It must be admitted that the evidence is indirect. The rate of oxidation of ammonia by bacteria in the soil is more rapid than the rate of formation, so that ammonia is practically never found in the soil in more than minimal amounts (1 or 2 parts per 1,000,000); indeed, the only evidence of its formation was for a long time the fact that no compound other than ammonia could be oxidised by the nitrifying organism. It has, however, since been shown at Rothamsted that ammonia accumulates in soils in which the nitrifying organism has been killed. Nothing is known of the mechanism of the oxidation of ammonia beyond the fact that it is biological; the reaction is not easily effected chemically at ordinary temperatures. Possibly the organism assimilates ammonia at one end of a chain of metabolic processes and excretes nitrates at the other. Or, the reaction may be simply a straight oxidation for energy purposes, the ammonia changing to hydroxylamine and then to nitrous and nitric acids. The nitrate does not remain long in the soil. Some is taken up by the plant and some is washed out from the soil. Part, however, either of the nitrate itself or of one of its precursors is converted into an insoluble form: probably it is changed into protein by the action of micro-organisms; it then goes through the whole process once more. These are the general outlines; they present no particular chemical difficulties. When we come to details, however, there is much that cannot be understood. First of all, there is the slow rate at which complex nitrogen compounds disappear from the soil in comparison with the rate of oxidation of the carbon. Thus, in the original plant residues, there is some forty times as much carbon as nitrogen: before they have been long in the soil there is only ten times as much carbon as nitrogen; this seems to be the stable position. What is the reason for this preferential oxidation of the carbon? No explanation can yet be given. [Illustration: FIG. 24. X-axis: 1887-8 1890-1 1900-1 1910-11 Y-axis: ℔ per acre] An equally difficult problem arises in connection with the length of time the process will continue. Decomposition of the nitrogen compounds never seems to be complete in the soil; it dribbles on interminably. In the year 1870 Lawes and Gilbert cut off a block of soil from its surroundings and undermined it so that the drainage water could be collected and analysed. The soil has been kept free from vegetation or addition of nitrogen compounds from that time till now; yet it has never failed to yield nitrates, and the annual yield falls off only very slowly (Fig. 24). This same peculiarity is seen in the yield of crops on unmanured land: it decreases, but very gradually; even after eighty years the process is far from complete, and there is no sign that it will ever come to an end. TABLE XVII.--APPROXIMATE LOSS OF NITROGEN FROM CULTIVATED SOILS: BROADBALK WHEAT FIELD, ROTHAMSTED, FORTY-NINE YEARS (1865-1914.)