Research on 128 children is presented as evidence of the superiority of a new clustering measure (X) over a more traditional "chance clustering" measure (sigma) as a means of organizing material to be learned to facilitate recall. X is shown to meet three criteria for a measure of a developmental process: X is more highly correlated with recall than sigma: X is more sensitive than sigma to the relationship between age and experience in the development of the clustering strategy: and X is sensitive to the partitioning of the stimulus list while sigma is not. X is discussed in terms of its ability to concurrently take into account both input and output dimensions of the task. By ignoring the qualities of the list to be learned, "chance clustering" measures treat clustering only as a phenomenon of the recalled list rather than as an adaptive strategy evaluated within the context of the dimensions of the task as it was presented. Research subjects were 128 elementary school students who performed recall tasks. (Author/MV)