[28] The sparks of rust are produced, owing to the fact that the volume of the oxide of iron is nearly twice that of the volume of the iron, and as the heat evolved is not sufficient to entirely melt the oxide or the iron, the particles must be torn off and fly about. Similar sparks are formed in the combustion of iron, in other cases also. We saw the combustion of iron filings in the Introduction. In the welding of iron small iron splinters fly off in all directions and burn in the air, as is seen from the fact that whilst flying through the air they remain red hot, and also because, on cooling, they are seen to be no longer iron, but a compound of it with oxygen. The same thing takes place when the hammer of a gun strikes against the flint. Small scales of steel are heated by the friction, and glow and burn in the air. The combustion of iron is still better seen by taking it as a very fine powder, such as is obtained by the decomposition of certain of its compounds--for instance, by heating Prussian blue, or by the reduction of its compounds with oxygen by hydrogen; when this fine powder is strewn in air, it burns by itself, even without being previously heated (it forms a pyrophorus). This obviously depends on the fact that the powder of iron presents a larger surface of contact with air than an equal weight in a compact form. [Illustration: FIG. 30.--Mode of burning a steel spring in oxygen.]