where m is the mean migration rate. The term m in (A1) vanishes with deterministic migration, and this is the only difference between the two diffusions that they compare in their figure 4. However, if we ignore selection, which we agree has no effect on the infinitesimal variance in the weak-selection limit, the exact expression for Var(p tϩ1 ͦ p t ) is easily computed for their model as: We wish to address some of the important issues raised by Coyne et al. (2000) in their critique of our Perspective on the contrasting views of Fisher and Wright (Wade and Goodnight 1998). Many of the points that they raise illustrate important differences between their view and ours on the content of evolutionary theory and on the experimental approaches used to test and investigate evolutionary hypotheses in laboratory and field studies. We believe that a controversy like this can be useful because it helps to identify gaps in the current understanding of evolutionary geneticists and to frame questions for future research. However, we also believe that, ultimately, it can be resolved only by continuing to collect more experimental data and by integrating the theoretical and empirical findings of the past 20 years more fully into existing thought. Only by broadening the range of data and types of experiments considered relevant can features like epistasis and population genetic structure be understood. By moving from parsimony to true understanding, we hope to resolve the differences between the two schools of thought, which have their roots in the Fisher-Wright debates (e.g., Provine 1971, Lloyd 2000.Empirical studies cited in our Perspective have had a formative influence on our views of Fisher and Wright, the relationship between interdemic and group selection, and the role of gene interactions (epistasis) in evolution. As Coyne et al. (2000) point out, our 1998 paper, by design, is a ''Perspective'' and we do not respond directly to the specific objections to Wright's shifting balance theory (SBT) that they raised in an earlier paper (Coyne et al. 1997). Instead, we reexamine the theories of both Fisher and Wright in light of new theoretical findings and new empirical data gathered in the 60 plus years since the original theories were put forward. We conclude that both theories, at least in their idealized original versions, have difficulty incorporating important features of natural populations (see table 2 in Wade and Goodnight 1998) because of the assumptions that differ between them. We argue that this limits the application of each theory to different domains. We conclude that many of the recent theoretical and empirical findings are not, strictly speaking, explicit in either theory and can be viewed as ''enriching'' both of them, especially Wright's. Many of these recent findings were discovered by ourselves and our colleagues in attempts to evaluate more rigorously some of the ideas that Wright expressed only in words and also in attempts to provide an empirical foundation to the study of adaptive evolution in genet...