252 THE STRUCTURE AND AFFINITIES OF HERDMANIA CLAVIFORMIS. 



They are undoubtedly comparable with the "Facher" first described by Kowalev- 

 sky ('74) in the postabdomen of Amaroucium. The membranes themselves are 

 exceedingly delicate. Flattened nuclei are found in them here and there, though 

 these are not numerous (PL XIX, Fig. 13, mb.). I describe these as membranes be- 

 cause the assumption that they are essentially of this nature accounts best for the 

 retention of the free cells in distinct masses and columns. It is probable, however, 

 that in reality they are composed of a much flattened network of fine fibres. 



The columns are not altogether indefinite in arrangement. The most constant 

 are the four situated on the ventral side of the body in relation with the two epicar- 

 diac tubes. Each tube has typically one on its ventral and one on its dorsal side. 

 Between the two on the ventral side there exists a large, distinct, constant space. 

 The membrane fining this space, which is of the same character as that already de- 

 scribed as enveloping the columns, is recognizable in nearly all sections, even where 

 the free mesenchyme cells are absent. This channel is always empty except for small 

 irregular masses of coagulum here and there. This I designate the subepicardiac 

 space (Fig. 8, sb'e'cr.). Other spaces around the viscera, evidently of the same 

 nature, are recognizable in many zooids; but they are less constant. These are 

 probably blood-spaces, though the almost entire absence of cells within them is diffi- 

 cult to understand on this supposition, for the possession of a great quantity of cells 

 of many varieties is one of the most characteristic features of ascidian blood. 



The free cells of the mesenchyme evidently play an important r61e of some sort 

 in the economy of the animal — as they do in so many other ascidians. The usual 

 supposition that this r61e is in connection with the nutritive processes is probably 

 correct, though direct evidence of this is wanting. Kowalevsky ('74) described the 

 abdomen of Amaroucium as being filled with fat cells, and Caullery ('95, p. 70) who has 

 given some attention to the development of the granules in Circinalium is also 

 inclined to consider them to be fat. 



A series of the cells becoming gradually filled with the yolk granules is shown 

 in Figures 17 to 17c. Before the granules appear the cells are as a rule nearly globular 

 (Fig. 17). Not infrequently, however, they display irregularities in the form of angles 

 and short broad processes (Fig. 17a). The cytoplasm is peculiarly dense and homo- 

 geneous, there being no fibrillar or alveolar or granular structure discoverable even 

 with the best powers (Zeiss oil immersion 2.0 mm., 1.30 ap.), at least so far as con- 

 cerns well-preserved tissues treated by the ordinary histological methods. In these 

 cells the cytoplasm stains uniformly and distinctly with hematoxylin after formalin 

 fixation. The first recognizable step toward the development of the granules con- 

 sists in the appearance within the cytoplasm of several round clear areas which do 



