On Errors and Approximations

Abstract

In physics, as in other sciences, approximation is the rule of the day. The theoretical work involved in preparing an experiment and the interpretation of the experiment’s results are approximate in nature. Much purely theoretical work also involves approximation of one sort or another, while laboratory work yields results that are reproducible at best to within certain error limits (in a sense to be discussed below). The choice between different theoretical models for a given application depends critically on the degree of approximation required, and publication of an experimental datum without appropriate error limits is strongly discouraged as being essentially meaningless. It is therefore rather surprising that philosophy of science has so far paid scant attention to what is evidently a central matter in scientific practice.