Sidney van den Bergh's 1999 Review (The Early History of Dark Matter) gives the landmark Zwicky paper as "Zwicky, F. 1933, Helvetica Phys. Acta, 6, 110", four years before the 1937 ApJ paper most folk remember (On the Masses of Nebulae and of Clusters of Nebulae, Astrophysical Journal 86, 217 - PDF).

Nevermind, in the 1937 paper Zwicky not only introduces an application of the virial theorem as a means of estimating the mass in a (rich) cluster, but also gravitational lensing (he also discusses rotation curves in spirals!).

This 'virial theorem' method continues to be used, and provides an estimate of the mass in individual (rich) clusters. Naturally, it comes with caveats (for example, the cluster must be 'dynamically relaxed', or close to it), so corroboration of these cluster mass estimates, obtained from the application of methods using quite different physics, would be nice.

So, what is the 'virial theorem' method?

Observationally, one obtains the redshifts of as many galaxies - in the cluster of one's desire - as possible. The dispersion of these redshifts (crudely, the value of the standard deviation of the distribution of redshifts) is related to the (total) mass of the cluster, via the virial theorem.

Of course, the estimate depends upon Newtonian gravity - if galaxies in rich clusters don't respect Newton, then the virial theorem won't work (the difference between GR and Newton, in this domain, wrt the virial theorem, is trivial).

The kicker is, as Zwicky found for the Coma cluster, that there's (apparently) far more mass in the cluster than you'd expect, simply by 'counting (optical) photons' - i.e. from the kinds of stars that we know and love, from our observations of our own Milky Way galaxy.

So, where is all the 'missing mass'? Is it (all) 'non-baryonic, dark matter'?

Stay tuned!