The debate between realists and antirealists is too often taken to extremes. Critics of Laudan’s pessimistic induction and van Fraassen’s constructive empiricism overkill with Putnam’s miracle argument, the associated doctrine of preservation realism this has inspired and Psillos has pursued, and the idea of inference to the best explanation. A moderate realism, more in tune with the common-sense scientific realism often expressed by working scientists, is sufficient to counter the excesses of antirealism. Although often cited by realists of all creeds, common-sense realism is deeply at odds with the efforts of preservation realists to save abandoned terms of science with a putative reference on the strength of an application of the causal so-called theory of reference. On the other hand, “atom” and “molecule” are, it seems, terms that have “preserved their reference”, so that what we mean by them is what at any rate Dalton, if not Democritus, meant by them. This, I take it, is what common-sense scientific realism claims, and it is supported by Kripke’s and Putnam’s dabblings with extant terms of science such as “water”, which has led them to suggest that their extension has remained fixed. This paper takes a middle course, arguing against the claims of preservation realism regarding both defunct terms such as “caloric” and, challenging commonsense realism, some extant terms, namely “atom” and “molecule”. When the present reference of these terms came into effect is deferred to another paper.