The Origins of Surnames
The etymology of surnames is another field of interest. The origin of surnames is often not transparent. Some surnames come from other languages: English surnames may be originally Cornish, Welsh, Gaelic, Latin, or French. Some pertain to occupations that have disappeared in the modern era. And some have mutated from their original form into variants, especially when their meanings have become opaque.
Smith has remained firmly Smith, with few variants, in part because everyone knows what a smith is and how to spell it. But another medieval occupational surname, Arbalistarius, recorded as a surname in the Domesday Book of 1086 and derived from the Latin arcus (bow) and ballista (catapult), has no meaning to the untrained modern ear. Thus it has mutated into the forms Arblaster (held by 450 people in England and Wales in 2002), and Alabaster (held by 468 people).
Similarly the medieval surname Cholmondeley, derived from a place-name in Cheshire, England, came to be pronounced as either Chomley or Chumley. As the name spread farther from its namesake locality, several variants developed, with the following numbers of holders in 2002:
Cholmondeley
141
Cholmondley
11
Cholmeley
18
Chomley
8
Chumley
94
Chamley
335
Champley
60
This ramification stems in part from these originally elite surnames’ being carried down the social ladder by social mobility over generations, to be held by people who were illiterate and had no knowledge of the history or meaning of the surname.
Surnames ending in -ville constitute a set of high-status English surnames. Most of them date from the Domesday Book, and they indicate that the bearer’s home estate was in Normandy. Thus we have Baskerville, Mandeville, Sackville, Somerville, and Turberville, to name a few. Many of these have also ramified into variants ending in -field instead of -ville. Since field is not a synonym for the French ville, this transformation again presumably reflects the downward social mobility of the surname, to the point where it was held by people who had no idea what a ville was. This shift, of course, is a central plot element in Thomas Hardy’s Tess of the d’Urbervilles. Tess Durbeyfield is the daughter of uneducated rural smallholders who learn that they are descended from the d’Urbervilles, a now-extinct Norman noble family. Variants of the original surname Turberville were held by the following numbers of people in 2002:
Turberville
203
Turberfield
209
Turbervill
22
Turbefield
48
Turburville
67
Turburfield
22
Turbyfield
12
For the years 1394–1858, we can compare the average status of the -ville and -field variants. There are 74 probates for the -ville variants, compared to 254 marriages before 1837, a rate 7.5 times the average. In contrast, there were three probates of the -field variants, compared to 93 marriages, a rate only about four-fifths of the average before 1858. Interestingly, there were two probates for another version of the surname, Turbervylde, which perhaps represents a step in the transition from -ville to -field.
Many books have been published on surname etymology. A comprehensive source for England is Reaney and Wilson’s Dictionary of English Surnames, with sixteen thousand entries. There are also some online dictionaries of etymology, but these are of very questionable reliability. One is the Internet Surname Database. This is quite extensive in the number of surnames it covers and the history it provides. But for the surname Smith it confidently asserts: “Of pre 7th century Anglo-Saxon origins, it derives from the word ‘smitan’ meaning ‘to smite’ and as such is believed to have described not a worker in iron, but a soldier, one who smote.”22 The tax rolls for the poll tax of 1381 include, for many taxpayers, not only the surname but also the occupation. A large number of men called Smith are also described as having the occupation of smith, suggesting that for these individuals the name must have had a more recent origin and was descriptive of their occupation.
1 Public Profiler, n.d.
2 Ward et al. 2012.
3 U.K., Office of National Statistics 2002.
4 Australian Government, “Search for Australian Surnames,” http://pericles.ipaustralia.gov.au/atmoss/falcon_search_tools.Main?pSearch=Surname.
5 Institut de la Statistique Québec, “Les noms de famille,” www.stat.gouv.qc.ca/donstat/societe/demographie/noms_famille/index.htm.
6 Statistics Sweden, “Namnsök,” www.scb.se/Pages/NameSearch.aspx?id=259432.
7 Statistics Sweden, “Namnstatistik,” www.scb.se/Pages/ProductTables____30919.aspx.
8 Public Profiler, “Great Britain Family Names,” http://gbnames.publicprofiler.org/Surnames.aspx.
9 General Medical Council 2012.
10 The register also shows where physicians trained.
11 Nursing and Midwifery Council, “Search the Register,” www.nmc-uk.org/Search-theregister.
12 Law Society, “Find a Solicitor,” www.lawsociety.org.uk/find-a-solicitor/?view=solsearch.
13 Legal Hub, “Law, Expert Witness, and Bar Directory,” www.legalhub.co.uk/legalhub/app/main.
14 American Medical Association, “Doctor Finder,” https://extapps.ama-assn.org/doctorfinder/recaptcha.jsp.
15 See, for example, “Social Security Death Master File,” on the privately run website http://ssdmf.info/. The Ancestry.com site, however, offers much greater ease of use.
16 Longevity is calculated for those dying at age 21 and older to limit the effects of differential fertility rates across populations on measured longevity. With higher fertility, proportionately more young people are at risk of death, and hence estimated average longevity is biased downward.
17 It is also easy to calculate longevity in the United Kingdom for the years up to 2005 from databases available on Ancestry.com, as another way of estimating surname average social status.
18 Australian Health Practitioner Registration Agency, “Registers of Practitioners,” www.ahpra.gov.au/Registration/Registers-of-Practitioners.aspx?m=Search. This database can be searched for surnames beginning with any string but returns only fifty results per search.
19 National Archives, n.d.
20 FamilySearch, n.d.
21 Census of Ireland 1911.
22 Internet Surname Database, “Last Name Origins,” www.surnamedb.com.
DATA SOURCES FOR FIGURES AND TABLES
To avoid cluttering the text with references and citations, this section details the data sources for the figures and tables.
Chapter 1
Figure 1.1. Jeff J. Mitchell / Getty Images.
Figure 1.2. Gregory Clark.
Figure 1.3. Corak 2013, figure 2. For Canada: Miles Corak, personal communication, July 27, 2012. For India: Hnatkovska, Lahiri, and Paul 2013. For South Korea: Ueda 2013. Gini for income: World Bank, n.d.
Figure 1.4. Hertz et al. 2007, table 2; World Bank, n.d.
Figure 1.5. © National Portrait Gallery, London.
Figure 1.6. The conventional estimates are the average of those shown in figures 1.3 and 1.4. The surname-group estimates are described in subsequent chapters.
Chapter 2
Figure 2.1. Photo by Tage Olsin / Wikimedia Commons.
Figure 2.2. Almenberg and Dreber 2009, 178.
Figure 2.3. List of noble surnames from Riddarhuset 2012. The stock of different types of surnames as of December 31, 2011, was derived from Statistics Sweden, Surname Search, www.scb.se/namesearch. The population share of the aristocratic surnames was then 0.6 percent. The share of aristocratic surnames over time was calculated from the trend in the share of surnames beginning with Adler-, Af-, Ankar-, Ehren-, Gripen-, Gyllen-, Leijon-, Lillie-, Munck-, Oxen-, Reuter-, Ridder-, Silfver-, Stiern-, and von for men born in the years 1810–2009 and th
ose dying in the years 1901–2009. Federation of Swedish Genealogical Societies 2011.
Figures 2.4–2.6. Federation of Swedish Genealogical Societies 2011.
Figures 2.7 and 2.8. Kalenderförlaget 2008a,b,c; all 2008 tax returns for the kommuns of Botkyrka, Huddinge, Haninge, Nacka, Täby, and Stockholm.
Figures 2.9 and 2.10. Swedish Bar Association 2013. Attorneys were divided into two birth cohorts, 1930–59 and 1960–88. The intergenerational correlations of status in figure 2.10 are based on the assumption that attorneys represent the top 1 percent of the status distribution. In figure 2.10 the relative representation is shown on a logarithmic scale.
Figure 2.11. Physicians in 2011: Sverige, Socialstyrelsen 2011.
Figures 2.12 and 2.13. Physicians registering 1972–2010: Sverige, Socialstyrelsen 2011. 1939–71: Sverige, Socialstyrelsen 1972. 1890–1938: Widstrand 1939. Population share of different surname types estimated from Federation of Swedish Genealogical Societies 2011. In both figures the relative representation is shown on a logarithmic scale.
Figure 2.14. Uppsala University, masters’ theses, 2000–2012: http://uu.diva-portal.org/smash/searchadthe.jsf.
Figure 2.15. Matriculating students, 1942–62: Elvin 1956; Uppsala Universitet 1954; Göteborgs nation 1967.
Figures 2.16 and 2.17. Brenner and Thimon 1971; Odén 1902; Elvin 1956; Uppsala Universitet, 1954; Edlund 1979; Karlberg 1908; Lundin 1882; Sjóstróm 1897, 1901, 1904, 1907, 1908. In both figures the relative representation is shown on a logarithmic scale.
Figure 2.18. Royal Academy of Sciences: for 1739–2012, Wikipedia, “Lista över ledamöter av Kungliga Vetenskapsakademien,” http://sv.wikipedia.org. For 2012, Maria Asp Dahlbäck, archivist of the Royal Academy, personal communication, June 25, 2012. Royal Academy, 1779–2012: Wikipedia, “List of Members of the Swedish Academy,” http://en.wikipedia.org. Royal Academy of Music (1771–2012): Wikipedia, “Lista över Musikaliska Akademiens ledamöter,” http://sv.wikipedia.org. The relative representation is shown on a logarithmic scale.
Figure 2.20. Statistics Sweden 2009, table 6, 20; Kalenderförlaget 2008a,b,c.
Chapter 3
Figure 3.1. The rarer surnames of the rich in 1923–24 were identified from “Taxpayer Listings” 1924, 1925. The rarer surnames of the Ivy League students in 1850 and earlier were identified from Chapman 1867; College of William and Mary 1941; Columbia College 1865; Harvard University 1915; Maxwell 1917; Princeton University 1908; Raven 1909; Yale University 1910.
The physician frequencies are all derived from the listings of physicians in American Medical Association 2012. Only physicians graduating from U.S. and Caribbean medical schools are included. Ward et al. (2012) give surname frequencies in 2000 for surnames held by a hundred or more people. Surname frequencies for rarer surnames, such as those of the rich and the Ivy League students, were estimated from Public Profiler’s World Family Names database, whose data for the United States are based on a recent sample of surnames. The implied numbers of surnames in 2000 were estimated from their frequency per million in this source by looking at the relationship for the same class of surnames for those held by 100–120 people in the 2000 census.
Figure 3.2. Public Profiler, n.d.
Figure 3.3. See sources for figure 3.1.
Figure 3.4. Physicians were allocated to thirty-year cohorts by year of medical school graduation listed in American Medical Association 2012. To estimate the relative representation of surname types among doctors in each generation, we need to know the numbers of people bearing each surname type in the corresponding birth cohort. The average age of medical school graduation is assumed to be 25.
Starting from surname stocks in 2000, the cohort size of the black surnames in earlier years was estimated by assuming that this follows the same trend as the black population as a whole. To estimate this trend we use the share of blacks in the total population age 0–9 at the time of each census to obtain a measure of the black share in each cohort of doctors. If the black surnames used here are representative of the black population, the share of each cohort increased over time 1950–2010.
For the New France group, we check the cohort surname frequency by looking at the share of deaths in the Social Security Death Index for these surnames (Social Security Death Index, n.d). For a group of average social status, the Death Index should indicate the relative frequency of births for a surname all the way from the nineteenth century until recently. The Death Index indicates that New France surnames were a constant share of births for the period 1900–1979. Thus for these surnames, we assume the shares for all cohorts to be equivalent to the 2000 census share.
However, because elite groups have lower mortality rates, the Death Index overestimates the population share of these groups in earlier birth cohorts and underestimates it for more recent. In order for a birth in 1900, for example, to be recorded in the Death Index, the person would have to have died in 1962 or later. And for a birth in 1980 to have been recorded, the person would have to have died at age 32 or younger. Thus it is potentially misleading to use this source to measure the cohort frequency of Jewish surnames, the surnames of the rich of 1923–24, and the Ivy League students.
Instead, for the Jewish surname group, we assume that their share of the population changed in a manner consistent with general changes in the representation of Jews in different age groups, as measured in 2004 (Tighe, Saxe, and Kadushin, 2011). This shows the U.S. Jewish population age 20–29 substantially declining by decade from the 1950s but rising again in the past decade.
For the other two rare surname groups we have no source other than the Death Index, but we must be aware that this source may not yield an accurate estimate of the stock of 25-year-olds in these groups for earlier years. The Death Index suggests that the relative frequency of each surname group was declining over time.
Figures 3.5 and 3.6. Crissey 2009, table 1; Pew Forum 2008, 56.
Figure 3.7. The figure is drawn assuming that the Jewish mean status is an estimated 0.62 standard deviations above the mean and the black mean status 0.27 standard deviations below the mean, with each group having the same variance in status.
Figure 3.8. See sources for figure 3.4.
Figures 3.9 and 3.10. We assume that attorneys and physicians are distributed proportionately across states. We then use the ratio of physicians to the general population for these states to estimate for each group the fraction of their population we are observing.
Attorney stocks in 2012 from listings of the state bar associations of Alabama, Arizona, California, Colorado, Connecticut, Florida, Georgia, Illinois, Louisiana, Maine, Maryland, Massachusetts, Michigan, Minnesota, Mississippi, New York, North Carolina, Ohio, Oregon, Pennsylvania, Texas, Utah, Vermont, Washington, and Wisconsin. Surname stocks as for table 3.1 and figure 3.4.
The relative representation here of a surname like Katz is defined as the number of attorneys with the surname Katz compared to the number of Katzes in the 2000 U.S. census, divided by the number of attorneys with the surname Olson or Olsen compared to the number of Olsons and Olsens in the census.
Figure 3.11. American Medical Association 2012; Ward et al. 2012. Surnames used were those whose holders were 95 percent or more white.
Figure 3.12. Intermarriage rates all obtained from Ancestry.com, based on the following sources. Connecticut: Marriage index, 1959–2001. Maine: Marriages, 1892–1996. Massachusetts: Town and vital records, 1620–1988. Oregon: Marriage indexes, 1946–2008. Vermont: Marriage records, 1909–2008. State Franco-American population shares are from US Census Bureau 2010.
Figure 3.13. See sources for figure 3.4. The relative representation of Japanese surnames among physicians is measured using all Japanese surnames held by at least a thousand people in the United States in 2000.
Tables 3.1 and 3.2. See sources for figure 3.4.
Table 3.4. Suzuki 2002, table 3, 265.
Chapter 4
Figures 4.1 and 4.2. For this chapter and the next, two main sources a
re a database of the surnames of those attending Oxford and Cambridge in the period 1170–2012 and a database of those whose estates were probated at the Prerogative Court of the Archbishop of Canterbury, 1384–1858. For the latter, the index of the court wills in series PROB 11 at the National Archives was used to construct a database recording the name, status (from titles such as Sir, Lord, and Gentleman), and the date of probate. The database contains 903,438 such probates for England. The Oxford and Cambridge database contains the name and year of arrival at the universities of all known students (and faculty in earlier years) from 1170 to 1889, and thereafter a large sample of known attendees. Brasenose College 1909; Cambridge University 1954, 1976, 1998, 1999–2010; Elliott 1934; Emden 1957–59, 1963, 1974; Foster 1887, 1891, 1893; Venn and Venn 1922–27, 1940–54; and Oxford University 1924–2010. For the years 2010–12, student surnames were derived from the e-mail directories for Oxford and Cambridge (www.ox.ac.uk/applications/contact_search; http://jackdaw.cam.ac.uk/mailsearch/). Women students at Cambridge for 1860–1900 were identified from the Cambridge Alumni Database (http://venn.lib.cam.ac.uk/acad/search.html), which contains records for 781,474 persons.
Figure 4.3. British Library, MS. Harley 4866.
Figure 4.4. Locative surnames were defined as those ending in -ton(n)(e), -tu(n)(e), -don(n)(e), -dun(n)(e), -dg(e), -ham(m)(e), -land(d)(e), -bur(r)(y)(e), -ber(r)(y)(e), -bur(r)i(e), and ber(r)i(e). Their share at Oxford and Cambridge was calculated as in figure 4.1. Their share in the population for each period from 1538 to 1837, before national registration of births, deaths and marriages, was estimated from marriages recorded in parish registers and other sources as transcribed for the FamilySearch website. This approach was preferred to counting births, since infant and child mortality differs by social class. It assumes the same marriage rate for high- and low-status groups. For the years 1837–2012, surname frequencies were estimated from three sources: England and Wales, Register of Marriages, 1837–2005; the 1881 census of England and Wales (Schurer and Woollard 2000); and the Office of National Statistics database of surname frequencies in England and Wales in 2002 (U.K., Office of National Statistics 2002). For the years 1200–1837, surname frequencies were estimated by projecting backward the trends observed in marriage records for the period 1538–1600.
The Son Also Rises Page 32