If one is to relativise truth then an immediate query is: "to what?" A frequent answer is: "to a framework of beliefs, a conceptual scheme, or something of that sort." More usually, such a view is defended within an anti-realist/coherence host theory of truth (see, e.g., Young 1994). If located within a realist/correspondence host theory, then this is usually as a prelude to such relativism's dismissal (see, e.g., Whyte 1993). Unusually, Jack Meiland has attempted to defend a version of (what I shall call for short) framework relativism within the broad constraints of a realist/correspondence theory of truth (Meiland 1977). In a previous paper in this journal (1994) I outlined and defended an explication and elaboration of Meiland's correspondence truth relativism (and I assume knowledge of that paper's content for this brief note). The version that emerged was weak (in Chris Swoyer's sense; see Swoyer 1982: 92, and my 1994) and relativised truth not to a substantive framework of belief but to a conceptual scheme (or something similar). This might seem a poor result for the relativist. The more interesting versions are strong (again in Swoyer's sense) and/or, rather than relativising truth to a mere conceptual scheme (or language), relativise it to a framework of substantive propositions (or beliefs). As in the last paper, I mean this latter distinction in much the same way as outlined by Israel Scheffler (1967: 36). What, then, is wrong with strong relativism and/or relativisation to a framework of substantive beliefs? I have remarked that intractable difficulties face such views (1994: ) and, in what follows, I will outline why and defend my choice of the weak conceptual scheme variety as the version of correspondence truth relativism most deserving of further study.
Abstract In a previous paper, attempting to explicate a viable truth relativism within a realist/correspondence theory of truth and using something like a framework of belief or a conceptual scheme as the relativiser, I remarked that "I (was) persuaded that having a body of substantive assertions (a theory, say) as the relativiser has intractable difficulties..." (Davson-Galle 1994: ). In what follows I attempt to substantiate that judgement.
 Using the weak/strong and framework of substantive belief/conceptual scheme distinctions gives us a matrix of four thesis types (with the latter distinction relabelled to avoid holist overtones).
 Strong propositional relativism is both the most radical and most objectionable of the four types. If the host theory is a realist/correspondence theory then, not just the propositional web, but the world, is in truth-making role. If there is just one world and some proposition, P, is to be about it then strong relativism's demand that P be true relative to one propositional web (W1) and false relative to another (W2) seems incoherent (see Newton-Smith 1982 and 1981: 37). Nor would it help to eliminate the "one world" assumption and, constructivist style, have propositional webs not so much the "measure of all things" but makers of worlds (or, less dramatically, but still pluralistically, corresponders to some members or other of a set of pre-existing multiple realities). That web PW1 creates/corrresponds to Reality 1 and P is true of that reality, while PW2 does not create/correspond to Reality 1, but does, perhaps, create/correspond to Reality 2 and P was not true of that second reality, seems nothing of interest to truth relativism. It would seem that the prior task for any proposition in such an ontologically lush referential situation would be to settle which world was under discussion. If P is about R1 then it won't be true of R2 simply because it isn't about it. Perhaps some other, counterpart, proposition will be false of R2 but that isn't P. In short, all that is manifest is an issue of reference clarification, not relativism. We don't have P true relative to PW1 (and of its attendant R1) and false relative to PW2 (and of R2). We simply have P1 true, simpliciter, of R1 and P2 false, simpliciter, of R2. Referential ambiguity is no source of relativism.
 Keeping the relativiser as propositional web but shifting to weak relativism is also unpromising. The whole point of weak relativism is that P, though able to be stated in terms of one framework, is not statable in another's. Given this, the apt relativiser is not a propositional web but a categorial web. If some P is statable in terms of one propositional web and not another, then this would be in virtue of those webs employing different concepts, not in virtue of them being different in their substantive propositions. If, however, one can't have strong propositional web relativism, and weak propositional web relativism collapses away to weak categorial web relativism, why not the other of our possibilities - strong categorial web relativism?
 On this suggestion, one is to have the same P being true relative to one categorial web, CW1, but false relative to another, CW2. On the face of it, such a thesis looks impossible. If there are two webs of categories then how is one to get the same P expressible in each? Much will depend on how one individuates webs and how much conceptual overlap two webs can have, yet be deemed different. With zero overlap, that is with untranslatable schemes, "same P" is impossible. On the other hand, if the criteria for web individuation were such that there were to be enough conceptual/categorial overlap for each to have the resources for expressing P, then if it is the same proposition picking out the same state of affairs in the world, then how can the rest of the categorial web manage to mediate the truth/falsity making power of the world such that P is true relative to CW1 but false relative to CW2? (This is, after all, meant to be correspondence truth relativism under explication.) Which brings me to weak categorial web relativism: is it any less open to objection than its cousins proved above to be? I think so, but why?
 Before proceeding further, I would like to make a distinction among possible relativisers here. Consider three users of "witch": the "witchist", the "a-witchist" and the "agnostic" about witches. They utter three propositions, respectively:
 So, is the relativiser to be a mere categorial repetoire or should it also be referentially operative? Most of those inclined to speak in terms of truth relative to a conceptual scheme, or whatever, seem to have the latter alternative in mind. One has one semantic agent carving up the world using one set of categories and another employing another. Now, as our target here is weak relativism, the point is that p is to be simply not expressible in the other web's terms. Clearly the sort of contrast that our above three agents have in their webs, even their operative webs, is not enough to deny them the conceptual resources to express the other's propositions. That they share categorial repetoires seems to guarantee this. So, appeal is to be made to some more conceptually foreign categorial web before one has the right sort of alternative relativiser. But might we not demand of the relativisers that, for us to speak of P being true relative to CW1 but not to CW2, CW1 and CW2 be not just distinct categorial repetoires, but also be deemed (positively) referentially operative respectively by two cognitive agents? Complying with this extra demand creates problems, however, as the web becomes a sort of propositional web.
 The first of these concerns the intelligibility of the denial of an operative web's categorial commitments. Take the witchist web as CW1 (and contrast it with some foreign/alien web which doesn't possess the concept of witchness at all). Now what of the a-witchist's P2? If "witches exist" (P4, say) is expressible, then, one would have thought, "witches do not exist" should be too. And, if P4 is to have its truth value (weakly) relative to some web then, one would have thought, P2 would have its truth value relative to the same web. Yet with one world both can't be true. Let us say further that the world is unable to be categorised by "witch." Presumably then proposition P4 is false (relative to the witchist's CW1) and P4 is true (again relative to CW1). Yet this scenario is tantamount to rejecting part of the witchist CW1's categorial commitments. But if "witch" is not positively categorially operative then why would one demand that P2's truth value be relative to a web in which "witch" was (putatively) positively referentially operative? (Nor does it help matters if we start with the a-witchist's CW2). Having the categorial web as mere repetoire escapes these paradoxical scenarios.
 Another difficulty for operative webs as relativisers concerns a criterion of satisfactoriness raised by Harvey Siegel (1987:12). If one is to have a proposition, the world, and a web as the three relata of a three place relation of relative truth then the relata had better be distinct. Explicating Meiland's three place relation with positively operative categorial webs as relativisers fails to meet this requirement. Say the proposition were to be P4 and CW1 were to be the relativiser. CW1's commitment to the referential success of "witch" and P4's statement that they exist are not sufficiently distinct to meet Siegel's criterion. The difficulty does not arise if we revert to mere categorial repetoires as relativisers. So, if framework correspondence truth relativism is to be given a satisfactory explication, then only weak categorial repetoire relativism escapes initial scrutiny and is worthy of further study, some of which occurred in my previous paper in this journal.