We investigate the effects of finite sky coverage on the spectral resolution Δℓ in the estimation of the CMBR angular power spectrum Cℓ. Using an uncorrelated mesh approximation in Fourier space, we obtain quasi-independent estimates of the power spectrum, and the cosmic/sample variance of these estimates is also calculated. The effect of instrumental noise is also considered for prototype interferometer and single-dish experiments. By proposing a statistic for the detection of secondary (Doppler) peaks in the CMBR power spectrum, we then compute the significance level at which such peaks may be detected for a large range of model CMBR experiments. In particular, we investigate experimental design features required to distinguish between competing cosmological theories, such as cosmic strings and inflation, by establishing whether or not secondary peaks are present in the CMBR power spectrum.