Galaxy clusters are beautifully simple, but also fantastically complicated structures. For many years, astronomers have treated these systems as spherical cows, but simulations and observations have repeatedly shown that clusters exhibit triaxial rather than spherical shapes with nice relaxed dynamics (are virialized). Many cluster mass estimators assume spherically symmetric velocity fields (i.e. you measure the same velocities of cluster galaxies regardless of which side you observe from), but if the shape is anisotropic it’s probable the velocities are as well. This makes it crucial to measure the degree of triaxiality of clusters in observations to constrain its impact on mass estimates.The authors sought to show that velocity anisotropy exists by testing for an azimuthal (angle on the sky) dependence of the projected velocity dispersion. To do this, they used a stacked sample of galaxy clusters from the Sloan Digital Sky Survey (SDSS). Stacking is a common technique of taking many clusters with a similar property (in this case galaxy richness or number which is a proxy for mass) and adding them together to make a composite system with many hundreds more galaxies than any one system alone. This gives much better statistics and makes a result more robust.
Because the authors are looking for azimuthal variations, they fit each cluster with an ellipse and stack them with their major axes aligned. They then estimate the projected velocity dispersion for galaxies closer to the stacked minor axis, and a separate velocity dispersion for galaxies closer to the major axis. Because they have a stacked sample with many galaxies, they are also able to measure the velocity dispersion as a function of radius for these two azimuthal directions. The authors find that the galaxies along the stacked system’s major axis exhibit a velocity dispersion which is 6% higher than the average of the system. The presence of such an effect suggests that clusters exhibit a prolate velocity ellipsoid on average, and the velocity ellipsoid is related in orientation to the halo shape. This confirmation raises questions about cluster dynamics such as how the large scale structure of the cosmic web and infall process affects observables like velocity dispersion.