I am not terribly happy with A's formulation if you have translated him correctly, but I certainly prefer it to Russell & Whitehead's formulation something like: "Three is the set of all sets that have three members." It seems so material and applied.  I am not sure why three should be a set at all.