The Son Also Rises Read online

  What is true for the 1800–1829 rare surnames that show up at Oxford and Cambridge is true for every such cohort of rare surnames. Figure 12.7 shows the path of educational status for cohorts of rare surnames found at the universities in the thirty-year periods starting in 1800, 1830, 1860, and 1890. All of them show the symmetrical inverted-V shape predicted by the simple law of mobility.

  These findings show two things. The first is the apparent stability of status persistence rates across a whole variety of social regimes in England between 1680 and 2012. The second is the very deep roots of elites at any given time. The rare surnames overrepresented in 1890–1919, for example, are already three times overrepresented at Oxford and Cambridge in 1680–1719. The modern elite has a long history of overrepresentation at the universities.

  A Tale of Two Pepyses

  Can all social mobility in England from 1500 to the present be explained by our simple law of mobility? Some examples of persistence seem remarkable even in a world where the overall persistence is in the range 0.75–0.8. Consider, for example, the Pepys family, mentioned in the introduction. This is one of the prosperous rare surnames we identify among the cohort dying in the years 1858–87. Evidence from parish records of marriages and baptisms suggests that from the sixteenth to the eighteenth century, no more than forty Pepyses were ever alive at one time. Such rare surnames always flirt on the edge of extinction, as the famous Watson-Galton result predicts. Pepys, with only eighteen holders in 2002, seems to be edging closer to annihilation.

  Pepys seems to be a surname of modest origins. There is no record of any Pepys in the medieval records of substantial property owners such as the Inquisitions Post Mortem (see chapter 4) or among members of Parliament. The name does not appear in the probate records of the PCC before 1620. The marriage and baptism records suggest only that it originated in rural areas of Cambridge and Norfolk. The Pepyses are sons of the soil.

  Yet since at least 1496, Pepys has consistently been a high-status surname. In the years 1496–1699, there could have been no more than fifty-six Pepys males who reached age 21. Yet twenty-eight of them attended Oxford or Cambridge, at a time when fewer than 2 percent of all men attended these universities. The family’s university attendance rate declined somewhat in the eighteenth and nineteenth centuries but was still twenty-two out of an estimated seventy adult male Pepyses, thirty times the rate for the general population. In the twentieth century, that attendance rate has declined further, but it is still more than twenty times that of the general population.

  Can this five-hundred-year record of educational attainment be due solely to random forces, or is it due to a special Pepys legacy or advantage, such as the fame of the surname, that transcends the ordinary pull of mediocrity? The answer suggested here is that exceptional as the Pepys family record is, its centuries-long pattern of overrepresentation at Oxford and Cambridge, with a persistence rate of 0.8, is also precisely what we would expect. Figure 12.8 shows the expected arc of rise and fall of the status of the surname over many generations. There was nothing special about the Pepyses except for the fact that between 1450 and 1650 they had a lot of random good luck.

  This random luck lay in good fortune in the shuffling of the genes on reproduction and good luck in the underlying characteristics of the women that Pepys men married. The rise of the Pepyses was not due to anything other than their skill and abilities. They were not the bastard offspring of kings, the recipients of patronage, or the inheritors of caste privilege. They prospered under medieval Catholicism, through the Reformation, and then under Puritanism, through the Restoration, the Glorious Revolution, and into the Victorian era. They prosper now. The General Medical Register in the United Kingdom shows four Pepyses.

  FIGURE 12.8. Occurrence of surname Pepys among Oxford and Cambridge students, by century.

  In contrast to the long arc of success for the Pepys family, there is another family that is likely related, with the similarly rare surname of Peeps, which is completely unremarkable. As with the Pepyses, the earliest parish records of births and marriages of the Peeps family appear in Cambridge. It is likely, given the vagaries of early spelling, that the Peeps and Pepys lineages had the same ancestors. Two Peepses are recorded at Cambridge in the 1530s, but none since. This is not in itself remarkable, as in 1881 there were only forty-six Peepses, and by 2002 that number had shrunk to twelve. While the PCC court records forty-nine probates of Pepyses between 1620 and 1858, not a single Peeps will was probated in this court.

  But the divergent fates of these two surname lineages through the generations have no greater meaning than as examples of the way that random shocks affect the underlying social competence of families. This is not the story of Cain and Abel, of those blessed and cursed in the eyes of God. There is nothing special about these families: their trajectories simply demonstrate that the law of social mobility tends to produce a long arc of privilege or want for those who end up at the extremes of the status distribution.

  This is all just the operation of chance. The Peepses, should the name survive, will sometime in future eons have their time in the Sun. The Pepyses are almost certainly destined for mediocrity. But the persistent effects of accidents of chance alarm many people. The idea that the abilities and status of our ancestors twelve generations in the past can predict our chances of entering university, being a doctor, or becoming wealthy somehow violates the sense that a fair society should offer equality of opportunity for all in the current generation. This issue is considered below.

  Qing China

  Another case that demonstrates the counterintuitive effect of the law of mobility on the rise of families is Qing China (1645–1905). Using surname–place of origin combinations, such as the Fan family of Ningbo, it is possible to track social mobility in the Republican era (1912–49) and the present (see chapter 9). From the records of juren, successful candidates on the provincial exam in the imperial era, in South Jiangsu and North Zhejiang, the status of these surnames can also be traced all the way back to 1645 and the beginning of the Qing era.

  What is the earlier history (1645–1870) of the surname–place of origin combinations identified as elite, based on their numbers of juren in the period 1871–1905? Figure 12.9 shows the relative representation of these surnames by generation from 1661 to 2010 for both Jiangsu and Zhejiang. It illustrates exactly the pattern in the earlier years (1721–1870) that the law of mobility would predict. For both sets of surnames, there is a near-symmetrical rise of the names in status over the previous seven generations. For Zhejiang, this increase implies an underlying persistence of 0.81, very close to the persistence parameter of 0.78 estimated for the decline of status in the Republican era.

  However, the pattern is not as predicted for the years 1661–1720. The relative representation of these surname–place of origin combinations is as high or higher in 1661–90 as in 1721–50. One possible explanation is that rare surnames declined in frequency with time. Absent better information, it is assumed here that these surname–place of origin combinations were the same share of the population in 1661–90 as in 1912–49. Experience in England, for which we can track surname frequencies all the way from 1538 to 2010, shows that a rarer surname in the period 1912–49 tends to account for a larger share of surnames in 1661–90. Surnames that were even more infrequent in 1661–1690 typically disappear by 1912. If these Chinese surname–place of origin combinations were, on average, twice as frequent in the population in 1661–90, then the puzzle would disappear.

  FIGURE 12.9. Relative representation among juren of 1871–1905 surname–place of origin elites, Jiangsu and Zhejiang, 1661–1990.

  FIGURE 12.10. Relative representation among juren of 1781–1810 surname–place of origin elites, Jiangsu and Zhejiang, 1661–1990.

  Figure 12.10 shows the same pattern for surname–place of origin combinations identified as elite in 1781–1810 from their appearance in the juren lists of this period. We see the predicted symmetry, but in a much looser-fi
tting way than in England. Again, difficulties in fixing the population share of these surname–place of origin combinations over time could explain the roughness of the fit.

  Thus the Qing juren data offer qualified support for the idea that status dynamics for elite families always follow the pattern found in England and shown in figures 12.6 and 12.7. Overall, this simple and counterintuitive prediction of the social law of motion holds up well.

  1 Remember that social status here is measured such that the average is zero.

  2 Dickens’s father, however, was at least lower middle class. He was a clerk in the Royal Navy unable to manage his finances: as a result, he was consigned to debtors’ prison and young Charles to the blacking factory.

  3 Handloom weaving was far from the bottom of the occupational hierarchy in preindustrial Britain. Weavers were skilled craftsmen who often owned their looms and cottages. But by the time of Carnegie’s birth in 1835, handloom weaving had gone into rapid decline, devastated by competition from factory weaving. Carnegie’s father reacted entrepreneurially, pursuing better economic prospects by moving the family to the United States.

  4 For this purpose we define the top and bottom of the status distribution as at least three standard deviations from the mean status, roughly the top and bottom 0.1 percent.

  5 See appendix 2.

  6 The incidence of the surnames in the population as a whole is estimated from their frequency in marriage registers.

  7 Their relative representation is even higher in the period in which they are identified, but in this period their relative representation is higher than implied by their true underlying social status because of the prevalence of positive errors (see appendix 2).

  8 As before, we define a rare surname as one held by forty or fewer people in the 1881 census.

  9 The period 1800–1829 is not included in either the forward or backward estimation because the social mobility observed between that period and the adjacent generations is the ordinary type, as opposed to the underlying persistence rate we seek here.

  THIRTEEN

  Protestants, Jews, Gypsies, Muslims, and Copts

  Exceptions to the Law of Mobility?

  THE CASES EXAMINED ABOVE INDICATE THAT across a broad range of societies and epochs, there appears to be a general rule of social mobility. All groups feel the pull of regression to the mean. Variations in social position are maintained by random shocks to families’ underlying social competence.

  It is assumed, however, that each population has a given, similar, distribution of talent, and that elites and underclasses are simply draws from this pool of God-given talent. Within societies, social mobility acts as though it were a biological phenomenon, as if the factors affecting mobility were genetically inherited.

  Some features of social mobility, however, seem to defy such simple quasi-biological laws. The law of mobility explains why individual families become elite or underclass, and the surprising dynamics of that process are laid out in the previous chapter. But it cannot explain how large social, religious, and ethnic groups in some societies attain and maintain high social status or are condemned to persistent low status. Such an outcome is not contemplated in the law of mobility.

  How did Jews emerge as a social elite in Eastern Europe and the Middle East in the Middle Ages or earlier? How did the Gypsy or Traveller population of England end up at the bottom of the social ladder in the sixteenth and seventeenth centuries? Why are Christian minorities typically economic elites in the Muslim world?

  The emergence of elite social groups has often been linked to their embrace of religious ideologies that privilege and foster the aptitudes and aspirations favorable to social success. Thus Maristella Botticini and Zvi Eckstein argue that the rise of the Jews as an educated elite in the Middle East between 70 and 700 CE was driven by the emergence within Judaism of a religious ideology that emphasized that each male should learn to read the Torah, the book of laws.1 Adherence to this emerging religious precept molded the history of the Jewish people in the preindustrial world. It gave them an advantage in the world of commerce, finance, and scholarship and transformed them from farmers into traders, scholars, and financiers.

  Protestantism in early modern Europe similarly emphasized that people should be able to read the Bible for themselves, rather than have their religious knowledge mediated by a priestly caste. The effect of this tradition on literacy rates has been invoked to explain why Protestants enjoyed higher economic status than Catholics in preindustrial Europe. If this explanation is correct, and religious or ethnic affiliation indeed plays an independent causal role in the social competence of families, then the law of mobility will fail to predict many outcomes.

  The law of mobility also fails to explain the persistence of status over centuries of some groups, with no regression to the mean. Why are Jews still a social elite in most societies more than two thousand years after their emergence as a distinct group? Why are Brahmins still an elite and Muslims still lower class in Bengal? How did Copts and other Christians in the Middle East and North Africa maintain above-average social status for more than a millennium after the Muslim conquests?

  Here it is shown that these outcomes are, in fact, compatible with the law of mobility detailed in the chapters above. Incompatibility would arise if religious ideology itself could change the social competence of families. An alternative explanation for the emergence of elites and underclasses is that religions tend to recruit selectively (either positively or negatively) from the existing pool of talent in a society. And the duties that different religions impose on their followers may dictate from where in that talent pool their adherents are drawn.

  Why Are Jews Unusually Successful?

  Botticini and Eckstein highlight that by 1490 CE, the Jewish population was a modest subgroup of descendants of a much larger parent population. They date the emergence of an emphasis on literacy for all Jewish males to the period around 70 CE. Assuming that Jews in the period 65–1490 CE had the same net fertility as the surrounding populations, by 1490 only about 10 percent of the descendants of the parent Jewish population of 70 CE was still Jewish (see table 13.1). The rest of the Jewish population had converted to other religions, probably mainly Christianity. These conversions occurred in environments where forced conversion was rare.

  TABLE 13.1. Jewish population as a percentage of parent groups, Europe, North Africa, Asia Minor, Arabia, and western Asia, 65–1492 CE

  Botticini and Eckstein explicitly present the decision to stay with Judaism or convert to one of the many competing sects as an economic one.2 What drove the decision was, first, the occupation of the family, because literacy is assumed to have had an economic value only to traders and craftsmen. There must also have been some selection based on talent, though this is not explicitly modeled, and the authors are ambiguous on the importance of this feature in determining conversion.3

  Botticini and Eckstein’s main idea is thus that a religion that emphasized a duty of literacy attracted adherents among those who engaged in the urban occupations of trade and manufacture. But if the adoption of such urban occupations was driven by the underlying talents of different families, as seems likely, then Judaism would also have been retained by the most talented among the earlier Jewish population. Much of their evidence is consistent with the possibility that conversion from Judaism was mainly driven by the social competence of families. They observe that “passages by early Christian writers and Church Fathers indicate that most Jewish converts to Christianity were illiterate and poor.”4

  One possible test of whether selection based on ability has shaped the fortunes of the Jewish population is the economic status of Jews in the modern world. Suppose the current Jewish population is just a random subset of descendants of an original Jewish population of 65 CE. Now that literacy is universal, and inculcated by the state, Judaism should offer no economic and social advantage. But even a hundred years after the arrival of universal literacy, Jewish populations are
still heavily concentrated in the upper parts of the status distribution in every society they reside in. What could be the source of this abiding economic advantage in the modern world, if not that Judaism selectively retained the talented from its parent populations?

  Selective Conversion

  Nice evidence of the tendency of religions to selectively attract adherents is provided by the experience of Ireland before Partition. Irish society is notable for the long-standing differences in status between the Catholic and Protestant populations. From the seventeenth to the twentieth century, the Protestants, a settler population largely established in the seventeenth century, remained the elite. There was a sharp and continuing social division between Catholic and Protestant, particularly the Presbyterian Protestants of Scottish origin, and seemingly impermeable barriers to social intercourse. From the arrival of the largely Scottish Protestant settlers in Ireland in the seventeenth century until the present, these two communities apparently developed in splendid isolation from one another. Thus we would expect the Protestant population now to be exclusively the descendants of Protestant settlers, and the Catholic population exclusively the descendants of the native Irish.

  Surname evidence, however, suggests that there have been considerable exchanges of population between the two religious groups.5 If we take a sample of surnames of exclusively Scottish origin and look at the religious affiliation of their holders in the 1911 Irish census, we find that a full 14 percent of them were Catholic.6 And similarly, if we take a sample of native Irish surnames, once exclusively Catholic, 12 percent of the holders in 1911 were Protestants.7 There was thus a significant two-way movement of population across an apparently impenetrable religious divide.