Abstract
We apply a uniform procedure to select very soft sources from point sources observed by Chandra in four galaxies. This sample includes one elliptical galaxy (NGC 4967), two face-on spiral galaxies (M101 and M83), and an interacting galaxy (M51). We report on some intriguing results, including the following: 1. We have found very soft X-ray sources (VSSs) in every galaxy. Some of these fit the criteria for canonical supersoft sources (SSSs), while others are somewhat harder. These latter have characteristic values of kT ≲ 300 eV; we refer to them as quasi-soft sources (QSSs). We found a combined total of 149 VSSs in the four galaxies we considered; 77 were SSSs and 72 were QSSs. 2. The data are consistent with the existence of a large VSS population, most of whose members we cannot observe because of the effects of distance and obscuration. The total VSS population of sources with L > 10 <sup>37</sup> ergs s <sup>-1</sup> in each galaxy could be on the order of 1000. 3. Whereas in M31 only ~10% of all X-ray sources detected by Chandra are VSSs, more than 35% of all detectable X-ray sources in the face-on galaxy M101 fit the phenomenological definition of VSSs. This difference may be due to differences in N <sub>H</sub> between typical lines of sight to sources in each galaxy. 4. SSSs can be super-Eddington for Chandrasekhar-mass objects. 5. We find evidence for SSSs and QSSs with luminosities of 10 <sup>36</sup> ergs s <sup>-1</sup> <L < 10 <sup>37</sup> ergs s <sup>-1</sup> . These sources have luminosities lower than those of the ∼30 soft sources used to establish the class of SSSs. 6. In the spiral galaxies Ml01, M83, and M51, a large fraction of the SSSs and QSSs appear to be associated with the spiral arms. This may indicate that some SSSs are young systems, possibly younger than 10 <sup>8</sup> yr. 7. In addition to finding hot white dwarfs and soft X-ray binaries, our method should also be efficient at selecting supernova remnants (SNRs). A small fraction of the VSSs in the spiral arms of M101 appear to be associated with SNRs.