Group selection is said to occur when the traits of groups that systematica
lly out-reproduce competing groups eventually come to characterize the spec
ies. Evolutionists have long disputed over the degree to which group select
ion is effective-that is, over the degree to which social group characteris
tics can be attributed to selection on these characteristics. The intractab
ility of this. controversy arises from three ambiguities in the natural sel
ection metaphor that manifest themselves when that metaphor is shifted to t
he group level: (1) uncertainty about what constitutes the analogue for "fl
ock" in the group level metaphor; (2) uncertainty about how to identify the
group "parents" of offspring groups; and (3) uncertainty about what consti
tutes a group trait for the purposes of group selection. When group selecti
on is specified as a theory about the evolution of emergent properties of g
roups through differential group productivity mediated by quantitative inhe
ritance of group traits, these ambiguities disappear.