368 THE SCIENTIFIC PAPERS OP 



leaves a larger nett area of metal to resist the strain. Then again, 

 if the rivets are put the same distance apart as they would be put 

 in an iron plate 20 per cent, thicker, it is natural that these rivets 

 stand too far apart for this reduced thickness of the plate, and 

 thus throws a buckling strain on the plate. I would suggest that 

 in riveting steel plates together, the distance between rivet and 

 rivet should be the same as that which would be adopted if a thin 

 iron plate were used in fact, the reduction should be propor- 

 tionate. The moment you use one material for the plate and 

 another material for stitching the plate together, there is a dis- 

 proportion between the two elements introduced which naturally 

 must tend to weaken the whole. Perhaps I may be allowed to say 

 one word before I sit down with regard to corrosion. Certain 

 laboratory experiments which have been tried seem to show that 

 steel corrodes faster than iron, whereas most of the working results 

 that have come to my knowledge seem to show the contrary, and 

 that steel wears longer. We have heard to-day of steel ships from 

 Mr. Martell himself, that have stood wear and tear exceedingly 

 well. But I have no doubt that some steel may be introduced 

 which is liable to corrode at a greater rate than iron would ; and 

 although I perhaps have no right to speak absolutely and definitely 

 on the subject, I believe it is the result of an excess of manganese 

 in that steel, and it would perhaps be well for steel users to turn 

 their attention to that subject. Manganese is an exceedingly con- 

 venient agent in order to give to steel, or homogeneous iron, the 

 high degree of ductility which is desirable ; but it is by no means 

 a necessary element, because at least the same ductility, and the 

 same strength, may be obtained without an appreciable percentage 

 of manganese in the material ; and my experience, which is some- 

 what limited as to the wearing quality and resistance to corrosion 

 of the two materials, goes distinctly to prove that with an increase 

 of manganese beyond a very narrow limit the steel becomes more 

 corrosive, does not weld with the same facility, and is not so abso- 

 lutely uniform as when a little manganese is used. 



