I just received Patrik Eklund's message containing the statement "Yet, Gödel uses "provability" to create new sentences, and simply opens up that bag of sentences, and throws in these new ones." Presumably others have already responded to this in messages that haven't yet reached me, but, just in case they haven't: Much of the work in Gödel's paper on the incompleteness theorems is devoted to making sure that what Eklund wrote here is not the case. Gödel shows how statements about provability can be encoded as arithmetical statements, which are not at all new but were available (in what Eklund calls "that bag") all along. Andreas Blass [For admin and other information see: http://www.mta.ca/~cat-dist/ ]