The history and theoretical role of the concept of a "replicator" is discus
sed, starting with Dawkins' and Hull's classic treatments and working forwa
rd. I argue that the replicator concept is still a useful one for evolution
ary theory, but it should be revised in some ways. The most important revis
ion is the recognition that not all processes of evolution by natural selec
tion require that something play the role of a replicator.