46 SECTIONAL ADDRESSES. 



The apparatus employed for this purpose is shown among the exhibits 

 from the laboratory. Originally designed for zinc, it is now being used 

 extensively for work on iron and steel. 



Oxygen passes through water or salt solutions to the immersed metal 

 either by diffusion or convection, but the latter mode of transference is 

 by far the more effective at more than very shallow depths. Convection 

 currents may arise in a salt solution owing to four different causes : 

 (1) temperature changes, (2) density changes produced by evaporation at 

 the surface layer, (3) density changes produced by differences of oxygen 

 concentration, (4) mechanical agitation. The apparatus employed for 

 these quantitative experiments is immersed in a thermostat and corrosion 

 occurs in a closed space within it, so that the effects of temperature changes 

 (1) and evaporation{2) are practically negligible, and special precautions are 

 taken to prevent agitation (4). Accordingly, by removing oxygen from 

 the neighbourhood of the metal, the corrosion process produces convection 

 currents of the third category due to changes in concentration of oxygen. 

 The velocity of these convection currents depends on the difference in 

 density between the solution saturated with oxygen at the liquid surface 

 and the solution next to the metal. Assuming that the latter solution 

 contains very little oxygen, the velocity of convection will probably be 

 proportional to the solubility of oxygen in the liquid, but the amount of 

 oxygen carried by the current is also proportional to its solubility. Hence, 

 the rate of corrosion (y) should be proportional to the square of the 

 oxygen solubility (x), a relation which is expressible by the equation 

 y=kx^. This assumption has been verified for on plotting the observed 

 rates of corrosion against oxygen solubility one obtains curves of parabolic 

 form. 



Hydrogen evolution due to the interaction of water or salt solution 

 with metals such as zinc or steel is of greater importance than is generally 

 supposed. Determinations of the hydrogen liberated during zinc corrosion 

 have shown that a very small amount of impurity has a considerable 

 influence on the amount of gas evolved. In N/10,000 potassium chloride 

 measurable quantities of hydrogen are obtained from 99.99 per cent, 

 zinc, whereas no hydrogen was detected from zinc of spectroscopic purity. 

 The proportion of zinc corrosion due to evolution of hydrogen increases 

 with concentration of potassium chloride, and with 2N-solutions it 

 amounts to 17.4 per cent, of the total corrosion. 



When all the foregoing factors are taken into account, successive 

 corrosion experiments exhibit a high degree of reproducibility, and the 

 curves indicate that duplicates differ from their mean value by 1 per cent, 

 or even less. This constancy indicates that the corrosion of zinc and 

 allied metals is not inherently erratic, but is quite a suitable subject for 

 physico-chemical investigations. 



Atmospheric Corrosion. 



Investigations of various types of indoor and open-air corrosion and 

 of protective oxide films, previously conducted under the auspices of the 

 Atmospheric Corrosion Research Committee of the British Non-ferrous 

 Metals Research Association, were taken over by the Department of 

 Scientific and Industrial Research in July 1927. This work was con- 



