The concept "trochiac meter" was taught to 180 college Ss by means of eight treatment conditions. The independent variables involved presenting a definition (D) or instances (E) or both combined with attribute definition (A) and/or atrribute prompting (P). Dependent variables were corrected classification and specified classification errors. Hypotheses consisted of prediction of particular errors for each treatment. Six of the eight hypotheses were supported at significant levels. The most effective condition for promoting correct classification consisted of D + E + A + P. The last effective condition consisted of D or E alone. (Author)