EXPEPJMEXTAL WORK TVITH EAW EOCK PHOSPHATE. 19 



7. Kind of crops groAvn and selection of seed. 



8. Rate of aj^plication and uniform distribution of pliosphat-es. 



9. ]SIethods of comparing raw rock with other phosphates. 



10. Effect of other fertilizers. 



11. Xnmber and distribution of plots. 



12. Duration of experiment. 



irsiFOKiriTT or expeeijie>'tai. field. 



The greatest care should be exercised in selecting a field contain- 

 ing soil of a uniform character throughotit, for unless the soil on the 

 various plots has the same productive power or the probable dif- 

 ference in its natural fertility on these plots is determined, the yields 

 obtained from the application of fertilizers may have little or no 

 meaning. A soil survey of a field is of great importance, but stich a 

 survey unless supplemented by borings may not give sufficient in- 

 formation, since an apparently uniform soil may so vary in depth 

 fi^om place to place as to catise wide differences in the productivity 

 of the various plots. Hall and Russell ^ state that a simple ret sensi- 

 tive method of determining the uniformity of a soil consists in as- 

 certaining the percentage of moisttire in samples collected from va- 

 rious parts of a field at the same time to the same depth. An even 

 more sensitive test, it is said, consists in determining the percentage 

 of nitrates in such samples. 



Piper and Stevenson- state that it is difficult to say what is the 

 ordinary error due to soil inequality. Hall and Russell^ state that 

 the Rothamsted records show that there is an error of 10 per cent on 

 plots where the past treatment has been tiniform and general con- 

 ditions were favorable for experimental work. 



In conducting fertilizer experiments in the field attempts have 

 been made to determine the so-called " natural fertility *' of the 

 plots subsequently treated with fertilizers. Offhand the most logical 

 method aj>pears to be to complete at least one rotation of the sys- 

 tem to be employed during the experiment without the addition 

 of any fertilizer whatever. The usefulness of this method of valuing 

 the plots, however, appears very doubtful. In fact, it might well 

 be asked. " Is there such a thing as nattiral fertility under field con- 

 ditions when these conditions vary so greatly from year to year?" 

 Lvon ^ has shown that when corn. oats, and wheat were grown for 

 several years on the same fields no definite relation was shown be- 

 tween the yields of the various plots from year to year. 



iChem. Xe^s, 102. ISO (19101. 



* standardization of Field Experimental Methods. Froc. Am. See. Agron., 2. 70-76 

 (1910). 



'Error of Experiment in Agricultural Field Tests. Chem. Xews, 102, SO (1910K 



* Experiments to Estimate Errors in Field Plat Tests. Jour. Am. Soc, Agron., S, 

 S9-114 (1911). 



