EVALUATION OF HAIR DYES 451 transmission, or the intensity of illumination, a given per cent difference in color will give a finite a.c. voltage to the meter. In effect, the sensitivity of the photocell is varied inversely with brightness of light reaching it, just as the iris of the human eye and accommodation of the brain act to decrease the eye sensitivity when a brighter object is viewed. The range of automatic accommodation of the instrument is comparable to that of the eye, but it is faster and has no fatigue factor with time. To obtain the tristimulus values, all one has to do is to insert into the light beam, prior to its reaching the photocell, the desired primary filters. The electronic system automatically transmits the a.c. output through a component necessary to express the difference between sample and standard in terms of reflected light. The per cent values, or readings, of the instrument are designated by the letters X, Y and Z, which are known as the tristimulus values, each letter representing the amount of one of the primary stimuli. The results obtained from the instrument indicate that, when two sam- ples have the same values of X, Y, Z, the quality of the reflected light from the samples for the specified illuminant is approximately the same. There- fore, dye classification by the equivalence of the tristimulus values appears to give sufficiently good approximations for matching. It is not our desire to go into the physical or optical principles necessary to describe reflectance or color absorbancy instruments. Nor do we intend to delve into the mathematical procedures required for the definition of tri-color stimuli used in the colorimeter. This may be found, for those who may be interested, in the many excellent publications on the subject (2-4, 6). The dyed human hair swatches, which are customarily used to evaluate color pick-up visually, proved to give poorly reproducible results when they were read on the colorimeter. This was probably due to the non- uniform affinity of the hair for the dye materials. It was decided that a suitable substitute be made for human hair. By experimentation, it was found that the monochromatic absorption of a dye on certain textile fabrics, as on the hair, is proportional to the concentration of the dye present. Textile fibers which were found to be unsuitable included synthetics, such as cellulose acetates and nylon, silk, cotton and linen (8, 9). Wool, be- cause of its keratin protein composition, was found to be satisfactory (10). Many of the woolen materials that were tried also had undesirable qualities. Cashmere, flannel and some worsteds gave a surface with too great a nap. A tightly woven wool worsted, obtained from Test Fabrics, Incorporated, New York City, was found to give excellent and reproducible results. Samples of the cloth were dyed, employing the same controlled conditions that are used in the dyeing of a test subject's hair. Great care was taken to maintain the following conditions constant: dyeing time, dye bath

452 JOURNAL OF THE SOCIETY OF COSMETIC CHEMISTS concentration, temperature, washing solutions and drying procedures. The colorimeter is calibrated by a standard which is provided with the instrument. Then the dyed wool patch is instrumentally compared with an undyed, identical patch of wool. The tristimulus values are recorded and graphed on a specialized graph paper provided for the purpose. The standard wool patches used for a particular color are produced from a re- tained color sample which has been previously accepted by a color expert in our Dye Color Center. This standardized color has satisfactorily been applied to the hair of many subjects. When, for example, we dye our wool patches with a warm brown dye, we may find that our readings will be plotted above that of our standard. This indicates that the color take-up is too slight and produces a brighter hue or weaker color. This has been confirmed on the dyeing of natural hair. The indication, that more of the dye material must be added to our production batch, is obvious. If we over-correct with too great an addition of dye material, our readings are plotted below our standard. This also has been confirmed on a subject's head. By exercising great care and from the experience of many observations and adjustments, a good colorist may usually make a proper correction in a single addition or dilution. The method of control outlined has been primarily presented for the control of finished products. This, however, is not the only use which may be made of this system. Specifications for raw materials may be formulated by the use of this form of colorimetry. The approximate purity of individual dyes may be estimated. The effectiveness of bleaching prior to applying dyes can readily be revealed. It would appear that we have diminished, to a great extent, the problem of color control on hair dyes. Actually this is only partly true. An old axiom, that all control chemists know well, states, "The control is only as good as the standard used." Our standard, the dyed textile patch, has a distressing fault. We may perform a test of a sample against a standard and obtain a near perfect match. However, after an extended period of time, they will not match because of fading of either the sample or the standard, or both. The fading of a standard patch will necessitate a complete restandardization cycle and presents an entirely new problem. We have decided that this can be corrected by the preparation of colored ceramic tiles which will give the identical values as those of our wool patches. We are attempting, at the present time, to correlate our work with this idea. The use of the tristimulus colorimeter to conduct the routine control of hair coloring products is in no way to be considered a means to replace the expert colorist. The expert is still needed to establish the standards by which the instrument may be calibrated. The colorimeter will, however, relieve him of the time-consuming and tedious drudgery involved in the