IN AN INVESTIGATION OF THE POTENTIAL OF TEACHING MACHINES IN THE JOB CORPS MATHEMATICS PROGRAM, EXISTING JOB CORPS ARITHMETIC MATERIAL WAS PREPARED FOR INSTRUCTION BY MACHINE AND PROGRAMED TEXT, AND THEN FIELD TESTED. REVISIONS WERE MADE, AND A PROGRAMED MANUAL FOR INSTRUCTORS WRITTEN, AFTER WHICH A NEW FIELD TEST WAS RUN. IN THE INITIAL FIELD TEST, THE TEACHING MACHINES WERE GREATLY PREFERRED TO THE PROGRAMED TEXT AND SHOWED SLIGHTLY HIGHER ACHIEVEMENT GAINS. IMPROVEMENTS PRECEDING THE SECOND FIELD TEST INCLUDED A REVISION OF THE TEXT, IMPROVEMENT OF THE ADMINISTRATIVE ASPECTS OF THE SYSTEM, ADDITION OF REMEDIAL TUTORING, AND DEVELOPMENT OF A NEW TYPE OF STUDENT RESPONSE FORMAT FOR PROGRAMED TEXTS. THIS NEW FORMAT GAVE IMMEDIATE PEEK-PROOF STUDENT FEEDBACK, IMMEDIATE DIAGNOSTIC FEEDBACK TO THE INSTRUCTOR, AND COULD BE USED DIRECTLY AS INPUT DATA FOR ITEM-ANALYSIS. THE SECOND FIELD TEST SHOWED THE LEARNER-ORIENTED PROGRAMED TEXT TO BE MORE EFFECTIVE THAN THE TEACHING MACHINES AVAILABLE. NO COMPELLING DATA SUGGESTED THE WIDE-SPREAD ADOPTION OF THESE MACHINES IN CONSERVATION CENTERS. APPENDICES INCLUDE DATA FROM PHASES II AND IV AND CORRELATION DATA. (BB)