Although more college students are enrolled in freshman composition classes than in any other, there is little clarity of purpose in these writing courses. In such a field, where outcomes must be expressed in terms of student writing skills acquired (or not acquired), few well-defined experiments have been reported. This investigator, in an attempt to discriminate among levels of performance in student themes, faced the problem of establishing a reliable evaluation technique. The techniques she finnaly adopted are described in this report. Having set up a grading system, she then sought to learn if one way of teaching grammar was more effective than another. Of two sections in the grammar course, one became the control group, taught in the traditional way, while the other, the experimental group, received three hours of language laboratory instruction and one hour in a recitation-discussion class. Tables show the correlation of rank and raters in grading the themes for each group, and pre- and post-test scores for each class. It appeared that gains were greater in style than in mechanics. The comparisons showed that theme improvement was not appreciable related to the teaching method or to the quarter of instruction. The students did improve their techniques after instruction in grammar and style. The method of grouping showed a significant upward shift in style, but an insignificant one in mechanics. (HH)