INSPECTION OF FERTILIZERS. 



The inspection of fertilizers by the Maine Experiment Station 

 for the year 1891 has required the analysis of sixty-four brands 

 of mixed fertilizers and two brands of bone, involving the selec- 

 tion of one hundred and thirty-five samples. The number of 

 brands inspected in 1890 was sixty-four. In order to secure 

 these samples, it has been necessary to send an agent into nearly 

 every section of the State, and during the present season he has 

 occupied the equivalent of the entire time of a month in constant 

 travel. Samples were taken in thirty-two cities and townships, 

 often at several points in the same township. An effort is made 

 to begin this work in March, but it is usually found that new goods 

 have been shipped to but few places so that a successful canvass 

 of the State can not be carried on until April, therefore the com- 

 pletion of the sampling and analytical work can not be reached 

 until late in June. It will be noticed that three samples of each 

 brand have not been secured in all cases. 



In general this has been owing to the following causes • Selling 

 of the fertilizer at but very few points, and finding onl} 7 the goods 

 held over from last year's sales in the hands of nearly all the 

 agents visited. 



SELECTION OF SAMPLES. 



Samples for 1891 were selected by Mr. W. J. Holden, an 

 agent of the Station, acting under its instructions. The sam- 

 ples were drawn from three or four packages, mostly one hundred 

 pound bags, so that in all cases where three samples were taken 

 the analysis represents from nine to twelve packages of the 

 goods. 



The drawing of the samples is accomplished by means of a 

 sampling tube which can be made to reach every portion of the 

 package, and as several drafts are made from each package, it is 

 readily seen that the method of taking samples is a very thorough 

 one, and there is no good reason for supposing that the contents 

 of the glass jar that is forwarded to the Station does not fairly 

 represent the goods sampled. In every instance a sample exact ly 



