1894.] Images following Visual Impressions. 139 



these disks was made as bright as possible, and the other very 

 feeble. By the side of the bright violet disk the feeble one often 

 seemed to be of a greenish-yellow hue, though, when seen alone, it 

 was distinctly violet. 



Experiment 4. 



In a similar manner a feeble violet and brighter greenish-yellow 

 were placed side by side, but, however much the intensity of the 

 former was diminished, it could never be made to assume a blue 

 colour at all comparable to that possessed by the recurrent image of 

 the greenish-yellow. Nor did it appear bluish-green beside orange- 

 yellow or orange-red. 



While, therefore, the result of Experiment 3 is consistent with the 

 contrast hypothesis, that of Experiment 4 appears to be opposed to 

 it ; but there is so great a difference in the circumstances of the two 

 kinds of observation, the one involving a deliberate comparison of the 

 colours of two stationary luminous disks, and the other an estimate 

 formed while a disk and its recurrent image were in rapid motion, 

 that the opposing evidence cannot be regarded as conclusive. 

 Another experiment was therefore devised. 



Experiment 5. 



The original screen with one small aperture was placed at D, and 

 two spectra were projected upon it in such a manner that a green 

 ray from one spectrum, and a red ray from the other, passed through 

 the aperture, forming red and green images which were exactly 

 superposed upon the screen B. The colour of the single image thus 

 formed could, by suitable regulation of the limelights, be made 

 greenish -yellow, yellow, or orange-yellow, these colours being, of 

 course, not simple ones, but compounds of red and green. Now, red 

 by itself gives no recurrent image whatever (this was verified before 

 proceeding further by shutting off the green ray), while green by 

 itself gives a violet recurrent image. The question to be decided 

 was whether the green, when accompanied by the inert red, would 

 give a violet recurrent image as if it were alone, or whether the com- 

 pound colour formed by the combination greenish-yellow, for 

 example would be attended by a blue or bluish-green recurrent 

 image, just as if the compound were a simple spectrum colour. 



The latter was found to be the case. The same hue of greenish- 

 yellow, Avhether a simple spectrum colour or a compound of red and 

 green, was always attended by a blue ghost. When the red ray of 

 the compound was shut off by a screen, the ghost instantly became 

 violet : when the screen was removed it at once resumed its blue 

 colour. 



