A software fault categorization scheme was produced in conjunction with the development of a medium-scale commercial product. The errors that were detected at the various phases of the life cycle were classified and analyzed in terms of the scheme. The results of this study include little that was not previously known about the software development life cycle in general, or testing in particular. The authors have not convincingly shown the superiority of their scheme over existing error categorization schemes. In addition, their poor choice of nomenclature (such as presence to denote “how a fault was corrected," or use to indicate “the operation being performed in a data handling fault") makes the scheme unattractive. I do not believe that this new scheme will become widely adopted.