I called it ‘conventional’ because this does accord roughly with normal usage and because I could not plausibly anticipate future arguments. In the last chapter I characterized it as legislative, mainly because it needs to be treated as legislative for testing purposes. But I am not claiming that it ought in all contexts to be called ‘legislative agreement’ or ‘legislation’ or anything to do with the enacting of law at all. On the contrary, for most purposes the connotations of the word ‘legislative’ are descriptively less appropriate than those of the word ‘convention’. For ‘legislation’ suggests laws which must be explicitly written down before they are valid, while the laws which convention decrees are usually unwritten laws; they are informal and implicit in the sense that respectable people are supposed to conform to them without being told to do so. And, when the agreements which guarantee logical neces­ sity are effectively doing their job, there is no point in demanding that they be made explicit; they can safely be left as unwritten laws. It is only when disagreement appears on the superficial level of a (nonfactual) discussion that we need to suspect lack of agreement at a deeper level, and hence to bring hidden premises into the open.