Ramaty High Energy Solar Spectroscopic Imager (RHESSI) observations of 18 double hard X-ray sources seen at energies above 25 keV are analyzed to determine the spatial extent of the most compact structures evident in each case. The following four image reconstruction algorithms were used: Clean, Pixon, and two routines using visibilities-maximum entropy and forward fit (VFF). All have been adapted for this study to optimize their ability to provide reliable estimates of the sizes of the more compact sources. The source fluxes, sizes, and morphologies obtained with each method are cross-correlated and the similarities and disagreements are discussed. The full width at half-maximum (FWHM) of the major axes of the sources with assumed elliptical Gaussian shapes are generally well correlated between the four image reconstruction routines and vary between the RHESSI resolution limit of ∼ 2 up to ∼ 20 with most below 10 . The FWHM of the minor axes are generally at or just above the RHESSI limit and hence should be considered as unresolved in most cases. The orientation angles of the elliptical sources are also well correlated. These results suggest that the elongated sources are generally aligned along a flare ribbon with the minor axis perpendicular to the ribbon. This is verified for the one flare in our list with coincident Transition Region and Coronal Explorer (TRACE) images. There is evidence for significant extra flux in many of the flares in addition to the two identified compact sources, thus rendering the VFF assumption of just two Gaussians inadequate. A more realistic approximation in many cases would be of two line sources with unresolved widths. Recommendations are given for optimizing the RHESSI imaging reconstruction process to ensure that the finest possible details of the source morphology become evident and that reliable estimates can be made of the source dimensions.

We have used Ramaty High Energy Solar Spectroscopic Imager (RHESSI) modulation profiles in the 25 -300 keV range to construct high-fidelity visibilities of 25 flares having at least two components. These hard X-ray visibilities, which are mathematically identical to the visibilities of radio imaging, were input to software developed for mapping solar flares in the microwave domain using the Maximum Entropy Method (MEM). We compared and contrasted the MEM maps with Clean and Pixon maps made with RHESSI software. In particular, we assessed the reliability of the maps and their morphologies for future investigations of the symmetry of bipolar electron beaming in the sample set.