Years ago, someone on Twitter quoted Thomas Jones as he was giving a lecture at some genealogy event: “Genealogical proof is not a vote. The most censuses in agreement do not win.” I liked the quote and saved it. But I don’t completely agree.

I was working with someone who is almost my cousin to straighten out a family of her cousins. We had worked on this before, but her list of kids was a little different than mine, especially the order the kids were born. I had a bunch of vital records for them along with many censuses.

It was time, once and for all, to straighten it out. So I opened up a blank spreadsheet and charted them. I started with the censuses, just looking at the kids who were living with the parents.

Pretty quickly, I could see where I had gone wrong. My database was following the 1900 census, where the birth dates were listed in more detail, rather than just the age. But that was the outlier of all of them. The other censuses all agreed on the ages, showing each kid aging between four and six years between censuses, each census five years apart, whereas 1900 was all over the place in comparison. Nathan and Gussie were the biggest problems, as Nathan was far out of order and Gussie didn’t exist in any other record.

I then went on to look at other records I had. There were some kids who never showed up in the census because they died young in between the years. I had several birth and death certificates and one or two other records with exact dates for ten of them. Nathan’s death certificate matched with his age in 1905. And everything matched with the later censuses. The 1900 census was the outlier.

It turned out that my cousin had the kids’ birth order more correct than I did because I had used the earliest census more, and that one turned out to be quite wrong. When I work on families with no vital records, or people that didn’t know their actual birth dates, I tend to use the earliest information I can find, figuring it’s maybe the most accurate. In this case, that was a bad idea.

So the four censuses, along with more than ten other records, outvoted the one. The most censuses in agreement won.

Maybe Thomas went on to say that you need additional supporting documentation along with the censuses, which is what I did. Maybe I need a better source for his quote.