Full Idea
There is a suspicion that Frege's definition of 5 (as the set of all sets with 5 members) may be infected with circularity, …and how can we be sure on a priori grounds that 4 and 5 are not both empty sets, and hence identical?
Gist of Idea
If '5' is the set of all sets with five members, that may be circular, and you can know a priori if the set has content
Source
comment on Gottlob Frege (works [1890]) by José A. Benardete - Metaphysics: the logical approach Ch.14
Book Reference
Benardete,José A.: 'Metaphysics: The Logical Approach' [OUP 1989], p.104