Chris Wardle wrote:

> How does one define a polygon for a
> settlement mentioned in a mediaeval document but which might be dispersed
> but which might equally well be nucleated  (and if it were nucleated we
> don't know where it was)? Another example might be the metal detector find
> where one cannot be certain if the object was in the field it was reported
> to be found in, in the next field or in the next parish. I have adopted the
> convention of placing sites such as these in a polygon which is a circle
> which is 500 metres in diameter. This is a rather crude. Has anyone any
> better ideas?

Perhaps it helps to decompose these questions into the several kinds of
information we'd like to be contained in culture-historical map symbology:
- firstly, there is the nature of the thing being represented: is it an object
(point-like), monument (point-like in some cases but not in others, may or may
not be visible on the ground), a linear or an areal unit?
- second, does it have clear geographical boundaries? There will not be a
problem with point-like data or modern legal/administrative entities, but as
Chris points out some 'sites' may be vaguely defined.
- thirdly, to what extent (accuracy, precision) are these boundaries known? If
SMR's could represent objects as *reported* from a particular field rather than
*definitely* coming from that field (say by using a dashed or dotted outline
rather than a continuous line), that would already express the fact that
provenance is uncertain. For older records, the precision of co-ordinates could
equally be expressed graphically, eg by the use of a 100 m diameter 'fuzzy
circle' where the location of an object has been recorded only to the nearest
hectare grid point.

So, if the geographical symbol set could express the nature of the thing itself,
the nature of its position/boundaries, and the uncertainties associated with
that position/boundaries, wouldn't that suffice to represent just about all you
need? I recognise that SMR's also deal with information that is so vaguely or
badly defined that none of the above may apply, but in such cases GIS may not be
appropriate tools to represent such information!

Martijn van Leusen