I'm writing code for my 41CV to replicate the DATE function in the 12C (the code that came with the 41CV for calendar math isn't as rigorous as that used by later calculators).
The 12C manual shows the formula used to convert dd.mmyyyy inputs into absolute days. So once you input, say, March 3 2004 into the y register, and 10 as the number of incremental days, you can calc what the new absolute day figure is for 10 days after March 3, 2004. But converting this back to 3,13,2004 is a bit of a puzzle.
Furthermore, it appears to me that the formulas used even by the later calculators isn't exactly right: Years evenly divisible by 400 are leap years, but years evenly divisible by 100 aren't; of course, years evenly divisible by 4 are. The formula HP uses on the 12C, 19BII, etc. doesn't take this into account, so when it calculates the number of days between 31 Dec 1899 and today, it'll be off by 1 day.