I liked your comment re generated dates.

As you say 0 to 99 are problematic as they will be cast into 20th or 
21st century.

 >9999 can't be entered and display is truncated, but ~sometimes~ can be 
generated and stored eg

[] = DATEVAL("31-Dec-9999") + 1

But these fail:

DATEVAL("1-Jan-12345")            wants DATE
[] = DATEVAL("1-Jan-12345") + 1   wants ALPHNUMERIC

And as "we" have nothing better to do today:

DATEVAL("1-Jan-0001") - 1               31-Dec-0
DATEVAL("1-Jan-0001") - 367             30-Dec-1899
DATEVAL("1-Jan-0001") - 367 + 100*366+1 16-Mar-2000

> 
> Just to clarify some of yesterday's notes, I think the Date-values for 
> the first century (0000-0099) can be generated only internally, via 
> Date-operations, because I think any attempt to input raw dates 
> xx/xx/0000 to xx/xx/0099 will be forced into the 1900-1999 range.
> 
> And, if we had nothing better to do!, we could then use further date 
> arithmetic to calculate dates PRIOR to 31.12.0000, and we could see what 
> PDoxDOS thinks of those!  ;-)
>