We construct in detail a 2 + 1 dimensional gauge field theory with fin
ite gauge group. In this case the path integral reduces to a finite su
m, so there are no analytic problems with the quantization. The theory
was originally introduced by Dijkgraaf and Witten without details. Th
e point of working it out carefully is to focus on the algebraic struc
ture, and particularly the construction of quantum Hilbert spaces on c
losed surfaces by cutting and pasting. This includes the ''Verlinde fo
rmula.'' The careful development may serve as a model for dealing with
similar issues in more complicated cases.