"All good science consists, and all good philosophy consists, of lucky oversimplification or, if you prefer the term, idealization" wrote Popper a little while ago (1974, p. 976). Granted that "some oversimplifications are very much better than others" (op. cit., p. 977), it still must be stressed that "the complaint that somebody has oversimplified (unsupported by any specification of the serious problem thereby missed) is, a priori, always true, and therefore empty" (op. cit., p. 1045). After all, "if one does not learn from the simple cases, one is bound to go wrong on tackling the more complicated" (loc. cit.). And, anyway, "nobody can say, at every place, all things at once" (op cit., p. 1188); so that if we eschew simplification in one dimension we will only have to intensify it in another. Thus when Popper and I, in discussing verisimilitude, 'speak as though the class of all true statements about the world were an unproblematic totality' (Rosenkrantz, p. 197), we are not so naive as to suppose that scientific languages are not constantly evolving. See for example Popper (1959), p. 378: "I do not believe in hampering scientific language by preventing the scientist from using freely, whenever it is convenient, new ideas, predicates, 'occult' concepts, or anything else." (The reader is asked to accept my lack of naivety on trust.) But how far is it really necessary to take this fluidity of scientific language into account when discussing truthlikeness? When faced with theories A and B can we not combine the languages wherein they are formulated into a language adequate for them both? Then if there were some persuasive way of comparing A and B for truthlikeness in this mutual language, that would be a solid achievement, even though dependent on a certain modicum of idealization. In critical vein we might go on to ask whether truthlikeness comparisons were persistent under all extensions of the language employed (see Harris, 1974, pp. 164f., for relevant results about Popper's theory). But in fact it has not yet been possible to formulate any such canon of