Comparing the frequency of typical events with that of unusual events allows one to test whether the cosmological density distribution function is consistent with the normally made assumption of gaussianity .
To this end , we compare the consistency of the tail-inferred ( from clusters ) and measured values ( from large-scale flows ) of the rms level of mass fluctuations for two distribution functions : a Gaussian , and a texture ( positively-skewed ) PDF .
We find that if we average the recent large-scale flow measurements , observations of the rms and the tail at the 10 h ^ { -1 } \mbox { Mpc } scale disfavor a texture PDF at \sim 1.5 \sigma in all cases .
If we take the most recent measurement of the rms , that from Willick et al .
( 1997b ) , the comparison disfavors textures for low \Omega _ { 0 } = 0.3 , and disfavors Gaussian models if \Omega _ { 0 } = 1 ( again at \sim 1.5 \sigma ) .
Predictions for evolution of high temperature clusters can also be made for the models considered , and , as is known ( e.g. , Henry 1997 ) , strongly disfavor \Omega _ { 0 } = 1 Gaussian models , while we find \Omega _ { 0 } = 1 marginally disfavored in texture models .
Taking the suite of tests as a whole , and using all of the quoted data , it appears that textures are strongly disfavored and only the low \Omega _ { 0 } Gaussian models are consistent with all the data considered .
But given evidence for the internal inconsistency of the observational data , had we only used the recent Willick et al .
results , the strength of our conclusion would have been reduced to the \sim 1 \sigma level .