Bob Allkin,

Gregor Hagedorn's original posting to the list explains the brief we received from TDWG. That would be a good starting point - you can get it (and all previous list correspondence) downloaded by sending the appropriate message to the server {can anyone tell Bob what that message is?}

This is my take on your questions.

• who will use this,

Potentially anyone who needs to capture and manage descriptive data, but this will have to wait on development of appropriate tools. Initially, and in a more limited scope, it will serve as an interchange format between existing programs e.g. Lucid, DELTA, XID etc (assuming the developers of those programs come to the party and write I/O functions.

• to store what types of information

I think initially we need to be flexible enough to be able to store any type of morphological or anatomical data. Some chemical descriptive data could fit in here as well, at least simple "chemical A: present/absent" statements. I suppose the main thing we seem to be leaving out for the time being is genome, sequence and protein data - too specialised for our present purpose?

• for what kinds of description (for identification as with DELTA) for

cladistics, for publishing to other audiences

Initially, I think we should aim at descriptions and interactive ID. This is a problematical area, that I think should be explored more. The idea of massaging one data file into two or more different products (e.g. natural-language and keys) is very attractive, but surprisingly problematical, since the structure of data needed for the two purposes is often subtly different. Of course, doing just this is the basis of the DELTA system, but we may need to do it in a more sophisticated way.

As for cladistics, this is more problematical. In the early days of the list there was much enthusiasm about our standard subsuming NEXUS. I think I'm right in thinking that the NEXUS people thought this was a bit of a joke - NEXUS is widely used, is widely supported and seems to be highly successful, while ours doesn't even exist yet! The problems inherent in the multiple-product model become even more alarming when you try to maintain one data file for both description/identification and phylogenetic analysis. My personal view is that we should leave cladistics out of the scope at least for the time being. Anyone else out there want to comment on this?

• to do what (for exchanging data between different databases, as a public

format upon which independent applications will run).

Initially, as an interchange format for existing applications, but if it works, hopefully a new generation of applications will flow.

I'd appreciate it if someone could point me at a summary of your previous deliberations on this.

Other contributors - is this a fair summary?

Cheers - k