It was only when, after a close examination of the sternal apparatus of one hundred and thirty species, which he carefully described, that he arrived (pp. 177-183) at the conclusion - astonishing to us who know of L'Herminier's previous results - that the sternum of birds cannot be used as a help to their classification on account of the egregious anomalies that would follow the proceeding - such anomalies, for instance, as the separation of Cypselus from Hirundo and its alliance with Trochilus, and the grouping of Hirundo and Fringilla together.