THE LIFE-HISTORY OF NUCDLA DELPHINODONTA. 359 



These branches become trough-shaped, with the open part of 

 the trough directed away from the inner plates. Finally, the 

 free edges of the troughs come close together, as described 

 in connection with the other set of filaments or plates. 



The chitinous material at the bases of the two sets of 

 plates also becomes trough-shaped, and has the open portion 

 of the trough directed away from the plates. Thus the 

 chitinous support of the gill consists of two series of troughs, 

 bent so as to form tubes, each of which is connected by one 

 end to the side of a larger trough that runs lengthwise of the 

 gill. The whole might be compared to a large trough with a 

 series of spouts leaving each side, the individual spouts of 

 the two series being placed opposite each other. Later, 

 bridges are built across the main trough in the intervals 

 between the side spouts. The whole system is in direct 

 communication with the blood-spaces of the gill, but 

 probably is not concerned with the circulation of the blood. 



The two sets of plates do not lie parallel to eacli other, 

 but they grow away from each other at an obtuse angle. 

 The inner plates grow almost in an anterior direction, and 

 the outer plates grow laterally and a little posteriorly, so that 

 the angle formed by the two sets of plates on the two sides 

 of the gill is visible when the animal is viewed from the side. 

 The suspensory membrane, formed by the growth of the 

 mantle at the base of each gill, makes it possible for the gill 

 to take up this position. 



The filaments begin to grow into flattened triangular plates 

 about the time that the fourth division of the gill is formed. 

 This is accomplished by slow, unequal growth, and throws 

 no light on the phylogeny of the gill. It seems to be a matter 

 of individual opinion whether each of the plates should be 

 considered to be homologous with a descending filament of 

 an ordinary Lamellibranch gill, or whether it should be con- 

 sidered to be homologous with both a descending and an 

 ascending filament. 



The adult structure of the gill of Nucula has been so care- 

 fully and accurately described by others, that were it not for 



