COLOR VISION IN FISHES 483 



paper techniques. Sporadic investigation continued, with training meth- 

 ods and filtered or spectral lights emphasized more and more. 



Goldsmith came to the support of Frisch with a report on Gobius 

 fluviatilis and Gasterosteus aculeatus. The former of these species was 

 red-shy, the latter red-loving, as were also some young plaice which she 

 tested. Goldsmith's experimental results were practically worthless, for 

 in an attempt to prove hue-discrimination she fell into the brightness 

 trap in a new way : where so many others had assumed that equal bright- 

 nesses for the human would be equal for the fish, Goldsmith assumed 

 that equal energies would be equally bright. That idea was alright for 

 its time ; but she proceeded to equate the energies of her lights by adjust- 

 ing their intensities until they darkened photographic plates to the same 

 extent in the same exposure time. The visible spectrum of any camera 

 film is of course very different from that of any eye. With such stimuli, 

 Goldsmith established that Gasterosteus preferred red>yellow>green-> 

 blue, and concluded that the choices were made on a basis of hue. Her 

 one permanent contribution was in finding that a fish trained to come for 

 food to colored forceps would persist in examining empty forceps bearing 

 the training color for as long as four days after a previous test. 



No new reports appeared until just after World War I, when those 

 of White and Reeves — the latter perhaps the most important single con- 

 tribution to date — were published. White worked on Umbra limi and 

 Eucalia inconstans, using pigmentary colors. She found that grays and 

 white lights were scarcely discriminated as to intensity, and that after 

 training to one of two colors neither species could be confused by any 

 intensity of the negative stimulus. Umbra discriminated between red and 

 green, red and blue, and yellow and green. Eucalia could discriminate 

 red from green, but not blue from yellow. White's steps in albedo were 

 coarse, but she reasoned that since the discrimination of intensity was 

 so poor there was no need of seeking any more perfect match in bright- 

 ness than the fish was able to make. Criticized on this ground, she re- 

 peated her work (as Hineline, 1927) but with a technique actually 

 inferior to her original one. Misled like Goldsmith by a prevalent notion 

 that equal energies should arouse equal subjective brightnesses in any 

 and all animals, she obtained filters equated within a few per cent in 

 total visible energy transmitted. With these, she found that Umbra was 

 able to discriminate red (?i660-700m^) from green (A,510-550m[l), red 

 from blue (X400-450m[i) , red from yellow (A,560-600m(l) and (with 

 difficulty) yellow from blue, but could probably not distinguish blue 



