Rundle et al. (1998) recently attempted to test the founderflush model of speciation by Carson (1968) and the genetictransilience model of speciation by Templeton (1980a). They correctly noted that these two hypotheses for founder-influenced speciation are not identical, but do require similar demographic situations and can co-occur (Carson and Templeton 1984). The hypotheses they state they are testing are therefore the combined founder-flush/genetic transilience theories of speciation. As will be evident later, their experimental design precludes tests of the founder-flush model, but their design does remain a valid experimental test of genetic transilience. Genetic transilience is particularly amenable to experimental testing because the primary motive behind the genetic transilience theory was to apply population genetics "to the problem of speciation in such a manner as to greatly augment the explanatory and, more importantly, the predictive powers of speciation theories" (Templeton 1980a, p. 1011). The model of genetic transilience when published in 1980 was, and continues to this day, to be a rarity in the speciation literature because it not only explained some observed patterns of speciation, it also predicted when founder events were unlikely to lead to speciation. Indeed, the theory of genetic transilience requires many conditions to be satisfied to make speciation likely after a founder event, leading to the overall conclusion "that genetic transilience is an unlikely mode of speciation and that the vast majority of founder events have no significance whatsoever with regard to the process of speciation" (Templeton 1986, p. 500-501). Although rare in a global sense, genetic transilience is readily testable because it makes specific and detailed predictions about the conditions in which speciation associated with founder events are likely. The number of predictions is quite lengthy (Templeton 1980a,b) and summaries are given in tables 1 and 2 of Templeton (1980a). The only predictions that will be discussed in detail in this comment are those that directly relate to the arguments of Rundle et al. (1998).