26 JOURNAL OF THE SOCIETY OF COSMETIC CHEMISTS Before considering the methods in use for testing chemical durability, it is as well to mention briefly the nature of the processes involved in the decomposition of glass by water. It is not a simple case of solution, and the testing of glass is not merely the determination of its solubility in water. On the contrary, the process is a highly complex one, involving the penetration of the glass by water and the subsequent decomposition of the complex silicate mixture, with formation of substances wholly different from those originally present. The attacking medium is not always water, however, and frequently the resistance of glass to the action of aqueous solutions is of importance. In many cases, solutions are dilute enough for differences caused by the dissolved material to be insignificant, but this is not always true. Whether or not such substances are present, the action will tend to go on until decomposition of the glass is complete. The presence of an acid or alkali will often profoundly affect the rate of decomposition and the effect will be dependent both on the composition of the glass and of the dissolved material. Decomposition of most glasses by water results in the liberation of alkali and the liberated alkali may accelerate further decomposition of the glass by dissolving away the silica framework and thus exposing fresh glass to attack, If, however, the reagent be continuously removed and replenished, or if it remains acid, the attack will slow down owing to the formation of a surface film, richer in silica than the original surface. This film may be indicated by faint iridescence. For these reasons, some types of glass, notably those containing a high percentage of silica and boric oxide, are more resistant to acid solutions than to water, and more resistant to water than to alkalis. On the other hand, glasses containing a low percentage of silica, such as are encountered in the optical glass range, are more rapidly attacked by acid solutions. In general, however, the rate of decomposition of glasses by alkaline solutions is much greater than by water, or by acids, and the glass may, in prolonged contact with hot caustic alkali solutions, appear to be unaffected yet be dissolving as a whole. The alkali silicates formed by the reaction of glass silica and reagent alkali are soluble in water and dissolve, thus in no way forming a protective coating. But ff the solution becomes loaded with extracted salts, or if certain salts are added deliberately, the reaction can be slowed down. Silicates, aluminares, and zinc salts exert this influence, and are used in commerce to decrease the attack of alkaline cleaning agents on glass. In a similar manner, concentrated solutions of acids as well as salts have been considered to have no direct action as such, the attack being due exclusively to water. This view naturally leads to the conclusion that a large amount of acid weakens the attack by diminishing the concentration of water, a conclusion borne out by experimental results. It is evident, therefore, that the term "solubility" has no meaning in the case of glass and there can be no quantitative figure for the solubility of even

AND PROPERTiEg OF GLASS CONTAINERS 27 the simplest glass in water. All that can be measured and compared, is a rate of attack, and all the measurements that have been made represent not solubilities but rates of reaction. Rates of reaction, however. differ from equilibrium values in being highly susceptible to slight differences in experi- mental conditions, and unfortunately many of the data in the literature have been obtained under poorly defined experimental conditions. Com- parison of the results of one worker with those of another is rarely possible. EXPERIMENTAL METHODS Since it is not possible to measure definite solubilities of glasses, the suitability of a glass for a particular purpose is best determined by actual service, or by exposing it to conditions simulating as nearly as possible those that will be met. Thus, for optical glasses, actual use in the finished instru- ment is the ultimate criterion and exposure to the elements under service conditions probably the next best. Such tests require a long time, and in addition they lack one prime requisite of a test of practical value in the development of resistant glasses. This requisite is that the test shall give such quantitative information that the glass maker can have a categorical answer to the questions whether or not a given change in composition or treatment is benficial, and which of several possible changes is the one which will give the most useful glass. It is in the attempt to provide such a method that much of the information that we have on the chemical proper- ties of glass, has been obtained. Most of the methods that have been used for testing the weathering qualities of glasses are variants of a few main ones. All have been developed from the same basic thesis, namely, that the verdict of actual service may be anticipated by accelerating the corrosive action by the use of higher tempera• tures, and by using sensitive methods to detect and measure the resulting action. Several of these methods have been combined for each has its faults as well as its virtues, and no method can be safely chosen as the best to be applied universally. The choice of method is largely determined by the use to which the glass is to be put. Methods suitable for differentiating between glasses to be used in optical instruments may be far too mild to discriminate adequately between types of chemical glassware. Thus, exposure to moist air would have little effect on Pyrex, etc., but it affords a suitable method for studying optical glass. One method which has been applied extensively to optical glass is the iodeosine test of Mylius, which forms the basis of the so-called hydrolyric classification. A surface of known area is immersed in a solution of iodeosine in ether for one minute, the excess iodeosine is rinsed off with ether, the alkali salt of the dye is dissolved in water, and the amount of iodeosine combined with the alkalis is estimated colorimetrically, by comparison with