Canada, Quebec: The Institut de la Statistique Québec reports the frequencies of the five thousand most common surnames in Quebec.5 Statistics Canada, however, has not produced any surname frequency listings for Canada as a whole.
Sweden: Statistics Sweden has a searchable database giving the frequency of every surname in Sweden at the end of the preceding calendar year, updated annually, from the population register (maintained by the Swedish tax agency).6 Statistics Sweden also reports the list of the one hundred most common surnames and their frequency in the previous two years, again updated annually.7
Another resource, maintained by the Public Profiler team at University College London, is the Great Britain Family Names website.8 This shows the surname distribution in Britain by county in 1881 (from the census) and 1998 (electoral register). However, this data set includes only surnames with one hundred or more holders in the 1998 electoral register.
Surname Status
We have seen that status tends to be persistent across many generations for surnames. There are a number of ways of inferring the current statuses of surnames.
ENGLAND AND WALES
For England and Wales, one means of ascertaining surname status is to look at the probate rate for a surname relative to the average probate rate for all surnames in a given period. For surnames associated with wealthier groups, the probate rate is higher than the average, and for poorer surname groups it is lower. The website Ancestry.com lists all probates in England and Wales for 1858–1966. (Information for subsequent years can be obtained only by going in person to the Probate Registry in London.) Using this source to look for infrequent surnames in the probate records for 1926–66 and comparing them to data for 1996–2012 gives the results shown in table A3.1.
The average probate rate for all surnames in England and Wales in the earlier period was 39 percent. The surname Smith had a 37 percent probate rate, implying a slightly lower-than-average status. However, the probate rate for Smyths was 74 percent, implying much higher status. The even more elite-sounding Cave-Brown-Cave had an 84 percent probate rate. Goodhart, an elite surname from the early nineteenth century, topped the charts at 100 percent. (There were in fact thirty-nine recorded Goodhart deaths and forty-three probates; some of the Goodharts presumably died outside England and Wales.) The Cornish Boscawen was also elite, with a 70 percent probate rate.
TABLE A3.1. Probate rates, England and Wales, 1926–66 and 1996–2012
In contrast, the table includes other surnames associated with the English Traveller/Gypsy community (though only a fraction of the holders of these surnames would identify as Travellers). These all have lower-than-average probate rates.
As noted, to get the current probate rates for these surnames after 1966 requires a visit to the Probate Registry (one of the least welcoming data repositories I have ever encountered—visit there at your peril). Even though the current data are hard to obtain, the probate test for surname status is the most accurate available, because nearly half of British wills are probated. Thus the records reveal status differences even for relatively rare surnames. However, they tell us people’s status only at the time of death. Since the average age of death for people in England and Wales is now in the late seventies, these data show the status of people one or two generations before the current one.
Another convenient way to check for the status of surnames is to compare their frequency in high-status occupations with their frequency in the population as a whole. Medical professionals are a good source of this type of information because they now represent both a substantial fraction of the population and a high-status occupation. Moreover, most countries maintain publicly accessible medical registers.
In the United Kingdom, the General Medical Council maintains a register of licensed physicians.9 But because there are only 2.8 domestically registered physicians per thousand of population, this test can measure only the status of more common surnames, or groups of surnames, with five thousand or more holders. Table A3.2 shows the number of physicians per thousand population, as measured by the 2002 Office of National Statistics surname list, for a standard average surname, Clarke. In comparison, we can see that physicians with a high-status surname like Smyth are much more numerous, occurring at a rate of 8.6 per thousand. For surnames common in the English Traveller community, such as Loveridge, physicians are less than half as frequent as in the general population. So for surnames held by five thousand or more people, or surname groups of equivalent size, the Medical Register generally enables a realistic assessment of the current average status of the surname.
TABLE A3.2. Numbers of physicians and nurses by surname types, United Kingdom
One anomaly of these data is that because large numbers of physicians of foreign origin work in the United Kingdom, any foreign physician’s surname that is common outside the United Kingdom may incorrectly appear elite on this measure. The surname Schmidt, for example, shows up at a rate of nearly thirty per thousand population, suggesting that Schmidt represents an elite in Britain. An examination of the first names of these physicians, however, shows they are mostly of German nationality, and we can infer that Schmidt is not an elite German surname.10 However, the heavy overrepresentation of Cohens shown in table A3.2 indeed comes mainly from Cohens of British origin.
Nursing and midwifery is another field commonly subject to registration. The Nursing and Midwifery Council in the United Kingdom also maintains a register.11 The advantage of this data set for determining the status of surnames is that it is larger: there are 10.3 registered nurses per thousand population for common domestic surnames, compared to only 2.8 for physicians. Thus for rarer surnames this group is less subject to random fluctuations as a measure of status. The disadvantage is that because nursing is not such a high-status occupation, it does not delineate high- and low-status surnames as clearly. Thus we see in table A3.2 that physicians are 3.1 times as frequent among the high-status Smyths, but nurses are only 2.4 times as frequent. Also nurses are about two-thirds as common as average among the lower-status Traveller surnames, compared with physicians, who are less than half as common.
Another profession that provides public registration information is attorneys, a category that in the United Kingdom includes both solicitors and the smaller group of barristers. The U.K. Law Society maintains a directory of solicitors.12 For common surnames, there are about 1.8 solicitors per thousand surname holders. Barristers are listed in a commercial directory, but there are only about 0.3 barristers per thousand people.13 In both categories, Smyth shows up as a higher status surname with 3.7 solicitors per thousand, and 0.54 barristers. The Traveller surnames listed in table A3.2 appear among solicitors at a rate of 0.9 per thousand, half the expected rate. However, their frequency among barristers is 0.4 per thousand, higher than the average rate. This finding emphasizes the importance of random elements when looking at small numbers.
UNITED STATES
In the United States, estimating the current status of surnames is more difficult than in the United Kingdom because the fifty states, rather than the federal government, have jurisdiction over many aspects of life. Thus the United States has no national probate index, and typically professions, such as attorneys, are licensed at the state rather than the national level. The American Medical Association does maintain a national register of physicians that has a million names on it. But the publicly available website listing these names is cumbersome and essentially unusable for systematic research into surname status.14 The AMA does sell printed and CD versions of its directory that contain much more useful information about individual physicians, such as medical school attended and date of medical school graduation. It is thus a useful, but expensive, source for examining status differences among U.S. surnames.
FIGURE A3.5. Longevity versus surname, United States, 2012.
Because social status differences are strongly associated with longevity, one quick way to estimate status levels of surnames in the United States is by cal
culating average adult ages at death for different surnames. Ancestry.com offers the Social Security Death Index, with information on 92.5 million deaths, available online, though completely free versions are also available.15 This records the dates of birth and death of most of the U.S. population with a Social Security number who died between 1962 and 2012. Thus it records a very large share of all adult deaths in recent decades.
Figure A3.5 shows the average longevity of those age 21 and older dying circa 2007, calculated from the Social Security Death Index in the United States and holding the surnames Katz, Suzuki, Doherty, Hebert, Washington, and Begay(e).16 These are characteristic surnames held by people of Ashkenazi Jewish, Japanese, Irish, New France, black, and Native American ancestry, respectively. Note the great differences in average longevity. The average lifespan of adults with the surname Katz was 80.2 years, compared to 64.6 years for those with the surname Begay(e)—a difference of 15.6 years. Heberts, whites of New France descent, still live three years less than Dohertys, whites of Irish descent.
FIGURE A3.6. Longevity versus occupational status for surname groups, United States, 2012.
These longevity differences correlate strongly with general differences in social status. Figure A3.6 shows, for example, the numbers of physicians in the AMA register per thousand holders of these surnames, versus adult longevity.17 The longevity calculation is less susceptible than occupational measures of status to biases caused by migration of the highly skilled.
AUSTRALIA
The Australian Health Practitioner Regulatory Agency maintains a website that lists all registered medical practitioners: physicians, nurses, midwives, dentists, optometrists, chiropractors, pharmacists, osteopaths, physiotherapists, psychologists, and podiatrists, among others.18 As in England and the United States, surname types show wide variations in representation among physicians. For common surnames of English origin, there are 3.5 physicians per thousand people on the voter roll. But for exclusively indigenous Australian surnames, the rate is zero. For the Jewish surnames Cohen, Katz, and Levy, the rate averages 22 per thousand.
The History of Social Status Using Surnames
ENGLAND AND WALES
For England it is easy to assess the status of surnames back to 1858, and indeed back as far as 1538, using probate records, though the accuracy of the assessment depends on how rare the surname is. The simplest way to measure status is to compare the probate rate of a surname with that of the average surname. This can be done on Ancestry.com for any surname back to 1858, using the national probate and death records.
Figure A3.7, for example, shows the relative probate rates compared to the average for the surnames Loveridge and Doherty/Dougherty. Loveridge, as noted above, is a surname held by many in the Traveller community. As the figure shows, since about 1900, Loveridge has had a probate rate no higher than 80 percent of the average, reflecting the low education, earnings, and wealth of many Loveridges. But interestingly, before this period, Loveridge had a higher-than-average probate rate. Thus Loveridge cannot always have been a surname with a large Traveller share of holders. The steady downward movement in the measured social status of the surname presumably reflects an increasing share of Loveridges who are Travellers. Traveller families have many more children than the average family in England, and the Loveridge surname experienced unusual growth between 1881 and 2002. For the average native surname in England, there were 90 percent more holders in 2002 than in 1881; for Loveridge this gain was 382 percent!
FIGURE A3.7. Probate rates of Loveridge and Doherty/Dougherty, 1860–2012.
In contrast with the decline in status of Loveridge, we see a steady rise in the status of the surname Doherty/Dougherty, of Irish Catholic origin. Initially this surname appears in probate records at only 27 percent the rate of the average name, but by 1980–2009 the frequency had risen to 84 percent.
The main source for earlier probate records, from 1538–1857, as noted above, is the Prerogative Court of the Archbishop of Canterbury (PCC). Since the wills provided in this court represent a much smaller fraction of the population before 1858, this measure is useful only for common surnames or larger surname groups. An index to these probate records is available online at the Public Record Office.19 It contains 980,000 probates from England and Wales for the period 1394–1858.
For the surname Boscawen, discussed above, we see twenty-five probates in this index. To discover whether that is a large or small fraction of all Boscawens, we need an idea of the number of holders of the surname in these years. One convenient source of data for this purpose is the free FamilySearch website, run by the Church of Jesus Christ of Latter-Day Saints.20 A rich source for demographic information before 1837, when national registration of births, deaths, and marriages was instituted, is parish church records of baptisms, burials, and marriages. Volunteers from the Mormon church have transcribed large numbers of these records, particularly those of baptisms and marriages, for England in the years 1538–1837. The site includes records for eighty-three million baptisms in England and twenty-five million marriages, though these include many duplicate entries. The data imply that the average ratio of probates to marriages was .039.
For the surname Boscawen, there are only twenty marriage records in the FamilySearch index. This implies a ratio of probates to marriages of 1.25, thirty-two times the average rate. Thus Boscawen was once a very high-status surname. In contrast, Loveridge has nineteen probates and 649 marriages, for a ratio of 0.029. Thus Loveridge in its early history was a modestly low-status surname.
If we take a famous family from the Middle Ages, such as the Berkeley family, progenitors of the famous philosopher Bishop George Berkeley (1685–1753) and thus the namesake of the University of California, Berkeley, we find the surname overrepresented in these records. There are 122 Berkeley probates and 142 Berkeley marriages, for a ratio of probates to marriages twenty-two times the average. Surnames derived from Berkeley, such as Barclay (now of course embodied by the giant Barclays banking corporation), are also overrepresented: for Barclay we find 145 probates and 568 marriages, producing a probate rate that is 6.5 times the expected rate.
Other surnames that still show up as high status in the period 1394–1858 on this test are those of the Norman conquerors of 1066. Thus, for example, the surname Mandeville appears in the Domesday Book of 1086 as belonging to a substantial landowner. For this surname the PCC index shows 120 probates, compared to 157 marriages, a rate twenty times higher than expected.
The kinds of surnames that tend to be revealed as lower status before 1858 include patronyms and toponyms. Williamson, for example, shows 688 probates but 23,400 marriages for a ratio of 0.029, only three-quarters that for the average surname. Toponyms are names such as Meadow(e)(s) that indicate the location of the holder’s dwelling in the community. Meadow(s)/Meadowe(s) had 128 probates, 4,826 marriages, and thus a ratio of probates to marriages of .026, or just over two-thirds of the average.
A second way to estimate the average social status of surnames in the past, which can be done for England and Wales for deaths from 1866 and later, is to calculate the average adult longevity of those dying with the surname (see n. 16). In all periods, longevity is closely linked to social status. From 1866 on, the death records in England and Wales indicate age at death. Another means of estimating surname status in earlier years in England and Wales is to examine the occupations reported in the census from 1841 to 1911. These data, however, have typically not been digitized by sites such as Ancestry.com (because occupational status is peripheral to the primary interests of genealogists), so using them can be extremely time consuming.
TABLE A3.3. Surname types and characteristics, Ireland, 1911
IRELAND
Ireland has generally poor records of historical surname distributions and status. The original returns of the censuses of 1821 through 1851 were largely destroyed by an explosion at the Public Record Office in 1922 during the Irish civil war. The original census returns for 1861–91 ha
d earlier been destroyed by administrative action. The censuses of the Irish Free State from 1926 on still have not been publicly released.
However, the National Archives of Ireland has digitized the household census returns in their entirety for all of Ireland from the censuses of 1901 and 1911, and these are freely available on the Internet.21 These censuses reveal for each person indications of literacy and occupations, from which it is possible to infer average surname status. They also reveal the religious affiliation of each person. Thus if we compare classically Gaelic and indigenous Irish surnames, such as Doherty/Dougherty, and surnames that arrived with the settlement from Scotland in the seventeenth century, such as Buchanan, we observe a large difference in status between such surnames in 1911. Table A3.3 shows results for men age 18 and older. Interestingly, in Ireland, the fraction of the holders of a surname in 1911 who were Catholic is a good indicator of its average social status.
UNITED STATES
Finding useful sources for the history of surname status is again more challenging for the United States than for England. There is no equivalent to the probate registries for England, the evidence on age at death, or the national counts of marriages, births, and deaths from parish records and the national registry. There is, however, copious material from the censuses of the period 1850–1940. This generally supplies information on occupations, and sometimes, as in 1860 and 1870, on the value of real and personal property. However, again Ancestry.com has not generally digitized occupations and other material considered ancillary to its customers’ interests in genealogy.
The Son Also Rises Page 31