ABSTRACT
However, it must first be appreciated that all the elements of quantitative geography must not and ought not to be lumped under one heading. It is not our purpose here to discuss essentially inductive, that is statistical, methodology;
we restrict ourselves to mathematical modelling on the basis that this has a more direct contribution to make to the evolution of geographical theory in the long term. This distinction, between the statistical and the mathematical, has usually not been well understood. Another area of weakness has been that there appeared to be little explicit connection between what mathematical modelling had to offer to geographical theory and what might be called the classical contr ibutions of such authors as von Thünen, Weber, Burgess and Hoyt, Christaller and Lösch-and these authors have provided the basis of much geographical textbook writing both before and after the advent of radical geography. (Perhaps their works constitute a neoclassical geography?) This was partly because the classical modellers had ventured into areas where the 1960s modellers had not the expertise to tread; and, more simply, because the effort of understanding what each perspective contributed to the other was not made. It could be argued that the contributions of modelling in the 1960s and early 1970s, exciting though they were at the time and useful though they remain in many ways, did not in fact address the central problems of geographical theory. However, this position has now changed and a brief articulation of the new contribution is a major purpose of our discussion.
