442 FRANCIS B. SUMNER 



ing that produced on a color-wheel by combining two parts of black 

 and one part of white. ^* The gray was not, it is true, perfectly 

 neutral, being somewhat 'cold,' i.e., tinged with blue. That this 

 fact played no part in the results seems likely from my later experi- 

 ments (p. 461). The side of the jar toward the window was left 

 transparent, the opposite side being covered with white cardboard 

 to increase, by reflection, the illumination of the bottom. In 

 addition to this, a reflecting screen of white cloth, inclined at a 

 suitable angle, was poised above. 



The white-bottomed jar had walls which also were painted white, 

 but the light was largely cut off from its interior by a cylinder of 

 sheet iron, painted black, which encircled the jar and projected 

 upward for a distance of 12 cm. above its top. 



That the bottom of the gray tank was actually far lighter than 

 that of the white tank, i.e., reflected far more light, is readily 

 seen from fig. 11/-126 (plate 13), which reproduces a photograph 

 taken in the laboratory with the jars arranged as nearly as possible 

 in the same manner as during the test. 



Several experiments were made with these jars, with more or 

 less instructive results. Only one of these tests, however, was so 

 clear cut and decisive that it deserves to be described in detail. 

 Specimen no. 11, which had previously been kept for a long period 

 upon white and other pale backgrounds, but was in the present 

 case taken from the black magnetite sand (after twelve days) was 

 placed in the gray-bottomed jar. It became appreciably paler 

 within an hour, much paler within a day, and, after two days 

 ''probably no darker than the gray bottom." Specimen no. 12 

 which had likewise been on various backgrounds, but came, in 

 this instance, directly from gray, was placed in the (shaded) white 

 jar. This specimen grew noticeably paler in the course of a day, 

 and the pallor increased for a day or two more. (The color- 

 book was here referred to for comparison.) 



At the end of four days, the two specimens were examined under 

 identical conditions of illumination, when it was found that no. 

 11 (on the gray bottom) was decidedly darker than no. 12 (on 



^* Contrary to what one might suppose, such a gray is far from being dark. 



