The first point to be determined was the sequence of colours to be expected from the ordinary image under these conditions. This was ascertained by a series of forty experiments made by W. at the outset of the research; the other subjects being then practised as stated above, till their accounts of the course of the image were consistent. Throughout the research at the beginning of each hour's work, and at intervals during its progress, two or three observations were taken of the ordinary unmodified image. The results obtained were almost perfectly uniform. After the momentary positive same-coloured image which appeared immediately on closing the eyes, there was an interval of five or six seconds, when a positive image came again. This image was at first rather fluctuating in colour, patches of red and green sometimes appearing on it, but in a second or two the panes of the window filled out with sky blue, the window bars remaining dark. This blue stage then passed into a stage of vivid green, the image sometimes but not always disappearing between the two stages. The green image usually disappeared and reappeared five or six times, growing paler, almost whitish, in colour towards the end. This neutralising of the green tint seemed to be due to the gradual emergence of the complementary colour, for the image next assumed a deep red tone, while the black bars became luminous and slightly greenish, the light appearing first as a crack through the length of each dark bar. Here, of course, was the transition from the positive to the negative image. The red stage, after undergoing several fluctuations, gave place to a deep blue image with yellowish bright lines, which lasted longer than any of the preceding, growing gradually darker until it became indistinguishable. By one of the subjects (O.) this final image was described as dark green, or sometimes as bluish green. No other important exceptions to this sequence were noticed with practised subjects, and it corresponds to that observed by Helmholtz under similar conditions. Experiments made with ten and fifteen seconds' exposure also confirm Helmholtz' statement, that after a certain limit has been passed the duration of the stimulus does not affect the course of the image. Variations in the intensity of the stimulating light did, however, influence the colour changes of the image. The sequence just described occurred when the sky was unclouded, or when it was covered with thin clouds and presented a uniformly white surface. When the illumination was much diminished, the sky uniformly and heavily clouded, the blue and green positive images were lacking.
