ASEXUAL REPRODUCTION IN SAGARTIA 219 
Siphonoglyphs 
The condition of these groups as regards number of siphono- 
glyphs is represented in table 13. These are referred to as 
'unselected' specimens, but the conditions under which the 
collecting was done and the counts made allow of a considerable 
degree of selection in addition to the normal errors of random 
sampling. Each lot consists of a portion of a natural group of 
individuals collected at one time from a closely restricted 
locality, such as a single stone or a few stones of similar quality 
lying near together. Each lot was either taken immediately 
into the laboratory and examined or, when collected during the 
winter, taken to Cambridge and studied there. In order to 
determine the number of siphonoglyphs it was, of course, 
necessary to wait until the animals expanded, exposing the oral 
disc. Naturally, also, many of the specimens were expanded 
at one time and the removal of a few whose condition was 
noted caused many others to contract, necessitating much delay 
in completing the count. There was opportunity for uncon- 
scious selection because many were open at a time and because 
some show their condition clearly at a glance while others require 
close examination to reveal the state of their siphonoglyphs. In 
a few cases sections were necessary to determine the number of 
siphonoglyphs. Experience shows that external examination is 
not wholly reliable for this purpose and that, in case of any 
irregularities in the mouth region (cf. fig. 1, 4), only sections 
can give certain information. In each lot some specimens per- 
sistently failed to expand; some left the stones and were lost; 
and some underwent division probably as a result, in part, of 
the change in environment. The errors from all of these sources 
together undoubtedly render this table useless for any precise 
statistical analysis. Furthermore, the great variation among 
different lots indicates that, even if the accuracy of the numbers 
shown could be depended upon, the total numbers are much 
too small. Nevertheless, some idea of the relative magnitude of 
diglyphic, monoglyphic, and other classes may be obtained from 
the numbers given. As shown in the table, these groups gave 
