Emerson and Halpern (1986, Journal of the Association for Computing Machinery33, 151–178) prove that the Computation Tree Logic (CTL) cannot express the existence of a path on which a proposition holds infinitely often (fairness for short).

The scope is widened from CTL to a general branching-time logic. A path quantifier is followed by a language with temporal descriptions. In this extended setting, the said inexpressiveness is strengthened in two aspects. First, universal path quantifiers are unrestricted. In this way, they are relieved of any temporal quantifiers such as of those in $\mathtt{AU}$ and $\mathtt{AR}$ from CTL. Second, existential path quantifiers are allowed with any countable language. Instances are the temporal quantifiers in $\mathtt{EU}$ and $\mathtt{ER}$ from CTL. By contrast, the fairness statement is an existential path quantifier with an uncountable language. Both aspects indicate that this inexpressiveness is optimal with respect to the polarity of path quantifiers and to the cardinality of their languages.

The prevalent interpretation of Gödel’s Second Theorem states that a sufficiently adequate and consistent theory does not prove its consistency. It is however not entirely clear how to justify this informal reading, as the formulation of the underlying mathematical theorem depends on several arbitrary formalisation choices. In this paper I examine the theorem’s dependency regarding Gödel numberings. I introduce deviant numberings, yielding provability predicates satisfying Löb’s conditions, which result in provable consistency sentences. According to the main result of this paper however, these “counterexamples” do not refute the theorem’s prevalent interpretation, since once a natural class of admissible numberings is singled out, invariance is maintained.