112 T. IJIMA : HEXACTINELLTDA. I. 



relation to their mode of multiplying themselves, which is by- 

 splitting and gradual separation into two of an originally single 

 beam, as was first shown by F. E. Schulze ('95, p. 24). More- 

 over, the continually recurring rise and fall in the value of 

 fio;ures in the two columns make it manifest that the number of 

 beams in both systems, but especially in the transverse, is sub- 

 ject to certain, often very considerable, variations according to 

 individuals. Under such circumstances there is but little prospect 

 of accurate inductive inferences being made from the annexed 

 table, unless it be known what allowances to make for variations 

 in individual cases, which is certainly not known. Nevertheless, it 

 seems to me that the general trend of figures in columns I and 

 II is anticipated by certain facts observed in the manner of the 

 arrangement of the beams. 



Firstly as regards the transverse system, it may be considered 

 as a general rule that the distance between each two beams is 

 widest where the sponge-body shows greatest ventricosity, i. e., at 

 or somewhat below the middle of its length. Toward the base, 

 the interval usually lessens somewhat or may remain nearly the 

 same. So also toward the upper end in old specimens. In 

 young specimens, on the other liand, it grows superiorly on the 

 whole gradually and continually less and less, until, at the 

 marginal zone close to the juncture with the sieve-plate, a con- 

 dition is reached which is strikingly different according as the 

 new-formation of transverse beams is taking place or not. It is 

 in that zone of comparatively small specimens only, — say of a length 

 under 70 mm. or 80 mm., at any rate of not over 100 mm., — that 

 there exist indications of the transverse beams undergoing active 

 multiplication. I have before called attention to the primitive 

 character of this outwardly smooth marginal zone, which is yet 



