Escape From Rome

by Walter Scheidel

  After the 680s, the Umayyad regime moved away from sinecures, tying receipt of tax income to actual service in a standing army. For a while, the Abbasids levied taxes even more vigorously to fund court and military, and Baghdad became the focal point of a more centralized fiscal network. Later Abbasid caliphs increasingly relied on outsiders—Turkic- and Persian-speaking cavalry—to hold down local elites and extract revenue. This gave rise to ethnic conquest regimes that carved up the empire along the fault lines generated by earlier fiscal regionalization. In the Iraqi core, taxation did not collapse until the tenth century, when the military elite began to obtain income directly from farmers.72

  The original Arab conquests had established a pattern that continued over the long run: successive regimes of regional armies were superimposed on and parasitic upon more continuous civilian government. Under the first two caliphates, local elites of the Sasanian period retained their estates and performed fiscal functions for the center. And even when soldiers eventually gained the right to extract income (iqta‘) from individual bits of land, the central authorities usually retained sufficient fiscal capacity to protect their power in the face of widespread attempts to privatize allocations and seek shelter from taxation by ceding land to influential patrons.

  In no small measure, this was made possible by the fact that iqta‘ remained restricted to military beneficiaries, whereas civilian landowners, denied similar entitlements, continued to work with state agents. Unlike in medieval Europe, where landownership coincided with military power and central government was crowded out by localized interests, medieval Middle Eastern urban wealthy elites lacked military muscle and iqta‘ remained a gift of the state: together, these constraints kept public service an attractive option and limited the breakup of states to regionalization instead of wholesale devolution into smaller units. Moreover, recurrent conquest from the outside facilitated the reallocation of gradually privatized military grant holdings.73

  Overall, starting conditions were logically consistent with different outcomes. In medieval Western Europe, tax erosion coupled with land assignments and localization of power resulted in intra- and interpolity polycentrism and low state capacity. In the Arabo-Turkosphere, maintenance of fiscal infrastructure and enforcement of service obligations sustained reasonably strong regional states run by detached conquest elites. In China, the rebuilding of the tax state and centralized control paved the way to large-scale empire and greater consolidation in elite circles.

  Conditions in South Asia are less well known but likewise conformed to this template. The universalist aspirations of the ancient Maurya empire have to be taken with a lot of salt, and it is quite clear that the Gupta empire, at its peak in the fourth and fifth centuries CE, relied heavily on tributary kings it sought to contain with the help of resources drawn from the Gangetic core. Southern India remained politically fragmented. The resultant regionalization of fiscal structures meshed well with the dominance of multiple regional states in the second half of the first millennium and the early second millennium. It was only the later Sultanate of Delhi and the Mughal empire that pursued more centralizing objectives, albeit ultimately with limited success. State formation oscillated between Middle Eastern–style regional states and larger empires of the Roman, Chinese, or Ottoman mold.74

  The fact that specific configurations of fiscal and military institutions correlate with polity scale—fragmentation in Europe, unification in China, and intermediate outcomes in the Middle East and North Africa region and South Asia—suggests that the fiscal “sinews of power” played a major role in shaping the geopolitical landscape. The causal linkages behind this correlation are straightforward: at the end of the day, it was simply not possible to build large empires without sufficient centralized bundling of energy inputs in the form of material resources (taxation) or military labor (conscription).75

  In China, imperial restoration was invariably driven by military action: an effective, unified military organization came first. Even if some Chinese dynasties later relaxed demands, they always started out with strong centralized assets. In post-Roman Europe, by contrast, fiscal decay drained the state of its strength. It took many centuries for intensifying competition to revive broad-based taxation. But by then, as we saw in chapter 6, competitive fragmentation had already become too deeply entrenched to give way to hegemonic empire. Instead, state capacity had been divorced from, and indeed become antithetical to, that very concept: a new type of extractive dynamic, focused on interstate conflict, intrastate integration, and strategically developmental policies, had been born. Both traditional empire and taxation had to wither first to make this possible.76

  So far, my account has tracked specific events and developmental sequences, focusing on what we might call proximate causes: the interplay of preexisting conditions and the preferences of particular conquest regimes, modes of military mobilization, and the resilience of fiscal extraction. These factors were critical in skewing outcomes in a particular direction, toward or away from large-scale empire. Yet it is one thing to identify such variables and assess their impact. It is another altogether to account for their relative prevalence. The persistence of certain trends in the long term, from 1,500 to more than 2,000 years—enduring fragmentation in post-Roman Europe, serial imperial reconstitution in China, and more hybrid outcomes elsewhere—raises the very real possibility that the institutional and organizational features that sustained these robust trends were themselves embedded in and molded by more fundamental preconditions. Chapters 8 and 9 explore the question of what these ultimate causes might have been.

  CHAPTER 8

  Nature

  GEOGRAPHY

  GEOGRAPHY IMPOSES basic constraints on the scope and scale of human social interaction. State formation is no exception. Going back at least to Charles-Louis de Secondat, Baron de La Brède et de Montesquieu, observers have sought to explain European fragmentation and Chinese unity with reference to the nature of the terrain. Back in the middle of the eighteenth century, the baron declared:

  In Asia they have always had great empires; in Europe these could never subsist. Asia has larger plains; it is cut out into much more extensive divisions by mountains and seas; and as it lies more to the south, its springs are more easily dried up; the mountains are less covered with snow; and the rivers being not so large, form smaller barriers.1

  That much of this is dubious—what about the Himalayas, or the Yellow and Yangzi rivers; and don’t rivers connect rather than divide?—ought not deter us from developing this line of inquiry. One does not have to be a geographical “determinist” to acknowledge that the physical environment matters—and by now we can do better than Montesquieu.

  Articulation: Coastlines

  In much of Europe—especially in its Latin western parts—land and sea are entangled in complex ways. Jared Diamond was not the first to consider its “highly indented coastline,” coupled with multiple peninsulas, conducive to political fragmentation: these features are simply the inverse of Montesquieu’s “much more extensive divisions” of Asia. East Asia’s coastline is smooth, with Korea as its only significant regional peninsula. Islands in close proximity to the mainland—Hainan and Taiwan—are much smaller even than Ireland, and Japan and the Philippines, each of them larger than the British Isles, are farther away.2

  Even though critics of geographical perspectives on history cannot deny that this is true, they doubt its relevance. Thus, in his account of “why Europe conquered the world,” Philip Hoffman notes that Europe’s islands were not immune to naval invasion, and its peninsulas did not develop into coherent states earlier than other regions. However, his reference to Italy as an example of intense fragmentation during much of the post-Roman period is qualified not only by the stability of the Kingdom of the Two Sicilies in the more isolated south but more importantly by the precocious unity of Roman Italy. Moreover, the key issue is not so much unity within peninsulas as their relationship to larger imperial formations: it is wo
rth noting that with the partial exception of the Mongol period, Korea, the only major peninsula in East Asia, was never fully ruled from China. More generally, coastlines did have a discernible effect on state formation: Britain, Ireland, Denmark, Italy, Sweden, and even France (more on which below) are all well delimited by the sea.3

  But these are just details. Cherry-picking is not helpful. Once we address this question in systematic quantitative fashion, Europe does indeed emerge as a serious outlier within the Old World. In this respect, David Cosandey’s work has produced striking results. Almost half of the surface area of what he labels “Western Europe”—generously defined as Europe west of what used to be the Soviet Union—is located on peninsulas and another tenth on islands. Conversely, the aggregated peninsular and insular shares of China, India, and the Middle East and North Africa region range from 1 percent to 3.6 percent.4

  Because of this, Europe’s coastline is much longer than that of East and South Asia: 33,700 kilometers for “Western Europe” as opposed to 6,600 kilometers in China and 7,300 kilometers in India. This in turn means that Mandelbrot’s fractal dimension—an index of complexity bounded at 1 (lowest) and 2 (highest)—is higher for “Western Europe” (1.24 and 1.42 without and with islands, respectively) than for China (1.13 and 1.26) and India (1.11 and 1.19). The latter two are overall more compact than Europe west of Eastern Europe—a landlocked region eventually subsumed within a single very large land empire, Russia. Even without factoring in other features, Western—Latin—Europe’s relative physical complexity should have made it more likely for stable and smaller polities to emerge there than elsewhere.5

  Integration: Mountains and Rivers

  Mountain ranges likewise contribute to physical segmentation, and ruggedness more generally imposes additional cost on communication. Although it has been claimed that Europe is not overall more rugged than China, it is the relative intensity of compartmentalization that matters most. The Alps, Pyrenees, and Carpathians are relatively high compared to mountain ranges that can be found in China east of Tibet: the first two generally rise above 1,500 meters. Moreover, well less than half of Western Europe lies less than 300 meters above sea level, mostly England, Ireland, northwestern France, northern Germany, and Poland, as well as the Po Valley (figure 8.1).

  By contrast, “core China” (excluding Tibet, Xinjiang, Inner Mongolia, and Manchuria) enjoys conditions that favor greater connectivity. Much of that area lies less than 300 meters above sea level, and elevations of 1,500 meters are rare, confined in the first instance to the northwest where the state of Qin developed sheltered “within the passes” (figure 8.2). Core China is an excellent illustration of Montesquieu’s “much more extensive divisions by mountains.”6

  Mountains helped define the territorial features of European state formation, not only for the aforementioned islands and peninsulas but also for France, whose Pyrenees borders have been quite stable since the Frankish period but which throughout the Middle Ages and into the early modern period was unable to absorb its more mountainous southeast. Chinese development, by contrast, was framed above all by its largest rivers, which created two very extensive basins that are not separated by major natural barriers and could be connected even with sixth-century CE technology. As a result, to quote Diamond one more time, China “very early became dominated by two huge geographic core areas of high productivity, themselves only weakly separated from each other and eventually fused into a single core.”7

  FIGURE 8.1   Altitude profile of Europe.

  FIGURE 8.2   Altitude profile of East Asia.

  This sanguine statement may seem to give short shrift to several rather important points. There already had to be a powerful empire before the task of fusing could be undertaken by building the Great Canal. In the long term, maintaining the canal system required even bigger efforts. The lack of tidal rivers other than the Yangzi hampered connectivity along the coast. And whenever China splintered—up to the twelfth century CE—it did so between these two core areas. Fusion took hard work and was not a given.

  Nevertheless, the Chinese river basins forcefully support the notion that geography matters. Relatively modest physical obstacles between them had a palpable effect on state formation and could only be overcome by centuries of repeatedly enforced political unification. If that was the case in China, would we not expect more serious obstacles to have had an even greater impact on Europe?8

  Moreover, China’s riverine interconnectivity was only the beginning. The Yellow River was navigable for about 600–800 kilometers inland, similar to the Nile below the First Cataract, and the Yangzi for 1,100 kilometers east of the Gorges. Between and beyond them, the Great Canal, multiple smaller rivers, and feeder canals created, in the words of environmental historian John Robert McNeill, “a huge fertile crescent united by cheap and safe transport.… No inland waterway system in world history approaches this one as a device for integrating large and productive spaces.”9

  This was not a one-way street from geography to empire: it was a dialectical process in which the physical environment and state formation, both contingent and acting upon each other, fostered ever-stronger path dependence. Thus, again according to McNeill, “The durability and resilience of the state depended in large part on Chinese geography, but Chinese ecology in turn depended on the state to an unusual degree.” The imperial state assumed responsibility for big waterworks and flood control. Canals, dam, and paddies required constant maintenance, and the more invested, the more at stake. In this precociously anthropogenic environment, decay would have been very costly and not readily reversible, unlike in rainfall-fed Western Europe. In consequence, Chinese society became “unusually dependent on demographic and political stability, and unusually vulnerable to disruption by neglect.… No other major society so locked itself into a situation demanding constant intensive maintenance to prevent sharp ecological degradation.”10

  Even as statements such as these might seem to propound an updated version of Karl Wittfogel’s model of hydraulic despotism, they help us identify the features that contributed most to long-term development. Unlike in China, rivers played a less dominant role in Europe. The two longest ones, the Danube and the Rhine, are shorter than the Yellow and Yangzi Rivers and connect less of Europe. They also flow in different directions, thereby raising the logistical penalty on moving toward the central regions.11

  Concentration

  The cumulative effects of coastlines, ruggedness, and river basins support a simple observation: Europe consisted of multiple smaller core regions whereas China initially had just one—the Central Plain—and then two, with the Yangzi basin added into the mix. Increasingly interconnected, the northern basin consistently remained politically and militarily dominant.

  As I argue in the next section, proximity to the steppe was part of the reason for the latter. But other natural endowments played a similarly important role. The Central Plain was not only flat but highly productive. Loess deposited by sandstorms up to 250 meters deep in the west—in Shaanxi, Western Shanxi, and Southern Gansu—was then carried by the aptly named Yellow River into the Central Plain where it boosted agricultural yields.12

  Whoever was in charge of this core also controlled China: it is not by accident that virtually all imperial unifications proceeded from this area. Its demographic dominance was similarly pronounced: for a long time, it towered over other parts of China not only in population number but also thanks to the state’s ability to count and burden its concentrated and accessible residents.13

  No such “natural” core existed anywhere in Europe. It was divided into a number of smaller pockets of development, alluvial plains separated by mountains (the Alps and Pyrenees), marshes (in the northern Low Countries), forests, and the sea (around the British Isles and Scandinavia). Following in the footsteps of others, Eric Jones made much of this configuration: the Paris and London basins, Flanders, and the Po Valley jointly yet severally represented the High Middle Ages. Montesquieu’s
“larger plains” of Asia are conspicuous by their absence.14

  It is easy to overestimate the impact of these differences. Rome, after all, had succeeded in uniting the aforementioned zones alongside others all the way to the Nile valley. And we cannot take it for granted that these conditions would predictably result in a particular type of state system: Jones’s observation that “enough states were constructed each about its core and all of a similar enough strength to resist the logical conclusion of the process of conquest and amalgamation: a single unified European state” is mere description, not an explanation.15

  Moreover, large plains as such were not a sufficient factor. Had they been, empire ought to have been more durable and hegemonic in India than it turned out to be in China. Huge northern river basins created a farm belt from Pakistan to Bihar as early as the first millennium BCE. As Victor Lieberman notes, by 1700 the North India plain contained some 60 percent of India’s total population, compared to the 25 percent or 30 percent of the Central Plain’s share of China. India was made up of fewer and smaller micro-regions than China. However, it lacked an integrating water-support system like China’s that would have allowed the northern state to expand south. Connectivity mattered as much as plains did.16

  Interspersion of arid and cultivated zones farther south added to South Asia’s fragmentation, and although the western deserts of Sind and Rajastan and the arid central Deccan plateau were not obstacles as formidable as the major mountain ranges of Europe, they did foster separation. The more difficult terrain and the dispersion of agrarian zones in the south correlate well with the prevalence of smaller polities there. Empire-building invariably proceeded from the north to the south, and southern state formation was often secondary to that process. In addition, the arid Deccan areas supported warrior communities and armed pastoralists that were capable of challenging the northern basin empires, in ways unknown in China. In sum, these traits help account for intermediate outcomes in terms of political centralization: more than in Europe but less than in China.17