GAYs of Wisconsin

Replies to This Discussion

Permalink Reply by John Barrett Robb on August 19, 2009 at 3:30pm

In a word: "Guy". Especially if your guy was Scotch-Irish. "Gay" was generally pronounced "Guy" in Scotland and in many parts of Britain at least as far back as the early 18th century, when the Scotch-Irish began to come to America in numbers, and the spellings were virtually interchangeable in the early records of frontier PA and VA where most of them settled. When spelling crystallized in America in the early 1800s, some held on to the phonetic, version, Guy.

My Scotch-Irish Gays were among the first settlers of SW Virginia - there by 1740 - but my particular branch ended up in Gays Mills, Wisconsin, a town built around the mills they founded there in the 1850s and 1860s. My line of John Gays were particularly literate (one was a Rockbridge Co VA justice for about 30 years), and there was another, earlier, Samuel Gay who was also a county justice in Rockbridge's parent, AugustaCo, which on paper stretched to the Mississippi River until the late 1770s when VA began to hive some other counties off from it. Thus, in spite of the fact that practically all the settlers of these counties were either Scotch-Irish (or Germans) who pronounced the name "Guy", the name was generally spelled "Gay" from the beginning in this backwoods area. Ironically, there was a stream which ran through my John Gay's property, and it is, perversely, named "Guy's Run".

Another important line of (originally) VA Gays was founded by Henry Gay, who came as early as the 1660s to the VA tidewater, probably from England, and many of the records for these folks, and their descendants, used the "Guy" spelling, a few, no doubt, to the present day.

"Guy" is also an independent surname with a different derivation, but in America, and probably in Scotland too, there's a lot of overlap between these surnames, including, no doubt, not a few whose surname was originially, and legitimately "Guy", who have simply adopted the more "correct", or at any rate, prevalent spelling.

You will find a lot more about this on my GAY Surname page ( the all-caps spelling stands for all the two possible spellings of the surname (and there are of course a few other truly exotic variants) including in the left column, links to maps showing the differential distribution of the "Gay" and "Guy" surnames in Britain.

Permalink Reply by Janeane on August 20, 2009 at 6:03pm

Thank you for an interesting and informative article on the Gay surname. I would have thought a Soundex search would have included that variation, but I tried again just in case it didn't. I guess my people didn't play by the rules! I still don't see them in the early census. I appreciate your reply. It might come in handy later!

Permalink Reply by John Barrett Robb on August 21, 2009 at 10:04am

Finding people in the Ancestry.com census indexes (which I assume is your source) is highly problematic in ways that go way beyond the suitability of Soundex for that purpose. Just for the record, there are far better phonetic algorithms than Soundex (Metaphone, Metaphone II, and BMPM) and Ancestry's default choice of Soundex, was unfortunate - and typical. I say typical, because, for another example, when this company decided a couple of years ago to piggyback on the success of the pioneer DNA testing company, FTDNA, they chose what was then the most common marker panel across the industry, rather than build on the scientific knowledge which had accumulated by then, to design their own.

The principal problem with Soundex, is that it lumps too many letters together which in fact have rather different sounds. All of these, for example, are treated as one: C G J K Q S X Z. So that Soundex searching brings up way too many false positives. It also misses many phonetic equivalents for certain surnames. I do generally agree with the Soundex rule of ignoring vowels, with the possible exception of "y". When I'm searching a book index, I've learned that it's best to assume that any vowel or combinantion of vowels might be used, and where the source isn't prohibitively long, I even scan through all the names which begin with a known first letter.

Another problem with Soundex is that it's keyed to the English language, and fails miserably for so many American melting pot surnames; BMPM, on the other hand, is an algorithm which can be tailored to particular languages or smorgasbord of languages, based on a set of customized rules for that language set. The search algorithm which the LDS came up with for the IGI, by contrast, is quite sophisticated - much better than Soundex. Why didn't Ancestry use that?

The main problem with Ancestry searching, though, is that it doesn't even (correctly) implement even primitive DOS wildcard searching, let alone the far more powerful UNIX regular expression style languages, which they might have done. Ancestry search help claims that "?"s can be used to stand for interior letters, just like DOS, but I've never been able to make "?"s work at all. Worse, and totally inexplicably, Ancestry requires that search arguments ending in the other wildcard, "*", begin with at least the first 3 letters of the name, a particularly nonsensical restriction given all the other search filtering they added later (the ability to match on State-County, a range of birth years etc.).

I almost never use Ancestry Soundex. Instead, I try to fashion a set of (partial) name variants which collectively encompass all the variants I've encountered for that particular surname in my paper trail searching, plus the ones I can imagine as close phonetic equivalents. Surname variants are all about phonetics.

With GAY it's pretty simple: "gay" and "guy" cover at least 99% of the cases. However, if I really can't find a name - and a principal cause of search failure isn't the inadequacy of the search strategy, it's the rampant misreadings by Ancestry indexers - I just scan through every page of the census sections where I expect to find people of my target surname. Believe it or not, I used to do this even before I had broadband and it took up to 2 minutes to load a page (I'd read something during the wait time)!

For the later censuses, where many category filtering options have been added, I do a lot of nameless searching - just look for people living in Iowa, say, who were born in Pennsylvania between 1832-1842. Or I search on matches on first name(s) (and/or initials), plus categories, with surname left blank, usually truncating the first name to the first three characters (and I'd truncate it more if Ancestry would let me) to bring up all the spelling variants. Then I scan through the people who come up and check out the more promising ones, or sometimes all of them where my search category set is sufficiently restrictive.

Indexing searching is a high art, and there are always new tricks to be learned. Over time, with experience, one develops a particular bag of tricks for each surname.