In the previous two sections, I have presented evidence that supports the presence of a large number of faint galaxies in clusters, but have not shown what type these galaxies are. The scale-lengths are consistent with either a dSph or dIrr interpretation (recall from Section 2 that these galaxies occupy the same part of the Binggeli diagram). They are not consistent with the faintest galaxies being low luminosity compact ellipticals like M32. These faint galaxies have scale-lengths comparable to or smaller than the seeing, so that their detailed morphologies cannot be used to tell whether they are dSphs or dIrrs.
The best discriminant between these galaxy types in the absence of spectroscopic information is the color of the galaxies. In all the distant clusters shown on Figure 1, I found that the colors of the faintest galaxies were consistent with them being dSphs and not dIrrs (the K corrections are given in ref. 41). Other authors (see the list of references in Section 5) conclude similarly. In Virgo, the galaxies are sufficiently nearby that morphological information can be used - both Sandage et al.  and Phillipps et al.  find that the faintest galaxies that they detect are dSphs as well.