Insider's View
Explaining SAS Date Values
posted: January 4, 2007People have asked many times why the base date for SAS date values is January 1, 1960. Is it John Sall’s birth date? (No, John is a little older than that). Is it the date of the first release of SAS? (No, that happened almost a decade later). Is it the date of the previous harmonic convergence? (Well, not quite).
The basic algorithm for the SAS date processing came from an article in Computerworld dated January 14, 1980, which was written by Dr. Bhairav Joshi (who was, at the time, a professor at SUNY-Geneseo). SAS co-founder John Sall adapted this algorithm for implementing the date-time processing algorithms that are still in use by SAS today, and that I currently maintain in my development role. However, the base date of January 1, 1960 was chosen years earlier by another SAS co-founder, Tony Barr.
Tony was looking for a timestamp that would pre-date most electronic records that were available in the early seventies, so he selected the 1960 date.
Tony also wanted to ensure that the timestamps would fit into less storage space, and selecting a more recent base date than the IBM base January 1, 1900, would allow this. The date January 1, 1960 was also close to the release date of the IBM System 360 architecture (which is recognized to be in April 1964). Remember that, at the time, SAS was only running on the IBM mainframes, hence the 360-centric mind-set.
Other systems have used similar decisions for January 1 at the decade boundary. UNIX has used January 1, 1970, and Microsoft has used January 1, 1980. Who knows? Maybe some of them got the idea from SAS!
These days, if your birth date is a negative SAS date value like mine, then you’re an old codger. So maybe it wasn’t such a good idea to choose that date after all!
