We analyze the metallicity distributions of globular clusters belonging to 28 early-type galaxies in the survey of Kundu & Whitmore ( 2001 ) .
A Monte Carlo algorithm which simulates the chemical evolution of galaxies that grow hierarchically via dissipationless mergers is used to determine the most probable protogalactic mass function for each galaxy .
Contrary to the claims of Kundu & Whitmore , we find that the observed metallicity distributions are in close agreement with the predictions of such hierarchical formation models .
The mass spectrum of protogalactic fragments for the galaxies in our sample has a power-law behavior , n ( { \cal M } ) \propto { \cal M } ^ { \alpha } , with an index of \alpha \simeq - 2 .
This spectrum is indistinguishable from the mass spectrum of dark matter halos predicted by cold dark matter models for structure formation .
We argue that these protogalactic fragments — the likely sites of globular cluster formation in the early universe — are the disrupted remains of the “ missing ” satellite galaxies predicted by cold dark matter models .
Our findings suggest that the solution to the missing satellite problem is through the suppression of gas accretion in low-mass halos after reionization , or via self-interacting dark matter , and argue against models with suppressed small-scale power or warm dark matter .