The example that Matt Emerton cited here:

prompted me to become interested in how one proves that odd two dimensional dihedral Galois representation are modular. This is the first case of the strong Artin conjecture for two dimensional representations and I feel like understanding it would be helpful in getting some sense for why Galois representations are modular. Emerton mentioned that the theorem was proved by Hecke; according to Gelbart's review of the Serre/Deligne paper on Galois representations attached to weight 1 modular forms; the dihedral case follows from Hecke's work on theta series attached to binary quadratic forms.

Chandran Singh Dalawat give some more detail on the example that Emerton gives on pp. 5-6 of his article titled Splitting Primes, citing an article of Serre for still more detail. I have some glimmerings of how this goes in the case under discussion; in that case one needs to show that the Artin L-function is 1/2 of the difference of two theta series; presumably one uses class field theory for the splitting field viewed as a cubic extension of the quadratic subfield. The two quadratic forms used to define the relevant theta series correspond to the nonprincipal ideal classes of the quadratic subfield. But I don't see exactly how it should go.

I've seen references to

J.-P. Serre, Modular forms of weight 1 and Galois representations. In: Algebraic
Number Fields (1977), pp. 193–268 = Œuvres/Col lected Papers III, Springer-
Verlag, Berlin, 1986, pp. 292–367.

but given that the result goes back to Hecke it seems like there should be expositions along classical lines from an earlier time (1930's-1960's). and I haven't been able to find them. Does anyone know such a reference?