This is the second half of the fourth part (I, IIa, IIb, III, IVa) of our honestly-who-knows-how-many part series laying out some general guidelines for how pre-modern armies are organized. Last week we took a look at how the leadership of these armies was drawn from the existing elite classes of their societies. As with other factors, armies cannot help but recreate their civilian social structures on the battlefield, with the result that the military leadership class often mirrored, where it wasn’t simply identical with, the civilian ruling class.
Now it will not surprise you that pre-modern military literature, written almost entirely by elites in the leadership class for other elites in the leadership class, is very interested in leadership and tends to attribute all military success or failure to the fellows in charge. Modern military science, however, has increasingly recognized the importance of combat motivation in individual rank-and-file soldiers.1
As noted last time, we can break this broader concept of combat motivation into two parts: morale and cohesion, where morale is the soldier’s attachment to the cause and belief in its possible achievement and cohesion is a soldier’s attachment to his fellows. What combat motivation studies have tended to show is that morale will get a soldier to the battlefield, but will not keep him on it; it is cohesion instead, the ability of a unit to stay together under pressure that is essential for enduring the terror of combat.
That said, different armies and different societies generate cohesion in different ways, as we’ll see. Now this is one component of this series that is a bit of an exception: so far I’ve been hammering that most societies are going to be pushed very strongly towards specific kinds of recruitment, organization and leadership as a result of their civilian organization. Meanwhile, as we’ll see, certain kinds of cohesion depend to a greater or lesser degree on specific social structures or assumptions, but it is a lot more common to see armies that simply have poor cohesion due to a mismatch of cohesive principles and military or social structures.
But first, as always, recruiting and maintaining large pre-modern armies is expensive! Much like many of those pre-modern armies, this project is supported by devolving the costs of my ruinous book-buying habit on to recruits readers. You can help by spreading the word to new readers and by supporting this project over at Patreon. If you want updates whenever a new post appears or want to hear my more bite-sized musings on history, security affairs and current events, you can follow me on Bluesky (@bretdevereaux.bsky.social). I am also active on Threads (bretdevereaux) and maintain a de minimis presence on Twitter (@bretdevereaux).
The Nature of Cohesion
We’ve talked about cohesion a few times here, so before getting into the systems that can generate cohesion, we can be fairly brief on what cohesion is. As the word implies, cohesion is the ability of a unit to hold together under pressure, but this is one of those concepts with a simple definition but more complex application.
On the one hand, cohesion is one of those ‘harder said than done’ principles. When a group of any animals is subjected to peril, the instinct is to flee and for humans, who are not herd animals, the instinct is to scatter, to flee individually. The goal of cohesion is instead to have individuals respond to peril by seeking the safety of the group rather than the safety of individual flight, so that intense fear causes the group to solidify rather than disperse. That is, as we’ll see, a challenging-but-not-impossible emotional response to stimulate, as humans are pack animals. That said, it has to be an emotional response, something sub-rational, because it needs to functions in conditions of intense fear where rational thinking is ineffective.
So you can think of cohesion, in its simplest form, as a kind of social or mental conditioning that aims to alter individual’s fear response from, ‘split up for safety’ to ‘group up for safety’ at an emotional, sub-rational level.
And then on the other hand, that has complex implications for battlefields in which humans generally fight as units rather than as individuals. Not all humans fight this way, I should note: warriors in a ‘first system’ ‘pounce-and-flee’ warfare system – as we see with hunter gatherers or early agriculturalists – often fight in a far more atomized, skirmishing style with no regular formations and no expectation that individuals will hold their ground or cohere under pressure. Instead, warriors in such a system can fall back and regroup relatively freely. But complex agrarian (even non-state complex agrarian) and pastoralist armies do fight as larger units, with complex group tactics that demand individual combatants work together and stick together much more tightly.
Meanwhile, if the unit remains cohesive – if the individual men remain bound together as a single mass – then their leaders can more readily maneuver the whole mass into action. After all, if the presence of peril is psychologically pushing each soldier to cohere for collective safety, then an officer can use that cohesion to push the group as a unit, potentially even moving it into greater peril. This is the foundation of all shock tactics – at least, all of them involving an even remotely equal opponent – because delivering ‘shock’ (that is, advancing into close combat in a group) demands advancing into peril, which is a think that is normally psychologically quite hard to do. As Ardant du Picq notes, “Man does not enter battle to fight, but for victory. He does everything that he can to avoid the first and obtain the second.” Getting fellows to walk into range of an opponent’s spears (potentially walking through the range of their slings, bows or muskets to do so) requires moving directly against human psychology. But if the whole mass of men coheres, the mass may be moved forward: group psychology succeeds where individual psychology would fail.
Of course cohesion is equally relevant – for the same reasons – for getting men to stand under fire, either in a tight formation or in more dispersed skirmish order (as in modern warfare). Once again, the normal human response to being in a ‘zone of danger’ is to leave it, so the cohesion response is required to get men to stay in the ‘zone of danger’ – the unit is in the zone, the men have cohered to the unit, so the men remain because the unit remains. Now, absolutely there are a small number of people in any group who have the right psychological mixture to do these things alone, but armies cannot be built out of a tiny handful of saints and sociopaths, they must be built out of regular men.
So cohesion serves several related purposes on the battlefield: it keeps men from fleeing but it is also the means by which they can be pushed into offensive action, which is to say it both keeps them from fleeing but also gets them fighting, all as a product of the action of holding them together. Even if a unit must withdraw, cohesion allows them to withdraw as a unit – maintaining fighting capability (which in turn deters an enemy from pursuing too aggressively) – rather than breaking apart into a rout (which an enemy can pursue aggressively).
So how does cohesion happen? High cohesion is most often the product of tight bonds between the men in a unit, though those bonds can be achieved in a range of ways. That said, as we’ll see, certain ways of producing cohesion can also produce a kind of corporate cohesion, where the cohesive principle ends up rooted in the identity of the soldiers (which they share), which creates a kind of cohesion that persists even when soldiers do not necessarily know their fellows (but when they do know that their fellows are also soldiers and so share the same formative, cohesion-building experiences; that’s why I frame it as ‘corporate’ in nature).
So cohesion is important, but it is also fundamentally unnatural: it must be produced somehow, it cannot be assumed. In practice there are two ways (or categories of ways) to produce cohesion: it can be produced in civilian society or it can be produced ‘in the ranks.’
Cohesion Dyed in the Wool
One way to achieve strong cohesion is to have it produced organically through civilian society. Tight social bonds are the key mechanism for producing this kind of cohesion: it derives from serving and fighting next to men who you know well in peace time and who will continue to know you well after the fight is done. The social pressure not to let these fellows down – fellows you will have to live with long after the smoke clears – is one of the few pressures that is as strong as the fear of dying. Or, as I sometimes put it, the one thing humans might fear more than death is shame. But for shame to work, it has to come from a peer group the individual will have to live with.
In practice, this kind of cohesion then tends to come one of two ways: from a small community or from a small, tight-knit group within the community.
The classic example of the former case are various forms of civic militias, like Greek hoplites. The hoplite phalanx of a Greek polis was generally divided into sub-units by tribe or neighborhood (phyle or deme) so that men served next to their friends, neighbors and family members in the phalanx as it was drawn up. Of course in some poleis, this sort of closeness was institutionalized through common dining groups, but it seems pretty likely that regular social dining with the men who would be near you in the phalanx was common across Greek poleis. Consequently, when a hoplite stood in the ranks, he knew he was being observed by everyone (at least, all of the men – though he could be assured his mother, wife and children would hear of his conduct too!) in the world who mattered to him.
The result is the creation of what is termed in military science a primary group: the small, face-to-face social group around a soldier, which tends to be tight and cohesive. Instead of generating that group ‘in the ranks’ – impossible for an amateur militia force that only formed up when there was a battle to fight – these sort of citizen militias instead form the primary group during peacetime and then use it as a basic building block in wartime to achieve that cohesion. As a result, a civic militia like a hoplite militia could be cohesive despite having little or no formal training or collective drill and relatively little in the way of command and control structures. Their cohesion was organic, derived from underlying social structures, rather than needing to be intentionally manufactured through training.
The alternative for ‘organic’ cohesion is the creation of a smaller unit of society that is similarly cohesive, often in the context of a much larger (and thus more anonymous) society which struggles with the kind of ‘small town’ cohesion of the Greek polis. This, of course, is generally the cohesion of a warrior class: the overall society may not be small enough for everyone to know everyone else well enough to generate that kind of tight social pressure, but the warrior-class might be. The aristocracy broadly construed, for instance, of even a large medieval European kingdom might consist of only around a thousand or so households. The kingdom might not be organically cohesive, but the aristocracy could be – though as we will see, knightly cohesion was reinforced through training and drill as well. One of the most striking examples of this kind of cohesion in action to me has always been the performance of the English huscarls at Hastings (1066): even as the rest of the army crumbled and the battle was lost, this group of vocational warriors held together, despite tremendous pressure.
Hopefully, you are also beginning to see how some of what we’ve laid out in these posts can ‘line up’ to produce strong military systems. A civic militia, for instance, can benefit greatly from these tight social bonds: in the ranks, everyone knows everyone else and the unit is led by elected officers who are the same men who lead the community in peacetime. The whole unit can thus draw on the familiarity and strong social bonds without needing to build those connections from scratch. The same, of course, goes for the retinue of a Big Man, whose retainers all know each other (and will continue to do so after the battle because they’re all socially connected) and who all know and already respect the authority of the Big Man who commands.
Now there is another requirement to making this kind of cohesion work effectively: there has to be important social value placed on military performance. For the militia man in a medieval schuttersgilde or a Greek hoplite or a knight riding out in his lord’s retinue, fighting bravely was crucial to their status in the community. Their friends, family and neighbors would be watching and would care if they stuck together or ran away. But for that to work there has to be some status for these men attached to the performance of military service.
A village can have very close-knit social bonds, but if the peasants living there don’t consider the role in society to be fighting, don’t value military service and don’t see themselves as personally involved in the wars of their rentier-elites, it will be almost impossible for those rentier-elites to actually harness the latent social cohesion of the village. Those villagers won’t hold together under pressure on the battlefield, they’ll scatter together – their social cohesion will be addressed in the ways they find socially valuable (like supporting each other with food during hard times). One common ‘failure state’ of military organization is where a political leader raises a population expecting to get organic cohesion but – because the society has not invested the social value in the fighting that those kinds of people do (so they may value knightly warfare but not the infantry warfare of peasants, for instance) – the cohesion simply isn’t there and the force collapses when put under strain. Levies drawn from the underclasses of a society – peasants swiftly formed into a local militia, for instance – often collapse this way under pressure, because the society at large (which is to say, the elites) have not dedicated the social value or resources to build organic cohesion.
In addition, this kind of organic social cohesion is very hard to ‘scale up’ since it relies on relatively small-scale tight-knit social bonds. One solution, however, was for large states to recruit the component units of their armies from specific locales, in what is termed a regimental system. Under a regimental system, soldiers do not move from one unit to another but rather are raised and spend their entire service career (which may be short-service or long-service) in a single unit. Often, when this system was combined with volunteer or short-service units (as in, for instance, the American Civil War or the British army in the First World War), units were intentionally raised from small geographic areas: individual states, cities, towns and so on. While the regimental system itself is a modern phenomenon, similar sorts of systems existed earlier; the Roman socii system seems to have functioned on a similar basis: Roman socii stayed in units recruited from their relatively small communities.
A larger army can thus try to harness some of that local cohesion by recruiting local units and keeping them intact, although this also creates organizational challenges: those soldiers are unlikely to be as willing to serve outside of their original units. They was told, you see, that they’d be serving with the 2nd Maine and the 2nd Maine only. That can cause issues if the original unit has been attritted so badly that it is no longer really functional (but cannot readily be merged with others), but also could create problems because it could also geographically concentrate casualties. The famous, grim example of this was the 11th Battalion of the East Lancashire Regiment (the Accrington Pals), all recruited from the same town, suffered 80% casualties in just thirty minutes of fighting at the Somme (Jul. 1, 1916), practically obliterating an entire generation of the town.2 The British abandoned the recruiting of ‘Pals Battalions’ after the Somme in no small part for that reason.
However, as you might be picking up, not every society generates strong enough organic cohesion to orient a military around (or the ‘warrior class’ that does is too small to form the bulk of the military). So then the question becomes how to create cohesion.
Training Cohesion
And it turns out it is possible to train cohesion into a unit.
What is striking is that historical societies talk about this process using a range of different understandings for how exactly they are inculcating bravery through training. Greek and Roman writers often frame this sort of training not necessarily as teaching soldiers specific, coordinated motions (the swinging of a sword, the throwing of a javelin), even though we know that the Romans, at least, did drill that sort of thing, but rather as rendering soldiers inured from hardship by putting them through hardship, by having them undertake strenuous duty regularly with strict punishments for failure (and generous rewards for success).
By contrast, early modern European drill, though it equally involved a lot of strenuous work, was understood more at the time as trying to produce a ‘mechanical soldier.’ The early modern European aristocrat largely concluded that the raw lower-class clay he was forced to work with for his army was temperamentally unsuited for bravery (these fellows were mostly stuck up snobs) and so could only be made brave by having the necessary actions for battle – loading and firing firearms, moving in formation – drilled over and over again until they became mechanical muscle memory. They use the same methods as the Romans – repetitive training of basic actions under as realistic circumstances as possible, with strict punishments for failures of discipline – but understood the mechanism of action to be quite different.
Now I want to be clear here what I mean when I say drill. Drill in this context is training through strenuous repetition of simple actions in preparation for combat. Not every army drills! There’s also a distinction here between armies that value “collective synchronized discipline” – the ability of soldiers to move together in synchronized action – and armies that don’t (you can use muskets in collective, synchronized musket lines but you can also use them independently in skirmish formations, for instance).3 It is likewise possible for armies to insist on physical conditioning in ways that do not produce group psychology effects, the difference between training together and training individually.
Now to be clear, obviously rigorously drilling soldiers in the manipulation of weapons is going to train what we can call ‘skill at arms’ – the effective manipulation of complex weapon systems – but we’re interested in the psychological conditioning here.
And the way modern thinking understands this same process is rather different than either ancient or early modern aristocratic thinking, in part because modern researchers and scholars have bothered to occasionally ask common soldiers about it. While the concept of unit cohesion has existed, I’d argue, since antiquity, it really only started receiving really systematic study after the Second World War, with the main conclusion of those early studies being that unit performance was heavily sensitive to the tight bonds in small units. Which is to say soldiers were motivated to fight for their ‘buddies’ at the squad, platoon and company level and that this social bonding effect was the primary source of unit cohesion.
The question, of course, was how a military went about manufacturing those sorts of social bonds so that units cohered under pressure. But of course the military wasn’t approaching these questions with a blank slate: in many ways the western military tradition (hardly alone) had already developed a system for manufacturing unit cohesion – what it was coming to realize is that it had never really understood how that mechanism functioned.
Instead, while the role of drill in training specific skills is important, it also imposed upon each batch of recruits intense conditions of shared suffering combined with creating a barrier to entry into the final group (one had to endure the initial drill and training process). In short, taking a bunch of potential soldiers and throwing them all into the same difficult situation (physical training and drill) they have to overcome together over an extended period of time (weeks or months, if not longer) is a way to ‘speed run’ the social bonding process to create a strong, cohesive unit.
One of the risks this approach is thought to run, however, is that rotation or high casualties can erode that cohesion. The soldier has cohered, after all, to a ‘primary group’ consisting of the men he trained or served long-term with, but that bond can be ruptured as members of the primary group are killed or rotated out of the unit (or the soldier is rotated into a new unit where they don’t know anyone). One of the responses to this reality is to avoid individual soldier rotations (such rotations in the Vietnam War were famously blamed for low morale and discipline) and instead rotate whole units, but of course there is nothing doing about casualties and the need to replace them.
The thing is, while certainly some units suffer reduced cohesion and thus reduced combat performance as they take casualties and – in the accidentally memorably bloodless phrasing of one of my graduate school colleagues – “the primary group disintegrates” that also quite clearly doesn’t happen to every military. It is a bit of a grim (and very modern) example, but the Wehrmacht (the German army) in WWII remained cohesive despite absolutely staggering casualties by war’s end and the effective impossibility of maintaining meaningful primary groups in the face of those losses.
In a similar vein, new assignments or distant posts often seem substantially less disruptive on cohesion in long-service professional forces. And indeed, mercenaries and long-service professionals can show remarkable cohesion under strain, even when the men in question haven’t necessarily been trained together. Xenophon’s Ten Thousand, recorded in his Anabasis, provide a notable example of this: the Ten Thousand were a mercenary force composed of men from ten different smaller mercenary units, scattered over the Greek world and in most cases we have to imagine the men who joined were not joining together with their neighbors but as deracinated mercenaries. Yet the Ten Thousand show extraordinarily strong cohesion under pressure, both in pitched battles but also under the steady strain of a long march under continuous threat.
That sort of ‘plug and play’ cohesion also shows up in long-service professional armies as well as other kinds of armies that have simply been in the field for a long time as the result of very long wars. One can see it, to take a modern example, quite clearly in the performance – in both victory and defeat – of British regular troops during the Anglo-Zulu War. At Isandhlwana (Jan 22, 1879), a British column was effectively annihilated when enveloped by a larger Zulu force but most of the regulars remained cohesive to the end even when it was clear the battle was lost. The same intense cohesion among the regulars enabled an unlikely victory the next day at Rorke’s Drift (Jan 22-3, 1879) when a badly outnumbered garrison post held off a large Zulu attack. Of course we should also note the remarkable cohesion of the Zulu (in their case a mix of organic cohesion and cohesion developed over frequent campaigns) in these battles, where they were able to persist on the attack despite shockingly heavy casualties.
What I think is going on here is that over time these fellows are being bound not only to a primary group, but also a corporate, professional identity, effectively a professional version of the corporate identity of the ‘warrior class’ discussed above. On the one hand, all of these ‘old soldiers’ still have shared experiences, because even though they were not drilled together, they were all drilled the same way and they all ‘know their business’ the same way. At the same time, they have a common connection to a shared identity nonetheless: they’re all soldiers, not merely by circumstance but by identity. Being a soldier is not something that has happened to them, but something that is part of them. And of course those elements – difficulty of entry, commonalities in experience in the group, a shared sense of belonging – are all elements that encourage group cohesion.
Now again, in both cases, the actual mechanism for producing this psychological and social bonding effect is drill and discipline, which of course also have other effects (as all training does) and are often done for those other purposes, developing cohesion as an accidental (but very necessary) byproduct.
Now as you might gather, the research on training and unit cohesion is, at this point, voluminous; this can only have been an overview. But I think, at least for thinking about worldbuilding (and also a general framework in which to understand the specialist literature), thinking in terms of two categories of ‘manufactured’ cohesion – primary group cohesion (a result of a group being trained together and undergoing hardship together) and corporate identity cohesion (a result of a shared sense of professional or corporate identity among a much larger group, married to shared experiences of training and hardship). However remember: your subjects may view the matter quite differently, understanding the improved combat performance to be the product of something other than the psychological conditioning that modern research focuses on.
Cohesion and its Absence
That leaves us with really four major avenues for a unit to develop effective cohesion: the organic cohesion of small communities (up to the level of a small town, but not generally much larger), the organic cohesion of selective groups (like warrior classes or specific guilds or social clubs) with in a larger, less cohesive society, manufactured primary group cohesion that results from training a unit together and having them endure hardship together and manufactured corporate identity cohesion, a result of a shared sense of identity among a larger group that have a sense of shared hardship.
Note, of course, that there’s no reason these mechanisms for cohesion cannot be combined. Roman citizen armies in the Middle Republic were recruiting with at least some substantial organic cohesion from the citizen class (and the socii in their even smaller, potentially more cohesive groups), which was then reinforced by strict discipline and the experience of military life, effectively layering small community cohesion on top of manufactured primary group cohesion. Likewise, a medieval knight (or other mounted warrior) is already coming with the organic cohesion of a warrior class, but knights also trained together extensively in small units (the conroi, a unit anywhere from a half-dozen to couple-dozen mounted warriors) and would fight in battle as part of that unit, layering that primary group cohesion over top that warrior class cohesion.
The thing is military units are not always cohesive. This is a bit of a difference from the previous sections in this series. Military units, after all, that exist presumably have to have been recruited, supplied, organized and paid somehow or they wouldn’t exist. But there’s no reason a society cannot form up an army or a unit within an army that simply never coheres will and as a result performs poorly on the battlefield. Indeed, we’ve discussed both fictional and real world examples of poor cohesion or combat motivation in the past.
Now the student of history generally comes at these questions knowing the outcome: we typically know that an army either collapsed under pressure or cohered under pressure and are seeking to understand why. Each society’s system for generating cohesion is going to differ in some respects – indeed, you may well have different cohesive principles at play in different parts of the same army for different units raised differently – but a basic sense of the varying ways that cohesion can be generated can help one begin to work out why this or that historical army might have performed well or poorly, creating hypotheses to then investigate more closely in the evidence.
For worldbuilders, the task is somewhat inverted because the armies are fictional and their battles yet unfought. So instead the task is thinking about the societies one has built and the ways they could generate cohesion and giving them military and social institutions to match. Or, crucially, giving them military performance to match. So if your story is meant to feature a fearless, highly effective sort of army, you probably ought to think at least a bit about why they cohere together so well (and not think about it in just bland ‘because they’re super soldiers’ way). Alternately, if you want a military to surprisingly underform, think about flaws or weaknesses in its cohesion – perhaps the a Big Evil who hasn’t given enough thought to psychology or has simply assumed that he can rely on organic cohesion without having done anything to foster it, for instance.
Perhaps most importantly, remember that cohesion is a properly of the regular soldiers of an army, rather than their leaders. Highly effective armies combine capable leadership with high degrees of cohesion. But cohesion can turn against leaders as well – after all, a unit that has strong group bonds will be cohesive when facing the enemy, but it will also be cohesive in a mutiny.4
We’ll probably have a shorter fireside post next week, but after that, we’ll close out this series looking at some real and fictional examples of how all of these elements – civilian society, recruitment, supplies, organization, cohesion, leader and so on – fit together to make coherent military systems (or in some fictional examples, incoherent ones).
- Note especially the impact of Ardant du Picq’s, Études sur le combat (‘Battle Studies’ in English; 1870) and the problematic but influential S.L.A. Marshall, Men against Fire (1947). It is obligatory to note that the modern verdict on the latter is generally something like ‘S.L.A. Marshall‘s work is shoddy where it isn’t outright fabrication, but he happens to be right.’
- They were not the only ‘Pals Batallion’ so badly hurt; the Sheffield City Battalion was about as badly mauled.
- On this concept of collective synchronized discipline, wee W. E. Lee, Waging War (2016), 238ff for a good primer.
- Some of this comes out in D.M. Machado, Voluntas Militum: Community, Collective Action and Popular Power in the armies of the Middle Republic (300-100 BC) (2023) for Roman armies – the tight social cohesion and civic habits of Roman citizen-soldiers show up in their group agency and collective political power. The same is true of professional soldiers in early modern Europe, something that comes out in L. Staiano-Daniels’ The War People: A Social History of Common Soldiers during the Era of the Thirty Years War (2024) quite well, but equally crops up here and there in G. Parker, The Army of Flanders and the Spanish Road (2nd ed. 2004).
