After each run the carbamate-urea cake had a decided reddish color due to the solution of chromium and other constituents of the metallic samples. All the samples were visibly affected by the exposure, the effects ranging from a slight tarnishing of the surface to an almost complete disappearance of the sample. The results of the two tests are listed in Table I. In the 4-day run triplicate samples of chrome steel, in different positions in the bomb, lost between 4.5 and 6.9 mg. per sq. cm. per day. This agreement is regarded as evidence of an ability to duplicate results, and practically eliminates the possibility of the rate of attack being influenced by the position of the sample in the bomb. In this run approximately half the samples showed either excessive corrosion or pitting. Because of the excessive corrosion and the possible influence of some of the products of corrosion upon the other samples, it was not considered worth while to calculate the results of this test in terms of penetration per year. The more promising metals were therefore more carefully studied in a second test. 

The 7-day test may be assumed to indicate the results which would be obtained in actual service. Of the selected metals included in this test, eight show less than 0.01 inch (0.25 mm.) penetration per year, with three others showing from 0.02 to 0.03 inch (0.5 to 0.76 mm.). There is no obvious connection between the composition of these samples and their resistance to this corrosion. This is especially true of the steel samples. It seems quite probable that heat treatment may have an important bearing on their corrosion resistance. The eleven samples showing less than 0.03 inch (0.76 mm.) penetration per year cover a wide range of mechanical properties. Of the metals tested, Rezistal 4 (sample 9), an austenitic chrome-nickel steel, is probably the best all-round metal for the construction of bombs, heads, etc. Duriron and Corrosiron (samples 1 and 8, respectively) are resistant to corrosion, but are brittle and almost unmachinable. Stellite (sample 10) is very hard and is unmachinable, but is tough and might be used for valve parts, finished by grinding. Illium G and Pioneer (samples 3 and 7, respectively) are both promising, particularly for valve parts and fittings. Both are reasonably hard but are machinable. Both are special alloys and probably too expensive for bulky construction. Lining the interior of the bomb with silver, lead, or nickel would be preferable to a tin lining. Chromium plating would be especially applicable to moving valve parts or in places where extreme hardness is desired. Invar (sample 11), a high-nickel steel, has rather a high rate of penetration.