Experimental History

A few weeks ago I teamed up with my friend David Fitzgerald once again to talk about historical method for a gaggle of godless kids at Camp Quest West. David and I dressed up in silly costumes and did a skit or two. One girl loved my hand-made ivy crown so much I was happy to let her have it after the show, but sadly we were so busy we forgot to get any pictures. Oh well. Anyway, the gist of our presentation was that the scientific method also applies to the field of history, and in fact history is really just another science, with its own peculiarities like every other field.

In more "typical" sciences, people go to the lab and run experiments, or go out and make observations of the phenomena they want to document or explain. But history is all in the past, so you can’t run experiments on it, or observe it anymore. So how do you apply the scientific method to it? Well, the same way, actually. Of course there are already explicitly historical sciences (geology, cosmology, paleontology, archaeology), so obviously "it's all in the past" hasn't stopped scientists before.

But human history presents its own peculiar problems. And yet, so does every science. Psychology faces completely different methodological challenges from particle physics, as particle physics does from zoological anatomy, and thus the actual methods in each branch of science will vary greatly. But they all share the same basic structure that I outline and discuss in my book Sense and Goodness without God (pp. 213-26), and they share many common elements of method besides that. Ultimately, human history is closer to criminal forensics in scientific genre, though still a field all its own, certainly in terms of the methodological toolbox needed to "do history" scientifically.

In short, all a scientist really does is adduce a theory or hypothesis about what happened in any given case (an experiment, a daily observation, whatever), then deduces from that what else would be observed it if were true (and, just as importantly, what should not be observed if it is true, or in other words, what would be observed it it were false), and then we go out and make many more observations, and from all the evidence we get, we inductively determine the relative probability of the hypothesis being true. Then we subject our work to peer review and replication by other experts, to confirm our observations and conclusions are valid and genuine, until we develop a consensus of experts in the field. The whole process aims at reducing the probability of deception and error in our conclusions. And all the other little refinements in method or technique, at each step, in every distinct field, are directed to that same end.

History works the same way. A historical event, like an atom or magnetic field, can't be observed. We can only observe its effects, and then infer the underlying facts, more or less precisely and with more or less certainty, depending on how much information we have access to. Like atoms and magnetic fields, a historical event causes things to exist that would not have existed otherwise. Evidence is left behind or destroyed. The course of events turns out one way rather than another. By observing the effects of an event that survive (what we call "historical evidence") we can infer what caused that evidence, in other words, what happened in history.

Of course, the biggest difference is that in most history, especially ancient history, we have a lot less data than any other science enjoys access to, even the more historical of them (like geology or cosmology). But this only means we will be a lot less certain of the results, and must often be a lot less ambitious in what we can claim to know about our past, but once that caveat is embraced and understood, everything works out the same way. To claim that a certain collection of evidence does not prove a particular historical claim, is in effect to propose an alternative theory of how that evidence came to exist, just as in every other science.

The kinds of evidence available to historians vary in nature and reliability, from each other and from the kinds of evidence available in other scientific fields. I discuss this at a fairly basic level in my book (pp. 227-52). But just as "laboratory experiments" are what most people imagine when they think of the "scientific method," so "research" (literary and archaeological) is what most people imagine when they think of the "historical method." Yet in fact much of science is built on observations that cannot be controlled or reproduced in any lab. Astronomy, for example. Or geological theories about the formation of the Grand Canyon. Likewise, there are (more or less) controlled experiments in the field of history.

I'm going to talk about the latter today, giving more examples than I did at camp, because I've run into so many lately I just have to talk about them! In "experimental history" you basically recreate a historical circumstance and see what happens, and from that you can infer things about what happened in the past. This employs the same general principle that science also depends on: the argument from analogy. What works yesterday works the same today, what works on Earth works on Mars, what works in London works in Paris, what works today worked the same a thousand years ago. A physicist would not demand that Newton reproduce a demonstration of his laws of motion in Paris on the assumption that physics works differently in London, nor do we demand that Newton's experiments be replicated every morning on the assumption that the laws of physics can change any day.

We accept this because we have abundant reasons to believe such deviations are highly improbable. Sometimes, it is true, analogies don't hold. Yet because a violation of this rule is so rare, you need really good evidence to believe things were different or have changed before you can conclude that any given analogy doesn't hold. The rate of fall of dropped objects is indeed lower on Mars than on Earth, and the elevation of the planet Mars is indeed different in London than it is in Paris. Yet we've accumulated plenty of evidence confirming that the circumstances in these cases are relevantly different. In fact, once we add those differing circumstances into our calculations, the corresponding differences in result vanish, and the analogy holds after all. Newton's laws are actually the same on Mars as on Earth, and in fact this explains why objects fall faster here than there. And once we take into account the sphericity and rotation of the earth, the actual celestial declination of Mars turns out to be identical with respect to Paris or London. And so on.

So, too, in history. I have already discussed this in some detail (see my 2006 essay What Can We Infer from the Present about the Past?). So I won't belabor the point. Let's get to some examples of experimental history instead, since that's the interesting bit. Of course these all relate to my own field, ancient history, and ancient science and technology in particular, since that's what I know well. There are no doubt examples in other fields.

What the Hell is a Tettarakontareme?

At Camp Quest I opened with the example of the Trireme Project. Around the 5th century B.C. the Greeks invented the trireme, which revolutionized naval warfare and contributed to the rise of the Athenian empire, and the subsequent defeat of Persia, two events that were rather decisive for us secularists, since without them the Age of Greek Rationalism might never have happened, and the development of modern Science, Logic, Human Rights, and Democracy might have been delayed considerably, possibly indefinitely, since these all subsequently began and originally grew only in Greece.

But in all this lay a historical problem. No trireme has ever been recovered, and surviving depictions and descriptions of them from antiquity are frustratingly incomplete or imprecise. So what exactly was a trireme? We know the word means something like "three rower" and that triremes were propelled by 170 oarsmen and about 120 feet long and employed a ram as their principal weapon. From all this it was reasonably surmised that "trireme" meant there were three decks of oars, one above the other, with six rows of oarsmen in all.

But soon after that a naval arms race began. There is a lot of interesting stuff about that arms race, and the best scientific discussion of it all is in Foley and Soedel's 1981 article "Ancient Oared Warships" in Scientific American (April issue). But here the troubling problem was that bigger and bigger warships were built and fielded. First came the quadriremes (four-rowers), then the quinquiremes (five-rowers), then sextiremes (six-rowers), and on and on, all the way up to the monster of them all, the tettarakontareme, a "forty rower." What the hell!?

Obviously, the argument from analogy operates here well enough to conclude that there is no damned way they fielded a ship with 40 decks of oars. So if that's not what the "reme" refers to, then are we also wrong about what a trireme was? And what, after all, is a tettarakontareme? Enter experimental history. First a hypothesis: the 1981 work of Foley and Soedel tentatively solved the problem by applying science to it and proposing (rather persuasively) what these ships really were like. But then how do you test it? Well, you build one.

In the late 80's and early 90's the Trireme Project was launched, and hypothetical trireme designs were realized, actually built and sailed. This allowed historians to experimentally test a trireme's maneuverability, speed, weight, and other important data, and to rule out some things, and rule in others. Ultimately, the tests proved that the trireme probably did have three vertical banks of oars, but that adding any more banks would have been increasingly unmanageable and unlikely. Meanwhile, the theories of Foley and Soedel made even more sense on the same evidence. So these experiments lended strong support for them.

Their solution was that after the trireme, larger ships kept the same basic structure of three decks of oars (which in a trireme were manned by six rows of oarsmen, one man to an oar), but then added rows of oarsmen. So the quadrireme put two men per oar on the top deck, to give added power where it was most needed, then the quinquireme put two men per oar on the top two decks, and so on. All this increased the size of ship that could still be propelled at an effective ramming speed, which allowed larger complements of catapults and marines to be carried, and that changed the course of battles.

Then the largest ships started putting three men per oar, and then four, sometimes more, which mechanically required installing steps for some oarsmen to march up and down to work their end of an oar. With this information, historians took a second look at literary descriptions of the tettarakontareme and concluded it must in fact have been a giant war catamaran, with two hulls connected by a battle bridge, all propelled by three decks of oars on each side of each hull, up to eight men to every oar. This monstrosity never saw battle. Though it did see service, its function was more to serve as a deterrent: the fact that the king of Egypt had it was intended to scare the shit out of everyone else.

The Antikythera Computer

In 1900 sponge divers recovered the contents of an ancient shipwreck dating to the 80's B.C., whose cargo included crusted bits of geared machinery. This is in fact what my avatar is an x-ray of (as I explained in a previous blog entry Why the Bizarre Avatar?). But this also offers an example of experimental history: historians figured out what this machine was by building one. They observed the parts that were available (including Greek inscriptions on some of the parts, which turn out to have been something like an instruction manual) as data to infer or rule out hypotheses about all its surviving, broken, and missing parts, and their arrangement. Then they built some examples and experimentally confirmed what it had to have been: an astronomical computer for predicting solar and lunar eclipses, lunar phases, solar positions in the zodiac, and probably planetary positions as well.

Horse Shit

A lot of experimental history has been done with horse equipment. Two I know well are harness experiments and cavalry experiments.

In the first case, it had long been claimed (and is still, in the writings of historians not up on their game) that the ancient system of harness impeded economic and technological progress, because it choked horses and thus prevented the exploitation of the greater power afforded by them, over mules, donkeys, and oxen. Although it is often not mentioned that horses consume a lot more fuel than these, so the decision which to use did not simply hinge on harness technology. Nevertheless, the Romans have been called stupid for not having figured out the inefficiency of their harness, while the invention of the modern horse collar has actually been credited as an inevitable result of Christianity (I won't bore you with the twisted logic behind that one).

Well, J. Spruytte figured this was so much armchair horseapples, and actually made an ancient horse harness and tested it. Turns out it doesn't choke horses and in fact does not impair their pulling power at all. Judith Weller gives a good online summary of this and other issues related to ancient harness in Roman Traction Systems. The horse collar may have afforded certain advantages, but none that would be at all decisive in the progress of civilization. Western history did not turn on its invention. So much for that theory.In the second case, one question people often wonder at is how the hell the Greeks and Romans fielded such effective cavalry without a stirrup. The cavalries of Trajan or Alexander the Great, or even the Persians, were legendary, effective enough to decide an entire battle, and heavily relied on tactically for nearly a thousand years, before anyone thought, "Hey, you know what would be a good idea...?"

Lynn White, who also came up with the Horse Collar of Christ theory (and many other inept boners like it), also hugely exaggerated the importance of the stirrup in the course of Western history, assuming that somehow the medieval heavy cavalry that depended on it were somehow a miracle (again of Jesus, in a roundabout way) that were essential to the rise of the Western world. Scholars who know better have taken him to task for such an absurd claim. Indeed, if he had seen how easily heavy cavalry could be defeated by mere sticks in the movie Braveheart, he might have thought twice about his silly idea. The Romans had so decisively defeated waves of war elephants (a far scarier military unit than stirruped heavy cavalry) with the similar tactic of the caltrop, that the elephant ceased to be employed at all, ever again, in the history of warfare. I don't imagine the legions would have had any greater trouble with mounted knights.

Anyway, stirrupless cavalry were certainly effective, since evidence of their effective use in antiquity is quite extensive. Hey, I hear tell the American Plains Indians fielded formiddable cavalries that kicked some major ass, without even saddles. Maybe had it not been for guns and germ warfare, things might have turned out differently for them? At any rate, the question remains: how much does not having stirrups affect a rider in combat, and what kinds of combat does this affect?

Well, Ann Hyland found out...by experimenting. She set a horse up with Roman gear and attempted various cavalry maneuvers with it and reported on the results. She found that a strirrup does afford some advantages, though not as many as once thought (and in fact it presented some disadvantages--like getting caught in them). Most importantly, she also tried out a Roman four-horned saddle we know had been invented by the early 2nd century A.D. from extant military reliefs, and found something surprising: it negated almost all the disadvantages of fighting without stirrups. Though it would not have helped mount a medieval-style heavy cavalry, it would have increased the fighting effectiveness of light cavalry to the same level the stirrup allowed. Thus, in fact, the Romans didn't need Jesus to invent an improved method of mounting cavalry.

And the Rest

I could add many other examples like this. In the 60's A. Gerhardt Drachmann built numerous ancient machines and instruments according to the available literary descriptions of them, in order to test their functions and properties, as well as the manufacturing knowledge and abilities of ancient scientists and craftsmen. Michael Lewis recently did the same for ancient astronomical and surveying instruments, ascertaining their accuracy, for example, and matching it with the known accuracy of Roman construction projects (such as some remarkably precise inclines built into aqueduct systems). John Riddle tested the efficacy of ancient drugs according to ancient pharmacological treatises, finding, for example, that they had in fact discovered several effective birth control drugs and abortifacients. John Healy tested some of the more suspect chemical claims in Pliny the Elder's Natural History, proving that, contrary to previous assessments, Pliny actually knew what he was talking about.

Others have done something similar in the textiles industry, building various models of ancient looms and testing their products against surviving samples of ancient cloth, to ascertain what sort of loom technology they had, and perhaps even date its development. Recently I read a study of ancient wine technology in which historians reconstructed ancient wine production methods in an actual vinyard, according to surviving technical manuals from the time, and then ascertaining the productivity and efficiency and other features of these ancient practices. But one of the most interesting examples is an ongoing experiment in ancient shaft furnace design, in which ancient smelting technology is being tested for its production and efficiency attributes, teaching us things about the choices made regarding labor and economics in the Roman iron industry, and aiding in the interpretation of archaeological finds, such as determining the amount of waste slag produced per furnace. Students every year man the furnaces and smelt iron, accumulating more and more data. Fun!

Some cases can be rather fanciful. I saw an episode of Myth Busters where they tested, using only technology known to have been available at the time, whether Archimedes could have burned attacking ships with parabolic mirrors. Parabolic burning mirrors were well known at the time and scientific treatises were written about them and the scientific basis for their operation. Though the literary evidence already confirms that Archimedes' use of them as a weapon was a medieval legend, testing whether it was still possible by conducting experiments (in Archimedes' Death Ray) still qualifies as doing experimental history.

But my favorite example is an episode of NOVA (Pyramid). Unfortunately this was years ago and they've done so many episodes since with a similar title, I can't find the exact one I saw anymore. But they went to Egypt with a bunch of historians and archaeologists to test their various theories of how the blocks were lifted into place, and all met with considerable difficulties, until an Egyptian stone mason walked by and was astonished at how silly their ideas were, and then showed them how it was done, blowing them away with a very simple, effective method.

Intriguingly, that might be somewhat moot now. It has recently been proposed that the pyramids were at least partly constructed from a pre-Roman form of poured concrete, and the evidence for this is somewhat impressive. I'll wait and see if this theory survives peer review and persuades the experts, but one thing is certain: experiments will certainly play a part in confirming or refuting it, or fleshing it out.