On Tue, 30 Nov 2010 10:56:02 -0500, Ray Denenberg, Library of Congress wrote

> So I take your question to be what does that mean. I suggest the 
> following definition: "An estimate which is asserted to be possibly 
> correct, and if not, close to correct."

As an estimate its always correct in this context. A questionable
estimate--- keeping to our symbols--- would be (1985-12~)?
One can talk, of course, of approximation error or the discrepancy between
the expressed approximate value and the known exact value but in our models
given the preference for precision and accuracy we don't have the exact date
and time or we'd express it. Approx. date/time, I'd suggest, is used only
when the "exact" (or something close enough to be accepted by speaker and
listener as "exact" within the agreed upon precision) date/time is unknown.
The assertion is thus only that its "close enough" for consensus but possibly

> I believe that the use case for approximate is satisfied by this 
> definition, but for those who are not satisfied I have offered the 
> extension mechanism that I mentioned so that the estimate may be 
> qualified by precision.
> --Ray


Edward C. Zimmermann, NONMONOTONIC LAB
Basis Systeme netzwerk, Munich Ges. des buergerl. Rechts
Office Leo (R&D):
  Leopoldstrasse 53-55, D-80802 Munich,
  Federal Republic of Germany
Umsatz-St-ID: DE130492967