Richard Lewontin's (1970) early work on the "units" of selection initiated
the conceptual and theoretical investigations that have led to the hierarch
ical perspective on selection that has reached near consensus status today.
This paper explores other aspects of his work, work on what he termed "con
tinuity" and "quasi-independence". that connect to contemporary exploration
s of modularity in development and evolution. I characterize such modules a
nd argue that they are the true units of selection in that they are what ev
olution by natural selection individuates, selects among, and transforms.