Frege claims that there is a close connection between proof and conceptual
analysis. Specifically, he claims (a) that some propositions can be proven only
on the basis of conceptual analyses, and (b) that this is the motivation for the
analysis of arithmetic that he provides inGrundlagen. It has been argued that
Frege's practice shows that he can't be taken at his word here, and that the
so-called analyses provided in Grundlagenare not in fact appropriately so
called. This paper has two purposes. The first is to clarify the Fregean
understanding of the connection between analysis and proof, and to argue that
Frege's foundational work in arithmetic is indeed centrally concerned with
conceptual analysis. The second is to clarify the role of formal systems in
demonstrating relations of logical entailment when proof, entailment, and
analysis are understood in the Fregean way. It turns out that the role of models
in assessing relations of logical entailment looks significantly different from
the Fregean point of view than it does from the more modern Hilbert/Tarski point
of view, and that some central metatheoretic results are of considerably
different significance from the two different viewpoints. A central motivation
for this work is the view that when we lose sight of the now-unfashionable
Fregean point of view, we lose something significant.