XIll. ACTION SPECTRA AND ABSORPTION SPECTRA 443 



Color vision offers problems of considerable difficulty in studying 

 the spectral vision of animals lower than man ; failure to understand 

 these seems often to have led to imtenable conclusions. In photopic 

 vision (cones) in man, one finds that, above the threshold, the matching 

 of different hues is a function of factors other than the magnitude of 

 the exciting radiation. For example, if two monochromatic beams 

 from different parts of the spectrum fall upon the same area of the 

 retina, the color sensation evoked differs from that evoked by either 

 beam alone and the threshold of perception for this color may be 

 quite different from those for the separate beams. As a result it is not 

 possible accurately to determine an action spectrum for human pho- 

 topic vision bj^ matching monochromatic beams unless the wavelength 

 differences are very small. In animals other than man, there is no 

 direct waj^ of knowing whether colors are perceived as such or not. 

 Thus, if one attempted to determine the action spectrum by matching 

 the effect of monochromatic beams in eliciting some end point, he 

 might expect to obtain in the case of animals that detect colors 

 an action spectrum that did not accurately reflect the absorption 

 spectrum of the light absorber. In fact, he might anticipate that 

 quite different action spectra would be obtained when different 

 methods of measurement are used. This seems to be the case with 

 regard to the honey bee, about which there exist two different sets of 

 determinations giving spectral sensitivity curves differing almost 

 completely except for the spectral range {11,23). A clue to the dis- 

 crepancy appears when one examines the biological criteria used; 

 in the one case the measurements consisted in the differential deter- 

 mination of the bee's response to two monochromatic beams; in the 

 other, the response to a monochromatic beam was measured in a 

 comparable manner against a polychromatic beam that was not spec- 

 trally homogeneous (black body radiator). 



When one reviews the many factors that may affect the action 

 spectriim and may interfere with an exact interpretation of its mean- 

 ing, he may be inclined to wonder at the value of such determina- 

 tions, for the few instances of close agreement between action and 

 absorption spectra that may be cited are certainly the exception 

 rather than the rule. But, nevertheless, such measurements con- 

 stitute a necessary part of the study of any photobiological mecha- 

 nism. The question is not whether attempts to measure action spectra 

 should be made, but how carefully it is profitable to make tliem in a 

 given case. Sometimes the nature of the system under study is such 



