If one morning the mail should bring you a letter containing nothing but a green leaf, you would not be able to make anything of it; you could record it as a simple fact, but it would not "mean" anything to you. On the other hand, the curious occurrence would have the character of a communication, it would be an actual message, if the leaf were accompanied by some explanation or if you had received some instruction concerning it. It might be a leaf someone promised to send to you from his garden, or there might be a note saying "I found this on my desk" or "please observe the colour of this leaf" or "this is the colour I spoke of yesterday" etc. In all these cases the object itself enters into language as part of it, it has exactly the same function that a picture or a description or any other sign would have: it is itself a symbol in the symbolism called "language". The only peculiarity of this case is that the symbol has the greatest possible similarity to the signified object.
Nothing can prevent us from making the signs out of which we construct our language as similar to the signified objects as we wish; this is even the most natural procedure, and when the human mind first invented writing it consisted of little pictures (hieroglyphics, Chinese characters). Gradually it was seen that similarity between object and symbol is quite superfluous and that convenience and practical utility are the only things which matter. If in order to denote a certain shade of green we use a little patch of colour together with our written words we use the same method as those ancient writings: we avail ourselves of the similarity of colour in the same way as they did of the similarity of shape.
It would be a mistake to think that by using samples as symbols in the way just described we had succeeded in communicating content and had avoided the indirect method of expression. This can be seen by referring to the arguments of section 8, and if you agree with them you will admit that we have no possibility of saying that the reader of the "sample writing" will have "the same content" in his mind as the writer of it. Although the "sample symbolism" is very useful for certain purposes, it cannot be said to be in every respect the most perfect language, it does not fulfill its function any more correctly than a verbal language can. There can be no doubt, for instance, that a scientific description of a colour in terms of wave lengths and other physical data (perhaps including the physiological state of the
