JOURNAL OF THE SOCIETY OF COSMETIC CHEMISTS restricted application, and uncertainty in the interpretation of results. Such techniques include the alloy-tin couple test of Kamm and Willey (7), the anodic polarisation method of Hancock and Mayne (8), the over-voltage- intercept method, potentiostatic polarisation, electrical capacity deter- minations as used by Tomashov, Mikhailovsky and Leonov (9), and zero charge potential studies as described by Antropov (10), and Brasher (11). The equipment needed for the above methods is not complex. The main requirements are a high impedance millivoltmeter, a microammeter, an adjustable low current source such as that described by Tomashov et al (9), an accurate balance, and a metallurgical microscope. CORROSION IN THE VAPOUR PHASE Little is as yet known of the mechanism of corrosion in the vapour phase, although research is being pursued (12 - 14). Unless the atmosphere is quite dry, a thin film of moisture is always present on the surface of metals at room temperature. The thickness of this film varies according to humidity, but is usually about one micrometre thick. It is, of course, very much thicker in atmospheres of 100% RH when physical condensation occurs. Electrochemical corrosion processes occur in these thin films, but they are somewhat different from those occurring in bulk solutions of the same composition, since such controlling factors as gas diffusion, diffusion coefficient of dissolved substances, and dielectric constant are considerably modified by action of surface forces on the water molecules, and other intermolecular reactions. Inhibitors for vapour phase corrosion are known, although their mode of action has not yet been established. Such an inhibitor must have an appreciable vapour pressure so that it is transferred to the metal surface, which it must then protect. One theory is that the inhibitor undergoes hydrolytic dissociation, the products of which evaporate and recombine in the water film (13). Electrochemical processes in thin water films (160-350 [•m) have been studied by Rosenf'ld (12) and thinner films (about 1 Mm) by Mindowicz and Puchalska (13) using a microelectrode. This technique is as yet not well developed, but may eventually lend itself to evaluation of vapour phase corrosion in a metal container. APPLICATION OF METHODS No single test will provide adequate information for all corrosion

PRINCIPLES OF CORROSION OF METAL CONTAINERS 25 problems occurring in packaging. Each case has its own requirements, for which relevant tests must be selected. The results must be interpreted both with a knowledge of the limitations of the test, and in the light of informa- tion gained from other tests. The amount and reliability of corrosion data which can be deduced will vary considerably from one case to another, depending on the complexity of the system under review. We have found that the best approach in the first instance is to establish the polarity relations between the metals occurring in the system. This is conveniently done by using a cell in the form of an inverted T. Metal plates can be sealed at either end of the cross-piece, and the 'stem' is used to fill the cell, and to give access for a reference electrode, if required. The magni- tude and direction of the current flowing between the plates can then easily be measured. A similar cell has been used by Koehler (15) for potentio- static studies. If a reference electrode is used, polarisation resistance measurements may be made at the same time. When this basic information has been obtained, together with conductivity and pH measurements, other tests can be applied if considered necessary. Potentially dangerous situations occur when a large cathodic area is in contact with a small anodic area, for under these conditions anodic attack is intensified due to the disproportionate sizes of the areas. For instance, a large area of oxide film on a metal can act as a cathode, although the forma- tion of such films is less likely in organic solutions. Tinplate consists of large areas of tin in contact with small areas of tin/iron alloy or steel at flaws in the tin coating, and if these flaws are anodic, pinhole corrosion will occur. However, although tin is more noble than iron in the electrochemical series, it is fortunately anodic to steel in many solutions, and consequently there is only slow solution of the tin since there is so small an area available for the equivalent cathodic reaction. If the presence of tin in the product is undesirable it may be considered necessary to apply a lacquer coating, but this should be done with caution since in some cases corrosion could become more acute. Lacquer should not be used as a protection if the tin is cathodic to steel, since most com- mercially applied lacquer systems contain flaws to a greater or lesser extent, and pinhole corrosion could still occur. Given a new product and container combination it is not usually possible to state whether or not it will be satisfactory in practice only from these tests. Their main value lies in providing comparative results, from which the most promising of a number of formulations may be selected. In