Collections: Total Generalship: Commanding Pre-Modern Armies, Part IIIc: Morale and Cohesion

This is the conclusion of the third part of our series (I, II, IIIa, IIIb, IIIc) looking at the role of the general in commanding pre-gunpowder armies in battle. Last time we looked at how junior officers, when empowered to act independently, could give armies a degree of flexibility and reactiveness on the battlefield but didn’t necessarily increase the amount of actual control the general exerted over them.

This week we’re going to finish up this analysis by looking at the rank and file soldiers themselves and the issue of morale. Film and video games generally present a binary portrait of morale: soldiers are either 100% following orders and prosecuting the battle vigorously or else running away in disordered terror and never anywhere between. But in fact the morale and cohesion of armies are complex and unpredictable things, with many possible outcomes. Moreover, popular fiction across many genres tends to represent armies as far more willing to fight on through high casualties as they were historically.

If this post has improved your morale, please share it. If you want to experience a sense of unit cohesion with the other ACOUP patrons, you can enroll in their serried ranks by supporting 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.

Gaming Morale

Most strategy/tactics games do not simulate morale at all, of course. In Starcraft or Age of Empires, all soldiers fight on until killed; they only retreat under orders (by the player) and never attempt to surrender. We’re going to set those sorts of games aside for the moment, however, and focus on games that do attempt some kind of morale simulation.

For games that do simulate morale, it generally exists in a binary state: units are either in good (enough) morale and will follow any and all orders the player gives them or they are ‘broken’ and will retreat uncontrollably to the rear. The binary itself is odd since, as we’ll see, morale and cohesion can exist in a lot of states. But it also has gameplay implications: because the shift from ‘fine’ to ‘broken’ is so immediate and severe (especially in games where it is unit-wide), the tendency in game design is to make the break-point very high.

The Total War series is probably the most popular strategy game series where morale plays a major role and provides a good example of the binary framework. As is the norm for such games, morale (or ‘leadership’ depending on the game) is a binary: units either execute your orders with full enthusiasm or run away uncontrollably with no binary space in the middle. There is a touch more complexity than this, but only a touch: units can either be ‘broken’ or ‘shattered;’ both flee uncontrollably, but ‘broken’ units can eventually reform if their morale improves. The way the game determines if a unit should flee is based on that ‘leadership’ or ‘morale’ stat: every unit has a different starting value (enhanced by gaining combat experience) which is then positively modified by things like a nearby general or friendly units and negatively modified by things like losses and danger; if they value drops below zero and stays there long enough the unit flees.

Now on the one hand, units in Total War consider a lot of things to determine their current morale: unit casualties, army casualties, the rate of unit losses,¹ routing friendly units, unit fatigue, the presence or absence of friendly units on the flanks and the presence of strong enemy units,² all matter to varying degrees. On the other hand, unit morale is generally set so high that most of these factors are at best indecisive unless they are stacked together with substantial damage to the unit. That is, it isn’t enough to flank a tired unit with a strong enemy and rout a unit next to them if the unit in question hasn’t taken substantial casualties they aren’t going anywhere.

To take some extreme examples, I set up some test battles in Total War: Warhammer II.³ First, I took the most bog standard infantry unit, the Empire Swordsmen – a solidly run-of-the-mill unit with neither particularly good nor particularly bad leadership (though the unit had 0 experience, so we are at the low end of ‘standard’) – and smashed them into a line of Har Ganeth Executioners, one of the strongest and most dangerous anti-infantry units available to see how long it would take for the line of swordsmen to run away.⁴ The answer, as shown below, was about 50% total army casualties – 1,168 models killed before the remaining 1,113 ran away. Then for the sake of setting a lower bound, I re-ran the experiment mashing an entire army of Bretonnian peasant mob – a unit noted on its unit card as having ‘low leadership’ – into an army of Swordsmen for a similarly lopsided fight; the peasant mob obediently charged the swordsmen and only retreated after having sustained roughly 1/3 losses.

This is, to put it mildly, not how actual armies behaved on the battlefield. But it is precisely how these units generally behave in Total War games. Indeed the community term for units like the peasant mob are ‘tarpits:’ cheap and weak units with lots of models and health which can hold down more powerful units simply by the time it takes to cut through them, especially if a nearby general or hero can provide the peasants with just a touch more morale. It’s a use-case for ‘chaff’ infantry with in fact depends on even extremely low leadership units sticking in the fight until they’ve taken significant casualties. And by and large they will do that; peasant mob’s 36 leadership is enough to keep them fighting even if they are charged from behind (-14) by a more powerful enemy unit (-4) with a bunch of their allies running away (-6) at the same moment their general dies (-10) if they haven’t yet taken any losses. And this is a peasant mob, the lowest leadership human⁵ unit available as far as I know. Any one of those things historically, as we’ll see, was sufficient to rout far more cohesive armies.

Now to be clear that doesn’t mean these morale modifiers don’t matter at all (at least on normal difficulty), they do. The key thing you are trying to do in a Total War battle is inflict losses to grind down enemy morale on all of their units and then use one or more of these ‘shocks’ to ‘spike’ that morale negative to rout the unit. But trying to rout even extremely low morale units without first grinding down their health is basically impossible. Indeed, it is so difficult that the game is set with a not-quite-hidden ‘backstop’ to the morale system: the ‘army losses penalty’ – a morale penalty which kicks up to absolutely massive values once the game decides that an army’s losses are too severe. Players on difficulties higher than normal will be very familiar with this because at any of the higher difficulties in the game, ‘army losses’ is just about the only way to actually rout high leadership units. But even on lower difficulties the ‘army losses’ penalty is necessary to avoid the odd spectacle of even very low leadership units that, while undamaged themselves, have just watched the entire rest of their army be crushed, inexplicably sticking in the fight.

Other games with similar ‘binary’ morale tend to be no less absurd in terms of how courageous they make even very basic units. Here’s a recent battle of mine from Mount and Blade: Bannerlord. The screenshot might be a bit hard to read if you are unfamiliar with the game, so let me decode.

Bannerlord handles large battles by having each army get a proportion of its forces on the field at start and then replacements ‘trickle’ in as casualties are taken. These two armies were on the same general order of magnitude, one army of 1000 and another of around 1400 (the former having much more cavalry).⁶ As you can see the defending army, out of its 1000, sustained 472 KIA (the red skull on the right) and 503 WIA (the broken shield on the right) before the surviving 37 troops retreated (the flag on the right). They inflicted only 119 casualties total; this was a preposterously, crushingly one-sided fight. Indeed, because of that trick effect only 436 of the 1,400 troops available to the attacker were even engaged. And yet the defender waited until they had sustained an astounding 96.3% casualty rate before attempting to run away.⁷ The wild thing is that the defending army featured a lot of cavalry (a typical Khuzait army, for those that play the game), which could have run away very effectively!

Not all games are quite so extreme, of course. Battles in Paradox titles – I’ll use Crusader Kings III as an example here because its the one I’ve played most recently – tend to feature casualty rates that are often a lot closer to historical norms. Looking through my own recent games, lopsided battles between armies of equivalent size tend to inflict around a quarter to a third losses (presumably both KIA and serious WIA since the ‘killed’ figure reflects all men not able to return to service), often with most of those casualties occurring during the ‘retreat’ phase of the battle – which is correct. The exception here are ‘stack wipes’ where an army is so wildly overmatched that it is defeated completely; I think we may take this to mean in most cases that the army collapsed and dispersed rather than that 100% of its troops were killed. Nevertheless a third to a quarter losses is closer to right, but actually probably still about triple the general average. It is not a good sign when the closest we seem to get is roughly three times the historical norm.

I’ve focused on game morale systems here, but of course this blends over into film as well, where the ‘mooks’ often charge the heroes seemingly utterly heedless of their losses – frequently despite the fact that the last identical group of mooks to do so just got taken apart before their very eyes. And invariably they do this until they are so beaten that they switch to the other binary state, simply running away.

Actual armies have far more than two states of morale and behaved in far more dynamic, unpredictable and interesting ways!

Morale and Cohesion

The first problem with this ‘binary model’ of morale is that it assumes just a single factor (‘leadership’ or ‘morale’) but in practice we ought to be thinking about at least two different ingredients here: morale and cohesion.

Morale is the commitment the combatants have to their leadership and their cause. To simplify a bit, we might say that soldiers with good morale believe three things: that their cause is a worthy one, that they are on the road to success and that their leaders have a good (enough) plan to achieve final victory. Poor morale can result from a breakdown in any of those three elements: troops might for instance believe both in their goal and its eventual possibility but not in their leaders to produce it (this seems to have been the case, for instance, in the French Mutiny of 1917, discussed below). On the other hand, regardless of the charisma of leaders, few people come to a war intending to die in it; if the cause appears impossible, morale will sink regardless. And armies that do not believe in the cause at all are extremely difficult to motivate by other means.

On the other hand cohesion is the force that holds a specific unit together through the power of the bonds holding the individual combatants to each other and/or to their (generally junior or non-commissioned) officers. There are a lot of ways to build that cohesion: people are generally unwilling to abandon neighbors, close friends and relatives, for one. They are also reluctant to expose themselves to shame at home for having done so; shame is one of the few things people fear as much, if not more than, death. For armies that can’t rely on that sort of organic cohesion, it can be built by reconstructing the soldier’s unit as his primary social group. Drill can do this: it creates an experience of shared suffering and achievement which bonds the soldiers together creating strong ‘artificial’ cohesion.

These two ingredients have different roots, but they also function differently. The formulation that has always stuck with me is one from James McPherson’s For Cause and Comrades: Why Men Fought in the Civil War (1998): morale (McPherson discusses it under the heading of ‘the Cause’) will get men into uniform, it will sustain them on large marches and cold nights and it will get them to the battle, but it will not get them through the battle. Instead, cohesion (the ‘comrades’ of the title) gets men through the terror of actual combat, when fear has driven ‘the cause’ far from mind. But of course cohesion isn’t enough on its own either, since it provides no reason to advance or attack or really to do anything at all except stick together.

Adding further complication to this, morale and cohesion are not, as they often exist in games, inherent properties of a unit, but rather emergent properties of the interactions of a whole bunch of individuals. In a strategy game, units exist primarily as extension of the player’s will; in film units typically exist as extensions of their commander’s or the main character’s will (note how common it is that right as the hero begins winning his duel with the villain, so too his army begins winning the battle). But of course actual armies are composed of lots of humans, each with their own individual will and agency.

Those humans are continually making calculations about risks, goals and survival. It’s not hard here to see why, by the by, morale won’t carry troops through high risk conditions: if your only goal is to survive to experience the end-state of the war, then it is always in your interest to let someone else do the dying; it doesn’t serve your end to stay in a high risk position. By contrast, if you are held there by the fear of shame if your close comrades see you run, that still applies. Thus these calculations get progressively more ‘primal’ as the sense of danger rises (fear makes a mess of those higher brain functions), but they do not stop.

Complex Outcomes

The result is a wide range of complex and unpredictable outcomes, rather than the neat binary state we tend to see in video games or film, where everyone is either 100% committed or 100% running away.

Consider, for instance, an army with high morale but with weak cohesion. This is actually so common an occurrence that it actually becomes tricky to point out specific examples; its most common expression are units at the beginning of a war that despite being highly motivated ‘for the cause’ nevertheless fall apart quickly when the fighting starts. A classic example is the First Battle of Bull Run (1861); both army’s morale was evidently high (they both thought they were about to win a brief and glorious war!) and that morale carried them into the battle just fine. But when it became evident that the US Army wasn’t winning the day, that morale was insufficient and the new and not yet sufficiently trained army lacked the cohesion to hold together, leading to a panic and collapse where the units themselves dissolved in a frenzied retreat.

What is striking in this example is that the army didn’t dissolve, because while cohesion had failed, morale had not. After the frenzy of the retreat war off, the army mostly reformed; only 1,216 US soldiers were reported missing after the battle (of 18,000 or so engaged), so most of the men found their way back to their units (one way or the other). Meanwhile the public and political leaders remained committed to the war, redoubling recruitment and enlistment; that commitment both reflected and influenced army morale: victory was still desirable and still possible. And so the army reformed to try again. The American Civil War is replete with examples like this where unit cohesion failed, but those same units after embarrassing retreats reformed and rejoined the army, often to put in solid service in later battles. Of course penalties for desertion and inducements for service also might play a part in this, but as I think McPherson (op. cit.) demonstrates, for both the United States and the Confederacy, ‘the cause’ was the driving factor. Morale brought the armies back together and sustained them through the campaigning – but it was not enough to hold together in battle until cohesion had been built within units.

What about an army experiencing a collapse of morale but with high cohesion? Here the particulars of the breakdown in morale matter a lot. A collapse of cohesion produces a fairly predictable result: the unit falls apart under the immediate stress and flees. But a collapse in morale, especially where cohesion is high means the unit stays together and generally continues to act in concert. The fact that they still act in concert means whatever they do will have all of the force of that cohesion, but the collapse of morale makes that action unpredictable.

Units may, for instance, hold together but refuse to attack or refuse to attack vigorously, especially in cases where the troops have lost faith in their leaders, but not in the cause. A classic example of this outcome is the 1917 French Mutiny. The French Army, having sustained very heavy losses in the previous years of the First World War responded to the failure of the Nivelle Offensive in April 1917 with a general mutiny that rapidly spread over much of the front. The immediate cause was the ‘crash’ in morale after the great optimism for the offensive (it was supposed to win the war) was dashed by its failure. What’s striking is that the mutinous soldiers in question didn’t simply desert en masse; they didn’t just go home, even after it was clear that the officers had lost control and more than half of the French army was refusing orders. Instead of retreating or rebelling, the mutinous divisions mostly just refused to advance; they had lost faith in the strategy and the generals, not the cause.⁸

In this case it also impacted the solution. While there was an immediate effort to crack down on the soldiers, what really seems to have alleviated the mutiny was a change in command: Nivelle was dismissed, replaced by the far more popular Petain⁹ and he promised a more careful strategy which wouldn’t waste the lives of his men (along with better pay and leave). The French Army had lost faith in its leaders, so a change in leader was able to restore the situation, at least to a degree.

Alternately, a highly cohesive army whose morale fails may turn on its leaders completely; this is particularly notable for armies that are at best weakly attached to the fortunes of the state or society they serve. Thus for instance after the end of the First Punic War, Carthage’s armies, composed mostly of mercenaries, turned on the state and instead formed the backbone of a widespread North African rebellion against Carthaginian rule (for more on this, check out Dexter Hoyos, Truceless War (2007); alas it is not cheap so try your library). A similar early modern example is the behavior of the Army of Flanders in 1576: having been left unpaid for an extended period due to the financial difficulties of the Kingdom of Spain, the army responded by electing new leaders and then brutally sacking its own regional capital, Antwerp. What’s striking here is that the army did this while functioning in perfect order (albeit against one of their own cities) – being able to execute a capable storming attack on a defended, fortified city – because cohesion was not the problem and the troops knew their business.

More broadly because morale is the force that holds the army together when it is out of danger, cohesive armies with low morale can suffer from contagious mass desertion where the army itself seemingly melts away admits a wave of mutinies, as with for instance the collapse of the Russian Army in 1917.

Alternately, morale failures can result from situations where the army itself is still a potentially effective fighting force, but victory itself is no longer possible. The most common way this happens in the pre-modern world is in wars of succession: even if the army is still functioning, if a faction’s proposed heir dies, victory is no longer possible. Thus for instance even though as a pure engagement the Battle of Cunaxa (401 BC) was indecisive, the fact that Cyrus the Younger (claimant to the Achaemenid throne) was killed meant that most of Cyrus’ army collapsed (though its Greek contingent, the famed 10,000 stuck together and fought their way out of the Achaemenid Empire, though one assumes had ‘vanishing away into the countryside’ been an option they’d have done that). Strategy games often have a big morale penalty for ‘general killed’ (which for some reason doesn’t remove your control over the army usually; Mount and Blade is an exception here – if your character is incapacitated, you do lose control of your army) which often seems to hearken to this kind of collapse, but it is worth noting that the army collapsed not because it lost any old general, but because that general was himself ‘the cause!’

A similar but more detailed interaction of morale and cohesion is visible with the Battle of Hastings (1066). William’s initial push up the hill failed and the rumor started in the army that he had been killed; since the war was a succession war that would mean the loss of any hope of victory. William had to ride along his line with his helmet off to show his face so that his men could see he was still alive and victory was still possible to reorganize them for another effort. On the other side, the death of Harold seems to have caused the collapse of the English line but in different ways. The troops of the fyrd seem to have collapsed and fled, melting into the countryside (failure of morale and cohesion), but the highly cohesive mass of the huscarls held position around Harold’s body, cohering in a single mass to the end.

Morale and cohesion can thus fail in different ways, sometimes leading to units that will respond to some orders but not others, or units which remain cohesive and act in concert but are no longer under the control of their officers. Perfect obedience and panicked rout are two of the possibilities, but only two states in a broad spectrum. And of course it is also possible for combatants to be too aggressive, attacking without orders, sometimes under the leadership of their officers, sometimes not. This sort of aggressiveness was itself a core part of the Roman ideal of virtus which in turn required disciplina to retain (at least until a Pelignian threw a flag at someone. Then all bets were off).

Now I do want to briefly note there are some games that do try to express some of these complexities, but they are mostly turn-based rather than real-time games with much smaller audiences. The old Sierra Civil War Generals¹⁰ titles, for instance, simulated both morale and organization; units with low morale would first refuse to charge (but still initiate rifle exchanges, though often half-heartedly), then refuse to attack, then refuse to advance at all (but could be ordered to dig in in their current position), then retreat in good order, then rout. More recently, some of Byzantine Games’ turn-based titles like Pike and Shot have units move through levels of cohesion (Steady/Disrupted/Fragmented/Broken), each of which lowers the combat ability of the unit; fragmented units can’t charge and may give ground while remaining minimally combat effective. But these more detailed considerations of morale are very rare in real-time titles.

When Does Morale and Cohesion Fail?

But this common binary system creates another related problem: because the player goes from having 100% control to 0% control in a single instant switch, the threshold for that switch is set very high. Armies often have to be attrited down to close to half their original numbers before units even begin considering retreat. The result are armies that are ahistorically accepting of combat casualties, with both skews the sense of how lethal warfare was but also how battles tended to flow.

For instance, as noted above even ‘peasant mob’ can be made both to charge and to receive a charge from basically any unit in a Total War game. Indeed, that’s the point: the way you are supposed to use ‘tarpit’ units like peasant mob is to have them absorb damage from more expensive units while your own expensive units deal blows to the enemy. Except that on actual pre-modern battlefields, units frequently broke before receiving a charge. For instance, here is how Thucydides describes the Battle of Mantinea (418, Thuc. 5.72.3-4):

As soon as they came to close quarters with the enemy, the Mantinean right broke the Sciritae and Brasideans, and bursting in with their allies and the thousand picked Argives into the unclosed breach in their line cut up and surrounded the Spartans, and drove them in full rout to the wagons, slaying some of the older men on guard there. But if the Spartans got the worst of it in this part of the field, it was not so with the rest of their army, and especially the center, where the three hundred hippeis, as they are called, fought round King Agis, and fell on the older men of the Argives and the five companies so named, and on the Cleonaeans, the Orneans, and the Athenians next them, and instantly routed them; the greater number not even waiting to strike a blow, but giving way the moment that they came on, some even being trodden under foot, in their fear of being overtaken by their assailants.

(Emphasis mine. Translation is R. Crawley from the The Landmark Thucydides, ed. R.B. Strassler, my only minor change was to leave hippeis untranslated; these fellows are the elite guard of the Spartan kings. Though called hippeis (‘horsemen’) they fought on foot.)

In short on the left (to the Spartans) side of the field, the Spartan forces (for what each of these groups is, see my series on Sparta) collapsed almost instantly on contact, while on the right side, the opposing Argive (et al.) force collapsed before contact, at the very onset of the Spartans. That’s not uncommon for units or indeed even entire armies with weak cohesion or morale to collapse before contact or at the very moment of contact. After all, the onset of an enemy is very scary and even with your armor and all of the cohesive elements designed to hold you into the fight, everything in your monkey brain in that line is telling you to run away; indeed, as we’ve discussed, ‘shock’ cavalry relies on this feature of human behavior in order to be effective, since such cavalry is almost useless against infantry that remains dense, cohesive and in good order.

If anything, by the by, this tendency of units to break before contact becomes even more common in the gunpowder era. Part of the reason armies kept doing bayonet charges into the 1800s is that they could be very effective. While it was very hard to get a unit with the cohesion and courage to make a bayonet charge, for those that could, it was extremely rare for the defenders to wait to receive such a charge: without armor or shields, casualties were likely to be extremely severe. And there is something in our barely evolved monkey-brains that really understands someone running at us with a very sharp stick on a more fundamental, primal level than it understands artillery or gunfire. It was thus relatively rare for bayonet charges to result in an extended melee: instead the defender typically poured fire into the attacker who either broke and fled or else delivered the charge in which case the defender broke and fled with those who could not flee in time being killed or surrendering.¹¹

Of course not every unit ran away before striking a blow and casualties did happen. Figuring out the average casualties taken before an army collapsed is in most cases beyond the evidence, though. The problem is that most losses occur after an army breaks and runs, since combatants who are fleeing no longer are operating as a mutually protecting unit or actively defending themselves. But we don’t get casualty figures in our sources for the moment when an army breaks, we get figures for the end of the battle itself (and these are often of questionable reliability). Still, these ‘total losses’ if we keep in mind that ‘pursuit casualties’ are included, can give us some sense of the ‘range of the normal’ as it were.

Peter Krentz did a general survey of Greek hoplite battle casualties¹² and concluded that attested loss rates (only men killed; we have no WIA figures for Greek hoplite battles) at 5% for winners and 14% for losers; this leads to the generally quoted average of roughly 10% losses in a hoplite battle. Strikingly, the variation is pretty tight; there are few examples where any side suffered more than 20% losses. Now of course we need to be alert to the potential problems with the source material here though at the same time sources tend to exaggerate casualties, not minimize them. If we assume that loss rates might be close to even until the moment when an army breaks we might say that hoplite armies tended to collapse somewhere between perhaps 5-10% losses (and sometimes much earlier than that).

Nathan Rosenstein¹³ attempted a similar analysis (albeit his interest was more demographic) for the Romans stretching from 200 to 168 (the period where we have the best evidence) and notes that the averages of his figures for those years were 4.2% losses for winners and 16% for losers. Again one must stress the potential unreliabilities of the sources (which Rosenstein discusses) and also note that the period from 200 to 168 was a period where Rome did a lot of winning and not a lot of losing; the reported casualties for Rome’s enemies when they lost were often a fair bit higher but of course we must be careful there of exaggeration. Still this broadly tracks with the Greek figures and suggests perhaps a general rule, though it must be noted that there are outlier battles, particularly where the impossibility of retreat produced extreme casualties for the losers (something more common in Roman warfare where everyone was a lot better at pursuit than hoplite armies were).

So pre-modern armies generally seem to have fallen apart at or before 10% losses. That is somewhat sooner than modern military units (‘combat ineffectiveness’ rules of thumb vary, but they’re often between 10 and 20% losses), but as we’ve noted before shock combat exerts a very different morale and cohesion pressure than being under extended fire does.¹⁴ That’s still substantially lower than something like Crusader Kings III’s c. 30% or so average loss-rate for defeated armies and massively, wildly lower than the 50% or 90% losses-to-break rates implied by games in the Total War and Mount and Blade series (much less the 100% rates implied by many other real time strategy games).

Now it is not hard to understand why many video games set the thresholds for army collapse much higher. The design philosophy here is one that empowers the player and having an army collapse early in a battle from seemingly random or uncontrollable causes would be badly disempowering. Controlling low leadership armies under these circumstances would feel like trying to sword-fight with a pool noodle. But that’s precisely the point: controlling most pre-modern armies was like that. Likewise it’s not hard to imagine why film, with its preference for Big, Decisive Battles That Decide Everything tends to depict battles with very high casualties, the heroes and villains fighting on until one or both is utterly ruined. But once again, actual battles rarely turned out this way.

That has, as I see it, two implications, one small and one large. The small implication is that the standard practice in many of these games to ‘bog down’ an enemy’s expensive units in a mass of cheap forces so as to spare your own expensive ‘damage dealers’ from taking damage – a cornerstone of the ‘match-up’ model we discussed at the beginning of this series – doesn’t really function on an actual battlefield. Advancing low quality, low cohesion infantry without support will simply cause them to break for the rear without bogging down or potentially without even engaging their higher quality opponents. Peasant mobs know they’ll lose to dismounted knights, they will not, as a rule, stick around to put the matter to the test except under extreme circumstances.¹⁵ If you want infantry to hold against dangerous enemies, it generally has to be high quality, cohesive infantry. Attempting to do your ‘cost savings’ by cheaping out on your infantry ‘anvil’ was a good way to find yourself with nothing but your surrounded and soon to be defeated hammer.

The broader implication is for understanding the nature of pre-modern warfare in general, that casualties were, as a rule, relatively low compared to what we might expect. Even in our sources, which love to exaggerate the scale of victories, battles which result in the total and utter destruction of an army are rare and produced by special circumstances (typically the army is incapable of retreat for some reason). Consequently in these societies men in the ‘combatant class’ (whatever that was) were likely to participate in and survive more battles, including battles they lost.

Conclusion

At the beginning of the series, we introduced a ‘Total War‘ model of army command. That model was heavily focused on active command, with the general responding to changing battlefield conditions. The purpose that active control was put to was to create specific unit-to-unit match-ups (as well as using ranged units to specifically target and remove dangerous enemy units) rather than to maintain a contiguous front or coherent formation for the whole army. Those match-ups in turn play into a ‘rock-paper-scissors’ view of tactics where the key is matching enemy units with their specific counters in one’s own army.

And what I hope I’ve shown here is that almost none of that model of generalship was possible under the conditions that pervaded in warfare before the advent of gunpowder (and indeed, for some time after it). The general himself almost never had the information necessary to target specific units of the enemy maneuvering on the battlefield; the rare exceptions (like Leuktra (371) or Gaugamela (331)) were typically only possible because the general could guess in advance where those units would be because of a standard enemy disposition or where the general could respond to their position because he was personally leading the unit deployed against them (like the companion cavalry). Instead because dispositions had to be decided with extremely limited information, most armies deployed cautiously in ‘standard’ formations designed to minimize risk rather than maximize ‘match-ups.’

Which as well enough because even if a general somehow did get a ‘bird’s eye’ view of the battlefield, he generally lacked the capability to communicate those orders or have the army act upon them. Rather than having a nearly limitless set of opportunities to make changes to the army’s battleplan, the general was heavily constrained, having only a number of key ‘decision points’ where he could intervene and in those points in turn having only a limited ‘McDonald’s Menu’ of orders he could give – limited based on the training and capacity for synchronized discipline in the army. For most armies, this essentially limited the general to hitting “go” and hoping for the best, perhaps leading his own small unit more directly or making a show of himself in a way intended to inspire bravery in the rest of the army.

Actual responsiveness to evolving conditions didn’t come from the general at all, but was an emergent property of junior officers empowered to make independent decisions combined with armies that had sufficient training and discipline to act on those decisions in the moment. Such armies could be very effective, but they were also difficult to produce (as were the capable junior officers) and so a relative rarity. Even so, these armies didn’t always win: discipline and distributed command was an advantage, but only one factor of many. Tyche still rules the greatest part of the affairs of men.

And finally, even the plans of generals or their orders when they did get through were at best implemented imperfectly by armies that acted not as extensions of their commander’s will but as collections of humans with their on agency, courage and fears. Units ordered to charge sometimes refused; units ordered to hold sometimes broke before contact. Armies sometimes collapsed from sudden panics before much of the battle had been fought. Alternately, units sometimes put in reserve impetuously attacked anyway. An army itself is a rarely restrained mob and many generals, including quite good ones, sometimes lose control of those restraints.

The point here, I should stress, is not to reduce ancient or medieval armies to dumb mobs of humans mindlessly hacking away at each other. There was order to this chaos and there was method to this madness: these were, after all, thinking humans who tried to plan and tried to organize and generally succeeded about as well as humans do today. Instead, the point here is that the strictly controlled battlefield of film and video games is an illusion.

That has itself all sorts of implications but perhaps the biggest one is a blow to the ‘great general’ and his place in our mythology. The fact is the greatest captains often have at best limited influence on events. As we’ve noted, at Bibracte, Caesar seems to have played no role in the brilliant two-direction attack that won the battle; had his centurions been less creative and less adaptive, Caesar would have lost regardless of his considerable command skill. Indeed, when it comes to the Romans it is hard not to miss that they conquered the Mediterranean not with a succession of brilliant commanders but with a long line of modestly competent ones. Producing so many minimally competent generals was enough because brilliance was never the determining factor in any of Rome’s wars.

And that’s where we’re going to go next: what exactly it was that those modestly competent generals were doing outside of battles? Logistics mostly, it turns out. But while we’ve talked a fair bit about logistics in terms of what makes army movements plausible, we haven’t talked about the actual mechanics of feeding (and watering) an army on the march. So that is where we’ll turn to next; I had planned that originally as part of this series but I suspect I’ll break it out into its own series.

But first, next week has the Fourth of July, so we’ll have a bit of a detour for that.

1. implemented generally as a morale penalty based on the percentage of the unit killed in the last few seconds
2. Specifically stronger enemy units that are faster or as fast as the unit in question.
3. Mostly because it was the one I had installed at the time
4. I also sent their no-skills general on vacation at the back of the map, so they had no morale support from that. If the general had been with the line, this would have taken even longer.
5. Across the Empire, Bretonnia, Cathay and Kislev rosters
6. The model limit was at 500, my proportion is about 300, so the enemy was coming in groups of 200, they had about 1000 total troops, so it isn’t hard to back calculate that I had around 1,500 troops; it was actually slightly less than this because my characters ‘tactics’ skill tilted the on-the-field advantage in my favor.
7. The same laws of Gods and Men that require me to inform you that I have beaten both Dark Souls and Elden Ring now also require me to inform you that this battle was fought on ‘realism’ difficulty, where the player and their troops get no bonuses. That ‘budget tercio’ I showed a few weeks ago is really effective if you have – as you can see I have here – buckets of top-tier imperial heavy infantry and Battanian foot archers. Though I will admit, not all of my battles are so lopsided.
8. An interesting parallel to this were examples of United States units late in the Civil War during the Overland Campaign which would advance but due to low morale generally refused to charge, making only a show of attacking fixed fortifications at places like Cold Harbor. Cohesion kept them together, faith in the cause kept them in uniform, but the failure of faith in their leaders kept them from attacking vigorously.
9. Who, for whatever he would be in the years following, was in this moment, as he had been before in 1916, the savior of France. It is difficult to imagine any human being whose final reputation could have been more improved by having been fatally struck by a bus in 1939.
10. Would really like this game ported to modern operating systems, it was quite well put together.
11. As an aside, you can see the impact of this in bayonet training into the early 1900s: there is a lot of emphasis on attacking with ferocity, with a big fearsome shout. But while there is certainly some emphasis on bayonet fencing, it often seems decidedly secondary. The point of the training was as much if not more to build the necessary will and courage to make the charge as it was to build up the manual skill of wielding the weapon; a real contrast with the priorities in combat training of the pre-gunpowder era.
12. in P. Krentz, Krentz, ‘Casualties in Hoplite Battles’, GRBS 26.1 (1985)
13. In Rome at War (2004)
14. And of course ‘combat effectiveness’ is measuring something a bit different than if a unit is routing.
15. And yes, those circumstances do happen, but they’re rare.