My interest in robustness and contingency accounts for the relative prominence in this book of another tactic, overt consideration of counterfactuals. In a very basic sense, counterfactual reasoning is a necessary ingredient of any historical account that seeks to rise above the level of bare description: there is, after all, “absolutely no logical way to make causal inferences without simultaneously making assumptions about how events would have unfolded if the causal factors we consider crucial had taken on different form.” Thus “we are all counterfactual historians”—and that “we” covers pretty much every person, not just professional historians.37
Even so, historians all too rarely highlight counterfactual reasoning in their research. This is a great loss. Explicit counterfactuals force us to confront the weaknesses of deterministic as well as revisionist assumptions, however implicit they might be: the notion that deviations from what happened might have proven short-lived and some approximation of actual outcomes would have happened anyway, or, conversely, that minor contingencies could have produced massive divergences from observed history. Merely to think about this makes us more careful about causal inferences. Just like comparative history—of which counterfactual history is a more exotic variant—“what-ifs” are a valuable means of assessing the relative weight of particular variables.38
The key question must be this: How little change would have been enough for history to have taken an alternative path—in the nontrivial sense of altering outcomes enough to be visible and to make a difference in developmental terms? Procedurally, this question calls for adherence to what has been called the “minimal-rewrite rule”: the least amount of tweaking of actual history and avoidance of arbitrary intervention.
Ideally, the direction of a counterfactual change should preserve “consistency with well-established historical facts and regularities, consistency with well-established generalizations that transcend what is true at a particular time and place, and consistency with well-established laws of cause and effect.” This does not rule out recourse merely to space aliens and asteroid impacts but also to historical actors that display anachronistic or contextually implausible behavior. The closer the change hews to what could well have happened at the time—the more informed the counterfactual scenario is by what actually did happen—the more reasonable it is.39
In devising counterfactuals, it is essential to be clear about putative connections—to specify antecedents and consequents—and to ensure that connecting lines are logically consistent. One problem in particular is difficult to avoid in practice: counterfactuals inevitably generate second-order effects that complicate the prediction. The more they add to the complexity of counterfactual scenarios, the lower the overall probability these scenarios will be compared to that of any given link within them: the whole exercise becomes more tenuous and frail. Although the problem of complexity can be a function of design—if, for instance, we introduce multiple changes at once—more commonly it is simply a function of time: the farther we project ahead of actual history, the less we are able to control the thought experiment. Counterfactuals work best in the short term.40
I follow best practice in identifying critical junctures at which things either might well have gone differently or would have needed to have gone differently in order to generate significantly different long-term outcomes (that is, before a particular trend had become firmly locked in). Unlike much of the existing literature, however, I do not start with some ostensibly plausible change to explore its likely ramifications. Instead, I ask, as I must, how much would have had to go differently at a certain point to bring about change on a large scale—in this case, either the abortion of Roman expansion (chapter 4) or the restoration of Roman-style empire in post-Roman Europe (chapters 5 and 6).41
That these are very substantial divergences from actual history makes it easier to judge their plausibility because they often tend to be incompatible with the dictates of the minimal-rewrite rule: if it is not feasible to obtain dramatically different outcomes without straying far from what might plausibly have happened at the time when the counterfactual change is made, historically observed developments are revealed as having been fairly robust. This robustness helps contain the ever-present risk that we design counterfactuals that support our own preconceived notions of what was likely to have happened.42
The odds of Rome’s failing to build a mighty empire steadily declined as time went by: whereas early changes could have derailed it from this trajectory, at later junctures it becomes more challenging to devise plausible pathways to a significantly alternative reality. The same is true for post-Roman Europe and the ascent of modern development in terms of economic growth and scientific and technological progress: trends that up to 1500 might quite readily have been aborted became more difficult to change during the following two centuries. In the eighteenth century, this exercise would require even more dramatic rewrites, and it becomes well-nigh impossible in the nineteenth. The real question is just how much this trend owed to compound and reinforcing effects, and how far these effects reached back in time: my own answer, of course, is that their roots were very deep indeed.43
Comparison is essential for establishing the European anomaly (chapter 1) and for explaining it (chapters 7 through 9) and its developmental consequences (chapters 10 through 12). Counterfactuals are essential in testing the robustness of what happened (chapter 4) and what did not happen (chapters 5 and 6) in Roman and post-Roman Europe, respectively. The epilogue takes the counterfactual approach even further, sharpening our appreciation concerning what exactly it is the modern world owes to the ancient past.
The final result is a book that is quite varied in content and perspective, moving as it does back and forth between ancient history, modern history, comparative historical sociology, and history that did not even happen. For this and other reasons it is bound to irritate: classicists and humanists of all stripes for giving short shrift to the (positive) legacy of the classical world; culturally and microscopically inclined historians by focusing on the big picture of state formation and economic development; even more historians by foregrounding the influence of ecology and geography; and most historians by being irremediably “reductive.” I might even be taken to task for eschewing conventional indictments of the “West,” or indeed of the very concept—a label for which I have little use. For balance, I also expect to annoy social scientists by dredging up proverbial ancient history and by relying on a great deal of qualitative reasoning.44
This is exactly as it should be. While it may be rare for such diverse elements to be brought together between two covers, that is the whole point of the exercise—to forgo business as usual and to experiment. Those challenging my argument will have to do so by drawing on a similarly broad canvas—or better still on an even broader one, or by showing why the canvas is too broad, or the wrong one. Any such critique will inevitably have to wrestle with a familiar conundrum: how to go about addressing the Very Big Question of why the world has turned out the way it has (so far). The more productive disagreement my book arouses, the better it will have done its job.
Yet needless to say, disagreement is not what I am after. I wrote this book to establish, as firmly and comprehensively as I could, two simple points: that interlocking forms of productive fragmentation were of paramount importance and indeed indispensable in creating the specific set of conditions that gave birth to modernity, and that the divergences that precipitated this outcome in only one part of the world but not in others were highly robust. In the end, only Western Europe and its offshoots fit the bill: had our “Great Escape” not begun there, it would most likely not have happened at all.
PART I
The European Anomaly
CHAPTER 1
Patterns of Empire
MEASURING IMPERIAL DOMINANCE
HOW COMMON was it for empires to dominate large parts of entire continents? How rare was it for imperial hegemony to end in persistent fragmentation? Just how unusual was the
European experience of state formation? All these questions call for comparative evaluation. They can only be answered by examining how often empires swept all before them—or at least came close.
This is harder than it might seem. The boundaries of empire are notoriously difficult to define. Imperial rule was sometimes highly indirect in nature, relying on vassal regimes to manage subordinate populations. On occasion, imperial centers lost effective control over areas that nominally remained under their authority to local elites or warlords. We need to simplify: my survey is predicated on the assumption that an empire was formally unified for as long as no overtly independent polities had emerged in its territory.1
But even if we accept this fairly generous definition of imperial rule, we still the face the problem of how to measure the degree of imperial dominance in a particular region in a way that allows for systematic comparison across several continents and millennia. Existing scholarship on the scale of empire focuses on the territorial size of polities, on the amount of land they controlled and how this changed over time.2
And not without reason: space is undoubtedly a critical variable. Fernand Braudel famously labeled distance the “number one enemy” of civilization. Far-flung empires continuously struggled to maintain communication and to exercise power over diverse terrains: their very survival depended on this.3
Even so, territorial reach may not be the single most important factor in assessing the relative weight of particular imperial formations. If we want to understand the role of hegemonic empire in shaping social, economic, and intellectual development, it is people who matter most. Geographical reach was not always matched by demographic heft: whereas very populous empires generally tended to be fairly large, not all large empires were necessarily endowed with sizable populations.
Steppe empires in particular sometimes extended over thousands of kilometers and multiple modern time zones without holding sway over more than a small fraction of the subjects of smaller agrarian polities, and acquired demographic weight only if they expanded into more densely settled areas of sedentary civilization. This phenomenon was not limited to nomads: in 1815 the Russian empire ruled half of the terrestrial surface of Europe but only a quarter of all Europeans; and a hundred years later it claimed more than a third of Asia’s landmass but fewer than 3 percent of its people.4
I thus focus on population as the most meaningful measure of imperial success. The most basic approach is to measure (or rather estimate) the proportion of the total population of a particular macro-region that was ruled by the most populous polity in that region. The larger its share in the overall regional population, the closer the leading polity came to enjoying effectively monopolistic power, and the more often or the longer this was the case, the more dominant was the role of hegemonic empire in the history of that region.5
Anyone who is even vaguely familiar with historical demography will appreciate that this seemingly straightforward exercise is something much easier said than done. The population size of early societies is not normally reliably recorded or otherwise empirically known. Census counts such as those that have survived from early China are rare and not without problems. For the most part, all we have are estimates—which are often little more than guesses—by modern scholars who have not followed consistent standards. I explain my own method and the steps I have taken in order to ensure a measure of consistency across space and time in the technical note at the end of this volume.6
I must stress that the following calculations are to be understood as a very rough guide to demographic conditions in the past—and the farther we move back in time, the rougher. Values that differ by a few percentage points are just as plausible as the ones in my charts. Even so, they are useful as long as it would seem impossible to envision significantly different levels of imperial dominance (of, say, over half instead of a third of a given regional population): my estimates are generally unlikely to be wrong to an extent that would affect the overall shape of the pattern. It is in this narrowly circumscribed manner that they provide a reasonably solid foundation for global comparison.
I have identified four macro-regions: Europe, the Middle East and North Africa (MENA), South Asia, and East Asia. A fifth one is a hybrid that overlaps with the first two, namely the area once claimed by the Roman empire at its peak, which I refer to as the “Roman empire region” (figure 1.1).7
Taken together, these areas had long been the core zone of human macro-social evolution. They cover more than a quarter of the earth’s land surface but house a much larger share of global population, around 60 percent today. Their demographic predominance was even more pronounced in the past: 2,000 years ago, these four regions housed at least nine out of every ten human beings alive at the time, and about four in five both 1,000 and 500 years ago.8
They differ in size, albeit only to a moderate degree. At 10.18 million and 11.84 million square kilometers, Europe and East Asia form spatially equivalent units. Due to the presence of substantial arid areas that did not play a major role in historical state formation, the MENA region is more difficult to define: its nominal size of 12.59 million square kilometers could easily be reduced by about one-third by bracketing out the Algerian, Libyan, and Egyptian deserts and the Arabian Rub’ al Khali. (While other regions also feature marginal terrain, such as the Gobi Desert or subarctic Europe, they do so on a comparatively smaller scale.) South Asia covers 4.53 million square kilometers, similar to the Roman empire that claimed between 4 million and well over 5 million square kilometers depending on how much desert area is taken into account. In terms of size, the five regions under review thus vary by a factor of between two and three.9
FIGURE 1.1 Macro-regions of state formation.
Much the same used to be true with respect to population: figure 1.2 shows that although East Asia was generally more populous than any of the other four regions, it was only rarely more than twice as populous. The MENA region of the past 1,000 years has been the only exception.10
This broad similarity in orders of magnitude ensures that we do not end up comparing apples and oranges by lumping together regions of greatly different size or population. For this reason, substantially smaller regions such as Southeast Asia, Mesoamerica, and western South America have not been included here; I briefly consider them later on.
FIGURE 1.2 The population of South Asia, Europe, the Middle East and North Africa, and the “Roman empire region” as a proportion of the population of East Asia, 200 BCE–2000 CE, at centennial intervals (East Asia = 1). Source: Based on McEvedy and Jones 1978.
PATTERNS OF EMPIRE
The Mediterranean and Europe
Over the long run of history, the position of the Roman empire within the area it had come to claim at the height of its power in the first few centuries CE remained unique. It had been without precedent: in the fifth century BCE, the Achaemenid Persian empire had ruled no more than 30 percent of the population of the same region. Having thoroughly dominated the scene from the first century BCE to the end of the fourth century CE, Roman preeminence ceased only after an abortive attempt at restoration in the sixth century.
Subsequent efforts to rebuild a “Roman” empire proved unsuccessful: Charlemagne managed a modest resurgence around 800 while Ottonian rule produced merely a minute uptick two centuries later. Ottoman expansion in the early modern period likewise failed to improve on the Achaemenid population share of 2,000 years earlier, and later blips under Napoleon and Hitler remained both relatively modest and utterly ephemeral (figure 1.3).11
FIGURE 1.3 The proportion of the population residing in the area covered by the Roman empire at its peak that was claimed by the largest polity in that area, 450 BCE–2000 CE (in percent).
KEY: Achaemenid empire: 450, 400, 350 BCE; Ptolemaic empire: 300, 250 BCE; Roman empire (Seleucid empire): 200 BCE; Roman empire: 150, 100, 50 BCE, 1, 50, 100, 150, 200, 250, 300, 350, 400 CE; Byzantine empire (Western Roman empire): 450; Byzantine empire: 500, 550, 6
00, 650; Umayyad empire: 700, 750; Frankish empire: 800, 850, 900, 950; Fatimid empire: 1000, 1050; Holy Roman Empire (France): 1100, 1150, 1200, 1250; France: 1300, 1350, 1400, 1450, 1500; Ottoman empire: 1550, 1600, 1650, 1700, 1750; France: 1800, 1812, 1815, 1850, 1900, 1933; Germany: 1943; United Kingdom: 1945, 1950; Egypt: 2000.
Roughly 40 percent of the subjects of the mature Roman empire resided outside Europe. Even so, it is clear that the observed pattern was in the first instance the result of developments in Europe rather than in the Levant or northern Africa. The level of Roman imperial dominance in Europe was unparalleled. At the peak of their power, the Romans controlled between three-quarters and four-fifths of the European population, even though they claimed no more than a quarter of the continent’s land surface.12
After the disintegration of the western half of the empire in the fifth century CE, the corresponding share of the eastern Roman empire plummeted to 20 percent or 30 percent. Ephemeral consolidation under Charlemagne was followed by an entire millennium of persistent polycentrism as the most populous power did not normally rule even a fifth of all Europeans. Neither Napoleon nor Hitler managed to match the reach of the Romans even for a few years, and Russia’s ascent up to the mid-twentieth century was eventually checked by the collapse of the Soviet Union in 1991 (figure 1.4).
FIGURE 1.4 The proportion of the population of Europe claimed by the largest polity in that area, 250 BCE–2000 CE (in percent).
Escape From Rome Page 4