Escape From Rome Page 5
KEY: Roman empire: 250, 200, 150, 100, 50 BCE, 1, 50, 100, 150, 200, 250, 300, 350, 400 CE; Western Roman empire: 450; Ostrogothic kingdom/Visigothic kingdom: 500; Byzantine empire (Frankish empire): 550; Frankish empire: 600, 650, 700, 750, 800, 850; Frankish empire: 900, 950; Holy Roman Empire: 1000, 1050, 1100, 1150, 1200, 1250; France: 1300, 1350, 1400, 1450, 1500; Spanish Habsburg empire: 1550, 1600; France: 1650, 1700, 1750, 1800, 1812; Russia: 1815, 1850, 1900, 1933; Germany: 1943; Russia: 1945, 1950, 2000. In 1933, 1945, 1950: Russia = Soviet Union.
Demographic change was insufficiently dramatic to call this very long-term perspective into question. Even though the three-quarters of Europe outside the former Roman borders gradually filled up, they never quite caught up with the original core and even today account for only a little more than half of the European population. For most of the post-Roman period, up to the sixteenth century, the proportion of all Europeans who lived in the territories once held by Rome did not differ greatly from that back in antiquity (figure 1.5). Thus, the distribution of population in medieval and early modern Europe was not somehow profoundly different from that in Roman Europe: continuities outweighed shifts over time.
FIGURE 1.5 The population of the area claimed by the Roman empire at its peak as a proportion of the population of Europe, 200 BCE–2000 CE (in percent).
The Middle East, South Asia, and East Asia
The pattern of one-off near-monopolistic empire followed by enduring polycentrism that can be observed in the area once held by the Roman empire and more specifically in Europe is completely different from conditions in the other three macro-regions. Over the course of the past 2,500 years, the MENA region went through four distinct phases of imperial consolidation, most notably under the Achaemenids and the Umayyad and early Abbasid caliphates. Roman and Ottoman rule resulted in a slightly lower degree of demographic concentration, made similar by the fact that it extended over more or less the same territories.
Two further imperial projects failed, that of the Sasanians in the early seventh century (after defeat by the Romans) and of the Seljuqs in the late eleventh century (due to political fission). Although levels of imperial preeminence did at times drop to post-Roman European levels, most notably in the period between 1100 and 1500, renewed concentration took place until the dismantling of the Ottoman empire ushered in a new state system that is already showing signs of strain (figure 1.6).
FIGURE 1.6 The proportion of the population of the MENA region claimed by the largest polity in that area, 700 BCE–2000 CE (in percent).
KEY: Assyrian empire: 700, 665, 650 BCE; Egypt: 600 BCE; Median empire: 550 BCE; Achaemenid empire: 500, 450, 400, 350 BCE; Seleucid empire: 300, 250, 200, 150 BCE; Parthian empire: 100 BCE; Roman empire: 50 BCE, 1, 50, 100, 150, 200, 250, 300, 350, 400, 450, 500, 550, 600, 610 CE; Sasanian empire: 626; Roman empire: 630; Umayyad empire: 650, 700, 750, 800, 850; Tulunid empire: 900; Roman empire: 950; Fatimid empire: 1000, 1050; Seljuq empire: 1092; Fatimid empire: 1100; Fatimid empire (Rum Seljuq empire): 1150; Ayyubid empire: 1200; Mamluk empire (Mongol empire): 1250; Mamluk empire (Ilkhanid empire): 1300; Mamluk empire: 1350; Mamluk empire (Timurid empire): 1400; Mamluk empire: 1450, 1500; Ottoman empire: 1550, 1600, 1650, 1700, 1750, 1800, 1850, 1900; Egypt: 1950, 2000.
A similar pattern can be found on the Indian subcontinent. Once again we are able to identify four episodes of hegemonic empire, under the Maurya, the Sultanate of Delhi, the Mughals, and the British empire that eventually gave birth to the modern Indian state (figure 1.7). Compared to the MENA region, dominant empires were less durable and for much of history alternated with formations that claimed approximately half of the overall population of South Asia, alongside intermittent periods of more intense fragmentation.
It is worth emphasizing that the population estimates for this region are fraught with margins of error that are particularly large even by the modest standards of this exercise. For the earlier stages of South Asian history, my reconstruction is bound to underestimate the relative demographic weight of the northern cores of the region that were located in the Indus and Ganges basins.13
FIGURE 1.7 The proportion of the population of South Asia claimed by the largest polity in that area, 500 BCE–2000 CE (in percent).
KEY: Maghada: 500, 450, 400, 350 BCE; Nanda empire: 325 BCE; Maurya empire: 300, 250 BCE; Satavahara: 200 BCE; Shunga empire: 150, 100 BCE; Saka empire: 50 BCE, 1 CE; Kushan empire: 50, 100, 150, 200; Gupta empire: 300, 350, 400, 450; Gupta empire/Hephthalites 500; Harsha empire: 647; Chalukya of Badami empire: 650, 700, 750; Pala empire: 800; Pratihara empire: 850, 900; Rashtrakuta empire: 950; Chola empire: 1000, 1050; Chola empire (Western Chalukya empire): 1100; Chola empire: 1150; Ghurid Sultanate: 1200; Sultanate of Delhi: 1236, 1250, 1300, 1350; Vijayanagara empire (Bahmani Sultanate): 1400; Vijayanagara: 1450; Sultanate of Delhi: 1500; Mughal empire: 1550, 1600, 1650, 1700; Maratha empire: 1750; British empire: 1800, 1850, 1900; India: 1950, 2000. Dashed line: no entries for 250, 550, 600.
Because of this, polities that were based in the northern reaches of South Asia such as the Saka, Kushan, and Gupta empires as well as the more ephemeral Harsha and Pala empires probably accounted for a somewhat larger share of the total population than shown in the graph (an adjustment indicated by the upward arrows in figure 1.7).
The easternmost macro-region, East Asia, has been characterized by much stronger dominance of hegemonic empire than any of the others (figure 1.8). Its leading polities attained population shares of 80 percent or 90 percent for much of the past 2,200 years. Polycentrism in the wake of the Han/Jin collapse—the period of the “Sixteen Kingdoms” and the “Northern and Southern Dynasties”—was quite prolonged but never as intense as comparable interludes in other regions.14
FIGURE 1.8 The proportion of the population of East Asia claimed by the largest polity in that area, 250 BCE–2000 CE (in percent).
KEY: Qin: 250 BCE; Western Han: 200, 150, 100, 50 BCE, 1 CE; Eastern Han: 50, 100, 150, 200; Northern Wei: 225; Jin: 265, 280, 290; Liu Han: 330; Former Yan: 366; Former Qin: 376; Eastern Jin: 400; Liu Song: 440; Northern Wei: 500; Liang: 535; Northern Zhou: 580; Sui: 590, 600; Tang: 650, 700, 750, 800, 850, 900; Later Liang: 920; Northern Song: 960, 980, 1000, 1050, 1100; Southern Song: 1150, 1200, 1250; Yuan: 1300, 1350; Ming: 1400, 1450, 1500, 1550, 1600; Qing: 1650, 1700, 1750, 1800, 1850, 1900; China: 1933; Japan: 1943; China: 1950, 2000.
Moreover, subsequent intervals between monopolistic super-states proved much more short-lived. The longest of them, in the twelfth and thirteenth centuries, was more apparent than real: most of East Asia’s people were controlled by just two great powers—the Jurchen Jin in the north and the Southern Song—that divided China proper between them. Much the same was true for parts of the fourth and fifth centuries, and also much later during World War II, when the Republic of China and the empire of Japan claimed almost the entire population of the region.
Owing to China’s enormous demographic weight, this East Asian profile was single-handedly generated by Chinese state formation (figure 1.9).
FIGURE 1.9 Comparison between the proportion of the population of East Asia controlled by the largest polity in East Asia and the proportion of the population of China controlled by the largest polity in China, 250 BCE–2000 CE (in percent).
KEY See Figures 1.8 and 1.10.
Comparisons
All of these findings are based on a formalistic definition of imperial rule. Throughout this survey, population has been assigned to polities wherever those polities claimed to hold sway. Such claims need to be taken with more than just a grain of salt: premodern empires did not normally govern people the way modern states do. Rulers and central authorities were often remote, and effective power was dispersed across multiple layers of intermediaries and local elites.15
Yet in the end, this does not greatly matter here. Various forms of delegational and indirect rule were the norm for most of the period under review, a situation that only reall
y changed in the past few centuries (in Europe) or even more recently (in the other macro-regions). For this reason, broadly comparable conditions applied throughout the roughly 2,000 to 2,500 years of premodern history that form the core of my survey, at least if viewed in ideal-typical contradistinction to more inclusive modern state formation.
To be sure, not all premodern polities were alike. Periods of relatively—that is, by the standards of the time—successful centralized control alternated with those of effective segmentation as the power of the central authorities dwindled away. It can be difficult to draw clear lines: the Abbasid caliphate or the Holy Roman Empire persisted for centuries after they ceased functioning as anything like unified polities, if indeed (in the latter case) they had ever done so.
This affects the overall profiles in subtle but not insignificant ways. In Europe, my emphasis on formal state claims serves to underestimate the effective extent of polycentrism that obtained in certain periods: treating medieval France or Germany as single imperial entities is generous at the best of times. A more hard-nosed assessment of political realities would push the population share of dominant polities far below the 20 percent mark indicated in figure 1.4. In the MENA region, even though I have classified breakaway kingdoms from the Abbasid caliphate as separate states, a similar problem emerges in the later stages of Ottoman rule when the sultan’s suzerainty had often become purely nominal, most notably in the Maghreb prior to European colonization and in Egypt under the rule of the Muhammad Ali dynasty. In South Asia, the actual reach of the Maurya empire in particular remains very much an open question.
In East Asia, central government control repeatedly frayed long before alternative states were openly established on the territory of existing empires: the late Eastern Han and late Tang periods are classic examples. In the case of China, it is possible to illustrate the mismatch between mere claims and the actual exercise of imperial control by comparing the estimated share of the actual population that was formally claimed by the largest polity with the share of the actual population that was captured by census counts (figure 1.10).
This comparison shows that the Period of Disunion of the fourth through sixth centuries CE resulted in a more sweeping loss of state power than simple population estimates would suggest. It also suggests that Chinese authorities under the late Tang and the early Northern Song found it difficult to register their subjects, and highlights the dramatic erosion of state control under the Mongol and Ming dynasties.16
FIGURE 1.10 Actual population and census population of the largest polity in China as a proportion of the total population of that region, 250 BCE–2000 CE (in percent).
KEY: Actual population: Qin: 250 BCE; Western Han: 200, 150, 100, 50 BCE, 1 CE; Eastern Han: 50, 100, 150, 200; Northern Wei: 225; Jin: 265, 280, 290; Liu Han: 330; Former Yan: 366; Former Qin: 376; Eastern Jin: 400; Liu Song: 440; Northern Wei: 500; Liang: 535; Northern Zhou: 580; Sui: 590, 600; Tang: 650, 700, 750, 800, 850, 900; Later Liang: 920; Northern Song: 960, 980, 1000, 1050, 1100; Southern Song: 1150, 1200, 1250; Yuan: 1300, 1350; Ming: 1400, 1450, 1500, 1550, 1600; Qing: 1650, 1700, 1750, 1800, 1850, 1900; China: 1933, 1943, 1950, 2000. Census population (Bielenstein 1987: 11–140): Western Han empire: 2 CE; Eastern Han empire: 105, 140, 157; Wei 264; Jin: 280; Northern Zhou: 579/580; Sui: 609; Tang: 705, 754/755; Northern Song: 1006, 1050, 1100; Southern Song: 1193; Yuan empire: 1291; Ming empire: 1402, 1450, 1500, 1552, 1602; Qing empire: 1750, 1800, 1850, 1901.
While graphs such as these are the most suitable means of visualizing long-term patterns of imperial state formation, overall scores of political concentration are better capable of capturing the scale of difference between various macro-regions. In table 1.1 I calculate the average proportion of the population of a given region that was under the control of the most populous polity in that region, both over the full chronological range of my surveys and for major subsets.
These cumulative scores reveal very substantial variation among these macro-regions. Europe has the lowest score overall, even if the massive break between Roman and post-Roman conditions might make it seem a rather meaningless statistical artifact. East Asia’s overall score is more than twice that for Europe and shows only moderate fluctuation over time: the secular trend toward greater concentration reflects the absence of prolonged disunion from the second millennium. The other macro-regions occupy an intermediate position. Roman state formation had a moderate effect on the MENA region, as did Islamic state formation in South Asia from the thirteenth century onward.17
European history has produced the most imbalanced profile overall, with a score that was almost three times as high in the Roman period as it has been since then. Similarly striking discontinuities are absent from the other macro-regions. As the graphs show, a 20 percent ceiling became the norm in post-Roman Europe whereas demographic dominance of the top polity only rarely dipped to such a low level elsewhere.
If we look at the twenty-six regular fifty-year intervals in the post-ancient period (from 650 to 2000), the most populous powers in Europe fell short of this 20 percent threshold no fewer than 19 times (or 73 percent), compared to not even once in East Asia, 4 times (15 percent) in the MENA region, and (probably) 10 times (38 percent) in South Asia. Conversely, in the same period the most populous empires in both the MENA region and South Asia cleared a 50 percent threshold 10 times each (38 percent), compared to zero times in Europe and 24 times (92 percent) in East Asia.
Very broadly speaking, the profiles for MENA, South Asia, and East Asia are all variants of the same underlying pattern, one of dominant empires interspersed with periods of deconcentration. They differ only with respect to the relative durability of the leading empires and the length of the intervals between them. South Asia occupies one end of this spectrum: relatively short-lived imperial formations alternated with sometimes substantial phases of abatement. East Asia is located at the opposite end, characterized for the most part by highly dominant and fairly durable (multicentennial) empires that were separated by increasingly brief periods of reconfiguration. The MENA region lies in between, with empires not as pervasively dominant as in East Asia yet more resilient than in South Asia.
Smaller regions in other parts of the world add little of substance to this picture. Polities in the Pre-Columbian New World operated on a much smaller demographic scale. In Mesoamerica, uncertainties surround the political reach of Teotihuacan and the nature of the Toltec polity. In the end, the eventual ascent of the Aztec empire and subsequent universal Spanish colonial rule across Central America snuffed out any semblance of polycentrism for hundreds of years. In the Andean region, the extent of Tiwanaku and Wari rule in the second half of the first millennium remains unclear. A period of fragmentation in the early second millennium preceded the rise of the Inka empire that captured what must have been a very large share of the total population of western South America until it too was absorbed into the global Spanish colonial empire.
Pre-Columbian cultures generally had not had enough time to reach the point at which they were capable of sustaining large empires: increasingly dominant polities had only just emerged when the Europeans arrived, an event that arguably preempted what might well have been the beginning of a pattern of concentration and fission similar to that found in several of the Old World macro-regions.
Southeast Asia is also a much smaller region: up to the nineteenth century it was roughly an order of magnitude less populous than Europe, South Asia, and East Asia. Since the later stages of the first millennium CE, episodes of imperial expansion alternated with times of more widespread polycentrism. On the continent, the Angkorian Khmer empire occupied a dominant position from the ninth through the fourteenth centuries, followed by several coexisting major powers (Ayutthaya, Khmer and Lan Xang), short-lived Taungoo Burmese expansion in the late sixteenth century, more intense fragmentation in the eighteenth century, and dominance by the Rattanakosin kingdom of Siam around 1800.
In Malaya and Indonesia, the Srivijay
a empire exercised hegemony from the seventh through the thirteenth centuries, succeeded on the islands by the Singhasari empire of the thirteenth century and the Majapahit empire of the fourteenth and fifteenth centuries. Dutch colonial rule eventually took over in the eighteenth and nineteenth centuries.
Yet anything resembling hegemonic empire was always conspicuously absent from this region. Even during the most noteworthy peaks, demographic imperial dominance remained rather limited: perhaps one-third of the region’s population under Angkor and Rattanakosin, and less for other imperial ventures. State formation never quite bridged the divide between continental Southeast Asia north of the Malayan peninsula and the Malay archipelago. In this respect, this region represents an outlier, different from MENA, South Asia, and East Asia as well as Mesoamerica and the Andes thanks to the absence of hegemonic empire, but also different from Europe in that this type of empire did not arise even once. In chapter 8, I consider whether the sheltered peripheral position of this region contributed to this outcome.
For ecological reasons, other parts of the world never supported large imperial states. Only a few substantial regions have thus produced sufficient relevant information to warrant systematic investigation. Because of this, our sample is just barely large enough to allow some basic generalization. What does emerge from the record is a widespread pattern of highly dominant universal empire alternating with periods of deconcentration. This back and forth is well documented in the Middle East and North Africa and in South Asia, and to a lesser degree in the more developed parts of the Americas. East Asia likewise belongs in this category, even as near-monopolistic empire increasingly crowded out other arrangements.