This is the third part of a three part (I, II) look at some of the practical concerns of managing pre-industrial logistics. In our last post, we outlined what ‘foraging’ actually entailed – how armies got supplies both from friendly populations but also from neutral or hostile populations. In particular, we focused on the considerable damage that armies could not fully avoid doing (not that many tried very hard) when moving over the countryside. Because of their limited ability to transport supplies, armies were forced to draw those supplies locally, to the considerable misery of the local rural population. For the average peasant, an army came upon them like a rolling calamity.
I want you to keep that misery in mind because we’re going to shift perspectives here to the perspective of the general in charge of campaign operations, but in the back of your mind you should also be thinking about what the calculations the general is making are going to mean for the underlying populace that will have to endure his decisions.1 This week we’re going to look at those decisions and how the interaction of foraging and marching shapes where an army can go, how fast it can get there and how large it can be. For the most part, because of the limits of space, that is going to mean simply covering the generalities of these concerns (with some models to walk through the interactions); this post is already terribly long and in any event getting much more specific than this is going to essentially require looking at the specific details of individual campaigns.
And before we march onward, if you want to support my logistics, you can do so via Patreon. 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.
A Good Road
The first concern for the general is to determine the route of march and the options here might be surprisingly limited. Just how limited, I suspect, often comes as a surprise to many modern students of pre-industrial campaigns; these days when we travel we have GPS route-finders that usually give several different possible paths, from highways to back roads. Even in the days before ubiquitous GPS, with a good road map, you could chart dozens if not hundreds of possible routes in a long distance journey. In many games, this freedom of movement is greatly amplified: marching armies, imagined to be no longer restricted to roads, are effectively able to march anywhere not bounded by obvious natural barriers (e.g. the more recent Total War titles, Mount & Blade, and Battle Brothers). Roads may supply a speed bonus (often a quite limited one), but are otherwise optional. Historical armies were far more limited.
Armies almost generally had to move over longer distances via roads, for both logistical and pathfinding reasons. For logistics, while unencumbered humans can easily clamber over fences or small ridges or weave through forests, humans carrying heavy loads struggle to do this and pack animals absolutely cannot. Dense forests (especially old growth forests) are formidable obstacles for pack and draft animals, with a real risk of animals injuring themselves with unlucky footfalls. After all the donkey was originally a desert/savannah creature and horses evolved on the Eurasian Steppe; dense forest is a difficult, foreign terrain. But the rural terrain that would dominate most flat, arable land was little better: fields are often split by fences or hedgerows which need to be laboriously cleared (essentially making a path) to allow the work animals through. Adding wagons limits this further; pack mules can make use of narrow paths through forests or hills, but wagons pulled by draft animals require proper roads wide enough to accommodate them, flag enough that the heavy wagon doesn’t slide back and with a surface that will be at least somewhat kind on the wheels. That in turn in many cases restricts armies to significant roadways, ruling out things like farmer’s paths between fields or small informal roads between villages, though smaller screening, scouting or foraging forces could take these side roads.
(As an aside: one my enduring frustrations is the tendency of pre-modern strategy games to represent most flat areas as ‘plains’ of grassland often with a separate ‘farmland’ terrain type used only in areas of very dense settlement. But around most of the Mediterranean, most of the flat, cleared land at lower elevations would have been farmland (something you can see fairly clearly in the city maps I’ve included here), with all of the obstructions and complications that implies; rolling grasslands tend to be just that – uplands too hilly for farming.)
The other problem is pathfinding and geolocation. Figuring out where you off-road overland with just a (highly detailed) map and a compass is sufficiently difficult that it is a sport (Orienteering). Prior to 1300, armies in the broader Mediterranean world were likely to lack both; the compass (invented in China) arrives in the Mediterranean in the 1300s and detailed topographical maps of the sort that hikers today might rely on remained rare deep into the modern period, especially maps of large areas. Consequently it could be tricky to determine an army’s exact heading (sun position could give something approximate, of course) or position. Getting lost in unfamiliar territory was thus a very real hazard. Indeed, getting lost in familiar territory was a real hazard: Suetonius records that Julius Caesar, having encamped not far from the Rubicon got lost trying to find it, spent a whole night wandering trying to locate it (his goal being to make the politically decisive crossing with just a few close supporters in secrecy first before his army crossed). In the end he had to find a local guide to work his way back to it in the morning (Suet. Caes. 31.2). So to be clear: famed military genius Julius Caesar got lost trying to find a 50 mile long river only about 150 miles away from Rome when he tried to cut cross-country instead of over the roads.
Instead, armies relied on locals guides (be they friendly, bought or coerced) to help them find their way or figure out where they were on whatever maps they could get together. Locals in turn tend to navigate by landmarks and so are likely to guide the army along the paths and roads they themselves use to travel around the region. Which is all as well because the army needs to use the roads anyway and no one wants to get lost. The road and path network thus becomes a vital navigational aid: roads and paths both lead to settlements full of potential guides (to the next settlement) and because roads tend to connect large settlements and large settlements tend to be the objectives of military campaigns, the road system ‘points the way.’ Consequently, armies rarely strayed off of the road network and were in most cases effectively confined to it. Small parties might be sent out off of the road network from the main body, but the main body was ‘stuck’ on the roads.
That means the general does not have to cope with an infinitely wide range of maneuver possibilities but a spiderweb of possible pathways. Small, ‘flying columns’ without heavy baggage could use minor roads and pathways, but the main body of the army was likely to be confined to well-traveled routes connecting large settlements. But which routes to take? That leads to the next concern…
The general now has to pick the route where the key question is supply (particularly food supply) as we discussed earlier. Because of the ‘tyranny of the wagon equation,’ there needs to be sufficient food and fodder supplies along the army’s route to sustain it as it moves and accomplishes its objectives. Determining what routes are viable is thus a complex calculation (some of the variables of which will only be known to approximation), dependent on how fast the army will move, the local population density, the efficiency with which the army can forage or demand supplies from that population, the time of year, the army’s capacity to carry those supplies, how long the army needs to spend doing actual combat operations (like sieges) at the end of the march and most importantly how large the army is (and thus its need for supplies). Because the army can carry some supplies (frequently around two weeks) these variables mostly function as questions of the ‘rolling average’ which, if they fall below a certain level, will put the army into severe difficulties.
Let’s take our sample Roman consular army as a starting point: a relatively lean army with 19,200 combat effectives, 4,000 non-combatants and 9,800 animals which thus needs 61,850kg of food and hard fodder daily. Such an army might be able to forage 10% of the available food along their direction of march while still moving at speed; for now we’ll say the speed of march is 12 miles per day (we’ll come back to this in a moment, but that pace is not insanely fast but fairly brisk). With a fairly chunky cavalry detachment (2,400 cavalry) the army can probably forage out ten miles in either direction of its zone of march, so it sweeps an area close to 240 square miles per day. If we’re marching right around the harvest we might assume that each person in the countryside has something like 250kg of wheat or equivalent food, including fodder.2 Thus getting the daily 61,850kg required means each day’s 240 mile foraging zone needs to contain at least 618,500kg of grain or a population of at least around 10 per square mile.
Is that good? To get a sense of that we have to get a sense of what kinds of population density were normal in pre-industrial and pre-modern societies in the broader Mediterranean. Luuk de Ligt3 accepts population densities for Roman Italy overall between 70 and 95 people per square mile (his figures are in square kilometers so I am converting to keep consistency), but with tremendous variation inside of that, with dense regions like the ager Romanus and Campania around 110 people per square mile but places like Etruria and Umbria as low as 20. Italy, however, was one of the densest populated areas in Europe in the ancient and medieval periods (the other being the medieval Low Countries); rural population densities in Europe more broadly over large areas will have tended to range from 25 to 50 people per square mile, falling below that in rough upcountry regions (e.g. the population density of Scotland in the medieval period may have hovered between 10 and 15 people per square mile).
So right after the harvest our army of 19,200 effectives can go almost anywhere there is a road or good path and rural settlement. Great! Except of course the amount of food and fodder in the countryside isn’t static. Hay and oats – key sources of fodder for our animals – ripen and are harvested beginning in May, wheat comes in in June or July and barley in between the two. Consequently the available supplies (food and fodder) peak in those months and immediately begin falling. By October perhaps a quarter to a third of that supply is gone (plus by this point carefully stored seed grain is going back into the ground!); suddenly we’re looking at something like only 175kg of food supplies per person in the countryside, which means our army needs a population density of around 15 per square mile. Worse yet, if we plan to winter in the field (that is, we’re not dismissing the army home), we have to think about stockpiling for the conditions in January and February when forage becomes very scarce indeed. Realistically to do that, even with our fairly compact army, we need a population density of 20-25 people per square mile in order to build up winter stockpiles to last through the lean months. At these levels, low density areas with lots of pastoralism and not a lot of towns are effectively off-limits: the army would have to fall back to denser farmland for the winter. And indeed, we see this exact behavior in many ancient armies that stayed in the fields year round that had to fight in less densely settled regions – they campaign in the restless uplands and forests but always with an eye towards dropping back to winter quarters in densely settled river valleys or coasts.
But we’ve made some fairly favorable assumptions about our army: it is fairly fast and fairly compact. We’ll deal with the impact of speed in a moment, but what if our army is bigger? Well, this is initially fairly simple if we assume a constant movement speed: double the size of the army and that doubles the food and fodder requirements which requires double the population density. So a double consular army of 38,400 effectives (plus 8,000 non-combatants and 19,600 animals) requires 123,700kg of food and fodder and thus a post-harvest population density of 20 per square mile over the campaign area, rising to around 40 if the army doesn’t want to have to disperse in the fall. That army suddenly has some fairly severe limits on where it can go outside of the couple of months right after the harvest; routes that run through less densely peopled territory are no longer options for that army.
But what if, instead of doubling the size of our fairly compact army, we assume an army whose non-combatant load looks something more like the Army of Flanders, where the non-combatants of the campaign community might number around half as many as the combatants (and in some cases equal their numbers). Suddenly our 19,200 effectives army now has roughly 30,000 mouths to feed; managing the baggage of the non-combatants probably adds a few animals too, perhaps bringing us up modestly to 12,000 or so. Now we need 77,500kg of food and fodder and thus summer population density of 13 people per square mile and a year round density of 26. Doubling the size of this army doubles those figures, requiring a harvest density of 26 and a year round density of 52 – such an army would either be limited to campaigning in only densely populated corridors or it would have to march divided to sweep a larger area of foraging. As hard as it is to exclude non-combatants from an army, they inflict much the same logistical load as combatants, limiting the operational capabilities of the army to which they are attached.
But that treats the size of our army and the speed of its march as independent variables. Unfortunately for our army, they are not independent variables, rather (with many other factors considered) big armies move slower. The problem is…
We’ve dealt with this problem before from other angles, but armies do not move as a theoretical ‘point mass’ like in a physics problem (or how they are represented in many video games), rather their elements take up road space, one after the other. An army on the march is thus not a battle line or a compact mass but a long, snaking series of columns. That in turn has substantial impacts on its rate of movement, because the front of the army has to march past the back at the beginning of the march and the reverse has to happen at the end.
So how much space might an army on the march take up? I’m going to be drawing some numbers here from J.A. Moss, Manual of Military Training (1914), because I happen to have a (well used) copy of the 1917 edition. It calls for a rank distance (the distance between the back of the man in front and the chest of the man behind him) of 40 inches, which is fairly standard; adding in the width of the soldier themselves and their pack each rank takes up around 1.25m (c. 4-5 feet) of road space. Formed six soldiers wide (which will, with minimal intervals, get us to the width of most single-carriage roadways, which what we are likely to be dealing with), our model army’s infantry component (16,800) would have 2,800 ranks to it moving in column.
But we can’t just march the army in one single mass like that. Intervals between units are necessary for control (so that officers can keep an eye on their whole units and ensure soldiers aren’t wandering off, straggling, injured, sick, etc.) and also to avoid ‘accordioning’ into each other as units ‘tailed out.’ Moss (1917) assumes intervals but does not specify them; Lockwood, Organization and Tactics (1895) offers the rule of thumb of allowing an extra 10% road space for a battalion, 15% for a division and 20% for an army corps. So instead of one infantry body of 16,800 men, we can divide it into 35 cohort-sized ‘battalions’ of 480 men each. Each of these battalions thus has 80 ranks taking up 100m of road space, plus an extra 10m for a unit interval to the next unit. Our infantry thus take up 3,850m or 2.4 miles of road space.
Cavalry takes up more space because horses are big. Lockwood offers that each individual horse takes up a depth of 3 yards and a front of around 1 yard (compared to 28 inches for an infantryman), so the cavalry might ride in column four abreast instead of six. Once again we also need unit divisions; for the sake of simplicity we’ll use a rough ‘squadron’ of 100 cavalry (a touch bigger than a typical cavalry squadron but there is a lot of variation here too: the Romans divided their cavalry into turmae of 30 each, which were quite small units), which is going to form 25 ranks of four cavalry each requiring 2.75m of road length (about three yards), so 68.75m plus our interval (10%) makes roughly 75m per squadron. With 24 of these squadrons to make our 2,400 cavalry they take up some 1,800m of road space, or 1.1 miles.
And then we need the baggage, which here is being moved by 5,000 mules and 2,400 spare horses. The mules will be a bit more compact than the horses per animal, but we’re still probably looking at something like 2.25m per mule (and 2.75m per horse) set four abreast on the road. That’s 1,250 ranks of mules (2,812.5m) and 600 ranks of horses (1,650m) for a total of 4,462.5m taken up by the baggage train. In practice while a some of that will be in a large mass in the center of the army (see below), a lot of these animals are going to be split up into small groups at the tail of their units (each infantry ‘battalion’ should have 80 mules – 45m of them – at its tail). With so many possible configurations, it seems tricky to try to calculate open intervals for the baggage train; perhaps (in small bits spread over the army) they might add something like 440m minimum, so their total road length is just short of 5,000m (3.1 miles). As for the non-combatants, our army is quite lean and many of those non-combatants are in practice guiding the mules and the horses; we can include them here for now.
So that’s our army, a snake-like beast 6.6 miles (10,650m) long, winding through the landscape. Now we can think about…
Unladen human walking speed on flat terrain with long strides is just over 3mph; it will not surprise you to learn that for a large army marching more precision is required. Going back to Moss (1917), he assumes the length of step of the average man is 30 inches (in practice, keeping up with standard march rates and thus standardizing gait is something soldiers will learn by training and experience) and one of the chief jobs of a commander on the march is determining the correct ‘rate of march’ – Moss even recommends the unit commander (at the front since the front unit sets the pace for everyone behind) time the pace every so often to maintain a steady marching pace. The standard paces for marching under loads are between 88 and 110 steps per minute, which come out to 2.5 to 3mph.
The army does not, however, move continuously at this speed. Moss (1917) sets out the guidelines for rest: a 15 minute half after the first three-quarters of an hour (with an effort to do this away from settlements), followed by a 10 minute halt every 50 minutes (adjusted to 45 minutes march and 15 minute halt in tropical or arid climates). In very hot environments, Moss (1917) notes you ought to also halt for three or four hours during the heat of the day (so around 1-3pm in most cases), but we’ll assume we’re not marching in that hot of a climate. But that means our formation, humming along at 3mph is actually only moving every 5 minutes out of 6 or 3 minutes out of 4; that means our actual average speed is 2.25-2.5mph, even though when moving we achieve a 3mph march.
But we also do not move for eight straight hours at this speed seven days a week. Marching soldiers under load straight through the whole day will, understandably, exhaust them; this may be necessary in some cases (where it is called a ‘forced’ march), but it is rarely desirable. After all, the army is supposed to fight at the end of this and has to reach that destination ready for the exertion of battle or siege. Moss (1917) recommends that “as a rule, foot troops do not start before broad day-light; mounted troops, when practicable, about an hour after broad daylight” (1219), in part to avoid losing the valuable rest of the early morning hours and in part because trying to move before the sun is fully up and visibility is best is silly. Now that’s obviously a time that is going to vary a fair bit by latitude and time of year, with anywhere from 10-15 hours of proper daylight to work with.
So let’s say we are on the Mediterranean coast of Europe (I’m picking Marseilles for my times here) in early May. The ‘false dawn‘ is going to start around 4:30 and we’ll have enough light to actually do things by around 5:30 or so; remember options for artificial light here are very limited (laterns and campfires; torches4 are more common in fiction than actual use) so we need actual sunlight to do much of anything. And there are things to do! Soldiers need to wake up, eat breakfast, handle personal hygiene and dress, break down their tents and the camp and fill canteens (the necessity of filling canteens with water in the morning is something I have seen in every infantry marching manual I have ever looked at for obvious reasons), all of which takes time. In the United States army, ‘reveille’ – when soldiers can be expected to actually fall in with units – is set at either 6 or 7 am and its easy to see how this fits as that’s almost precisely the ‘broad daylight’ of marching, while using the dawn hours to get ready. Cavalry needs an extra hour because, of course, the horses.
So it’s 7am and we start marching! Well, not quite. Someone starts marching. But the back of the army now has to wait for the entire rest of the army to march past it in sequence. That’s going to take time! Recall our army is 6.6 miles long and moves at 2.5mph; it’s going to take a bit more than two and a half hours before the last elements of the army begin leaving the camp in ideal conditions (this factor shows up a bit less in 1800s manuals because by then it was very common for armies to march divided; this was very rare for earlier armies). So it will probably be almost 10am before the entire army is moving. But now we have our problem in reverse form: the front of the army must stop and set the camp somewhere the back of the army can reach before the sun goes down. Night marches, as Moss (1917) is noting even in the 1900s, are difficult and perilous. So while the sun starts to set around 8:30 or so, we’re going to lose that two and a half hours again on the back end; the front of the army must stop marching around 6pm at least so the rear of the army has time to catch up before sunset.
In practice it is yet worse than this because of course we need to set up the camp again at the other end of the march, which is also going to eat precious time. And while the day gets longer in the summer, soldiers do not magically become supermen of endless endurance in the summer (if anything, as the heat rises, the reverse is true: marches must become shorter) meaning that they need time not merely to set up camp, but also rest, cook food, attend again to hygiene, the maintenance of weapons and equipment and so on. You can get more marching hours out of the day but only by eating up those rest hours which the army needs to be combat fit. This is why, as Moss (1917) notes, “forced marches seriously impair the fighting power of even the best troops” (1225).
So we have perhaps a standard marching day: the army wakes around 5:30-6am. The front is formed up at 7am and properly starts moving (after scouts and screens head out) at 7:30 or 8am, the back starts marching at 10am in the best case; the army halts around noon to eat (long marches on empty stomachs are a recipe for fatigue casualties; I cannot stress enough that while physical fitness and training obviously matter in all of this, heat, dehydration, hunger and exhaustion do not care how many reps you can do and the general has to get his whole army to the battlefield ready to fight). The back of the army probably needs to be making camp by 5:30pm or so (to have time to setup the camp – especially if it is a fortified camp – before nightfall), so that means the front of the army has to stop around 3pm.
So while some part of the army is probably moving at every point between 7:30am and 5:30pm, the actual effective marching time – initially 10 hours minus one for lunch – is reduced by 2.5 to just 6 or 7 hours if absolutely everything went right. At c. 2.5mph average, our army would make 15-17.5 miles in a day. But of course everything isn’t always going to go right; if it’s hot (in summer, recall) our speed drops to 2.25mph and our distance in a day to 13.5 to 15.75 miles. If the roads are bad, we lose yet more time, especially if the roads narrow at any point, forcing the army to stretch out even more. I want to stress we are still mostly working with best-case situations, but we’re already looking at figures right around the 12-15 miles per day that most rules of thumb consider the upper end of long-term sustainable marching base for infantry moving along good roads.
Now let’s fiddle with our assumptions a bit. What if the army carries a lot more baggage? This could be because the army needs to carry more supplies to get past areas where forage options are limited (either by low population or by the presence of enemy forces) or because the army is having to move a large siege train. If we, for instance, remove our mules (so save 2,812.5m of road space, but lose c. 650,000kg of carry capacity) and replace them with 700 heavy wagons (c. 1,000kg carrying capacity per), we can expand our logistics (because we’ve dropped from 5,000 mules to perhaps 2,800 horses and likewise the number of required non-combatants to handle the mules has gone from 3,400 to just 700 wagoneers) and have some space for a heavy siege train. At best we might cram each wagon into something like 12m of road space (American Civil War rules of thumb was 80 wagons to a mile, or each wagon taking up 66 feet or 20.1m), so our new wagon train is now 8,400m long. That’s 5.2 miles of baggage alone, so now our army is a stunning 16,237.5m long (10 miles). In very good roads we might double up the wagons and get down to just 12,037.5m or (7.5 miles).
Which in turn means our ‘lost marching time’ extends from just 2.5 hours to 3-4 hours (be careful not to double-count the loss, subtract either from the front or the back to get the correct number of marching hours, not both), leaving each unit with just 5-6 hours of actual marching, reducing our daily rate of march to between 11.125 and 15 miles per day, under ideal conditions if everything goes right and not one of those seven hundred wagons breaks down (of course the same was true of having not one of 5,000 mules become injured or stubborn as mules are wont).
Likewise, if we take our original army and double its size, either by adding more combatants or by loading it up with more non-combatants (now so numerous that they are not all simply handling the mules but are filling up even more space in the line), that doubles its road length demand out to 13.2 miles. That means our ‘lost marching time’ is suddenly 5-6 hours leaving just 3-4 hours of total effective marching time. Suddenly our army is crawling at between 6.75 and 10 miles per day – under best case conditions! – even though our ‘start’ and ‘stop’ times have not changed at all.
In practice we’re actually still being a bit too nice to our army: we’re assuming they march every day. But while humans can more or less do that, pack and draft animals cannot. In practice for a sustained march these animals need to be rested one day out of six or so or the strain will render them useless in the long term. So our lean, compact army is really moving around 12.5-14.5 miles per day over a six-day average with one day of rest under normal conditions (a bit faster if they push it, a bit slower if they don’t) and our double-sized army could be moving as slow as 5.6-8.3 miles per day.
The key observation here is that everything you add to the army slows it down. A good general has some options to mitigate this (but they all come with risk), but the fundamental rule remains. Adding more baggage to carry artillery or supplies slows the army down. Adding more soldiers to try to improve your chances of victory slows the army down. Adding more non-combatants to the ‘campaign community’ slows the army down, though remember from the earlier two posts, depending on the organization of the army and the expectations of the soldiers, some significant number of non-combatants may be effectively unavoidable.
Sink or Swim
And now that we have our range we can return to some of our independent variables and see how they collide. Recall that during the summer our compact, mid-sized 19,200 soldier army could go anywhere where the population density was at least 10 or so people per square miles.
Now let’s see what happens when we double the size of that army. First, it’s food needs, as we’ve noted, double from 61,850kg to 123,700kg of food and fodder, so the number of people it needs to forage in a day (at a 10% forage rate) goes up from 2,474 to 4,984. And you are thinking, “Yes, we’ve done this math and we know this ends up with them needing around 20-25 people per square mile after the harvest.” Ah, but our forage area has also changed.
Because our army isn’t clipping along at 12 miles per day anymore, it’s slowed down to an average of perhaps 7 miles per day. That in turn means our foraging zone isn’t 20 miles by 12 miles, but 20 miles by 7 miles: we’ve gone from 240 square miles of forage to just 140. So our required population density has actually leap from just over 10 to 35.6 people per square mile, more than three times as high because the increased nutrition demand and decreased speed have compounded. And that is, you will recall, the situation immediately following the harvest. By October that figure is going to have risen to something like 46 people per square mile; the year-round campaigning requirement would be close to 70 people per square mile.
Those aren’t impossible numbers, but they are going to pose the general of this army with some difficult choices. What will not work is overland external supply with more wagons; as discussed the animals eat food and so adding more wagons just slows the army down more and increases its food demands; that’s the tyranny of the wagon equation at work – the army cannot wagon its way out of a foraging pit. One option is to forage more intensively to raise the percentage of food in the countryside the army is pulling in; this poses twin problems. The first is that doing so generally means either slowing down or dispersing foraging parties for more effective searches; the former may be counter-productive, the latter may be dangerous. In either case, foraging beyond 10% is likely to cause fairly immediate and sharp starvation in the countryside; pushing that number higher is going to risk invoking Marshal Villars’ warning from last time, “If you burn, if you drive out the population, you will starve.” Not today, but tomorrow; if the army has to return home through this same route or intends to operate in this area again soon, depopulating it by over-foraging merely delays disaster. Nevertheless, faced with the exigencies of war, many armies did just this, ruining the local population to enable the movement of oversized armies.
Another option is to speed up, thus increasing the foraging area. The march could be lengthened, but that is going to have negative impacts not merely on morale but on combat readiness: after days of regular long marches under heavy loads soldiers will begin to suffer from exhaustion. Instead many generals opt to trim the baggage, trying to get rid of every sort of excess thing; this is common enough that it becomes a sign of good generalship in the ancient world – the good general enforces discipline, compels his soldiers to get rid of everything they carry that isn’t strictly necessary, maintains morale by doing the same himself and then for good measure kicks every unnecessary non-combatant out of the camp. Unfortunately for our general, we’ve already started with a very lean army, by pre-industrial standards, so there isn’t very much fat to trim (which is partly why most actual armies, which were not so lean, marched at speeds that tend to cluster at the bottom of these ranges we’re creating; the other big reason why is we’re assuming pretty good road conditions which will not have pertained generally). However a general can also push an army by lengthening the marching day and raising the pace; doing this for long periods will wear out an army quickly, but for quick, short-term solutions to supply problems (or for rapid operational maneuvering) there is a fair bit of flexibility in daily marching rate. Morale ends up as a major consideration here: motivated soldiers (either inspired by their general or their cause) may push harder on the march, allowing the army to sustain a higher rate of movement, sticking closer to the 10 or 12 miles per day and thus alleviating some of the supply issues.
A third, more complex option is to march divided, splitting up the army to move in several separate columns along different roads which can both move faster (because they are now two smaller armies) and can forage different farmers in different areas. The problem this poses is simple: coordination. An equal sized enemy who remains concentrated could engage one of those armies, win easily due to the 2-to-1 numbers advantage, then turn and obliterate the second army, what is known as ‘defeat in detail.’ To avoid that the general in overall command has to coordinate so that these armies march separately but arrive at the battlefield at the same time. This sounds easier than it is: the general may not know where the battle will be (because he does not know where the enemy army is) and he has to plan (at least) two routes of march which will get both armies to the critical space with a margin for error measured in hours at the end of a march which may be days or weeks (or months!) long. This can be done, but it is very difficult and very easy to completely ruin a campaign with a fairly minor mistake; in practice before 1800 or so, most armies only operated like this if they absolutely had to.5 In practice the threshold where armies end up forced to split up is somewhere above 50,000 effectives, depending on conditions.
Finally there is a fourth option available in only certain circumstances: naval (or river) supply. This is, all things considered, somewhat rare; most ancient and medieval polities were not up to the task of supplying large armies by sea regularly at long distance (though the Achaemenids, Athens, Carthage and Rome all do this at one point or other and if you are thinking, “gee, those seem like some of the highest state-capacity polities in the ancient world” why yes, yes they do.).6 Nevertheless ships and river barges offer an escape from the tyranny of the wagon equation: the food consumption of the crew is effectively trivial compared to the amount that they can move. The result is that an army at a decent port with access to the sea (directly or by navigable river) has access to effectively infinite supply throughput, assuming that the supplies can be assembled and safely shipped. The problem that most pre-modern polities would struggle to do that is a separate issue, but a significant one. One of these days we may come back and talk about Roman naval and riverine supply in more detail.
Underlying all of this is a central fact about armies before railroads: armies are like sharks, they must swim or they will sink.
Now Add Hostile Armies
Now so far we’ve been considering our army moving through a countryside devoid of actual resistance, but of course the enemy gets a vote! Foraging in enemy territory is a dangerous activity; troops that are searching farmhouses or hauling large amounts of supplies (or pillage) aren’t going to be in fighting trim. In many cases they may not even have their weapons and armor (e.g. Liv. 31.2.8; Caes. BG 4.32.5), since working light means being able to shift larger amounts of forage and pillage. Moreover as we’ve discussed getting an army into fighting formation quickly is hard; if your unit is foraging in a village and an enemy unit suddenly attacks, the chances of effectively defense are very low indeed. That’s why foraging parties had to be so large, so that some of the foragers could be on guard while others were doing the actual foraging. The Roman historian Livy has an interesting description of command failures leading to an ambushed foraging party, noting that the commander, Gaius Ampius, had failed both to sufficiently scout the area but also setting up sufficient strong-points from whence armed soldiers could protect and screen the unarmed ones (Liv. 31.2.8).
Of course those sort of arrangements are only going to work against small parties of enemies (like their own foragers and scouts); against the main force of an enemy army those scouts and guards really function just to give the army enough time to retreat and concentrate for a battle. Because of that ‘defeat in detail’ threat, if the enemy’s main field army is nearby and close enough in strength to offer battle, it becomes necessary to keep your own army concentrated; even if you don’t intend to offer battle as situations like that at Pydna show, battle may happen regardless. But of course as we discussed back then, generals here have at best limited and often badly out of date information about the composition of location of enemy forces; a general thus has to balance his need to move and feed his army with the knowledge that the enemy could show up at almost any time.
That starts, of course, with marching formations. We’ve described how it takes time for an army to get moving in the morning, but there’s an additional complication to this: the order of march isn’t itself random. Exact formations vary but there was a fairly standard setup: proceeding the army was often a screening force of scouts (often cavalry), followed by the vanguard (or ‘van’), a combat unit or several marching some distance in advance of the main force, often with minimal baggage. The general himself often moved with the vanguard because that’s where information would be quickest come by. Then came the main body of the army, typically with the baggage train in the middle sandwiched between combat elements (which typically composed the bulk of the army’s infantry). Those combat units might also (as above) have their own smaller baggage trains directly behind them in the marching order or it might all be centralized in the center of the army (the latter was the Roman practice). Finally there was a rearguard, more or less a mirror of the vanguard. That layout ensured the army could respond relatively rapidly to the sudden appearance of an enemy in front or behind. For extra security, light infantry skirmishers with minimal baggage might also be detached as ‘flankers’ moving off the road or along side roads so that the army couldn’t be surprised from those directions either.
Typical advice (this actually comes up in Moss (1917)) is that a commander needs to both rotate which units are in the vanguard and rearguard and that his subordinates need to rotate which part of their command is at the front and back of their units, regulating the pace. There are many silly things in the film Gettysburg (1993), but one moment I just love is Chamberlain’s exasperation on realizing on top of an already difficult morning that his regiment has been assigned to the front of the front brigade in the corps and so has to assign flankers; that rotation would have been regular in most armies. And what would have happened to create this scene, is that the commander of the V Corps (Major General George Sykes) opted to march with his first division in the front, which then its commander (Brigadier General James Barnes) opted to put the 3rd Brigade in front and its commander (Col. Strong Vincent) opted to put the 20th Maine in front and of course its commander (our Professor of Rhetoric, Joshua L. Chamberlain) would in turn have to pick where each of his companies would go. The main concerns for smaller units is that the front part of the unit regulates pace and the rear part of the unit will need to deal with stragglers (including heat and fatigue casualties) as well as keep up pace to avoid the unit ‘tailing out’ too badly as it marches.
Foraging opportunities would also be shaped by the enemy presence and this could be a crucial operational concern. Because foraging operations were so vulnerable the presence of a large enemy force could prevent them even if it didn’t give battle. This is often what is happening when one commander ‘bottles up’ another without ever giving battle – by simply saying close to but out of reach of an enemy army, the weaker or more timid force can massively complicate their logistics. Roman efforts against Hannibal in southern Italy after Cannae (so from 215-203) as a master-piece of this kind of maneuvering. The Romans after Cannae had accepted that further direct engagements against Hannibal were unwise, but on familiar ground and close to their own bases of supply, Roman armies could operate freely where Hannibal needed to forage to advance. By simply staying close to him Roman armies could restrict his ability to forage and thus his ability to maneuver at all, forcing him to remain close to his base of support in southern Italy.7 Of course such a strategy only prolongs the conflict, but the Romans had other armies that were chipping away at every other corner of Carthage’s empire.8
Armies in close proximity might still manage to forage by sending out very large foraging parties (e.g. App. Pun. 100, Caes. BAfr. 68; Bohemund’s large foraging party that engages Duquq of Damascus during the Siege of Antioch (1097-8) also seems to be an example) or by pointedly sending their foraging parties away from the enemy force (as with Antony and Octavian’s foraging efforts between the two engagements at Philippi in 42, App. BCiv. 4.122; given the size of the armies this must have been a desperation move to extend their operational endurance; they could not have been permanently supplied on that basis). Still at best the presence of an enemy army might cut the available foraging territory of an army on the march in half, which for armies of even a fairly modest size is going to push the required population above sustainable levels quickly. Of course that is happening to both armies, which is part of why ancient and medieval generals are so confident they can force their enemies to battle once the armies get close: both armies, unable to forage effectively, have ticking clocks in most cases. Exceptions do happen (especially when one army has naval supply), but for the most part the supply situation will eventually compel each army to either fight or retreat.
Absent a full field army, a fortified site (like a castle) with a strong cavalry detachment could achieve similar operational results in restricting enemy forage opportunities. Indeed, as we’ve discussed, this is how castles worked at the operational level: the presence of troops (specifically mounted troops) in the castle made effective foraging or administration of the land around it impossible for any invading force that didn’t also lay siege to the castle. Consequently an army looking to maneuver in the region had to neutralize the castle first or else swiftly march past it into less well-defended farmlands that could be foraged effectively. Fortified cities and hill-top towns (oppida) posed similar problems for Alexander the Great and Caesar respectively: if captured or made friendly (either by alliance or preemptive surrender) they were sources of supply, but such a settlement untaken could create supply deadzones and serious risks in the rear of the army. Indeed the emperor Julian gets himself into precisely this problem with an audacious march into Persia without fully securing the rear (Ctesiphon being too well defended), ending up in a situation where the army could neither safely advance nor retreat. Consequently even when the road was open there was often no alternative but to besiege and take these strong points.
All of which at last lines up most of the most pressing concerns a commanding general has when considering the march of an army. Of course the general does not have some of this information to the precision we’ve modeled it here: he does not know the exact population density of the regions he is going to march through and instead is probably thinking in analog terms: some areas are ‘rich’ some are ‘sparsely peopled’ and so on (which is to say a wise general is going to want a pretty big ‘margin of error’ on those daily foraging calculations); likewise the expected rate of march over a long period must on some level be an informed guess rather than an exact calculation.
Still, the general has to first select an objective (probably the capture of a major civilian center which will bring with it the countryside it administers), then figure a route of march to get there. He probably has a decent sense of how much food and fodder his army will need but at best an imperfect sense of what is going to be available in enemy territory or where enemy armies might be; that will demand improvisation on the march in response to changes in the army’s food supply. The army’s rate of movement is variable, but while pushing hard can solve logistics problems, it can also exhaust soldiers, deplete morale or involve leaving heavy (but important) baggage behind; marching fast to defeat in battle is a mistake many commanders make (e.g. Marcus Antonius (‘Antony’) at Mutina in 43, Crassus at Carrhae in 53, Antony against Parthia in 36, Harold Godwinson at Hastings in 1066, Antony at Actium in 31. Antony does this a lot. I confess I have never quite understood his reputation among some historians as a ‘great captain.’).
Indeed it isn’t uncommon to see skilled generals employing a variety of solutions in response to conditions at various points in a campaign. Once again, the reader who wants a vivid example would do well to simply read Caesar’s Commentarii de Bello Gallico. Caesar makes a dynamic mix of solutions to supply and movement difficulties, especially since the terrain of Gaul largely cut Caesar off from Mediterranean sources of supply beyond the very beginning of his campaigns. Caesar often divides his army to accomplish multiple objectives or to move quickly (although this comes very close to disaster, e.g. Caes. BG 6.32-42 where Caesar splits his army up and very nearly loses his camp as a result) and as noted always seems to be arriving at rich farm country or prepared stockpiles of supplies just as his own is running out (which in turn is allowing him to move light with minimal baggage and thus move fast, Caes. BG 1.23, 1.34, 1.48, etc). Honestly a good, up-to-date logistics-and-campaign history of Caesar in Gaul really is desideratum.9 At the same time, Caesar’s motivational ability clearly also plays a role: both his ability to get his soldiers to both maintain discipline and push themselves on campaign but also his ability to retain their confidence even during marches where food supplies might seem dangerously low (though it’s not clear if he let them know that; Caesar showed special attention to the days set for regular grain ration distribution possibly for this reason – knowing that missing a day might not cause instant starvation but would cause his soldiers to lose certainty in his leadership).
On the other hand, of course, in a contest against another other thinking humans, victory often depends on riding the line of failure as tightly as possible without going over; Caesar repeatedly comes within sight of logistics disaster when making rapid movements – his famed celeritas (‘speed’) – but always lands on sufficient supplies at the last minute. As you might imagine from how the baggage train slows an army down, a ‘flying column’ with just a few days food can move very quickly, but of course must achieve its objective quickly and return. Once again the element of calculated risk here is very high; when Alexander learned that Darius had been seized by Bessus (330BC) he launches himself in that direction with a flying column with just two days supplies (hoping to capture Darius, Arr. Anab. 3.21.6-10; on this see Engels op. cit. 83). Had functionally anything gone wrong (except for the one thing that did go wrong, which was Darius died before Alexander got to him), Alexander would have been in dire straits indeed.
Caesar was, of course, uncommonly good at this (with Dyrrhachium in 48BC being the one very notable time things don’t quite work out), but that disguises the complex web of calculation, map-reading (or scout-reading) and guesswork that goes into it. Most Roman generals, lacking Caesar’s skill, tended towards more careful and methodical campaigns which were the slow road to a certain victory; one of Rome’s notable advantages was the regularity with which it produced reliable if unexceptional generals who managed the difficulties of logistics and movement fine, if not exceptionally. It takes only a brief glance at the difficulties of Marcus Antonius, Caesar’s protegee, to see why this was valuable: Antony tries again and again to pull off bold, Caesar-like campaigns and really only succeeds once (Philippi in 42). Antony has a reputation in the sources as bull-headed and impulsive in contrast to Caesar’s bold but calculating nature and it really comes out in their campaign abilities; Caesar’s dice always seem to come up just right (too many times to assign to pure chance) whereas Antony takes similar risks and more frequently fails. The difference between their decisions was often very small (indeed, in many cases Antony appears to be taking the exact same risks) but at the ragged edge of the possible in logistics, a slight miscalculation can bring catastrophe.
In the end, I think there is some good sense to Clausewitz’ argument (drink!) that warfare cannot be boiled down to mere numbers – there being too many unknowns – and that what separates a serviceable general from an excellent one is the ineffable quality of being able to instinctively take in the situation and make the right risks, what Clausewitz terms ‘genius.’ At the same time I cannot help but note that for all of the flashiness of Caesar, the Romans rarely had brilliant genius; what they had a lot of was workmanlike generals whose military-political careers (those being conjoined in the Roman world) had prepared them for the difficult administrative task of making the army go, of managing marches, supplies, foraging and so on. Roman elites serves after all first as military tribunes (who frequently led foraging parties), then as quaestors (who managed supplies and pay),10 then later as praetors in command of smaller armies before being consuls in command of big armies (and of course the fact that all of these Roman elites also managed large estates with huge – frequently enslaved – workforces must also have prepared them). A steady supply of solidly adequate generals backed up by a superior military system – mostly superior in mobilization rather than actual fighting – was enough.
Meanwhile on the other end of the spectrum you have generals whose over-sized armies seem to wallow, practically paralyzed by the herculean effort to merely get them moving. As a younger student, I always found myself puzzled by this – how hard can it be to just march in a direction? Is the general just lazy? But what I hope you can see in this series is that there is actually quite a lot going on to keep a large army moving. A mediocre general (or a general with medicore subordinates or staff) with a very large army may find themselves overwhelmed by the coordination demands it imposes – who marches where? When? On what roads? How do you make sure they don’t run into each other and that there is enough food locally for each group? Who takes the front? At what rate of march? And what do you do when invariably units that you cannot directly observe (because the army is very big) move slower than you need or get lost? On top of this the general has to be managing the direction of march and also the supplies of food as well as the deployment of foraging parties. The number and complexity of these moving pieces only increases as the army gets larger, while the hours in the day for actually marching decreases as the army train lengthens, and so the army just crawls along, paralyzed by its own bulk.
Consequently, going back to where we began this series, while the general of a pre-industrial army has at best only limited control over how it fights in battle, his control over the army in the marches before the battle is much more extensive and coordination in this phase can shape the battle or siege to come. Because foraging and supply concerns are directly connected to army size, a military system (or individual commander) that is better at managing these issues can push larger armies further out, opening up new opportunities in the campaign space. By contrast a system (or general) that is poor at this kind of thing will be forced to accept either a smaller army or else may end extremely limited in options: either unable travel certain key routes with insufficient supplies or population or else functionally paralyzed by the size of the army itself.
Of course related to this the general mismatch between the size of armies and the states that fielded them in Europe during the early modern period (esp. 1500-1700 or so) made those armies increasingly destructive. As the transition to gunpowder warfare pressured states to deploy larger infantry contingents in their armies, the logistics systems which had worked in the late Middle Ages buckled under the strain, leaving the soldiers in question to supply themselves locally, with the catastrophic results we discussed in part II. Consequently while the increased effectiveness of these large armies was clearly welcomed by their rulers, by the mid-1600s it obviously became clear to European monarchs and their generals that the destructiveness of these forces had become an intolerable hindrance, leading to the (re)establishment of centralized state logistics in Europe which would in turn both motivate and feed off of increasing administrative capacity which in turn leads towards the establishment of the modern administrative state.
Finally, while this last post has been mostly focused on this question once again from the perspective of the general, we must also keep in mind that even well-run armies caused substantial misery among the rural population as they moved. Even an army not trying to cause damage, by pulling in such a massive amount of the local surplus, could push local communities into hardship. At some point we may come back to actual strategies of devastation apart from foraging, but of course if armies could push peasants to the brink merely by being around, they could do even worse if they actually intended to cause damage (though ironically the upper-limit to this sort of damage was often limited by the small size of these armies; administrative capacity cuts both ways and the very things that have made more modern armies capable of causing less damage merely by existing have made them more capable of causing much more damage by intent).
The ‘background noise’ of pre-industrial warfare was thus looted barns and crowded roads; the army had to do these tasks in order to sustain itself on the march, the general had to coordinate these tasks in order to keep the army moving and the rural population had to endure these tasks in order to survive.
- I should note, I am going to refer here to our imaginary general with masculine pronouns. Women commanding armies in this period in the broader Mediterranean was not entirely unknown but it was exceedingly rare; even in most societies where women might rule in their own name it was very common for army command in the field to be delegated to a male subordinate (women commanding the defense in the event of a siege was more common, though still rare). Again exceptions exist: Artemisia of Caria, Aisha at the Battle of the Camel, Matilda of Tuscany, etc., but only a relative handful over a very long period with very many army commanders. So it seems reasonable to use masculine pronouns for an occupation that was in this period performed overwhelmingly by males, though that does not erase the small number of women who did command armies, thus this footnote.
- Lots of adjustments here: they need 365-person-days of grain to survive per year, but there will also be fodder for animals and some small surplus for sale. On the other hand, the nutritional needs of an average soldier are much higher than the nutrition needs of the average rural person – which after all includes women and children with lower calorie demands (high activity adult male might require upwards of 3,500 calories per day where children, the elderly an less active folks might be below 2,000). I’m working my numbers from a model of a smallholding Roman farming family that I developed for my current book project.
- In Peasants, Citizens and Soldiers (2012)
- Of the actual flaming kind, not British flashlights
- Unsurprisingly, one of the few ancient Mediterranean polities to do this with any regularity and success were the Romans. This comes out very clearly in Julius Caesar’s campaigns in Gaul and also in the lead up to the Battle of Philippi, where in the latter case the armies were absolutely too large to march as a single formation. That said the Romans still only did this from time to time; for the far more common use of these kinds of operations in the pre-modern period one must look to the Steppe.
- “Why is Athens here?” So the Sicilian Expedition is a really big (and very disastrous) overseas operation, but seems to have dealt with supplies locally. The reports of reinforcements include men, horses and money, but not massive amounts of food. In practice the Athenian army was of a size that could be supported on the relatively rich grain-lands of Sicily. On the flip side, one might argue that provisioning the city of Athens substantially by sea during the Peloponnesian War is a feat sufficient to merit inclusion in this list and I tend to agree; thus Athens’ inclusion.
- This effort is treated in detail in P. Erdkamp’s masterful Hunger and the Sword (1998), which is, alas, very expensive.
- Which matters, to be clear: soldiers need a ‘theory of victory’ to sustain morale. ‘We bottle up Hannibal here until everyone involved dies of old age’ is not going to inspire anyone to put up with hardship, but ‘We hold Hannibal here while Scipio and his army sock it to the Carthaginians in Spain and Flaccus and his mates slowly reconquer southern Italy’ is much better.
- But this, this ain’t it, chief.
- At least in the Middle Republic. Caesar’s quaestors seem not to do this, curiously – something that may have to do with Caesar’s manifest near total command and logistics independence on account of being at odds with much of the Senate.