Your ancestor and their spouse are enumerated in census records over a period of fifty years in a time and place where the provider of the information is not listed.
Is it possible that each enumeration has a different informant (husband, wife, child old enough to answer questions, etc.)? Is it possible that the informant’s memory got worse over time or that their ability to answer questions got worse?
Kat says:
When my ancestor died in 1895, her obituary gave her age as 104. Tracing her in the census records, shows age 20-30 in 1830, 20-30 in 1840, 39 in 1850, 50 in 1860, 63 in 1870, and 84 in 1880. Obviously she was not 104. She lived with different children thru those years and they had to be the ones who supplied the wrong ages. One advantage to the 104 age was that this little country lady was able to get an obituary in the city newspaper!