We investigate whether classical combinatorial theorems are provable in ZF. Some statements are not provable in ZF, but they are equivalent within ZF. For example, the following statements (i)–(iii) are equivalent: (i) \(cf({\omega }_1)={\omega }_1\) ,
(ii) \({\omega }_1\rightarrow ({\omega }_1,{\omega }+1)^2\) ,
(iii) any family \(\mathcal {A}\subset [{On}]^{<{\omega }}\) of size \({\omega }_1\) contains a \(\Delta \) -system of size \({\omega }_1\) .
Some classical results cannot be proven in ZF alone; however, we can establish weaker versions of these statements within the framework of ZF, such as (1) \({{\omega }_2}\rightarrow ({\omega }_1,{\omega }+1)\) ,
(2) any family \(\mathcal {A}\subset [{On}]^{<{\omega }}\) of size \({\omega }_2\) contains a \(\Delta \) -system of size \({\omega }_1\) .
Some statements can be proven in ZF using purely combinatorial arguments, such as: (3) given a set mapping \(F:{\omega }_1\rightarrow {[{\omega }_1]}^{<{\omega }}\) , the set \({\omega }_1\) has a partition into \({\omega }\) -many F-free sets.
Other statements can be proven in ZF by employing certain methods of absoluteness, for example: (4) given a set mapping \(F:{\omega }_1\rightarrow {[{\omega }_1]}^{<{\omega }}\) , there is an F-free set of size \({\omega }_1\) ,
(5) for each \(n\in {\omega }\) , every family \(\mathcal {A}\subset {[{\omega }_1]}^{{\omega }}\) with \(|A\cap B|\le n\) for \(\{A,B\}\in {[\mathcal {A}]}^{2}\) has property B.
In contrast to statement (5), we show that the following ZFC theorem of Komjáth is not provable from ZF + \(cf({\omega }_1)={\omega }_1\) : (6 \( ^*\) ) every family \(\mathcal {A}\subset {[{\omega }_1]}^{{\omega }}\) with \(|A\cap B|\le 1\) for \(\{A,B\}\in {[\mathcal {A}]}^{2}\) is essentially disjoint.
A function f is a uniform denumeration on \({\omega }_1\) iff \({\text {dom}}(f)={\omega }_1\) , and for every \(1\le {\alpha }<{\omega }_1\) , \(f({\alpha })\) is a function from \({\omega }\) onto \({\alpha }\) . It is easy to see that the existence of a uniform denumeration of \({\omega }_1\) implies \(cf({\omega }_1)={\omega }_1\) . We prove that the failure of the reverse implication is equiconsistent with the existence of an inaccessible cardinal.