percipient) must be regarded as ever so much safer and more accurate than the presentation of a sample or the coloured object itself, for the latter may have undergone all kinds of changes when we were not looking (or, for that matter, even while we were looking), and the physiological condition of the percipient may not be at all what we expected it to be. The colour sample language can be understood only by people with normal eyesight, it will produce a certain colour perception in their minds, but it will not "communicate" any "colour content" to them.
Thus we may say in conclusion that we need Language for communication, there is no escape from it and consequently no possibility of "communicating content". We can introduce samples into our language, i. e. speak with colours about colours, with sounds about sounds etc. but Content refuses to enter into it. In so far as a sample can communicate anything it does not do so by its content, but because it is used as a symbol (i. e. as something whose signification must be indicated) and it functions in the same way as all symbols do. Signs remain signs, no matter how we fix their meaning. We can relate the symbol to the object we want it to symbolise by pointing at both of them simultaneously, or by agreeing that the sign shall exhibit a. well defined similarity to its object (as in the case of "samples"), or in some other way: in all cases it is entirely a matter of arbitrary agreement.
No fact can be an "expression" except by agreement. Nothing expresses anything by itself. No series of signs, whether consisting of human sounds, or marks on paper, or any other natural or artificial elements, is a "proposition" simply by its own nature, if by this word we denote something which "says" anything or has a "meaning". A series of marks can become a proposition only by virtue of some agreement which assigns a signification to the single signs and a grammar to the way in which they are combined.
12. On "Sameness of quality".
Several times during the course of the foregoing considerations I had to warn the reader that I was not expressing myself correctly, and had to ask his pardon for it. We shall now examine one of the most important cases in which our language was imperfect, and then see in general how we can guard ourselves against falling into error on account of such imperfections.
