In ‘Wittgenstein on Language and Rules’, Professor N. Malcolm took us to task for misinterpreting Wittgenstein's arguments on the relationship between the concept of following a rule and the concept of community agreement on what counts as following a given rule. Not that we denied that there are any grammatical connections between these concepts. On the contrary, we emphasized that a rule and an act in accord with it make contact in language. Moreover we argued that agreement in judgments and in definitions is indeed necessary for a shared language. But we denied that the concept of a language is so tightly interwoven with the concept of a community of speakers (and hence with actual agreement) as to preclude its applicabilty to someone whose use of signs is not shared by others. Malcolm holds that ‘This is an unwitting reduction of Wittgenstein's originality. That human agreement is necessary for “shared” language is not so striking a thought as that it is essential for language simpliciter.’ Though less striking, we believe that it has the merit of being a true thought. We shall once more try to show both that it is correct, and that it is a correct account of Wittgenstein's arguments.