172 WATER AND FOREST. 



ing from 35 per cent, below to 16 per cent, above the average seasonal 

 rainfall of 16.56 inches for the entire 27 years. The extreme low points 

 of the Los Angeles precipitation curve are situated 22 years apart, 

 being 4.49 inches for the season 1876-77, and 5.6 inches for the pres- 

 ent season up to date. (June, 1899.) 



The fluctuation to which the average annual rainfall is subject 

 was very exhaustively considered by Mr. Binnie, M. Inst. C. E., in a 

 paper read before the Institution of Civil Engineers, London, in 1892. 



ijTom a close analysis of records of forty-two stations at various 

 parts of the world, covering periods of from 50 to 97 years, he drew 

 the following conclusions: 



That, for records of five years, the probable error in averages 

 ranged from minus 16 per cent, to plus 17.6 per cent., falling as the length 

 of the period was increased to about minus 2 per cent. and plus 2 per cent. 

 for periods of 30 years and more. And that: The least number of 

 years the records of which would give an average annual rainfall 

 that would not be materially altered by extending the record, would 

 be thirty-five years. He also concluded that dependence could be placed 

 on any good record of that duration to give an average rainfall correct 

 within 2 per cent. 



UNCERTAINTY OF RAINFALL RECORDS. 



These examples have been presented to illustrate the uncertainty 

 attached to any deduction based on rainfall records of short duration, 

 -■^s a case directly to the point, we have the French observations, made 

 about 50 to 75 miles south of Paris. The observations of one year 

 gave the precipitation over woods as 33 per cent, in excess of that over 

 open ground. Three years continuous observation changed this to 2 

 per cent. 



Long records for forest purposes are rare. This necessity for long 

 records is but one of the many obstacles in the way of arriving at ab- 

 solute comparisons of the relative precipitation over woods and open 

 ground. And after sufficient time shall have elapsed to reduce this 

 difficulty to a minimum, there will still remain the errors inherent 

 to measuring rainfall. There is great difficulty in obtaining two loca- 

 tions, the one covered with woods and the other bare, that are subject 

 to precisely the same conditions. Again, the rain gauges themselves 

 are not instruments of precision, and no conclusions can be more accu- 

 rate than the data upon which they are based. In scientific work it is 

 very necessary to fully realize and attribute a true value to data. Rain 



