Eight groups of ten junior college students were given an experimental training session with manually simulated teaching machines, each group being taught with a different mode of teaching machine operation. The variables were student response mode, size of steps between successive items, and branching procedure. A written test was given to all students after the training session, and again three weeks later. The same test was given to a control group which had no training with the teaching machines. The use of the simulated teaching machines led to a significant degree of learning by the subjects. Multiple choice and branching procedures required less time but did not affect test scores significantly. Small item steps required significantly more training time, but also yielded significantly higher test scores. Significant interaction between the mode of response and branching variables resulted from a high mean criterion score obtained by the constructed response--non-branching group. Statistical tables supplement the report. [Not available in hard copy due to marginal legibility of original document.] (JY)