The third point to notice,

112 J. A. FODOR
are no (explicit) indexicals, no demonstratives, none of that stuff. The third
point to notice, however, is that tokens of what Marco Polo wrote are not
evaluated with respect to literally all pairs of times. For example, it does not
make what Marco Polo said false that we can now do the trip in an hour and
a half by jet, or in seven seconds by manned satellite. Nor, for that matter,
does it tell against the truth of what Marco Polo wrote that, in the fifth century
B.C. (before they had turbocharged camels or whatever) it might have been
a forty days journey; or, indeed, that the journey might then have been
impossible.
What we do, in our relentless pursuit of unbloody-mindedness, is this:
we evaluate what Marco Polo wrote as though it had contained an indexical;
as though he had written something like "Kesmur is a province distant from
Bascia seven days journey by the means of transport now available". If, in the
course of making the truth conditions of tokens explicit, we now eternalize
this latter sentence by replacing indexicals with names of what they index,
wet get something like " 'Kesmur is a province distant from Bascia seven days'
journey' is true (as tokened by Marco Polo) iff Kesmur is a province distant
from Bascia seven days by the means of transport available circa 1275". Parity
of analysis would, of course, apply to a tokening of the same sentence by,
say, Alan Shepard; so that, for Shepard's token, the truth rule would run
" 'Kesmur is a province distant from Bascia seven days' journey' is true iff
Kesmur is a province distant from Bascia seven days' journey by means of
transport available circa 1980". The very same standing sentence would thus
be true when Marco Polo tokened it and false when Alan Shepard did. "What
we have discovered", you might be inclined to say, "is that 'journey' is
implicitly indexical". But, of course, we have discovered no such thing. All
that has happened is that we have conscientiously avoided bloody-mindedness
in deciding how implicit quantified variables shall be evaluated; in particular,
we have evaluated them with respect to the universe of discourse over which
the speaker may be presumed to have intended them to range. Thus does the
application of the Principle of Reasonableness in determining the universe
of discourse of implicit quantified variables sometimes contrive to simulate
the operation of implicit indexicals, thereby misleading unwary philosophers.
Just as we evaluate implicit quantifiers with respect to the relevantly
local times, so too we evaluate them with respect to the relevantly local
places', and that, I claim, is what makes kind terms seem indexical. Putnam's
indexical story, you will remember, went something like: "water is wet"
is used to express the belief that stuff of the same kind as this (indexing water)
is wet. The present view, by contrast, is that when somebody on Earth believes
that water is wet, he holds a universally quantified belief with approximately
the content: all the potable, transparent sailable-on, . . .etc., kind of stuff is
wet.16 And when somebody on Earth2 believes that water2 is wet, what he
holds is a belief with that very same content. So, according to this story, the
content of the belief that water is wet = the content of the belief that water2
is wet after all. But establishing the identity of the contents of these beliefs
is not yet establishing how their tokenings shall be evaluated (or, equivalently,
what the truth conditions on the tokens are). To do the latter, we apply the
Principle of Reasonableness in the form: evaluate universally quantified beliefs