I'm not very expert on this subject, but since no one has answered, and I was wondering just how hard it would be to find, I've spent a little time searching.
The conclusion? It's pretty hard to find. I don't have a satisfactory answer to your actual question.
According to this PDF slide deck, income was not solicited on the US Census until 1940, so that doesn't provide you with a whole lot of pre-1951 data.
While I'm not very expert in it, my understanding is that the Current Population Survey (CPS) is the best source for longitudinal data, as long as you don't need fine-grained geographies. However, it appears that the CPS also doesn't extend much before 1951 -- I infer that from this table detailing methodology changes, which begins in 1947. This page also has a note, "If you have any questions, please Call 301-763-3243" -- this may be a case where calling that number or someone in your a state Census Data Center in your state or finding a "data dissemination specialist" for your Census region is in order.
People forget that our tax dollars pay for people to help you with questions like this! Of course it would be awesome if we could just find the answer online, but... well, if you call and learn something, you could post the answer here as a signpost for the next person with a similar question. (Or you could encourage whomever you talk to to consider monitoring this site and fielding questions directly!)
Update: If I understand IPUMS correctly, I requested income data from all years from 1850-1950 and the only rows with non-blank values were for 1950. So it may simply be that it wasn't systematically collected before then.
