HYGROSCOPIC AGENTS AND THEIR USE IN COSMETICS I1 of humectants has been proposed by Livengood (21). This method rep- resents a desirable simplification for practical use. However, the hygroscopicity data presented are not in agreement with published data, and his proposed relationship of water content as a logarithmic function of relative humidity from 20 to 100% relative humidity does not hold over this humidity range. The semilog relationship is true over a limited range within an ac- curacy of about 4-5% H•O for most organic humectants. A revised form of the Livengood graph with a corrected family of curves on a limited scale is shown in Fig. 2. To ascertain the general behavior of an organic humectant the water- holding power at 50% relative humidity is determined by reference to Table 1. This value is then transferred to the 50% relative humidity vertical line on Fig. 2. A curve is drawn through this point, parallel to the curves in the graph and the approximate water reten- tion properties of the humectant at any desired humidity between 25% and 75% is then read off the curve. The graph and table may also be used to choose a humectant that will hold a desired amount of water at a given humidity by following the reverse of this procedure. Blends or mixtures of humectants apparently do not exhibit hygro- scopicities exactly in accord with the arithmetic average of their in- dividual hygroscopicities, as seen in Table 2. However, since moisture control by humectants is influenced TABLE 2--HYGROS½OPI½ITIES OF BLENDS OF HUMECTANTS R.H., % Water Held at Equilibrium-- 50/50 Arlex-Glycerol----. Found* Calculated t 30 8.7 8.3 5O 17.6 16.6 70 32.0 30.9 * Points from curve plotted from actual data. t Arithmetic average of points taken from curves plotted from actual data. more by the surrounding atmos- phere than by the over-all choice of humectant, an arithmetic average will usually .suffice for estimate cal- culations. There are occasions wherein blends of non-volatile and volatile humectants are used, for example, sorbitol and propylene glycol. Here, when a large surface to volume is exposed, the over-all hygroscopicity will change due to the change in composition and net remaining weight of the humectant. DYNAMIC HYGROSCOPICITY DATA Dynamic hygroscopicity studies have been carried out in our labora- tory utilizing the sample-on-sand technique. Our reason for using this method was twofold--since the sample is present in a thin film, the actual rate of moisture change of the product is not influenced by the rate of moisture transfer inside the solution, and the large area ex- posed provided a rapid gain or loss which allowed the study to be com- pleted in a relatively short time. The importance of the thickness of the film in rate measurements is graphically illustrated in Fig. 3, which shows data for several depths,