It looks to me that ISO8601 allows YYYY as a format? Have I misunderstood Ian’s problem with using bio:date (since he seems to have co-written it I probably shouldn’t argue)?

I think handling the YYYY birthdate as a range makes sense. What I don’t quite understand is why you don’t express it as an actual range (e.g. 1900-01-01 to 1900-12-31) – this would allow you to be more (or less) precise where this was possible? Or does the model you’ve adopted allow this anyway, and the use of YYYY where applicable is just more concise?

Re: The question of ‘circa’ and what it means. This page http://www.library.yale.edu/cataloging/music/pernames.htm suggests that if you knew the birthdate within 7 years, you would give a range rather than use circa. I suspect in reality circa is used more than this. Is what you need is some way of modelling a ‘margin of error’ or ‘degree of uncertainty’?

The other thing to note, going back to your previous post and mentioning natural keys, is that part of the point of adding dates into the catalogue records is not to be complete, but to offer ways of differentiating between people with same name – so once you have clean date data, you should consider incorporating the date into your natural key.