Words seem to me also more or less definitely colored, but the association of coloration with me arises solely from the letters of which the words are composed. The dominant letters, especially the initial letter, or the letters most conspicuous in pronunciation, give color to the word. Thus Rosalind, though containing but two green letters, has a dominant shade of green, as salvia or silica have of yellow. A pleasing variety in the colors of the letter tends to render a word attractive. Thus the words Vernon, Severn, and Exeter, with contrasted colors, are more attractive than such words as Patton, Hammond, or Armenia, in which the colors are few or not contrasted. This association of color is stronger than that of the names of the colors themselves, for these fail with me instantly to call up the colors they represent. Thus the word red seems decidedly green in its hue, and it seems unnatural to me that so many words beginning with R, as red, roth, rouge, ruber, rufus, and the like, should have come to mean red. The word blue is also largely green, while yellow is very far from the hue indicated by its meaning.
These letter colors seem for the most part not deep or vivid, but suggest transparent shades like the hues of colored stars, and they are often evanescent where the attention is fastened directly on them. The red, for example, is more like that of the planet Mars than that of a flaming torch. The shades of red vary somewhat, from the scarlet of X or Z, in which the colors seem most pronounced, to the reddish brown of a or n, in which the coloration is less conspicuous.
On the basis of these colors I would make the following classification of the alphabet, placing in each category the most positively colored letters first:
| Red, | X, Z, F, E, H, A, N. |
| Green, | R, L, B, T. |
| Blue, | V, D, Y, K, W, M, P, Q; the V of a violet shade, the M and P lead-color, the Q almost colorless. |
| Yellow, | S. |
| White, | O, C. |
| Straw-color, | G, U. |
| Blue-black, | I. |
In some cases, as in O, C, G, U, I, J, this supposed coloration is plainly derived from the forms of the letters themselves, the O inclosing most empty space, the I none at all. In some other cases, as E, F, or R, B, W, M, the resemblances of form in the pairs in question may have led to their taking place in the same category, the duller letter taking its place beside the brighter one which it resembles.
Similar associations take place with the numerals, although to me the coloration of figures seems less vivid than that of letters.
