The method of analysis of double spectrometer rocking curves developed by L. P. Smith is reconsidered. Although from Smith's very general viewpoint, six experimental curves are needed for a complete analysis, it is shown that for rocking curves from calcite, taken with the usual type of double spectrometer, it should be possible to deduce the shape of the single crystal diffraction pattern from two rocking curves, the (1,+1) and (2,+2). A method of modifying the equations of the instrument to allow for a simple type of mosaic structure is indicated.The equations have been applied to rocking curves of Mo $K{\ensuremath{\alpha}}_{1}$ from calcite, supplied by L. G. Parratt. The method requires resolution of the observed curves into Fourier components, and a numerical method of doing this is described. The reliability of the components obtained can be tested by predicting the (1,-1) curve with them and comparing with experiment. In this way it is found that the curves are consistent as regards Fourier components of long period and large amplitude, but inconsistent in the short period, small amplitude components. A single crystal pattern is deduced, based mainly on the observed (1,-1) curve, with asymmetry as indicated by the (1,+1) and (2,+2) curves. It indicates that the crystals used do not have the flat-topped Darwin-Ewald-Prins diffraction pattern. Possible causes of the short period discrepancy have been investigated, but an adequate explanation has not been found.