THIS STUDY SOUGHT TO DETERMINE IF HIGHLY DIRECTED (PROGRAMED) LEARNING IS SUPERIOR TO NONDIRECTED (DISCOVERY-TYPE) LEARNING. SPECIAL UNITS IN MATHEMATICS WHICH EMPLOYED A DISCOVERY METHOD WERE DEVELOPED AT THE FIFTH-GRADE LEVEL. NEXT, A PROGRAMED MODIFICATION OF THE BASIC UNIT OF INSTRUCTION WAS DEVELOPED. FINALLY, ANOTHER PROGRAM OF INSTRUCTION WAS DEVELOPED TO TEACH THE SAME INSTRUCTIONAL OBJECTIVES AS THE BASIC INSTRUCTIONAL UNIT, BUT WITHOUT EMPLOYING THE DISCOVERY METHOD. THE PROGRAMED MATERIALS WERE DEVELOPED SO THAT A HIGHLY DETAILED DOCUMENTATION OF THE PROGRAM DESIGN COULD BE PROVIDED. THE EVIDENCE OBTAINED INDICATED THAT THE DESIRABLE LEARNING OUTCOMES WHICH SOMETIMES RESULT FROM LEARNING-BY-DISCOVERY CANNOT BE ADEQUATELY EXPLAINED IN TERMS OF OPERANT CONDITIONING. IT IS POSSIBLE, MOREOVER, THAT SEARCHING BEHAVIOR MAY NOT BE LEARNED FROM A DISCOVERY-TYPE INSTRUCTIONAL EXPERIENCE AT ALL. AS A RESULT, THE PROJECT IMPLICATIONS DEAL MORE WITH DESIGNING INSTRUCTIONAL UNITS THAN WITH THE RESEARCH PROBLEM, LEARNING-BY-DISCOVERY. (GD)