56. But what if no such sample is part of the language, and we bear in mind the colour (for instance) that a word stands for?—"And if we bear it in mind then it comes before our mind's eye when we utter the word. So, if it is always supposed to be possible for us to remember it, it must be in itself indestructible."—But what do we regard as the criterion for remembering it right?—When we work with a sample instead of our memory there are circumstances in which we say that the sample has changed colour and we judge of this by memory. But can we not sometimes speak of a darkening (for example) of our memory-image? Aren't we as much at the mercy of memory as of a sample? (For someone might feel like saying: "If we had no memory we should be at the mercy of a sample".)—Or perhaps of some chemical reaction. Imagine that you were supposed to paint a particular colour "C", which was the colour that appeared when the chemical substances X and Y combined.—Suppose that the colour struck you as brighter on one day than on another; would you not sometimes say: "I must be wrong, the colour is certainly the same as yesterday"? This shows that we do not always resort to what memory tells us as the verdict of the highest court of appeal.