THE BIOLOGICAL CONDITIONS IN THE SOIL iii 



species, but not necessarily plants of different species. He thus ex- 

 plained the well-known fact that a rotation of crops is more effective 

 than a system of continuous cropping ; in a rotation the toxin excreted 

 by a particular crop is innocuous to the succeeding crops and disappears 

 from the soil before the same plant is sown again. 



The hypothesis was tested in a classical research by Daubeny at 

 Oxford (78), but could not be justified. Eighteen different crops were 

 grown continuously on the same plots, and the yields compared with 

 those obtained when the same crops were shifted from one plot to 

 another, so that no crop ever followed another of the same kind. No 

 manure was supplied. The results showed a gradual decrease in the 

 yield in almost every instance, and the decrease was generally greater 

 when the crop was repeated year after year on the same plot than 

 where it was shifted from one to another. Nevertheless the difference 

 between the yields in the two cases was not sufficient to justify any 

 assumption of the existence of a toxin, except perhaps in the case of 

 Euphorbia lathyris ; in the other seventeen cases it was attributed 

 to the more rapid removal from the continuous plots of the mineral 

 nutrients required by the plant. This explanation was supported by 

 analyses of the plant ash and of the soil — analyses which led to the im- 

 portant distinction between " available " and " unavailable " plant food. 



Pot experiments made by the writer at Rothamsted have led to the 

 same conclusions. Six crops of rye were grown in succession in sand 

 to which only nutrient salts were added so as to maintain the food 

 material at a constant amount. A seventh crop was then taken and 

 at the same time a crop was grown on perfectly fresh sand, on which no 

 crop had ever grown before, but which was supplied with an equal 

 amount of the same nutrient salts. There was no significant difference 

 in the two crop yields. A similar experiment was made with buck- 

 wheat, another with spinach, and a parallel series was made in soil 

 cultures. In all cases but one the result was the same; the 1910 

 weights were as follows : — 



Both sand and soil contained 2 per cent, of calcium carbonate. 



