PROBLEMAs part of a comprehensive study of the clinical usefulness of abbreviated intelligence tests, the authors and their colleagues have investigated the use of "scatter" as an indicator of intellectual impairment (2,3,4). The CVS Individual Intelligence Scale (a battery consisting of Comprehension and Similarities from the Wechsler-Bellevue Scale and Thorndike's fifteen-word vocabulary scale taken from the Stanford Binet list (5,6)) was selected for this extensive investigation both because of its high correlation with external criteria of intelligence and for its diagnostic potentiality. The subtests in this abbreviated battery were deliberately selected for their diagnostic possibilities, since vocabulary is known to be relatively insensitive to psychopathosis, and comprehension and similarities to be relatively sensitive to it.Diagnostic scatter, defined in this battery as the presence of a vocabulary score higher than the scores for comprehension and similarities, has been investigated by us to date in four clinical groups : schizophrenics, paretics, arteriosclerotics, and mental defectives; as well as in a number of normal control g r o~p s (~~3~~) .The simple criterion of both comprehension and similarities srores being below the vocabulary score was found to distinguish young normals adequately from schizophrenics of the same a g~(~,~) .In differentiating organic pathosis, however, one runs into the necessity of introducing some correction for the normal deterioration attendant upon mere aging. By complicating slightly our criterion for scatter, however, it appears possible t o solve this problem and to differentiate normal and pathological groups in both the middle and advanced age ranges(4).By a rational, experimental approach we have been able t o construct a brief intelligence measure which will also act as a rough screen for possible psychopathosis. Our success, we think, is attributable merely t o the fact that we have set diagnostic scatter as a goal in constructing our test battery and not attemptedposthoc t o wring diagnostic measures from test batteries not designed for the purpose. However, our success is purely relative and our screening is rough. While from 60 to 80 per cent of our clinical groups consistently show scatter, from 10 to 20 per cent of our normal subjects also show it. This false positive rate greatly reduces the efficiency of our measure as a screening device. It is therefore necessary to further investigate the nature of scatter in the hope of refining our predictive indices.One cue as to what might be causing false positives with our scatter measure came from our mentally defective groups, which consistently showed less scatter than our normal control groups. In view of the controversial literature on scatter in feeble-mindedness, we must point out that our defective groups were all simple, familial defectives uncomplicated by added emotional difficulties or any organic involvement. This finding suggested that scatter might be a function of intelligence level. Since vocabu...