Rule of Thumb for when Anglo-American surnames became standardized?

Question

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.)

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.