[contravariant]

I think they skip a few steps, but in this derivation im phi is exactly Z/2^nZ so every subgroup should have cardinality 2^k for some k.