106 JOURNAL OF THE SOCIETY OF COSMETIC CHEMISTS forward. Because the test material is always compared to a target reference standard, a match to the standard sets the lower range of the specification limit. The upper limit of the range can be determined by product development, marketing, plant capability, and importantly, by knowledge of consumer tolerance levels. The ideal process for establishing product sensory specifications is based on consumer data. Consumer testing should be used to determine 1) which product attributes influ- ence acceptance and 2) what the limits of consumer tolerance are around these key attributes. Such testing requires a range of samples that demonstrate the potential variability that might be experienced. These samples should be identified through the use of in-house sensory panels or descriptive analysis capabilities, and then exposed to consumers. Action standards or minimum performance criteria for sample overall rating, overall preference, or other overall acceptability measures need to be established, as well as values for any preselected "consumer sensitive" product attributes. The actual numerical value for the specifications depends on the specific rating scale employed and the product and attributes being controlled. The consumer acceptance data is key to establishing this value realistically. For example, if a tackiness score for a moisturizer is 2.0 and correlates with a consumer-use test overall rating score of 80% (% of consumers responding Excellent/Very Good), and a tackiness score of 3.0 yields an overall rating score of only 50%, the specification for tackiness should be set at an upper limit of 2.0 to insure a high level of consumer acceptability. In many companies, manpower resources and budgets preclude basing specification decisions on consumer data. In such cases, product management teams representing marketing, R&D, and manufacturing should evaluate the group of samples (such as would have been used for consumer testing) representing the range of possible product variability. They should identify the key probable consumer-sensitive sensory parame- ters and determine the variability that occurs around the selected attributes. They must then agree upon a specification beyond which unintended and unacceptable deviations from the original product concept or description will occur. It should become obvious, however, that relying on this method may eventually result in quality problems. It assumes that the management team knows the sensory properties that are important to consumers and the level of product reproducibility that consumers demand. Manage- ment-based specifications may be confirmed or modified based on ongoing consumer experience. Information obtained through consumer hotlines or letters is useful in monitoring whether specifications have been established appropriately. These sources of information are very useful and should not be overlooked as methods of monitoring quality in the field. CONCLUSION Reliance on a single expert to qualify production materials for release assumes that the expert's opinion reflects the opinion of all consumers and that, in fact, the expert's opinion can be imposed upon consumers. Evaluation of the alternate methods discussed and selection based on appropriateness in specific manufacturing circumstances will begin the evolutionary path away from the industry-wide practice of relying on an "expert" to predict consumer behavior.

SENSORY QUALITY 107 REFERENCES (1) H. Stone and J. L. Sidel, Sensory Evaluation Practices (Academic Press, San Diego, 1993), pp. 295- 304. (2) H. B. Heath, Source Book of Flavors (AVI Publishing Co., 1981), pp. 546-548. (3) J. S. Jellinek, The Use of Fragrance in Consumer Products (John Wiley & Sons, New York, 1974), pp. 146-154. (4) A.M. Munoz, G. V. Civille, and B. T. Carr, Sensory Evaluation in Quality Control (Van Nostrand Reinhold, New York, 1992), pp. 5-51. (5) C. Krueger, Color control in cosmetics, Cosmet. Toiletr., 96, 35-36 (1981). (6) H. R. Moskowitz, Product Testing and Sensory Evaluation of Foods (Food and Nutrition Press, 1983), pp. 567-584.