Using a uniform analysis procedure, we measure spatially resolved weak gravitational lensing and hydrostatic X-ray masses for a sample of 18 clusters of galaxies. We find a radial trend in the X-ray to lensing mass ratio: at r2500 we obtain a ratio M
X/ML=1.03+/-0.07 which decreases to MX/ML=0.78+/-0.09 at r500. This difference is significant at 3 sigma once we account for correlations between the measurements. We show that correcting the lensing mass for excess correlated structure outside the virial radius slightly reduces, but does not eliminate this trend. An X-ray mass underestimate, perhaps due to nonthermal pressure support, can explain the residual trend. The trend is not correlated with the presence or absence of a cool core. We also examine the cluster gas fraction and find no correlation with ML, an important result for techniques that aim to determine cosmological parameters using the gas fraction.
The rich cluster Abell 520 (z=0.201) exhibits truly extreme and puzzling multi-wavelength characteristics. It may best be described as a cosmic train wreck. It is a major merger showing abundant evidence for ram pressure stripping, with a clear offse
t in the gas distribution compared to the galaxies (as in the bullet cluster 1E 0657-558). However, the most striking feature is a massive dark core (721 h_70 M_sun/L_sun) in our weak lensing mass reconstruction. The core coincides with the central X-ray emission peak, but is largely devoid of galaxies. An unusually low mass to light ratio region lies 500 kpc to the east, and coincides with a shock feature visible in radio observations of the cluster. Although a displacement between the X-ray gas and the galaxy/dark matter distributions may be expected in a merger, a mass peak without galaxies cannot be easily explained within the current collisionless dark matter paradigm. Interestingly, the integrated gas mass fraction (~0.15), mass-to-light ratio (220 h_70 M_sun/L_sun), and position on the X-ray luminosity-temperature and mass-temperature relations are unremarkable. Thus gross properties and scaling relations are not always useful indicators of the dynamical state of clusters.
