This article describes an experiment undertaken to analyze the learning habits of Latin students in Germany, discusses the implications therein for programed instruction, and relates the advantages of such instruction. In the experiment, students were found to confuse similar words which had been presented together in instruction (e.g. "memor"--mindful of, and "memini"--to remember), often disregarding grammatical considerations in the process. Consequently, the author proposes that programed learning material not present similar concepts, nor a rule and its exceptions, at the same time, lest they be confounded. Furthermore, he finds presenting examples of a rule before stating the rule preferable to the opposite sequence. Among the advantages of programed instruction that are discussed are its use as homework (thereby saving class periods for other work,) its saving of time, and its self-motivating nature. The author warns that programed instruction must use material based on modern principles of the psychology of learning and not rely on outdated concepts. This done, however, he claims, presenting evidence from a three-year course, that students will make rapid and thorough progress through the major Latin works. (RS)