ASEXUAL REPRODUCTION IN SAGARTIA 
of four to twelve. As a matter of fact, the smallest number 
added in any case of complete regeneration as recorded in these 
tables was five and the largest number eleven. 
The average results of repeated divisions and regenerations 
may be derived from the data at hand. On the right of table 
9 are given the numbers of mesenteries regenerated in zones of 
different types. The same data are given, considering sepa- 
rately each lateral half of a regenerating region, on the right of 
table 10. As explained in connection with the latter table, the 
average number of complete mesenteries in regenerating zones of 
all types is 8.4. It is obvious, therefore, that repeated divisions 
into two parts followed by regeneration will tend toward an 
average of approximately 17 complete mesenteries. Table 11 
exhibits the number of complete mesenteries after regenera- 
tion for the specimens recorded in tables 3 to 6, and the average 
number of such mesenteries for each siphonoglyphic class and 
for the Whole. The latter average is 15.8. The fact that this is 
below the number toward which repeated division and regenera- 
tion tend, points to a still lower number of mesenteries in the 
form resulting from ontogenetic development. 
The average number of complete mesenteries, before division, 
of specimens represented in table 4 (including only those indi- 
viduals all of whose fission products were available for record) 
is 13.5, a number considerably below the average for all the 
specimens of tables 3 to 6. In each of the cases given in table 4, 
the two closely succeeding divisions resulted in rapidly increas- 
ing the number of complete mesenteries. Although it may be 
purely a coincidence that apparently multiple divisions have, in 
these observed instances, occurred in specimens with a low num- 
ber of complete mesenteries, it is possible that such divisions 
serve in an adaptive way to bring about a rapid increase. 
Whether this is correct or not, certainly the variations described 
for the processes of fission and regeneration are adequate to 
account for much wider variations in form of regenerated speci- 
mens than have been encountered. This indicates clearly the 
probability that there are correlations in these variations that 
are as yet unproved. 
