When a spectrum showing a fair degree of saturation is observed, it is seen that the point of greatest brightness lies in the yellows. As the intensity is heightened, this point moves toward the red end, and as it is lowered, it moves toward the blue. It will be seen that the relation between the different amounts of curvature for the different colors is the same as that between the different degrees of apparent brightness when the intensity of the colors in the spectrum is decreased. It is not that of the extreme case of the phenomenon of Purkinje, but when the point of brightness has moved from the yellow into the yellowish greens or decidedly to the right of the place it occupies in the normal spectrum. In that case yellow would be the color second in brightness. In our measurements the amounts of curvature obtained from yellow images were next in size to those of green. The red and the violet-blue which we used would therefore be about equal. It is a noteworthy fact, however, that when the intensities of light (1) and (2) are too great to give a maximum of curvature the amount of curvature obtained with the red is greater than that of the blue, while with the intensities which are too small (4) and (5) to give a maximum the blue curve is greater than the red. As yet it has not been possible for me to find an interpretation of these facts which would seem to meet all the requirements, and I should not wish to offer any explanation at present. The question of the possible connection of this phenomenon with Purkinje's is probably important for any explanation, though it is possible that the arrangement of the curves is merely a coincidence, yet this hardly seems likely, and it would seem as if an explanation of the connection would involve an attempt at explanation of Purkinje's phenomenon, and lead at once into the most doubtful problems of the theory of visual sensations. It is also noticed that the series of numbers obtained when the amounts of curvature of the different colored images, at the different intensities given in the foregoing table, are averaged up, and the curve of the average of all colors is thus obtained, that this average curve is very like that obtained for the white light. These curves and the series of numbers which they represent are here given. [Illustration] Average for all colors. 14.00 15.70 17.90 15.90 15.35 Curve for white light. 11.50 15.00 17.80 15.75 14.25