Hi Gregor,
I disagree with using "Individual" for sets of objects. It is surprising,
and
lacking any clear definition when to stop, that means a taxon is an
individual,
a collection is an individual, etc.
I refer you to the lengthy discussions on this list a while back, which covered all of this in great detail. Going "up" the hierarchy chain (e.g., beyond "School" or "Herd" or "Colony", etc.) -- to units such as "population" or "taxon" -- things get very messy. During the previous discussion, it seemed that everyone agreed that "taxon" was too broad and not useful for our purposes; whereas "colony", "school", "herd", etc. were absolutely necessary (lest we need to treat every polyp on a coral head as a separate database entry, and many use cases involve treating herd/pod/flock as an important unit to be able to track in exactly the same way that "individuals" are tracked -- which is why the definition of dwc:individualID is as it is). The border-line term is "population" -- which is cleary below the realm of taxon, but perhaps a bit too vauge and poorly defined to be regarded as the same class as an individual.
But the "upper" limit of what the class "individual" would encompass is not relevant to this discussion -- that's a separate issue. The topic here is in the "downward" direction (whole individuals, parts of individuals, parts of parts, etc.), and in that context, our experience has increasingly convinced me that treating such objects as a subclass of what is intended by "dwc:individualID" is a very flexible, useful, and powerful approach to modeling our data -- and also, in my mind at least, the most logically/conceptually defensible approach as well.
But again, I am not advocating that DWC be modeled after our approach. I'm just asking to what extent a conversation about the class "individual" (if, indeed, there is or needs to be such a conversation) should be overlapped with the conversation about "materlaiSample".
Aloha, Rich
P.S. Another very messy question -- if a dwc:individual class is established -- is how it relates to the concept of a "CollectionObject". In most cases, it's very straightforward -- individuals are either extracted from nature and placed in a collection (at the colony, lot, singleIndividual, or any sort of "part" level), or they are left in nature. Where it gets tricky is when you have CollectionObjects that are clearly multi-taxon in nature (e.g., a soil sample as per the proposed materialSample, or a single rock containing multiple fossil taxa on the same physical object). This is where Steve Baskauf and I focused much of our debate -- with me eventually coming around to his perspective that an "Individual" should represent what is believed to be a single taxon, and when it is discovered to include more than one taxon, it should be split to multiple individuals such that the 1:1 relationship of individual:purportedTaxon is maintained (see http://code.google.com/p/darwin-sw/wiki/TaxonomicHeterogeneity for a more extensive discussion). Perhaps this is where the notion of a "material sample" can help.