Comparisons are made between calculations based on the double-helix and side-by-side models of DNA and data taken from the available x-ray diffraction measurements of the B form of DNA. The SBS model invoked here is an improvement on the one which was published earlier. In our estimation the fibre diffraction patterns do not significantly favour either model. However, the DH model is incompatible with the on-axis form of the cylindrically-averaged Patterson function, whereas the SBS model is compatible (this result is relatively insensitive to appreciable modifications to the SBS structure)—the crucial question is whether the quality of the data is sufficient to make this finding significant. It is shown that the maximum expected effect on the fibre pattern, of exchanging A-T base pairs for G-C base pairs, is less than the probable experimental error.