POSSIBLE CAUSES OF THE GRASS EFFECT 285 



the surrounding soil, and much more conclusive results were 

 obtained by pot experiments. The value of pot experiments is 

 so generally recognised that we rarely hear now the objection 

 which used to be urged against them that they do not reproduce 

 the conditions prevailing in ordinary practice : it is true that 

 they do not do so, but it is therein that their value lies ; for by 

 modifying these conditions, and eliminating the multitude of 

 unknown factors which must attach to experiments in the field, 

 pot experiments simplify matters so far, that the influence of 

 one factor alone can be studied. The case of the water-con- 

 tents of the soil is a very good instance in point, for, when dealing 

 with trees grown in pots, it is easy to keep the soil in the same 

 condition as regards moisture, whatever the other conditions 

 may be, weighing the pots at frequent intervals, and adding 

 such water as is necessary to supply any loss. In that way it 

 was proved beyond question that the water-content of the 

 soil had nothing to do with the action of grass on trees, for 

 trees grown under these constant conditions, but with a surface 

 crop of grass, showed, at the end of two years, a total growth 

 amounting to only 24 per cent, of that shown by trees without 

 grass (XIII, 67). Nor could it be urged, however improbable 

 such a suggestion might be, that, though the water in the soil 

 was the same in both cases, the grass absorbed the greater part 

 of it to the detriment of the trees, for in some cases a layer of 

 fine copper gauze was spread four inches below the surface, so 

 that the grass roots could not penetrate down to the tree roots, 

 and the water was supplied from the bottom of the pot, thus 

 enabling the tree to get whatever it wanted before the water 

 reached the grass : yet the results were exactly the same, the total 

 growth after two years being only 24 per cent, of that of trees 

 without grass. 



The pots used in these experiments were of glazed earthenware, 

 holding about one hundredweight of soil, and in some subse- 

 quent experiments a device was adopted whereby the separation of 

 the grass-roots from the tree-roots was made more complete than 

 by the mere interposition of the layer of gauze between them ; for 

 the grass was grown in a separate tray made of sheet iron, coated 

 with paraffin wax, the tray being perforated at the bottom, 

 with a sheet of copper gauze covering the perforations ; the 

 tray simply rested on the soil, and could be removed at any time 

 (Fig. 39, A, p. 292). The water required was added to the tray, 

 and soaked down through it to the tree below. In the check 

 experiments, where no surface crop was present, the arrange- 



