There are good Q&A's on this site about how to identify alternative versions of a given surname, but I have a different question more related to this and this.

I know that when I'm looking at 18th century records, "William Allen" may very well be the same person as "William Allyn." But the further we get into the 19th century, I feel like this is less likely, and "Allyn" may represent a different family (or branch of the family) than "Allen."

I am working on a project where identifying potential kinship ties is important. I realize that best practice in any individual case is to find primary documents, which can confirm whether William Allen is actually the Wiliam Allyn who is related to Zachariah Allyn, etc. But I am working with several thousand records, and it won't be practical for me to do this for every case.

So: Do genealogists have any rule of thumb for when Anglo-American surnames became standardized in official records, such that alternate versions of a surname likely indicate a lack of close kinship? When did alternate surnames become signal rather than noise?

(It's worth noting that at the moment, I am looking at families in New England, not on the frontier. I'd imagine standardization hit Providence, RI before it hit the Great Plains. I am also only looking at established English families. I'm aware that non-English immigrant names were not standardized in the 19th century.)

  • I would argue that any system to automate the process of identifying kinship based on surname will fail. With common surnames (e.g. Smith) an index of potential kinship would be high regardless of actual kinship simply due to the fact that the surname is common. In any case, it matters very much what the source of the data is. Was it personally transcribed, or can it be assumed that transcription error is negligible? Also, I should point out that my own (English) surname is still frequently misspelled today, even though there are high levels of literacy.
    – Harry V.
    Commented Mar 5, 2015 at 19:08
  • @vervet: The names were manually transcribed, and double-checked by several people. I know that I can't prove kinship from two people having the same last name, but it would actually be useful for me to disprove kinship wherever possible.
    – two sheds
    Commented Mar 5, 2015 at 19:11

1 Answer 1


Based on my experience my "rule of thumb" would be about 1900, although spelling variations continue to this day even in highly literate areas. There are many surname phonetic algorithms, like Soundex (and more refined algorithms), which would be useful to disprove kinship, but would not be of great use to prove kinship. These algorithms are unlikely to be useful to detect very subtle spelling differences like Allen and Allyn, although they would detect that there is "potential kinship" between Allen and Allyn.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.