I should like to say that where surveyability is not present, i.e., where there is room for a doubt whether what we have really is the result of this substitution, the proof is destroyed. And not in some silly and unimportant way that has nothing to do with the nature of proof