2001ApJ...549..871S

We test the methods developed in previous papers for inferring the intrinsic shapes of elliptical galaxies, using simulated objects from N-body experiments. The shapes of individual objects are correctly reproduced to within the statistical errors; close inspection of the results indicates a small systematic bias in the sense of underestimating both the triaxiality and short-to-long axis ratio. We also test the estimation of parent shape distributions, using samples of independent, randomly oriented objects. The best results are statistically accurate, but on average the parent distributions are again slightly biased in the same sense. Using the posterior probability densities for the individual objects, we estimate the magnitude of the bias to be ≲0.1 in both shape parameters. The results support the continued use of these methods on real systems.