[12] Brooks takes a very different view of the nature of the parts in Salpa. He says, No. 7, p. 322, "The atrium of Salpa, when first observed, was composed of two broad lateral atria within the body cavity, one on each side of the branchial sack, and a very small mid-atrium.... The lateral atria do not however, as in most Tunicata, remain connected with the mid-atrium, and unite with the wall of the branchial sack to form the branchial slits, but soon become entirely separated, and the two walls of each unite so as to form a broad sheet of tissue, which soon splits up to form the muscular bands of the branchial sack." Again, p. 324, "During the changes which have been described as taking place in the lateral atria, the mid-atrium has increased in size.... The branchial and atrial tunics now unite upon each side, so that the sinus is converted into a tube which communicates, at its posterior end, with the heart and perivisceral sinus, and at the anterior end with the neural sinus. This tube is the gill.... The centres of the two regions upon the sides of the gill, where these two tissues have become united, are now absorbed, so that a single long and narrow branchial slit is produced on each side of the gill. The branchial cavity is thus thrown into communication with the atrium, and the upper surface of the latter now unites with the outer tunic, and the external atrial opening is formed by absorption." The above description would imply that the atrial cavity is a space lined by mesoblast, a view which would upset the whole morphology of the Ascidians. Salensky's account, which implies only an immense reduction in the size of the atrial cavity as compared with other types, appears to me far more probable. The lateral atria of Brooks appear to be simply parts of the body cavity, and have certainly no connection with the lateral atria of simple Ascidians or Pyrosoma. The observations of Todaro upon Salpa (No. 38) are very remarkable, and illustrated by beautifully engraved plates. His interpretations do not however appear quite satisfactory. The following is a brief statement of some of his results.