FISHERIES, GAME AND FORESTS. 509 



of Civil Engineers of England in 1892, and which may be found in Volume CIX. of 

 the proceedings of that Institution. 



In this paper Mr. Binnie takes up a number of long rainfall records and by- 

 analysis of them arrives at the conclusion that a rainfall record only five years in 

 length may be subject to a possible error of 32 per cent. ; a 10 years' record to an 

 error of 7.7 per cent. ; a 25 years' record to an error of 5.8 per cent. ; a 35 years' 

 record to an error of 3.5 per cent. ; a 40 years' record to an error of 2.9 per cent. ; a 

 45 years' record to an error of 2.6 per cent., and a 50 years' record to an error of 2.5 

 per cent. Hence, Mr. Binnie concludes that a carefully kept rainfall record thirty-five 

 years in length is as good a record for a longer period, and that safe conclusions may 

 be based upon such a record within the limit of an error of about 2.5 per cent. It is 

 carefully worked out conclusions of this character which indicate the uncertainty of 

 basing anything final on records of two or three years, and a thorough appreciation of 

 the importance of this proposition has led me to question the validity of the conclu- 

 sions of Messrs. Roberts and Vermeule. I doubt their conclusions, because analysis 

 shows that neither has used data of sufficient extent to indicate safe conclusions. 



In regard to the run-ofT of the Upper Hudson River, we have a fairly complete 

 record from 1888 to 1896, inclusive. By way of comparison of run-ofT with temper- 

 ature and precipitation' I have tabulated in tiie manner already detailed for the Sud- 

 bury, Cochituate and Mystic areas the mean temperature, precipitation, run-ofT and 

 evaporation for the years 1891 to 1896, inclusive, using the temperature and precipi- 

 tation of the Northern Plateau as given by the State Weather Service in its monthly 

 publication. In 1 891 the total rainfall was 42.96 inches, the total run-off 26.56, and 

 the evaporation 22.4 inches. In 1892 the total rainfall was 53.37, the run-ofT 

 33.08, and the evaporation 20.79 inches. A comparison of the two years 1891-92 as 

 regards the evaporation is exceedingly interesting. The rainfall in 1892 was 10.91 

 inches, while the evaporation was only 0.39 greater than in 1891 ; that is to say, as 

 already pointed out in discussing the Massachusetts data, the evaporation is persistent 

 at about the same figure whatever the rainfall may be. For instance, as illustrating 

 this position further we may take the year 1893 when the total rainfall dropped to 

 42.18 inches or within 0.78 inches of what it was in 1891, whereupon the evaporation 

 showed 20.27 inches or practically the same figure as in 1891. In 1894, however, 

 somewhat different meteorological conditions prevailed, and with a rainfall of 41.37 

 inches the evaporation rose to 22.00 inches. In 1896, when the rainfall rose to 45.21 

 inches, the evaporation was 22.33 inches. The mean evaporation of the six years 

 covered is 21.14 inches. 



We may say, then, as a tentative conclusion subject to correction up to the period 

 when we shall have at hand about thirty-five years' data, that the evaporation on the 



