SENSORY AND PHYSICOCHEMICAL PROPERTIES OF EMOLLIENTS 177 open discussion, taking into account the previously mentioned terms and the ones cited in the bibliographical references (8,11). Working conditions were noqnalized in order to train the panel of judges and obtain valid and reproducible results. This normalization included the definition of the selected terms, the establishment of the conditions of sensory evaluation, and the selection of references (anchors) to quantify the intensity of each attribute (Table I). Secondly, judges were trained to evaluate the attributes using structured 10-point scales of intensity, until consistent judgments from the panel were achieved (8,11). Sample evaluation method. A balanced complete-block experimental design was carried out for duplicate evaluation of the samples during six sessions (three samples in each session). Structured ten-point scales anchored with "nil" and "high" were used to describe at- tribute intensity. The test was carried out in a sensory laboratory designed in accordance with ISO 8589 (12). Before the test started a mixture of isopropyl alcohol and water (45/55) was used on the forearms of the judges in order to clean the zone, allowing it to dry. Three 4-cm- diameter circles were outlined on the internal side of the non-dominant forearm. The coordinator placed one drop of the sample in the center of one circle, and the judge rotated the sample with one finger in a circular manner 120 times within the circle, at a rate of two times per second and employing a metronome to control the rhythm. The evaluated attributes are shown in Table I. INSTRUMENTAL ANALYSIS The physicochemical properties selected were those that could be related to the capacity of emollients to spread and to the sensations they cause when applied to the skin. These physicochemical properties were spreadabili ty, viscosity, and superficial tension. Instrumental spreadability determination. Instrumental spreadability was determined by Table I Emollient Profile Descriptors Evaluation instance Attributes Description During application Difficulty spreading Difficulty in moving product over skin After application Gloss Amount or degree of light reflected off skin Residue Amount of product left Stickiness Slipperiness Softness Oiliness * Stafford Miller Ireland, LTD. ** Lab. Andr6maco. on skin Force required to separate a finger from skin Ease of moving two fingers over the skin Skin surface uniformity Type of residue (non-wet liquid that leaves residue) "Nil" "High" Baby oil Lanolin USP Corega®* Baby oil Untreated skin Hipoglos®** Baby oil Lanolin USP Lanolin USP Baby oil Corega® Glass Untreated skin Baby oil

178 JOURNAL OF COSMETIC SCIENCE gravity action at one-half minute and at one minute, by discharging 500 µl of emollient at 20° ± 1 ° C over a glass dish. Two perpendicular diameters of the area occupied by the sample at a half minute and at one minute (S0.5 and Sl) were measured and the mean calculated. Five determinations were performed for each emollient, and the homogeneity of variance of the data set was studied in order to verify that the values were comparable. Viscosity determination. The viscosity of the emollients was measured employing a Brook- field LVT viscometer. Measurements were carried out employing Spindle No. 1 at 60 rpm. Samples were thermostatized at 20° ± 1 ° C. Ten determinations were performed for each emollient, and the homogeneity of variance of the data set was studied in order to verify if the values were comparable. Surface tension determination. The surface tension of each emollient was determined by stalagmometry (13, 14). The method consists of weighing a drop of emollient obtained from the end of a calibrated capilar tube and determining the surface tension by means of a calibration plot. In order to draw the calibration plot, ten drops of known surface tension standards [diethyl ether, ethyl alcohol (absolute), chloroform, and carbon disul- fide (all HPLC grade), and distilled water} were accurately weighed in a milligram balance. The calibration plot that correlates drop weight with surface tension for each of the standards was obtained by regression analysis. A linear model with its y-intercept through the origin (direct proportionality), with a significance level of 95% (equation 1), was accepted: P 10 = 0.357 Xu (1) in which CT = surface tension (din/cm) and P 10 = weight of 10 drops (mg). Ten drops of each emollient were weighed and their corresponding surface tensions were determined by means of the calibration plot. All determinations were carried out at a constant temperature of 20° ± 1 ° C. DATA ANALYSIS A three-factor (assessor, sample, repetition) analysis of variance (ANOV A) was per- formed for all samples on the sensory data obtained. The mean rating and Fisher's least significant differences for each term were calculated by ANOV A. Principal component analysis (PCA) of mean ratings for each sensory and instrumental attribute was used to illustrate the relationship among variables and samples. All statistical analyses were performed using Statistica 5 .1 software (StatSoft Inc., USA). Linear partial least squares regression analysis (PLS) was used to analyze the relationships between sensory and physical matrices (15,16). PLS extracts a few linear combinations (PLS factors) from the physical data that predict as much of the systematic variations in the sensory data as possible. Because of the multivariate nature of the sensory data, PLS2 was performed ( 15), and all the sensory variables were correlated versus all the instru- mental variables simultaneously. Osten's F-test (17) was used to determine the number of significant (p 0.05) factors. The Gens tat statistical language Release 4.1, Numerical Algorithms Group (18) was used for these analyses. RES UL TS AND DISCUSSION ANOVA showed that between-repetition and between-assessor variations were not sig-