New York Agricultural Experiment Station. 389 
butter fats are practically constant for each individual cow under all 
circumstances of normal feeding. 
Must we then conclude that neither care of cow nor food can influence 
the quality of butter? Not at all; quality of butter depends more 
upon certain physical properties and flavors, which comprise scarcely 
more than traces of the butter, and which as yet have not even been 
identified by chemists, than upon the fixed oils which compose it. 
These principles, and also those which determine the color of butter, 
are unquestionably influenced by food and treatment. I believe, how- 
ever, that the first and most essential requisite in the manufacture of 
good butter is a butter cow. Without this one factor, success is impos- 
sible; with it and a skilled dairyman, success is certain and practically 
independent of the character of the ration fed, if sufficient nutriment 
is supplied. 
LYSIMETEE. 
Each year since the establishment of this Station, in 1882, records 
have been presented in the annual reports, of the monthly drainage 
water from each of the three lysimeters and of the chemical examina- 
tions made. Discussion of results has been deferred, hoping that 
sufficient data would reconcile the discrepancies which exist between 
the lysimeter results and what apparently takes place in outside soils. 
Increased data has not done this, and I am inclined to believe that no 
analogy exists by which evaporation, drainage and consequent loss of 
fertility in natural soils can be determined by lysimeters constructed 
in the ordinary way.* 
Before discussing our lysimeter results and the causes which are 
believed to render them unsatisfactory, I wish to consider the evapo- 
ration which takes place from large areas under natural conditions. 
For this purpose I assume that the difference between the rainfall 
upon a given water-shed, for a series of years, and the water which is 
discharged by the drainage streams, represents the evaporation. 
Unfortunately, there are but few water-sheds which have been studied 
with sufficient accuracy to be available for this purpose, but the dis- 
charge of streams which are depended upon for the water supply of 
cities have been quite closely determined. 
The data given in the following table are compiled from the reports 
of the Boston Water Board and from the Tenth Census of the United 
States, Vol. XVI : 
*See N. Y. Agr. Exp. Station report for 1882, p. 14. 
