Hmmmm.....

> MaterialSample doesn't necessarily have to be an occurrence and
> there could possibly be multiple occurrences in a sample (or vice versa).
 
> Thus we need the Identifier to track this (especially in the context of
> DwC represented as a flat-file) 

This is (again) 100% consistent with our interpretation of our "Individual",
as I discussed previously.  I don't care what the term is (I agree that
"Individual is potentially vague and misleading).  But whatever this "thing"
is, it always includes "material", but depending on how one defines
"sample", our "thing" may or may not always be sampled.

Again, I don't want to re-define MaterialSample, and I don't want to
obstruct it from being added to DwC.  But I *do* want to make sure I
understand its full meaning and scope, so I can conceptually "Venn-diagram"
it against the object that we call "Individual" in our model.

Let me ask a few questions (through declarative statements):

1) An instance of MaterialSample may include zero, one, or many taxa
represented among its material contents (zero taxa = no organisms).
Correct?

2) An instance of MaterialSample may represent a portion or subsample of a
whole organism, may be a whole organism, or may be a set of multiple whole
organisms. Correct?

3) An instance of MaterialSample always represents material that is
extracted from nature.  (e.g., a visual observation with no material
extracted from nature is not an instance of Material Sample.) Correct?

4) An instance of MaterialSample may represent a subsample of (i.e., be
derived from) another, more inclusive instance of a MaterialSample.
Correct?

5) If an instance of MaterialSample is associated with an Occurrence, it
would be an Occurrence that explicitly involved that instance of
MaterialSample (e.g., an Occurrence at which the MaterialSample was
extracted from nature).  Correct?

The last one probably needs a bit more elaboration.  Suppose a whole
organism was observed 20 times over 10 years, representing 20 different
Occurrences.  On the 5th (chronologically) Occurrence, a sample was
extracted from the whole organism, and we represent it as an instance of
MaterialSample. I would assume that we would also establish some sort of
relationship between the MaterialSample instance, and the 5th Occurrence.
But would we also directly associate the MaterialSample instance with the
first four Occurrences (the implication being that this sample had been part
of that whole organism during those first four Occurrences)?  What about
directly associating the MaterialSample instance with the other Occurrences
of the same whole organism that happened chronologically after the
MaterialSample was extracted from nature?

By "directly" associate, I mean "represent as a property of", rather than
some sort of inherited property through some sort of intermediate object
(e.g., the whole organism as a separate object).

Don't get bogged down on this being an Edge Case, because I can come up with
hundreds of other such examples, which, while perhaps being Edge Cases
individually, collectively add up to non-Edge Cases in aggregate.

Aloha,
Rich