Was it the character by which the frog is found to belong to the class of reptiles, which was first apparent? By no means. It appeared first, under the form and structure of a fish, and not under the form and with the characters of a reptile. The lowest form of vertebrated animals was first developed in the earlier changes of the egg, before the class to which that animal belonged could be recognized. (1849, p.28)

Likewise, embryonic starfish do not first develop the plates and suckers that mark their group, but rather the "forms which would lead us to mistake them for Polypi or Medusae" (p.28). (In Agassiz's system, the Coelenterata [including polyps and medusae] are a lower type of the embranchement Radiata, which contains Echinodermata [including starfish] as a higher group. This comparison does not violate the immutability of the four Baupläne.) Any competent naturalist can recognize whether a fetus will become a domestic cat (that is, he can determine the species of the embryo) before its generic characters appear (four molars in the upper jaw, three in the lower; retractile claws). Moreover, varietal characters (coat color) and even individual peculiarities (playfulness) precede the eruption of the molars. "In short, everything takes place in the reverse order from what it is supposed in this [von Baer's] system" (1849, p.28).^*

Agassiz then unveils his ambitious plans for recapitulation as a working doctrine:

There is a gradation of types in the class of Echinoderms, and indeed in every class of the animal kingdom, which, in its general outlines can be satisfactorily ascertained by anatomical investigation; but it is possible to arrive at a more precise illustration of this gradation by embryological data . . . The most special comparisons of these metamorphoses [in ontogeny] with full grown animals of the same type, leads to the fullest agreement between both . . . These phases of the individual development are the new foundations upon which I intend to rebuild the system of zoology. (1849, p.26)