Excel 2011 for Mac gives a date for February 29, 1900. But there is no such date because 1900 was not a leap year (not an integer multiple of 400). The WEEKDAY() function says 2/29/1900 was a Wednesday whereas it was 2/28/1900 that was a Wednesday. Further checking shows that all WEEKDAY() functions for January and February of 1900 are correspondingly off by one day, e.g. 1/1/1900 was actually a Monday, not a Sunday as Excel says. That one-day shift allows 3/1/1900 to be correctly designated as a Thursday, as are all subsequent dates. However, the insertion of 2/29/1900 throws off by one the date-sequence numbers used in calculations for all dates after 2/28/1900. The problem is a mistake rather than a misunderstanding because other centennial years that are not leap years are handled correctly. I conjecture that the problem arose when the 1900 date system for Macs was added to the 1904 system to achieve compatibility with Excel for Windows. But for true compatibility, wouldn't the Windows version need to harbor the same mistake?

I'm wondering if this bug was ever fixed in Excel 2016 for Mac. Is it present only in Excel 2011 for Mac or is it present also in any of the Windows versions as well? I don't have access to Windows machines. It's hard to believe that for an app as widely used as Excel that this is not a known flaw, although I failed to find mention of it in this forum (but I'm new here and not facile with searching). I've kept up with upgrades for Excel 2011 and the bug persists, so I suspect Microsoft has chosen to ignore it. That would be a sad departure from the care that seems to have been taken with Excel to insure correctness otherwise.