In attempting to meet school-board mandates for competency-based testing in composition, educators must devise the most acceptable testing programs they can. This paper describes a design (the Paul Diederich system) for testing students' writing skills, which yields statistically reliable data on individual students, and reports on a New Orleans project in which the Diederich system was implemented in one middle school, three junior high schools, and three senior high schools. The paper explains the program procedures, which involve the collection of four writing samples from each student and the rating of the papers by eight teachers during a school year; points to some useful data that emerged from the testing program, and to its effectiveness as an inservice tool; and notes problems in administering it, which led to the conclusion that the total design is not practical for mass testing purposes. THe paper then shows how a modified version of the Diederich system was devised for use in New Orleans schools. In the modified design, the total student population in each school is tested only once. Test scores are used, in conjunction with teacher recommendations, as the basis for placing students with poor writing skills in remedial programs. (GW)