238 PERCY W. COBB 



brighter than the test field; (3) and the condition in which the 

 field was the brighter. Practical light distributions of which 

 these may be looked upon as extreme types might be said to be: 

 (1) a well diffused and uniform illumination, presenting no more 

 than moderate contrasts, (2) the conditions under which one 

 attempts to look out into the night from a well-lit room, or 

 attempts, in daylight, to see an interior through a window or 

 open door, in such a way that the opening covers a comparatively 

 small extent of the visual field and (3) a local, shaded light, 

 close to the work, the rest of the space being in darkness. 



Now as to the second of these conditions, it is to be said that 

 a large increase in the difference threshold is to be noted when 

 the surroundings are much in excess of the test field in brightness; 

 the increase amounting to a large deviation from Weber's law. 

 Visual acuity, under like condition behaves in a similar way. 

 On the other hand it is to be said that such conditions are in no 

 way a fair representation of any practical lighting conditions 

 which have to be considered by the illuminating engineer as 

 relevant to the comfort, efficiency or hygiene of the visual organs. 



In the third case, that of a bright test field seen in dark sur- 

 roundings, the writer's results show an increase in the value of 

 the difference threshold, over its value under the standard mean 

 conditions, of 55 to 70 per cent, according to the interpretation 

 put upon the results. These figures are the means for three 

 observers, and it is to be remembered that the experimental 

 conditions are typical and extreme. The test field of less than 

 three degrees angular extent is much smaller than any page 

 or sheet ordinarily viewed at reading distance, and the surround- 

 ings, in the experimental arrangement, were as dark as technical 

 devices could make them, which is not apt to be the case in the 

 analogous practical situation. 



Moreover the threshold is small (as large as 1 per cent of the 

 test-field brightness in a single instance only), and the divergence 

 of the values of the same, derived from separate series of judg- 

 ments, is represented by a mean variation of 13 to 25 per cent 

 of the mean. This degree of precision was adequate to make 



