92 BLIND VERTEBRATES AND THEE* EYES. 



compartments contained but one opening, only one compartment bordering each. 

 A strong positive reaction toward violet is indicated, and a strong negative reaction 

 toward pink and blue. The totals were : violet, 25 ; blue, 6 ; green, 1 1 ; yellow, 13 ; 

 orange, 14; pink, 7. 



To test these results, the second aquarium, with but two compartments and three 

 specimens, was used. The specimens are marked A smallest, B medium sized, c 

 largest. In series I, 20 out of 24 events occurred in the red. The windows were 

 interchanged, transposing colors, when, in series II, out of 24 events 13 occurred 

 in the red. This indicates a decided positive reaction toward the red. In series 

 III, 16 out of 20 events occurred in the red. A new aquarium was substituted, with 

 the windows side by side, looking toward a west window. Out of 17 events (series 

 IV) 13 occurred in the red. The colors were then interchanged, so the fishes would 

 be compelled to change compartments in order to be in the same light. In this 

 series 27 out of 29 events occurred in the red. These series give conclusive evi- 

 dence that the affinity of these fishes is strongly in favor of the red. It may also be 

 noted that the smallest specimen was most frequently found in the blue. 



BREEDING HABITS OF AM BLYOPSIS. 



The eggs are laid by the female, to the number of about 70, into her gill cham- 

 ber. Here they remain for perhaps 2 months, till the yolk is nearly all absorbed 

 and the young fish has attained a length of about 10 mm. If at any time a female 

 with young in her gill pouches is handled, some of them are sure to escape. This 

 was observed and gave rise to the idea that this species is viviparous. 



We owe the first observations on the breeding habits of Amblyopsis to Thomp- 

 son, who states that a fish "was put in water as soon as captured, where it gave 

 birth to nearly 20 young, which swam about for some time, but soon died * * * they 

 were each 4 lines in length." It is unfortunate that the highly interesting suppo- 

 sition of Thompson that they were viviparous has gained common currency. 



Putnam adds to the above, judging from some data in his possession, that the 

 young are born in September and October, and further along remarks that they 

 are "undoubtedly" viviparous. 



The first young I obtained were secured on May 9, 1896. The little fishes 

 could move actively for a few moments, but as they were encumbered with much 

 yolk, they soon settled to the bottom and remained quiet. A large number of old 

 ones were in the water in which the young were found, and the mother of this lot 

 was not identified with certainty. Another lot of young obtained on September 5 

 of the same year were much farther along in their development. Some were pre- 

 served and others placed in various aquaria, where one lived to be 10 months old. 

 As before, the parent was not with certainty determined, simply because it was 

 taken for granted that they were viviparous and the ovaries only were examined. 

 Two other lots of young were obtained on June 5, 1897. O ne 0I these lots was in 

 the stage of the first lot obtained, with a large amount of yolk still present, while 

 in the other lot the yolk had almost entirely disappeared. These had been carried 

 in the gill cavity of the mother, and it became evident either that the fishes were 

 not viviparous at all, or that their viviparity was not nearly of the pronounced 

 character hitherto supposed. 



On March 11, 1898, 29 individuals were captured. Four were females with eggs 

 in their gill cavities. The youngest stage among these was at the end of segmen- 



