Statistical Methods in Chemical Analyses

Abstract

IT is well established that many of the distributions met in biological work approximate to the normal, and the modern developments of statistics are associated with this fact. Most of the techniques are dependent for their validity, to a greater or less extent, on the goodness of the normal equation as a description of the experimental distributions. Since there cannot be any a priori reasons for expecting distributions in chemical work to be normal, and there seems to have been very little investigation of the natures of the distributions actually occurring, application of methods which presume normality may be highly dangerous.