THE INTRINSIC SHAPES OF STELLAR-SYSTEMS

I compute the estimated distribution function f(q) for the apparent ax is ratio q of various types of stellar systems, using a nonparametric kernel method. I then invert f(q) to find the distribution of intrinsi c axis ratios, using two different hypotheses: first, that the stellar systems are all oblate, and second, that they are all prolate. The sh apes of globular clusters in our own Galaxy are consistent, at the 99% confidence level, with both the oblate and prolate hypotheses. The sh apes of dwarf galaxies in the Virgo Cluster are consistent, at the 99% confidence level, with the prolate hypothesis but inconsistent with t he oblate hypothesis. The shapes of star clusters in the Large Magella nic Cloud, of ordinary elliptical galaxies, of brightest cluster ellip ticals, and of galaxy clusters are all inconsistent, at the 99% confid ence level, with both the oblate and prolate hypotheses. The globular clusters in our Galaxy are older than their half-mass relaxation time and are most likely rotationally flattened oblate spheroids. The other stellar systems considered are generally younger than their half-mass relaxation time and thus are triaxial bodies flattened by anisotropy of their velocity dispersion.