Recapping the Fall of the Roman Empire: The Cavalry Problem

It's summer! God speed the plough! Unfortunately, with the hand of every good man and woman turned out to bring in the mangelwurzels and vetch, there's no-one to staff the university library on Sundays. (The alternative explanation, that the university needs to save money, is self-refuting nonsense. The objection that there is an acute shortage of actual researchers using the library, although supported by notoriously unreliable visual evidence, is equally crazy. We're spending enough money on the research-education infrastructure that we must be getting results.)

The "I. K. Barber Learning Centre." It used to be the boring old Main Library used to stand, but we got rid of the stupid stacks in favour of a "state-of-the-art system [which] drastically reduced space, which allowed for the integration of classrooms, offices, informal learning environments, group rooms, reading rooms, and even a climate-controlled vault for rare books." Which is nice, I guess. Well, not nice in the sense that automated retriveal is anywhere near as good as open stacks or in the sense that a schoold with declining enrollment needs more classroom space. But nice in the sense that the university gets to build a Real Big Building. Otherwise, it would be stuck with only building all of the other Real Big Buildings that it is building. Oh! Oh! I have a theory about where the decline in consumer spending might be coming from!

Meanwhile. . . .

St. George's School charges $20,000/year in tuition, and is a bucolic fifteen minute bike ride from Main Library. Does "bucolic" mean rain-drenched forests? Maybe not. Not many buildings in sight, is what I'm saying. In a city where a detached home  on a city lot is going for seven figures. Oh, well. Whatcha gonna do? Develop the 2000 acres of Pacific Spirit Regional Park? Then where would Vancouverites go to get back to nature?

Shorter ironically couched rant; the library was closed on my last day off, and that's why this is a recap.



First, in the spirit of recapping, let's enjoy Bernard Hill absolutely killing it as Theoden King, followed by Dino de Laruentis' re-enactment of Ney's great charge at Waterloo.

And then this old favourite.

 Dan Herlihy clearly should have given a better, longer speech. I can't find my copy of Return of the King, but I am pretty sure that Theoden gives the count of the muster at Dunharrow on the third day as 8000 riders. Having hoped for 10,000 (it turns out that a lot of the Rohirrim had better sense than get cooped up at Helm's Deep with their king), he is out of time and rides for Minas Tirith on March 10th, 3019, T.A. The battle of the Pelennor is on the 13th, so even given cavalry's high attrition rate, the force we see us probably still around 8000 sabres --more than 50 squadrons of horse, in two cavalry corps if we're going to be all TOE about it, or three large divisions in a single corps based on Theoden's orders session, such as it is. (Also, it turns out that the military climax of Lord of the Rings is a winter campaign. Sauron is a bad guy, but one hell of a logistician.)

Ney's charge is the latest of the series of actual, real-world corps-level cavalry charges that distinguish the big battles of the late stages of the Napoleonic wars, beginning at Eylau and ending at Waterloo, Napoleon's armies prove capable of throwing anywhere upwards of 9000 to perhaps 12000 riders into a single cavalry charge, and while Wikipedia calls Sobieski's 20,000 sabre "charge" out of the Wienerwald the largest cavalry charge in history, the scare quotes should  signal my respectful disagreement that anything like a coordinated, massed charge by 20,000 cavalry occurred in the afternoon of 12 September 1683.  (Also, I hope they paid Richard Overy one hell of a lot of money for this project. Also some more, in a net bit of symmetry, Theoden's march on Minas Tirith and his charge at the Pelennor Fields is clearly modelled on Sobieski's at Vienna.) Eight thousand sabres in a single charge is Peak Cavalry Warfare, requiring lots of old-timey staff work, and is emphatically not something you expect to see the fyrd of Rohan putting together.

Organisation, you say. Pff, you say. How hard can it be? This is, it turns out, a surprisingly hard question to answer, mainly because there's something about horses that evidently rots the brain, he said, uncontroversially. 

"The military importance of Adrianople was unmistakeable; it was a victory of cavalry over infantry. The imperial army had developed its attack on the position of the Goths, and the two forces were hotly engaged, when suddenly a great body of horsemen charged upon the Roman flank. It was the main strength of the Gothic cavalry, which had been foraging at a distance; receiving news of the fight, it had ridden straight for the battle-field. Two of Valens' squadrons, which covered the flank of his army, threw themselves in the way of the oncoming mass, and were ridden down and trampled under foot. Then the Goths swept down on the infantry of the left wing, rolled it up, and drove it upon the centre. So tremendous was their impact that the legions and cohort were pushed together in helpless confusion. Every attempt to stand form failed, and in a few minutes left, centre and reserve were one undistinguishable mass. Imperial guards, foederati and infantry of the line were wedged together in a press that grew closer every moment. the Roman cavalry saw that the days was lost, and rode of without another effort. Then the abandoned infantry realised the horror of their position; equally unable to deploy or to fly; they had to stand to be cut down. It was a sight such has had been seen once before at Cannae, and was to be seen once after at Rosbque. . . . Such was the battle of Adrianople, the first great victory gained by that heavy cavalry which had now shown its ability to supplant the heavy infantry of Rome as the ruling power of war. During their sojourn on the steppes of South Russia the Goths, first of all Teutonic races, had become a nation of horsemen. Dwelling in the Ukraine, they had felt the influence of that land, ever the nurse of cavalry, from the day of the Scythian to that of the Tartar and Cossack. They had come to consider "it more honourable to fight on horse than on foot;" and every chief was followed y his war band of mounted men. Driven against their will into conflict with the empire, they found themselves face to face with an army that had so long held the world in fear. The shock came, and probably to his own surprise, the Goth found that his stout lance and good steed would carry him through the serried ranks of the legion. It had become the arbiter of war, the linear ancestor of all the knights of the middle ages, the inaugurator of that ascendancy of the horseman which was to endure for a thousand years."

See, Athens is where all those pilosopher dudes were? And Attila is a barbarian who conquers the place? And the Athenians get all undignified and give him fruit and stuff? There's a moral here. Needs more naked chicks.

So, basically, for five centuries or so, the Roman legions had been awesome; then, at Adrianople, they weren't! "Gothic heavy cavalry" was awesome instead! So then everyone said, "Let's not do infantry any more. Let's do cavalry!" And for a thousand years, everyone proceeded to forget that no good infantry can ever be upset by a cavalry charge. (Especially not good, true Scotsmen infantry.) If you can't draw the appropriate conclusion from the dates as given,   Charles Oman will draw it for you, a little later on: English longbowmen are even more awesome than knights. What happened? Oh, Parliament, rule of law, middle class, capitalism: that kind of stuff. It's an inoculation against Horse Related Chronic Dementia --one of the two long term disabilities caused by being rich and enjoying fox hunting, the other being gout.

And that's how it came to be that between Classical infantry armies and Renaissance early modern ones, there is a thousand years of barbarian darkness of knights charging each other with no regards for tactics. People were stupid. And then they got smart! (The Dark Ages were dark,  you see.) Revising the historical account to put infantry back into Medieval armies, ,as modern scholars like to do, doesn't help as much as would seem at first glance. 

Putting cavalry back into modern war does, however. Especially once conceding gunpowder weapons, actually paying attention to the role of cavalry on the Napoleonic battlefield makes things seem much more like a continuous trend in which the crucial break was between ancient Roman warfare and later medieval, and not between medieval and (early) modern.

Lynn White long ago resolved this problem to everyone's satisfaction by arguing that the stirrup was an early Medieval invention. Before the stirrup, a rider couldn't couch a lance, therefore couldn't charge, as a knight does, and achieve the huge kinetic energy advantage that allows a mounted knight to kill a firebreathing dragon 

win battles with Roman legions. White's formulation being a bit silly, revisions have been issued in which it is saddles, or perhaps harness, or, who knows, boots or horseshoes or, in general, something horsey (bridles are the bits that go in the mouth, right?) that made the difference.

So, technological progress? But, then, during the Punic wars, it seems that cavalry could decide battles. So the real anomaly is the Late Republic/Empire? Probably. The question is the nature of the anomaly. I am going to go with, "Horses eat stuff."  

Since it's only been two weeks since I took the trouble to type it all in, I am going to resist the temptation to reproduce my summary tables of the state of English agriculture in 1875 here, however much I want to cram it all down the throat of the people trying to fight global warming by getting rid of steak.

And more importantly, dairy fats. 

In 1875, four years after the decennial census showed 3.1 million "persons engaged in agriculture" in the United Kingdom (so including the Republic of Ireland), there were 1.8 million horses "in" agriculture, 6.3 million acres in wheat, barley and rye; 4.2 million in oats, and 6.0 million in hay, and 23.7 million in permanent pasture. This would have produced a little over 2 million tons of grain, 2 million tons of oats, 2 million tons of good quality hay, and about the same amount of grass forage.

Googling about looking for figures for the British harvest of 1870, I find this, from Richard N. Adams, Paradoxical Harvest: Energy and Explanation in British History, 1870--1914, just so that you have an idea of what kind of diet this agriculture was supporting, and this (fact free) report on the harvest, so that you can get an anecdotal sense of the effects of drought, early frosts, wireworm and etc.

  

Unfortunately, it turns out that pretty much everyone on the Internet does what I'm doing: multiply the acreages in the Encyclopedia Britannica with the yields given there to extrapolate the British (English) harvest. You can do the same if you'd like. If not, one more bit of plagiarism here, this time from Steve Broadberry and team's attempt to estimate English GDP growth from 1270 to 1870, available as an (automatically downloading) pdf here, thanks to Mark Overton. 

If you are wondering why the numbers differ by so much, apart from higher productivity, the amount of land in fallow drops dramatically with the increasing use of crop rotations, something we can either  account for with an old fashioned, "They invented clover and turnips in 1700" explanation, or , my preferred alternative, accept that we just don't know enough about medieval populations and production to generalise in a useful way.

For the record, my takeaway is that it is demand, not supply, which dominates here. Horses supply demand for agricultural products. They might, in fact, be a particularly important form of demand. Governments, which have a history of being too alert to Very Serious Things to care very much for irrelevancies like peasants dying in the fields, can be diverted from Very Serious Things by the right incentive. Say, your cousin declaring himself emperor out in France, or the Persians coming across the border. Unlike peasants dying of hunger and black rot, this kind of thing might end with you getting killed.

That insight tends to lurk just behind the surface of the discussion of Adrianople. The dominance of infantry gives us the Roman legion. The dominance of cavalry gives us feudalism. 

Edward Freeman says what?

The basics, and thank Heavens for the Internet, which allows me to plagiarise from something called the "Horsechannel:"

With a gut designed for almost non-stop grazing, horses should consume between 1.5 to 2.5 percent of their body weight daily. This means that an average 1,000-pound adult horse should consume 15 to 20 pounds of hay per day, with additional grain supplements depending on condition and activity level. In their wild state, horses' natural forage is a mixture of grasses, clovers, grains and the occasional herb or woody shrub. This provides a variety of minerals and vitamins, as well as fat and protein levels that vary with the seasons. Feeding a single type of hay year-round eliminates this nutritional diversity. The trick is to recreate the natural balance as closely as possible while still meeting all of the nutritional demands created by an artificial training regime. This is commonly done through the feeding of grain "concentrates.

In practice your normal Nineteenth Century or other traditional farmer will feed 20lbs of hay and 5lbs of oats in the English climate, and throw in an exciting variety of additional foods as "physic." A stud farmer, faced with the daunting knowledge that the hire of a stallion means that he has launched into an investment that will not be realised for four long years, takes the normal anxiety about proper treatment of an animal which cannot tell you how she is feeling, and takes it to new heights.  A pre-modern farmer, unequipped with Science? His anxieties are on another level entirely.

It is thus perhaps not surprising that some Iron Age horse industries simply turned the animals out on the range and then rounded up what they needed. Anneli Sundqvist has kindly put up an ethnoarchaeological investigation of horse-rearing practices in Iron Age Scandinavia (pdf), and concluded that they were almost certainly the same as those still practiced in the Nineteenth Century. "Free-roaming" horses were left to themselves on suitable range land, and rounded up as needed.

We do not know just how extensive free-roaming, hands-off horseraising was in the Roman world, whether of adorably fuzzy Norwegian ponies or the North African barbs that somehow became the ancestors of the American horse in a way that has nothing to do whatsoever with long-standing close colonial American trade relations with (ick!) Africa. It seems likely that the Romans had their equivalent of the Yorkshire moor ponies, etc. (Did you see that the Rohirrim banner is the White Horse of Yorkshire? That Tolkien.) . 

However, the way to bet, given the persistence of local variations in Roman equine populations noted in C. J. Johnstone's biometric study, is "a lot." Local working equine populations were derived largely from the persistent native stock, which was selected to the particular environment, and not to the requirements of the breeder. At least, that's Caroline Willekes' explanation.

This does not --quite-- mean that the Roman horse industry was dominated by free-range animals. It means that the genetics, thus the population, is. The free-range genetic base is, however, under-utilised compared with stud-raised animals. The increase in average animal size in Roman times has to do with diet, not breeding.

I'll talk about diet in a moment, but first want to parenthetically comment that, in this context,  Charles Oman's account of Adrianople is silly. The Crimean steppe is the "nursery of cavalry," but of light cavalry. Steppe horses are selected for hardiness, not size. In fact, if you want gigantic mounts suited for heavy cavalry, you simply have to abandon free roaming in favour of stall feeding. This isn't to say that Gothic nobles on the Ukrainian plains weren't producing big horses by stall feeding; only that this does not work as an explanation for the numerous Gothic cavalry, although it might be an explanation for their numerous cavalrymen. 

On to diet: Horsechannel's recommended diets for hard-working animals like race horses are:

i) 15 lbs. oat or barley hay
10 lbs. alfalfa hay
10-12 lbs. sweet feed
top dressed with 2 cups oil

Or:
(ii) 25 lbs. alfalfa/grass hay
5 lbs. complete pellets
10-12 lbs. crimped oats
or barley
Top dress with 1 lb.
rice bran

The standard 2--2.5% gives you about 30lb of assorted grass and grains. As you can see, a working horse pushes up to 40! Wild forage is not going to come to anything like this calorie itnake, and the result will be a rangy pony, more suited to being an Iron Age chieftain's battle taxi than a mighty charger ready to ride for Gondor.

The diet of working horses isn't all that relevant to a stud farmer, but the demands of pregnancy, nursing, and animal growth are also considerable. In the first three years of its life from conception to two-year-old, a foal puts on about 90% of its adult weight and height and of the muscles and bones that will carry it through life, much of it channeled through the foal's dam. Beginning in the second trimester, a brood mare is at the high end of the recommended diet, between 2.25% and 2.5% of body mass each day. For a typical, 1200lb mare, almost 30lb of fodder daily, with supplementary protein and mineral fortification. For the nursing year, when a foal puts on 3--5lb per day, a mare's diet tops out in the same range as that for a working horse, about 3% of body weight. Since this is well above the amount of hay it can eat, dietary supplements are inescapable, and the farmer should splurge on the best quality hay available.*

Again, and because I cannot emphasise it enough. We are talking about grass and grain, preferably the digestively easy oats, and protein supplements via leguminous fodder. Crop rotations are implicated with the kinds of things you want to sell. Wikipedia tells us that Pliny tells us the cultivation of alfalfa was introduced to Greece by the Persians in 490BC, and cites Palladius recommending a ten-year alfalfa rotation. (And speaking of Latin agricultural writers, my Internet sources cite Columella on Roman stud farms.) The difficulty with legume crops remains constant: they're good for the soil, but you have to find something to do with them. Find a path to a legume rotation, and you will be able to produce more from your fields; but even even the most desperate slave is going to be hard to feed on legumes --not least because a regime that produces so much soy, bean and lentils is also likely to produce a great deal of meat and dairy, so that there will be alternatives.

Remember the 1970s, when food shortages were going to force us to tear up all the good  crops and replace them with legume-heavy rotations? And the problem would be, who wants to eat all that high-roughage, tasteless soy and lentil product? The future was so different in those days.

Let's do a thought experiment. Modern police forces may take a horse at the end of its second year, and retire them when they turn fifteen. These are outlier cases, however, and the basal attrition requirement might well be a hand-waved tend percent. For every 100 cavalry, ten new horses are needed each year. A stud farmer aiming, very optimistically, to win this contract without wasting any resources, needs to have 10 each brooding and milking mares and 10 second-year foals in his operation, plus as many of his third year foals as are late bloomers. That's an absolute minimum of 1000lb of high-quality fodder every day! Recall that you're only going to take 2200lb of grain or --at a very optimistic three mows a year, 12,000lb of green fodder off of a cultivated acre of land in a year. Assuming a long fallow, perhaps a total of two acres: an acre each of grain and green fodder every two weeks, week in, week out, for an entire year! That's the landholding (exclusive of paddock and stall space) of two prosperous farming families. Two hides, in the old Anglo-Saxon sense, before considering the maintenance of the labour. Four hides? Or perhaps a full quarter of a square mile of land.

These are pretty quick, off-the-top-of-my-head calculations, intended more to show scales of magnitude than to guide you in your new career as a stud farmer. Those scales should suggest that stud farming for an army is a big business. But this isn't even the beginning of it! Biometric archaeology has confirmed the ubiquity of the mule in the Roman world and shown that the legions were absolutely dependent on mule transport. Now that Lefebvre's experiments have been shown to be flawed, and we have discarded the idea of oblivious Roman teamsters and muleteers choking their desperate animals with primitive harnesses, we can conclude that the harnesses known tell us more about Roman land transport than their practical common sense. Specifically, there wasn't much call for heavy traction in the Classical period.

So, not much deep ploughing of heavy soils, nor big waggons and big horses on macadamised roads --this is probably not the place to start on the famous Roman road-- and not much in the way of military baggage that can't be broken down into mule loads.Also, no hay waggons rolling about. We're deep in "hystersis" territory, here. A stud farmer is a major consumer of fodder crops, and apparently a Roman farmer who gets into fodder crops is banking not only on selling his product to someone, but also on not ploughing the land again for another ten years.

The choice to go for alfalfa is about markets, and labour management, and capital. It requires "certainty" about the future. It boosts production, but without guaranteeing future markets that will give a return on capital.

So, leaving agronomy aside for the moment: what happened at Adrianople, specifically? The temptation is to start a long way back, and to see 411 as the culmination of a long-run crisis. On the other hand, we know a lot about what happened in 411 because of the literary renaissance of the Theodosian period, in which so many historians and other preserved writers (Palladius, for example) were active. Literary renaissances don't usually happen at the peak of the crisis. Pressed to pinpoint when they do happen, I would say, at the peak of an expansionary boom.

Still, David S. Potter wants to write a book about The Roman Empire at Bay, and chooses 180--395 as his timeframe.  Without going back to Commodus, Potter does make an arresting point by starting early. The details of the Severan civil wars suggest that the Roman army, the dominant insitution, the core of the state, the world-dominating colossus, has feet of clay, etc. It's not very good, is what he's saying. Search and destroy missions beyond the western frontiers aside.

Well, apart from waging civil war; which isn't that high an agonistic barrier, given that the legions are fighting each other. and it doesn't matter how useless they are. The Severans, who get into the first  Persian war in --a while, and this part isn't a particularly well developed part of the thesis-- try various things. At one point Caracalla even experiments with raising a few Macedonian-style, pike-armed phalanxes, which ancient writers tend to ridicule, but seems like a good, if premature idea. The pike isn't the queen of battle for nothing, but on the contrary because men with long pointy sticks beat men with short pointy sticks. You do have to move the pikes around, and the best solution for this of which I am aware is to equip pike companies with long-bed wagons, with the Severan army probably didn't have, and which would raise significant horse supply-related issues if it did. (The possibility that we should be talking about legions' combat effectiveness in terms of baggage trains brings us around to equids pretty quickly, of course.)

Potter moves, in his stately way, past mid-third century crisis (when the Roman legions definitelly couldn't fight Persians), to the army reforms of Diocletian-or-Constantine to new Persian wars, to Julian the Apostate's catastrophe. This is, he tells, us the practical end of his narrative, even if we have five endings to go before the Emperor Theodosius passes the harbour bar on his way to the West. No army, no way to beat the Sassanians, no revenues. Here, finally, is the crisis from which the Romans cannot recover.

Here, it bears repeating, is the oldest explanation of the fall of Rome: Persians lead to cavalry, cavalry leads to feudalism. Feudalism means no Roman Empire --oh, wait, no, what about the east Romans, I know, we'll change their name oh, look, 1453, I think we're done here.  Kaysar-i-Rum? Look, over there! It's Columbus and Petrarch, doing the Robot!

Fine. But whenever the imperial misadventures of our modern day go astray --often exactly where Julian's did-- modern people like to talk about how much money the "military industrial complex" is making, and darkly allege that the motivations for imperial war come from below, and have more to do with war profiteering than the alleged rewards of imperialism as such. David Potter prefers explanations in terms of the remoteness of the Emperor, which reduces the flow of patronage and leads discontented, mainly Gallic and British, armies declare rebel emperors. This kaisernahe explanation has had a good run in the modern history of the Holy Roman Empire, and I doubt that it is wrong, but "patronage" is just another way of extracting state revenues for one's own personal benefit. Persian war or civil war, both proft the Classical-era military industrial complex. Or the military-agricultural complex.

So we're post Julian. The new Roman army, we're all agreed, is inescapably bound to a higher ratio of horse-to-foot than the legion's twenty-to-one, which is actually entirely familiar from more modern eras as the proper ratio of divisional cavalry to divisional infantry. The huge masses of charging cavalry at Waterloo are organised into "corps," because they are corps --groups of divisions of exclusive cavalry. It is far from clear that at any point in the history of the legions, that the ill-defined forces of allied cavalry did not bring the ratios up to a much more modern level. Concede that they did, and you still have an argument about the actual ratio, which can be seen to fall as low as five-to-one, even though the "ideal" ratio in open terrain is probably two foot to one horse.

At the specific moment of Adiranople, it seems that the Romans and Goths had about the same number of men within a very generous range of between ten and fifteen thousand men, and that anywhere from three to five thousand might have been cavalry. These revised numbers leave us in no position to argue that the deaths at the Battle of Adrianople killed the Roman army.

The Roman economy, on the other hand. . . How many horses were lost at Adrianople? How many could the Roman economy replace? The early modern ratio of hands to horses is one-to-twenty at best, and that would give the Roman empire in its entirety 3 million horses, requiring perhaps 250,000 or so to replace annual losses, as working careers go on longer than military ones. Five thousand horses, if that is a good number for Adrianople, is a surprisingly large disruption in the normal economy of horse production.

And that is before we get to the famous Gothic wagon fort of Goths at Adrianople. Oh, you say. "Like Mongols!" And it is true that the nomadic peoples of the steppe, that nursery of cavalry, moved pasture in big wagons, whcih are known from the Roman Neolithic on, and, incidentally, if they had no other use in the west, would have been good hay wagons. However, it is all too easily forgotten that the yurt-dwelling, wagon-driving nomad of the Eurasian steppe is something of an aristocrat, lording it over fishers, hunters and farmers --and, more importantly, trading with them. It is not a self-sufficient lifestyle, in other words. The wagons exist because the nomads are producing a surplus for trade.

What's the Goths' excuse? Again, a point which can be generalised. Fragmentary and anecdotal notes tell me of a Roman army of 15,000 men and 540 wagons in 499; and that Attila's army retreated into a wagon fort after being defeated at Chalons-sur-Mer in 451. These might be quite small wagons, and, racking my brain for an explanation for their numbers and prominence, I wonder about the number of arrows which might have been carried along on campaign.

Finally, I want to move back to Potter's chronological framing. He sees an extended crisis from Julian's defeat through Adrianople. A more cynical reader might notice that Adrianople is described, in crucial respects, as similar to Cannae, but also to Decius' defeat and death at the Abritta. In contrast is Claudius Gothicus' first victory, and for propaganda purposes I do not even have to issue a caveat about Claudius only being known as the apical ancestor of the Constantinian dynasty from Constantinian propaganda. Beating the Goths is a Constantinians' job. No wonder that Valens, an opponent of the dynasty, is killed instead.

From that perspective, the cause and effect of Adrianople is internal. It is caused by Valen's personal deficiencies, and it brings the Theodosian dynasty to power; and the Theodosians reach out to the Constantinians, out of power since Julian.

And as for Julian, he is gone because he got an entire Roman army destroyed out there in Anbar Province in 363. Fifteen years is a long time for a crisis, but there are not lacking modern writers (Matthew Innes, notably), who think that Julian's defeat set in motion a recruiting crisis. For one thing, in this new age, landlords are unwilling to give up men to be soldiers. As Innes points out, the simultaneous expansion of "deserted lands," (Agri decumantes) makes this reluctance easy to understand. Even if it is hard to explain except in terms of depopulation, entangling us once again in a long-run explanation of Roman decline, since a falling population under the late Empire has to have roots going back a generation or more.

Except. . . As at Adrianople, the same question: How many horses were lost? As for the recruiting, another question: constant efforts to recruit "barbarians" from beyond the frontier do not explain where the barbarians' horses are coming from. If the're free roaming ponies from outside, so much for the future market of Roman stud farmers.

So I'm proposing a profound derangement of the Roman rural economy as a result of a relatively small excess of mortality in the army's cavalry horses. This makes more sense than might be apparent once the cost of raising horses is taken into account. It does not explain why future episodes don't have the same, deranging effect. The answer, it seems to me, is an agricultural economy turned around to producing horses. (Where's that "duh" giff. again?) It means horse traction, horse ploughing and new field regimes.

A feudal revolution, if you like.


*This also applies to dairying mares, so if  you like your koumiss, or just Beef Stroganoff according to Grandma Stroganova's old family recipe.