Collections: Shield Walls and Spacing: Hollywood Mobs and Ancient Tactics

This week, we’re going to take a look at a different aspect of ancient infantry tactics: how heavy infantry shield formations work. While I’ve framed this around ‘shield walls,’ not every kind of shielded heavy infantry fought that way and in practice the line between what is a ‘shield wall’ and what isn’t comes down to formation structure and can be quite fuzzy. For the sake of this post, we’re going to frame the topic rather more broadly: how do infantry formations of heavy infantry with shields who are mutually-supporting function? That can mean overlapped shields, but as we’ll see, it usually doesn’t.

This is an element of ancient and medieval fighting that popular culture almost invariably gets wrong, with formations that are typically both too dense, too complex and trying too hard to be clever. We’ve critiqued a few of these before: the preposterous tactics of Game of Thrones come to mind, but fit into a broader trend of pop-culture writers trying to overcomplicate and outsmart classic battle tactics. So lets instead talk about some of our best-attested close-order infantry formations: how dense were they? How did they function? And what role did the shield play?

As always, if you like what you are reading, please share it as I rely on word-of-mouth to find readers! And if you really like this and want to allow me to be able to buy my own shield (a thing I have wanted to do for some time), you can support this project over at Patreon; supporters at the patres et matres conscripti level even get to vote on future topics (like our recent look at Greek and Phoenician colonization). 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, assuming there is still a Twitter by the time this post goes live. I am also on Bluesky (@bretdevereaux.bsky.social) and (less frequently) Mastodon (@bretdevereaux@historians.social).

Shields on the Screen

As I implied above, the depiction on screen of how a shield wall functions is usually quite bad. There are two common problems. The first is are shield walls that are much too much like walls. Often the shields are vertically stacked to make a rather literal wall, as with this example from The Last Kingdom:

That in turn may be taken up to eleven by having some kind of complete shield coverage, sometimes as a stationary dome of shields (as in 300) or in some kind of take on the Roman testudo, as in Troy (2004), pictured below. In this case, the formation is presumably meant to provide safety from arrows.

You can probably already see the immediate problem: these shield walls often cannot move and you can’t actually fight in them (indeed, in the example from Troy, they have to break the shieldwall to actually come into close combat). They also, perhaps somewhat paradoxically, trade that mobility for a lot more protection than they need. As we’ll see, you just don’t need this much shield coverage in order to render your infantry formation tolerably safe from arrows, while on the other hand giving up the ability to move is an enormous disadvantage because of course it means an opponent can simply keep shooting you.

The other problem is one I’ve seen more of lately which is a formation that has dedicated two-handed shield-bearers supported by dedicated two-handed pikemen. This is, to my knowledge, not a formation ever used in history. It would be absurdly vulnerable, given that the moment either the pikeman or the shieldman is killed, that entire file stops functioning. Some armies did have units which were composites of different troop types (this is common in Han-dynasty Chinese armies, for instance, and a standard feature of warfare in the Near East in the early iron age), but never in a way that makes each totally reliant on the other providing either all of the offense or all of the defense. ‘Shieldbearers’ of various types were effectively always spear-and-shield infantry who could defend themselves in close combat. That said, there is a historical pike-and-shield formation, the Macedonian sarissa-phalanx, and we’ll talk about how that actually works (it does not include dedicated shield-bearers).

In all cases, these formations are extremely dense, with men typically both shoulder-to-shoulder with no lateral space but also with very tight rows (that is, very close front-to-back). This is a more complex issue, but as we’re going to see, shield walls don’t need to be this tight and I think on balance generally weren’t.

None of these formations work particularly well, displaying a poor grasp of what is possible in a chaotic, swirling battlefield. So we’re going to look at what heavy infantry formations were actually like, with a focus on some of the best attested examples from the ancient Mediterranean (some of which are shield walls and some of which are not).

Now as I look back on this post, before hitting ‘publish,’ I see that it comes in essentially two parts: a quite long and rather technical discussion (with diagrams that I spent far too much time making!) of shield coverage and formation spacing, followed by some rather briefer conclusions at the end about how these formations actually worked. I’ll not cut the technical stuff – we are here for pedantry, after all – but if you want to skip through it, the ‘how to actually shieldwall’ part of the essay is at the end, under the section heading, “How to Actually Shieldwall.”

A Boy and His Shield

Let’s start by talking about the basic dimensions of these shields. The Greek aspis, the classic shield of the hoplite, was about 90cm in diameter; some later Macedonian shields (used by pike-wielding phalangites by means of a shoulder-strap) were a little bit smaller, but only a little. The Roman scutum during the republic was about 120cm (tall) by 60-65cm (wide), curved so that the soldier could put his body into it.¹ The later rectangular imperial-period version of this shield was a touch smaller; the shield recovered from Dura Europos measures 105.5cm by 41cm. The oval La Tène shield clearly could vary quite a bit, but was in the same basic size range, with the most intact example from the site of La Tène itself (where deposition in water resulted in the wood being preserved) looks to have been roughly 110cm by 51cm. Moving into the Middle Ages, the ubiquitous center-grip early medieval roundshield used by Franks, Angles, Saxons, and Scandanvians (read: Vikings) of various kinds varies a lot in size but tends to fit in the range of around 75-90cm in diameter, which you may note is the same rough size as the aspis.

The point I am trying to make is that large, battlefield shields tend to vary within a relatively narrow range of sizes and basically two major shape categories, either round or oblong (a category which fits both ancient oval shields but also later kite shields, which have broadly similar dimensions, being around 90-100cm tall). Oddly, one repeat feature in popular depictions are square – not rectangular, but square – shields wielded in the manner of round shields; these, as far as I can tell, did not exist (it would have been a fairly silly shape to have a flat, square shield). Of course those categories are reductive and there are all sorts of variations within them – the degree to which the shield is dished or curved, the grip-type, the presence of metal bosses or rims, other reinforcements, and so on. Those variations are meaningful and often point to how a shield was intended to be used in its context.

But as noted, the shield wall is a very widely used combat tactic, almost certainly the most widely used agrarian infantry formation and so it is going to depend on the features that are common to these shields, which is the basic size, being in the neighborhood of 75-90cm tall and either round (and thus 90cm wide) or oblong and around 50-60cm wide.

And the first thing we ought to note is that this is actually quite a lot of protection. These shields are big, indeed often bigger than the props used to represent them in films or video games. To get a sense of how big, let’s try a convenient little diagram here of various shields sized against a person. Note that my little person-diagrams are meant to show width-at-shoulders (a bit more than 45cm (1.5′); the standard way to measure width-at-shoulders is from the top right above the joint, but of course the arm does not have zero-thickness, so a person’s shoulders, arms held down, are a bit wider than this). An actual soldier occupies more space than this, usually around three feet, accounting for both his limbs and weapons – however when thinking about protection, it is the trunk of the body we care about, thus the width-at-shoulder diagram. But of course in a fight, each soldier is going to turn his body away, advancing the shield and refusing the rest of his body; turned like this, these shields cover basically the whole body.

And of course these shields also have different shapes. The aspis and the medieval round shield are, well, round meaning, which means when held defensively, they’re going to be wider and offer more coverage around the central torso and narrow around the shoulders and the hips. This isn’t as big of a disadvantage as it seems at first – one thing to always keep in mind is that, biomechanically, all close-combat strikes must originate from shoulder-height. The further they stray from that height, the more power and reach are lost. Meanwhile, as we’ve discussed before, the chest and head are the key parts of the body that have to be most protected and this even a rounded shield does well.

The other thing to note in terms of what kind of coverage is going to be required is that no one stands in combat with their chest pointed straight at the enemy. Instead, a fighting stance places the legs apart and their body nearly perpendicular to the line of battle, presenting a narrow target (with their shield arm turned out towards the enemy). The right side of the body might be brought forward for a strike, but the ‘neutral’ position when actually in close combat is as shown in the lower set of diagrams up there.

That in turn has implications for the angles a shield protects from. Even held close to the body, a good shield doesn’t just protect from dead ahead, but offers a pretty wide arc of protection because the body is turned narrow-face towards the enemy. But shields don’t need to be held passively this way: even ‘strap-gripped’ shields like the aspis (with its porpax and antelabe grip) have a pretty wide range of movement in an arc in front of the body. Center-grip shields (like the La Tène shield, the Roman scutum or the early medieval roundshield) can be potentially even more mobile in front of the body, as they can be advanced much further out. Efforts to reconstruct medieval shield fighting – necessarily speculative as most of our fighting manuals come from the later Middle Ages when shields were less central to combat – tends, reasonably in my view, to extrapolate from buckler-fighting treatises and from period artwork that on the offensive shields were often extended out to open attack routes. But even then, because they’re big, a shield used offensively in this manner could simultaneously close off huge regions to attack, as seen in these diagrams here:

As an aside, someone just looking at those diagrams might wonder why you would ever advance your shield like the figure on the right. And the answer is, if you can keep your opponent in the ‘blocked’ zone (the darker blue), you can use the shield to intercept and turn aside their weapon, move into their fighting space (‘into measure’ in fencing terminology – into range to land your own blow) and use your shield to move their shield out of position to apply a strike. Its an aggressive stance, but also an effective one and it appears not infrequently in artwork. You could also, of course, strike with the rim of your shield or the central metal boss (the grey nub in the middle) with great effect, and we know that was done too.

But the point I want to make with these diagrams is that a soldier alone with that shield can use it both offensively and defensively and also create a fairly wide arc of relative safety against a single opponent. The single opponent is important, because notice how crucial moving the shield is to achieving that comprehensive protection – a second opponent might either time their attack to catch the shield out of position or else they might be unnoticed entirely as our shield-bearing infantryman focuses on the first. But so long as someone else is occupying that potential second enemy, our shielded fellow has a wide lateral arc of protection from just the shield! That in turn has an implication for these formations: the fellows covering your flanks do not need to be right next to you, and as we’ll see in most fighting formations, they weren’t.

And of course from there we also need to consider vertical protection. And here, period artwork actually does us a lot of favors because they frequently show shields at scale with the figure in profile or straight on to give us a good sense of how much of the body they covered. And the answer was by far most of it. The Roman scutum is surely the most protective – at c. 120cm and with an oblong shape (and a curved shape) it can cover basically the entire body from ankle to shoulder, while its center grip makes it easy to pivot around the body or lift up to cover the face. We’ll get to, in a moment, why I think the Romans went for this rather extreme form of shield in a moment. But it is not so different from the others. With a 90cm diameter, a hoplite aspis is going to cover a hoplite from just above the knee to the shoulder; a medieval roundshield has a similar level of coverage (though again let me stress that these shield types vary a fair bit in size and these are rough average sizes; some are bigger, some are smaller!).

And because, as mentioned any close-combat attack originates, mechanically, from shoulder-height, that’s quite good coverage. Striking feet or the lower-legs with a sword or a spear is doable, but tricky and the long extension to hit those targets means the attacker is going to be vulnerable themselves (especially as spears are often used overhand in such close combat and so are striking down from above shoulder height and so would have to pass through the shield to get to the legs). Making matters worse, feet and legs move a lot in combat, making them harder to hit and worse yet, strikes to the limbs may not be lethal or even debilitating and it is a poor trade if, to score a glancing blow on someone’s shin, you open yourself up for a lethal blow to the neck or face. The one situation where you might be really worried about hits to the legs not protected by the shield would be in the case of incoming ranged fire, like arrows, shot at close range (because at long range, the missiles will be plunging downward) – put a pin in that for a moment, we’ll come back to it.

However this system does neatly explain why you do not need dedicated, pure-shield-bearers (with no spear or sword) as one often sees in TV or video games: one-handed shields are perfectly capable of protecting the whole body fairly well. There’s simply no need for some super-jumbo shield that requires both hands or the whole of a soldier’s attention.

But with that all laid down we can now turn to the shield wall as a formation and how multiple shield-carrying infantry interact with it. And let’s start with:

Phalanxes

When thinking about how a shield wall functions in battle, we have to think of it as a formation. Now this is a narrow path to walk with pitfalls on either side: on the one hand, these were more-or-less regular formations; these are not mobs. The men in these kinds of infantry formations formed up roughly in square grids, they had somewhat standard spacing within those grids, they could move together in at least very basic maneuvers (like ‘march forward’) while more or less keeping formation and so on. On the other hand, these men are not robots and are often not extensively drilled or trained in maneuvers, so everything about these formations is approximate. As we’ll see, spacing is often determined by rough body-part measurements (like the length of an arm or forearm) rather than someone coming out with a measuring stick and in actual formations there’s a fair bit of space for jostling and movement, because of course there is. So the first trap is assuming too little organization, like these armies are little more than an unruly school class on a field-trip, while the second trap is instead imposing upon them the ultra-regimented values – rarely achieved in practice – of early modern European gunpowder armies with their ideal of the ‘mechanical soldier.’²

There are a few key ‘variables’ for any given formation within this framework. First, there is the basic question of depth: how many men does each file (that is, each line from the front to the rear of the formation; files are vertical, rows are horizontal) have? Even within military systems, there’s a lot of variance. For instance, as Roel Konijnendijk has noted,³ even among Greek hoplite armies – an identifiable if politically fragmented military system – there is a wide range of reported army depths, though clean multiples or divisions of 6 and 8 are very common and 8 is the most common, suggesting those might have been typical depths. Historically speaking, one does see greater depths than this (the standard Macedonian sarissa(read: pike)-phalanx depth was 16), but rarely shallower ones. Depths of 2 or 3 – very common in Hollywood – are very rare on actual battlefields. Depth provides both strength by having men who can replace someone who falls in front but also by providing a morale-strengthening effect, building confidence from a deep formation.⁴

Formation width is often more variable, with a lot of formations having a set depth and then simply being as many men wide as could be managed, but we do see formations that have ‘standard’ widths as well. The Roman maniple (manipulus, ‘handful’) was the basic building block of a legion’s fighting line in the Middle Republic and was 120 men standard, typically arrayed six-deep for battle, so it had a formation width of 20 and a depth of 6 (though as we’ll see, it could probably also shift to a tighter file with a depth of 3 and a width of 40). Likewise, the syntagma, the 16-by-16 pike-square of the Macedonian and later Hellenistic armies, which was the standard building block of their battle line. That said, we’re less concerned with width here for now.

The other two measurements that matter are what we might term file width and row depth. ‘File width’ is the amount of horizontal space each person in the file takes up, while row depth is the amount of vertical space each person in a row takes up. Put another way, file width is how we figure out how closely soldiers were to the soldiers to their right and left, while row depth is how close they were to the ones in front and behind.

Those questions are tricky to figure out, however, both because they are much less often reported in our sources (many cultures, including that of Greek polis armies, have no attested figured) and because even when they are it can be really tricky to figure out what exactly is being counted and there are a few ways to express the idea. To take file width, you could count ‘tactical space’ (the space from the edge of the next person on the left to the next person on the right, the total zone in which the person in the middle can move), or perhaps the space from one right shoulder to the next left shoulder (the ‘interval’ or open space). I am going to focus on what we may properly call ‘file width’ which instead measures from one right shoulder to the next right shoulder here so that we have a standard measurement. That’s useful because this is a measure which counts every centimeter of frontage just once, without missing any. But that’s often not how file width is described in our sources, so often some conversion is necessary. Row depth has all of the same problems in counting and is even less commonly described in our sources.

The curious may be wondering, “why not just use artwork?” The answer here is fairly simple: ancient and medieval artwork is not typically realistic in its positioning of figures in a scene nor is it intended to be realistic (the artists are not bad, they are trying to do something other than represent realistic file width): figures instead tend to cluster together, overlapping often in unrealistic ways for the sake of artistic convention or composition or simply to fit all of the figures in the ‘frame.’ When we do have attested file widths, as we’ll see, artwork very much does not reflect them as we’ll see really clearly when we get to Roman file widths in a second.

That leaves us with literary sources: we do not have attested file widths or row depths for the hoplite phalanx or for formations like the Gallic shield wall, but we do for the Roman legion and the Macedonian phalanx. Polybius treats both formations (Polyb. 18.29 and 18.30) while a pack of what we might term military philosophers, most notably Asclepiodotus, offer discussions of the Macedonian formation. Difficulties assail us immediately: Polybius is the only eye-witness to these formations (the later military writers are writing about Macedonian warfare as an intellectual exercise and by their time, no one has fought that way in a century or more) and the interpretation of his passage is tricky, for Polybius seems to say that their spacing is the same and then immediately that it is very different. Neither formation is strictly speaking a shield wall (though the Macedonian one does create a wall of shields!), as we’ll see, but the spacing guidelines will get us somewhere, so let’s discuss them.

We can start with the Macedonians. Polybius gives the file width – for the soldier and his arms and the empty space together – as three feet (c. 90cm). Peter Connolly⁵ argues, persuasively in my view that Polybius is using all of his measurements a bit roughly. No one, after all, is pulling out a ruler in order to get the spacing right when forming up for battle. Instead, Connolly notes that Polybius provides a slew of measurements in the passage all of which are fairly clean multiples of a ‘double cubit,’ which is the length of one’s arm (the length of your forearm is a cubit, add your upper arm and make a fist and you now have about two cubits). Connolly supposes – and I think this is right – that Macedonian troops lined up by holding one arm level with the ground (with a balled fist) and placing it on the shoulder of the next man, creating a very rough two-cubit interval (technically a bit more).⁶ That leaves just one more interpretive question – are these men holding their arm out to the right or left (and thus we’d need to add the width of their shoulders) or are they instead lining up by putting their arm to the back of the man in front of them when marching and turn left- or right-facing into combat formation.

If the latter seems silly, it shouldn’t. We can be pretty sure, from Roman roads, among other things, that armies tended to march in column (roads are narrow) using their combat-files as marching rows (so that to shift into fighting formation, everyone just had to face right or left). That made deploying armies easier too and that ‘trick’ for getting quickly from column into a fighting line gets used a lot in history. In this case, I think the Macedonians form up with that two-cubit interval as a vertical spacing in column (with men shoulder-to-shoulder marching to use up the road) and then simply pivoted (which also conveniently creates the closely spaced rows in the combat formation which Polybius describes in the same passage); for reasons that will become clear, I think the 90cm file width (remember: right shoulder to right shoulder) is both correct and an inherited trait of the Macedonian phalanx that it owes to the earlier hoplite phalanx. That means that we’re not adding the shoulder-to-shoulder length but just the chest-to-back to get our file width, which comes out neatly to about 90cm, a figure I want you to put in your pocket for a second.

And what do you know, but our later military writers also record this file width. In particular Ascleopiodotus records three standard file widths: an open order width of c. 180cm (6 feet), a standard ‘compacted’ (πύκνωσις) width of 90cm (3 feet, keep in your pocket) and then a super-compact ‘shields together’ width of 45cm (1.5 feet) called συνασπισμός (lit, “shields together”). The other writers echo these standards and you may immediately notice the same neat multiples of cubits: this is a one cubit, two cubit or four cubit setup. Which is to say, again using the quick form-up trick, it is half-an-arm (hold your elbow to your side, extend your forearm, that’s your spacing), one arm or two arm (have your mate also extend his arm, move to when your arms meet).

Now we can diagram the tighter two of these formations (I think open order is, for now, pretty self-explanatory) and see some things.

Note that, compared to the aspis in the early diagrams, which was the hoplite aspis, the Macedonian aspis is a bit smaller, usually around 75cm. On the top of the diagram, we can see the relative position of the standard ‘compacted’ width marching forward with sarissa raised as they would while maneuvering (once you lower the sarissa points, you can’t readily turn the formation). At a casual glance, the interval makes perfect sense: with the shield and the pikes, this is as about close as you can get together without being in each other’s way. If you lower pikes in this formation, you get the middle diagram, with the dark line indicating where the front man’s sarissa is. But the whole point of this formation is that the front five sarissai can get in on the action and clearly they can project through that open space in the intervals, angled with each row so that no one’s sarissa-butt (which is going to be a few feet behind the fellow holding it for balance reasons, and these are held at waist-height) is poking into anyone (which would be the problem if you angled them or tried to have everyone’s sarissa placed vertically over the next).

Then you have the ultra-close synaspismos formation. And just based on the size of the shield, the normal size of a person and the need to have space for the sarissa, we can see that this is basically as tight as the formation can possibly be. Because it is exactly twice as tight as the ‘compacted’ formation, shifting between the two would be easy: just have the back half of every file move into the interval between the front half of every file. The Macedonian phalanx even has junior officers at this exact spot in the line to make the process work. At the same time, this formation is so tight that you can see how hard it is going to be to make it work. and it almost immediately makes sense why this ultra-tight formation is not the standard fighting formation (that is, according to our sources, the c. 90cm ‘compacted’ formation!). The shields have to be about about their maximum usable forward arc simply to fit everything; with all five sarissai leveled through the tiny space left, you aren’t going to be able to move them very freely. And remember, these weapons are supposed to be moving, jabbing at approaching enemies, trying to find gaps in shields and armor. That said, in this synaspismos, the level of shield coverage is very high from basically all directions, which would make this formation very resistant to attack (particularly by ranged weapons, since melee attack will be discouraged by the hedge of sarissai). You could see it as a closed formation to resist ranged fire, but you’d want to loosen it again to actually fight.

It is thus worth noting even here – we’ll see it even more clearly in a moment – that the most dense formation was not, in actual battle conditions, generally the best, because soldiers even using pikes (much less swords or spears) need space to move both themselves and their weapons in order to fight effectively. This is one of those cases where the video-game logic (pack as many men as close together so they can all deal their 1d10 piercing damage) doesn’t actually work on real battlefields where the physics are every bit as complex as real physics.

But of course the sarissa-phalanx was an unusual creature, a pike-and-shield formation. What it does demonstrate is the pointlessness of Hollywood formations where pikemen (without shields) are protected by men who carry nothing but shields, but what about shield walls? Here, I actually think the evidence we’ve run through is useful because of how the sarissa-phalanx developed: it was an innovation on the hoplite phalanx, which was a traditional shield wall. In particular, what I suspect is going on here is that the Macedonians when developing their arms started with a standard file-width of 90cm (and thus a natural double-width of 180 and half-width of 45cm) they inherited from the hoplite phalanx (which has a roughly 90cm shield!) and then, shifting to a pike formation had to make a smaller shield to make a tight formation work with the shields and the pikes. Remember, no one is measuring the synaspismos out: you simply form a normal 90cm file width file and then have the back half of the file advance forward into the open interval.

No source attests the width of the hoplite phalanx, but if we work from this assumption, the results broadly speaking make sense.⁷ Measuring out a 90cm file width for a hoplite is easy: the file width is roughly the diameter of the shield. Line up so your shield is just about touching the shield of the hoplite to your right and the formation is set. Assuming a normal interval around 90cm (because again, the diameter of the hoplite aspis is not perfectly fixed and these formations are not perfectly regular) also makes sense of Thucydides’ famous remark that hoplites in the advance tended to drift rightward to seek more protection from the shield of the man to their right (Thuc. 5.71). For that drift to happen, there needs to be open space for each hoplite to drift into.

Still at this interval, each hoplite is still supporting his neighbor, even though the shields do not overlap, because he is still closing off that route of attack by occupying it and forcing the attention of any enemy attempting to move into it. And Greek writing resounds with the importance of holding a position in the line. Thus Agesilaus supposedly quipped to a physically lame Spartan going to war and thought he needed a horse, “war has need, not of those who run away, but of those who stand their ground?” (Plut. Mor. 210f) and of course the famous Spartan aphorism of Demaratus that a soldier carries their shield for the entire line (Plut. Mor. 220a). Likewise Tyrtaeus’ praise that, “It benefits the whole community and state, when with a firm stance in the foremost rank a man bides steadfast, with no thought of shameful flight, laying his life and stout heart on the line, and standing by the next man speaks encouragement. This is the man of worth in time of war.”⁸

The aspis, like all of these shields is big and can cover pretty much the whole body in fighting stance; Tyrtaeus actually says this too, calling “Let every man, then, feet set firm apart, bite on his lip and stand against the foe, his thighs and shins, his shoulders and his chest all hidden by the broad bulge of his shield.”⁹ So it isn’t necessarily the needed overlap of shields that matters here, but the presence of the hoplite on each side who both physically occupies the space that makes up your vulnerable flanks but also by their own threat forces the opponents diagonal from you to face them and not you. In short, the support being offered is as much psychological – both for you and the enemy – as it is physical.

At the same time, denser formations were clearly employed. We do get a few scattered references to synaspismos dealing with the older hoplite phalanx (e.g. Xen. Hell. 3.5.11, 7.4.23, though only the latter is used literally of a formation), so the clearly could form up tightly as well. And if you look at the diagram above, the spacing barely works out; because hoplites held their spears overhand, it is fine if there is no space for them amidst the men. It’s possible the hoplite synaspismos was a bit looser than this. Still, again, it can be little wonder that this seems not to have been the common fighting formation.

Now lets jump over to shield formations in the western Mediterranean.

The Tortoise and the Hispaniensis

And this brings us to the various multi-level shield walls or shield domes or variations on the testudo-formation. Now the testudo, a Roman formation which used the large Roman scutum to create a moving ‘box’ of shields (on the top, front and sides) for a very high degree of coverage was a real thing. You can see an example from the Column of Trajan below. But while this was a Roman formation, it was hardly the normal Roman formation and in fact one thing we hear quite a lot in the sources was that the testudo was bad for fighting in. Instead, the main purpose of it was in situations where the Romans would have to endure sustained ranged attack. That was quite rare in field battles, so the normal use-case for the testudo was in sieges, allowing infantry to approach walls and gates with minimal losses.

Even without the testudo, the popular vision of Roman fighting is the use of really dense shield walls. Note for instance the really densely packed formations in Gladiator (2000):

Likewise, the opening scene of HBO’s Rome, which I actually like a few things about, but these guys are really densely packed. Here’s an overhead shot while they’re changing out the front rank (a thing the Romans did, although how they did it is going to become a lot more obvious in a second):

The Romans did not fight this way, at least, not in the late Republic and early Imperial periods.

And to see how, we can go back to our question about spacing and back to that same difficult passage of Polybius. Fortunately, I think Michael Taylor has largely succeeded¹⁰ in getting what Polybius is trying to say. Polybius comments of the Romans that “Each Roman with his equipment also occupies three feet” but that their style of fighting means they also “clearly require a space and flexibility between each other, so that each soldier must have three feet from the man to their flank and rear, if they are to be effective.”¹¹ What confused scholars in that passage is that Polybius seems first to say Roman spacing is the same, and then that it is more spaced out. What Taylor correctly recognizes is that what he’s actually saying is that a Roman soldier physically occupies as much space as a Macedonian soldier, but requires more open air to move in.

On the other hand, the easy reading here would be to assume that Polybius is imagining three foot open intervals between each soldier, a file width of six feet (three ‘double cubits,’ or c. 1.8m), which is way too wide. Instead, Taylor argues that Polybius is expressing tactical space, but that measure – as noted above – double counts some space. If we account for that, a tactical space of six feet (the space from the right shoulder of one soldier, through a middle soldier to the left shoulder of the next, so interval + body + interval again) produces a file width of 4.5 feet (right shoulder to right shoulder, so interval + body). That’s a wider interval than the Hellenistic phalanx (or indeed, the hoplite phalanx), c. 135cm to their c. 90cm (remember, these measurements don’t need to be exact, they’re not lining up by ruler), but it is a natural enough spacing to achieve quickly, because you can just line up holding out your right arm at touch distance to the left shoulder of the next fellow and get the spacing roughly right.¹²

That produces a spacing system that looks like this:

And it produces exactly the sort of effect Polybius then goes on to describe: a single Roman soldier might have to deal with ten sarissai (Polyb. 18.30.9) in order to close with a Hellenistic phalanx (though of course readers of our previous post may note the Roman has weapons with a reach advantage: his two pila, which could be a real deadly threat even out of reach of those points). Consequently, I feel fairly confident this reconstruction is basically right. Polybius is, of course, fudging the details a bit; the Roman interval is not twice the Macedonian one, so while some Romans may face two files of Macedonians, others will face just one (though in practice in that looser fighting order, the Macedonians may aim their sarissai and so focus their ‘pokes’ on an advancing Roman).

But what I want to note here, coming back to the Hollywood tactics, is how wide these intervals are. Once the Romans turn into their fighting stance, shield forward and body turned at an angle, the shield both completely covers their body from a fairly wide angle (because they can put a good portion of their body into the dish of the shield) but it also leaves a LOT of open space.¹³

Now the immediate response is to assume this is a terrible vulnerability – couldn’t an enemy rush into the gap and flank these men? But of course an enemy who tried to do this would immediately be flanking himself, with both Romans on his right and left, and two more (the next row) in front able to move forward and engage him. Small intervals like this are not generally fatal to these sorts of formations – you only need to avoid such intervals if you rely either on overlapping shields or on pike coverage. Meanwhile, the wide intervals give the Roman soldiers a lot of space (remember: I have not included arms in these drawings, but they’re kind of long!) to work their swords, which can both cut and thrust effectively, as Polybius notes (2.30.8, 3.114).

But what I think is really instructive is how the Romans use tighter formations. As Taylor notes,¹⁴ we see repeatedly in the sources that the Romans will pull into a tight formation, typically to resist missile fire, before having to loosen up again to actually fight. Perhaps the most famous instance of this is at the Battle of Carrhae (53) where Cassius Dio reports (40.22) that the Parthians were able to put Crassus’ legions in a bind: if they closed ranks to resist arrows, the Parthian shock cavalry charged, taking advantage of the fact that the Romans couldn’t fight well in close combat formed up that close. But if they spread out to fight, the cavalry withdrew and peppered them with arrows. But that’s hardly the only example, with Romans closing up against missile fire and then widening out to actually fight in Spain (Livy 28.2), and against Macedonian skirmishers (Livy 42.65) and during siege operations in Sparta (Livy 34.39). And given the spacing it seems to me that you’d achieve this shift the same way the Hellenistic phalanx does: the back half of each file advances into the open interval between the front halves of each file, creating a twice-dense but half-deep formation. To stretch out again, you simply need enough space for the ‘front’ half of each file to advance again while the rear halves hold position.

Strikingly, to me at least, the Roman scutum seems almost perfectly sized to almost but not quite overlap in this formation. I think that in turn connects with its design: the strong curve means overlapping shields are not required to get good protection against missiles, just close-set shields. By contrast, Gallic shields were flat and we know they sometimes overlapped them because Caesar tells us it happened (Caes. BGall. 1.25). Notably, Caesar describes this overlapped-shields formation of the Gauls as a phalanx (he uses the Greek word, transliterated into Latin). La Tène shields are a bit narrower than Roman ones generally (though the physical evidence for this is limited, but our literary and artistic sources support the assumption too), around 50-55cm wide, so an overlapping formation would have to be quite tight – in fact, it would probably have been right around the 45cm file width of a synaspismos formation. At the same time, we know they didn’t always form up this close as our sources note (Polyb. 2.30.3; Livy 38.21.4) that the lack of a curve to the La Tène shield could be a vulnerability given its narrowness when faced by missile weapons; that only makes sense if the shields were not overlapping as these unfortunate Gauls were being showered with Roman javelins.

How to Actually Shield Wall

I think the problem a lot of popular depictions face is that they know this was an effective tactic and so they go looking for ‘clever’ ways to make it work. This is actually one my recurring gripes with depictions of battle in fiction: the desire for clever, counter-intuitive or unexpected solutions to battles. Like most tactics, shield walls are effective, but not particularly clever: it is simply a basic principle (infantry with shields offering mutual support) extrapolated out into a large formation. So how do you actually make a shield wall work? 

Well first you start with one kind of soldier, because part of the reason these formations come with depth is so that the death of one soldier in the line doesn’t rob all of the others of the protection their neighbors bring: you do not want a cascade collapse the moment the first fellow catches an arrow and no one around him can do his job.

Next, you need to get these infantry into formation, which is going to be a line wider than it is thick. The classic way this was done was to line up men, in their spacing for the march and then march them out, taking a 90-degree right (or left) turn when you reach the intended deployment area, thus turning your front into your flank and your rows into files. Once in position, every soldier makes a 90-degree pivot and you now have a fighting line. There are quicker and more complex ways to form from column into line, but that turn is the easiest way and as you can see above, when you assume that’s how the thing is being done a lot of things start to make sense. These armies move slowly (foot pace) and aren’t hard to see, so it is rare – unless that is a lot of cavalry involved – that one army is going to surprise another before they can form a basic line in this manner.

The spacing of your heavy infantry formation is going to be approximate, not mechanical, but it will be at least somewhat regular. These fellows aren’t machines, but this army also isn’t a mob, even if it isn’t composed of professional soldiers. Spacing out at arm’s length or shield’s width is not hard to do; likewise having the back half file fill the interval to make a closed formation is also not that hard.

But how much spacing? It probably won’t be shoulder to shoulder. The preference for these sorts of ultra-dense formations in fiction is itself, I suspect, a product of early modern gunpowder tactics, which eventually did become that tight in order to maximize fire over frontage (but also in a style of fighting where no one needed to move laterally!) – but it should not be imported back into primarily melee formations. Instead, as noted above, standard fighting intervals tend to be upwards of 80cm or more, which is to say there tends to be about as much open air along a single row as human body (since shoulder width (around 45cm) occupies about half of a 90cm file width). For formations using axes and swords over spears, you might expect a bit more width to give space for those weapons.

If the danger is attack by missiles – arrows, javelins, etc. – the formation may close up to meet it, forming a dense shield wall with adjacent or overlapping shields. But here it is worth noting the danger generally has to be sustained missile fire over some period of time, both because that dense formation is bad for fighting in, but also because large shields already provide a lot of protection. These shields are big enough that a soldier can generally be safe simply hunkering behind their own. Moreover, arrows are not machine guns; from Alexander to The Lord of the Rings, all sorts of films love arrow volleys with devastating lethality, but in practice infantry regularly marched through heavy arrow fire or stood under it for extended periods. Shields – and for that matter, armor – work. Infantry advancing in fighting order in the open against arrow-shooting enemies may close up briefly, but they may also stay in fighting order and simply push through the arrows, relying on their armor and big shields to protect them. What they will not generally do is assume an immobile or nearly-immobile shield-dome or box, because that just means spending more time under fire.

The other thing worth noting here is that when actually fighting, these formations are not designed for a rugby scrum, which is in turn suggestive of the tactics used. The Romans felt free to fight with a standard file width and row depth of around 135cm, which should tell you that they didn’t expect most battles to come down to physical shoving. I should note this is also why modern riot police tactics (and some modern sport reenactments) can be an imperfect guide to the function of these formations: billy clubs and HEMA blunts are not sharp swords and spears and so the behavior of the formation can be quite different. Now you might charge and try to knock an opponent around with an initial impact and the mass movement of bodies doubtless could produce a ‘press’ or ‘crush’ (especially, I imagine, on initial impact), but in order to fight effectively, these fellows are going to space out a bit, often far enough apart that their shields are only barely touching or not even touching at all.

Instead, the communal protection of the shield wall is about the ability first of each soldier to threaten the space on the flanks of their neighbors (so that enemies cannot blindside them from the flanks) and then also to be above to move their shield or weapon secondarily to intercept blows to that neighbor. That protection is likely to be more about movement, than a static wall of unmoving, interlocked shields and the support has as much to do with presence as it does with physically blocking attacks.

Because in the end as the Romans show, a looser formation can be made to be savagely effective. But what is crucial is the psychological impact of the shield wall, not its mechanical impact: it encourages confidence, it reminds men that they stand next to their family, friends and neighbors, that they will be shamed to throw down their shield and run away. It is ironic to me that Hollywood writers love trying to outsmart the physics of the battlefield because the victors in these kinds of battles often had more to do with motivations, emotions and character – the sort of things good writers should actually be concerned about.

The battle is always as much in the head as in the hand. Even when – perhaps especially when – the thing you have in your hand is a big damn shield.

1. The width measurement is end-to-end, not the length of the arc, so it is comparable to the aspis measurement. We have one surviving shield of this type, the Kasr el-Harit shield, but Polybius also describes its shape and size, Polyb. 6.23.2.
2. On this latter trap, see especially R. Konijnendijk, Classical Greek Tactics (2017), which had an extended discussion of how the assumptions of early Prussian writers – who assumed a regimentation similar to or even exceeding their own, famously regimented armies and in so doing erred substantially.
3. op. cit.
4. Consequently, within a military system, commanders willing to stretch a line thin are often expressing some confidence in their troops’ training or commitment to the cause.
5. “Experiments with the Sarisa” JRMES 11 (2000)
6. The greater significance of this realization is that it clarifies about how big the cubit is that Polybius is working with – around 80cm or so, which then lets us interpret the measurements he gives subsequently for the length of the sarisa – the Macedonian pike.
7. The most sustained discussion of hoplite phalanx file width I have to hand is Schwartz, Reinstating the Hoplite (2013), 157-167, though the overall thesis of that book has not found wide adoption among scholars currently studying hoplites.
8. Trans. M.L. West
9. Trans. M.L. West again
10. M.J. Taylor, “Roman INfantry Tactics in the Mid-Republic: A Reassessment” Historia 63.3 (2014)
11. Taylor’s translation, op. cit..
12. Just to clarify, the difference is that the Romans are probably lining up with their arm held out horizontally to the right, so that the file width is (shoulder width + arm length) which comes to roughly 45cm + 90cm (c. 135cm), whereas the Greek system, I suspect, is doing this holding the arm forward touching the back of the shoulder of the fellow to your front, thus removing nearly all of the person’s width and getting an interval of just the arm length, which is that 90cm double-cubit, plus or minus a few centimeters.
13. Interestingly, on that fighting stance point, Michael Taylor has also been talking recently about how the fighting stance for the Republic – with the larger gladius Hispaniensis – differs modestly from the fighting stance in the imperial period with shorter imperial-era gladii, mostly in terms of the angle of the sword blade.
14. op. cit.