PS. Hausdorff's (1916) proof that every uncountable Borel set has a perfect subset is perhaps the longest known for this particular result, but isn't it the earliest nontrivial application of the WF induction (or rather recursion) ? -- V.Kanovei