Now, you might ask at this point: why not defuse some of this tension with a “no first use” policy – openly declare that you won’t be the first to use nuclear weapons even in a non-nuclear conflict?
For the United States during the Cold War, the problem with declaring a “no first use” policy was the worry that it would essentially serve as a “green light” for conventional Soviet military action in Europe. Recall, after all, that the Soviet military was stronger in conventional forces in Europe during the Cold War and that episodes like the Berlin Blockade (and resultant Berlin Airlift) seemed to confirm Soviet interest in expanding their control over central Europe. At the same time, the Soviet use of military force to crush the Hungarian Revolution (1956) and the Prague Spring (1968) continued to reaffirm that the USSR had no intention of letting Central or Eastern Europe choose their own fates – this was an empire that ruled by domination and intended to expand if it could.
The solution to blocking that expansion was NATO, the North Atlantic Treaty Organization. Not because NATO collectively could defeat the USSR in a conventional war – the general assumption was that they probably couldn’t – but because NATO’s article 5 clause pledging mutual defense essentially meant that the nuclear powers of NATO (Britain, the United States, and France) pledged to defend the territory of all NATO members with nuclear weapons. But just like deterrence, mutual defense alliances are based on the perception that all members will defend each other. Declaring that the United States wouldn’t use nuclear weapons first would essentially be telling the Germans, “we’ll fight for you, but we won’t use our most powerful weapons for you” in the event of a conventional war; it would be creating a giant unacceptable asterisk next to that mutual defense clause.
So the United States had to be committed to at least the possibility that it would respond to a conventional military assault on West Germany with nuclear retaliation (often envisaged as a “tactical” use of nuclear weapons – that is, using smaller nuclear weapons against enemy military formations. That said, even in the 1950s, Bernard Brodie was already warning that restraining the escalation to general use of nuclear weapons once a tactical nuclear weapon was used would be practically impossible).
Bret Devereaux, “Collections: Nuclear Deterrence 101”, A Collection of Unmitigated Pedantry, 2022-03-11.
April 23, 2026
QotD: The problems of a “no first use” nuclear weapons policy
April 17, 2026
QotD: The decline of cities in the late western Roman Empire
This was the world the Romans knew and the world they were most comfortable governing. Consequently, while the Romans were utterly uninterested in “civilizing” anyone, when they conquered areas which weren’t urbanized, they tended to found cities or encourage local urbanization in order to create the administrative structures through which the Romans could extract revenue most efficiently.
As mentioned above, the Romans generally wanted these cities to be mostly self-governing. While at conquest, the Romans found themselves managing a bewildering array of different styles of local urban government, over time a mix of Roman administrative preference and cultural diffusion tended to produce a fairly similar set of civic institutions. City governments, which also administered their rural countryside, were run by a town council which consisted of the wealthiest notables of the town – the curiales – in much the same way that the Roman upper-class had dominated the running of the city during the Republic. Roman authority generally protected the curiales and their wealth from the sorts of popular uprisings that tempered many Greek oligarchies in the classical period and in return the curiales managed the population and the collection of taxes for the Romans.
The curiales both managed the town affairs and were also expected to use their own wealth to fund public activity and works: maintain temples and baths, fund religious rituals and festivals, and so on. Through the first and second century, that process was mostly responsible for providing the cities of the Roman Empire with the impressive collection of often still-visible public works they boasted: baths, theaters, amphitheaters, aqueducts, temples, courthouses, public spaces and so on. While some of these structures were little more than the public posturing of the elites, many of them were open to the general public and will have represented, in as much as anything before the industrial revolution could, meaningful improvements in the lives of regular people.
While most of the wealth of any of these cities was derived from the rents and taxes extracted from their agricultural hinterlands, these cities also substantially lived off of trade and markets. Because the local city typically housed the local market, they were the obvious point for local products to enter the stream of provincial-wide or empire-wide trade or for distant imports to reach their final customers. We’ll come back to this next time when we discuss trade and the economy, but for now I want to note that this trade provided a fair bit of the economic vitality of these cities but also that it did in fact reach down beyond mere luxury goods into the basic staples that even the relatively poor might buy.
The decline and fall of these Roman cities is most extensively described in J.H.W.G. Liebeschuetz’ aptly titled, Decline and Fall of the Roman City (2001). Given his title, as you might imagine, Liebeschuetz is in the “decline and fall” camp, arguing that the classical city which defined the Roman world largely did not survive it. Regional patterns differ, with Liebescheutz identifying three “patterns”: I) Western and Central Anatolia, II) Syria, Palestine and Arabia, III) the west, including North Africa).
We’ll deal with the situation in the east in just a moment, so let’s focus here on the cities of the west, which were at the start generally smaller, less wealthy and generally far younger than those of the east (with some exceptions in Italy). Decline sets in fastest and is most severe in Britain, with the final collapse of the cities coming as early as the 360s, whereas in North Africa, the classical city doesn’t seem to tip into decline until after 400.
While each individual region and indeed each city will have been subject to its own unique conditions, a few basic causes seem to have been active everywhere to some degree. First, the crisis of the third century seems to have fundamentally disrupted empire-wide Roman trade, which then stabilized at a lower level for the fourth century, before declining precipitously in the fifth. That first decline seems to have been somewhat offset by the increased demands of imperial administration and in particular the centralized taxation in-kind and movement of goods which had to move through cities. Peter Brown describes the late Roman state as, “the crude but vigorous pump which had ensured the circulation of goods in an otherwise primitive economy” (The Rise of Western Christendom, 2nd ed., 13). We’ll return to this when we discuss the shape of the economy next time, but for now it works as a crude, but vigorous description of that facet of the late Roman economy.
At the same time, as Liebescheutz describes, the role of the curiales steadily atrophies in the fourth century. On the one hand, much of the authority and power of being on the council was steadily eroded as those functions were pulled upwards into the imperial bureaucracy. At the same time, members of the curial class who sought imperial office could get immunities from the progressively more severe taxation which otherwise often fell on the curiales and so the imperial elite often crowded out the curiales when it came to wealth and prestige in the community. As they lost both control and responsibility for their cities, the curiales‘ investment in public works and monumental architecture also ceased (though local elites do invest in church-building and monastic foundations), leading to the decay of the physical urban centers.
Finally, the warfare of the fifth century had its impact, though as Liebescheutz notes, it cannot be presented as a sole cause simply because many urban areas were already clearly in decline when conflict hit. In the case of Britain, the cities were gone by 420, decades before the arrival of any invaders. Nevertheless, political instability and violence in the fifth century seems to have delivered death-blows to ailing communities, especially in the Balkans and along the Rhine.
The end result was that in the West, urbanism declined severely between the fourth and sixth centuries. Rome, once a city of a million people, collapsed down to a population of just 80,000. Arles, which had been a thriving Roman city with an amphitheater, an aqueduct, a chariot-racing track, a theater and full city walls shrunk so severely that the remains of the city moved inside its amphitheater, repurposing it as a new set of city walls, with the town square in the middle and houses built in the stands. While many towns survived in their new, shrunken and impoverished form, urbanism in Europe outside of the Eastern Roman Empire would largely have to be reinvented during the High Middle Ages, (though with some key institutional survivals from the Roman era and often rising out of the diminished remains of Roman cities). Instead, the society of the early Middle Ages was overwhelmingly rural in both population and focus. If on politics we have a bit of a mix between decline and continuity, when it comes to the cities that made up the old political system, the “decline and fall” knight strikes a clear blow: the system of social organization that characterized the ancient world practically vanished and would have to be redeveloped centuries later. The institutions that had maintained it (like the curiales) largely vanished, replaced in some cases by local “notables” and in other cases by ruralization.
Bret Devereaux, “Collections: Rome: Decline and Fall? Part II: Institutions”, A Collection of Unmitigated Pedantry, 2022-01-28.
April 11, 2026
QotD: Kingdoms were not “nations” in the Middle Ages
It is thus a bit odd that the regions of Westeros are seen by its inhabitants as being clear and unchanging. For instance, the Reach has borders, those borders do not move and everyone in those borders is loyal to House Tyrell. This is not how medieval rule works. The borders of, say, France, shifted over time (some places we consider “obviously” part of France were added only quite late, like French Flanders or Provence) as the ability of the French king to control those regions changed. For long periods of the Middle Ages, large parts of France were effectively controlled by the Kings of England (because they were also Dukes of this or that French duchy).
The idea that France, or Germany or Italy was a distinct, permanent entity with its own existence apart from a given royal family – more than just a space on a map – which comprised a people, their language and the government of those people, this is a modern phenomenon. Indeed, one may argue, it – that is, the nation-state – is the modern phenomenon.
Bret Devereaux, “New Acquisitions: How It Wasn’t: Game of Thrones and the Middle Ages, Part III”, A Collection of Unmitigated Pedantry, 2019-06-12.
April 5, 2026
QotD: The structure of a typical polis government
Because this week we’re interested in the politeia of a polis, that’s going to mean we’re mostly focused on the politai, the citizens, who we discussed last time as one of the key building blocks of the polis. Now, as we noted last time, it’s important to keep in mind that the politai are not all of the people in the polis or indeed even very many of them: women, children, resident foreigners, native members of non-citizen free underclasses and slaves were all set outside the politai and often had no means of gaining entrance. We’re going to talk about all of those folks in more depth in the third part, where we’ll look at the status layer-cake of polis society. But for now I just want to note that all of those people are there, even if they won’t figure very prominently in this discussion of the structures of polis government.
Now we’ll explore this question of how a polis was governed: first laying out the standard elements of a polis constitution, which as we’ll discuss were surprisingly similar from one polis to the next. Then we’ll deal with variations in how those elements are structured, which the Greeks understood to define the differences in the three kinds of constitution that a polis might normally have: oligarchy, democracy and tyranny. Then […] we’ll look at what sort of magistrates a polis might have and what their jobs might be as well as the structure of the legal system a polis might have.
THis is going to mean that we’re discussing the “constitutions” of poleis, but I want to be really clear here at the start that these are almost never written constitutions. So when I say “constitution”, understand that we mean this in the broad sense of “the actual makeup of the state’s institutions” rather than in the narrow sense of “a formal set of instructions for the running of the state”. Some poleis did actually have the latter (the oldest we have that I know of is a constitution established by Ptolemy I Soter for Kyrene in 322; the fact that this is a constitution dictated by a king to a subordinated polis should signal how odd it is), but they seem to have been very rare.
Bret Devereaux, “Collections: How to Polis, 101, Part IIa: Politeia in the Polis”, A Collection of Unmitigated Pedantry, 2023-03-17.
March 30, 2026
QotD: The Revenge of the Archaeologists
It is thus relevant that over the past half-century or so, it has so happened that effectively all ancient historians have had to develop a strong grasp of archaeological data; we don’t all necessarily learn to do the excavation work, of course (that’s what archaeologists do), but pretty much all ancient historians at this point are going to have to be able to read a site or artifact report as well as have a good theoretical grasp of what kinds of questions archaeology can be used to answer and how it can be used to answer those questions. This happened in ancient history in particular for two reasons: first, archaeology was a field effectively invented to better understand the classical past (which is now of course also used to understand the past in other periods and places) so it has been at work the longest there, but also because the sources for ancient history are so few. As I like to say, the problem for the modern historian is taking a sip of meaning from the fire-hose of evidence they have; but the challenge of an ancient historian is finding water in the desert. Archaeological data was a sudden, working well in that desert and much of the last two decades of ancient history has been built around it. Other fields of history are still processing their much larger quantity of texts; why dig so deep a well when you live next to a running river?
The result, in ancient history, has been what I tend to refer to as “the revenge of the archaeologists”. Not, mind you, revenge on medievalists, but in fact revenge on a very specific ancient historian and classicist, Moses Finley. Moses Finley was, from the 1950s to the 1980s, one of the most prominent classicists and his work touched on many fields, including the study of the ancient economy. Finley, writing in the 1960s was generally skeptical of the ability of archaeology to provide useful answers about the ancient economy (he preferred to understand the question by probing the mentalities of the Greek and Roman elite). Archaeology, Finley thought, was frequently over-interpreted and could never give a representative sample anyway; as he quipped in his 1965 article “Technical Innovation and Economic Progress in the Ancient World”, “we are too often victims of that great curse of archaeology, the indestructibility of pots”, a line for which, as far as I can tell, he is still quite unforgiven by some archaeologists.
As if in response, the archaeologists have spent the subsequent almost-six-decades proving again and again the tremendous value of their discipline by, among other things, utterly burying Finley’s The Ancient Economy (1973) under a mountain of archaeological data. It turns out the mentalités of aristocrats who largely hated merchants were not a good barometer of the activities of those merchants.
But you may now guess how this is going to play out in the discussion of Late Antiquity. The ancient historians come to the question ready to think in archaeological terms and ask what archaeological data can do to clear up these questions. Scholars of Late Antiquity trained as medievalists on the other hand, may or may not be well versed in archaeological methods or data (to be clear, some medievalists very much are versed, including prominent voices on the “change and continuity” side of this debate! But it is also very possible to be a “pure text” medievalist in a way that I don’t think I know a “pure text” ancient historian younger than sixty) because their field has not been forced, by dint of the paucity of sources, to revolve so heavily around archaeological data and because the archaeological data on the Middle Ages is not yet as voluminous as that on Classical Antiquity.
As I noted in the first post, beginning in the 1970s, what James O’Donnell calls the “reformation in Late Antique studies” launched a long overdue reassessment of Late Antiquity and the impact of the Fall of Rome – what we’ve called the “change and continuity” argument. I bring up all of this to note that the “counter-reformation” – what we’re calling the “decline and fall” argument – that really emerges beginning in the 90s is in many ways an extension of the “revenge of the archaeologists” in Classical studies (and especially the ancient economy) into the field of Late Antiquity. Indeed some of the scholars are the same (e.g. Willem Jongman) and many of them enter the debate on Late Antiquity as an extension of the debate about the Roman economy (in part demanding that “change and continuity” Late Antique scholars acknowledge things now generally considered “proved” by ancient historians about the earlier Roman economy).
In my own experience, particularly in more informal conversations, the methodological difference that interaction creates between ancient historians – for whom it has long been almost entirely settled that in a “fight” between archaeological evidence and effectively any other kind, the archaeological evidence “wins” – and medievalists for whom archaeology is a much less central part of their method (in part because their textual sources are more extensive) can lead to situations where the two sides of the debate talk past each other.
But when it comes to questions of demographics, economics and the conditions of life for the sort of people who rarely figure in our sources, archaeological evidence – although it is often incomplete and hard to interpret – offers the possibility of decisive answers to questions that otherwise would have to live entirely within the realm of speculation.
Bret Devereaux, “Collections: Rome: Decline and Fall? Part III: Things”, A Collection of Unmitigated Pedantry, 2022-02-11.
March 24, 2026
QotD: Citizens of a polis
Not everyone born in a polis was a member of the politai. Women could be of citizen status (and thus able to bear citizen children in poleis where that was required), but they could not be citizens at all. Being the male child of citizen parents was generally the core requirement of citizenship and in a democratic polis that was generally enough, but oligarchic poleis typically imposed wealth qualifications for political participation so not everyone born to citizens might themselves be a polites if they ended up too poor to meet the requirements. The terms astos and aste (ἀστός and ἀστή), “townsman” and “townswoman” respectively, might be used to make this distinction between the politai and people who were “merely” natives of the polis but barred for whatever reason from political participation. These distinctions become a lot more meaningful when you realize the point Aristotle is making defining the polis this way: if the polis is a community of politai then the residents of a polis (the physical space) who are not citizens are not members of the polis (not merely, we might imagine, non-participatory members).
Now the politai themselves also existed in subdivisions. We’ve mentioned division into demes or neighborhoods; while notionally geographic, demes could become hereditary (and indeed did become so in Athens). In Sparta and some poleis on Crete, citizens were divided into mess groups (syssitia or andreia). But by far the most common and important such division was into “tribes” or phylai (φυλαί, sing. φυλή), inherited kinship groups that often formed the largest subdivision of the politai of a polis, with even very small poleis having attested divisions into phylai in some cases (e.g. Delos as noted by M.H. Hansen in “Civic Subdivisions” in the Inventory). The politai might also be subdivided by other groupings like phratria (brotherhoods) and indeed a polis might have multiple such groupings, either neatly nested (as in Athens’ demes sorted into thirty trittyes sorted into ten phylai to make up the citizen body) or they might confusingly cross-cut each other.
There’s another key distinction between the politai – or at least men who might be politai – which isn’t a legal distinction but nevertheless matters for understanding how the Greeks imagined civic governance: the distinction between the few (hoi oligoi) and the many (hoi polloi). The few were the economic elite of the politai – the wealthy landowners – and the dominant group in oligarchies. A few terms might signify this group: “the few” (οἱ ὀλίγοι – hoi oligoi) or “the best” (οἱ ἄριστοι – hoi aristoi), or “the rich” (οἱ πλούσιοι – hoi plousioi) and can also be part of the meaning of the appellation “beautiful and good” (καλὸς κἀγαθός = καλὸς καὶ ἀγαθός – kalos kagathos) which translates more idiomatically to something like “gentleman” with an implication of both good conduct (especially in war) and high status. At its broadest reach, the few might consist of those politai with enough wealth to serve as hoplites, though it seems in most cases this group is understood much more narrowly and might be defined by heredity in addition to wealth in some cases.
In contrast to the few were, of course, the many. Once again a few terms might signify this group: “the many” (οἱ πολλοί – hoi polloi or οἱ πλῆθος – hoi plethos) or “the poor” (οἱ ἀποροῖ – hoi aporoi) or the people (δῆμος – demos), the last of which gives us the word democracy – rule by the demos. At its narrowest extent, these are all of the people too poor to serve as hoplites but who would otherwise be politai; in fact in a democracy they are politai, but in closed oligarchies they may not be. More broadly the concept of the demos can encompass all of the politai, both wealthy and poor, especially in a democratic context. Nevertheless the Greeks often understand these two groups as oppositional and non-overlapping: the politai composed of “the few”, with money and high status lineages and “the many”, without that, but with far greater raw numbers.
As we’ll see, it is that distinction – between “the few” and “the many” which the Greeks used to define the different forms of polis government, what they called a politeia (πολιτεία), which we might translate as “constitution” with the caveat that these are not written constitutions. And that’s where we’ll go next: now that we have our subdivisions, we’ll discuss next week the different ways they are organized and governed.
Bret Devereaux, “Collections: How to Polis, 101: Component Parts”, A Collection of Unmitigated Pedantry, 2023-03-10.
March 18, 2026
QotD: Feeding a Roman Consular army
Let’s deal with animals next. Each contubernium (“tent group”) of six soldiers likely had its own mule, so that’s 3,200 mules for the army, plus some additional number for the siege train and any army supplies; perhaps around 5,000 total (see Roth, op. cit. on this). On top of this we have horses for the cavalry; this will be rather more than 2,400 since spare horses will have been a necessity on campaign. Judging by Roman barley rations for cavalrymen (presumably intended to feed the horse) it seems a good guess that each cavalryman had one spare; for later medieval armies the number of spares would be substantially higher (at least three per rider). But for our lean army of Romans, that’s just 4,800 horses. An early modern army might require quite a few less mules (replacing them with wagons), but at the same time it is also probably hauling both field artillery and siege guns which demand a tremendous number of draft animals (mostly horses). My sense is that in the end this tends to leave the early modern army needing more animals overall.
Next the non-combatants. The mules will need drivers and the cavalrymen likely also have grooms to handle their horses, which suggests something like 3,400 calones [slaves or servants] as an absolute minimum simply to handle the animals. Roth (op. cit., 114) figures one non-combatant per four combatants in a Roman army, while Erdkamp (op. cit. 42) figures 1:5. Those figures would include not merely enslaved calones but also sutlers, slave-dealers, and women in the “campaign community”. Taking the lower estimate we might then figure something like 4,000 non-combatants for a “lean” Roman army, with many armies being more loaded up on non-combatants than even this. And while estimating the number of non-combatants for Roman armies is tricky, we actually have some figures for pre-modern armies to give a reference. Parker (op. cit. 252) notes units of the Army of Flanders (between 1577 and 1620) as high as 53% non-combatants, including women in the campaign community; one Walloon tercio in 1629 was 28% camp women on the march. It is tempting to compare these but caution is necessary here – both Roth’s and Erdkamp’s estimates are heavily informed by more modern armies so the argument would be circular: the estimates for the Romans look like later armies because later armies were used to calibrate estimates for the Romans.
That gives us an army now of 19,200 soldiers, 4,000 non-combatants, 5,000 mules and 4,800 horses. Roman rations were pretty ample and it seems likely that many of the calones did not eat so well but the ranges are fairly narrow; we can work with an average 1.25kg daily ration per person normally, with the absolute minimum being the 0.83kg daily grain ration following Polybius (Plb. 6.39.12-14, on this note Erdkamp op. cit. 33-42) if the army was short on supplies or needed to move fast eating only those buccelatum [hardtack] biscuits. That’s a normal consumption of 29,000kg per day for the humans, with the minimum restricted diet of 19,256kg for short periods. Then we need about 2.25kg of feed for each mule and about 4.5kg of feed for each horse (we’re assuming grazing and water are easily available), which adds up to 11,250kg for the mules and 21,600kg for the horses.
And at last we now have the scale of our problem: our lean army of 19,200 fighting men consumes an astounding 61,850kg (68.18 US tons) of food daily. It also consumes staggering amounts of water and firewood. In order to move this army or sustain it in place it is thus necessary to ensure a massive and relatively continuous supply of food to the army. Failure to do that will result in the army falling apart long before it comes anywhere close to the enemy.
Bret Devereaux, “Collections: Logistics, How Did They Do It, Part I: The Problem”, A Collection of Unmitigated Pedantry, 2022-07-15.
March 12, 2026
QotD: Roman armies of the middle and late Republic
It is hard to tell if the Roman triple-line (triplex acies) fighting system created the demand for synchronized discipline or if the Romans, having already developed a tradition of drill and synchronized discipline, adopted a fighting style that leveraged that advantage. Probably a bit of both, but in any event our evidence for the Roman army before the very late third century is very poor. By the time we truly see the Roman army clearly (c. 225 BC) the system seems to already [have been] in place for some time.
A Roman consular army was a complex machine. It was composed of an infantry line of two legions (in the center) and two socii “wings” (alae) to each side, along with cavalry detachments covering the flanks. Each of those infantry blocks (two legions, two alae) in turn was broken down into thirty separate maneuvering units (called maniples, generally consisting of 120 men; half as many for the triarii), which were in turn subdivided into centuries, but centuries didn’t really maneuver independently. In front of this was a light infantry screening force (the velites). So notionally there were in the heavy infantry of a standard two-legion consular army something like 120 different “chess pieces” that notionally the general could move around on their own and thus notionally the legion was capable of fairly complex tactical maneuvers.
You may have noted that word “notionally” because now we get into the limits of drill and synchronized discipline, because this isn’t a system for limitless tactical flexibility of the sort one gets in video games. Instead, recall that the idea here is to create coordinated movement and fighting (the synchronized discipline) through rigorous, repeated practice (drill). Of course one needs to practice specific things. Some of those things are going to be obvious: a drill for marching forward, or for turning the unit or for advancing on the charge.
In the Roman case, a “standard” battle involved the successive engagement and potentially retreat of each heavy infantry line: first the hastati (the first line) formed a solid line (filling the gaps) and attacked and then, if unsuccessful, retreated and the next line (the principes) would try and so on. Those maneuvers would need to be practiced: forming up, then having each maniple close the gap (we don’t quite know how they did this, but see below), the attack itself (which also involved usually throwing pila – heavy javelins), then retreat behind the next line if things went poorly. It’s also pretty clear from a battle like Cynoscephelae (197) or Bibracte (58) that individual maniples or cohorts (the Romans start using the larger 480-man-cohort as the basic maneuver unit during the second century BC) could be “driven” over the battlefield to a degree so there were probably drills for wheeling and turning.
Now even in this “standard” battle there is a lot of movement: maniples need to open and close gaps, advance and retreat and so on. This is what I mean by saying this army is a complex machine: it has a lot of moving parts that need to move together. The men in a maniple need to move together to make that mutually-supporting line and the maniples need to move together with each other to cover flanks and allow retreats. In terms of how the individual men moved, I’ve tended to think in terms of a “flow” model akin to this video of South Korean riot police training, rather than the clunkier Spartacus (1960) model.
But once an army has practiced all of these drills, it creates the opportunity for great improvisation and more complex tactics as well. Commanders, both the general but also his subordinates, can tell a unit to perform a particular maneuver that they have drilled, assuming the communication infrastructure exists in terms of instruments, standard shouted commands and battle standards (and note [that] Roman methods of battlefield communication were relatively well developed). That, for instance, allowed Aemilius Paullus to give orders to his first legion at Pydna for each of those maneuver units to either push forward or give ground independently, presenting the Macedonian phalanx with a tactical problem (an unevenly resisting line) it did not have a good solution for (Plut. Aem. 20.8-10). Having good junior officers […] was required but it wasn’t enough – those officers needed units which were already sufficiently drilled so that their orders (to press hard or retreat and reform in this case) could actually be carried out by soldiers for whom the response to those calls had become natural through that very drill.
At the same time I don’t want to give the wrong impression: even for the Romans battles where there was this sort of on-the-field improvising led by the general were uncommon (though not extremely rare). For the majority of battles, the legionary “machine” simply pushed forward in its standard way, even when – as at Cannae (216) – pushing forward normally proved to be disastrous. Just because an army can fight flexibly doesn’t mean it will or even that it should.
Bret Devereaux, “Collections: Total Generalship: Commanding Pre-Modern Armies, Part IIIa”, A Collection of Unmitigated Pedantry, 2022-06-17.
March 6, 2026
QotD: Operations, strategy, and tactics
[…]
Strategy is the upper layer of military analysis. Fundamentally strategy concerns the identification of final objectives, the way those objectives can be achieved and the resources to be used to achieve those objectives; these three components of strategy in US doctrine are termed “Ends, Ways, and Means” respectively. Strategy is thus the “big picture” thinking behind an action, including the decisions to both commence hostilities and end them.
[…]
Tactics are the lowest layer of military analysis. Tactics concern the methods to be used to win battles. Things like flanking, suppressive fire, ambushes, etc. are tactics. A military’s tactical system is often spelled out in doctrine. In theory, operations is designed to deliver forces to battles in such a way (positioning, comparative force, etc.) that their tactics can win those battles, while strategy should aim to ensure that winning those particular battles will achieve the desired political end (whatever concessions are desired). It is important to distinguish actions which are strategy (designed to directly produce a desired end to the conflict) from those which are merely tactical (designed to achieve a local success or advantage in a given engagement). It is important when assessing failures in war to distinguish between strategic failures (typically a failure to come up with realistic goals and the means to reach them), operational failures (e.g. logistics failures or unreasonable maneuver timetables) and tactical failures (e.g. failure to use combined arms effectively).
Bret Devereaux, “Miscellanea: A Very Short Glossary of Military Terminology”, A Collection of Unmitigated Pedantry, 2022-03-25.
February 28, 2026
QotD: The “Balance of Terror” in the missile age
This also meant that a devastating nuclear “first strike” could now be delivered before an opponent would know it was coming, or at least on extremely short notice. A nuclear power could no longer count on having enough warning to get its nuclear weapons off before the enemy’s nuclear strike had arrived. Bernard Brodie grappled with these problems in Strategy in the Missile Age (1959) but let’s focus on a different theorist, Albert Wohlstetter, also with the RAND Corporation, who wrote The Delicate Balance of Terror (1958) the year prior.
Wohlstetter argued that deterrence was not assured, but was in fact fragile: any development which allowed one party to break the other’s nuclear strike capability (e.g. the ability to deliver your strike so powerfully that the enemy’s retaliation was impossible) would encourage that power to strike in the window of vulnerability. Wohstetter, writing in the post-Sputnik shock, saw the likelihood that the USSR’s momentary advantage in missile technology would create such a moment of vulnerability for the United States.
Like Brodie, Wohlstetter concluded that the only way to avoid being the victim of a nuclear first strike (that having the enemy hit you with their nukes) was being able to credibly deliver a second strike. This is an important distinction that is often misunderstood; there is a tendency to read these theorists (Dr. Strangelove does this to a degree and influences public perception on this point) as planning for a “winnable” nuclear war (and some did, just not these fellows here), but indeed the point is quite the opposite: they assume nuclear war is fundamentally unwinnable and to be avoided, but that the only way to avoid it successfully is through deterrence and deterrence can only be maintained if the second strike (that is, your retaliation after your opponent’s nuclear weapons have already gone off) can be assured. Consequently, planning for nuclear war is the only way to avoid nuclear war – a point we’ll come back to.
Wohlstetter identifies six hurdles that must be overcome in order to provide a durable, credible second strike system – and remember, it is the perception of the system, not its reality that matters (though reality may be the best way to create perception). Such systems need to be stable in peacetime (and Wohlstetter notes that stability is both in the sense of being able to work in the event after a period of peace, but also such that they do not cause unintended escalation; he thus warns against, for instance, just keeping lots of nuclear-armed bombers in the air all of the time), they must be able to survive the enemy’s initial nuclear strikes, it must be possible to decide to retaliate and communicate that to the units with the nuclear weapons, then they must be able to reach enemy territory, then they have to penetrate enemy defenses, and finally they have to be powerful enough to guarantee that whatever fraction do penetrate those defenses are powerful enough to inflict irrecoverable damage.
You can think of these hurdles as a series of filters. You start a conflict with a certain number of systems and then each hurdle filters some of them out. Some may not work in the event, some may be destroyed by the enemy attack, some may be out of communication, some may be intercepted by enemy defenses. You need enough at the end to do so much damage that it would never be worth it to sustain such damage.
This is the logic behind the otherwise preposterously large nuclear arsenals of the United States and the Russian Federation (inherited from the USSR). In order to sustain your nuclear deterrent, you need more weapons than you would need in the event because you are planning for scenarios in which some large number of weapons are lost in the enemy’s first strike. At the same time, as you overbuild nuclear weapons to counter this, you both look more like you are planning a first strike and your opponent has to estimate that a larger portion of their nuclear arsenal may be destroyed in that (theoretical) first strike, which means they too need more missiles.
What I want to note about this logic is that it neatly explains why nuclear disarmament is so hard: nuclear weapons are, in a deterrence scenario, both necessary and useless. Necessary, because your nuclear arsenal is the only thing which can deter an enemy with nuclear weapons, but that very deterrence renders the weapons useless in the sense that you are trying to avoid any scenario in which you use them. If one side unilaterally disarmed, nuclear weapons would suddenly become useful – if only one side has them, well, they are the “absolute” weapon, able to make up for essentially any deficiency in conventional strength – and once useful, they would be used. Humanity has never once developed a useful weapon they would not use in extremis; and war is the land of in extremis.
Thus the absurd-sounding conclusion to fairly solid chain of logic: to avoid the use of nuclear weapons, you have to build so many nuclear weapons that it is impossible for a nuclear-armed opponent to destroy them all in a first strike, ensuring your second-strike lands. You build extra missiles for the purpose of not having to fire them.
(I should note here that these concerns were not the only things driving the US and USSR’s buildup of nuclear weapons. Often politics and a lack of clear information contributed as well. In the 1960s, US fears of a “missile gap” – which were unfounded and which many of the politicians pushing them knew were unfounded – were used to push for more investment in the US’s nuclear arsenal despite the United States already having at that time a stronger position in terms of nuclear weapons. In the 1970s and 1980s, the push for the development of precision guidance systems – partly driven by inter-agency rivalry in the USA and not designed to make a first strike possible – played a role in the massive Soviet nuclear buildup in that period; the USSR feared that precision systems might be designed for a “counter-force” first strike (that is a first strike targeting Soviet nuclear weapons themselves) and so built up to try to have enough missiles to ensure survivable second strike capability. This buildup, driven by concerns beyond even deterrence did lead to absurdities: when the SIOP (“Single Integrated Operational Plan”) for a nuclear war was assessed by General George Lee Butler in 1991, he declared it, “the single most absurd and irresponsible document I had ever reviewed in my life”. Having more warheads than targets had lead to the assignment of absurd amounts of nuclear firepower on increasingly trivial targets.)
All of this theory eventually filtered into American policy making in the form of “mutually assured destruction” (initially phrased as “assured destruction” by Secretary of Defense Robert McNamara in 1964). The idea here was, as we have laid out, that US nuclear forces would be designed to withstand a first nuclear strike still able to launch a retaliatory second strike of such scale that the attacker would be utterly destroyed; by doing so it was hoped that one would avoid nuclear war in general. Because different kinds of systems would have different survivability capabilities, it also led to procurement focused on a nuclear “triad” with nuclear systems split between land-based ICBMs in hardened silos, forward-deployed long-range bombers operating from bases in Europe and nuclear-armed missiles launched from submarines which could lurk off an enemy coast undetected. The idea here is that with a triad it would be impossible for an enemy to assure themselves that they could neutralize all of these systems, which assures the second strike, which assures the destruction, which deters the nuclear war you don’t want to have in the first place.
It is worth noting that while the United States and the USSR both developed such a nuclear triad, other nuclear powers have often seen this sort of secure, absolute second-strike capability as not being essential to create deterrence. The People’s Republic of China, for instance, has generally focused their resources on a fewer number of systems, confident that even with a smaller number of bombs, the risk of any of them striking an enemy city (typically an American city) would be enough to deter an enemy. As I’ve heard it phrased informally by one western observer, a strategy of, “one bomb and we’ll be sure to get it to L.A.” though of course that requires more than one bomb and one doubts the PRC phrases their doctrine so glibly (note that China is, in theory committed to developing a triad, they just haven’t bothered to actually really do so).
Bret Devereaux, “Collections: Nuclear Deterrence 101”, A Collection of Unmitigated Pedantry, 2022-03-11.
February 22, 2026
QotD: The shift from “motte-and-bailey” construction to stone castles
To talk about how these designs work, we need to clear some terminology. The (typically thin) wall that runs the circuit of the castle and encloses the bailey is called a “curtain wall“. In stone castles, there may be multiple curtain walls, arranged concentrically (a design that seems to emerge in the Near East and makes its way to Europe in the 13th century via the crusades); the outermost complete circuit (the primary wall, as it were) is called the enceinte. Increasingly, the keep in stone castles is moved into the bailey (that is, it sits at the center of the castle rather than off to one side), although of course stone versions of motte and bailey designs exist. In some castle design systems, with stone the keep itself drops away, since the stone walls and towers often provided themselves enough space to house the necessary peacetime functions; in Germany there often was no keep (that is, no core structure that contained the core of the fortified house), but there often was a bergfriede, a smaller but still tall “fighting tower” to serve the tactical role of the keep (an elevated, core position of last-resort in a defense-in-depth arrangement) without the peacetime role.
While the wooden palisade curtain walls of earlier motte and bailey castles often lacked many defensive features (though sometimes you’d have towers and gatehouses to provide fighting positions around the gates), stone castles tend to have lots of projecting towers which stick out from the curtain wall. The value of projecting towers is that soldiers up on those towers have clear lines of fire running down the walls, allowing them to target enemies at the base of the curtain wall (the term for this sort of fire is “enfilade” fire – when you are being hit in the side). Clearly what is being envisaged here is the ability to engage enemies doing things like undermining the base of walls or setting up ladders or other scaling devices.
The curtain walls themselves also become fighting positions. Whether on a tower or on the wall itself, the term for the fighting position at the top is a “battlement”. Battlements often have a jagged “tooth” pattern of gaps to provide firing positions; the term for the overall system is crenellation; the areas which have stone are merlons, while the gaps to fire through are crenals. The walkway behind both atop the wall is the chemin de ronde, allure or “wall-walk”. One problem with using the walls themselves as fighting positions is that it is very hard to engage enemies directly beneath the wall or along it without leaning out beyond the protection of the wall and exposing yourself to enemy fire. The older solution to this were wooden, shed-like projections from the wall called “hoarding”; these were temporary, built when a siege was expected. During the crusades, European armies encountered Near Eastern fortification design which instead used stone overhangs (with the merlons on the outside) with gaps through which one might fire (or just drop things) directly down at the base of the wall; these are called machicolations and were swiftly adopted to replace hoardings, since machicolations were safer from both literal fire (wood burns, stone does not) and catapult fire, and also permanent. All of this work on the walls and the towers is designed to allow a small number of defenders to exchange fire effectively with a large number of attackers, and in so doing to keep those attackers from being able to “set up shop” beneath the walls.
[I]t is worth noting something about the amount of fire being developed by these projecting towers: the goal is to prevent the enemy operating safely at the wall’s base, not to prohibit approaches to the wall. These defenses simply aren’t designed to support that much fire, which makes sense: castle garrisons were generally quite small, often dozens or a few hundred men. While Hollywood loves sieges where all of the walls of the castle are lined with soldiers multiple ranks deep, more often the problem for the defender was having enough soldiers just to watch the whole perimeter around the clock (recall the example at Antioch: Bohemond only needs one traitor to access Antioch because one of its defensive towers was regularly defended by only one guy at night). It is actually not hard to see that merely by looking at the battlements: notice in the images here so far often how spaced out the merlons of the crenellation are. The idea here isn’t maximizing fire for a given length of wall but protecting a relatively small number of combatants on the wall. As we’ll see, that is a significant design choice: castle design assumes the enemy will reach the walls and aims to prevent escalade once they are there; later in this series we’ll see defenses designed to prohibit effective approach itself.
As with the simpler motte and bailey, stone castles often employ a system of defense in depth to raise the cost of an attack. At minimum, generally, that system consists of a moat (either wet or dry), the main curtain walls (with their towers and gatehouses) and then a central keep. Larger castles, especially in the 13th century and beyond, adopting cues from castle design in the Levant (via the crusades) employed multiple concentric rings of walls. Generally these were set up so that the central ring was taller, either by dint of terrain (as with a castle set on a hill) or by building taller walls, than the outer ring. The idea here seems not to be stacking fire on approaching enemies, but ensuring that the inner ring could dominate the outer ring if the latter fell to attackers; defenders could fire down on attackers who would lack cover (since the merlons of the outer ring would face the other way). As an aside, the concern to be firing down is less about the energy imparted by a falling arrow (though this is more meaningful with javelins or thrown rocks) and more about a firing position that denies enemies cover by shooting down at them (think about attackers, for instance, crossing a dry moat – if your wall is the right height and the edges of the moat are carefully angled, you can set up a situation where the ditch never actually offers the attackers any usable cover, but you need to be high up to do it!).
Speaking of the moat, this is a common defensive element (essentially just a big ditch!) which often gets left out of pop culture depictions of castles and siege warfare, but it accomplishes so many things at such a low cost premium. Even assuming the moat is “dry”! For attackers on foot (say, with ladders) looking to approach the wall, the moat is an obstacle that slows them down without potentially providing any additional cover (it is also likely to disorder an attack). For sappers (attackers looking to tunnel under the walls and then collapse the tunnel to generate a breach), the depth of the ditch forces them to dig deeper, which in turn raises the demands in both labor and engineering to dig their tunnel. For any attack with siege engines (towers, rams, or covered protective housings made so that the wall can be approached safely), the moat is an obstruction that has to be filled in before those engines can move forward – a task which in turn broadcasts the intended route well in advance, giving the defenders a lot of time to prepare.
Well-built stone castles of this sort were stunningly resistant to assault, even with relatively small garrisons (dozens or a few hundred, not thousands). That said, building them was very expensive; maintaining them wasn’t cheap either. For both castles and fortified cities, one ubiquitous element in warfare of the period (and in the ancient period too, by the by) was the rush when war was in the offing to repair castle and town walls, dig out the moat and to clear buildings that during peace had been built int he firing lines of the castle or city walls.
Bret Devereaux, “Collections: Fortification, Part III: Castling”, A Collection of Unmitigated Pedantry, 2021-12-10.
February 16, 2026
QotD: How nuclear weapons were viewed right after WW2
At the same time, US strategists (particularly associated with the RAND Corporation) were beginning to puzzle out the long term strategic implications of nuclear weapons. In 1946, Bernard Brodie published The Absolute Weapon which set out the basic outlines of deterrence theory; he did this, to be clear, three years before the USSR successfully tested its first nuclear weapon in 1949 (far earlier than anyone expected because the USSR had spies in the Manhattan Project). Brodie is thus predicting what the strategic situation will be like when the USSR developed nuclear weapons; his predictions proved startlingly accurate, in the event.
Brodie’s argument proceeds as a series of propositions (paraphrased):
- The power of a nuclear bomb is such that any city can be destroyed by less than ten bombs.
- No adequate defense against the bomb exists and the possibilities of such are very unlikely.
- Nuclear weapons will motivate the development of newer, longer range and harder to stop delivery systems.
- Superiority in the air is not going to be enough to stop sufficient nuclear weapons getting through.
- Superiority in nuclear arms also cannot guarantee meaningful strategic superiority. It does not matter that you had more bombs if all of your cities are rubble.
- Within five to ten years (of 1946), other powers will have nuclear weapons. [This happened in just three years.]
All of which, in the following years were shown to be true. Consequently, Brodie notes that while nuclear weapons are “the apotheosis of aggressive instruments”, any attacker who used them would fear retaliation with their enemy’s nuclear weapons which would in turn also be so destructive such that “no victory, even if guaranteed in advance – which it never is – would be worth the price”. Crucially, it is not the fact of retaliation, but the fear of it, which matters and “the threat of retaliation does not have to be 100 per cent certain; it is sufficient if there is a good chance of it, or if there is a belief that there is a good chance of it. The prediction is more important than the fact.” [emphasis mine]
This does not “make war impossible” by any means, but rather turns strategy towards focusing on making sure that nuclear weapons are not used, by making it clear to any potential aggressor that nuclear weapons would be used against them. And that leads to Brodie’s final, key conclusion:
Thus, the first and most vital step in any American security program for the age of atomic bombs is to take measures to guarantee to ourselves in case of attack the possibility of retaliation in kind. The writer in making that statement is not for the moment concerned about who will win the next war in which atomic bombs are used. Thus far the chief purpose of our military establishment has been to win wars. From now on its chief purpose must be to avert them. It can have almost no other useful purpose.
To sum that up, because both the United States and its key enemies will have nuclear weapons and because their destructive power is effectively absolute (so high as to make any “victory” meaningless) and because there is no effective defense against such weapons, consequently the only rational response is to avoid the use of nuclear weapons and the only way to do that is to be able to credibly threaten to retaliate with nuclear weapons in the event of war (since if you cannot so retaliate, your opponent could use their nuclear weapons without fear).
That thinking actually took a while to take hold in actual American policy and instead during the 1940s and 1950s, the United States focused resources on bomber fleets with the assumption that they would match Soviet superiority in conventional arms in Europe with American nuclear superiority, striking military and industrial targets (“precision attacks with an area weapon”, a notion that is as preposterous as it feels) to immediately cripple the USSR in the event of war, or else aim to “win” a “limited” nuclear exchange.
Bret Devereaux, “Collections: Nuclear Deterrence 101”, A Collection of Unmitigated Pedantry, 2022-03-11.
February 10, 2026
QotD: The (historical) walls of Jericho
The problem standing between that attacking army and access was, of course, walls (though as we will see, walls rarely stand alone as part of a defensive system). Even very early Neolithic settlements often show concerns for defense and signs of fortification. The oldest set of city walls belong to one of the oldest excavated cities (which should tell us how short the interval between the development of large population centers and the need to fortify those population centers was), Jericho in the West Bank. The site was inhabited beginning around 10,000 BC and the initial phase of construction on what appears to be a city wall reinforced with a defensive tower was c. 8000 BC. It is striking just how substantial the fortifications are, given how early they were constructed: initially the wall was a 3.6m stone perimeter wall, supported by a 8.5m tall tower, all in stone. That setup was eventually reinforced with a defensive ditch dug 2.7m deep and 8.2m wide cutting through the bedrock (that is a ditch even Roel Konijnendijk could be proud of!), by which point the main wall was enhanced to be some 1.5-2m thick and anywhere from 3.7-5.2m high. That is a serious wall and unlikely the first defensive system protecting the site; chances are there were older fortifications, perhaps in perishable materials, which do not survive. Simply put, no one starts by building a 4m by 2m stone wall reinforced by a massive stone tower and a huge ditch through the bedrock; clearly city walls [were] something people had already been thinking about for some time.
I want to stress just how deep into the past a site like Jericho is. At 8000 BC, Jericho’s wall and tower pre-date the earliest writing anywhere (the Kish tablet, c. 3200 BC) by c. 4,800 years. The tower of Jericho was more ancient to the Great Pyramid of Giza (c. 2600 BC), than the Great Pyramid is to us. In short, the problem of walled cities – and taking walled cities – was a very old problem, one which predated writing by thousands of years. By the time the arrival of writing allows us to see even a little more clearly, Egypt, Mesopotamia and the Levant are already filled with walled cities, often with stunningly impressive stone or brick walls. Gilgamesh (r. 2900-2700 BC) brags about the walls of Uruk in the Epic of Gilgamesh (composed c. 2100) as enclosing more than three square miles and being made of superior baked bricks (rather than inferior mudbrick); there is evidence to suggest, by the by, that the historical Gilgamesh (or Bilgames) did build Uruk’s walls and that they would have lived up to the poem’s billing. Meanwhile, in Egypt, we have artwork like the Towns Palette, which appears to commemorate the successful sieges of a number of walled towns
So a would-be agrarian conqueror in Egypt, Mesopotamia or the Levant, from well before the Bronze Age would have already had to contest with the problem of how to seize fortified towns. Of course depictions like these make it difficult to reconstruct siege tactics (the animals on the Towns Palette likely represent armies, rather than a strategy of “use a giant bird as a siege weapon”), so we’re going to jump ahead to the (Neo)Assyrian Empire (911-609 BC; note that we are jumping ahead thousands of years).
Bret Devereaux, “Collections: Fortification, Part I: The Besieger’s Playbook”, A Collection of Unmitigated Pedantry, 2021-10-29.
February 4, 2026
QotD: The impact of quasi-official monotheism on the Roman Empire
That is part of the context where we should understand Constantine (r. 306-337). Constantine is famous for declaring the toleration of Christianity in the empire and being the first emperor to convert to Christianity (only on on his death-bed). What is less well known is that, having selected Christianity as his favored religion, Constantine – seeking unity again – promptly set out to unify his new favored religion, by force if necessary. A schism had arose as a consequence of Diocletian’s persecution and – now that Christianity was in the good graces of the emperor – both sides sought Constantine’s aid in suppressing the other in what became known as the Donatism controversy, as the side which was eventually branded heretical supported a Christian bishop named Donatus. Constantine, after failing to get the two groups to agree settled on persecuting one of them (the Donatists) out of existence (which didn’t work either).
It is in that context that later Christian emperors’ efforts to unify the empire behind Christianity (leading to the Edict of Thessalonica in 380) ought to be understood – as the culmination of, by that point, more than a century of on-again, off-again efforts by emperors to try to strengthen the empire by enforcing religious unity. By the end of the fourth century, the Christian empire was persecuting pagans and Jews, not even a full century after it had been persecuting Christians.
These efforts to violently enforce unity through homogeneity had the exact opposite effect. Efforts to persecute Arian Christians (who rejected the Nicene Creed) created further divisions in the empire; they also made it even more difficult to incorporate the newly arriving Germanic peoples, who had mostly converted to the “wrong” (Arian) Christianity. Meanwhile, in the fifth century, the church in the East splintered further, leading to the “Nestorian” (the term is contested) churches of Syria and the Coptic Church in Egypt on the “outs” with the official (Eastern) Roman Church and thus also facing persecution after the Council of Ephesus in 431. The resentment created by the policy of persecution in the East seems to have played a fairly significant role in limiting the amount of local popular resistance faced by the Muslim armies of the Rashidun Caliphate during the conquests of Syria, the Levant and Egypt in the 630s, since in many cases Christian communities viewed as “heretical” by Constantinople could actually expect potentially better treatment under Muslim rule. Needless to say, this both made the Muslim conquests of those regions easier but also go some distance to explaining why Roman/Byzantine reconquest was such a non-starter. Efforts to enforce unity in the empire had, perhaps paradoxically, made it more fragile rather than more resilient.
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.
January 29, 2026
QotD: Nitpicking the Roman army in Gladiator (2000)
In the process of Maximus riding up, the failure of negotiations and Maximus riding to join his cavalry, we get something of an overview of the Roman army and its position and both are wrong. Let’s start with the soldiers: we see a very clear distinction between two kinds of soldiers, the mail-clad auxilia, all archers, and the legionaries wearing the lorica segmentata and there appear to be about the same number of both groups. And here is where we first see the clear influence of the Column of Trajan (and to an unfortunately lesser degree, the far more appropriate Column of Marcus Aurelius) on the depiction, because this use of armor to distinctly signal the Roman citizen legionaries and non-citizen auxilia is straight from the Column of Trajan, completed probably around 113 and commemorating Trajan’s two Dacian Wars (101-102, 105-106).
What this sequence gets correct is that the Roman army was divided into those two groups, they were roughly equal in number (by this period, the auxilia probably modestly outnumber the legions in total manpower)1 and Trajan’s Column does use that visual signifier to distinguish them. This component is the crux of the verisimilitude that leads people to trust the rest of this sequence.
The problems start almost immediately from there. Roman auxilia were far more varied than what we see here in terms of equipment and tactics and only very few of them were archers. So let’s break down Roman auxiliary contingents. With all due caveats about the limits of our evidence, infantry auxilia outnumber cavalry by about 2:1 in attested auxilia units (auxilia were grouped into cavalry alae and infantry cohortes, generally of 480 men (sometimes around 800), but unlike for legionary citizen-infantry, these cohorts were not grouped into larger legions).2 So we ought to expect about a third of our auxilia to be cavalry, which is important because the cavalry detachments of Roman legions were very small (and mostly for scouting and messenger duties). Auxilia cavalry ranged in equipment and could include horse archers and even ultra-heavy cataphract cavalry, but most were mailed shock cavalry, equipped quite a lot like how Gallic or Germanic warrior-aristocrats or Roman legionary cavalry would be.
Of the remainder, the most common kind of infantry auxilia by far seem to have been heavy infantry, fighting in fairly heavy armor. These fellows get depicted in Roman artwork generally in mail armor, with flat oval shields (as opposed to the curved, rectangular imperial-period Roman scutum), spears and swords. These fellows, totally absent in this sequence are all over the Column of Trajan, with their flat oval shields being frequently seen (although one must distinguish them from Dacians who carry the same shield; the auxilia stick out for their mail and helmets). A bit less than 10% of auxilia units are attested as cohortes sagittariorum (“cohort of archers”). We also know the Romans used slingers within the auxilia, but as far as we can tell, not in specialized units; they may have been brigaded in with other auxilia cohorts. In either case, they appear in fairly small numbers. Finally, we also see on things like the Column of Trajan Roman allied or auxiliary units that are substantially lighter infantry: on the Column of Trajan, these are local troops shown wielding large clubs and stripped to the waist, presumably representing troops local to the Danube region, fighting in local (unarmored, with heavy two-handed weapons) style.
So whereas the army we see is a nearly even split between legionary heavy infantry and auxilia archers (with a small amount of legionary cavalry waiting for Maximus to show up to lead them), in practice a typical Roman field army would have far fewer archers, indeed around ten times fewer: not almost 50% of the force, but in fact probably a bit less than 5% of the force (since they’re less than 10% of the auxilia who would make up around half of a Roman field army). Meanwhile we’re simply missing the – by far – two most common sorts of auxilia cohorts, those of heavy infantry or heavy cavalry. This mangling of the structure of a Roman army is going to have implications when we get to Maximus’ overall plan for the battle as well.
Meanwhile, the legionary infantry are also much too uniform, literally. This is easily the most pardonable error, because what has happened here is that director Ridley Scott has copied the Column of Trajan but far too uncritically. After all, the Column of Trajan is not a photograph and thus has space for the artists producing it to take liberties, particularly in the name of imperial ideology and propaganda. In this case, showing large numbers of identically equipped soldiers, often moving in unified formation, serves the same rhetorical purpose in antiquity as it does today, suggesting an impressive, inhumanly uniformed and disciplined source. Moreover, the segmented Roman body armor, which we call the lorica segmentata (we don’t know what the Romans called it), was very distinctive to the legions, as it was the one armor that it seems like the auxilia probably (the evidence here can be tricky) didn’t share. And keeping the legions distinct from the auxilia also matters, as the legionary soldiers are higher status citizens who thus get “higher billing” in the imagery, as it were, than the auxilia. So showing all of the legions equipped neatly with this armor makes them seem distinct, impressive and uniform.
In short, it served Trajan’s image (and thus the artists aim) to suggest that all of his legions wore this armor.
Archaeology tells us quite clearly it was not so. Indeed, the lorica segmentata, so iconic because of its use in this way on the Column of Trajan, was probably the least common of the three major types of Roman legionary body armor in this period. The most common armor of the Roman legions was almost certainly still – as it had been in the Late Republic – mail, exactly the same as we see the auxilia wearing. We find fragments of Roman mail in legionary sites in all corners of the Empire and it remained common everywhere. To head off a standard question: no, it does not seem that the Romans ever got the idea to layer other defenses over mail, so when it was worn, it was the “primary” armor (worn over a padded textile defense called a subarmalis, but not under any other armor).3 We also see mail represented in Roman artwork, including on very high status soldiers, like senior centurions.
The next most common armor was probably scale armor, which we find very frequently in the East (that is, on the frontier with the Parthians/Sassanids) and often enough (if less frequently) in the West (that is, the Rhine/Danube frontier). We also know that some auxilia units wore this armor too and we see quite a bit of scale armor – wholly absent in this sequence – on the Column of – wait for it – Marcus Aurelius (completed c. 193). That’s the column that commemorates this war. Contemporary with this fictional battle. But it is less famous and somewhat less well-preserved than 70-years-earlier Column of Trajan, which they pretty evidently used quite a bit more of.
The lorica segmentata shows up the least often and – to my knowledge – effectively exclusively in the west on the Rhine/Danube frontier, where it is still probably not the most common (although it may have been more common than scale on that frontier). So what we ought to see in this army are legionaries who are marked out by their large scuta (the big Roman shield, by this period distinctly rectangular and also (as in the republic) curved), but in a range of mail, scale and lorica segmentata (with mail and segmentata being the most common, because we are on the Danube frontier, but scale hardly rare), along with auxilia divided into specialist cohorts (480 man units) each with different sets of armor and weapons: a few missile cohorts (archers, slingers), a lot more heavy infantry cohorts with spears and long shields, some lighter troops, and so on. The auxilia ought to be wearing basically every armor under the sun except for the lorica segmentata (which to my knowledge we’ve only ever found in sites associated with the legions).
Finally, these units are backed up by a whole load of catapults. We see two kinds, dual-arm arrow-throwing machines (which most folks would casually call ballistae) and single-armed pot-throwing machines (which most folks would casually call catapults), all of them in stationary mounts. Now on the one hand, “the Romans use lots of torsion-based catapults as artillery” is a true statement about the Roman army of this period, but on the other hand once again beyond that basic idea, most of this is wrong. Once again there’s an issue of verisimilitude here: the appearance of strange catapults and the true fact that the Romans used a lot of unusual catapults is likely to lead the viewer to assume some research has been done here and thus that these are the right catapults. For the most part, they are not.
We can start with the easy one, the larger single-armed pot-throwers. These are onagers, a late-Roman simplified single-arm torsion catapult, named for their fearsome “kick” (like an ass, an onager). These are popular favorites for Roman artillery, for instance showing up in both Rome: Total War and Total War: Rome II (both of which have main campaigns set during the Late Republic). There’s only one problem, which is that Gladiator (much less the even earlier Total War games) is set substantially too early for an onager to appear. Our first attestation of the onager is in Ammianus Marcellinus, writing in the last two decades of the 300s AD about the events of 353-378 (his work was broader than this, but only the back end survives). Vegetius, writing roughly contemporary with Ammianus also mentions them. But before the late fourth century, we don’t have any evidence for this design and it doesn’t show up on the Columns of Trajan or Marcus. So this isn’t just a little bit too early for these catapults but, given the evidence we have, around 150 years too early, the equivalent of having a line of M1 Abrams show up in a film about the Battle of Gettysburg.
What we do have are a number of twin-armed bolt or arrow-throwing machines and the Romans certainly had those, though what we see doesn’t match up well with what the Romans used. What we see is a single size of fairly large arrow-throwing engines, aimed upward to fire in fairly high arcs and built with large metal cases containing the torsion springs (generally made of hair or sinew, tightly coiled up; it is the coiling of these springs which stores the energy of the machine).
These two-armed torsion catapults came in a wide range of sizes and could be designed to throw either arrows/bolts or stones (the latter carved into spheres of rather precise caliber for specific machines). And we ought to see a pretty wide range of sizes here, from massive one-talent engines, which threw a 1 talent (26kg) stone and stood about three times the height of a man, to much smaller anti-personnel weapons (scorpiones) that were more like a “crew served” weapon than a large artillery piece. By Trajan’s time, the Romans had even taken to mounting these smaller crew-served engines on mule-drawn carts (called carroballistae) to allow them to be rapidly repositioned, something like early modern “horse artillery” (they were not meant to fire on the move; when we see them on Trajan’s Column, at least one of the operators is usually standing on the ground outside of the cart to winch the machine). These smaller machines, which would have made up the bulk of those deployed in a field battle, seem mostly absent in the sequence.
The result of all of this is that the Roman army presented in the opening moments of Gladiator manages to strike a remarkably unhappy balance: having just enough of the appearance of accuracy to decisively influence two decades of subsequent depictions of the Roman army without actually being particularly correct about anything beyond a very surface level. But subsequent pop-culture (again, I think Rome: Total War played a significant role here) would codify this vision of the Roman army – fire-throwing onagers, lots of auxilia archers, legionary rather than auxiliary cavalry, uniform use of the lorica segmentata – as the dominant model for quite some time.
But the army isn’t the only thing that’s wrong.
Bret Devereaux, “Collections: Nitpicking Gladiator’s Iconic Opening Battle, Part I”, A Collection of Unmitigated Pedantry, 2025-06-06.
- See figures in P. Holder, Auxiliary Deployment in the Reign of Hadrian (2003).
- There’s some complexity here because some infantry auxilia cohorts had small attached cavalry contingents too.
- I suppose I should note that is an odd exception for a type of very fine armor sometimes called lorica plumata (“feathered armor”) by modern writers where metal scales were mounted on mail armor (typically with extremely fine, small rings), rather than on a textile backing. This armor type seems to have been rare and must have been very expensive.
