This is the second part of a five part (I) series covering some of the basics of fortifications, from city walls to field fortifications, from the ancient world through the modern period. Last time, we looked as the ancient besieger’s playbook (both the motives and options for taking walled cities) through a case study of Assyrian siege methods. This week, we’re going to move forward several centuries to look at Roman fortifications, in particular how Roman methods of fortification changed to meet different objectives within different strategic environments, a good example of how what one is trying to do with a fortification alters the form that the fortification takes.
The Romans were, of course, building fortification for a very long time; we’re going to start with Roman fortified camps no later than the second century BC (probably also in use in the third century) and march all the way forward to Roman legionary fortresses in the fourth and fifth centuries AD, focusing on the ways that the changing strategic environment alters the overall aims of these fortifications which in turn influences their physical structure (but within a pre-existing framework established by past practice and assumptions).
As always, if you like what you are reading here, please share it; if you really like it, you can support me on Patreon. And if you want updates whenever a new post appears, you can click below for email updates or follow me on twitter (@BretDevereaux) for updates as to new posts as well as my occasional ancient history, foreign policy or military history musings.
Ace of Spades
Now when we talk about the evolution of Roman forts, it is important to be clear that the Romans (and their Mediterranean contemporaries) already had a well developed tradition of city fortification (and we’ll look at perhaps the greatest example of that tradition at the beginning of next week). Sophisticated walled cities in the Hellenistic period (323-31 BC) typically a stone curtain wall running the perimeter of the city (much like the one at Jericho; these could be very substantial – the 4th century BC walls of Rome were up to 10m high and 3.5m thick, made out of tufa), supported by projecting towers that might mount not only archers but also catapults and be further defended by a combination of defensive ditches and out-walls.
So the Romans have access (filtering in through the successor states; the Romans were voracious adopters of foreign military technology) to all of those techniques already, so the decision in some cases to not use them is a conscious choice (although in some cases the reason for that choice may well be ‘tradition and inertia’).
But we’re not going to start with Roman city walls, but rather with Roman marching forts. The degree to which we should understand the Roman habit of constructing fortified marching camps every night as exceptional is actually itself an interesting question. Our sources disagree on the origins of the Roman fortified camp; Frontinus (Front. Strat 4.1.15) says that the Romans learned it from the Macedonians by way of Pyrrhus of Epirus but Plutarch (Plut. Pyrrhus 16.4) represents it the other way around; Livy, more reliable than either agrees with Frontinus that Pyrrhus is the origin point (Liv. 35.14.8) but also has Philip V, an capable Macedonian commander, stand in awe of Roman camps (Liv. 31.34.8). It’s clear there was something exceptional about the Roman camps because so many of our sources treat it as such (Liv. 31.34.8; Plb. 18.24; Josephus BJ 3.70-98). Certainly the Macedonians regularly fortified their camps (e.g. Plb. 18.24; Liv 32.5.11-13; Arr. Alex. 3.9.1, 4.29.1-3; Curtius 4.12.2-24, 5.5.1) though Carthaginian armies seem to have done this less often (e.g. Plb. 6.42.1-2 encamping on open ground is treated as a bold new strategy).
It is probably not the camps themselves, but their structure which was exceptional. Polybius claims Greeks “shirk the labor of entrenching” (Plb. 6.42.1-2) and notes that the stakes the Romans used to construct the wooden palisade wall of the camp are more densely placed and harder to remove (Plb. 18.18.5-18). The other clear difference Polybius notes is the order of Roman camps, that the Romans lay out their camp the same way wherever it is, whereas Greek and Macedonian practice was to conform the camp to the terrain (Plb. 6.42); the archaeology of Roman camps bears out the former whereas analysis of likely battlefield sites (like the Battle of the Aous) seem to bear out the latter.
In any case, the mostly standard layout of Roman marching camps (which in the event the Romans lay siege, become siege camps) enables us to talk about the Roman marching camp because as far as we can tell they were all quite similar (not merely because Polybius says this, but because the basic features of these camps really do seem to stay more or less1 constant.
The basic outline of the camp is a large rectangle with the corners rounded off, which has given the camps (and later forts derived from them) their nickname: ‘playing card’ forts. The size and proportions of a fortified camp would depend on the number of legions, allied and auxiliaries present, from nearly square to having one side substantially longer than the other. This isn’t the place to get into the internal configuration of the camp, except to note that these camps seemed to have been standardized so that the layout was familiar to any soldier wherever they went, which must have aided in both building the camp (since issues of layout would become habit quickly) and packing it up again.
Now a fortified camp does not have the same defensive purpose as a walled city: the latter is intended to resist a siege, while a fortified camp is mostly intended to prevent an army from being surprised and to allow it the opportunity to either form for battle or safely refuse battle. That means the defenses are mostly about preventing stealthy approach, slowing down attackers and providing a modest advantage to defenders with a relative economy of cost and effort.
In the Roman case, for a completed defense, the outer most defense was the fossa or ditch; sources differ on the normal width and depth of the ditch (it must have differed based on local security conditions) but as a rule they were at least 3′ and 5′ wide and often significantly more than this (actual measured Roman defensive fossae are generally rather wider, typically with a 2:1 ratio of width to depth, as noted by Kate Gilliver. The earth excavated to make the fossa was then piled inside of it to make a raised earthwork rampart the Romans called the agger. Finally, on top of the agger, the Romans would place the valli (‘stakes) they carried to make the vallum. Vallum gives us our English word ‘wall’ but more nearly means ‘palisade’ or ‘rampart’ (the Latin word for a stone wall is more often murus).
Polybius (18.18) notes that Greek camps often used stakes that hadn’t had the side branches removed and spaced them out a bit (perhaps a foot or so; too closely set for anyone to slip through); this sort of spaced out palisade is a common sort of anti-ambush defense and we know of similar village fortifications in pre- and early post-contact North America on the East coast, used to discourage raids. Obviously the downside is that when such stakes are spaced out, it only takes the removal of a few to generate a breach. The Roman vallum, by contrast, set the valli fixed close together with the branches interlaced and with the tips sharpened, making them difficult to climb or remove quickly.
The gateway obviously could not have the ditch cut across the entryway, so instead a second ditch, the titulum was dug 60ft or so in front of the gate to prevent direct approach; the gate might also be reinforced with a secondary arc of earthworks, either internally or externally, called the clavicula; the goal of all of this extra protection was again not to prevent a determined attacker from reaching the gates, but rather to slow a surprise attack down to give the defender time to form up and respond.
And that’s what I want to highlight about the nature of the fortified Roman camp: this isn’t a defense meant to outlast a siege, but – as I hinted at last time – a defense meant to withstand a raid. At most a camp might need to withstand the enemy for a day or two, providing the army inside the opportunity to retreat during the night.
We actually have some evidence of similar sort of stake-wall protections in use on the East Coast of Native North America in the 16th century, which featured a circular stake wall with a ‘baffle gate’ that prevented a direct approach and entrance. The warfare style of the region was heavily focused on raids rather than battles or sieges (though the former did happen) in what is sometimes termed the “cutting off way of war” (on this see W. Lee, “The Military Revolution of Native North America” in Empires and Indigines, ed. W. Lee (2011)). Interestingly, this form of Native American fortification seems to have been substantially disrupted by the arrival of steel axes for presumably exactly the reasons that Polybius discusses when thinking about Greek vs. Roman stake walls: pulling up a well-made (read: Roman) stake wall was quite difficult. However, with steel axes (imported from European traders), Native American raiding forces could quickly cut through a basic palisade. Interestingly, in the period that follows, Lee (op. cit.) notes a drift towards some of the same methods of fortification the Romans used: fortifications begin to square off, often combined a ditch with the palisade and eventually incorporated corner bastions projecting out of the wall (a feature Roman camps do not have, but later Roman forts eventually will, as we’ll see).
Roman field camps could be more elaborate than what I’ve described; camps often featured, for instance, observation towers. These would have been made of wood and seem to have chiefly been elevated posts for lookouts rather than firing positions, given that they sit behind the vallum rather than projecting out of it (meaning that it would be very difficult to shoot any enemy who actually made it to the vallum from the tower).
When a Roman army laid siege to a fortified settlement, the camp formed the ‘base’ from which siege works were constructed (particularly circumvallation – making a wall around the enemy’s city to keep them in – and contravallation – making a wall around your siege position to keep other enemies out. We’ll discuss these terms in more depth a little later). Some of the most elaborate such works we have described are Caesar’s fortifications at the Siege of Alesia (52 BC; Caes. B.G. 7.72). There the Roman line consisted of an initial trench well beyond bow-shot range from his planned works in order to prevent the enemy from disrupting his soldiers with sudden attacks, then an agger and vallum constructed with a parapet to allow firing positions from atop the vallum, with observation towers every 80 feet and two ditches directly in front of the agger, making for three defensive ditches in total (be still Roel Konijnendijk‘s heart! – but seriously, the point he makes on those Insider “Expert Rates” videos about the importance of ditches are, as you can tell already, entirely accurate), which were reinforced with sharpened stakes faced outward. As Caesar expressly notes, these weren’t meant to be prohibitive defenses that would withstand any attack – wooden walls can be chopped or burned, after all – but rather to give him time to respond to any effort by the defenders to break out or by attackers to break in (he also contravallates, reproducing all of these defenses facing outward, as well).
Deciding to Stand
The end of the reign of Augustus (in 14AD) is a convenient marker for a shift in Roman strategic aims away from expansion and towards a ‘frontier maintenance.’ The usual term for both the ROman frontier and the system of fortifications and garrisons which defended it is the limes (pronounced lim-ees), although this wasn’t the only word the Romans applied to it. I want to leave aside for a moment the endless, complex conversation about the degree to which the Romans can actually be said to have strategic aims, though for what it is worth I am one of those who contends that they did.2 We’re mostly interested here in Roman behavior on the frontiers, rather than their intent anyway.
What absolutely does begin happening during the reign of Augustus and subsequently is that the Roman legions, which had spent the previous three centuries on the move outside of Italy, begin to settle down more permanently on Rome’s new frontiers, particularly along the Rhine/Danube frontier facing Central and Eastern Europe and the Syrian frontier facing the Parthian Empire. That in turn meant that Roman legions (and their supporting auxiliary cohorts) now settled into permanent forts.3
The forts themselves, at least in the first two centuries, provide a fairly remarkably example of institutional inertia. While legionary forts of this early period typically replaced the earthwork-and-stakes wall (the agger and vallum) with stone walls and towers and the tents of the camp with permanent barracks, the basic form of the fort: its playing-card shape, encircling defensive ditches (now very often two or three ditches in sequence) remain. Of particular note, these early imperial legionary forts generally still feature towers which do not project outward from the wall, a stone version of the observation towers of the old Roman marching camp. Precisely because these fortifications are in stone they are often very archaeologically visible and so we have a fairly good sense of Roman forts in this period. In short then, put in permanent positions, Roman armies first constructed permanent versions of their temporary marching camps.
And that broadly seems to fit with how the Romans expected to fight their wars on these frontiers. The general superiority of Roman arms in pitched battle (the fancy term here is ‘escalation dominance’ – that escalating to large scale warfare favored the heavier Roman armies) meant that the Romans typically planned to meet enemy armies in battle, not sit back to withstand sieges (this was less true on Rome’s eastern frontier since the Parthians were peer competitors who could rumble with the Romans on more-or-less even terms4; it is striking that the major centers in the East like Jerusalem or Antioch did not get rid of their city walls, whereas by contrast the breakdown of Roman order in the third century AD and subsequently leads to a flurry of wall-building in the west where it is clear many cities had neglected their defensive walls for quite a long time). Consequently, the legionary forts are more bases than fortresses and so their fortifications are still designed to resist sudden raids, not large-scale sieges.
They were also now designed to support much larger fortification systems, which now gives us a chance to talk about a different kind of fortification network: border walls. The most famous of these Roman walls of course is Hadrian’s Wall, a mostly (but not entirely) stone walls which cuts across northern England, built starting in 122. Hadrian’s walls is unusual in being substantially made out of stone, but it was of-a-piece with various Roman frontier fortification systems. Crucially, the purpose of this wall (and this is a trait it shares with China’s Great Wall) was never to actually prevent movement over the border or to block large-scale assaults. Taking Hadrian’s wall, it was generally manned by something around three legions (notionally; often at least one of the legions in Britain was deployed further south); even with auxiliary troops nowhere near enough to actually manage a thick defense along the entire wall. Instead, the wall’s purpose is slowing down hostile groups and channeling non-hostile groups (merchants, migrants, traders, travelers) towards controlled points of entry (valuable especially because import/export taxes were a key source of state revenue), while also allowing the soldiers on the wall good observation positions to see these moving groups. You can tell the defense here wasn’t prohibitive in part because the main legionary fortresses aren’t generally on the wall, but rather further south, often substantially further south, which makes a lot of sense if the plan is to have enemies slowed (but not stopped) by the wall, while news of their approach outraces them to those legionary forts so that the legions can form up and meet those incursions in an open battle after they have breached the wall itself. Remember: the Romans expect (and get) a very, very high success rate in open battles, so it makes sense to try to force that kind of confrontation.
This emphasis on controlling and channeling, rather than prohibiting, entry is even more visible in the Roman frontier defenses in North Africa and on the Rhine/Danube frontier. In North Africa, the frontier defense system was structured around watch-posts and the fossatum Africae, a network of ditches (fossa) separating the province of Africa (mostly modern day Tunisia) from non-Roman territory to its south. It isn’t a single ditch, but rather a system of at least four major segments (and possibly more), with watch-towers and smaller forts in a line-of-sight network (so they can communicate); the ditch itself varies in width and depth but typically not much more than 6m wide and 3m deep. Such an obstruction is obviously not an prohibitive defense but the difficulty of crossing is going to tend to channel travelers and raids to the intentional crossings or alternately slow them down as they have to navigate the trench (a real problem here where raiders are likely to be mounted and so need to get their horses and/or camels across).
On the Rhine and the Danube, the defense of the limes, the Roman frontier, included a border wall (earthwork and wood, rather than stone like Hadrian’s wall), similarly supported by legions stationed to the rear, with road networks positioned; once again, the focus is on observing threats, slowing them down and channeling them so that the legions can engage them in the field. This is a system based around observe-channel-respond, rather than an effort to block advances completely.
Refusing to Fold
As we move into the later Roman Empire, particularly after the Crisis of the Third Century (235-284 AD), we start to see changes in the form of Roman forts. Two things had been happening of the course of the Crisis (and in some cases before it) which transformed the Roman frontier situation. First, Rome’s enemies had gotten quite a bit stronger: in the west, long exposure to Rome had led the various ‘barbarians’ on the other side of the limes to both pick up elements of Roman military practice but also to form into larger and larger political units (in part in order to hold off Roman influence) which were more dangerous. In the east, the Parthian Empire had collapsed in 224 to be replaced by the far more capable and dangerous Sassanid Empire. At the same time, fifty years of civil war had left Rome itself economically and militarily weaker than it had been. Bigger threats combined with scarcer state resources enforced a more flexible approach to controlling the borders.
In particular, Roman forces could no longer be entirely sure they would possess escalation dominance in any given theater. Indeed, during the Crisis, with legions being peeled to fight endless internal wars between rival claimants had meant that major frontier problems might go under-resourced or even entirely unaddressed for years. While the reign of Diocletian (284-311) marked a return to Roman unity, quite a bit of damage had already been done and by the end of the third century we see changes in patterns of fortification that reflect that.
The changes seem fairly clearly to have been evolutionary, in part because many older legionary forts remained in use. Some of the first things we see are traditional ‘playing-card’ forts but now with the neat rectangular shape disrupted by having the towers project out from the walls. The value of a projecting tower (as we’ll discuss more in the next part) is that soldiers on the tower, because it projects outward, can direct missiles (arrows, javelins, slings, etc) down the length of the wall, engaging enemies who might be trying to scale the wall or breach it. Of course a fortress that is now being designed to resist enemies scaling or breaching large stone walls is no longer worried about a raid but rather being designed to potentially withstand a serious assault or even a siege. Defensive ditches also multiply in this period and increase in width, often exceeding 25ft in width and flat-bottomed; the design consideration here is probably not to stop a quick raid anymore but to create an obstacle to an enemy moving rams or towers (think back to our Assyrians!) close to the walls.
Over time, forts also tended to abandon the ‘playing-card’ proportions and instead favor circular or square shapes (minimizing perimeter-to-defend for a given internal area). And while even the original Roman marching camps had been designed with a concern to make it hard for an enemy to fire missiles into the camp – using the trench to keep them out of range and keeping an interval (literally the intervallum, the ‘inside the wall’) between the vallum and the buildings so that any arrows or javelins sent over the walls would land in this empty space – later Roman fortresses intensify these measures; we even see fortresses like the one at Visegrád shown below incorporate its internal structures into the walls themselves, a measure to make the troops within less vulnerable to missile fire in a siege; this style becomes increasingly common in the mid-fourth century. Finally, by the fifth century we start to see the sites of Roman forts changing too, especially in the western part of the empire, with forts moving from low-land positions along major roadways (for rapid response) to hilltop sites that were less convenient for movement but easier to defend (in the East, a lot of the focus shifts to key heavily fortified cities – essentially fortress cities – like Nisibis (modern Nusaybin), Amida, Singara and Dara. We’ll talk a bit more about how a heavily fortified late-Roman city might be protected in the next post.
In short, Roman forts in this late period are being designed with the ability to resist either serious assaults or prolonged sieges. This in part reflects a lack of confidence that the Romans could always count on being able to immediately force a field battle they could win; while Roman armies retained the edge through most of this period, the main field armies were increasingly concentrated around the emperors and so might be many days, weeks or even months away when an incursion occurred; local forces had to respond elastically to delay the incursion much longer than before until that army could arrive.
Now of course the downside to a focus like this on single-site defense (‘point defense’ in its most basic form) is that the enemy army is given much more freedom to move around the countryside and wreck things, where they would have been engaged in the older observe-channel-respond defense system much more quickly (Luttwak terms this ‘preclusive’ defense, but it isn’t quite that preclusive; the frontier is never a hard border). But of course the entire reason you are doing this is that the shifting security situation means you can no longer be confident in winning the decisive engagement that the observe-channel-respond defense system is designed for; you need to delay longer to concentrate forces more significantly to get a favorable outcome. Single-site defenses can do this for reasons we’ve actually already discussed: because the army in the fort remains an active threat, the enemy cannot generally just bypass them without compromising their own logistics, either their supply lines or foraging ability. Consequently, while some forts can by bypassed, they cannot all be bypassed (a lesson, in fact, that the emperor Julian would fail to learn, leading to disaster for his army and his own death).
And so the enemy, while they can damage the immediate environment, cannot proceed out of the frontier zone (and into the true interior) without taking some of these forts, which in turn will slow them down long enough for a major field army to arrive and in theory offer battle on favorable terms.
While it is easy to discount these shifts as just part of the failure of the Roman Empire (and we’ll come back to this idea, often presented in the form of a misquotation of George S. Patton that ‘fixed fortifications are monuments to the stupidity of man’ though what he actually said was merely that the Maginot line was such), they contributed meaningfully to the Roman ability to hold on to a vast empire in an increasingly more challenging security environment. At pretty much all stages of its development, Roman fortification on the frontiers was designed to allow the Romans to maintain their territorial control with an economy of force precisely because the Roman Empire could not afford to maintain overwhelming force everywhere on its vast perimeter. Rome wasn’t alone in deploying that kind of defensive philosophy; at any given point the northern frontier of China was guarded on much the same principles: the need to hold a frontier line with an economy of force because no state can afford to have overwhelming force everywhere. In both cases, the need for defense was motivated in no small by the impossibility of further offensive; in the Roman case, further extension of the limes would simply create more territory to defend without actually creating more revenue with which to defend it (this is why the Roman acquisition of Dacia and much of Britain were likely ill-conceived, but then both operations were politically motivated in no small part) while in the Chinese case, the logistics of the steppe largely prohibited further expansion.
This Roman system, combining local single-site defenses (which included a proliferation of walled towns as the population centers of the western empire frantically rebuilt their walls) with concentrated mobile field armies really only began to fail after the Battle of Adrianople (378), where to be clear the fortification system worked fine, the error came from the emperor Valens’ stupid decision to attack before his co-emperor Gratian could arrive with reinforcements (Valens was eager to get all of the credit and so he takes all of the blame).
Next time, we’ll actually begin to turn to what I suspect you all had in mind when I started writing this: point defenses aiming to defend an individual town or residence (read: castles and walled cities) and how they could be structured to resist attack in the pre-gunpowder era and we’ll get into the function of curtain walls, outworks, towers, citadels and so on.
- Please do note, not entirely constant. Changes in the organizational structure of the legions are reflected in the structure of the camp. For more on this and Roman camps in general, check out M. Dobson, The Army of the Roman Republic: The Second Century BC, Polybius and the Camps at Numantia, Spain (2008) and while Polybius insists that Roman camps never bow to the terrain, they clearly do sometimes. So they aren’t perfectly consistent, but they are remarkably consistent.
- One of these days I may get around to writing about the debate; in the meantime, for a ‘quick’ primer on it, look up J.E. Lendon, “Primitivism and Ancient Foreign Relations” CJ 87.4 (2002): 375-384. For a long primer on it, there is little substitute for E.L. Wheeler, “Methodological Limits and the Mirage of Roman Strategy” JMH 57.1 (1993): 7-41 and 52.2 (1993): 215-240).
- The formulation here and the thinking in terms of what the various elements of the Roman system were intended to accomplish goes back to E.N. Luttwak’s The Grand Strategy of the Roman Empire (1976), the book that initiated the long-running strategy debate mentioned above. It is not without its own problems and again I’d direct you to the two articles in the previous note to read about that. Nevertheless, as an overview of the Roman frontier defense system, I think the basic description that follows holds up (the main debate is less about the structure of the system and more about the intent of it in any event).
- The Romans win those rumbles more often than they lose, but it is hardly a one-sided affair!