212 THE STRUCTURE OF EVOLUTIONARY THEORY
about inheritance. No matter how exquisite or optimal the outcome for any one
organism, the results of the struggle cannot be imparted to offspring. (The capacity
for functional adaptation might, of course, be heritable and might evolve by
ordinary natural selection, but Roux never discusses this quite different issue.)
Weismann reacted to Roux's theory in a complex and ambiguous manner. He
always credited Roux as an antecedent of germinal selection (a reasonable
attribution, if only because an explicit metaphor of struggling parts can direct
another scientist's thinking to a truly selectionist theory, even if the original
proposal operated in a different domain). Weismann, particularly in his early work,
seems to credit Roux—incorrectly—as a true suborganismal selectionist:
"Functional adaptation is itself nothing else than the efflux of intrabiontic selective
processes" (1896, p. 15). Roux's theory, he argues in several passages, rests upon a
variational base, and is therefore Darwinian.
But, by 1904, Weismann had recognized that Roux's suborganismal struggle
could not operate as a theory of evolutionary change: "There is an essential
difference between personal and histonal selection, inasmuch as the latter can give
rise to adaptive structural modifications corresponding to the needs of the tissue at
the moment, but not to permanent and cumulative changes in the individual
elements of the tissue" (1904, volume 1, p. 248). "No one will be likely to suppose
that the distorted position of the spongiosa of a badly healed fracture could
reappear in the straight bone of a descendant" (ibid., p. 251).
Moreover, Weismann added, even the metaphorical linkages of Roux to
Darwin cannot be logically sustained. Most of Roux's examples do not include
competition among members of the same cell population (as in bone cells within
the developing femur), but between entirely different organs: liver cell with lung,
or kidney or heart. This process cannot be viewed as a struggle for existence at all,
but only as a sorting out of different "species" into their appropriate places: "The
struggle for existence and for descendants, in this case, is between two kinds of
cell which were different from the beginning, and of which one has the advantage
at one spot, another at another" (1904, volume 1, p. 248). Weismann then drew a
striking analogy* between different
*This remarkable passage anticipates our modern debates about the efficacy of species
selection. Weismann's analogy surely holds: this particular case involves no directional se-
lection, but only a sorting out and consequent balance among three species, each in its proper
place—just as lung, kidney, and heart develop where they should, and to their appropriate
size. However, if such a competition led not to balance and stability, but to differential birth
and death of the entities involved, then we could speak of directional species selection. This
argument cannot apply at a lower level within an organism, for lung cannot defeat liver
without destroying the entire system (but consider cancer as an event of this type).
Nonetheless, the higher level version remains potentially valid for competition among
species, for the success of one bird species over another will not cause an island to founder
into the ocean. Species selection does operate in this manner—and some critics (e.g.
Maynard Smith, 1988) have denied a creative status to this higher level because species
selection only sorts entities already shaped by organismal selection, a position that I shall
challenge in Chapter 8). After all, organismal selection also works only by sorting