Compromise and Conceit

My recent post on the case fatality ratio of the new Wuhan Coronavirus sparked a long discussion about the role of European epidemics in the colonization of the new world. There is a theory that after Europeans came to the new world (the Americas, Australia, etc) they brought with them diseases that went through the local populations like wildfire, killing huge proportions of the local populations because they were not previously exposed to these diseases, and so lethality was much higher and even simple diseases that Europeans were used to (like influenza) were highly destructive in these naive populations.

This theory sparked my statistician’s skepticism, and also my cynicism about colonial narratives. Europeans arrived in the Americas in 1492, an era not known for its highly advanced demography, and when they arrived counting the locals wasn’t their primary priority. Epidemiology wasn’t particularly advanced at that time either, and medicine incredibly poor quality, not to mention the difficulty of preserving accounts from that time. Furthermore, I don’t see any evidence that the mortality rates due to diseases like smallpox and plague have changed over time in western populations, and because our recent encounters (in the past 500 years) with immunologically naive populations have been very hostile it’s hard to believe that people bothered to adequately (let alone accurately) record what happened in that time, and it’s hard to imagine that there have been any actual, valid studies of immunologically naive populations in modern times.

Furthermore, there has been a major revisionist movement in the west in the past 20 years, which has tried to deny the reality of genocide in the Americas and Australia, and to cast the white invaders as innocent of any crimes, or at worst having made a few well-meaning mistakes. In Australia this has been spear-headed by Keith Windschuttle, whose Fabrication of Australian History series explicitly attempts to deny violence towards Aborigines and recast the destruction of Australian Aborigines as a consequence of disease and demographic decline. This has been pushed by national newspapers (The Australian, of course, fulfilling their role as propagandists for Satan) and our former prime minister, and its “success” has no doubt sparked similar narratives in other countries. There is even a counter-narrative in the Spanish world of the “Black Legend“, which dismisses claims of violence by Spanish conquistadores as propaganda by England and France. It’s very convenient for these people if they can claim that immunologically naive populations are especially vulnerable, and population decline due to violence is actually the consequence of disease. They can even claim that mass movements of indigenous populations occurred due to disease, not genocide. Handy!

This led me to ask two related questions:

1. Are immunologically naive populations actually subject to higher mortality rates when disease hits them?
2. Did disease kill the majority of the population in the Americas, and was that disease introduced by Europeans?

The first question can be answered by looking at the history of black death in Europe, and by genetic studies. The second depends on demographic and epidemiological data, and as I will show, there is none, and all the accounts are extremely dodgy.

The history of diseases in naive populations

A population that is naive to a disease is referred to as a “virgin soil” population, although it appears that this name is never used to describe European populations affected by the plague (which was imported from Asia) – “virgin soil”, along with terra nullius, is a concept reserved for the new world. In fact Europe was virgin soil for the plague in the 14th century, and experienced repeated and horrific epidemics of this disease from the 14th century to the 16th century, with smaller plagues later on. In total the black death is estimated to have killed 30-60% of the population of Europe, and to have precipitated huge social changes across the continent. That was 700 years ago, and yet today the case fatality rate due to plague remains 60%, so 700 years of exposure to this disease hasn’t changed European susceptibility at all.

We can also see this in influenza. The H1N1 epidemic of 2009 killed only 0.01% of people who caught it, even though it was a new strain of influenza to which people could be expected not to be immune. The Spanish flu probably killed 10-20% of people it infected, but it did not do an especially greater job in isolated communities who had never experienced influenza before. For example in Samoa it probably killed about 20% of the population, having infected 90%, which suggests it did not behave particularly egregiously in an unexposed population. Smallpox, which has existed for 10,000 years in humans, had a similar mortality rate over most of its history, with variations in this mortality rate primarily driven by the number of people infected and the quality of the healthcare system. There is some evidence that the mortality rate is lower in Africans, who had been exposed to it for longer, but if so this has taken 10,000 years to manifest, which suggests that in general infectious diseases do not behave differently in “virgin soil” populations, though they can be much worse in populations with inadequate health care or infection control methods.

It’s worth noting that many estimates of the impact of these diseases rely on extremely dubious estimates of population. Putting aside demographic methods of the 14th century, Samoa in 1918 was a colony managed by New Zealand, with a colonial management so incompetent that they allowed people to disembark from a plague ship flying a yellow quarantine flag, and then mismanaged the resulting epidemic so badly that everyone on the island got infected. Did New Zealand’s colonial administration have any incentive to accurately count the population before the epidemic? Did they accurately register newborns and elderly people, or did they only record the working age population? How good were their records? If the Samoa population is underestimated by a small amount then the mortality rate plummets, and conclusions about the effectiveness of the disease in this naive population are significantly changed. And was the population even naive? Were the NZ colonial administrators previously recording every influenza epidemic on the island?

These problems are an order of magnitude worse when we try to understand what happened in native populations.

How many Spaniards went to Mexico?

Accounts of the effect of epidemics depend ultimately on our knowledge of the population affected, and population estimation is a very modern science. How was this done in 15th century America, by people who were busy slaughtering the people we now wish they were counting? What was the variation in population estimates and who was recording population, how and why? Fortunately we have a partial answer to questions about how population was recorded, because a historian called David P. Henige wrote a book called Numbers from Nowhere: The American Indian Contact Population Debate, much of which can be read on google books, that makes a lot of strong criticisms of recording of population at that time. Sadly his specific chapters on over-estimation of epidemics are not available online, but he does provide an analysis of accounts by Spanish reporters of the numbers of Spanish soldiers present at certain actions on the continent. As an example, he reports on the number of deaths recorded during the noche tristes, an uprising in the city of Tenochtitlan in which the Aztecs rose against their Spanish occupiers and slaughtered them, driving them out of the city. Spanish accounts of that event – by people who were there – record the number of deaths as between 150 and 1170, with Cortes (the general in charge) recording the lowest number. Henige also notes accounts of expeditionary forces that vary by up to 10% between reporters who were on the scene, and may not even mention Indian attachments that probably far outnumber the Spanish forces. He reports on a famous Spanish reporter on the continent (las Casas) who misreports the size of the continent itself by a huge amount, and notes that a room that was supposed to be filled with treasure as tribute was given radically different sizes by different Spanish observers, as was the amount of treasure deposited therein. He also notes huge discrepancies (up to a factor of 10) in population estimates by colonial administrations in north America. He writes

If three record books showed Ted Williams lifetime batting average as .276, .344 and .523 respectively, or if three atlases recorded the height of Mt. Everest as 23,263 feet, 29,002 feet, and 44,083 feet, or if three historical dictionaries showed King William XIV as ruling 58 years, 72 years and 109 years, their users would have every right to be thoroughly bemused and would be justified in rejecting them all, even though in each case research could show that in each case one of the figures was correct. Yet these differences are of exactly the same magnitude as those among the sources for the size of Atahulpa’s treasure room that Hemming [an author reporting this story] finds acceptable

These are all relatively trivial examples but they make the point: almost nothing reported from the colonies in the 15th century was accurate. In the absence of accurate reporting, what conclusions can we draw about the role of infectious diseases? And what scientific conclusions can we draw about their relative mortality in virgin soil populations?

Scientific estimates of epidemic mortality in Latin America

A first thing worth noting about scientific reports of epidemic mortality in the Americas is that they often use very old sources. For example, this report of the environmental impact of epidemics in the Americas  cites McNeill’s Plagues and Peoples (1977), Dobyns’s estimates of population from 1966 and 1983, Cook’s work from 1983, and so on. It also relies on some dubious sources, using references extensively from Jared Diamond’s 1997 breakout work Guns, Germs and Steel. Some of these works receive criticism in Henige’s work for their credulity, and Diamond’s work has been universally canned since it was published, though it has been very influential outside of academia. Many of these works were written long before good computational demography was well established, and though it’s hard to access them, I suspect their quality is very poor. Indeed, McNeill’s seminal work is criticized for using the Aryan population model to explain the spread of disease in India. These works are from a time before good scholarship on some of these issues was well established.

Dobyns’s work in turn shows an interesting additional problem, which is that no one knows what caused these epidemics. In his 1993 paper Disease Transfer at Contact, (pdf) Dobyns reports on many different opinions of the diseases that caused the demographic collapse in south America: it may be smallpox, or plague, or Anthrax, or typhus, or influenza, or measles. Dobyns’s accounts also often note that people survived by fleeing, but do not ever consider the possibility that they were fleeing from something other than disease. Contrast that with accounts from north America 400 years later (such as the story of the Pince Nez reported in Bury My Heart at Wounded Knee), which make clear that native Americans were fleeing violence and seeking sanctuary in Canada. There is a lot of certainty missing from these accounts, and we need to be careful before we attribute population decline to disease if we don’t know what the disease was, and are relying on accounts from people who refused to consider the possible alternative explanations for the social collapse they are witnessing.

This is particularly complicated by recent studies which suggest that the epidemic that wiped out much of the Mexican population was actually an endemic disease, that jumped from local rats to the indigenous population, spread from the mountains to the coasts (not from European coastal settlements), and had symptoms completely unrelated to European diseases. In this account, a long period of drought followed by rain triggered a swarm of a type of local rat into overcrowded settlements of native peoples, where a type of hantavirus jumped from those rats to humans and then decimated the population. The disease started inland where the drought had been worse and spread outward, and it primarily affected indigenous people because they were the ones forced to live in unsanitary conditions as a consequence of slave-like working conditions forced on them by the invaders. Note here that the western invaders, presumably completely naive to this disease, were not affected at all, because the main determinants of vulnerability to disease are not genetic.

Further problems with the epidemic explanation for native American population loss arise from the nature of the transatlantic crossing and the diseases it carried. The transatlantic crossing is long, and if anyone were carrying smallpox or influenza when a ship left port the epidemic would be burnt out by the time the ship reached the Americas. In fact it took 26 years for smallpox to reach the continent. That’s a whole generation of people slaughtering the natives before the first serious disease even arrived. During that time coastal populations would have fled inland, social collapse would have begun, crops were abandoned, and some native communities took sides with the invaders and began to work against other native communities. In 9 years of world war 2 the Germans managed to kill 50 million Europeans, several millions of these due to starvation in the East, and created a huge movement of refugee populations that completely changed European demographics and social structures. What did the Spaniards do in 26 years in central America?

It is noticeable that many of the accounts from that time seem not to account for flight and violence. Accounts at that time were highly political, and often reported only information that served whatever agenda the writer was pursuing. Las Casas, for example, whose accounts are often treated as definitive population estimates, appears not to have noticed massive epidemics happening right in front of him. Others did not notice any possible reasons why natives were abandoning their fields and farms, and didn’t seem to be able to consider the possibility that something scarier than disease was stalking the land. The accounts are an obvious mess, with no reliable witnesses and no numbers worth considering for serious study.

Conclusion

Without good quality demographic data, or at least even order of magnitude accuracy in population estimates, it is not possible to study the dynamics of population collapse. Without decent information on what diseases afflicted local populations, it is impossible to conclude that “virgin soil” populations were more vulnerable to specific diseases. There is considerable evidence that disease mortality is not different when populations are naive to the disease, drawn from European experience with plague and global experience with influenza, and there is no solid evidence of any kind to support the opposite view in indigenous populations. Historical accounts are fundamentally flawed because of their subjectivity, lack of accuracy even when their interests are not threatened, and the unscientific nature of 15th century thought. A whole generation of conquistadores acted with extreme violence before dangerous diseases arrived on the continent, so many accounts of population collapse must reflect only war, but even after the diseases arrived it is likely that they were no more dangerous in native populations than they were in Europe, which by the 16th century was experiencing endemic smallpox that regularly killed large numbers of people (in Europe in the 18th century it killed 400,000 people a year). There is no reason to think that the Americas were special, or that their local population was especially vulnerable to this or any disease.

It is important to recognize that these issues – accurate diagnosis of disease, accurate estimates of numbers who died, and accurate population numbers – are not just academic exercises. You can’t put them aside and say “well yes, we aren’t sure what disease did it, how many people died, and what the population was, but by all accounts it was bad in the colonies.” That’s not how epidemiology works. You would never, ever accept that kind of hand-waving bullshit when applied to your own community. Nobody would accept it if the Chinese government said “yeah, this coronavirus seems bad, but you know there aren’t that many people affected, the population of Wuhan is anywhere from 1 million to 20 million, and we don’t even really know it’s not seasonal influenza or smallpox.” You would rightly reject that shit out of hand. It’s no different when you’re talking about any other population. We have no reason to suspect any special impact of epidemics in the Americas or Australia, and no reason to conclude that they were especially influential in the history of those regions compared to the violence inflicted on the locals – which we know happened, and we have many accounts of. To look at the accounts we have of disease in the new world, and conclude anything about them beyond “it happened” is to put undue confidence in very, very vague and very poor reporting. There is no empirical evidence to support many of the claims that have been made in the past 40 years – and especially, by genocide deniers, in the past 20 years – about the role of disease in the destruction of indigenous populations of the new world.

This matters for two reasons. First of all, it matters because it has interesting implications for how we think about the threat of disease, and how new diseases will affect naive populations when they jump from animals to humans (which is how almost all new diseases start). These diseases can be extremely dangerous, killing 30-60% of the affected people in some cases, but the reality is that for them to become pandemics they need to mutate to facilitate human-to-human transmission, and that mutation significantly reduces their mortality rates. It is rare for a disease that transmits easily to also be dangerous, and there is very little in the history of the human race to suggest otherwise. The Spanish flu pandemic of 1918 is perhaps the sole exception, and if so it should show just how rare such events are. We should, rightly, be concerned about coronaviruses, but we should also not expect that just because we’re naive to them they’re going to be extra dangerous. Diseases do what they do, and that is all.

But more importantly, we need to reject this idea that the catastrophe that unfolded in the new world between 1492 and 1973 wasn’t the fault of its perpetrators, white Europeans, and we need to reject even partial explanations based on epidemics. It was not disease that killed the people of America and Australia. There is no evidence to suggest it was, and a lot of reasons to question the limited evidence that some people present. The epidemic explanation is a nice exculpatory narrative, which tells us that even if white Europeans had approached the people of the new world with open minds and hearts in a spirit of trade and collaboration they would still have been decimated by our diseases. In this story we may have done some bad things but it doesn’t matter, because contact was inevitably going to destroy these fragile and isolated peoples. And this story is wrong. It isn’t just uncertain, it is wrong: there is nothing in the historical record to support it. If white Europeans had approached the new world in this spirit, there would have been a generation of trade and growth on both sides before the diseases struck, and then we could have helped them to escape and overcome the diseases we were familiar with, that were no more dangerous to them than they were to us. Their communities would have been better prepared to resist the social consequences of those diseases because they would not have been at war, and would not have been experiencing social collapse, overcrowding, starvation and poverty because of western genocidal policies. They would not have been forced into overcrowded and desperate accommodation on drought-stricken plains as slaves to Spanish industry, and the homegrown epidemic of 1545-48 would not have affected them anywhere near as badly. It’s important to understand that the tragedy that befell native Americans was caused by us, not by our diseases, and our diseases were a minor, final bit of flair on a project of destruction deliberately wrought by western invaders.

This other story – of diseases we couldn’t help but strike them down with, even if we had been pure of heart – is a genocide denier’s story. It’s self-exculpatory nonsense, built on bad statistics and dubious accounts of native life presented by biased observers. It is intended to distract and to deny, to show that even if we did a few bad things the real destruction was inevitable, because these frail and noble savages were doomed from the moment they met us. It is a racist narrative, racist because of its false assumptions about native Americans and racist because of what it assumes about the balance of mortality in the continent, racist for trying to pretend that we didn’t do everything we did. It is superficially appealing, both because it adds interesting complexity to an otherwise simple story, and because it helps to explain the enormity of what Europeans did in the Americas. But it is wrong, and it is racist, and it needs to be rejected. There is no evidence that epidemics played a major role in the destruction of native American communities, no evidence that native Americans were especially vulnerable to our diseases, and nothing in the historical record that exonerates European society from what it did. White Europeans enacted genocide on native Americans, and just a few of them happened to die of some of our diseases during the process. European society needs to accept this simple, horrible fact, and stop looking for excuses for this horrible part of our history.