``It is part of the grammar of the word `chair' that _this_ is what we call
`to sit on a chair' . . . ''  That you use this object _that_ way, sit on it _that_
way, is our criterion for calling it a chair.  You _can_ sit on a cigarette, or
on a thumb tack, or on a flag pole, but not in _that_ way.  Can you sit on a table
or a tree stump in that (the ``grammatical'') way? Almost ; especially
if they are placed against a wall.  That is, you can _use_ a table or a stump
as a chair (a place to sit ; a seat) in a way you cannot use a tack as a chair.
But so can you use a screwdriver as a dagger ; that won't make a screwdriver
a dagger.  What can _serve as a chair_ is not a chair, and nothing would
(be said to) serve as a chair if there were no (were nothing we
called) (orthodox) chairs.  We could say : It is part of the grammar of the
word ``chair'' that _this_ is what we call ``to serve as a chair.''

The force of such remarks is something like : If you don't know all this, and
more, you don't know what a chair is ; what ``chair'' ``means'' ; what
we call a chair ; _what_ it is you would be certain of (or almost certain of,
or doubt very much) if you were certain (or almost certain, or doubt very
much) that something is a chair.

``Certain,'' ``almost certain,'' will come up even under optimal conditions
(when, that is, the context is not one in which the object is, e.g.,
partially hidden, or in bad light, or in fragments) when in an unfamiliar
environment (visiting a foreign tribe, perhaps ; but you do you know it
is _a_ tribe?) you see someone more or less sit on something more or less
like (what we call) a chair, and perhaps hear the people say something
about it using the words we have already translated as ``chair'' and ``sit.''
Perhaps he doesn't _bend_ when he ``sits'' on it : it is nothing so much as a
plank stood on end, about the height and width of an average human being,
tipped and braced back slightly from the vertical, into which there
are fitted at right angles two pegs which, you discover, are to go under
the armpits, and a saddle peg in the middle to ``sit'' on.  We might feel
he's not so much sitting on the thing as hanging from it.  But he looks
comfortable enough.

Then what happens to my contention that when we are hesitant about
asserting a concept it is either because we haven't yet established its
criteria for ourselves or because we haven't got a perfect instance of it?
Surely I am in full possession of the criteria for something's being a chair;
and surely there is no question here of this thing being a _perfect_
example!  And what happens to my contention that in neither case do we
accommodate to the situation by saying ``It's certain'' _or_ ``It's almost
certain''?  With this ``chair,'' I feel not only that the uncertainty is
in me, but that the _fact_, so to speak, is uncertain : anyone ought to realize
that I am not in a position to say for sure, just as much as if the thing
were in fact mostly hidden or in bad light.  It _is_ not certain, maybe
very doubtful, whether it is one or not.

_The Claim of Reason_ p.71