Did parents with common surnames choose unusual forenames for their children?

1) Introduction

This note makes use of a large database of Victorian marriage registrations to check whether parents with common surnames chose unusual forenames for their children, and conversely whether common forenames were more frequently chosen by parents with unusual surnames.

2) Sample

The database was generated for an analysis of the frequency distribution of forenames, available as separate articles for males and females on this site. A summary of that research was published in an article in Family History Monthly, December 2002. It also forms the basis for an analysis of the frequency of occurrence of surnames, which has yet to be published.

All the marriages between 1867 and 1886 of brides and grooms whose surname began with the letters Ha were extracted from the FreeBMD database as at November 2001. At that time the coverage was about 73%. (5,7 million persons marrying, compared to the ONS total of 7,7 in this period). Surnames beginning with Ha were about 3,4% of all surnames at this time. Ha does not include any of the most common surnames such as Smith or Brown, but has a good scatter of names in the top 200. Table 1 summarises the distribution of the sample within the population:

Table 1 - Sample characteristics

generated with xls2html converter

*including homonyms such as Haynes/Haines, Haigh/Hague, etc. NB: IN 1853 the most common surname, Smith, was 1,36% of all names or slightly more than 4 times as frequent as Hall

Because of the large samples involved, the analysis here excluded the most frequently occurring surnames; only Hart, Harding, Hartley, Haigh, Hale and Hamer were included from the list above. A one third sub-sample of Hall, the most common name, was added to test for any bias - none was found, and the same was the case with Smith. In total the sample used for this analysis comprised (after correction for incomplete entries) 93732 surnames (excluding the 6693 sub-sample for Hall), almost equally split between men and women.

This sample of 93732 names covered a large variety of forenames, many of which only occurred once or twice. However, 13 of the male forenames and 21 of the female ones accounted for 80% of all persons. 53 male names and 63 female names accounted for ±95% of the sample. In total 507 male and 429 female names were identified. 90 male and 57 female names occurred only once or twice in the sample.

3) Diminutives and homonyms

To arrive at these totals of 507 and 429 separate names it was necessary to decide which diminutives (e.g. Daniel and Dan) and homonyms (names that sound the same but are spelled differently, e.g. Sydney and Sidney) should be considered as one name or as separate names. To a certain extent this is an arbitrary process. The Appendix summarises the classification used. In total only about 1200 persons, (a little more than 1%), including both males and females, had (registered) names that were classified as diminutives or homonyms.

4) Analysis

The relative frequency or infrequency with which a particular forename was combined with a particular surname was revealed by drawing up a table in which the actual frequency of an individual name, or of groups of names, could be compared with the expected level. The top forenames and surnames were compared individually, followed by groupings into "common", "unusual" and "rare" in the following way:-

Table 2 Classification of forenames and surnames

generated with xls2html converter

Tables 3 and 4 give an extract / condensation of this analysis.

Table 3 Male names

generated with xls2html converter

Table 4 Female names

generated with xls2html converter

* Excluded from the main analysis and added later as a ( +-) 8 % sub sample of Smith and a one-third sub sample of Hall to check for bias by analysing a very common and a common name

The tables do not suggest that there was any great tendency for parents with common surnames to choose forenames that were unusual or vice versa. For example, on average 14,5% of all parents named a male child William. The six most common surnames in the sample (Hart to Hamilton) with 19% of all persons, chose between 12 and 16% for William. The parents with the rarest surnames, the group of 5% of all persons (2441 persons with about 1000 individual surnames) chose nevertheless just as frequently,13,5%, for William.

Figures 1 and 2 illustrate these distribution patterns.

Figure 1 Forenames by rarity of surnames, Males

For example, the name John is relatively common as John Hartley, but relatively uncommon as John Hale. James, a Scottish name, is relatively common with Hamilton, also Scottish. The rarest forenames show no particular tendency to be used more or less for any particular type of surname. Even the stereotype name John Smith does not occur significantly less often than would be expected, 13% compared with 14% for all Ha names. With a population in 1875 of about 24.4 million persons in England and Wales, this suggests there were about 2200 John Smiths (24.4 x 1,36% (share of Smith name) x 0.5 (males) x 12,9% (share of males with forename John)). Even more common were William Smith with ± 2600 persons and Mary Smith with ± 2400. On the same basis there would have been ± 150 Emma Hamiltons; the 1881 census gives only 108 entries}

Figure 2 Forenames by rarity of surnames, Females

There are slightly more divergences for the female forenames in Figure 2, with Mary Hamilton for example being relatively a little more common than Sarah Harding or Sarah Hamilton. Some of these smaller differences are of course a result of sampling error, but tests at the 95% significance level show a number of interesting real differences for common forenames and surnames. Table 4 shows some of the more marked differences.

Table 4 Forename / surname combinations with marked divergence from average

generated with xls2html converter

*in order of forename frequency
** e.g. the name Richard Hambly occurs 4,2 times more frequently than its expected frequency, (actually 10 times out of 107 Hambly's or 9,3%, instead of the expected 2,2% since Richard occurs 1029 times in the sample of 46706. 9,3% / 2,2% = 4,2).

Amongst the most significant high frequency names are David Harris, with 46 occurrences instead of the expected 4,6 (453 Davids out of 46706 names (1%) times 472 Harris) and Alice Haworth with 64 occurrences instead of the expected 19,4 (1194 Alices out of 47206 names (2,5%) times 766 Haworths). Names that were avoided were Harry Harry with no occurrences against the expected 1, but this is scarcely significant. The absence of any Emily Hamers when 15 would have been expected is however notable. A number of graphs have been provided to show the degree to which the frequency of occurrence of a particular name varies across all the names considered; e.g. for the name Hall, which forenames differ from the expected levels for this surname? These graphs cover the most frequently occurring surnames and forenames for both men and women. In some cases the unusual combinations can be found on two graphs, e.g. James Hamilton under both " James" and "Hamilton"}

Appendix - Names treated as homonyms or diminutives

generated with xls2html converter

* also Lilley 10%

Names treated as separate names

Supporting graphs for individual names

generated with xls2html converter

{ © Donald Hatch, Bilthoven, Netherlands, 12 August 2003}