lamella around the margin of the gill and up to the free border 

 of the other lamella. Whether jSlaments are always continuous 

 in this manner or not has not been determined, but in the sec- 

 tions that have been examined the same number of filaments are 

 constantly present on the two sides of any given water-tube. 



The filaments are very similar in size and appearance, except 

 those that are concerned in the formation of the inter-lamellar 

 junctions. These are many times as large as the others and diflfer 

 decidedly in shape as well as structure. Those placed next to 

 these modified filaments are somewhat larger than the remainder 

 but they do not otherwise differ in appearance or structure. 

 The number of filaments concerned in the formation of water 

 tubes is not entirely constant. Nineteen of the small filaments, 

 "between the large modified filaments, is a very common number 

 "but as few as seventeen and as many as twenty-two have been 

 aioticed. No attempt has been made to determine the relative 

 Jiumber of filaments for each water tube in different parts of the 

 ■gill but the variations mentioned occur within a space of ten or a 

 dozen tubes. 



Each of the smaller filaments is composed of a layer of sur- 

 face epithetium that incloses some connective tissue and a large 

 blood space. The connective tissue is so arranged that quite 

 universally a strand of tissue extends across the blood space (fig. 

 21, fs.) from one side of the filament to the other, so in cross 

 sections of the filament the blood space appears divided into two 

 nearly equal portions. This Kellogg (14) has quite naturally 

 taken for a functional division that allows the blood to pass 

 down one side of the filameflt and back the other. That this 

 is not actually its function is indicated by injections of the vas- 

 cular system that I have made, and by the connections of the 

 blood -Spaces of the filaments to the aflferent and efiferent vessels 

 of the gill. There is every indication that the blood moves in 

 the same direction on each side of the partition, if it is a com- 

 plete partition. The only reason that I can suggest to explain its 

 constant presence is that each acts as a brace to keep the filament 

 from swelling into a cylinder with the pressure of the blood, and 

 so partially close and interfere with the flow of water through 

 the inhalent ostia. That there is great need for braces of this 

 character in filaments shaped like these, where they are not 



