I have, over the years, found interesting errors in censuses for people within and outside of my family. Many of these errors can be attributed to careless or misinformed informants, likely to be outside the family (neighbors, neighbor's child, or someone else unrelated), or to language difficulties on the part of the informant or the enumerator.
Spelling errors can be amusing. My father-in-law's name was Leonard Marshall Rhodes. Everyone, in and out of the family, knew him as Marshall. He was born in Tampa, Hillsborough County, Florida, 16 October 1919, according to his baptism certificate. In 1920, the family was still living in Tampa, and the young (less than one year old) Marshall was recorded as Lennon M. Rhodes. In those days of the Red Scares, it is probably a good thing the enumerator did not spell it "Lenin."
In the 1930 census, it was Marshall's sister's turn. Her name was Della Mae Rhodes, and she was named for her mother, whose name was also Della Mae. I wonder if it ever got confusing in that household.
Anyway, in 1930, young Della, five years old, was recorded as Kellie A. Rhodes. How that happened, I can't be sure. It seems a bit of a stretch.
My grandparents, Perry W. and Mary L. Reed, do not appear at all, as far as I have been able to determine, in the 1920 census. That may be because they were en route from Indiana to Pensacola, Florida.
My father, Arden Packard, born 29 April 1911 in Los Angeles, California, shows up twice in the 1930 census, either because the enumerator did not follow the rules, or because the informant did not have all the information. In 1930, according to the instructions to enumerators, if an individual in a household was in the military and not living at home, he or she would be recorded where they were stationed, and not at their home of record. But in Pasadena, California, in the household of my grandfather Walter H. Packard, my father was enumerated as being there. He was, however, in the United States Navy at the time, in training in San Diego, and was recorded there -- correctly -- as well.
My great-grandfather Charles Reed was also recorded twice, in the 1860 census. He was first recorded in the home of his father, Harvey Reed. A month or so later, it taking time for the census-takers to get around on horseback, Charles Reed was recorded in the home of a Vinson Nidey, who I believe was his boss. I am fairly certain this is the same Charles Reed, because the other data match up, and because in that same block where the Nidey family lived, there was the family of Francis Marion Wright, whose daughter, Clarissa Haney Wright, was very soon to become Mrs. Charles Reed.
There is one item in the Florida State census for 1935 that is probably less of an error than it is a function of the small spaces in which the enumerators recorded their data. The space for occupation was less than an inch long, and one fellow, who was probably a certain type of fisherman, was recorded as being by occupation a lobster.
Finally, I come to the most amusing error I have yet found, not in my family, but in the subject of study I do because I'm just intrigued with this guy. There was a family in Chicago, in the household of a baker named Peter Ness, the youngest child, a son. He was misrecorded in the 1920 census as Ella, daughter, 16 years old. Most likely, this individual never knew he, actually the son of Peter Ness, was so misidentified, and the mistake is probably attributable to either language difficulties between the enumerator and the informant (the parents were Norwegian immigrants), or a less-than-accurate informant from outside the family. Either way, this error would probably have been quite embarrassing to Eliot Ness, the prohibition agent famed in legend and history.