A STUDY OF FIVE DIFFERENT ESSAY EVALUATION PROCEDURES WAS CONDUCTED TO DETERMINE THE INTRARELIABILITY OF EACH AND TO USE THESE RELIABILITY SCORES AS A BASIS FOR COMPARING THE FIVE PROCEDURES. TEN THEMES WERE ASSIGNED TO FIVE RATING GROUPS, EACH USING A DIFFERENT EVALUATION PROCEDURE--(1) THE CALIFORNIA ESSAY SCALE, (2) THE CLEVELAND COMPOSITION RATING SCALE, (3) THE DIEDERICH RATING SCALE, (4) THE FOLLMAN ENGLISH MECHANICS GUIDE, AND (5) THE "EVERYMAN'S SCALE." THE RATERS WHO GRADED THE THEMES WERE CONSIDERED HOMOGENEOUS--ALL WERE UPPER DIVISION COLLEGE ENGLISH MAJORS WITH THE SAME COURSE IN ENGLISH METHODS. RESULTS INDICATE THAT (1) THE DIFFERENCES AMONG RATING GROUPS DID NOT CHANGE WITH THE SUBJECT MATTER OF THE ESSAYS, (2) THE ESSAYS RECEIVED SUBSTANTIALLY THE SAME SCORES FROM ALL FIVE RATING GROUPS, (3) THERE WERE HIGH INTERCORRELATIONS AMONG SYSTEMS, EXCEPT FOR THE DIEDERICH SCALE, AND (4) THE HIGHEST RELIABILITY SCORES WERE EVIDENCED BY THE ENGLISH MECHANICS SCALE, THE LOWEST BY THE CLEVELAND COMPOSITION RATING SCALE. THE HIGH RELIABILITY OBTAINED ACROSS DIFFERENT EVALUATION PROCEDURES MAY BE DUE TO THE HOMOGENEOUS NATURE OF THE RATERS RATHER THAN TO THE RATING SYSTEM. HYPOTHESES SUGGEST THAT A RATING SYSTEM WOULD HAVE ITS GREATEST EFFECT IN RAISING THE RELIABILITY OF GRADING WHEN USED BY A GROUP WITH HETEROGENEOUS TRAINING AND BACKGROUNDS, AND THAT RATERS OF HOMOGENEOUS TRAINING WILL BE EQUALLY CONSISTENT WITH OR WITHOUT A RATING SYSTEM. THIS ARTICLE APPEARED IN "RESEARCH IN THE TEACHING OF ENGLISH," VOLUME 1, NUMBER 2, FALL 1967, PAGES 190-200. (BN)