Determining the reality of X-ray filaments

A number of authors have reported filaments connecting bright structures in high-resolution X-ray images, and in some cases these have been taken as evidence for a physical connection between the structures, which might be thought to provide support for a model with non-cosmological redshifts. In this paper I point out two problems which are inherent in the interpretation of smoothed photon-limited data of this kind, and develop some simple techniques for the assessment of the reality of X-ray filaments, which can be applied to either simply smoothed or adaptively smoothed data. To illustrate the usefulness of these techniques, I apply them to archival ROSAT observations of galaxies and quasars previously analysed by others. I show that several reported filamentary structures connecting X-ray sources are not in fact significantly detected.