Building a Fantasy Army Part 7: Logistics
An army marches on its stomach. What this means is that while a fantasy writer probably should not spend words boring his readers with minutae of logistical arrangements, some thought should be given to logistics of raising, equipping and then supplying an army. It will make for a better and more interesting story: an interesting story requires an adversity to be overcome, and logistics of maintaining an army in the field are a very major problem for any army. They are a big (but not only) reason for why warfare was, for millenia, far more complex than smashing two big blocks of men together. A teleporting army subsisting on air – a depressingly common trope in fantasy – is not fun to either write or read about. An army where commander starts looking like a juicy steak if he mismanages logistics is a lot more interesting.
Militiamen and part-time professionals (such as feudal knights and Byzantine thematic soldiers) were usually expected to provide their own equipment. They might receive some cash allowance from the government (soldiers of themata did), but they were expected to procure their own equipment. In contrast, full-time professionals could be equipped by the state. Late Roman and early Byzantine soldiers were equipped in this way. Likewise, household troops of fyrd and feudal armies were typically equipped by their patron – be it a king or a major noble. But this was by no means uniform – mercenaries typically bought weapons and equipment themselves, as did Roman legionaries. These weapons would dull and break, which means replacements – or at least backup weapons – were required. That is what swords were for: they were, with few exceptions such as zweihanders, literal sidearms – something you used if your primary weapon broke. But on army scale, replacements for primary weapons were a must.
This naturally influenced the choice of weapons and armour. In general, where armour was supplied by the central state, there were two popular choices: no armour, or munitions armour. This munitions armour could then be segmented, lamellar, or plate armour. Roman lorica segmentata was such a standardized armour, and highly uncomfortable to wear. During Middle Ages, Romans opted for the lamellar armour as it had several advantages compared to mail. Armour was made of many small pieces of armour, which meant that pieces could be easily produced en masse by relatively unskilled workforce. Armour could also be easily adjusted to the wearer by taking pieces out or putting them in, allowing for a standardization of production (just remove several horizontal rows and put in more vertical rows for a short fat man, and the opposite for a tall thin one) – this was done regularly by soldiers. Ease of removing and adding plates also meant it could be easily repaired in the field by simply replacing damaged pieces. Similar reasons were behind Chinese usage of lamellar armour. In both cases, armour for officers had many more lamellae than armour for rank-and-file, and was also supplemented by mail.
In Western Europe, where armour was produced for individuals and bought by individuals, normal choice was mail, which was later reinforced by plates. This led to development of a coat of plates, which further evolved into two lines. On one side, plates got smaller and smaller, turning a coat of plates into brigandine. On the opposite side, plates merged and enlarged, leading to appearance of the full plate armour. At around the same time, in fact, other styles of armour surpassed the lamellar world-wide: brigandines in China, early plate armour in Japan and mail in the Ottoman Empire. European choice of large plates atop mail again may be explained by the fact that soldiers bought their own armour – and poorer soldiers opted for brigandines or simply gambesons. Ottoman troops likewise bought their own armour, while Chinese choice of brigandines may again be connected to the state supply of armour.
In Europe, there was significant similarity in armour, due to only a few production centres making it, and armour consolidated into two main styles: German (Gothic) and Italian (Milanese) style, and while some areas showed preference for a certain style, everybody used what was available. In the 16th century, a combination of increased usage of gunpowder weapons and introduction of standing, professional and centrally-supplied armies led to the appearance of Almain Rivet, one-size-fits-no-one half-plate armour intended for the pike infantry. It was generally worn wothout mail.
Such armour, while not contemporary with Byzantine thematic system or Anglo-Saxon fyrd, could easily be used to equip similar troops where they were available. In 15th century, infantrymen also did something similar by ordering plate armour but without leg protection: aside from being more lightly (and more cheaply) made in general, this additionally reduced both weight and cost of armour despite it not actually being munitions armour.
Greater problem are logistics of maintaining army in the field. A large army requires a massive amount of supplies. Overall, an army might be able to carry around 3 weeks of supply with it, and definitely no more than four – and this only in flat, nice terrain that allows mules to pull a large number of large wagons. If pack animals have to carry supplies (as is the case in mountainous terrain), then two weeks may be an expected limit. When it comes to foraging, around 25 kg of food may be pulled from each farmer without causing starvation. Now, main army may move at 10 miles per day, and foragers can reach area ten miles from each side of the army. This means that foraging can cover some 200 square miles. Population of England in 1450 was 2 million, for average density of 40 per square mile. This means that an army may count on average of 200 000 kg of food per day, or 2 800 000 kg over two weeks. This would enable an army of up to 100 000 men. In reality however an army would be far smaller, as population density is not everywhere the same, and any army had about as many noncombatants as combatants (and sometimes far more). Thus a practical field army was typically around 20 000 – 30 000 soldiers, especially since late medieval armies had far more cavalry by proportion (2/3 – 3/4 was cavalry), and thus required much more logistical support per soldier.
If army is 10 000 strong, with 6 000 cavalry and 4 000 infantry, this would mean 10 000 soldiers, at least 10 000 camp followers (assistants, armourers etc.), 6 000 war horses and 6 000 riding horses. Cavalry would require maybe 6 000 mules, with additional 500 mules for infantry. Large war horses require a large supply of food if relying on pasturage, around 15 kg per day, while normal war horse requires 9 kg per day, and same for pack or draft horse, while mule would require 7,5 kg per day. World War I regulations required 10 kg of oats and chaff for a charger, and 4,5 kg for a mule. If fodder was exclusively hard, then a horse may require 2,2 kg per day and mule 2 kg per day; a charger may thus require 4 kg per day. A soldier would also require 1,36 kg of grain per day. Mule also requires 20 l of water per day, and horse 30 l.
Basic requirements are thus as follows:
- 10 000 soldiers: 13 600 kg per day
- 10 000 servants: 13 600 kg per day
- 6 000 war horses: 24 000 kg per day (60 000 kg oats + chaff)
- 6 000 riding horses: 13 200 kg per day (29 700 kg oats + chaff)
- 6 500 mules: 13 000 kg per day (29 250 kg oats + chaff)
Total requirements come out to 77 400 kg per day. A standard pack horse or a mule can carry a maximum of 120 kg over any distance. Ridden cavalry horses would carry 34 kg of barley. As such, a horse with a rider can carry enough supplies for three days. Requirement of 77 400 kg per day would require 645 mules for each day of march. For two weeks, each mule would have to carry 63 kg of fodder for itself and 19 kg for its guide, leaving 38 kg for supplies; due to weight of a saddle and other accesories, this could be reduced to 30 kg. A march of two weeks requires 1 083 600 kg of barley, not counting fodder for mules carrying the fodder. Of this maybe 408 000 kg could be carried by cavalry. Remaining 675 600 kg would require 22 520 mules (on top of the 6 500 mules carrying equipment). Note that this is a very simplified calculation: in reality, one would need to account for water spilling, food spolling or falling down, wagons breaking down, mules and horses dying or being lamed… a fully cavalry army on the other hand could cover up to 65 kilometers per day, but in small groups and with no wagons.
Roman army at march during Titus’ campaign in Judea. Army is in strength of two legions plus auxilliaries. Legend of the order of movement: 1) light infantry scouts, 2) vanguard of one randomly selected legion plus a cavalry troop, 3) ten men from each century in the army, carrying tools for making a camp, 4) pathfinders / engineers tasked with clearing obstacles, 5) commander’s and officers’ baggage, escorted by cavalry, 6) commander with his bodyguard, 7) complete legionary cavalry, 8) mules carrying siege weapons, 9) higher officers: legates, tribunes, prefects, with select troops, 10) legions, with aquilla in front and trumpeters behind them; legionaries march six abreast, 11) mercenaries, 12) rearguard comprised of numerous infantry and heavy cavalry. Army would march 30 kilometers a day in normal conditions, and could cover over 50 kilometers a day in a forced march, when using a road. Off-road, 11 km per day was more typical.
Beyond the obvious wagon drivers, logistical support of the army will also include messangers, caretakers (veterinarians etc.), and other support staff. Depending on how organized army is, these might be conscripted civilians, civilian contractors or even members of the army. I mentioned weapons breaking in the second paragraph. It did happen, and while a backup weapon is good to see you through the battle, a replacement will eventually be necessary. This means that armies were followed by camp followers. These were not – contrary to what GRRM thinks – whores (though whores were included!). Majority of camp followers were far less fun but far more important. They were blacksmiths, carpenters, leather workers cooks and medical personnel – for men and horses alike. And they have to be both fed and paid.
Logistics would be similar for human-esque races. Dwarves would have logistical requirements similar to a human army – but due to short legs, they would have a much slower sustainable march speed. As a result, dwarven army would likely be able to cover about the half of the distance per day that a human army does, and would thus have lower (maybe little more than a half) operational radius, for the same amount of supplies. Dwarves are also likely to lack cavalry, making foraging more difficult. As a result, they are unlikely to mount any long-distance offensive campaigns, unless nearby rivers allow the access.
Elves would also be similar, as they are typically portrayed as “human+” in fiction – or at least as far as the demands side goes. But since they are extremely advanced, their supply side logistics would be far superior. Something like Tolkien’s lembas and miruvor would completely transform logistics – even if it was not utilized on a regular basis. Instead of having to move around with a massive wagon train, elven armies could be entirely on foot or mounted, thus allowing them far better mobility. Roman army’s forced march of 50 kilometers per day would likely be normal for an elven army marching on a road, and off-road march rate would be significantly increased. Assuming that water is not a problem, elves would only need to carry lembas. That would be a major help. During the Civil War, soldiers received 9 – 10 pieces of hardtack per day weighting a total of 1 pound (0,4 kg). With lembas, these 10 pieces would last for ten days. For comparison, Roman soldier’s ration was 1,5 kg per day, and twenty days’ rations were carried by the legion. This means that by using lembas instead of those rations, army could march for 75 days without resupply, giving it an operational range (not radius) of 3 750 kilometers.
Centaurs would have the problem opposite to the elves: they would basically have the caloric need of a human plus a horse. This means that requirements of an army would be 53 600 kg per day for soldiers, same or similar amount for servants, for a total of 107 200 kg per day (compare to 64 400 kg per day for human army discussed above). As a result, two-weeks march would require at least 23 600 mules, almost four times as many as for a human army of 10 000. In other words, a centaur army is likely to have operational radius about a quarter to a third of that of a human army.
Firewood is likewise significant requirements. It usually is procured locally – which can lead to a massive deforrestation wherever a large army camps for any longer period of time.
All of the above leads to several consequences:
First, warfare is hideously expensive. Conscripting ten percent of population in premodern warfare might have been theoretically possible, but was meaningless since such armies could not be sustained for long. As a result, relatively small but highly professional armies were preferred over masses of peasant levies. Actual field forces will not have been significantly reduced as they were constrained by logistics anyway, but fact that soldiers were professionals meant that they were far more effective than peasant levies will have been (closest thing to levy troops was Hungarian Militia Generalis, but these were still not peasant levies). Militia armies such as Greek hoplite armies are a good study of limitations of such a system: campaign season was limited by the need for citizens to help with the harvest, which also meant that the enemy could not be brought down simply by raiding him. Wars were fought literally several hours or days away from home, and were typically inconclusive due to inability ot hoplite armies to stick around long enough to bring about any decisive conclusion. Despite Attica being 230 km away, and 90 km from friendly Corinth, Sparta was incapable of sustaining operations there, and had armies disperse for lack of supplies. Hoplites were expected to provide their own rations, with Athenian standard being three days’ worth, suggesting that its ability to mount and sustain land campaigns was not any better than that of Sparta.
Second, size of a field army thus depended on sophistication of the logistical system. An army supplied from the field may be some 10 000 – 20 000 strong, at most, but even at that size it requires fairly organized logistical system. If system of supply is more primitive armies will be much smaller. Larger armies on the other hand require a highly sophisticated and centralized system of supply, able to secure supplies and organize supply depots in advance, or else will have to be broken up into multiple field armies – which then require coordination of movement and still risk being defeated in detail. Oftentimes, armies had to be split up regardless.
Third, large armies will tend to follow waterways – either rivers or seas. This can be seen from Ottoman wars: raiding parties of around 10 000 – 15 000 men could go basically whenever they wanted (but even they tended to stick to rivers, and typical raid was much smaller – on the order of 1 000 men). But larger armies, such as those that fought at Mohacs, always marched along the rivers. This is the reason for importance of Belgrade: it sat at the junction of Danube and Sava, and Danube also connects to Drava. Thus, its fall will have opened up a way for invasion of both Hungary and Croatia: Danube flows straight between Buda and Pest, Drava formed the entirety of Croatian-Hungarian border since 10th century, Sava leads to Zagreb, and Kupa leads from Sava to central Croatia (hence why fort of Sisak was built at junction of Sava and Kupa). So while on a map terrain might seem open, in reality armies were constrained to a network of paths not unlike modern road networks. Navigable rivers were lifeline of kingdoms and empires, but also pathways for invasion. Mountainous terrain meanwhile was a good defense – not only did it facilitate successful ambushes, but it also significantly limited the logistical support which large armies could receive. Basil II’s first campaign against Bulgaria in 986. is a good example of this problem: Emperor marched directly to Sardika (Sofia) in the heart of Bulgaria. But the siege dragged on for too long, and Basil found his logistics overstretched. While retreating towards Constantinople, his army was ambushed and destroyed in the mountain passes. His second campaign, starting in 1001., gradually nibbled at outer holdings of the Bulgarian state, strictly avoiding the overextension which proved so dangerous in 986. By 1004. Samuel had lost the ability to mount a successful attack, and Bulgaria was ground down over next ten years of sieges and raids. But in 1071., internal political pressure would force Romanos IV to pursue decisive victory much like Basil II had in 986., with similar but far more strategically disastrous result.
Fourth, army can be defeated without ever being engaged in a battle. If it can be cut off from any potential supply lines and its foraging parties neutralized, an army will starve to death (or, more likely, surrender) without ever fighting in a battle. This is what castles were for: they protected defender’s army and logistics (including his peasants, contra to what GRRM believes) while serving as bases from which invader’s army and supplies could be threatened. Offensive warfare meanwhile was this same approach but raised to the level of strategy: an attempt to destroy enemy’s ability to wage war (that is, logistics) by destroying his economic basis. An army that managed to be destructive and murderous enough could bring the enemy to battle he did not wish by generating political pressure: an example of this is Battle of Crecy, which was a culmination of English chevauchee. Battle of Krbava Field likewise resulted from Croatian army intercepting Ottoman raiding force. And if an enemy army is too powerful, cutting it from supplies can work – even before the soldiers start starving to death, morale will plummet as they lose supplies.
Fifth, most campaigns were not fought during winter. Winter campaigns were avoided when possible. Instead, campaigns were – again, contra Martin – mounted just before, during or just after the harvest season, when supplies could be procured locally. Still, winter warfare was not unknown or impossible: again, Hungarian-Ottoman wars are a good example, and Romans and Byzantines were also capable of campaigning year-round. But Hungarian experiences testify to the difficulties of such. In fact, the main reason why winter campaigns were so frequently mounted by Hungarians, including King Matthias, was that Ottomans – despite their ability to mount year-round raids – had difficulty in responding to those campaigns. Such measures were necessary simply due to massive differences in raw power between two states: even at beginning of the period, Ottoman Empire was more powerful of the two, and the gap only increased with time.
Feeding a Dragon
Now for the fun part, which is specifically fantasy logistics. I will assume a “typical” fantasy dragon, about the same size as dragon from the Skyrim trailer – about 8 human heights, or maybe 15 meters. Problem is weight: being a flying creature, dragon will have to be relatively light and/or assisted in flight by magic.
Canada goose is on average 90 cm long, has 156 cm wingspan, and weighs 4,5 kg. A 15 meter dragons would thus weight exactly 3 375 kg. However, due to different builds, closer weight would likely be 1 500 – 3 000 kg, with latter being typical only for extremely large specimens.
Now for food intake. Kleiber’s law states that the food requirements of various animals scale by the mass of the animal to the 3/4 power. This means that a dragon would require some 78 to 156 times the caloric intake of a goose. Minimum requirements of a goose are about 11 000 kJ/kg/day. For flight, however, energy requirement (based on a crow) would be around 17 watts per kg, or 25 600 – 51 200 W. Raw beef has around 3 300 kcal per kg, or 13 890 kJ.
Cooper’s hawk weights an average of 280 g, or 10 000 times less than a largish dragon. It eats 12% of body weight in one day, or 33,6 g.
Goose do not breathe fire (though they do come awfully close), which means that caloric intake goes way up.
As a result, following requirements can be calculated:
- Basal intake:
- Goose 1: 11 000 kJ/kg/day * 4,5 kg * 78 = 3 861 000 kJ = 278 kg per day
- Goose 2: 11 000 kJ/kg/day * 4,5 kg * 156 = 7 722 000 kJ = 556 kg per day
- Dragon 1: 17 W/kg * 1 500 kg * 4 hrs = 367 200 kJ per day = 26 kg per day
- Dragon 2: 17 W/kg * 3 000 kg * 4 hrs = 734 400 kJ per day = 53 kg per day
- Dragon 1: 304 kg per day
- Dragon 2: 609 kg per day
Medieval cow or ox weighted around 200 – 250 kg, but only 3/5 (120 to 150 kg) is usable. This means that a dragon would need anywhere between 740 and 1 110 oxen per year. If a peasant household could raise about two oxen per year, this means that 370 to 555 households would be required to support a single dragon. If around 5 people were present in the household, then a society with 10% urbanization rate would require 2 000 to 3 000 people to support a dragon. This would allow for 330 to 500 dragons per million people. But due to humans needing oxen as well, more likely number is 33 – 50 dragons per million people (a tenth of maximum – based on the medieval tithe).
It should be noted that these are very small dragons. A 30-ton Regal Copper (of Temeraire fame) would require 738 as many calories as a goose, or 36 531 000 kJ, as well as 510 000 W for flight. A dragon that flies for four hours per day would thus require 7 380 531 000 kJ per day, or 3 542 oxen. If flight requirements scale similarly to basal requirements, however, requirements would be 56 440 W for flight. This would give total of 849 267 000 kJ per day, or 408 oxen per day. This would lead to 150 000 oxen per year, requiring some 400 000 – 500 000 people to support a single dragon. But as before, more likely number is about ten times that, or 4 – 5 million people per single dragon.