Statistics—A Useful Tool for the Examination of Corrosion Data by Charles F. Lewis. CORROSION, Vol. 9, No. 1, 38-43 (1953) Jan.
Discussion by A. H. DuRose, Harshaw Chemical Comany, Cleveland, O.:
Mr. Lewis is to be complimented on a clear presentation of a subject which is usually difficult for a novice like myself to grasp. However for the specific data dealt with it seems that a few refinements would allow for better interpretation of the data and still maintain the clarity of presentation.
First, examination of corrosion data will usually show that the greater the degree of corrosion the more erratic it will become; in other words, a log-normal distribution. By plotting the unlogged data on probability paper it appeared that the high value 29 (in the iron-containing data) was a maverick and it was rejected. However by plotting the data on logarithmic probability paper a fairly straight line is...
