I think I've solved a bit of a mystery, but that has led to additional questions. It's all very complicated!
The mystery starts with the family genealogy, written by a first cousin of my great-grandfather. As I went though that genealogy, adding all of my great-grandfather's first cousins to my tree on Ancestry (which includes the author), along with their census records, and other things, I noticed that an apparent first cousin appears in the census records (which there are a lot of, since this is for the latter half of the 1800s, in New York State, where there were state census records in 1855, 1865, 1875, and 1892) but who is NOT in the published family genealogy.
Well, as I said, the author would be a first cousin to this mystery person, and she (the author) lists his brother, and even another brother (who isn't even in the census, because he died as an infant), but not him. I wrote off the apparent omission to perhaps a typographical error, or maybe the guy turned out to be a criminal or something.
To add to the mystery, the mystery man, Frank, seems to disappear from the records after the 1880 census, where he appears as a single man, aged 25, living with his family. I figured perhaps he just died young, and had no headstone, or something like that. But the 1880 census also describes his relationship to his father as "Ad son", which was my perhaps first hint about what might be going on - that apparently is an abbreviation for "adopted son".
Finally I got around to putting his mom's census records in, and in 1900 I see she's living with a widow who is a few years younger than Frank. Frank's mom, Sarah, is a widow, and her relationship to the younger widow, Electa, is said to be "M in Law". Well, that's great - mystery solved about what happened to the son (who apparently died) - EXCEPT, Electa's surname is NOT the same as that of Sarah's (and neither has been married twice!) So how does THAT work!?!
Well, Electa's married name, "Sherwood", sounded familiar, so I looked around and discovered that in 1855 - the first year Frank appears with Sarah (when he was one year old) - two of Sarah's relations have a child in their households with the surname "Sherwood", the relation for both being given as "Adopted". Both those relations live in the same town as Sarah, and both those Sherwoods disappear from the household by the next census, in 1860, even though they'd still be minors. Because of that, I assume those children really weren't adopted but were in some sort of foster child situation, so I left them off my tree. They're not listed in the family genealogy either.
Also, Sarah's little brother married a Sherwood (I actually suspect she was one of the two "Adopted" Sherwoods, but used her middle name instead of the first name used in the 1855 census).
Then I looked at Electa's findagrave entry, and it said her husband is "Frank B Sherwood (1854-1899)". Sarah's son was named Frank, and born about 1854.
So my theory is that Frank B Sherwood was adopted as an infant, and was raised by Sarah and her husband. In every census up through 1880, when he was 25, he's listed with Sarah's married name (Postel). But by the 1892 census he's married, and "Frank Sherwood", and he's buried under that name in 1899.
The question is, what do I do with this mess? Everything I have is basically circumstantial, even if it is compelling. Do I consider him an adopted (not foster) child, even though he ultimately went by his birth name, and isn't in the family genealogy? Do I put "Sherwood" on WikiTree as his "name at birth" (which is a problem, since I already have him as "Postel")? Do I ignore practically everything, because it's all kind of circumstantial? Told ya it's complicated...