Historical Constraints and the Evolution of Development 1087
right into the old wall of stymied practice—for the biology of his day knew no
methods for identifying the homologous generators that could mark a homoplastic
similarity as parallel rather than convergent. Unable to cash out his theoretical clarity
in actual practice, Simpson threw in the towel and admitted operational defeat (1945,
p. 9):
It is a complication that a third sort of process also produces similarities:
parallelism. The term is descriptive rather than explanatory and refers to the
fact that distinct groups of common origin frequently evolve in much the same
direction after the discontinuity between them has arisen, so that at a later
stage the phyla may have characters in common that were not visible in the
common ancestry but that tend, nevertheless, to be more or less in proportion
to the nearness of that ancestry. This proportional tendency distinguishes
parallelism from convergence, but the distinction is far from absolute. The
two phenomena intergrade continuously and are often indistinguishable in
practice.Simpson (1945, p. 10) also stressed the intermediate nature of parallelism in
phylogenetic inference, recognizing that even homoplastic characters usually record
reasonably close genealogical affinity (in their common origin from homologous
generators) in cases of parallelism, but must be regarded as confounders of affinity in
cases of convergence: "Homology is always valid evidence of affinity. Parallelism is
less direct and reliable, but it is also valid evidence within somewhat broader limits.
It may lead to overestimates of degree of affinity, but it is not likely to induce belief
in wholly false affinity. Convergence, however, may be wholly misleading, and a
principal problem of morphological classification on a phylogenetic basis is the
selection of characters that are homologous or parallel and not convergent."
In his 1961 book on Principles of Animal Taxonomy, Simpson continued to
express his frustration at the conceptual need, but operational impossibility, of
distinguishing parallelism from convergence. "The distinction of parallelism from
convergence is vital," he writes (1961b, p. 106). Fifteen years after his joint paper
with Haas, and their disagreement over geometrical versus causal definitions of the
terms, Simpson stated in frustration (1961b, p. 103): "Parallelism is the independent
occurrence of similar changes in groups from a common ancestry and because they
had a common ancestry. Some students (for example, Haas in Haas and Simpson,
19 46) have preferred a more purely descriptive definition, especially by the
geometrical model of parallel lines, symbolizing two lineages both changing but not
becoming significantly either more or less similar... Most taxonomists do, however,
consider that the term parallelism should be used only when community of ancestry is
pertinent to the phenomenon."
Simpson concludes his discussion (1961b, p. 106) with the clearest statement I
have ever read for citing homology of underlying generators as the basis of
parallelism, and on the joint operation of both overt selection and underlying
homology in the evolution of homoplastic structures by parallelism: