We have every reason, always keeping in mind that the goal is to test normality, to invite sufficiently learned people here to review critically the way the Jarque-Bera Test was constructed. Note that a former similarly proposed test is d´Agostini´s ?omnibus?: the sum of two parameter´s squared after normalizing, then a N(0,1) value. Though successfully attained this stage, every square follows a 1-degree of freedom Chi-squared, the sum a 2-degrees and finally the quantiles could be used as critical values. But, I wonder, to test what? Not surely the source Distribution is normal . . . It is impossible to go further than the SUM is, or is not, similar to that a normal Distribution provides. This conclusion is in complete desagrement with the current not criticized further refinement, i.e., failed to have normal intermediate stage it come on the paper: Precise finite-sample quantiles of the Jarque-Bera adjusted Lagrange multiplier test. Diethelm Wuertz and Helmuth Katzgraber (Dec. 2009). Surprisingly it seems that the critical values obtained through random simulated samples was considered sufficiently sound to ascribe normality, or not, to whatever sample tested this way. For me is indisputable that an unsolved main problem was shift to other that is rather second rank in importance. Is people comfortable with a procedure that is unable to answer what it was proposed? Or I am wrong?