The opportunity of watching these fishes in their natural element, and keeping many of them alive for hours or days in our glass tanks, was very instructive, and suggested comparisons not dreamed of before. Our arrangements were very convenient; and as the commander of the steamer allowed me to encumber the deck with all sorts of scientific apparatus, I had a number of large glass dishes and wooden tubs in which I kept such specimens as I wished to investigate with special care and to have drawn from life. One of the most striking changes made by J. Müller, in the classification of the spiny fishes, was the separation into a distinct order, under the name of Pharyngognathi, of all those in which the pharyngeal bones are soldered together. With these the illustrious German anatomist has associated a number of soft-rayed types, formerly united with the Pickerels and Herrings, and characterized by the same structure. It would thus seem that there is here a definite anatomical character easily traceable, by the aid of which a vast number of fishes might be correctly classified. But the question at once arises, Are these fishes truly related to one another, and so combined in this new order of Pharyngognathi as to include all which properly belong with them, and none others? I think not. I believe that Müller has always placed too much value upon isolated anatomical characters; and, while he was undoubtedly one of the greatest anatomists and physiologists of our age, he lacked zoölogical tact. This is especially evident with reference to the order of Pharyngognathi, for though the Scomberesoces have fixed pharyngeals like Chromides, Pomacentrides, Labroids, Holconotes, and Gerrides, they have no real affinities with these families. Again, the character assigned to this order is not constant even in the typical Pharyngognathi. I have found Chromides and Gerrides with movable pharyngeals; in the genus Cychla they are normally so. It is therefore not out of place to state here that the Chromides of South America are in reality closely related to a group of fishes very generally found in the United States, known as Pomotis, Bryttus, Centrarchus, etc., and usually referred to the family of Perches, from which they have, however, been separated by Dr. Holbrook under the name of Helichthyoids. They not only resemble the Chromides in their form, but even in their habits, mode of reproduction, peculiar movements, and even in their coloration. Cuvier has already shown that Enoplosus is not a member of the family of Chætodonts, and I may now add that it is a near relative of the Chromides, and should stand by the side of Pterophyllum in a natural system. Monocirrus of Heckel, which I consider as the type of a small family under the name of Folhidæ, is also closely allied to these, though provided with a barbel, and