92 



HOMING AND RELATED ACTIVITIES OF BIRDS. 



time of the response was also recorded/* The tests shown in table 25 were 

 made on successive days. 



Since the animals were holding to such a high percentage of correct choices, 

 I became suspicious as regards the stimulus to which they were responding. 

 I convinced myself speedily that there were no secondary criteria, such as 

 noise, rotating of the box, etc., which could be responsible for their reactions. 

 The animal would not resjiond in any definite way when the monochromatic 

 light was excluded. It became evident that the illumination on the plaster 

 surface was responsiljle for their actions. I next turned the slit far out beyond 

 the limits of the human spectrum (X = 8000) and found that the animals still 

 reacted correctly. Under the conditions of this latter test the jjlaster surface 

 was invisible to me, but after darkness adaptation it became faintly vis- 

 ible. This was undoubtedly due to the stray white light present in this simple 

 spectrometer. I had not supposed its intensity great enough to create dis- 

 turbance in animal work of this kind. Such a source of error is not mentioned 

 in the voluminous works on color responses in animals. 



This brought me into consultation with 

 my colleague, Professor A. H. Pfund, who 

 assured me that there was always a large 

 j'.mount of such light in both prism and 

 gi'ating spectra, and that there were two 

 methods of eliminating it : one by putting 

 behind the slit in the objective a filter of 

 such a nature that there would be trans- 

 mitted only the region of the spectrum 

 desired. In this way all diffuse white light 

 could be eliminated. The other method 

 was to use the jjurified s])ectrum of Helm- 

 holtz, i. e., to take the monochromatic 

 band as it issues from the first spec- 

 trometer and to pass it through another spectrometer. All diffuse white light 

 in the beam so chosen is thrown off into a secondary spectrum. Mention 

 (p. 88) has already been made of the set u]) of this auxiliary spectrometer. 



The first method is not satisfactory for use in experiments where it is neces- 

 sary to pass rapidly from one region of the spectrum to another. The second 

 method is the only thoroughly satisfactory one in such work as the tletermina- 

 tion of the relative stimulating value of waves of different length. 



Since the work on limits required only two filtei's — one at the red end and 

 one at the violet — it was at the time decided to be more convenient to use 

 lliem than to wait for the construction of the double spectrometer. By the 

 time the work on limits had been completed the second sjjectrometer was 

 ready for use. It is certainly advisable, when the ajiparatus is at hand, to 

 employ the "purified spectrum,'' even in the work on limits. 



*E\'idciitly increased intensity made re- 

 sponse easier. 



*The time of all responses was taken from this point on throughout the remainder of the experi- 

 ment. It is without general sijinificauce, so is not given. It aided the experimenter materially 

 in keeping adapt:ttion reasonably constant. As a matter of fact, as the limit is approached, the 

 time increases enormously. The chicks attempted to form the habit of waiting in H until adap- 

 tation brought the light over the threshold. This could be checked l)y waiting luitil the normal 

 time for reaction had passed, then closing the door, turning up the light for 5 seconds, then turn- 

 ing it out and repeating the rci:ul;ir routine of the test. The time of normal reaction varied from 

 about 2 to 14 secc)nd>. 



