This week we’re going to start a four-part look at the role of the pre-modern general or army commander, particularly in the context of a pitched battle. This is of course a vast topic, but we are going to focus not on tactical or strategic questions but on a lot of the nuts and bolts constraints which condition those questions: where is the general, what can he know, what can he see, who can he communicate with and to what degree can they follow his commands effectively? After all, a plan which is perfect save that it cannot be communicated or executed isn’t perfect at all.
Pre-modern generals – and I am going to use this word in a very broad sense to mean the overall commander of an army, even in societies where that figure is a consul or king or what have you – figure very prominently in media. While modern warfare scenes in film are often shot from the perspective of regular soldiers, pre-modern battles are almost always shown at least partially from the perspective of the commanders: Théoden, Jon Snow, Alexander, Darius, William Wallace, Robert the Bruce, Henry V and so on. And when I teach I sometimes use these scenes to point out some of the absurdities: the commonplace of the general signalling with hands or a shout to an army obviously too large to see or hear him, for instance.
Generals are also the primary frame for pre-modern pitched battles in games, at least on any scale larger than a small skirmish. Of course there are games that are entirely about strategy or tactics, but even in fantasy or pre-modern RPGs it is most common for the player’s character to be put in command of the Big Battle than to merely be a foot soldier in it (or more common yet, to be a mere foot soldier in the Battle We Lost in the opening but in command by the end when it is time for the Decisive Battle To Decide the Fate of the World). As you may have gathered from the title, we’re going to use the Total War series of games predominantly to represent the common cultural model of generalship against which we can compare the actual practice of generalship.
At the same time, as we’re going to see, these depictions – which for story reasons place strong emphasis on the control and agency the general supposedly has – for the most part fail to represent much of what a general can and cannot do in a pre-modern pitched battle, primarily by failing to take into account the physical and cultural constraints on the general’s actions. So we’re going to talk about those constraints and how they shape the job of generalship in a pre-modern army.
A note on scope before we begin: I am going to restrict this look to the broader Mediterranean in the pre-modern period. That means the Mediterranean littoral, plus the Middle East and Europe; many of these rules will apply elsewhere, but I am going to stick to where I know the sources. It also means the ‘pre-modern,’ by which I mean the period stretching from the end of pre-history (which varies place to place with the advent of written sources) to the beginning of the early modern period in c. 1450 or so. That distinction matters because technological changes after 1450 begin to alter some of these constraints, although generally slower than you might think.
And 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.
Fundamentally the tactical portion1 of Total War games (and indeed many real time strategy – RTS – games) is oriented around the concept of the ‘match-up.‘ What I mean by this is that the main tactical puzzle the player-as-general is trying to solve is to achieve a series of favorable match-ups: tie up enemy cavalry with spearmen that perform better against cavalry while engaging enemy spearmen with swordsmen that perform better against infantry and so on. This is a simplification of course: these games often have special abilities, charge impacts and other considerations that complicate the basic match-up based puzzle; but the base of the puzzle is the match up (particularly in PVP matches where player-generals lack the easily abusable predictability of a computer opponent).
That shapes the way a Total War battle shakes out. To run through it very briefly, in the pre-battle phase, the player places their units in a formation, though I use this term loosely as high-level play tends to revolve around very loose ‘checkerboard’ formations with very large intervals between units. In practice the army doesn’t maneuver as a formation much in Total War games, but instead each unit (which tend to be around company-sized; c. 100 troops or so) maneuvers independently of all of the rest.2 The match then starts and one or both armies will close distance and what the player is looking to do is to make sure that their units end up in those favorable match-ups where their inherent statistics will enable them to do more damage (measured in unit value) than they take.
What exactly makes a favorable match-up varies. Many units in these games have specific attack and defense bonuses which only work when fighting specific kinds of other units (e.g. ‘bonus against infantry’ or ‘bonus against large/cavalry’). Units have armor values and armor-penetration values, so a high armor unit will ‘trade well’ against an enemy with low armor penetration because they’ll take very little damage. Slow units are vulnerable to missiles, but some units have shields which negate a percentage of missile damage and thus ‘soak up’ missile fire, making it a waste to shoot at them if any other targets are available. The term for a unit which is particularly effective against a particular kind of unit is a ‘counter’ (as in, ‘spearmen counter cavalry’); if the match-up is very lopsided so that no conditions could really shift its balance, it is a ‘hard-counter.’ Of course, there are instances where the way to a favorable match-up may be to double-up on a single enemy unit and some units benefit greatly from charging into enemies, encouraging ‘hammer-and-anvil’ tactics, but it is really striking how often these match-ups exist effectively in isolation, a point we’ll come back to later.
But for now what I want to focus on is what that makes the player’s task of generalship about. Typically in a Total War game, once the battle begins, the player has nearly perfect information about an enemy formation. Terrain can create exceptions, but like most strategy games the Total War tactical information environment is binary: either you have no idea what enemy units are in an area, or you can see exactly what units are present: how many of them, their formation, what type of units they are, their position, facing and motion, their exact combat stats and often even how much damage they’ve sustained or even how tired they are instantly at a glance.
At the same time, players can issue immediate orders to their units, allowing rapid responses to new information. While units may still take time to move, they can be gotten moving almost instantly, so a player that sees the possibility of a favorable match-up (or the danger of an unfavorable one) can immediately begin shifting units around within the army to try to get the advantage. Consequently, the player’s main task is to quickly ascertain the enemy units position and type and then maneuver their own units as they (or the enemy or both) advance so that the match-ups (and to a lesser extent, positioning) that result are favorable.
And what is striking to me (and thus forms the basis of this series) is that this main task has functionally nothing to do with how ancient generalship actually functioned. This basic task of the most popular historical real-time strategy game series almost3 never shows up in the sources. Indeed for reasons that, as we proceed, will become clearer and clearer, it was impossible for most generals in most battles in the ancient world to do this once the battle had begun. It was more possible, as an aside, in early modern warfare but that is a period these games rarely simulate and indeed haven’t really returned to in earnest for more than a decade.
Instead what we’re going to see is that command in the ancient world (and indeed, in any environment) existed under a set of fundamental constraints which historical strategy games generally remove for ‘playability’ reasons. And I should be clear here: I understand why in many cases developers do this. The goal of these games is to be fun and it is seemingly axiomatic that ‘fun’ results from player empowerment and as we’ll see all of these constraints are profoundly disempowering in different ways. But the removal of those constraints tends to mean that even with historically inspired units and armies, the actual tactics and battles in these games don’t much resemble their historical inspirations because skilled players rely heavily on tactics and maneuvers which the historical inspirations for these armies weren’t capable of.
The same is true in different ways for film depictions of generalship. There are exceptions, of course, but in most films where an ancient or medieval general is prominent, the film-maker is either developing or revealing that character through a battle sequence, which in turn generally requires perceivable cause and effect between the general’s actions and the battle’s outcome. Moreover the need to build tension typically means that crucial decisions are delayed to create that moment where the heroic leader shouts ‘charge!’ or ‘fire!’ at just the right moment to win the day. Rarely does one get a battle scene in which all of the decisions of import were made at a council of war in the early, pre-dawn hours with the battle then left to simply play out according to plan and largely out of control of the participants.4 Instead characters in command make decisions on the fly which often could not have been done without planning.
We are going to look at those constraints in three parts. This week, we’ll look at the information environment that an ancient or medieval general had access to: what could they know and how could they know it? Then next week we’ll look at the ability of those same commanders to use that information to coordinate their army. Then in the third part, we’ll look at the ability of ancient armies to act on those commands effectively and quickly, as well as suggesting an overall battle model that might be more akin to how an actual battle was commanded. Finally, in a fourth bonus part, I want to look at a number of the other tasks generals were expected to do which are typically not featured (or at least not mechanically simulated) in these sorts of games much at all.
One of the fundamental structuring assumptions of the ‘match-up’ model of generalship is the ability of the general to adequately observe the enemy force in order to identify both its component units and their positions. In a Total War game, you an generally see enemy units quite easily and if you mouse over them a helpful tooltip will inform you exactly what kind of unit they are, often down to their precise statistics. Of course real battlefields do not come with zooming cameras or tooltips. So what kind of information might an ancient or medieval army commander have on the day of the battle?
Battles, of course, do not spring up out of nowhere but are generally the result of a campaign. Typically the attacking army’s goal was to ‘deliver a siege‘ to a key enemy population center as a way of gaining control over the surrounding territory (and possibly convincing other, smaller population centers, even fortified ones, to surrender preemptively). There are exceptions, of course; ancient Greek hoplite armies had such limited siege capabilities that they often didn’t attempt to deliver the siege at all but instead threatened agricultural disruption in order to force a battle. In any event, when both armies arrive at the battle it is because they have been maneuvering for some time, during which one might reasonably expect both commanders to be trying to get some sense of the other army’s strength, position and intentions.
Generally the commanders are going to have to rely on a broad category of intelligence we might collectively term ‘reports.’ Some of those reports might be from detachments (generally fast moving cavalry) from their own army on intentional scouting missions, but just as often the ‘reports’ were generated by civilians. Traders moving along the roads might be asked for information or rumors; friendly civilians in enemy territory (frequently of a political faction currently out of power that might hope for a change in their fortunes) might also send word of an army’s movement, its size or intentions. And of course as elements of the army passed through smaller population centers, they could collect whatever rumors were available. Of course one’s own scouts were the most reliable, but in order to be able to coordinate their movements with the army and send messages to it, scouts tended not to stray too far from the army, so information on the situation in regions where the army wasn’t often could only be had through less reliable channels.
Games struggle to reflect this sort of information environment. Often games present a ‘fog of war’ of sorts – areas where the player can see the terrain, but not the movement of enemy units – but the real fog of war in this kind of warfare was a lot more complex. First of all, the general didn’t know where it was. A lack of reports from a given region might mean it was quiet, but it might also mean that, for whatever reason, the general is blind to events there. Moreover, in most games what information the player does have is invariably accurate, but the very nature of reports is that they are very frequently wrong. Indeed, intentionally seeding false reports was a standard stratagem.5 The general in campaign is thus not looking at an incomplete but secure picture of the war on a map, but rather trying to make sense of a cloud of verbal or written reports (delivered as words, with all of their shades of meaning and difficulties of interpretation) many of which are inaccurate and some of which are blatantly false.
Reports of these sort also time time to travel. As I remind my students, from antiquity to the early 1800s, you are dealing with societies where the most sophisticated communications technology generally available was ‘man on horse.’ Now a messenger rider that is regularly changing horses at pre-appointed stations can move very fast indeed; in excess of a hundred miles a day. But reports are only going to move at that kind of speed when carried by official messengers on pre-set routes; generals in the field will rarely have use of that kind of message system. Instead, reports are going to move at the speed of merchants or the speed of rumor, which is going to be much slower, far closer to the movement speed of the enemy army. This leads to situations where generals often know what general region an enemy force is in and they may have a sense of its ultimate objective (because the report they have was that the army was in such-and-such town moving down such-and-such road ten days ago) but often not its exact position at any given time. At the same time, again, no reports doesn’t mean there’s no army somewhere; it just means there are no reports! Unlike in most strategy games, a pre-modern commander doesn’t really know where the fog of war is or what might be hiding in it.
For instance, one not-uncommon occurrence in ancient battles was the appearance of some detachment of an army being interpreted incorrectly as an entire second army appearing. For instance at the Battle of Delium (424 BC), the Athenian right-wing, having defeated the Boeotian forces opposite it, panicked at the sight of a relatively modest detachment of Boeotian cavalry sent by Pagondas (the Boeotian general) in a desperate effort to shore up his collapsing left wing; the Athenians thought this was an entire fresh army (Thuc. 4.96.4) and routed. Likewise, at the Battle of Tifernum (297) between the Romans and the Samnites (Liv. 10.14), the Roman consul, Fabius Maximus Rullianus panicked the Samnite army by having a small detachment of his infantry march around the battlefield ahead of time and reveal themselves only after the battle was joined, appearing like a freshly arriving army. What Fabius was counting on here is that there was another Roman army in the field (under the command of the other consul, Publius Decius Mus) but it was too far away to assist in the battle; but of course the enemy (and indeed, Fabius’ own soldiers) didn’t know that so the sudden appearance of the ‘advance guard’ of Decius’ army might inspire panic in the enemy and resolve in the Romans, which it did.
The other side of this problem is the ‘encounter battle’ or ‘meeting engagement,’ where two armies unaware that they are in close proximity run into each other. One such example, famously, was the Battle of Cynocephalae (197) between a Macedonian army under Philip V and a Roman army commanded by Titus Quinctius Flamininus. Neither Philip nor Flamininus knew exactly where the enemy army was, but both were working on reports about their general location (Plb. 18.18-21 which is also a great example of two armies attempting to maneuver operationally against each other for advantage while relying on reports) and so knew the enemy was fairly close.6 Thus both generals dispatched scouting forces to seize a large hill (called “the Dog’s Heads” or Cynocephalae) for observation and as covering forces; the scouting forces collided and when both commanders attempted to reinforce them (based on the reports those engaged scouts sent back) it precipitated a general engagement, which the Romans won. Of course it is striking that this is how things go when careful scouting is done; the alternative were the relatively rare (far rarer in history than in Total War games) true ambush battles like Lake Trasimene (217).
I can think of few games that model this kind of information environment. Perhaps the closest I’m aware of are Radio Commander where you have to rely on the spoken reports of forces in the field and HighFleet which forces the player to rely on a mix of signals intelligence (SIGINT) from intercepted transmissions to both radar and systems to detect enemy radar, systems which often provide incomplete or confusing information (“there’s something over there, but what is it and how many?”). That said, this information is generally correct, merely incomplete and in any case is instant whereas the information available to a pre-modern commander was often simply wrong or days or even weeks old by the time it was received.
All that said, one thing various reports might well inform a general is of the relative size and composition of his enemy, at least in the most general terms. That was a sort of information which was, at the degree of accuracy we are dealing with, less time sensitive. The mustering of a large army, after all, was a hard thing to fully conceal, as was its movement. Observers might well note specific kinds of soldiers and very roughly their numbers. This sort of information was never so precise as it tends to be in games, where the player often has an exact enemy order of battle at time of contact. But it is clear that generals often knew both when they substantially outnumbered their foe and when the reverse was true and could act accordingly. Especially once armies drew close, any capable scout could size up the camp and the number of campfires (although faking these signs was a standard stratagem too).
Moreover, while the precise location of an enemy army might be a mystery, its general location was often known and as a result, true encounter battles in the pre-modern world were the exception rather than the rule. Armies had the move have to forage for supplies, meaning that the presence of an enemy army in the neighborhood is news that is likely to travel fast, so two armies are unlikely to be very close to each other without knowing it.7 And as noted, armies tended to move with scouts, creating a ‘bubble’ of more accurate information which made accidentally running into an enemy army unusual. Unlike in modern warfare where just about any enemy that can see you can shoot you, we are dealing with armies whose weapons have effective ranges far shorter than human visual range, so in the absence of some sort of obstruction (like the hill at Cynocephelae) scouts are likely to observe each other long before they crash into each other and thus be free to report back on the presence of the enemy.
At this point the question is offering and accepting battle; I want to note that I am going to use the Battle of Pydna (168) as an example a few times here so I want to recommend for a second time the recent and excellent reconstruction of the battle published by Paul Johstono and Michael J. Taylor in GRBS. This is, to modern thinking, such an odd affair that it deserves some explaining; it is often missed even by modern scholars but is a commonplace of pre-modern battle and campaign narratives or mistaken for ritualism when it is just good sense. Armies, of course, do not march in fighting formation; to fight an army generally wants to form in a wide line, but roads are narrow, after all. No general relishes the idea of throwing soldiers straight into battle exhausted after a whole day of marching either, so the armies are likely to encamp near each other with the plan to offer battle on the following day. You may reasonably ask why one army doesn’t skip this step and roll straight in; this happens but generally rarely. Generals tend to encamp their armies in either fortified camps, on solid defensive positions or, where possible, both. So the attacker, exhausted from marching, would have to open the battle with an assault on well-defended position; hardly an appealing option in most cases.
So frequently final preparations for a battle were made with both armies encamped fairly close to each other, often separated by just a few miles of open field. Even if no battle occurred immediately the following day, armies were not idle: scouts and pickets would be set out to give advance warning of enemy movements or an attack on the camp, while foraging parties both food and water might also be dispatched (more on these a bit later). For logistics reasons (again, a bit later) both armies encamped this way are on a timer before their local supplies run out and they must move, so they cannot sit this way indefinitely (which both generals know about the other). At some point they must either fight or withdraw.
Thus the other thing an army in this situation might do: offer battle. One (or both) armies might draw up their fully array, in battle formation in front of their camp, essentially signalling “I’m up for a fight if you are.” Now of course generals are likely to form up on favorable terrain (which might mean the heights or areas where the width of the battlefield is favorable) and don’t like to fight on terrain favorable to the enemy, so you might end up in a situation where, over multiple days, each army forms up on terrain that is good for them and then refuses to fight on the other army’s ground. This happens for instance at the Battle of Mantinea (418) between Sparta (and friends) and Argos (and friends); the Argives formed up on a solid position and while the Spartans initially prepared to attack the king (Agis II) decided at the last minute not to and withdrew, refusing battle; the Spartans subsequently lured the Argives into less favorable terrain by damaging the irrigation system, forcing a battle on their terms, which the Spartans won (Thuc. 5.64ff). Likewise, at Pydna (168), both Roman and Macedonian armies meet on a narrow plain near Pydna, but the Romans – fresh from a long march and probably preferring a wider battlefield – refuse battle and withdraw, leading to fairly wild battle that occurs the following day. At Philippi (42), the two armies (Octavian and Antony leading the Caesarian force on one side and Brutus and Cassius the Liberatores on the other) encamped opposite each other for a number of days (we are not privy to the exact number) while Antony offered and indeed tried to force a battle. He got his battle on October 3, but it was indecisive, leading to another period of waiting, offers of battle and refusal culminating in a second engagement on October 23, by which point the armies had probably been camped within a few miles of each other for at least a month.
Offering battle gives the general in command some options in terms of time. He can, for instance, choose what time of day to offer battle, with the classic stratagem here being that a general who thought a battle risky might only offer battle late in the day, so that if the battle went poorly the enemy pursuit would quickly be ended by nightfall. Of course the reverse is just as true: a commander who is very confident and hoping for a decisive victory might prefer to commence the battle at dawn. He can also try to choose the ground; if cautious a general can form up on the presumably favorable ground near his camp. By contrast, offering battle in the low ground between the two camps8 was a sign of boldness and confidence. There might thus be a kind of negotiation of a sort; no one is passing notes back and forth, but each general can gauge the other’s intentions from how their army forms up. And of course a general who sees his opponent forming up on favorable ground near their camp, but who is supremely confident of his army (or rushed by a lack of supplies) might attack anyway, accepting battle on his opponent’s terms.
But to understand the information a general is working with when he makes the decision to offer battle, we have to think about what the process of actually forming an army into its battle array involves. In strategy games, this is generally an easy process: you click and units move exactly where you clicked. There is no confusion, no one gets lost, everyone understands your orders. But that isn’t how directing real groups of humans works in the real world. One exercise I do with my warfare classes is to form up the front row of students to make up a single half-file of a Macedonian sarisa-phalanx; without fail one needs to budget several minutes for this fairly basic task of “form a line one arm’s length apart facing this way.” The issue gets even more complex, of course, as the general is dealing with groups so large they cannot all hear his orders at once and in many smaller units each of which needs to know where it is going. In short, forming into battle array cannot be done ‘on the fly,’ this is a fairly complex, difficult task which takes time and planning.
So how was it done?
Generally plans for a battle, if any planning was to be done (we’ll talk about how armies responded when detailed planning wasn’t possible in the third part of this series) were drawn up the evening (sometimes in the very early morning) before in a ‘council of war’ where the commanding general met with his subordinates and advisors. Livy notes just such a council on the eve of the Battle of Pydna, for instance (Liv. 44.36.7-14) and it is very clear that such meetings were standard practice; the Latin term for all of the various people who ought to be included in such a council was the general’s consilium. Such a council provided an opportunity for the general to work out the details of exactly how they wanted to conduct the battle the following day so that all of their subordinate officers knew what to do. My own sense of descriptions of these councils in the ancient sources is that they were more often consensus-building tools rather than deliberative ones: the goal was to get everyone on the same page with the general’s original plan, not to submit the plan to committee.
The army then has to be gotten into formation. Since it is exiting a camp (which is going to have a narrow, road-width entrance) this entails reforming from a marching column leaving the camp into a battle line. The expedient to make this easier was to have a fighting depth (the number of men in a file) either equal to or in multiples of your marching width, so that a unit could exit the camp in column, turn either right or left, march to the appointed place and then everyone could simply right-face or left-face (that is, a 90-degree turn) and thus be in fighting formation. The relative position of units then, and thus the organization of the overall line of battle, would be determined by the order that units exited the camp. So at that council of war the evening or morning before the battle, the formation can be decided on and the order of march out of the camp into that formation set. All of that is going to make getting a huge body of thousands of men out of their tents, into their equipment and then into the right positions in the battle array possible in a reasonable amount of time.
But notice what the general does not know in that moment. He does not know if the enemy intends to accept his offer to battle, of course. He does know where their camp is, so he knows in what direction they will approach from, but he does not know what formation they will adopt, nor do they know what formation he will adopt. Because forming up for battle takes a while, it was generally impossible to wait to see what the enemy’s formation was before adopting your own: you’d be very vulnerable indeed if the enemy, fully formed, simply advanced on you while you were only beginning to form up (there having been a delay to plan your marching order and thus formation). So the basic question of the overall formation has to be made effectively blind to the enemy’s formation. The general might know their numbers, direction and even some of their composition (things like “the enemy has brought many war elephants” or “their army is much larger than mine”) but probably not their formation.
And this goes a long way to explaining why armies in the ancient or medieval world tend to have ‘standard’ battle formations. They cannot respond to specific enemy dispositions in advance anyway, so armies tend to form up in predictable ‘safe’ patterns. Typically that means the heaviest, least mobile forces (generally heavy infantry) are deployed in the center, with progressively lighter and thus more mobile units on their wings, where faster maneuver might be necessary. The planning at the council of war could take terrain into account; it was very common for armies to seek to ‘anchor’ one or both flanks on terrain unsuitable for fighting (rivers, marshes, hills, dense forest) in order to create a controllable battle-space or to deploy light infantry in difficult terrain where heavier infantry or cavalry might struggle.
Of course a general might well guess at this point how an enemy will form up over the terrain and seek to prepare for that, although they are only guessing. At the Battle of Magnesia (190), the Roman commander, L. Cornelius Scipio (the brother of Scipio Africanus) gambled that a constraining river on his left would mean that his opponent, Antiochus III (king of the Seleucid Empire) wouldn’t launch a heavy attack here (Liv. 37.39; App. Syr. 31). Consequently, he left only a small force on that side in favor of firming up his center and right. Scipio, by no means a poor commander, was in the event wrong in his gamble: Antiochus did end up placing his elite cavalry (the agema, the king’s cavalry bodyguard) on that side and broke through the thin forces Scipio had placed there, though the battle was salvaged by the Romans and their allies9 being victorious on the other side of the field and Antiochus’ inability to keep his victorious right from trying (and failing) to raid the Roman camp.10
Sometimes those formations were culturally loaded too; Greek hoplite armies, which were almost always composed of a number of allied states fighting together, generally deployed their hoplites in a single line (of variable depth, generally around 8), but position in the line signaled the importance and honor of each alliance member: the far-right was the position of highest honor, the far left the next-most important position, and then each other position was essentially ranked from right to left. Consequently the formation of hoplites might be as much about diplomatic ‘alliance maintenance’ and civic pride as it was about tactical expediency, which was part of by the Theban general Epaminondas’ decision to load up all of his elite Theban troops on the left wing at the Battle of Leuctra (371) was, in the Greek context, so innovative (this resulted in the Thebans crashing directly into the main Spartan contingent, which was Epaminondas’ goal).
Once the two armies have begun forming up, our general is going to be getting information a bit more directly. In an environment where the very longest ranged battlefield weapons had maximum ranges of a few hundred meters and effective ranges under 100m, armies can form up quite close to each other, enabling the general to potentially see quite a lot without worry that his position is immediately dangerous. And most pre-modern generals will have one key visibility advantage at this juncture: they’re on a horse (or a chariot, or an elephant), giving them an elevated observation position that allows them to see over the large bodies of infantry moving around. On the other hand, the general’s ability to see is restricted to where he is; he can move around but not infinitely fast (as we’ll see in the next part, that’s also an issue for giving orders).
Now I am sure some readers are eager to note that a number of games have taken this particular limitation into account! Some Total War games had difficulty settings which restricted the camera to be within a certain distance of either your general or friendly units. Take Command – 2nd Manassas (2006), a game we’ll have cause to revisit more than once, likewise restricted the camera (based on rank). And the Mount & Blade series (we’ll focus here and subsequently on the latest entry, M&B 2 Bannerlord) does this aggressively as well. But even these examples fudge the issue in two important respects: height and distance. Because our general doesn’t have an aerial camera or a zoom feature: he is working with the Mark One Eyeball (henceforth abbreviated as Mk1 Eyeball).
Height is easy to talk about in this context. As noted a general is generally going to have a modest visibility advantage by virtue of being on a horse, but games in third person generally place the camera even higher. Mount & Blade‘s camera generally hovers a few feet above and behind the player’s character, for instance, while Take Command and Total War both position the camera much higher even if its position remains restricted. That matters a lot of course both because obstructions (trees, fences, hills) tend to occur close to the ground and also because the higher up a point of view is, the longer its horizon becomes. But the Mk1 Eyeball is, frustratingly, stuck at eye-level.
More complex but also important is distance. In Mount & Blade, a very large royal army might have a few thousand soldiers in it, of whom only perhaps 500 or so (depending on how powerful your computer is) are in the battlefield at once. Total War battles are larger, but a ‘full stack’ (a maximum size army) is typically 20 units of around company size (c. 100 soldiers) each; depending on faction and game that tends to come to a bit less than 2,000 soldiers once units with far fewer troops-per-unit (cavalry, artillery, etc.) are accounted for. As a rule, recent Total War games have tended to cap the number of armies that can engage at once in a battle to two full stacks, so 40 units and typically quite a bit less than 4,000 soldiers (and indeed, large armies like this are so cumbersome and difficult to control that it is no wonder that Creative Assembly, the developers, haven’t tried to use improved computer power to make the armies any bigger).
But armies in large pitched battles are often much larger. The very largest pre-modern armies in the broader Mediterranean tend to be in the 80,000-100,000 range (Cannae, Magnesia, Philippi) though these are rare. Nevertheless, armies in the 10-40,000 range were relatively common for large forces expecting a major pitched battle (note that armies in the middle ages tend to be on the smaller end, whereas armies in antiquity will tend to be on the larger end). The standard Roman consular army in the Middle Republic was a pair of legions (4,800 men each) matched by a pair of socii alae, for a total army size just short of 20,000; this was effectively the floor for a major Roman army, since non-Italian allies or more legions might well be added to this if necessary. The total Roman force at Pydna was fairly typical at around 35,000 all told, facing perhaps 30-40,000 soldiers, both Macedonians and their allies. Medieval armies were often somewhat smaller; the English at Agincourt only had around 8,000 troops (to the French’s perhaps 15,000), though of course the size of a cross-channel expeditionary force was naturally constrained.
Now I should note here that of course there were a lot of engagements which were much smaller than this. In the ancient world there were still skirmishes between scouts and foraging parties which would involve only small detachments of these large armies. And in the Middle Ages, a lot of the endemic warfare was between relatively small forces in wars between more minor nobles. Of course, apart from Mount and Blade, that kind of lower-scale endemic warfare is something games generally do not simulate at all (and films generally only feature by accident when they fail to put enough people on screen for what are supposed to be ‘big’ battles). So we are going to remain focused on large-scale pitched battles here.
And the thing is, an army of 15,000 or 35,000 is a lot bigger than an army of 1,500. And I don’t just mean in terms of its numbers; it is physically bigger – it occupies more space. And here I can use Johstono and Taylor’s Pydna reconstruction (linked above) to make the point, aided by google streetview, because they’ve calculated and figured the likely positions of the various elements of the battle. Now of course they present the fight in a nice, neat map overlaid on the modern topography using Google Maps; these maps are lovely but of course they are the battle as seen by aliens from space not as seen from the ground. But thanks to the wonder of Google Street View, we can get a vague sense of what you might have been able to see from where, aided by the fact that, being mounted atop cars, the street view camera isn’t far from the height of a man on a horse.
Some brief background on the battle and the relative positions. The Macedonian army, under the command of their king, Perseus11 had occupied a blocking position on the road leading up to Pydna and through it into Macedonia proper. The Romans had arrived the previous day but their general, Lucius Aemilius Paullus, had refused an engagement on ground so favorable to the Macedonians (who had a sweet, narrow little position at the northern edge of the field) and had instead gone into camp a few miles to the south. Thus, after the two armies had encamped opposite each other, neither had accepted the offer of battle on the other’s terrain. However, they both had the same water source, a stream on the eastern edge of the battlefield. Thus, the day after the Romans arrived, parties from both armies going to get water ended up in a small scrap which, as both armies rushed to reinforce them, precipitated a general engagement.12 Thus each element of both armies came into action one by one, beginning on in the East and moving West as the two armies marshaled out of camp (generally very close, as Plutarch notes, to the Roman camp, in part because of where the skirmish started, but also because the Romans were initially pushed back). The Roman army was fundamentally in four roughly equal-sized major units (from left to right or west to east): the left wing (Sinistra Ala) allies, then the second legion (Legio II) then the first legion (Legio I, where Aemilius, the Roman commander was), then the right wing (Dextera Ala) of allies.
Plutarch (Livy’s account of this battle has a large gap that sadly covers most of the action) has a passage in this context about the fear that Aemilius experienced when he saw the Macedonian heavy infantry, the chalkaspides (literally ‘bronze shields’ – part of the Macedonian sarisa-phalanx which wielded the 21ft sarisa pike along with a shield held by a strap grip; the shields were not entirely bronze but faced in it, in contrast to the Leukaspides, presumably less elite troops whose shields were probably faced with hide, painted white) – advancing (Plut. Aem. 19.1-3). To get there, Aemilius will have led Legio I out of his camp, fortunately (for us) advancing directly over a road with street view, so we can get a sense of what he’s seeing:
The blue line there just coming over the opposite hill crest some 550m away is about where Aemilius’ screening force initially had been and where the chalkaspides were advancing rapidly; it is striking that the thing Aemilius can apparently see is the bronze-faced shields of the chalkaspides which makes a lot of sense: at this distance he can hardly have seen much else, but he will have known when the men in the phalanx readied their shields and lowered their sarisae because of the gleam of the shields (which is, by the by, exactly what Plutarch describes him as being able to see: that the Macedonians had readied their shields). Nevertheless it is only as his legion is beginning the attack that Aemilius is fully aware of what unit is directly in front of him and that it is advancing in good order (his subordinate, Scipio Nasica, forward with the screening force, seems to have had a better view, Plut. Aem. 18.5-8, and may well have rushed back to tell him – those reports again). It is, by this point, probably too late to send Legio I anywhere else; Aemilius seems to have concluded that on his right a detachment of allies Paeligni and Marrucini could keep his flank secure (he was right) and so he best plow forward into the oncoming chalkaspides head on.
But that is just Legio I, which, by the by, at 4,800 men will have filled the entire width of that shot above as they advanced (imagine them about where the one brown tree is halfway down on the right, but extending all the way from the left to the right and a little off the picture on both sides). What about his right wing? Well, it’s just on the other side of that building (which obviously wouldn’t have been there in 168)!
Oh, you can’t see the building? Let me zoom in. It’s on the other side of THIS building.
The one that is simple a white spec in the original image. That would be the closest flank of the dextera ala, which would probably extend another 500m or so beyond that. Of course Aemilius’ Mk1 Eyeball did not have a zoom feature; a battlefield-useful telescope was at least 1500 years away. At this distance, Aemilius might have actually been able to make out the war elephants deployed on that flank (Liv. 44.41.4), but not much else save that his far right flank was still there, if he could see the glint of their helmets over the crest of the hill. What about his left flank? Well, here there’s a problem: there’s a hill in the way. Not a very big hill but hills don’t need to be very big to be taller than a man on horseback at their foot. Still, if we jog about 400m down the road we can get a decent impression of the left flank’s view from the ground:
The circled building there is, in Johstono and Taylor’s reconstruction where the right edge of the Sinistra Ala and the left edge of Legio II will have come together. Where this picture is, we’re probably 200m or so closer than the center of Legio I would have been, but perhaps Aemilius could have galloped over here to get a quick view of how L. Postumius Albinus, in charge of his second legion, was managing. He can hardly have seen much except that they were still there and presumably holding their ground and even less of the left wing of his army – the closest edge of which was still 800m away from the far leftward edge of Legio I. So these images are essentially the views that Aemilius could get, assuming he had nothing else to do except gallop from one vantage to another trying to figure out what was going on (in practice we are told that instead he seems to have stayed with his Legio I and rallied them in the advance; Plut. Aem. 19.3. Doubtless this was a better use of his limited time).
Because here’s the thing: a Total War army (single, full-stack) is going to occupy a frontage in a game of usually around a few hundred meters. Whereas, as reconstructed by Johstono and Taylor, the Battle of Pydna – by no means the largest battle the Romans ever fought – takes place on a field perhaps 2.5 kilometers wide (a bit wider than it might otherwise have been because of the chaotic situation and the gaps). Not every battlefield was so sprawling, but even small battles might be quite wide. The Athenian/Plataean army at Marathon was just 10,000 hoplites, deployed eight deep except for 2 of the 10 Athenian tribes in the center which were just four deep, which with a little math suggests a front rank roughly13 1,475 hoplites wide. The spacing in hoplite warfare varied, from synaspismos (tight formation, literally ‘shields together’) probably giving each man about 1m of space (and overlapping the shields) to a looser spacing perhaps twice as much.14 So that combined Athenian-Plataean army, despite being only 10,000 hoplites, might have been anywhere from 1.5 to 3km wide, likely to the narrower side given the relatively small size of the plain in most places.
There are some places on Earth flat enough to allow a man on horseback to see the c. 3.8 miles to the horizon, but even in those few, exceptionally flat places, formations of men more than a mile distant are going to be barely visible blobs moving over the landscape. Enemy formations are going to be harder to see and further away. Video games don’t work on this scale and frankly film rarely does either; when a large army is shown on film, they are generally arrayed in such a way to allow the camera to capture all of them, but real armies do not form up with the aim of easy observation by the enemy.
It is striking, as an aside, that these games tends to simulate the command of an entire army by giving the player the task of commanding forces that are generally around the size of a standard maneuver unit15 like Roman cohort of the Late Republic (c. 480 men and about the max number of deployed soldiers on default settings in Mount and Blade 2) or a American Civil War brigade (c. 1000-1500 and about the size of an army in a Total War game). Actual armies, of course, would be composed of many of these maneuver units; a Roman legion is ten cohorts and a Roman army, as noted, consists of multiple legions.
Instead with actual army sizes, a general is limited in his ability to observe the battle mostly to the rough positions of the component parts of his army and the enemy’s army; he may not know where specific units are unless they are very distinctive because most of them are at distances where he might struggle to make out specific equipment (uniforms and battle standards might clear this up a bit, but by the time we’re getting armies in uniform ‘facing colors‘ we’re well into the early modern period). Good observation positions (ridges, hills) might enable longer sight-lines, but they’re not going to enable more detail because again, all anyone has is the Mk. 1 Eyeball. This is why the general with his headquarters on the hill behind the battlefield is much more of an early modern or modern thing: telescopes and binoculars enabled generals to observe much more of the battlefield if they could get a good vantage.16
Seeing the Battle
And all of this is before the general confusion of battle makes a mess of things. Without standard uniforms (and often even with them) even from a moderate distance making out one block of troops from the next is difficult; friends mistaken for foes and vice versa even within musket range is hardly an uncommon occurrence, for instance. More broadly, depending on the terrain, thousands of feet moving at once can kick up a lot of dust, further obscuring the field.
The upshot of all of this is that at each stage of the fight – the march to the battle, the night before, the forming of the battle array and then the battle itself – the information environment the general is forced to operate in is heavily constrained. In particular it is constrained in ways that make a lot of Total War generalship (and a lot of Hollywood generalship) at best deeply uncertain and more often effectively impossible.
At best if a general wants to ensure specific ‘match-ups,’ he is going to have to largely guess where the enemy might deploy those forces and arrange his both battle array – well before he can know the enemy’s disposition – to try to get those match-ups. Doing this was possible (as noted Epaminondas does this at Leuktra; Hannibal’s plan at Cannae likewise required guessing the Roman plan in advance), but tricky guess-work rather than something that could be improvised on the fly.
Think back, for instance, to the example from Pydna. By the time Aemilius can actually know for certain the formation his opponents had adopted (which was, in any event, somewhat unusual in that the presumably more elite chalkaspides were on the center-left of Perseus’ line rather than the center-right; it seems safe to assume this was a product of the rush to get the army out of camp and into formation), it is already too late for him to make any meaningful changes to his own formation. He is already committed to fight with the plan he has. That doesn’t mean, as we’ll see next time, that a general in this position couldn’t respond to sudden changes – Aemilius was at least aware that the Paeligni to his right were in trouble (Plut. Aem. 20.5-6), though he seems to gamble that they will hold; according to Plutarch he does respond to seeing the organization of the chalkaspides begin to break up by communicating to his officers that the component maniples of his legion should advance independently (so some giving ground, others pushing, to the degree they were successful, rather than presumably trying to keep one even and coherent line), aggravating the gaps into the Macedonian phalanx (Plut. Aem. 20.8-9).
As a result, while the general was certainly expected to think about and tweak the positioning of his forces, most ancient and medieval cultures had fairly standard battle arrays and formations which they essentially reused. It is by now more or less a commonplace to note, for instance, that Alexander’s dispositions at Granicus, Issus and Gaugamela essentially replay the same battle three times (refuse the left, phalanx anchors the center, Alexander leads the cavalry as the key striking force, breaking through on the enemy’s center-right rather than fully flanking them), exploiting the same weaknesses in the Persian armies that opposed him. The Battle of the Hydaspes is more complicated because of the difficult river crossing, but after that it the battle plan mostly another ‘Alexander-battle’ like the rest. The same is of course mostly true of hoplite battles or the Roman manipular legion: generals could and did innovate, but there was a ‘standard’ fairly ‘safe’ battle pattern for these armies and they more often stuck to it than not.
Instead the general’s role was if anything more complex and difficult: he had to manage an uncertain information environment, attempting to ascertain how his enemy intended to fight from incomplete and occasionally outright false reports and his own knowledge of how the enemy tended to fight battles. Approach marches and camp sites – ironically something Total War largely does not let you choose – mattered a lot more because they determined the terrain of the battle. But rarely did generals have the information necessary to make micro-adjustments to match known enemy dispositions (at least not before they might have become aware that a flank was crumbling, etc). Instead, they were forced to make decisions based on incomplete information and assumptions; in effect, they had to guess in advance.
Which is fine, because it was also fairly rare that commanders of pre-modern armies could exercise the kind of fine control of an army required to achieve perfect ‘match-ups’ even if they did know exactly where the enemy was. That isn’t to say the general necessarily had no role once the fight was started. Next time we’ll take a look at command: what is the general’s job in battle and how does it vary culture to culture?
- That is to say we are removing for the moment the strategic game including in many games unit production (or ‘macro’) as well as elements of games where a key aspect is simply feats of mouse and keyboard speed and dexterity (or ‘micro’) and focusing purely on the tactical elements here.
- Yes, I’m aware of groups with formation locking but again capable players do not simply formation-lock their entire army and move it into the enemy as a single unit.
- Important word
- Though I must call out the otherwise somewhat weaker depiction of the Battle of Philippi in HBO’s Rome for the wonderful moment when Antony essentially points out, to Octavian’s horror, that at this stage he has basically no control over the battle, so he can either sit and watch or charge in, as he likes. Though the opening giving of orders, as we’ll see next time is utter nonsense: two fellows shout and then someone lowers a flag no one can actually see and everyone understands that is the order to advance.
- I’m going to use this word a few times so it is worth defining: a stratagem (from Greek στρατήγημα, passing through Latin as strategema before arriving in English as ‘stratagem’) is a ruse or clever trick a general might use that is outside the more typical arts of generalship. This doesn’t mean that stratagems were always seen as dishonorable (on this note E. Wheeler, Stratagem and the Vocabulary of Military Trickery (1988)), but just that these were ‘special tricks’ one might play, usually involving disinformation or surprise.
- In part due to some recent bad weather limiting both armies’ ability to get a sense of their surroundings.
- A notable exception here were armies that were entirely mounted, even if they dismounted to fight. Such armies could, so long as their logistics held up, often effectively outrun the news of their coming. Such fully mounted armies larger than raiding parties are rare in agrarian societies, but common on the steppe, a crucial advantage to steppe nomads.
- Since setting camps on hills was often advisable, it is very common to have situations where both armies are encamped on different hills with the low ground contested between them.
- Chiefly Eumenes of Pergamum
- Put a pin in this, we’ll come back to it, but restraining victorious soldiers from rushing for the nearest loot was a perpetual problem for generals. Getting a wing that had broken through to turn on the enemy center rather than plunder their camp demanded discipline and command skill.
- Not the mythological figure, just named after him
- There is a tradition in our sources that this was a clever ruse by Aemilius; I think Johstono and Taylor are correct in rejecting this – it is clear from our sources that both armies had to scramble to respond to the battle once it started.
- Important word, given the unknowns here.
- Please note that hoplite spacing is much, much more speculative than Roman or Hellenistic spacing, which is also very uncertain so I am speaking here in broad generalities.
- That is, a unit which would be expected to maneuver independently. Many smaller organizational steps are purely administrative, while many larger ones are essentially operational in nature.
- And I suppose for those writing fantasy fiction, some magical means of visual amplification might well have the same effect.