Abstract

Thirteen measurements of the Compton profile of water carried out in five different laboratories are compared. The possible sources of error inherent in Compton measurements are analysed. Inconsistent normalization of the various profiles is found to be an important source of error. After renormalization of the profiles the results are found to be consistent to within plus or minus 2% of the peak height of the profiles. It is shown that all probable sources of error can be analysed in terms of five general types of error and further that the effects of all of these types of error on the shape of the Compton profile are almost indistinguishable from one another. The determination of the position of the Compton peak is found to be another important potential source of error in experimental measurements. Certain assumptions relating to the conversion to a momentum scale were found to be mistaken and are corrected. A comparison of currently accepted relativistic and non-relativistic expressions for the cross section revealed that they are not in fact significantly different. Multiple scattering is found to be the only major source of error which is not yet well understood. Each experimental measurement is assessed in detail with particular regard to the differences between the techniques used by the contributing groups. Proposals are made relating to the accumulation, processing and presentation of Compton data. It is concluded that a reproducibility of plus or minus 1% of the peak height of the profile is quite feasible and that in a differential experiment plus or minus 0.5% or even better is possible.