In this paper, I examine a number of alternative global structures for Newtonian space-time, and corresponding Newtonian theories of mechanics and gravitation. I argue that since these theories differ only with respect to questions concerning the relative distribution of inertial and gravitational forces, the choice between them is a matter of convention. Therefore, the global structure of Newtonian space-time is also a matter of convention. Since this result is based on a consideration of the nature of inertial and gravitational forces, (...) rather than on general reductionist principles, it is a "limited conventionalist" result. (shrink)