A perfect taxonomy is one which would perfectly express all the facts in the evolution and development of the various forms. It would recognize all the evidence from the three ancestral documents, palæontology, morphology, and ontogeny. It would consider structure and form independently of adaptive or physiological or environmental modifications. It would regard as most important those characters which had existed longest unchanged in the history of the species or type. It would regard as of first rank those characters which appear first in the history of the embryo. It would regard as of minor importance those which had arisen recently in response to natural selection or the forced alteration through pressure of environment, while fundamental alterations as they appear one after another in geologic time would make the basal characters of corresponding groups in taxonomy. In a perfect taxonomy or natural system of classification animals would not be divided into groups nor ranged in linear series. We should imagine series variously and divergently branched, with each group at its earlier or lower end passing insensibly into the main or primitive stock. A very little alteration now and then in some structure is epoch-making, and paves the way through specialization to a new class or order. But each class or order through its lowest types is interlocked with some earlier and otherwise diverging group. =Defects in Taxonomy.=--A sound system of taxonomy of fishes should be an exact record of the history of their evolution. But in the limitations of book-making, this transcript must be made on a flat page, in linear series, while for centuries and perhaps forever whole chapters must be left vacant and others dotted everywhere with marks of doubt. For science demands that positive assertion should not go where certainty cannot follow. A perfect taxonomy of fishes would be only possible through the study, by some Artedi, Müller, Cuvier, Agassiz, Traquair, Gill, or Woodward, of all the structures of all the fishes which have ever lived. There are many fishes living in the sea which are not yet known to any naturalist, many others are known from one or two specimens, but not yet accessible to students in other continents. Many are known externally from specimens in bottles or drawings in books, but have not been studied thoroughly by any one, and the vast multitude of species have perished in Palæozoic, Mesozoic, and Tertiary seas without leaving a tooth or bone or fin behind them. With all this goes human fallibility, the marring of our records, such as they are, by carelessness, prejudice, dependence, and error. Chief among these defects are the constant mistaking of analogy for homology, and the inability of men to trust their own eyes as against the opinion of the greater men who have had to form their opinions before all evidence was in. Because of these defects, the current system of classification is