Cosmic microwave background experiments ideally suited for testing the coherence of the underlying cosmological theory

We present a formalism for computing errors in estimates of the cosmic microwave background power spectrum valid for a broad class of single-dish and interferometric experiments. We then investigate which experiments are better suited to test the robust prediction that cosmic strings do not produce secondary Doppler peaks. We find that single-dish experiments are required to have a resolution better than 0.2°. Interferometers appear to be less restricted. A deep-patch scanning technique is always favoured.