The first paper I looked at ([Paraficz et al. 2012][1]) explains that the hot gas mass is determined from X-ray observations. The X-ray flux from an optically thin gas depends on the square of the gas density multiplied by its volume - if you can then estimate the volume you get the density and also the gas mass. Some details for the analysis of the Chandra X-ray observations of the Bullet cluster are found in [Close et al. (2006)][2], including how they model the geometry of the various components. They conclude that their gas mass estimate is good to 10 per cent. 

The masses of individual galaxies are estimated by modelling their luminosities through [Faber-Jackson][3] or (for spirals) [Tully-Fisher][4] scaling relations (see also [here][5]). These give the total galaxy mass, which would include dark matter. To estimate just the baryonic mass one just uses the mass to luminosity ratio for stellar material under the assumption that most of the baryonic matter is stars (a small correction could be made for gas, dust etc).

It is on this basis that it is claimed that the X-ray emitting gas contains *a similar* amount of mass to that associated with individual galaxies. *If* those galaxies have non-baryonic dark matter halos that dominate their total mass (which seems likely unless they have extraordinarily low luminosity to mass ratios) then I think this leads to the claim that about 90 per cent of the baryonic mass is in the X-ray emitting gas. If one is sceptical of dark matter and don't trust the FJ and TF scaling relations, then I guess you just take the luminosity of the individual galaxies, convert that to a stellar mass, and you would arrive at more-or-less the same number.

For the Bullet Cluster, gravitational lensing then reveals that the galaxies plus hot gas only represents 20 per cent of the total cluster mass (9 per cent in hot gas, 11 per cent in galaxies) and thus that 89 per cent of the total mass is not in galaxies and that only a small fraction of this is in the form of a hot baryonic gas.


  [1]: http://arxiv.org/abs/1209.0384
  [2]: http://arxiv.org/abs/astro-ph/0611496
  [3]: https://en.wikipedia.org/wiki/Faber%E2%80%93Jackson_relation
  [4]: https://en.wikipedia.org/wiki/Tully%E2%80%93Fisher_relation
  [5]: http://spiff.rit.edu/classes/phys443/lectures/tffj/tffj.html