Five thin sections of granite from Westerly, Rhode Island, were analyzed by each os five relatively inexperienced operators using the point-counter method of determining the mode. Each operator ran his slides in a different sequence. In a preliminary run the observed analytical error for the constituents exceeded the expected error; in a second run by the same operators, now more experienced, the observed analytical error was approximately the same as the expected error. Evaluation of variances showed order of analysis to be insignificant throughout, operator and slide differences to be slight. The means for all constituents in both runs were remarkably close to the means previously obtained by Chayes in an independent run on the same sections. The test definitely establishes the level of precision for the point-counter method of analysis on a broader base than was possible from Chayes' unchecked analyses. The convenience and rapidity of the method are already well known; its precision has now been thoroughly tested.