The most primitive reptiles that we know had no less than thirty-seven pairs and four single bones in the skull. The crocodiles have but twenty-four pairs and six unpaired bones in the adult; the turtles have twenty-two or twenty-three pairs and five or six unpaired bones; the lizards have at the most twenty-nine pairs and five single bones. But not all are the same. The crocodiles have three or four pairs that have been lost in the turtles; the turtles, one pair that is fused in the crocodiles; the lizards, several bones lost or fused in the turtles, and so on. All reptiles since Triassic times have lost four bones in the pectoral girdle, and all have lost some bones of the feet. The persistence or loss of bones furnishes many certain evidences of relationships and descent. Each order must have descended from ancestors that had the persistent bones; they could by no possibility have regained them when once lost.
The relative importance of all such characters in classification is, however, largely a matter of the classifier's personal opinion. No two persons see them from the same viewpoint and consequently no two persons whose opinions deserve consideration ever wholly agree as to the value of characters in classification. It is only in the gradual crystallization of opinions that stability finally results, and this crystallization is never complete. So long as science endures, new facts will be discovered to influence our opinions. Any system of classification, then, merely represents the present state of our knowledge and the consensus of the opinions of those best qualified to decide as to their value, more or less influenced by the classifier's individual opinions. No classification will ever be perfect, for perfection postulates complete knowledge. Fortunately, however, the increase of knowledge affects less and less the major principles, and more and more subordinate details.
