Ramsey eliminability and the testability of scientific theories

The role in scientific theories of theoretical terms, which do not correspond directly with observables, continues to be a matter of discussion and some puzzlement. If a theoretical term in a particular theory is definable, it can, of course, be eliminated directly without altering the semantic content of the theory. In order to make room for theoretical terms that are not definable, Ramsey (1931, Chapter DC), introduced the notion of eliminability. Roughly speaking, a theoretical term is eliminable, in the sense of Ramsey, if all the empirical claims of the theory can be made without invoking it. Ramsey apparently believed that in a properly formulated scientific theory all theoretical terms would be eliminable, but that not all need be definable. Sneed (1971), however, has produced examples of simple theory-like systems containing theoretical terms that are not eliminable in the Ramsey sense.