Kripke and Putnam have proposed that terms may be introduced to refer to theoretical entities by means of causal descriptions such as 'whatever causes observable effects O'. It is argued that such a reference-fixing definition is ill-formed and that theoretical beliefs must be involved in fixing the reference of a theoretical term. Some examples of reference-fixing are discussed e.g., the term 'electricity'. The Kripke-Putnam theory can not give an account of how terms may be introduced into science and then subsequently (...) be discovered to be non-referring. A modified account of reference-fixing is suggested in which terms such as Gilbert's electric effluvia' and 'phlogiston' can be introduced into science and then be found to lack a reference. (shrink)