The work recorded in the following paper was undertaken with a view to testing the hypothesis frequently brought forward that the islets of Langerhans of the pancreas yield an internal secretion whose function has some relation to the utilization of dextrose in the vertebrate organism. The primary cause of diabetes is, according to Macleod,1 the ' want of disruption of the dextrose molecule and its consequent non-utilization by the organism.' Further, it is well established from the experiments of von Mering and Minkowski and others that total extirpation of the pancreas induces violent diabetes in dogs, cats, pigs, and frogs, whilst the supplying of ordinary pancreas per os fails to alleviate the condition. Now, this relation between pancreas and diabetes is held by many to depend more particularly upon the epithelial cell groups, commonly termed ' islets of Langerhans,' which are a characteristic of the pancreas in all vertebrates. These islets have a rich blood supply, and have been held by various writers to be elements sui generis. Comparatively recently the view has been advanced (not for the first time by any means) that they are not a distinct tissue, but simply a phase of ordinary pancreatic tissue. The present is not an appropriate opportunity upon which to discuss the merits of these opposing views; that is reserved for another occasion. Reference is made to them here simply to clear the ground for what follows. One of us2 has shown that in Teleostei, at all events, these