Interactions may appear in an analysis of variance because of a poor choice of model. The effects of the independent variables in corrosion data seem to be multiplicative, which calls for a logarithmic transformation to convert this to an additive system. The simplification produced by such a transformation is strikingly illustrated in the analysis of a corrosion experiment. The experiment described was made to determine whether the occasional use of low purity water at low temperature would cause excessive corrosion in the cooling system of a power producing nuclear reactor. Three types of aluminum alloy were used in the corrosion tests. 2.1.1

This content is only available as a PDF.
Copyright 1959 by the National Association of Corrosion Engineers.
1959
You do not currently have access to this content.