The bulk of the H2 mass in galaxies is cold and furtive. The main tracer is the CO molecule, but the H2 / CO conversion ratio is very variable, according to the physical conditions in molecular clouds (density and temperature), but mainly with the metallicity.
Probably because of radial abundance gradients, the molecular gas traced by the CO emission is only observed to extend over the optical disk, while the atomic gas component is prolonged much farther out in radius. Nevertheless, the HI and H2 at the same radius are tightly correlated, at large-scale (kpc scale). They trace the same spiral structure for example. In the vertical direction, the two components are also well mixed: both reveal a constant vertical velocity dispersion with radius, of comparable amplitude. This suggests that the HI could serve as the tracer of the H2 component, that would then also extends far out in radius. The constant ratio between dark matter and HI surface densities in galaxies support this hypothesis. If cold molecular gas is a good candidate of baryonic dark matter, observational tests should be pursued: H2 UV absorption lines may be the best probe, and data from the FUSE satellite will certainly make big advances on the subject.