282 EIGHTH REPORT 1838. 



change of composition bj' contact with the fuel, and cast in 

 green-sand moulds into the form of parallelopip.eds, of 5 inches 

 by 5 inches x 1 inch thick, and of 5 inches by 5 inches x | 

 inch thick, respectively ; and at the same time a bar of 1 inch 

 square and 12 inches long was cast of each description of iron. 

 The whole of these specimens, whose surfaces are as nearly 

 equal as possible, were then weighed each to a single grain, or- 

 within about :f o Vo o °^ ^^^ weight of the piece, and inclosed in the 

 external frame of the box (fig. 1, as shown in Plate XVII.), No. . 

 This box is so contrived as to permit free access of air and sea 

 water at all sides, and while the specimens of iron are held fast 

 at four of their angles, they are freely exposed to the action of 

 these agents ; but each in a separate cell, for a reason to be here- 

 after mentioned. As any mode of numbering these specimens 

 would be inadmissible, if not impracticable, they are to be re- 

 cognised when reexamined solely by their place in the box. 

 The series commences at A. fig. 1, and reads from left to right, 

 going upwards, as more particularly described in the notes at- 

 tached to the tables. No iron used in the construction of the 

 box enters its interior, and the specimens, with the frame in 

 which they are arranged, can be lifted out at any time for in- 

 spection, without disturbance of their position or touching their 

 surfaces. This box, like the others to be described, is of stout 

 oak kyanized, which, although the researches of Lassaigne upon 

 this subject, showing that the combination of albumine and bi- 

 chloride of mercury is soluble in alkaline chlorides, renders it 

 probably of less service in sea water, yet is not likely to inter- 

 fere in any ^vay with the results of these experiments. 



69. The object of casting the parallelopipeds of two thick- 

 nesses, viz. 1 inch and j inch, is, that the " grain " or crj'stalline 

 arrangement, and proportion to the metal of " skin," as it is 

 technically called, varies with the scantling of the casting : 

 hence these thin castings will give residts discovering what 

 variety of British metal produces a skin best calculated to re- 

 sist corrosion, and what amount of variation of skin each sus- 

 tains by difference of thickness in casting. 



70. The castings were all poured, as nearly as possible, at 

 the same temperature, (the crucibles having been heated in 

 draught furnaces,) and all permitted to cool at the same rate. 



Their forms being all regular, and their dimension and weight 

 known on again weighing after taking up and cleansing from 

 adherent matter, a set of numerical results will be obtained, 

 giving the relative rates of degradation in sea and fresh water, 

 of most of the British cast irons per unit of surface ; thus ena- 

 bling the engineer to choose that which will be most durable in 



