114 I. IJIMA : HEXACTINELLTDA. I. 



to think that the hitter process may at times possihly take place, 

 some of the originally transverse beams becoming transposed into 

 the oblique. 



Secondly with respect to the longitudinal system (column II), 

 there is observable a general rise in the value of figures from 

 top to bottom of the column, indicating a continual addition to 

 its beams along with the growth of the sponge. This corres- 

 ponds with the fact that even in the oldest specimens there exist 

 here and there such longitudinal beams as seem to represent 

 different stages of splitting lengthwise and separating into two. 



If now the transverse beams should cease to multiply at an 

 early period while the longitudinal persist in multiplying, the 

 numerical proportion of both in young specimens must be quite 

 different from that in old specimens. This is likely tlie ex- 

 planation of the fact to be noticed in the table that, while down 

 to the specimen of 110 mm. length (No. 19) the excess of dif- 

 ference between the numbers of the two kinds of beams in each 

 specimen stands on the side of the transverse system (I), the 

 case is reversed in most of the remaining larger specimens. 



For E. simplex, F. E. Schulze ('95, p. 23), by counting and 

 comparing the number of the two kinds of skeletal beams in 

 half a dozen young specimens of various size, has reached the 

 conclusion that, during the growth of tlie sponge, the transverse 

 beams increase considerably in number while the longitudinal do 

 so to but an insignificant degree. To wit : the smallest specimen, 

 0Ö mm. long, had 25 transverse and 28 longitudinal beams 

 against 40 and 30 respectively of a specimen 110 mm. long. In 

 order to see how far Schulze's above conclusion can be verified 

 with E. mai'shalU, it would be necessary to take into considera- 

 tion only those young specimens of that species in wiiich both 



