Bothersome statistical terms are popping up more frequently these days in strength specifications. Phrases like "standard deviation," "coefficient of variation," "normal distribution" and "statistical evaluation" are begin used as though their meaning were obvious. Why should we be bothered with them? Are they tossed in as one more onerous requirement that will cost the constructor, concrete supplier and engineer more time and money? Or can the concepts these terms represent be used to the advantage of all those involved in concrete construction? Anyone who has ever had occasion to try to evaluate concrete strengths has most likely been vexed and puzzled by the spread or dispersion in test results. Groups of cylinder strengths measurements on concrete materials and proportions that are supposed to be identical never seem to agree as closely as we think they should. Known also as the "root mean square," the standard deviation is derived from test results. Standard specifications define a "test result" as the average strength obtained from two cylinders made from the same sample and broken as the same time. Sometimes an average of three cylinders is used. For any set of test results the standard deviation is found by the following operations: (a) find the sum of the squares of all test results and (b) divide this sum by the number of test involved. And ( c) square the average of the test results and subtract it form the answer obtained in (b). Finally, (d) extract the square root of the value obtained in (c). The application of these provisions in designing a concrete mixture to satisfy specifications is illustrated in Figure 2 found in the article which essentially shows that if controls are tight, that is, if standard deviation is lower, the average strength must exceed the required strength by a smaller amount.