Relative assessments of 10 attributes commonly used to describe the texture of liquids on skin were obtained for 14 liquids of widely varying rheological properties. The results were analyzed to show that these judgments can be predicted accurately from the assessments of just three attributes, smoothness, thinness, and warmth. Smoothness was found to be closely related to the forces of contact lubrication, and could be predicted from measurements of coefficients of friction. Thinness was found to be closely related to viscous forces, consistent with other predictions of assessments of thickness and spreadability. Warmth was not studied further. The results show how careful definition of the stimulus on the skin's surface can sharply simplify the interpretation of the psychophysical experiment.This work is in two parts. The objective of the first part was to devise a nonredundant vocabulary to describe texture perceived with the fingers. The following 10 attributes were selected as the initial vocabulary: thick, thin, spreadable, soft, hard, smooth, creamy, dry, warm, and cool. These attributes are the most important descriptors reported for cosmetics (Goldenberg & de la Rosa, 1971;Prall, 1973; Breuer, Note 1). This vocabulary contains considerable redundancy: For example, if a fluid is assessed as thin, it seems unlikely to be assessed as thick. The list is also uneven because it includes antonyms like warm and cool, but does not include rough, the expected antonym for smooth. Nevertheless, we decided to restrict the vocabulary to the words given because they are those most commonly used, and to delay investigation of the larger group of less cited attributes.Untrained observers were asked to scale these attributes, using the method of magnitude estimation and fluids of widely varying rheological properties. This wide range of properties was chosen in the expectation that this would effect a wide range of assessments. The observer's assessments were then used in a multiple regression analysis, from which a nonredundant subset of attributes could be selected. It was found that a subset of just three attributes could be used to accurately predict other attributes. This subset thus is a reduced vocabulary, suitable for further study.The second part of the paper shows how tactile assessments of fluid samples by human subjects can be quantitatively related to the forces acting on the skin's surface. This step, which is outside the J. E. Moon helped with some of the experiments. This work was partially supported by National Science Foundation Grant GK-32312-A2, and by Petroleum Research Fund Grant PRF-6376-AC-7. E.L.C. is supported by National Institutes of Health Research Career Development Award 1K04-Am-70461. usual range of psychophysical investigation, represents a scheme for generalizing empirical correlations like Stevens' law. The nature of this generalization is best suggested by the example of smoothness. Stevens and Harris (1962) showed that for sandpaper (smoothness) 0: (grit diametery">.( 1) This correlat...