by invoking a criterion of visual appearance against the fact of latent potential, but they could not survive an attitude that came close to denying process altogether.

Mendel's Resurrection, Haeckel's Fall, and
the Generalization of Recapitulation

The preformationist attitude had not destroyed recapitulation. After all, Weismann was an enthusiastic supporter of the biogenetic law and Roux, in public, was at least indifferent. Yet the shift of attention to the germ and its structure redirected the study of heredity and led ultimately to the Mendelian synthesis. Experimental embryology had abandoned recapitulation as unfashionable, but genetics would render it untenable.

Early Mendelians had the same general reasons as students of Entwicklungsmechanik for neglecting recapitulation. As a putative "structure" in the fertilized ovum (and elsewhere), the gene was a perfect particle to support the biases of preformationist thought about early organization. "Benign neglect" again played its role. Several famous Mendelians began their careers as recapitulationists in the tradition of speculative phylogeny. Bateson (1886), Morgan (1891), and Castle (1896) had sought the origin of vertebrates in the embryology of primitive chordates; Davenport (1890) had tried to unravel the history of bryozoans from the ontogeny of modern forms. After their conversion to Mendelism, most of these men never mentioned the biogenetic law in print.

But the main impact of Mendelism was much more specific: it ultimately disproved the two " laws" of evolution that recapitulation required for its general occurrence—terminal addition and condensation. The more astute recapitulationists had long recognized that a catalog of cases could provide no ultimate justification for their beliefs. As long as the mechanism of heredity lay shrouded in mystery, recapitulationists could always postulate a convenient and purely hypothetical set of laws to yield their preferred results. The laws of terminal addition and condensation were of this type. Beyond the vague analogy to memory pursued by some Lamarckians, they had never had any justification beyond the argument: (1) recapitulation is true; (2) these laws yield recapitulation as a general result; (3) these laws must be true. This argument is neither illogical nor circular (one can, in theory, test the first statement by accumulating more and more cases, and then argue that no other laws fit the second statement). It is unsatisfying because it can be vindicated only by induction from its results. While the mechanism of heredity remained unknown, there was no