2QQ RECENT PROGRESS OF THE THEORY OF VISION. 



the red seems much brighter in full sunlight, the blue in 

 moonlight or starlight. This peculiarity in our perception 

 is also made use of by painters ; they make yellow tints 

 predominate when representing landscapes in full sun- 

 sliine, while every object of a moonlight scene is given a 

 shade of blue. But it is not only local colour which is 

 thus affected ; the same is true of the colours of the 

 spectrum. 



These examples show very plainly how independent our 

 judgment of colours is of their actual amount of illu- 

 mination. In the same way, it is scarcely affected by the 

 colour of the illumination We know, of course, in a 

 general way that candle-light is yellowish compared with 

 daylight, but we only learn to appreciate how much the 

 two kinds of illumination differ in colour when we bring 

 them together of the same intensity — as, for example, in 

 the experiment of coloured shadotus. If we admit light 

 from a cloudy sky through a narrow opening into a dark 

 room, so that it falls sidesvays on a horizontal sheet of 

 white paper, while candle-light falls on it from the other 

 side, and if we then hold a pencil vertically upon the 

 paper, it will of course throw two shadows : the one made 

 by the daylight will be orange, and looks so ; the other 

 made by the candle-light is really white, but appears blue 

 by contrast. The blue and the orange of the two shadows 

 are both colours which we call white, when we see them 

 by daylight and candle-light respectively. Seen to- 

 gether, they appear as two very different and tolerably 

 saturated colours, yet we do not hesitate a moment in 

 recognising white paper by candle-light as white, and 

 very different from orange.* 



The most remarkable of this series of facts is that we 

 can separate the colour of any transparent medium from 



' This experiment with diffused white day-light may also be made with 

 moonlight. 



