LIGHT OF DIFFER EKT WAVE-LENGTHS BY FISH 15 



Before the day's series was begun these slips were shaken together and drawn by 

 chance, and the order in which they were drawn determined the positions of the 

 positive plate. It will be noted that there is here possibility for only four consecu- 

 tive tests to be made on one side. This procedure was followed to prevent the 

 formation of " position habits." In order to prevent any possible " alternating 

 swing," i.e. fish going first to one side and then the other, it appeared best occa- 

 sionally to modify this possible order of choices by substituting for two of the single 

 choice slips two with 3 left and 3 right. These gave the possibility of six conescu- 

 tive choices to occur on one side. 



When a fish had been given from five to twelve chances to discriminate between 

 the stimulus patches, it was closed back into the retention compartment by means 

 of partition B and was not used again until the following day's series. Care was not 

 taken to keep the number of trials per day exact. The number of food masses 

 seized and eaten was thought to be a better index of the physiological condition of 

 the hunger of the fish than the number of tests given. Trials with an individual 

 were, therefore, kept up until its behavior indicated that it was no longer hungry 

 (until it had eaten about 6 to 8, or 8 to 10 food masses). This largely depended 

 upon whether the fish were being tested every day or every second day. Care 

 was taken to place the food-bar in exactly the same position with reference t) the 

 magnets. The two masses of food on the ends were made as nearly as possible 

 of the same size and shape. Frequently, as the day's series progressed, new masses 

 were placed on the lighter end of the bar. Both ends of the bar were kept wet. 

 They were alike in appearance. 



During the tests the operator observed the fish from behind the yellow curtain 

 already mentioned. The stop watch and the string running to the sliding door 

 were manipulated together in one hand while the other hand was placed on the 

 switch so that no jar or body movement or motion of the curtain might indicate 

 the correct side. The time was taken from the raising of the door until the food 

 was snapped at or the fish was before the negative stimulus patch and less than 15 

 cm. from it. The path taken by the fish was then traced on a plan of the aquarium 

 floor, the time recorded, and other notes made. The fish was then closed back for 

 a second trial or closed out into the retention compartment while others were tested . 



Matching Brighness. 



Whenever it was necessary to match the brightness of two stimulus patches, 

 either direct judgment of the appearance of the patches was used or the flicker 

 photometer was employed. In the first case wdiere the Equality of Brightness Method 

 (Ives 1912) was employed one of the following graduate students in psychology, — 

 Miss Marion Bills, or Miss Z. Pauline Buck, — directed the changes to be made in 

 the variable plate until the two plates appeared to be equally bright. For assist- 

 ance in this matching my gratitude is acknowledged. 



The method for the use of the flicker photometer was as follows. The lamp- 

 hcuse with the frame and lenses was swung out past the end of the aquarium. The 

 flicker photometer was placed immediately beneath the partition separating the 

 lalvesof the lamphouse. Plates of ground opal glass, like the stimulus plates, 

 were set beneath the lamphouse, each at an angle of 45° facing the flicker photo- 

 rneter. The light that fell upon these plates of ground glass was that which would 

 give the stimulus patches when the lamphouse was in place, so that, as nearly as pos- 

 sible, the patches were matched against each other by this means. 



5. Determinations of the Intensity and the Distribution of Light 



In order to determine the relative brightness of the 

 colored lights employed, so that the conditions may be 

 duplicated in the future, the illumination used for the 

 blue stimulus plate was measured. The method was to 

 illuminate with a IIU volt tungsten lamp, at a suitable 



