If anti-realism is defined as the principle that all truths are knowable, then anti-realists have a reason to revise logic. For an argument first published by Fitch seems to reduce anti-realism to absurdity within classical but not constructivist logic.1 One might try to sever this link between anti-realism and revisionism in logic by giving either a modified version of anti-realism not vulnerable to Fitch's argument within classical logic or a modified version of Fitch's argument to which anti-realism is vulnerable within constructivist logic. A recent article by Dorothy Edgington in effect takes the first path; I have objected elsewhere.2 In this paper I discuss an ingenious attempt by Crispin Wright to take the second path.3 The anti-realist principle can be formalized as: