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 […] 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 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.

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 [part] 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).

Bret Devereaux, “Collections: Fortification, Part II: Romans Playing Cards”, A Collection of Unmitigated Pedantry, 2021-11-12.