Now the white light with which the red is diluted may be regarded as consisting of two parts, one of which is of exactly the same hue as the pure red itself, and the other an equivalent proportion of the complementary colour, which in the present case will be greenish-blue. The fact therefore really is that, as we pass along the scale from 100 to 0, the _total_ quantity of red in the mixture is not reduced to nothing, but only to one half, while at the same time greenish-blue is added in proportions increasing from nought at the extreme right to 50 per cent. of the whole at the middle of the card. The ordinates of the quadrilateral figure E D B F show the proportion of red, and those of the triangle E F B the proportion of greenish-blue, at different parts of the scale. Regarding the portion of the strip which lies above the point marked 0, as representing the zero of colour--that is, whiteness or greyness, which is essentially the same as whiteness--let us continue the diagram in the negative direction, gradually reducing the quantity of red until it falls from 50 per cent. of the whole at F to nothing at A, and at the same time increasing that of the greenish-blue from 50 per cent. at F to 100 per cent. at A. The resultant hue in the portion of the card between F and A will be greenish-blue, which begins to be perceptible as a very pale tint just to the left of F, and increases in purity as A is approached, at which point the colour will be entirely free from any admixture with white. We have in the scale thus presented to our imagination a pair of colours, each occupying one-half of the scale, and gradually diminishing in purity towards the middle line; here only, just at the stage where one colour merges into the other, is there no colour at all, and this region represents the fixed physical zero or standard from which is reckoned the purity of a colour corresponding to any other portion of the scale. The completed scale, it will be observed, though originally intended only for the case of red, turns out to be equally serviceable for greenish-blue: if we consider greenish-blue as positive, then the red, being on the other side of zero, must be regarded as negative. Any other possible pairs of complementary colours may be similarly treated.