The approach that we are proposing is that the descriptions are collected as atomic statements, and more traditional 'characters' can be discovered by analysis of this data (many characters are apparently a collection of atomic scores/states)

Our taxonomists find this quite a departure from how they compose and record their characters at the moment ( they recognize/discover and define a set of characters by looking at the variation that exists in their specimen, then create a scoring sheet/proforma that allows them to pick one of these alternative characters) - our system might be tweaked to allow them to work in a more character oriented manner if they precompose sets of statements as part of the proforma specification, and then score these alternates as present or absent.

I believe when you say "character" in the above you mean what in DELTA and SDD is considered a character "state"; is that correct? Do you consider the states to be completely unrelated, i.e. green and red and hairy and present are all on the same level? Or is this the collection of state the property or type? This is something I don't yet understand. In many aspects, SDD and DELTA treat states are potentially independent, i.e. characters have semantics, but are otherwise sets of states. The important deviation is the coding status itself, which always applies to the entire character. Can you say in your model that a property is inapplicable or unknown?

a major advantage of our system can be seen from some of your simple characters - eg growth habit: you have split this into two alternatives 1. Epiphytic or lithophytic habit vs 2. (not epiphytic or lithophytic) whilst this might make sense for a key, and is a DELTA-like representation, we would argue that if the ACTUAL growth habit was scored for each specimen as epiphytic, lithophytic, terrestrial, aquatic ( or concatenations of these ) far more accurate information would be recorded. For example, this would allow the same specimen description to be divided into other character sets if desired ( someone else may think that a key would work better if the alternates were soildwelling or lithophytic vs epiphytic, another person might want the alternates separately....if the description data had been recorded in the orginal two-alternate-character division, this data reuse would not be possible.

I think potentially the point your are making is very important, i.e. that for some (not all) "state sets" there may be different partitionings motivated by different goals. One potentially can split color into metallic colors, bright colors, earthen/brownish colors. However, I think the example above is not well choosen (I assume you refer to Kevin's General habit and Epiphytic or lithophytic habit. It is quite possible to have combinations of these. Epiphytic plants may be climber, herb, or grass- or sedge-like plant.

I hope this shows some of the salient features of our model...and how we think it would beneft working taxonomists.

(The rest of the email is unfortunately not readable to me, you seem to use some Microsoft Office or Outlook specific feature to construct the table)

LUCID CHARACTERS<?xml:namespace prefix = o ns = "urn:schemas-microsoft-com:office:office" />

STRUCTURE

PROPERTY/

STATEGROUP

STATES

Salt tolerance

· plants tolerating high salt levels (halophytes)

· plants not salt tolerant

Entire Plant

Ecological Adaptations

Halophytic

(there are a list of alternate states that could be scored, or NOT-halophytic is allowed)

General habit

and so on Gregor ---------------------------------------------------------- Gregor Hagedorn (G.Hagedorn@bba.de) Institute for Plant Virology, Microbiology, and Biosafety Federal Research Center for Agriculture and Forestry (BBA) Koenigin-Luise-Str. 19 Tel: +49-30-8304-2220 14195 Berlin, Germany Fax: +49-30-8304-2203 Often wrong but never in doubt!