edges of elephant’s natural range simply don’t apply to states closer to the source. Indian kings could (and did!) deploy elephants in far greater numbers than Seleucid or Roman armies could. In particular, North Indian rulers, rather than relying on long distance trade, could acquire elephants through trade relations with “forest peoples” in their own hinterland. We have reports of armies with not hundreds but thousands of elephants from, for instance, the Nanda or Maurya empires. Nevertheless, while these factors simplified elephant logistics, they hardly made the use of the animals cheap.
What Trautmann instead observes is that the rise of war elephants occurred specifically in the context of kingship in India. Indeed, elephants were associated with kingship through royal elephant hunts and domesticated elephants kept for show even before war elephants were developed. Around 1400 B.C. the chariot arrives in India, bringing with it a military aristocracy where the nobles – and the noblest of all nobles is, of course, the king – rode into battle.
(I keep finding myself recommending it, but I’ll again note – for a good rundown of the value of chariots as royal symbols more than battlefield weapons, check out chapter 2 of Lee, Waging War (2016).)
That was the context the war elephant emerged into. By the fifth century or so, the war elephant seems to be displacing the chariot as the quintessential vehicle of the warrior-aristocrat (and thus the ultimate warrior-aristocrat, the king). Interestingly, the Mahabharata (fourth century B.C., but with components that may date as early as the ninth) preserves some of this shift, with a mix of aristocrats on chariot and aristocrats on elephant. As chariots faded (they were tactically inferior to true cavalry which was arising at this time), elephants progressively became the vehicle for the important warriors.
It’s not hard to see the appeal. For the warrior-aristocrat, battle isn’t just about winning, but is also about social status and position. Put another way: why does anyone put up with warrior-aristocrats, who get to live in luxury and boss everyone around? The implicit reason (sometimes explicit) across cultures is that it is the martial prowess – typically the personal, physical combat skill – that justifies the existence of the military aristocrat. You need Sir-Better-Than-You (to use a European framing) because you need someone who has mastered a difficult combat art (mounted combat) and is very, very good at it.
The warrior-aristocrat needs to be seen being a warrior aristocrat. For this purpose the elephant (much like its chariot forerunner) is perfect. Fighting from the back of an animal is a difficult skill which requires a lot of training the common folk do not have time to do. It also requires being able to afford and maintain a very expensive military asset commoners cannot afford. And not only does it allow the warrior-aristocrat to have an out-sized impact on the battle, but it literally elevates him over his fellow men so he can be seen (and it could not have escaped anyone that this was a physical realization of his actual high status). So long as the elephant remained even moderately militarily valuable, it was a perfect vehicle for a warrior-aristocrat to display his power and prowess.
And even more so for the king. Not only can the king ride his own elephant, but with his vast resources, he can procure elephants for his retainers. What is more impressive than a warrior aristocrat who has his own elephant? A warrior-king who has hundreds or thousands of elephants and his own warrior aristocrats to mount them. The thing is, a king’s actual power derives from the perception of his power – showing off the king’s military might makes him more likely to be obeyed (in ways – like tax collection – which allow him to further enhance his military might). This isn’t just a vanity project for the king (though it is that too) – extravagant displays of royal power are a key component of remaining king (the key big-word idea here is legitimacy).
This pattern in turn becomes self-reinforcing: as kings use elephants to show off (and thus reinforce) their power, elephants become symbols of royal power all on their own. Trautmann (2015) tracks this spread, particularly in South-East Asia – as the Indian model of kingship spreads into that region, war elephants spread with it. Whereas in places where there is plenty of contact, but the institution of Indian-style kingship doesn’t spread, war elephants are used rarely, if at all.
This in turn answers another quandary: why war elephants appealed to Hellenistic (that is, the heirs of Alexander) monarchs. Macedonian monarchy was not a form of Indian kingship – it had grown up in Macedon and been influenced by exposure to the Great Kings of Persia all on its own – but it was very similar in many ways. Compatible, we might say. Macedonian monarchs did not ride elephants (they rode horses), but they did need to be seen demonstrating martial excellence before their armies, just like Indian kings. In that context, the display of wealth and royal power implied by fielding a large elephant corps could be powerful, even if the king himself didn’t ride on an elephant. This is, perhaps most vividly demonstrated with Seleucus I Nicator, who earned himself the nickname “The Elephant King” and even produced coins advertising that fact […] This tie between elephants and kings seems to have been quite strong. Trautmann (2015) notes that even within India, states without kings (oligarchies, independent tribes and cities, etc) only rarely acquired elephants and never in the same sort of numbers as kings. So even when elephants are cheaper – because they are close by – unless you need elephants as physical symbols of the power and legitimacy of the king and his warrior-aristocrats, they are largely not worth the effort to procure.
The one great exception is Carthage – by the time it was using war elephants, Carthage was a mixed republic (much like Rome), and yet employed elephants extensively. Unfortunately, we have no sense of if Carthage – like Rome – would have abandoned elephants given time. The earliest attestation we have of Carthaginian war elephants is 262 B.C. (although they would have encountered them earlier from Pyrrhus of Epirus) and Carthage is completely gone in 146 B.C. It is possible Rome simply caught Carthage in the same “trying them out” phase of elephant use Rome would undergo in the second century B.C. and that Carthage may too have largely abandoned war elephants had it not been destroyed.
Bret Devereaux, “Collections: War Elephants, Part III: Elephant Memories”, A Collection of Unmitigated Pedantry, 2019-08-09.
July 11, 2025
QotD: Pyrrhus arrives in Magna Graecia to support the Tarantines
There are a few reasons I think this is unlikely. First, sources enlarging armies to fit the narrative magnitude of battles is a very common thing. But more to the point, Pyrrhus has crossed to Italy with 28,500 men total and – as Plutarch notes – hasn’t had a chance to link any of his allies up to his army. That may mean he hasn’t even reabsorbed his scouting force of 3,000 and he may well have also had to drop troops off to hold settlements, secure supplies and so on. Pyrrhus’ initial reluctance to engage (reported by Plutarch) is inconsistent with him wildly outnumbering the Romans, but his decision to wait for reinforcements within reach of the Romans is also inconsistent with the Romans wildly outnumbering him. So a battle in which Pyrrhus has perhaps 20-25,000 men and the Romans a standard two-legion, two-alae army of 20,000 give or take, seems the most plausible.1
The two forces met along the River Siris at Heraclea on the coastal edge of Lucania, Laevinius having pushed deep into southern Italy to engage Pyrrhus. As usual for these battles, we have descriptions or partial descriptions from a host of sources (in this case, Plutarch, Dionysius of Halicarnassus, Zonaras, Florus) which don’t always agree, leaving the modern historian in a bit of a pickle. Generally, we assume that a lot of the later Roman narratives of a famous defeat are likely to have been tailored to try and minimize the embarrassment, either by implying the battle was closer than it was or that Pyrrhus was a very impressive foe (or both) or other “face-saving” inventions. Worse yet, all of our sources are writing at substantial chronological distance, the Romans not really having started to record their own history until decades later (though there would have been Greek sources for later historians to work with). Generally, Patrick Kent tends to conclude that – somewhat unusually – Plutarch’s moralizing focus renders him more reliable here: Plutarch feels no need to cover for embarrassing Roman defeats or to embellish battle narratives (which he’d rather keep short, generally) because his focus is on the character of Pyrrhus. Broadly speaking, I think that’s right and so I too am going to generally prefer Plutarch’s narratives here.
A fairly handy map of Pyrrhus’ campaigns (though some of the detail is lost in the big sweeping arrows). What is notable is, apart from Pyrrhus’ lightning raid into Latium in 280, he is almost invariably fighting in “friendly” territory, either in Lucania (Heraclea), Apulia (Asculum) or Samnium (Beneventum), the lands of his allies. Pyrrhus never fights an actual pitched battle on Roman-controlled territory, which I think speaks to his strategic intent: to carve out a kingdom in Greater Greece, not to conquer the whole of Italy.
Wikimedia Commons.The battle was defined by Pyrrhus’ use of terrain – Pyrrhus thought delay might be wiser (to link up with his allies) but left a blocking force on the river (the Romans being on the other side). The Romans responded by forcing the river – typical Roman aggression – but Plutarch at least thinks it caught Pyrrhus by surprise (he hadn’t fought Romans before) and so it leaves him in a scramble. He charges his cavalry (Plut. Pyrrh. 16.5) to give his main phalanx time to form up for battle resulting in what seems like a cavalry engagement near the river. Pyrrhus nearly gets himself killed in the fighting, but survives and falls back to his main infantry force, which then met the Romans in an infantry clash. The infantry fighting was fierce according to Plutarch and Pyrrhus, still shaken from being almost killed, had to come out and rally his troops. In the end, the Romans are described as hemmed in by Pyrrhus’ infantry and elephants before some of his Greek cavalry – from Thessaly, the best horse-country in Greece – delivers the decisive blow, routing the Roman force.
It is, on the one hand, a good example of the Hellenistic army “kit” using almost all of its tactical elements: an initial – presumably light infantry – screen holding the river, followed by a cavalry screen to enable the phalanx to deploy, then a fierce and even infantry fight, finally decided by what seems to be flanking actions by cavalry and elephants. Plutarch (Pyrrh. 17.4) gives two sets of casualty figures, one from Dionysius and another from Hieronymus; the former says that the Romans lost 15,000 to Pyrrhus’ 13,000 killed, the latter that the Romans lost 7,000 to Pyrrhus’ just a bit less than 4,000 killed. The latter seems almost certainly more accurate. In either case, the Roman losses were heavier, but Pyrrhus’ losses were significant and as Plutarch notes, his losses were among his best troops.
Even in the best case, in victory, Pyrrhus had lost around 15% of his force (~4,000 out of 28,000), a heavy set of losses. Indeed, normally if an army loses 15% of its total number in a battle, we might well assume they lost. Roman losses, as noted, were heavier still, but as we’ve discussed, the Romans have strategic depth (in both geography, political will and military reserves) – Pyrrhus does not. By contrast, Alexander III reportedly wins at Issus (333) with just 150 dead (and another 4,802 wounded or missing; out of c. 37,000) and at Gaugamela (331) with roughly 1,500 losses (out of c. 47,000). The Romans will win at Cynoscephelae (197) with just 700 killed.
This isn’t, I think, a product of Pyrrhus failing at all, but rather a product of the attritional nature of Roman armies: even in defeat they draw blood. Even Hannibal’s great victory at Cannae (216) costs him 5,700 men, according to Polybius (more, according to Livy). But the problem for Pyrrhus is that his relatively fragile Hellenistic army isn’t built to repeatedly take those kinds of hits: Pyrrhus instead really needs big blow-out victories where he takes few losses and destroys or demoralizes his enemy. And the Roman military system does not offer such one-sided battles often.
Nevertheless, Pyrrhus shows that a Hellenistic army, capable handled, could beat a third-century Roman army, albeit not cleanly, and that is well worth noting.
Bret Devereaux, “Collections: Phalanx’s Twilight, Legion’s Triumph, Part IIIb: Pyrrhus”, A Collection of Unmitigated Pedantry, 2024-03-08.
1. I should note, this is Kent’s assessment as well.
July 5, 2025
QotD: Roman provinces under the Republic
While the Romans mostly improvise this system for just a handful of provinces – most of the basic patterns of Roman provincial governance develop in just the first four provinces1 – that system becomes the customary way Roman magistrates [consuls and praetors]and promagistrates [proconsuls and propraetors] handled the overseas provinces they were assigned to. Consequently, it was replicated over and over again through Rome’s steadily expanding empire. By the time the Republic collapses into the Empire, Rome will have not four provinces but fourteen; by the end of the reign of Augustus, as the Roman Empire largely took the borders it would mostly hold for the next four centuries, there were just under thirty provinces. Yet the way Rome will govern these provinces largely continued to hold to model established in the Republic, at least through to the Severan Dynasty (193-225 AD), if not further.
As a result, the improvised system the Romans developed for those first four provinces would end up being how the vast majority of people in the Roman empire would experience Roman governance.
Bret Devereaux, “Collections: How to Roman Republic 101, Addenda: The Provinces”, A Collection of Unmitigated Pedantry, 2023-11-03.
1. Sicily, Corsica et Sardinia, Nearer Spain and Further Spain, gained in the period from 241 to 197 and Rome’s only provinces until the addition of Macedonia in 147.
June 29, 2025
QotD: People in the past believed their own religion
It is an old mistake – Polybius makes it of the Romans, writing in c. 150 BC. Polybius notes that Roman religion was “distinctly superior” for maintaining the cohesion of the Roman state, but that “they (= the ruling class) have adopted this course for the sake of the common people” (Plb. 6.56.6-8), by which he means for the sake of the “fickle multitude” which must be “held in invisible terrors and pageantry”. Polybius, I should note, immediately contradicts himself – one benefit of this religion, he says, is that Roman magistrates (read: elite Senators) do not steal from the public treasury even when they are not watched, which would seem to imply that even many very elite Romans believed their own religion (also, irony alert: those famously incorruptible Romans …). Poor men, after all, are not elected to watch the treasury.
Fortunately, Polybius is not our only source on Roman culture, so it is possible to say – and I want to be clear here – Polybius is wrong (on this point at least). Roman elites took their religion very seriously. There are exceptions, of course – Julius Caesar was an Epicurean philosophically, although this does not seem to have altogether disrupted his performance of the duties of Rome’s highest priesthood (he was Pontifex Maximus). Nevertheless, such philosophically minded men were generally regarded with scorn by even the Roman elites (Cicero heaps such scorn on Cato in the Pro Murena in what were clearly intended as friendly “ribbing” before a mostly elite audience), who nevertheless greatly valued religious character (in contrast, Cicero is quick to accuse his hated enemy Catiline of irreligion before the Senate). Roman leaders vowed temples and sacrifices in duress, and built and gave them in victory. Lavish dedications to the gods from all levels of Roman society confirm the overall picture: the Romans believed their religion. Yes, even the elite Romans (there is, I should note, some complicated in that elites tended to separate what they viewed as religion from common “superstition”, but that’s a topic for another day.) And, if anything, the evidence for elite “buy-in” to Medieval religion is far more voluminous.
Yet none of the main characters of Game of Thrones, apart from perhaps Catelyn Stark, is devoted to their region’s traditional religion in any private sense. I cannot recall at any point any character protesting an action on account of religion – no marches held up by high holy days (a regular occurrence in ancient and medieval literature), no groups, individuals or places religiously exempt from violence. The Faith has certain rules about sexuality, but they’re rules that are not only flouted privately by individuals in the show, but also quite publicly by the last ruling dynasty for centuries.
Moreover, none of the ruling characters seem to have any religious functions – or if they do, they simply ignore them most of the time. Keep in mind, these people believe that their gods – be it the Seven or the Old Gods – are powerful divine beings who are directly and immediately interested in the world and who act on that interest. Keeping such beings happy is vitally important – it is not an afterthought. No society, so far as I can tell, has ever believed there is enough gold or enough armies to save you if you have enraged the gods. There is simply no off-setting an angry divinity (and before the smart guy asks, “what about a favorable divinity?” the Greeks had an answer for that – go read Euripides’ Hippolytus. It ends badly for the mortals).
The lack of religious duties or functions for characters who are kings is particularly surprising. Kingship has three core roles in almost all human cultures where the institution appears: kings are 1) chief judge, 2) chief general, and 3) chief priest. That third role appears more or less prominently in almost all societies. In ancient Egypt and (at times) in Mesopotamia, kings were held to literally be either earthly incarnations of major gods or minor gods in their own right. Roman Emperors held the office of Pontifex Maximus, and took over as the chief priest of the state, before becoming gods on their deaths. The Chinese emperor was the “Son of Heaven” and was tasked with maintaining the right relationship with the divine (the “mandate of heaven”). The emperors of Japan are purported to be direct descendants of the goddess Amaterasu (and they have a family tree to back it up).
The relationship between medieval kings and the Church was more complicated, because of the existence of a clear religious head (the Pope) outside of secular authority, but medieval kingship retained a strong sense of religious purpose. Coronation rituals often involved the clergy and were essentially religious rituals, because kingly power was still thought to be bestowed by God. Kings, in turn, had a special role in keeping their kingdom in right relationship with the divine, both through just rule (which included protecting the Church) and through the performance of religious rituals. Those rituals, of course, worked both ways: both believed to be religious efficacious (as in they helped bring about God’s good favor), but also valuable tools of building royal legitimacy (hat tip to my colleague Elizabeth Hassler’s excellent dissertation on the topic of holy kingship). In some places (England and France most notably), kings were thought holy enough to be able to heal certain diseases miraculously by touch – a king who couldn’t was clearly insufficiently pious.
Bret Devereaux, “New Acquisitions: How It Wasn’t: Game of Thrones and the Middle Ages, Part II”, A Collection of Unmitigated Pedantry, 2019-06-04.
June 23, 2025
QotD: Recruiting and organization under the “Marian reforms”
Sallust reports that Marius, “after he saw that the spirits of the plebs were aroused, he swiftly loaded ships with supplies, pay, weapons and other requirements; with them he ordered Aulus Manlius, his legate, to set out. Meanwhile himself he enrolled soldiers, not according to the mos maiorum [‘the customs of the ancestors’] from the census classes, but making use of whoever wished to go, mostly the capite censi [‘those counted by heads’ = the propertyless poor or proletarii]” (Sall. Iug. 86.1-2, trans mine). Plutarch repeats this report, that Marius violated custom by enrolling men who didn’t meet the property qualification for military service (Plut. Mar. 9.1).
There are a few oddities here to start, though. First, Sallust quickly notes that this resulted in Marius having an army rather larger than what the Senate had actually authorized (Sall. Iug. 86.4) and that’s actually quite a neat detail that may explain part of what’s going on here because this has, in a way, happened before. In 134, Scipio Aemilianus was elected consul for the second time (illegally, again) with a mandate to end the frustrating Roman war against the Celtiberian stronghold of Numantia in Spain. The Senate, however, denied Scipio authorization to raise fresh troops, to which Scipio responded by enlisting some 4,000 volunteers to replenish his legion; Appian says this was done with the consent of the Senate, but Plutarch’s brief note on it sure implies Scipio Aemilianus is end-running around Senatorial efforts to stifle him (App. Hisp. 84; Plut. Mor. 201A-B). And this too was hardly the first time for this sort of end-run; Scipio Africanus (what is it with Scipiones!?) back in 205 agitated for his invasion of Africa to end the Second Punic War and was given the province of Sicily with authorization to go to Africa if he thought it necessary, but the Senate registered its displeasure by refusing to let him levy troops, at which point – wait for it – Scipio took volunteers, equipping and financing his force through the socii and even building a fleet that way (Liv. 28.45.9-12).
In short, the Senate sometimes tried to trim the sails of generals it was displeased with – and Marius reportedly had gotten elected on a campaign platform of “to hell with the Senate” (Sall. Iug. 84.1) – by limiting the size of their armies or refusing to allow them to conduct a levy. And since 205 (a century before Marius), popular generals had occasionally juked this effort by the Senate by instead calling for volunteers, which the Senate could not stop. Marius is not doing something new in taking volunteers to supplement an army through the levy.
He also doesn’t keep doing it. After Marius wins in Africa with his volunteer-supplemented army (the bulk of which of course were still recruited through the dilectus under Metellus), he returns to Italy to take over the war against the Cimbri and Teutones but he doesn’t keep up the volunteer force, instead taking command of his predecessor Rutilius Rufus’ normally levied army (Front. Strat. 4.2.2). In practice, Marius probably took volunteers in part for that first army because the Senate was diverting available levy manpower towards the early phases of the Cimbric War (or at least that was a convenient excuse to kneecap him) – a series of costly military disasters for Rome which likely soaked up much of the manpower the Senate was willing to raise. Once Marius has access to that “primary” stream of manpower generated through the dilectus, he uses it and seems to stop using volunteers.
But what of recruiting the capite censi? Well, that isn’t quite new either, although it surely wasn’t typical. For one, it wasn’t that the poor absolutely never served; Polybius notes that the capite censi served in the fleet (Polyb. 6.19.2). But we also see non-assidui (assidui being the term for those wealthy enough to be liable for normal conscription) in a range of other emergencies. Livy reports in 329 a “crowd of sellularii [men who work sedentary trades, literally, ‘stoolsmen’], a type least suited for military service, were called into the army” (Livy 8.20.4), though the historicity of this report is questionable given the early date. In 296, Etruscan entrance into the Third Samnite War causes a draft of “not only the freeborn or the iuniores took the oath, but cohorts were made of seniores and centuries of freedmen” (Livy 10.21.4). Gellius (16.10.1) quotes Ennius reporting the proletarii were pulled into the armies in 280, presumably in response to Pyrrhus’ victory at Heraclea. And during the Second Punic War the Romans pulled out all of the stops, recruiting debtors and men convicted of capital crimes (Livy 23.14.3), enrolling slaves into the army (called the volones; you free them first and then draft them, Livy 27.38 and 28.10, Val. Max. 7.6.1) and as noted above, taking volunteers more generally.
As an aside, if you are wondering why the Romans seem in some of these to skip recruiting freeborn capite censi and go straight to freedmen and enslaved people, I think there are two answers here for this period. First, many of the available freeborn poor are probably already in service in the fleet. Second, there probably aren’t that many of them. Recall our chart of Roman social classes – the capite censi in the third century is quite small, almost certainly outnumbered by enslaved persons in Italy. But the population of Italy was rising over the third and especially second century and without adding new farmland, those new freeborn Romans may have swelled the ranks of the capite censi, leading to a much larger propertyless class by the late second century or the first century.1 Consequently, there may have been a lot more capite censi worth recruiting by Marius’ day, when Rome no longer needed to keep a large navy at sea (not facing any naval powers in its wars) and the number of capite censi having risen.
Finally, Marius does not mark the end of the Roman dilectus! Evidently Roman conscription persisted at least to the end of the Roman civil wars, as Suetonius reports Augustus (perhaps when he was still Octavian) inflicting the traditional penalty of being sold into slavery for draft-dodging on a Roman eques who cut the fingers off of his two sons to make them ineligible for military service (Suet. Aug. 24.1). Indeed we have attestations of the dilectus in 55, 52, 50, 49, AD 6 and AD 9.2 Even once the army is fairly clearly primarily a volunteer force, at least notionally the ability to hold a levy when necessary to fill the ranks remained “on the books” and Trajan (r. 98-117 AD) holds at least one levy because he punishes a father for the same reason Augustus had done (Dig. 49.16.4.12). So the traditional dilectus remained a thing Roman leaders could do well into the empire. In practice it seems safe to assume the system by the mid-first century is substantially ad hoc, as the census straight up doesn’t happen from 69 BC to 28 BC, which would make it hard to actually enforce the property requirements. But the process doesn’t stop in 107 and there’s no reason to suppose from 107 to 69, with the census being regularly conducted, that most annual levies were not conducted along traditional property lines.3
So the most we might say is that a one-time crisis expedient in earlier periods slowly becomes a standard way to supplement legions and then the standard way to recruit them, with the old normal method of the dilectus instead becoming the unusual way to supplement in a crisis. It’s unclear exactly when that shift-over point happens, but it sure isn’t in the career of Gaius Marius, who sits clearly in the “volunteers as a crisis response” side of the issue.
And what of the notion that Gaius Marius introduced both citizenship as a reward for service as a regular bonus and also that he instituted the paying of soldiers at the completion of a campaign to render them loyal? Well on the latter point, the Romans had been distributing spoils to the soldiers at the end of a campaign as a lump-sum payment since the beginning. This is exceedingly well reflected in Livy’s accounting of the years from 201 to 167 (where we have a nice continuous burst of Livy), see for instance Livy ::deep breath:: 30.45, 31.20, 33.23, 33.37, 34.46, 34.52, 36.40, 37.59, 39.5, 39.7, 40.34, 40.43, 40.59, 41.7, 41.13, 45.40, 45.43.4 And the idea that Roman victories might seize land which would then be settled as Roman coloniae, creating new land for Roman settlers was also not new (Wikipedia has a convenient list of Roman coloniae). So Marius is simply promising to do a thing Roman commanders regularly did, essentially saying, “serve with me, because I’m going to win and victory will make us rich”. Which is exactly the reason volunteers rushed to serve with Scipio Africanus and Scipio Aemilianus: they anticipated a lucrative victory for such well-regarded commanders.
And by now you may well be asking, “but wait, then when does the system change?” Because after all, I said that by the early empire, we can pretty clearly see an army primarily composed of professional, long-service volunteers who receive substantial retirement bonuses and are permanently stationed on the frontiers. Who is responsible for that? And in response, I give you, this guy.
It’s Augustus. It was always Augustus. Or at least I should say that is my view, given the evidence. Older scholarship – I think here of Keppie (1984) in particular – tended to assume that because most of the big changes happened with Marius (but we’ve seen they don’t) that Octavian/Augustus probably made only minimal changes to the military system he inherited from Julius Caesar. I don’t think that’s correct. I think if we look at the evidence in more detail it becomes clear that Augustus is the “break” (though not a clean break by any means) and that in fact we need to start regarding Augustus as a military reformer of some significant scale rather than merely the codifier of a Caesarian military system (though he probably does that too).
Augustus, after all, institutes regular bonuses for discharge, establishing a treasury funded by a regular tax to meet the expense rather than simply promising that he would win a lot and so soldiers would get rich off of their share of the booty (Res Gestae 17). And it’s not hard to see the problem he’s responding to – the massive military buildup of the Roman civil wars had left Octavian, as the victor, with the red-hot potato of hundreds of thousands of soldiers who were promised the spoils of victory, including large numbers of men who didn’t win but who, if not settled down somehow would disrupt the state (RG 3). Earlier in the civil wars, Octavian had used proscriptions and land confiscations to solve this problem but as emperor, he needed a permanent solution, thus the establishment of the aerarium militare and its discharge bonuses (praemia). Before that, you simply had generals promising to feast their soldiers off of the property of the vanquished; the civil wars had only changed that in that the vanquished were now Romans. It also establishes a standard length of service, creating that professional, long-service army.
There’s a related issue which is the fate of the citizen equites and the velites. Caesar’s armies in Gaul seem to have neither, so the assumption was that the shift to recruiting proletarii meant that these wealth-based distinctions (the richest Romans serve as equites, the poorest as velites) dropped away, leaving a uniform heavy infantry legion. And in a schematic it makes sense: both roles are absorbed by the auxilia and indeed Caesar makes use of a lot of Gallic cavalry auxiliaries. But as François Gauthier recently pointed out,5 it’s not all clear that the velites really did vanish in the late-second/early-first century. Cicero still refers to to them writing in the 40s (Cic. Fam. 9.20; Brut., 271) and their apparent absence in Caesar’s writing may well just be an accident of Caesar’s avoidance of technical language. Caesar doesn’t generally talk about hastati or triarii much either; he prefers milites (“soldiers”). Likewise, it’s clear the citizen cavalry – the equites – survived Marius; as Jeremiah McCall notes, we have good evidence for citizen equites at least as late as the 90s BC and suggests the citizen cavalry probably vanished in the 80s as a result of the Social War and Sulla‘s Civil War.6 It surely did not happen in 107 or 104.
Meanwhile the auxilia as a mature part of the Roman army really only emerge under Augustus, and not even right at the beginning of his reign either. Roman armies needed cavalry and light infantry to function, so once again we may not be looking at a clean break but rather a period of transition as a result of some generals preference for (non-Italian) allied or auxiliary cavalry and light infantry and the formalization of that system not in 107 with Marius but again in 27 with Augustus.
Marius is also sometimes credited with the idea of extending citizenship to non-citizens who served, which is a catastrophic misreading of one episode in his career. For one, this gets read as meaning that Marius extended citizenship to all of the Italians in his army or that he made it standard to do so. Note for instance this line pulled from Wikipedia:
Finally, Marius granted citizens of the Italian allies (Etruria, Picenum, etc.) full Roman citizenship if they fought for Rome and completed a period of service in the Roman army.
[Dated] 6/25/2023, specified in the hope this page changes to be less wrong.
And that’s very much not right either. We have evidence for only a handful of citizenship extensions by Marius. In particular, of his army he extended citizenship to just two cohorts (c. 1,000 men) from Camerinum (Plut. Mor. 202D, Cic. Pro Balbo 46.). I can only assume this gets misunderstood because some writers don’t know their unit sizes, but Marius had 32,000 men in his army at Vercellae (101 BC), probably something like half of which were socii. These two cohorts were a comparatively tiny fraction. Marius also seems to have selected a very small number of his other socii veterans for citizenship (Cic. Pro Balbo 48), but there was no blanket grant of citizenship. Of course there wasn’t, this issue remained substantially unsolved until the Social War (91-87BC); if Roman levies had been calmly minting new citizens out of thousands of Italians through the 90s, there would hardly have been a cause for the Social War.
Instead, citizenship as a reward for service is an artifact of the imperial period and the auxilia. The Roman use of non-Roman, non-socii troops to supplement their armies was not new, but it emerged as a formalized, permanent part of the Roman army not during the civil wars – where such units where both ad hoc but also not nearly so numerous – but under the reign of Augustus, coming to form about half of the army by the end of his reign (Tac. Ann. 4.5; on the emergence of the auxilia, see I. Haynes, Blood of the Provinces (2013)). Indeed, as Haynes notes (op. cit. 49), it is actually only under Tiberius (r. 14-37) that we get direct evidence of citizenship grants to auxilia and the practice even then seems at least somewhat irregular (though it comes to be regularized).
In short that, the notion that Gaius Marius instituted the pattern of granting citizenship to serving non-citizens on discharge is simply wrong; that’s not in our sources. That doesn’t become consistent until Tiberius well over a century later. Gaius Marius did recruit volunteer capite censi into his army once but didn’t make a habit of it and as such isn’t a major reformer so much as a key step in a slow process of change which reaches its decisive point probably under Augustus, more than half a century after Gaius Marius died. He wasn’t the first to do either thing, whatever our sources say.
Bret Devereaux, “Collections: The Marian Reforms Weren’t a Thing”, A Collection of Unmitigated Pedantry, 2023-06-30.
1. For more on the dynamics of this, see N. Rosenstein, Rome at War (2004), as this is part of his central argument.
2. For textual references, see Brunt, Italian Manpower (1971), 636-7.
3. We do not know how that process would have accounted for the massive expansion of the Roman citizen class due to the Social War. But evidently it did!
4. These were happily already compiled by Brunt, op. cit., 394.
5. “Did velites Really Disappear in the Late Roman Republic?” Historia 70 (2021).
6. J.B. McCall, The Cavalry of the Roman Republic (2002), 100-113..
June 17, 2025
QotD: What is a “tank”?
And we should stop to note that as with any question of definition, this one too is language-sensitive. The exact confines of a term vary from one language to another; kampfpanzer, for instance is not necessarily an exact synonym for “tank”.
In any event, the basic demands of early tanks were dictated by the realities of the Western Front: a tank needed to be able to resist small arms fire (particularly machine guns), deliver direct supporting fire itself, it needed to be able to move on the muddy, artillery-flattened ground and it needed to be able to cross a trench. This last requirement – the need to be able to both climb a parapet (usually c. 4ft) and then cross over an 8ft wide trench – was significant in the design of early tanks.
Those factors in turn dictated a lot of the design of early tanks. The armor demands of resisting small arms fire meant that the vehicle would be heavy (and indeed, as soon as tanks appeared amongst Allied troops, their German opponents began introducing more powerful bullets, like the K bullet and later the 13.2mm anti-tank round fired from the Mauser 1918 T-Gewehr). And here is the first advantage of tracks. The weight of a vehicle is distributed along all of the area of contact it has with the ground; with tires that area is limited to the bottom of the tire so the total area of ground contact is fairly low, which is fine for most vehicles.
But tanks are heavy. Really heavy. Even something like the Renault FT could mass around 7 tons and by later standards that would be classified as a tankette (a “mini-tank” as it were); by WWII, medium tanks often clocked in around 30 tons. If you put a vehicle like that on tires, you are going to create a LOT of pressure on those small points of contact. That might still be OK if you are just going to drive on roads and other firm surfaces which can take the pressure. But remember: tanks were designed for the Western Front, which looks like this.
Fortunately for the landship committee, this wasn’t a new problem: farming tractors were also heavy and also had to operate in churned up (in this case, plowed) soft soil; the heaviest of these vehicles had much the same problem and the solution was continuous tracks or “treads”. When kept properly tensioned – tune in, by the by, to Nicholas “The Chieftain” Moran’s YouTube for more than you ever want to know about track tension – the track distributes the weight of the tank across the entire section of the track touching the ground, which reduces the ground pressure at any given point, allowing a big heavy tank to roll over terrain where even a much lighter wheeled vehicle would get stuck.
This is one of those points where the functionality of a tank (what a tank does) has such a strong influence on design that the design implications of the functionality become part of the definition: a tank has to be heavily armored and has to be able to move off-road and as a result has to be tracked, not wheeled. One might be able to imagine some sort of exotic technology that might make it possible to do all of the things a tank does without tracks, but we don’t have that yet.
The other factor was fire. I’ve mentioned this before, but one of the significant background factors of WWI is that a lot of the belligerents misjudged the kind of artillery they’d need for a general European war. Not to get too deep into the weeds here, but most of the belligerents expected a relatively rapid war of maneuver and so thought that light, direct-fire artillery like the famed French ’75 (the Matériel de 75mm Mle 1897) would be the most useful. Those guns could be moved quickly and could deliver a lot of quick firepower on static or moving formations of enemy infantry in support of friendly infantry.
The problem is that in the conditions of trench warfare, those guns – as they were configured, at least – were far less useful. They were, first off, much shorter in range which meant they had to be brought dangerously far forward to do their direct fire role – often so far forward they could be engaged by enemy rifles and machine guns. This was compounded by the fact that direct fire at range was ineffective against trench works (which are dug down into the earth). But at the same time, the value of rapid firing (because these lighter guns could fire a lot faster than the heavy, indirect fire artillery) direct fire artillery remained high, if only you could get it to the fight.
This was also a problem a tank could solve: as a mobile, armored platform it could move a rapid-firing direct fire gun forward without immediately being knocked out by enemy small arms to support the infantry. There is, I should note, early complexity on this point, with both “male” (heavy direct fire cannon focused) and “female” (machine gun focused) tanks in WWI though in the end “hermaphrodite” designs with both capabilities (but much more focus on the main cannon) triumph, so that’s what we’ll focus on.
And that gets us the fundamental role structure for tanks: enough armor to resist enemy small arms (but with the understanding that some weapons will always be effective against the tank), enough mobility to cross the churned up battlefield and some direct fire capability to support the infantry crossing it at the same time.
Bret Devereaux, “Collections: When is a ‘Tank’ Not a Tank?”, A Collection of Unmitigated Pedantry, 2022-05-06.
