At the outset, the reader should be cautioned not to expect too much, for the task of determining phylogenies fairly bristles with difficulties and encounters many unanswered questions. Even when the evidence seems to be as copious and as complete as could be wished, different observers will put different interpretations upon it, as in the notorious case of the Steinheim shells. (In the Miocene beds of Steinheim, Wurtemberg, occur countless fresh-water shells, which show numerous lines of modification, but these have been very differently interpreted by different writers.) The ludicrous discrepances which often appear between the phylogenetic "trees" of various writers have cast an undue discredit upon the science and have led many zoologists to ignore palaeontology altogether as unworthy of serious attention. One principal cause of these discrepant and often contradictory results is our ignorance concerning the exact modes of developmental change. What one writer postulates as almost axiomatic, another will reject as impossible and absurd. Few will be found to agree as to how far a given resemblance is offset by a given unlikeness, and so long as the question is one of weighing evidence and balancing probabilities, complete harmony is not to be looked for. These formidable difficulties confront us even in attempting to work out from abundant material a brief chapter in the phylogenetic history of some small and clearly limited group, and they become disproportionately greater, when we extend our view over vast periods of time and undertake to determine the mutual relationships of classes and types. If the evidence were complete and available, we should hardly be able to unravel its infinite complexity, or to find a clue through the mazes of the labyrinth. "Our ideas of the course of descent must of necessity be diagrammatic." (D.H. Scott, "Studies in Fossil Botany", page 524. London, 1900.)