BULLETIN OF THE BUSSEY INSTITUTION. 311 
These results illustrate the great importance of keeping in view 
the mechanical improvement of clay soils in all processes of fer- 
tilizing and tilling them ;—a point upon which much light has 
recently been thrown by the writings of Prof. Hilgard.* 
There is no need of discussing at any great length the arguments 
upon which the use of one and another of my chemical solutions 
were based. As the work proceeded the tendency has been to sim- 
plify the tests, and I would now use a much smaller number of 
chemicals than I dared restrict myself to at first. The follow- 
ing example will illustrate well enough the general line of argu- 
ment. In 1873 experiments were tried with six solutions, namely : 
1. Phosphate of Potash, to show if there was any nitrogen in the 
soil, available as plant-food. In case the soil was known to con- 
tain nitrogen, sulphate of potash would be used instead of the 
phosphate of potash, and the purpose of the jar would then be to 
exhibit the phosphoric acid in the soil, in comparison with jar 
No. 2. . 
2. Nitrate of Potash: To exhibit the presence of phosphoric acid 
in the soil; to show if any advantage is gained by adding potash 
or available nitrogen, and to contrast with jar No. 4. 
3. Rain-water: To contrast with the other jars, and particularly 
with No. 1. 
4. Nitrate of Lime: To show potash, phosphoric acid and the 
need of lime, and to contrast with jar No. 2.. 
5. Nitrate of Potash and Phosphate of Potash: To see if the soil 
needs lime. 
6. Nitrate of Lime and Phosphate of Potash: As the only 
‘¢ complete” mixture. 
It was held that the rain-water contained, beside some lime, 
enough chlorine and sulphuric acid for all practical purposes. In 
case magnesia or other essential element of plant-food were ab- 
sent from a soil, the fact would be indicated by the failure of the 
plants in jar No. 6, and new experiments could then be undertaken 
to meet that special case. The chief defect in so limited a num- 
ber of solutions as the foregoing is that we cannot get from them 
that mass of corroborative proof which is naturally afforded by a 
larger number of tests. When a single solution indicates some 
fact we are not sure but that this indication may be accidental, 
but when the results from two or three different solutions all point 
to one conclusion we are quite ready to accept it. 
* <¢ American Journal of Science,” 1879, 17. 205. 
