Dragonfly 44 became famous as a galaxy of almost pure dark matter: an ultra-diffuse Coma-cluster galaxy with the stellar mass of a dwarf but a velocity dispersion of about 33 km/s and an early count of roughly 80 to 100 globular clusters, both pointing, under standard scaling relations, to a Milky-Way-class halo and a dark matter fraction near 99.99 percent (van Dokkum et al. 2016). The globular cluster count carried much of the weight, since GC numbers correlate tightly with total halo mass. The claim is now disputed at its foundation: Saifollahi et al. (2021), reanalyzing the same Hubble imaging with stricter membership criteria, count only about 20 globular clusters, a number normal for dwarf galaxies, and argue DF44 is not extraordinary; subsequent work continues to contest the counts, the dispersion-based mass, and the inferred halo.
The episode cuts in both directions for ΛCDM. If the original numbers stand, the model must explain how a dwarf-mass stellar system occupies a Milky-Way halo, a catastrophic failure of the stellar-to-halo-mass relation and of every feedback prescription governing it, and DF44-class objects would constitute a distinct formation channel the model lacks. If the recount stands, the model instead inherits a methodological warning with cosmological reach: GC-count halo masses, used widely across the UDG population, can be wrong by factors of several, unsettling the broader census of dark-matter-rich and dark-matter-poor UDGs on which several small-scale tensions rest.
The standing is genuinely unresolved: kinematic tracers, GC counts, and stellar population analyses of DF44 disagree among themselves, deeper imaging campaigns continue, and the UDG halo-mass distribution, from DF2-class deficient systems to DF44-class claims of excess, spans a range no single ΛCDM formation channel covers.