CHEMICAL SCIENCE. 229 



ing from cream the greatest amount of butter in the shortest possible 

 time is between 57 and 60. With the same apparatus, by varying 

 the temperature, the duration of the operation may be varied in the 

 proportion of ten to one, while very variable amounts of butter are 

 produced. 



SOIL ANALYSIS. 



In a review of the agricultural chemistry of the Geological Sur- 

 veys of Kentucky and Arkansas, communicated to Silliman's Journal 

 (Sept., 1861), by Prof. S. W. Johnson, the author makes the follow- 

 ing common-sense observations on the general value of soil analyses 

 to agricultural science. He says : 



Years ago, following the teachers of agricultural chemistry in this 

 country and England, we believed that soil-analyses were adapted 

 to be of exceeding use to farmers. Having practised analytical chem- 

 istry sufficiently to undertake the work, we proceeded, when on a 

 vacation visit, to collect some farm soils for the purpose of applying 

 our skill and knowledge. On putting down the spade and post-auger 

 into the drift overlying the lowest Silurian of Northern New York, 

 we were at once struck with the difficulty of procuring an average 

 specimen. The soil, for a depth varying from two to six inches, 

 was quite fine, but below that depth largely mixed with gravel. On 

 comparing different samples taken from a small area, it was plain 

 that the soil was not a fit subject for analysis. The relative quanti- 

 ties of organic matter, as indicated by the color of the surface of 

 small stones some quartz and granitic, others slate and limestone 

 of several geological members were astonishingly variable. Here 

 we found the soil sandy, there it was clay. To take a sample from 

 one place was to do obvious injustice to the sixty-acre field. To take 

 it from a dozen places would not render the selection of a fair sample 

 any more certain. Then, as to depth, was it proper to go down six 

 inches, one foot, or how far ? Had the field been a bed of iron ore, 

 assays of a dozen samples taken from different parts would have indi- 

 cate'd very satisfactorily the general value of the deposit, would have 

 served as data for buying and selling the property, because the worth 

 of an un worked bed of such ore depends less upon its content of iron 

 than upon external circumstances which affect the extracting of the 

 metal. Had the field been covered with rich-dressed copper ore to 

 the depth of six inches, it would have been necessary to divide it up 

 into small parcels of a few tons, average these carefully, and as care- 

 fully assay each one. No one would risk purchasing a hundred thou- 

 sand tons of copper ore on the analysis of one or of a dozen samples, 

 because it is impracticable to intermix or average such a mass of ma- 

 terial as that a dozen samples shall accurately represent it. 



We hold it, therefore, as the first objection to soil analysis that to 

 procure a specimen which accurately and certainly represents a field 

 or district, is practically impossible in a majority of cases, and, if 

 possible, requires a series of analyses to prove the fact. This argu- 

 ment applies with the greater force when we consider how small a 

 proportion of the ingredients of a soil are of any immediate use in 

 feeding crops. The really active nutrient matters of a soil are not 



20 



