wrong number of days between 1/2/1900 and 1/3/1900 ( gives 29 and must be 28!!!). The year 1900 is not a leap year.
is the day function wrong?
A.C. WILSON commented
Excel dates before March 1, 1900 are, and always have been, erroneous. See this discussion about how Excel copied Lotus-1-2-3's error, on purpose so that files converted from Lotus-1-2-3 to Excel would be consistent with each other:
No, it is not.
Multiplan/Excel was competing with Lotus 1-2-3, and 1-2-3 had a programming mistake (not actually their fault, they sort of were doing the same thing Excel chose to do) in this matter so Excel (Lord, I wonder if they remember those non-Mighty days?) found it expedient to program their date system to simulate the mistake existing in 1-2-3.
Mac users have a different date system altogether simply to avoid this whole issue. Not address it head on and damn the consequences, we're Macs!... No, they just avoided it altogether by starting dates that can exist, so to speak, in 1904, avoiding the leap year mistake being made vis-a-vis 1900.
So no one really had any courage and we live with the results today. And clearly, will for thousands of years to come.