276 REPOBT— 1889. 



also bent double cold, but broke in the radius with the least blow. 

 P (3'46 per cent. Si) was much stiffer, and bent only to a right angle. 

 G and H (4*49 per cent, and 6'63 per cent.) would not bend at all, and 

 were exceedingly brittle. These bending tests were confirmed by 

 Mr. Turner's experiments with bars of the same size ; up to D specimen 

 the samples bent to an angle of 180° with ^" radius. 



Pieces from the bars used for bending tests were also tested for 

 weldability, but entirely without success. The writer's experience has 

 always been that silicon is quite fatal to welding, notwithstanding that 

 the contrary might be expected from the fact that silica is of such 

 material assistance in welding wrought iron. 



As regards water-quenching or hardening, samples A, B, C, and D 

 ("24 per cent, to 2'18 per cent. Si) were unaffected by even the highest 

 heat. Even if plunged at welding heat into water made specially cold 

 no hardening beyond a surface stiflFening took place, nor did their tough- 

 ness seem impaired by this treatment. Specimen E (2'67 per cent. Si) 

 was heated to an ordinary yellow heat and plunged into cold water at 

 about 70° F. This piece was much stiffened, but only broke on being 

 bent double. Another piece of the same material heated to a welding 

 heat, and plunged into water at about 52°, also proved very stiff, and only 

 broke when bent double ; F (3'46 per cent. Si) : this sample was only 

 stiffened by being water- quenched at a welding heat. It was just as 

 brittle as before, and had not hardened, being easily touched by a file. 

 In this respect, therefore, i.e. as to being toughened by water-quenching, 

 this material differs from manganese steel. The heating did not cause 

 much alteration in fracture, the crystallisation being still open and 

 coarse. H (5'53 per cent.) was quenched both at ordinary heat and at a 

 welding heat, and although the surface was skin-hardened, upon being 

 fractured it was easily filed. These tests, therefore, clearly prove that 

 silicon does not confer the same property as carbon does upon iron, i.e. 

 of becoming hardened when dipped hot into a cooling medium. It 

 should also be here stated that the whole of the samples were subjected 

 to high heats, even to a welding temperature, without falling to pieces ; 

 in fact, as regards this point they behaved more like mild steel ; apparently 

 proving that silicon itself does not cause iron to become red-short. 



That silicon steel has a certain kind or degree of softness or lack of 

 body compared with carbon steel is especially brought out in the com- 

 pression tests (Table III.), where sample E, although apparently very 

 hard and brittle (high in the scale of Turner's hardness tests, viz. 33), 

 crushed up 38 per cent, of its length under a compression load of 100 

 tons per square inch. A correspondingly hard temper of carbon or 

 tungsten steel would not shorten more than 20 per cent., or, if hardened, 

 would remain unaltered. A very mild steel, containing not more than '20 

 per cent, of carbon, would not shorten much more than this sample, con- 

 taining over 21 per cent, of silicon. At any rate, therefore, silicon is not 

 so powerful a hardener of iron as is carbon. Mr. T. Turner has tested 

 tliis material in his sclerometer, or hardness-testing machine, the results 

 of which are given under this head in Table I. 



In order to test this steel in the shape of wire, samples E (2"67 

 per cent. Si) and G (4'49 per cent. Si) were reduced to rods, and Messrs. 

 Shipman & Co. of Sheffield kindly undertook to draw them into wire. 

 The G material, though readily rolled into rods, would not, however, 

 draw, nor did annealing soften or give it the requisite ductility. In the 



