Why does a still from Kelly’s Heroes front a post about the late Republic/Early Empire legions? The answers, my friends, is science. Well, some science, at least. (Around 4.800 words, estimated reading time: 24 min.)
How physically active Roman legionnaires were is critical to the history of Roman military logistics.
The legionnaires’ level of physical activity indeed determines their caloric needs which, in turn, determines how much foodstuff the Legions needed to get through the supply line and source locally through foraging and requisitions (and when the latter two failed, buys from local merchants) to keep the legions functional.
On the face of it, it seems that the proper way to address this question is to eyeball the level of physical activity of legionnaires, count calories, and check a calories table for the nutritional value of the food we know the Legions ate. But that’s where things get tricky: the methodology that backs the received view of the caloric intake of Roman legionnaires is notably different and rests on an assortment of questionable comparisons with modern soldiers and companion assumptions to this comparison.
I’m far more interested in the level of activity of legionnaires and how they could maintain it rather than by their caloric intake. But the received view of the caloric intake encapsulates an estimate of the legionnaires’ physical activity level which is most likely wrong. So, before I address the latter and substantiate my hypotheses, I have to take care of the former.
And so, here we go.
Calories Counting, the Lazy Way
The reference in the field of Roman logistics is Jonathan Roth’s The Logistics of the Roman Army At War (B.C. 264 – A.D. 235), published in 1999.
Roth dedicates the first chapter of his Logistics of the Roman Army at War to obtaining an estimate of the caloric content of standard-issue Roman rations based on: (1) the energy content of food; and: (2) some anthropometric estimates, including US Army Recommended Daily Allowances for infantry soldiers from the 1960s and the average size, weight and age of the Roman legionnaires. From this, Roth derives a figure of how much calories a legionnaire would need on a daily basis. Roth’s methodology is definitely weird. But before I get to that, a few words of praise.
Roth goes into great length to dispel nutritional myths that other scholars had previously bought into. For instance, that wheat loses some of its nutritional value when baked into bread. Or that the legionnaire’s predominantly vegetarian diet, with about 80% of its calories coming from wheat and pulses (legumes), would have been missing in quality protein.
This no-nonsense approach breaks down when Roth follows with a pair of half-baked assumptions which are encapsulated in the quote on the left-hand sidebar. Slightly rephrased, we get:
Roth’s Fundamental Assumption (RFA): the average daily caloric cost of fighting in an ancient army and in modern infantry are by-and-large the same.
Roth’s Auxiliary Assumption (RAA): the average caloric cost of downtime in an ancient army and in modern infantry are by-and-large the same.
I left implicit in (RFA) and (RAA) parameters of age and bodyweight, but shorter and older individuals need fewer calories than taller and young ones, respectively. (RFA) is not really problematic, and I won’t say much about it.The real issue is (RAA), because the non-fighting part of legionary life would have been at times as calorie-intensive as the fighting part, and sometimes more.
But before we get to that, let’s first look at Roth’s methodology.
Roth’s Comparative Methodology
In methodological terms, (RFA) and (RAA) are what philosophers of science call empirical hypotheses: they are about something observable in principle (hence “empirical”) and they can be supported or undermined by the data (hence “hypotheses”). Roth’s own words lump together both (RFA) and a claim that it is supported (“there is no reason to think that…”), so our first step is to look at Roth’s reasons to think that (RFA) lacks any good counter-argument.
These reasons, as it turns, are also well-entrenched empirical hypotheses:
- The Recommended Daily Allowance (RDA) for the US Army in the 1960s are normative values for infanterying. This hypothesis allows Roth to consider the US Army RDA as an average caloric requirement for ancient warfare. The RDA for an average Army recruit (a 175cm/75kg 19-years-old) is 3.600 kcal, and Roth proposes to adjust it from anthropometric data about the legions.
- Shorter, lighter, older people need fewer calories overall than taller, heavier or younger ones. I mentioned that one already, which was derived from statistical values and is built-in in the US Army RDA. For instance, the RDA for a 25-yr old of the same stature as the reference recruit drops to 3.200 kcal.
- The bulk of Roman legionnaires were in their late 20s-mid 30s. This hypothesis is derived from both literary and archaeological sources.
- The average Roman legionnaire was shorter and lighter than the average US Amy soldier, whereby Roth settles for estimate average height and weight of 171cm and 66kg (respectively).
Roth’s methodology boils down to adopting the US Army tables as gospel (1-2) and reading values for Roman legionnaires (3-4). This is of course conditional on the energy expenditure of fighting as ancient and modern infantry being equivalent, a hypothesis that I did not number because Roth never makes it explicitly (but which is, when all is said and done, most likely correct).
That’s about it. That’s simple, elegant, methodologically very economic, and extremely powerful. Indeed, from the above, one can deduce that the ideal intake for a 30 yrs, 171cm and 66kg legionnaire is 3.000 kcal. Furthermore, data from modern military medicine show that a temporary severe caloric deficit (around 50% for a few days) does not affect a soldier’s performance. Assuming this, Roth deduces that the caloric intake of legionnaires could have often have dropped down to 1.600 kcal during periods of inaction without compromising the legionnaires’ fighting chances.
Given what archaeological sources tell us of the composition of standard-issue rations and what we know of the caloric value of their components, this yields an estimate of the bulk and weight of the military rations. Down the line, we can get an estimate of the constraints (procurement, transport, storage, and their associated cost) the corresponding amount of foodstuff would impose on the supply line. But that’s Roth’s problem, not ours.
There are no major complaints about (1), (2), (3) or even (4). There are some minor issues with (1), but I’ll leave for an aside; and (2) which only holds for basal metabolism, not the metabolism of effort, but that’s a red herring given how (RAA) underestimates the level of effort of legionnaires. There’s nothing wrong with (3) and (4) goes to the credit column because it runs against the received opinion of historians and archaeologist at the time Roth wrote his book (although it’s probably dumb luck, see aside below).
Roth’s estimates for height and weight. Roth’s book was published in the late 1990s, when the consensus among historical anthropologists was that the height and weight of pre-Industrial Era populations could be estimated by projections of our older statistical data (mid-19th century, thanks in particular to census during Napoleonic Wars) and contemporary data about developing country. Roth seems to have ignored this debate and simply taken for granted the figures from Vegetius’ Epitoma (a minimum height of 171 cm for recruits in the legion). Vegetius does not mention the weight explicitly, so Roth cross-referenced the average weight of US soldiers of 171cm according to US Army tables and came up with the 66kg figure. Eventually, Vegetius’ figure was supported by Geoffrey Kron (2005), “Anthropometry, physical anthropology, and the reconstruction of ancient health, nutrition, and living standards” published in Historia: Zeitschrift fur Alte Geschichte. Kron estimates the average height of Italians during the Roman Era (500 BCE-500 CE) to 168cm based on direct calculations rather than projections from modern data and argues convincingly against the use of modern comparisons. He also adds that his data set is biased by inclusion of older individuals (people get shorter with age) and that the average height of military recruits would have beeen a good 3cm higher. From this reasoning, we fall back Vegetius’ 171cm. Kron gives no estimate for the weight tho, so I’ll stay with Roth for the moment.
Periods of inaction, really?
Roth’s carefully-built theoretical edifice is not as solid as it seems. In fact, there is something genuinely weird in how Roth estimates caloric expenditure while not fighting. There may ultimately be nothing wrong with (RFA), but (RAA) is way off base.
Downtime for post-WWII soldiers is factored in the US Army RDA: the recommendation of 3.600 kcal (or 3.200 kcal for a smaller, Roman-sized soldier) corresponds, in fact, to maintenance of base metabolism and ideal body mass for the average recruit. Then again, modern army barracks have water and central heating, etc. When these amenities are not accessible (e.g. when the army is advancing), there are still water trucks, field messes, and in the worst-case MRE. But everything a modern soldier gets effortlessly during periods of ‘inaction’ was an occasion of physical activity for ancient soldiers. No running water means daily water runs. No central heating means daily wood chores. Rations issue in the form of grain means daily milling, kneading, and baking.
Roth gives a cursory treatment to milling (p. 48-49) and reports that, based on experimental archaeology data, it would have taken about 100 minutes/day to mill the daily grain ration of a 10-man contubernium into a coarse flour using the standard-issue handmill (carried by one of the contubernium‘s pack animal.) Roth then remarks off-handedly that “[s]uch work is very strenuous and in order to get a reasonably fine meal, it must be repeated two or three times” (p. 49) which amounts to 5 hours of milling (!). That’s substantial, but Roth totally fails to factor in any caloric cost for this strenuous activity in his daily calculations.
Why so? Let’s find some clues. The first is that “[o]ne should not assume that each soldier literally did his own grinding and baking each day” (p.48) because it’s a contubernium duty that was probably carried primarily by the (two) support personnel. The second is that “[t]his work was probably not done every day, but rather enough grain was ground at one time to last for several days.” (p.49). So, the charitable interpretation of Roth’s oversight is that Roth seems to think that the caloric cost of ‘milling days’ evens out with the absence of extra cost on non-milling days. That makes sense, of course, but only to a point.
Now, I don’t know what your experience with a handmill is. If you have none, I recommend reading this short blog post from a member of the Association Européenne de Reconstruction Antique. It does not mention the level of effort of this activity, but according to my own experience  there is no way that the support personnel could actually spend n*100 minutes hand-grinding in a single day even if stuck to one-time, coarse-milled flour. At least, without tremendous loss of power output, which would delay the process by additional minutes. So, on “grinding day”, the eight soldiers of the contubernium would have had to lend a hand to the milling. Or, more likely, every day was grinding day and everybody would lend a hand, bringing milling down to 10 min/man.
Beyond the effort of milling, the problem is how (RAA) discounts the cumulative cost of daily chores during “periods of inaction”. I already mentioned the water and firewood situation, but remember that for a Roman army on the move (but not fighting, which counts as inactions) there would have been at times some daily or nearly-daily marching, dig-ditching, fence-raising, tent-setting, field-cooking, etc.
In fact, the cumulative caloric cost of inaction for an ancient army might at times have exceeded the caloric cost of fighting. This invalidates (RAA) with potentially other dire consequences for Roth, since any revision of his calories figure implies down the line a revision of logistics constraints.
Fortunately, Roman soldiers had ways to supplement their caloric intake without straining the logistics. For instance, they could forage, hunt, make cheese, and (when available) buy food from sellers that would follow the army train. None of these would have set off their caloric balance (cf. aside below about foraging and hunting). And so the logistic constraints for higher caloric needs may not have always been worse than assumed by Roth. But we certainly should not trust him, because (RFA) is questionable and (RAA) is flat-out false.
Calories Counting, Mk II
While Roth’s (RAA) fails to factor some well-known, heavy caloric costs, straightening the record is not that complicated.
In fact, the record had been set straight at about the same time Roth was putting together the material for his book. Unfortunately, the odds that Roth would have accessed that straight record are extremely low. And judging by (RAA) and his other conclusions, he clearly didn’t.
The last section tracked some “missing calories” but recognized that some of the calories hidden by (RAA) could be obtained without additional logistics constraints, at least on the supply train. Daily chores such as hand-milling, firewood and drinking/cooking water runs, carrying fodder for the contubernium‘s pack animals, and the likes, would not offset Roth’s figure by more than 250-300 kcal. That’s nothing that a few nuts and some full-fat cheese can’t provide.
But then, there’s work that Roth acknowledges but does not consider from the energy expenditure standpoint. Less common than hand-milling, and arguably not performed daily, this type of work had a high energy cost that throws off all of Roth’s calculations out of the window. Let’s have a look at it.
Sub sole otiari
With (RAA) Roth seems to have a pretty narrow view of what counts as “action” and “inaction” where “action” amounts to combat and where anything else amounts to idleness. Picture Kelly’s Heroes Oddball’s unit before they provide Kelly with “a very nice edge”, and I’d bet that you’ll have something close to Roth’s mental image of a legionnaire during “periods of inaction”.
That picture certainly applies mutatis mutandis to the Numidian legions Q. Caecilius Metellus took command of in 107 BCE before he whipped them back into shape. But I already mentioned (in the second installment of this series) how Metellus disciplined his troops with a regimen of daily marches, camp-setting, and cooking which under Marius would become a training standard. In fact, Roman military doctrine holds that that idleness breeds mutiny, and mutiny was the one thing at Roman officers feared more than the legionnaire feared their officers.
When taking this doctrine into account, Roth’s calculations begin to seriously break down. First, there’s the impact of loaded marches on the caloric intake of the legionnaires. Legions in campaign marched a lot and seldom established long-duration camps for other purposes than logistics. But soldiers stationed in those camps would still be drilled often with both loaded marches and weapon training, according to guidelines issued by the camp commander. Here, the examples of P. Cornelius Scipio not-yet-Africanus 4-day training periodization after the conquest of New Carthage in 216 BCE and (again) of Q. Caecilius Metellus not-yet-Numidicus in 107 BCE ( both covered in the second installment of this series) stand out again.
And there is more. For instance, the original muli Mariani who dug the original Canal d’Arles à Fos (also called Canal d’Arles à Bouc) while waiting to attack the Cimbri and Teutones (On the latter, see Loaded Carries in the Analytic Fitness™ Dictionary). More generally, heavy construction work was a common occupation of “inactive” legions (in Roth’s sense of “inactive”). For additional examples, C. Julius Caesar had his Gallic legions built pontoon bridges over the Rhine and Trajan’s troops contributed to build Colonia Ulpia Traiana Augusta Dacica Sarmizegetusa at the end of the Dacian Wars. That was what long periods of Rothian inaction would look.
And, adding insult to injury, Chapter 4 of The Logistics of the Roman Army at War features a 4-page section titled “Logistical infrastructure” (pp. 214-218) with three subsections: Roads, Bridges and Canals, where Roth acknowledges all of the above. But for some reasons, Roth remains absolutely silent and apparently unconcerned with the energy cost of such work. Fortunately, others have addressed it.
“The Legionnaire, an unacknowledged athlete”
Researching for this series, I stumbled upon a 20-yrs old review study from Emile Fornaris and Marc Aubert on Roman military nutrition published (in French) in 1998 in Histoire des Sciences Médicales and republished by the French website Theatrum Belli (Le Légionnaire Romain, cet athlète méconnu). Unlike Roth, Fornaris and Aubert do not proceed ‘top-down’ deductively from general principles, but ‘bottom-up’ adding piecemeal caloric expenditures of the activities of the legionnaires. The section on the evaluation of the overall energy expenditure of legionnaires is remarkably compact and is translated below by yours truly.
Extant ergonomics data allows evaluating the energy expenditure of a soldier. For [Roman] legionnaires: 5-hour loaded march with a 20kg load: 3.000 kcal; setting up camp with one hour of earthwork or tree felling: 400 kcal; assorted chores: 250 kcal. To these expenditures, one must add the non-compressible cost of base metabolism, that is, 1.600 kcal. Adding up those values, the grand total is 5.250 kcal.
Here’s for another estimate: the forced march of 74 km, performed by C. Julius Caesar’s legionnaires in 24 hours, without load, with a 3-hour rest, would correspond to an individual expenditure of over 6.000 kcal.
As for public work, in the absence of quantitative data, one can estimate that they would correspond to earthwork and masonry work performed by workers at the beginning of the 20th century before the introduction of power tools. Their caloric cost has been evaluated between 4.000 and 5.000 kcal per day.
Today, only athletes, military recruits and a few manual laborers actually achieve an energy expenditure of 4.000 kcal per day. (Fornaris & Aubert 1998, p. 163)
There cannot be a starker contrast between the methodology exemplified above and Roth’s reliance on US Army RDAs from the 1960s. The immediate consequence is that Roth’s figures are off by possibly more than what soldiers could compensate on their own. In fact, they might be fine for a day of ‘action’, but that’s a topic for another post (if you’re not fed up with calories, pun intended).
An army of (body)builders
The nutrition of Roman soldiers deserves better treatment than it has so far received from mainstream scholarship.
Whatever the other merits of Roth’s book, his methodology for estimating the caloric expenditure of legionnaires is wanting. I paraphrased his assumptions as (RFA) and (RAA) and in view the arguments previously exposed, the verdict is something like that:
- (RFA): Questionable, but likely correct. On average, modern infantry soldiers carry heavier equipment than ancient soldiers, but its use is not as caloric-intensive and they usually carry it on shorter distances (save for special operation forces, see Old School Strength (III):Ad Optimam Valetudinem Fingendam). Effort-wise, this may very well even out, validating (RFA).
- (RAA): Provably false. Even in the best case scenario, that is stationed in a permanent or long-term camp, on days without training (when there where such days, as with P. Cornelius Scipio in New Carthage, 216 BCE) and fed at a sustainable deficit, the caloric expenditure of an ancient soldier would have been 250-500 kcal higher than that of a modern one, due to unavoidable chores.
Logistic-wise, the Roman army could have done with “deficit” rations, leaving to foraging, hunting, and cheesemaking to balance out calories in and calories out. But in the rather common circumstances when legionnaires were used as earthworkers, stonemason, etc., guaranteeing adequate caloric intake for the Legions’ construction work would have posed serious logistic dilemmas.
From the example of Marius, we know that public work was one of way officers used to maintain peak physical fitness in the legions. Officers must have been conservative with both the workload and the food supply, so that kind of work would not impair the legions’ ability to fight. Marius definitely set an example for that with the fossa Mariana.
We also know that Marius’ example was not an isolated case and that legions were physically ready for undertaking large-scale engineering work while campaigning and recover fast from it. Ceasar’s Gallic legions were not worse for the wear after building the bridge over the Rhine, and soon after ere able to double back extremely fast when they learned that the Sicambrians and Suevians were moving eastward to meet them, tearing down the bridge in passing.
Finally, we know that legions undertook public work in the aftermath of a campaign as a counterinsurgency tactic ensuring swift colonization and (possibly) preventing soldiers from roaming the conquered country. Case in point, the construction of Ulpia Traiana Sarmizegetusa, ordered by Trajan at the end of the Second Dacian War (105-106 BCE). Within five years, the new Dacian capital was settled by veterans, and since it was built according to Roman standards, those veterans probably did not merely boss around Dacian masons and carpenter for five years drinking posca, eating cheese, and catching some rays (you know).
All of the above suggests that Romans were adept ad supplementing the diet of legionnaires with high-calories food one way or another. If I were a betting man, my money would be on some sort of trail mix (dried fruits and nuts) from foraging, and high-fat dairy (from the legion’s livestock). This seems quite logical, given that the legionnaire’s rations included a disproportionate amount of carbs and could have lead to deficiencies in liposoluble vitamins, in particular from the B-group (as noted by Fornaris & Aubert).
Anyway, what we are left with is evidence that the Roman legions maintained strength, endurance and athleticism through a combination of military drills, manual labor, and high-calories nutrition. The optimal caloric intake for them would have been in the high 4.000 kcal, with Roth’s top-end figure (3.600 kcal) closer to a sustainable deficit if not supplemented by hunting and foraging. As noted by Fornaris & Auber, the minimum sustainable without loss of performance for a short period would have been around 2.000 kcal.
In a future post, I’ll address what I think this regimen would have made our legionnaires look like, but [spoiler alert] they were certainly not crassi fututores.
^ The myth that baking bread reduces the caloric value of wheat is tackled explicitly (p. 48) and sourced to Donald Engels (1978) Alexander the Great and the Logistics of the Macedonian Army, Berkeley: University of California Press (p. 124). It leads to an incorrect estimate of how much bread would be necessary to cover a 3.600 kcal daily ration (more about this figure later): 1.6 kg according to Engels vs. 1.3 kg according to Roth. The myth that a predominantly vegetarian diet is poor in quality protein is tackled implicitly (p. 43), noting that “One should note that the combination of various elements of the ration resulted in a higher nutritional value than each individual element. For example, grain and beans eaten together provide protein [sic]”. Of course, grain and beans eaten alone also provide protein, but we’ll let that slip, as Roth probably means “complete protein”.
^I have some experience of Roman handmills from back in the days where I briefly dabbled in Roman and Etruschean archaeology (something like 1994-1995). I was not in the same shape as I’m in today, and lacked in particular in power-endurance in the upper body, but I had some gym experience and at the time, I would have rated the action of a handmill at about the same level of effort as rowing on an ergometer, which is Vigorous on the MET scale (6-8 MET/h). Accounting for the lack of significant lower-body action, I would be incline to settle for a conservatinve 5 MET/h, which is the transition zone between Moderate and Vigorous (“Moderately Vigorous Plus” for fine-grain MET scales).
^ Take, for instance, how C. Julius Caesars’ engineers build their largest pontoon bridge under the Rhine in merely 10 days. Those of Cesar’s legionnaires assigned to the task must have been logging and building quite intensively and could probably not have built for 10 days straight in a caloric deficit. On the other hand, Ceasar’s troops numbered 40.000, but only so many could participate to the construction at one time. Organizing a rotation and foraging and hunting parties would not have been an issue.