Wednesday, April 4, 2012

The Art of Not Being Governed, II: Flight Considered As a Lack of Litter

So there was a black-haired girl who belted them. She was the partner of the Governor of California, so I thought of her as a personification of the golden state of the west, and here's a song with gold and silver in it. Also, quite dispensably, the Eagles. 

Then something happened. I tell a story to myself about that, about how she's older now and doesn't like to sing so hard. It's obviously wrong, but blaming time and age works for me. It works for all of us; that's why we tell our little story about empires aging, albeit vaster and more slow. Rome: it's a story about how we all get older. Or, more likely, it's a compelling historical event that keeps getting turned into A Hugely Relevant Story About How Right Now Is Just Like The Fall of Rome, And We Better Quit It, Or We'll All Be Sorry.  That's a summary of this post, if you hadn't guessed.

Before I get to the meat of it, though, another one, just like that: 

Neo-cons have been getting English-speaking powers into stupid wars for so long that I think there was a Kagan who didn't get his job through nepotism pushing for an Afghan war in early 1878.  As always, things got sticky. A British force and accompanying journalists got to enjoy winter in Kabul in 1879—80 under semi-siege by  self-appointed "warriors of the faith" (ghazi). Taking time out from the little war, journalist Howard Hensman went to the Kabul Arsenal to find out  how the Afghans had been spending their large British subsidy. It turned out that the arsenal was making perfectly satisfactory steel breech-loading rifled Armstrong pattern guns. Extra Oriental colour was provided by the information that the guns were bored out, rifled and polished at the water mills Deh-i-Afghan, run by a “Hindustani” named Muah Khan, who learned his art from a “Negro named Belal,” who was in turn the apprentice of an Iranian who came to Kabul early in the century. The arsenal had to put in a major research and development effort to produce modern fuzes, but had much less trouble setting up a percussion cap factory, and I do mean factory. (Here, 322--24.)

As often in Afghan interventions, controlling Kabul just meant more trouble with Kandahar and Herat. When the occupation settled for backing Kandahar, it basically invited a  Herati invasion of the Helmand Valley by Herati General Hafizullah Naib at the head of 11 battalions of infantry and 32 guns. No Kandahar government, no matter where its masters kept their palace,could tolerate Heratis in the Helmand, so a British column of three infantry battalions, two cavalry squadrons, 2 half batteries and a pioneer field squadron set out to meet him, with

Hafizullah Naib managed to outmaneouvre the British, taking the granaries of the village of Maiwand. This would have allowed him to take winter quarters in the upper valley, so the British advanced to expel him, thinking themselves outnumbered by as much as 2476 men to 6000. (Hensman, 462ff.). Those were not, however, outlandish odds for Oriental warfare, so when the survivors of the battle returned to report over 50% British/Indian casualties, all were duly appalled. It came out right in the end, but people were left wondering how it happened, although the answer is in the general's account of the battle, describing an artillery duel of a battle won, not surprisingly given the disparity of force, by the Afghans.

It's just that this is not the answer that we have. The London papers didn't like this version, so they added a vast, vague hordes of 10,000 ghazi to the Herati army, and Lord Roberts later upped the  ante to 25,000. It's not that I'm saying that the ghazis weren't there, although I strongly suspect that they weren't. It's that we know what won the battle:  superior Afghan firepower produced by a not-far-behind-cutting-edge military industrial complex. The fact that fabulists turn it into a victory for a general with half the East at heel just shows how much the average pretentious commentator likes to quote Houseman.

People keep talking about when the "Great Divergence" happened, when the West gained that crucial tech level on the Rest and officially became H. Sap.'s vanguard of progress. (Mood music.) One person likes the French Revolution, another the Scientific Revolution, yet another the Reformation. My scholarly grandfather spent his career pushing the crucial date back to the Middle Ages, if not Constantine the Great. Jared Diamond wants to blame continental drift, near as I can make out.

Me? I'm saying that it can't have happened until after Maiwand. So, you know, all those attempts to explain the West's advantage "culturally"? Blah blah essentialism racism Sokal Hoax naturalistic epistemology blah. 

Okay? Glad I got that out of my system. What if we do this another way? By this I mean that instead of  arguing endlessly about culture and technology, let's get the hell into the tank with Ester Boserup and Nathan Rosenberg. I'm not going to argue, I'm going to propose. Let's treat technology as purely endogenous to the economy. Technological progress is nothing but capitalised learning by doing. Basically, the technological level of a given society is a function of its population, nuanced by the fact that while any work of the hand will come to be done more efficiently as people do more of it, capital has to be invested in the process to realise productivity improvements. This isn't even a claim about human nature; but rather about microeconomics, a field I'm so poorly equipped to argue in that it's not even funny. That's why I say, let's take it as a hypothesis. Let's drive our tank around the ruin of the Roman Empire today, and, if I don't get distracted in the mean time, around the longrun history of Europe next week. 

Walter Scheidel says that he can break out the surviving Roman data well enough to show that it fits model life tables compiled from early 20th Century data from several countries that combined good statistics with bad public health. Thus, Romans, and there is surprisingly little difference between rich and poor here, had a life expectancy at birth (female) of between 21 and 25, with men living on average three years more. Women had 6.5 completed pregnancies on average, corresponding to virtual population homeostasis (doubling time 350 years). Low remarriage rates built in a significant reserve fertility, but even raising the completed pregnancy rate by a full 1 only reduces the doubling time to 140 years. Conversely, raising the mortality rate only slightly, corresponding to a drop in female life expectancy at birth below 20, and population decline sets in. And it, unlike population growth, will be quite rapid. The fact that we have never seen this in a well-documented context, except in modern times and perhaps in the case of Ireland after the Potato Famine but before artificial contraception, is telling in itself.

Not to telegraph my point, we can certainly say that Roman population was hugely stable.  I know that I say this a lot, but it remains true. Given what we know about human frailties, exogenous demographic shocks have to be controlled and limited by internal processes, or there would be no humans at all.  I mean, it might be happening, but we can't allow the population of the Roman Empire to decline in the Antonine Plague, and in the fifth century, and in the Plague of Justinian. If demographic growth went into the negatives that frequently, our species would be extinct by now. That's something that happens. We've avoided it, so I'm saying that we are, so far, a fit and adaptive species. It follows, then, that apparent evidence for population decline at the end of the Roman Empire is  a priori probably evidence of something else.

This is a tough one, however. The historians who argue for population decline are pretty vehement on the subject, and they have strong evidence in the form of field walks that use surface scatter, mainly of pottery fragments, to assess the number of sites by period. Fewer sites datable to the fourth century in the West, and to the sixth/seventh in the East strongly suggests that the dissolution of Roman state power in each region is linked to demographic decline, perhaps preceding it, or during the episode itself. We generally assume that the decline is linked to a decline in Roman spending.

It is even hairier to try to estimate Roman GDP, but the fit with modern subsistence economies suggests that the Romans had a very low per-person surplus income to draw. Combined with our handwaves in the direction of totalling up Roman state spending, and there is good reason to think that virtually the whole surplus product of the Roman imperial economy went to funding the Roman army. Population decline would have had an immediate impact on the military budget. 

It was once fashionable to excuse this on the basis that the Empire was surrounded by ravening barbarians intent on invading the empire and destroying civilisation. I think, however, that the current trend is to call that complete horseshit. I mean, the Empire did get invaded a lot, but that wasn't the usual reason the army crossed the frontiers. Foreign wars were a particular strategy in an operational inventory deployed by one Roman political actor against another. 

What does that mean to us? That the army's where we're going to find economic/technological drivers, and they'll be most strongly at work when there is most opportunity for capital to fund skill improvements. This opens the way to reverse the causality. One way to do that is to make population decline a result of a fall in the military budget. Chris Wickham makes this argument. The fall of Africa blows a hole in the army budget, rendering it a less potent instrument of domination, allowing the peasantry to relax in its reproductive duty, since it no longer needs to sacrifice health and happiness to produce so many children as labour to meet the extortionate demands of the state. I think Scheidel, even if he hasn't addressed this directly, would not be impressed by flipping cameralism over to make libertarianism. (I'm also throwing a flag over that whole "A Vandal army of 80,000 including soldiers and families crosses the Strait of Gibraltar in 430AD and shows up in front of Annaba by August" play. However Carthage came to fall to the Vandals, at no point in our story as it meets final acceptance will there be room for a barbarian army to walk from Tangier to the Tunisian border in a single season. Or possibly at all.)

So what does a Boserupian hypothesis about the fall of the Empire look like? That a rising population, leading to new technologies, is undermining the Roman state. We have two candidates here: Horden and Purcell long ago promised to deliver us a technological revolution based on barrels. Look, guys, if an inability to prove this is holding up the second volume of Corrupting Sea, you have my permission to drop the argument. I'll buy the book if it consists of a thousand pages of "All work and no play make Peregrine and Nicholas dull boys."

So we're not getting the promised deep history of the barrel for now. Too bad. Could the barrel  have had a vast, disruptive effect on quotidian life in the west, one that we can defensibly situate around 376AD? Yes, barrels are known from as early as 100AD, and a set of cooper's tools dating to c. 100BC have been found, but the point of a Rosenbergian concept of technological change as learning-by-doing is that it  isn't enough for there to be an invention. It has to penetrate society cooper by cooper to have its full effect. It can only go at the rate of cooper training, and this is in itself an endogenous process. 

In the crudest analysis, the lifetime surplus generated by an individual cooper's productivity advantage over an equivalent investment in pottery-making has to suffice to cover the cost of training a replacement for that  cooper and at least part of the cost of training another lineage of coopers. This could take awhile unless some institution intervenes to divert surplus capital into training even more coopers. The army? According to this potted history of the barrel (we historians of technology take what we can get!), they're quite prominent on Trajan's column. So, yes, the idlers of the Roman Army had some chance to learn the fine art of the cooper. Can we push the argument, invent a massive intensification of the process in the mid-300s? Oh, heck, why not? As long as we're spinning a theory rather than pegging the facts to the vivisection table, we'll say that the Persian wars of the last half of the 300s, with their constant sloshing of tribal levies from Germany to Persia and back again are mobilising military logistics in unprecedented ways and, incidentally, training numerous coopers.

The old technology-changes-everything argument for the collapse of the Roman Empire in the east features the domestication of the camel. And, Lord knows, if there's one field of Late Antique history badly in need of shaking up, it's the story of the rise of Islam. (Hey, look, it's Patricia Crone, up to her old tricks again!) The idea that camel caravans might have been just coming into use in 600AD and disrupting the old Roman trade networks with catastrophic results is an old argument, and contested by ever-earlier dates for camel domestication. Now, I'm not enthused by any field where "Biblical archaeologists" show up. The Interweb crowing about how archaeology has disproven the atheists who think that the story of Abraham is wrong on account of including camels does not exactly inspire. On the other hand, I'm not going to argue with the archaeologists. If they push domestication back before the Iron Age, there's only so much ground you can make up arguing about the difference between domestication and close herding. 

Where you can go is over the same ground I've suggested with the rise of the barrel. It's not the existence of camel harness and saddles, but of a sufficient supply of good teamsters. Douglas Porch might be that most awful of things, a popular military historian, but his Conquest of the Sahara is a great account of just how hard it was for the French to project military power into the desert fastnesses. It proved impossible for the French to train cameleers. The skills were just too  hard to acquire quickly, and too pointless for a metropolitan Frenchman to learn, and the Sahara was eventually won by local cameleers adequately attached to the needs of the French state. Here's the flip side of Maiwand: Western imperialism being baffled by a technology that was beyond its economic/educational capacity to master. There are technologies that the West could not deploy, territories that were impenetrable to it. It's lucky for universal imperialism that the internal combustion engine came along when it did.

So we might have a story of technological disruption here, after all. Starting with Khosrau's attack on the Empire, or perhaps even a dimly-seen economic crisis lying behind Phocas' coup (and we shouldn't underrate the first overthrow of an emperor in the east in three centuries), something was going wrong in the East. Whoever the Qurayshi actually were (if we get in the tank with Crone as well as Boserup --it's a big tank!) they are the proximate agents of a rapid realignment of the Middle Eastern economy around the camel caravan. Camelry war -or whatever other mechanism will work-- leads to more and more camel caravans, and in turn to a total destabilisation of higher-level economic activities. The new alignment has its own new capitals, elites, and thus religion. 

I think I can get my head around that. What about barrels? The field walks that ostensibly demonstrate a decline in population in the west might, obviously, actually show people fleeing the state. Since they can't go to up into the mountains in any systemic way, the flight might take the form of a lighter and more transient lifestyle. A more extensive reliance on transhumant pastorage, seasonal fishing, or other economic niches that might deserve a more extensive accounting will do for this.

The problem is that we have a phenomonological problem here. The site surveys rely on potsherd series for their chronological framework. That is, we can identify the date of a site detected by surface scatter because the surface scatter is mainly potsherds, and we have a chronology of pottery styles that is so good that we can identify the date of a site by the style of the potsherds. As the quantity of surface scatter goes down, we not only cannot recognise an occupied site, but also have more trouble dating it. 

Another point arguing for the radical simplification or flattening of Roman society in the west is the disappearance of wheel-thrown pottery. It's even more telling that this happens in Britain by the first third of the fifth century and in Gaul north of the Loire in the latter third. About the time that the Roman army stopped operating in either place, and taxes stopped being levied. Okay, sure; but aren't the two related? The army can't very well operate where it can't tax. 

I'm belabouring this point because Chris Wickham makes  a huge point about the way in which the end of the taxing authority produced profound social change in its own right. He talks about the end of state-derived exploitation and verges on arguing the end of exploitation period. The critique of this libertarian argument (also present in Scott)  is that local elites can exploit quite as well as the distant state. Granted; but if the tax collector stops demanding payments in coin, the local farmer can stop participating in the market in order to acquire the coins with which to pay taxes. This is a net loss if the market is working in a neutral way to make everyone richer. If, on the other hand, the market is a rigged game where the farmer loses (which, I've argued, is what wall building and coin hoarding imply to me), we do have a story in which the end of taxes (and the state, and the Roman army) is a net gain. 

If I'm going to fit the barrel (and, of course, many other late-Roman technologies) into this story in a Boserupian way, the late Roman population has to be rising, not falling. The new technologies, however, have a disruptive effect. They make it possible for the population to escape exploitation. Sure, they may be living in shanties pitched in hidden hollows, high meadows, and firm spots in the midst of fenland rather than well-found huts round about London, but Scheidel's demographic evidence suggests that life sucked either way, so no great loss. 

So can technology be an alternative to James Scott's mountain habitats, his state-repelling geographies? Of course! In a Boserupian account, the tool kit required to exploit a habitat (state-free or not) is just a technology. That is, Scott's account of the history of the Zoma as the history of not being governed is a history of technology. Is a late-Roman living site with no potsherds, but only long-since rotted-away broken barrel or two a state-free geography, even if it happens to be a tinker's camp somewhere near London? I don't see why not.

*Arguably, with this move, all human history dissolves into a history of technology, and university history departments should fire everyone who is not a historian of technology. See, there's a reason that people talk about faculty politics!)   


  1. I think I'll be a little obnoxious here and throw you a bunch of questions all at once:

    Why would the Vandals have marched from Tangier to Annaba, when they could just hop back on their boats and hug the coast? Why bother with the marginal lands of Mauretania when you could head straight for the breadbasket of the empire?

    On which subject, all other things being equal and in the absence of fertiliser imports, export agriculture slowly impoverishes the soil by sending nutrients far away faster than they can be replaced by other processes. After several centuries of grain boats leaving Africa (and not, I assume, bringing back sewage), surely it would have been able to support fewer troops anyway?

    Given that camels were well-established in North Africa by the late Roman period - and trans-Saharan trade had already begun, at least on a small scale - how do they fit in to the collapse of Roman authority in Africa? It would be tempting to invoke military use, but there doesn't seem to be a lot of evidence for this (or so Mattingley says), and I suspect another factor is more important: with a camel herd, you can make a reasonable living on land too marginal for non-camel-owners to even seriously consider, which would certainly facilitate efforts to "escape exploitation."

    I don't see the problem with hypothesising frequent population downturns; what's to stop subsequent upturns from making up for them? Animal populations are strikingly prone to booms and busts, so it's not as if that's absurd on the face of it.

  2. On rereading, I see your argument about the downturns - but it still looks suspicious. In the aftermath of a downturn, we might expect improved hygiene in the cities (if part of the hygiene problem came from overcrowding) and improved nutrition (if there was a land shortage beforehand), and expect the downturn itself to disproportionately affect the weakest (assuming it results from epidemics / starvation / etc. rather than war.) So in the aftermath of a downturn life expectancies might well be a good bit higher than the long-term average, letting the population go back to normal faster. Dunno how applicable the conditions are here.

  3. On downturns, the argument I got at the breast argues that homeostasis requires a recovery mechanism from demographic crisis. Less crowding might mean better hygiene. (Not unrelated: giardia sucks.) Fewer people might mean higher living standards.

    Of the two, I'd probably go with the latter, only nuanced. Before considering population density, hygiene first requires social discipline. Do people poop in the woods if the camp boss doesn't get a latrine dug? Of course they do.

    Second, location matters. That's the lesson of the whole "malaria" issue, where people were aware of adverse health outcomes in regions with "bad air" long before they were able to disaggregate the causes and understand the different reasons that poor drainage leads to a sickly population. Nineteenth century engineers detected sickly towns and healthy ones without being able to move people out of the former, and Peter Fleming has tried to detect evidence of this same disparity of public health outcomes in sub-Roman cemeteries.

    The second claim is traditionally Malthusian. The land isn't producing enough food at peak population (and we'll imagine that we've made some kind of argument about labour inputs to cover the big hole here), so fewer people means more food. Okay, sure, but what kind of mouths have we lost? The epidemiological argument is that they're likely to be the old and the young, or unproductive mouths. That's an economic booster right there, in response to which the population mobilises its reserve fertility. Young widows, no longer needing to look after parents and children, remarry.

    So to make a long story short, we're not lacking potential mechanisms to explain homeostasis. We're not clear if people get to move up out of the fever swamps, but some version of the vulgar Malthusian account looks to be right.

    The issue here is the scale of possible deviations from the homeostatic mean. The Malthusian version, mainly gesturing at pre-Revolutionary America, assumes rapid population change. People (or, if we're faithful to the Reverend Malthus' account, Catholics) are capable of reproducing very quickly. What we know from the life table investigations is that this is wrong. That is why, among other things, we need an alternative explanation for the peopling of the United States.

    The further extrapolation is that historic population deviations can't be as great as they're meant to be in the sources. For epidemics, for example, we can hypothesise that they might have had the mortality rates as well as the huge social impact of better-documented Nineteenth Century epidemics. Out with 30% population mortality, in with 8%.

    None of this means that we can discard the hypothesis of a late Roman population decline. It just gets less likely, and, if it happened, less severe.

  4. Now that we've done the boring stuff, the cool stuff: those darn Vandals.

    First off, we can agree that if 80,000 Vandals got from Spain to eastern Algeria, they got there by sea. No horn helmeted blonds standing on a mountain in the Rif and waving their longswords in the direction of Carthage: "Come on, we've got to make it before Augustine dies!"

    The only profound improbability in the by-sea theory is that on the basis of sixteenth century warfare along the littoral, I doubt that Roman Hispania could have mustered the bottoms to lift an army of 80,000 (including camp followers) from Andalusia to Algiers. That is, however, a big improbability. We can, of course, allow a smaller army.

    Here's the bigger problem: by land to the number of 80,000 is what our sources give us. This is the classic problem with revising classic sources. Once we're flying alone, why are we tied to any specific element of the old narrative?

    If the Vandal army was quite small, then there's not much separation between the classical account and Pohl's Traditionskernel, a small group of Vandals around whom a larger ethnogenesis takes place.

    Once we have ethnogenesis, we have one of Scott's "art[s] of not being governed." It looks like the Vandals mainly cut off the delivery of taxes in kind to Rome. If those were particularly onerous, we have a quite sufficient explanation for why the indigenous North African elite might have Vandalised.

    Now, in the case of the collapse of the late Roman state, we have pretty clear evidence of the rise of inland "Moorish" states in resistance to Justinian's restored government. That looks like a case for camels. It also looks like a spontaneous, bottom-up repudiation of the late Roman economy. This is why I want to suggest that the "prosperity" of late Roman Africa is probably another of the Empire's rigged games. I'll talk about farming next time. I don't think that interregional nutrient balances are the way to go, but I do think that there's agrotech issues here.

    Also, horses.

  5. From my language-centric perspective, the fewer Vandals the better. As far as I know, no one has ever identified a single Vandal loanword in North Africa (Berber or Arabic). It's clear that some speakers of a Germanic language really did show up (, but they seem to have shifted to Latin rather quickly.

    The Moorish states were already showing up during the Vandal period, surely?

    You know, since you mentioned the peopling of the US again - you haven't made any allusion to the genetics of this, but as it happens, there are several haplogroups nearly or completely absent from Europe but common in Native Americans. So has anyone checked to see how much Native American DNA Yankees have?

  6. I think that there's some debate about what late Roman Moorish states would look like. Building mountains out of an epigraph that links one burial site to a guy actually mentioned in late Roman history, we can vaguely see one family presenting themselves as patricians to the Romans and as ...Moors(?) to other Moors?

    On the other hand, there's a push to see the Vandal state as an only barely not-Roman late Roman province, and the big changes being indicated by Procopius' lugubrious account of the troubles of the reconquered province. Again, there are some who accuse Procopius of exaggerating. When, in doubt, I turn to my copy of Wickham, where I'm told that the export infrastructure of African Red-Slipped pottery producers linked to oil farms only shows signs of recession at the end of the Vandal period. I like the idea of the camel pastures of the pre-Sahara as a refuge from the Roman state. But even if it holds up, I'm right now agnostic about when it started to function as such.

    And there have been mitochondrial studies of the proportion of Indian markers in White Americans. I've even found one on the Internet, but I'm not finding it now. IIRC (dubious -discuss), it showed that a minority of Americans have a varied proportion of native ancestry, and a majority have none. This looks like the result for Black ancestry as well, and confirms me in my theory (surprise, surprise!) that pur laine North Americans are descended from a multiracial ethnogenesis event, while most of the genetic lineages of modern white North America crossed the Atlantic under steam.

    I'll post a link to the study when I find it, but Combo C at Tang's Noodle House is calling me loud and clear right now.