Nat asked:

In what sense MUST water be H2O?

============

This is a natural necessity which needs support from the doctrine of essentialism, and you might want
to look at Kripke's book Naming and Necessityon this. On Kripke's view, if we imagine a world in
which there is something which looks identical to water and is used for the exactly the same purposes
for which we use water, but has a different constitution such as XYZ rather than H2O, we wouldn't
say that the stuff was water. Kripke thinks there can be "fool's water" as there can be "fool's gold" and
this would be water which is not H2O. On the other hand, if we found H2O had different identifying
marks than those we use to identify water at the moment, that would still be water. Natural kinds are
identified and defined by their molecular structure, so water is and must be H2O. This differs from
tables and chairs, for instance, since these could be wooden or metal. That water must be H2O relies
on the essentialism since we can only hold this if we take it that water necessarily has the property of
being H2O.

An alternative view would be that water is the stuff in the sea and rivers which we wash with and drink
regardless of its chemical constitution, and even if its constitution changed, we would regard it as
water. Of course, before it was known that water is H2O people had beliefs about it and referred to it,
so the problem that arises is how they could have had the true belief that the "water level in the river
is high this year".

The essentialist statement that water must be H2O is a "de re" proposition and an a posterioritruth.
Necessity in such propositions relates to the property of a thing and this is in contrast to "de dicto"
propositions where a whole statement is taken to be necessarily true in the sense that it cannot be
false. It could be false that water is H2O in that it could have had a different constitution, so it is not a
de dicto necessity.

Necessity is also defined as truth in all possible worlds, which is again to claim that a proposition
could not be false, but "water is H2O" can be true in this sense depending on what we take a possible
world to be. For Kripke, when we talk about water in another possible world, we mean our water
which is H2O since water is identified with its molecular structure, but for other philosophers we mean
the stuff in the sea and rivers, whatever its constitution.

Rachel Browne

There's no reason at all that water must be H2O. In fact, there is a great deal of water that is not. It
contains salt (NaCl) and various other compounds and ions. We call this complex "water", do we not?
Now, if you start by defining"water" as H2O, then, in order to be consistent with this definition, you
must use the term "water" or "pure water" to refer to that compound. On the other hand, if, in the
question above, you are referring to the same compound by two terms, and have explicitly said that
beforehand, then, again, in order to be consistent with your use of those terms, water and H2O must
refer to the same existent substance.

But both of these are contingent on your use, definitions and suppositions. There are no "musts"
about language. The problem here is that one is caught in a kind of ambiguity between language and
substance, sense and reference, if you will. In order to compare two substances to see if they are the
same, do we use their names? Not originally, nor should we. We should make physical tests of their
characteristics, and if they match (closely enough), we know they are the same substances (i.e.,
exemplars, tokens, of the same physical types). Thus, if what we're calling water is a substance in a
bottle, and what we're calling H2O is one in another, and we compare those substances, then, yes,
water is H2O, and if that identity is fundamental enough (the same molecular structure), we say,
assuming consistent laws of chemistry, etc., that water must be H2O. But then we have a problem
with the next bottle, with the lake outside the room, etc. We have to assume, based on appearance,
that what seems to be water (or what seems to be H2O) in those locations actually is... and since we
have independently determined that water is H2O, we also assume that those substances are also
H2O (or water), given constant laws of chemistry, etc. But now we have the general problem of
induction andthe more specific problem(s) of identifying similar substances (the lake might be a
mirage, glass, or it might be alcohol), where the latter is a more classical empirical problem.

You have touched here on several very complex issues, in language and in induction, and the
meta-problem of distinguishing (inasmuch as possible) those issues. I can't possibly do justice to this
question here; in order to have reasonable background for this you need to read analytic philosophy,
philosophy of science, and about the actual practice of science. Then there are the
counter-arguments to the analytic position... and on and on. Putnam has a lot to say about this issue
(e.g., Putnam, H. (1994). "Sense, nonsense, and the senses: an inquiry into the powers of the human
mind." The Journal of Philosophy91(9): 445-517; also, Putnam, H. (1973). "Meaning and reference."
The Journal of Philosophy70(19): 699-711).

Steven Ravett Brown

If we embrace scientific realism, and we discover that the underlying reality of that wet, colorless,
odorless, neutral, thirst quenching liquid we call 'water' is that it is constructed of one oxygen and two
hydrogen atoms bonded in just this way, then in that sense, water must be H2O.

Tim Sprod