Vanessa; when we developed the facsimile schema for TEI we did consider the case of polygons but in the end we excluded it (and other features) in order to keep the schema simpler, reasoning that a simpler schema would achieve wider support, even if it were problematic for certain cases (such as Bouvard & Pecuchet - I loved that book!).
However, we left a way open to achieve what you want. So I want to second what Martin said, but also to describe this other option, which is to build a custom TEI schema in which you add a new atttribute to att.coordinated. You might define a new attribute called e.g. polygon, to contain a list of the vertices of your polygon. In this way, you could use the existing att.coordinated attributes to model the minimal rectangular region which encloses the shape (the so-called "bounding box"), but also to explicitly describe the shape in detail.
From: TEI Text and Graphics SIG discussion list on behalf of Le Rolle Vanessa
Sent: Thu 6/12/2008 2:34
To: [log in to unmask]
Subject: TEI and polygons
In my training, I'm operating on the project of a critical edition
electronic of the notebooks of Flaubert's "Bouvard et Pécuchet".
We are planning to use TEI. We're now thinking about a method to develop a
close correspondence between text and image, and to cut the sheet in several
The problem is that the corpus is a modern manuscript and so areas or zones
are very irregulars.
We found that TEI only allows rectangles ("att.coordinated") and we need for
the project to make polygons.
I would be thankful for any information about this case, and if other
digitalization projects have met this problem, how are they dealing with?
Vanessa Le Rolle
Institut des Sciences de l'Homme - CNRS UMS 1798
Service d'Ingénierie Documentaire