14 



HOW WE SEE 



luminances above 0.1 millilambert (log = 

 — 1), i.e., above the cone threshold. At 

 luminance levels below 0.1 mL (milli- 

 lambert), however, the curves diverge. 

 The yellow and green curves are not shown 

 liere, to keep from cluttering up the chart, 

 but the yellow curve falls between the red 

 and white curves and the green between the 

 white and blue ones. Without attempting 

 any detailed theoretical explanation of 



see objects only if they are more intense 

 than the Hght the eye had been adapted to. 

 Although these very data have been used by 

 Hecht (35) in support of his photochemical 

 theory of vision, this inconsistency appears 

 to have escaped him. 



Further Research Needed. These data 

 provide an excellent illustration of the 

 difficulty produced by the lack of a con- 

 sistent system of photometry for luminances 



3 



O - 



uj V) 



< UJ 



is 3 



UJ - 

 u-o: 2 

 Ox o 



Z Z 



< Z 



-2 



-6 



-4 



-2 



LUMINANCE OF ADAPTING SURFACE 



(IN LOG MILLILAMBERTS ) 



Fig. 8. The luminance of the just visible light imme'diately after lights of various luminance levels 

 are turned off. (Data from Blanchard, 3) 



these data, it is easy to see that the Purkinje 

 shift is at work here again. 



For our purposes, it is important to 

 notice that when the eye has been adapted 

 to high luminance levels, e.g., 1000 mL, 

 it can see luminances about one one- 

 thousandth as intense immediately after 

 it is plunged into darkness. When it has 

 been adapted to a luminance of 1 mL 

 (log = 0), it can see a light one one-hun- 

 dredth as intense upon being plunged into 

 darkness. The one difficulty with this set 

 of data becomes immediately apparent at 

 the lowest luminance values. If we were to 

 trust the data implicitly, it would seem that 

 at very dim adapting levels, the eye can 



below the cone level. The problem worried 

 Blanchard, as is evident from his discussion. 

 He says (3, pp. 85-86) : 



In order to express the results consistently 

 in the same unit of brightness it is necessary to 

 take into account the Purkinje phenomenon. 

 If two fields of different color are illuminated 

 to the same apparent brightness and both cut 

 down by equal amounts, the brightness will 

 not decrease in the same ratio. For example, 

 red will grow darker much faster than blue. 

 But at very low intensities it is impossible to 

 measure brightness by any photometric means, 

 and without having a definite measure of the 

 Purkinje effect for the different colors, the only 

 feasible way of expressing relative intensities 

 is in fractions of a certain measured intensity 

 above the brightness at which the effect sets 



