Here we have all the same conditions as before, but we apply a chi square probability cut of .01. This should remove some amount of real tracks (roughly 1/100) but it should significantly increase momentum resolution or other SNR if the bad fits in the first layer are bad enough.