Soundex Code Calculator

Why census indexes used Soundex

Names were written down by enumerators, clerks and ship’s officers who spelled what they heard. Smith appears as Smyth and Smithe; Schmidt becomes Smit. To find families despite this, the US Works Progress Administration indexed the 1880, 1900 and 1920 censuses, and parts of 1910 and 1930, with Soundex in the 1930s and 1940s, a code that groups names that sound alike. The National Archives still publishes the rules, and Soundex remains a search option in many databases. Knowing a family’s code lets you search the original microfilm indexes and check variants that a modern wildcard search would miss.

The rules

A code is the first letter followed by three digits. Keep the first letter. Code the remaining consonants: B, F, P and V are 1; C, G, J, K, Q, S, X and Z are 2; D and T are 3; L is 4; M and N are 5; and R is 6. Ignore A, E, I, O, U, Y, H and W. When letters with the same code are next to each other, or separated only by H or W, code them once; this includes the first letter, so Pfister is P236, not P123. When a vowel separates them, code both: Tymczak is T522, because the Z is skipped after the C but the K after the A is coded. Stop at three digits, and pad short codes with zeros: Lee is L000 and Lloyd L300.

Worked examples

Ashcraft codes as A261: A, then S = 2, H is ignored and the C, also 2, is separated from the S only by H, so it is skipped; R = 6; F = 1. Ashcroft gives the same A261, so both spellings are filed together. Robert and Rupert are both R163, while Rubin is R150. Smith, Smyth and Schmidt are all S530. Washington is W252 and Gutierrez G362. McDonald and MacDonald are both M235. A common misconception is that Soundex groups all variants; it does not group names whose first letters differ, so Cohen, C500, and Kohn, K500, are filed apart, and you must search each initial separately.

Prefixes and quirks

The National Archives notes that surnames with prefixes such as Van, Von, De, Di, La and Le may be coded with or without the prefix, so a family may appear under either code: De Vries is D162 as written and V620 as Vries. The calculator shows both when a prefix is written as a separate word or with an apostrophe. Some 1880 and 1900 indexes were compiled with slightly different rules for H and W, so if a family is missing, try the code with those letters treated as separators. Mac and Mc are coded as written; they are not treated as prefixes.

How to use it

1. Enter the surname as it appears in the record, including any prefix.
2. Read the Soundex code, and the code without the prefix if one is shown.
3. Enter a second spelling to check whether the two are filed together.
4. Search the index under every code that applies, including different first letters.