[stuart] EXPERIMENTAL WORK WITH FERTILIZERS 171 



If now, we could succeed, as outlined before, in eliminating all 

 variables, other than the amount of the fertilizer required, we could 

 locate exactly the minimum for each and every point in the triangular 

 base of the prism. From each of these points we could erect a wire 

 cut in correct length to represent the minimum for that point. The 

 result would be a bundle of wires of different lengths standing up inside 

 the prism and completely covering the triangular base. We can now 

 join up the uneven ends of these wires by a surface. 



Judging from the results of fertilizer experience, it is likely that 

 the shortest wires would be at some point towards the centre, from 

 where they would lengthen towards the sides and corners. The 

 surface now formed by joining the ends of the wires would somewhat 

 resemble an inverted cone inside the prism. The apex of this cone is 

 the answer and from it is read off the fertilizer required and the cor- 

 rect amount. 



V. THE PLAN APPLIED TO FERTILIZER EXPERIMENTS. 



1. Control of the variables. 



At the start it would be wise to restrict the work, in so far as is 

 possible, to certain constants. For this the greenhouse, in the case 

 of certain crops, might enable us to conduct some of the preliminary 

 work. Although of course no absolute readings could be made from 

 this work, the larger differences might be ascertained and the work, 

 now narrowed down considerably, transferred to actual conditions of 

 the field. In the greenhouse we can fairly effectively control all the 

 variables and reduce these to constants — soil, water, temperature, air, 

 etc., etc., Also it would permit of greatly enlarging the number of 

 tests to be made of the various points in the triangular diagram. 



Having now eliminated from discussion many of the mixtures of 

 fertilizers, we could turn to the field work. We could still control the 

 variables somewhat by following uniform practices and by resorting to 

 as many averages as possible. As an example we might choose one of 

 the commonest rotations, such as roots (first year), grain, seeded with 

 timothy and clover (second year) and hay (third year). Observations 

 could be made throughout the complete rotation and due regard had 

 to the "after effects" of fertilizer, the sum of all to be taken as the 

 reading. In passing, it seems possible that, by such means, the best 

 treatment for each crop in the rotation could be worked out, rather 

 than the usual practice of fertilizing in excess but one crop in the rota- 

 tion and relying on the residues to benefit other crops. In this way 

 costs could be properly proportioned and charged up equitably against 

 each crop. 



