In two replications, 97 and 122 Williams College students in elementary psychology classes were given two lists of word pairs with instructions to give a discrete free associate to each pair. List A contained word-pairs designed to increase the frequency of the normative primary responses and List B contained pairs designed to increase the range of responses. The results showed that significantly fewer different responses and significantly higher frequencies for the primary responses were given to List A than to List B. Compared to the single-word norms, responses to List A had higher, and to List B, lower, commonality. Two factors were used in constructing List A: (1) the stimulus pair denoted or connoted a third word; (2) in common speech, a third word frequently follows the stimulus pair. The List B pairs lacked these factors, or had them in a low degree.