The result was encouraging, and led to the probability of the nitrogen being altered in some way, or of the presence of some new component of the atmosphere. An experiment was therefore begun on a larger scale, the atmospheric nitrogen being passed backwards and forwards from one large glass gasholder A to another B, through a tube filled with magnesium heated to redness G, to absorb nitrogen; over red-hot copper oxide (_a_) (_b_), so that any carbonaceous matter such as dust should be oxidised to carbon dioxide and water; and these, if produced, were absorbed by placing in the train of tubes, one filled with a mixture of soda and lime F and I, to absorb any carbon dioxide which might possibly be formed, and two filled with pentoxide of phosphorus D and H, to dry the gas, so that water-vapour, carried along with the gas from the gasholders (which contained water) might be removed before the gas passed over the red-hot magnesium; for water acts on hot magnesium, forming oxide of magnesium and hydrogen, and the gas would have become contaminated with the latter had this precaution not been taken. The process was continued for ten days, by which time most of the nitrogen had become absorbed. The apparatus was then somewhat altered, so as to make it possible to work with a smaller quantity of gas; but the tubes destined to absorb nitrogen, hydrogen, etc., were filled with the same materials as before. In a few days more the volume was reduced to one-seventh of what it had been when the transference to the smaller apparatus was made, or about one-eightieth of the original volume of the atmospheric nitrogen taken.