1918] Michael: Behavior of Salpa democratica 269 



According to the stratification hypothesis this table ought to show 

 an approximate equality between the abundance of solitary forms in 

 the warmer water and their abundance in the colder water, and also 

 between the abundance of aggregate forms in the warmer water and 

 their abundance in the colder water. This is obviously not the case. 

 The relations shown by table 10 are essentially the same as those shown 

 bj' table 4, solitary forms again appearing as most abundant and least 

 frequent in the warmer w'ater, while aggregate forms appear as most 

 abundant and most frequent in the colder water. Not only is this true, 

 but the difference in abundance of the solitary forms in the warmer 

 and that in the colder water has been increased from 93 in table 4 

 to 161 in table 10. Therefore, although the temperature did decrease 

 with increasing distance from the coast, the two generations were so 

 distributed with respect to position as to make it quite impossible for 

 the differentials revealed by tables 4 to 6 to have been due to stratified 

 areas of warm and cold surface water, or what amounts to the same 

 thing, to differences in position. 



3. Ostw^s.ld's Viscocity Theory 



The inadequacy of this theory to explain vertical migrations has 

 been pointed out before (Michael, 1916, p. xiv). It is not my pur- 

 pose, therefore, to discuss the theory in detail. But, as someone is 

 sure to claim that variation in viscocit.v of the water induced by 

 variations in temperature is responsible for the differentials observed 

 in the surface distribution of Salpa democratica, it is necessary to 

 forestall this claim. The matter will be considered on the basis of 

 three alternative assumptions: (a) that solitary forms or short chains 

 have the same specific gravity as aggregate forms or long chains; (b) 

 that solitary forms or short chains are heavier than aggregate forms 

 or long chains; and (c) that aggregate forms or long chains are the 

 heavier. 



It needs no argument to demonstrate that if, the first case (a) were 

 true, differentials in distribution of the two generations would be 

 impossible if due solely to viscocity effects. For if owing to an increase 

 in temperature, the viscocity decreases sufficiently to cause the indi- 

 viduals of one generation to sink, those of the other generation, being 

 of the same specific gravity, must also sink. 



In the second case (b), a decrease in viscocity might cause solitary 

 forms or short chains to sink while aggregate forms or long chains 



