Census data is the weirdest source material for genealogical research and I will tell you why: it’s the easiest to come by and it’s frequently mentioned as the best place to start with one’s family research. But it’s also among the most inaccurate—perhaps the most inaccurate—data available.
Census-takers didn’t fact-check. How could they? People who answered the door for the enumerator—people who may not have been even related to the people the census-taker was looking to enumerate—frequently gave wildly inaccurate details. Birthdates, birth places and immigration years were fudged or guessed. Ages were wrong. Nicknames were supplied instead of given names. Spelling errors popped up everywhere.
I had a problem. I was tracking back a family as far as possible using a combination of the federal census (taken nationally every 10 years since 1790) and the New York state census (taken in Brooklyn every 10 years between 1855 and 1925). I knew my family’s data from the 1865 state census was accurate to a very high degree of certainty.
But the data from what seemed to be the same family in the 1860 federal census was sketchy. I knew I was comparing two families from the same large geographic swath of Brooklyn but neither street names nor addresses were recorded on either of the censuses, so I couldn’t match up those. The given names of the family’s two children matched but they were common names. Some of the ages seemed off. The spelling of the family’s surname was mangled—or was it the correct spelling for a different family altogether? The head of the household had the correct job title but I knew his name was Edward and this census clearly listed it as “Geo,” presumably George. And his wife was listed without any first name at all, just “Mrs.” I got a strong whiff of lazy nineteenth-century census-taker.
I’d let this fishy record sit because I didn’t want to deal with it. But it nagged me. Then I remembered something forehead-smackingly obvious. Census-takers, whether federal or state, didn’t track the families of their district however they pleased. They traced their way around blocks consistently and methodically, with the happy side effect of preserving next-door neighbors. Here’s a diagram from the Census Bureau’s Instructions to Enumerators showing how to canvass a city block for the 1920 census:

That’s when I realized the solution to my problem: in both censuses, I could try matching-up the family just before (and/or after) the one I was trying to prove. Only five years had passed between the two censuses; many families likely had stayed put. And they had. Thanks to my family’s next-door-neighbors, the Watkins’.

A snippet of the 1860 federal census is on the left. The 1865 New York state census is on the right. The names don’t completely match up and the ages aren’t all right—and see how the 1860 census taker cuts the same corners as he did with the family I was tracking, such as listing “Mrs.” (twice!) instead of a given name. But the similarities are numerous, and there are others, including job titles and countries of birth, that I’ve cropped out. I’m now convinced the family I thought I found in 1860 is the same one as in 1865.
(diagram via Source: A Guidebook of American Genealogy (3rd Edition, 2006), edited by Loretto Dennis Szucs and Sandra Hargreaves Luebking; census snippets via Ancestry Library Edition)






























































