In this study, an evaluation was made of a second grade experimental written composition program based on the social and psychological nature of children, on principles of written composition, and on effective classroom procedures. The distinguishing feature of the experiment, which involved 82 children, was the use of chalkboard compositions to motivate and instruct the students and to help them evaluate their own writing. As a control, another group of 82 second grade children from similar backgrounds were taught in a traditional composition program. Three raters determined the evaluation criteria for the children's writing and systematically rated the compositions from both groups for mechanics and quality of content. By using covariance analysis, it was found that (1) raters can achieve high reliability when they develop their own criteria and practice using it, (2) although group treatment seems to affect the quality of written composition, sex differences do not, and (3) the higher adjusted mean scores of the experimental group indicates that the experimental method of teaching written composition was more successful than the traditional method. (Author/JB)