216 TABLES AND RESULTS OF THE PRECIPITATION, 



to year; in the New England States and along the sea-board as far south as 

 Washington, the probable uncertainty in any cme annual amount is 12 per cent., 

 and the same liolds good for the Ohio and eastern part of the Mississippi Valleys, 

 and in the region of the Lakes ; along our southern Atlantic coast, and along the 

 Gulf coast, the probable error is 17 per cent; west of the Mississippi it is 23 per 

 cent. ; in California, Oregon, and Washington Territory, it is 19 per cent. ; and at 

 Sitka, Alaska, but 8 per cent, of the annual amount. 



The figures in the last column of the above table vary from about + 1 to about 

 ± 2^ inches, and on the average indicate a probable error of +1| inch in the re- 

 sulting annual amount from tlic longer series. In the construction of the annual 

 hyetal curves on the accompanying chart, results from series as short as four years 

 were admitted, and upon the average these curves must be considered as being 

 liable to a probable error of ± 2 or 3 inches, less or more, according to location, as 



indicated by the ratio - . 

 R' 



The probable uncertainty of the hyetal curves of the extreme seasons may be 

 estimated at J of that given above. 



Returning to the consideration of a secular change in the annual amounts, we 

 might, for limited localities where the rain-fall is nearly the same, examine for 

 every year the difference from the mean amount ; but for our purpose it becomes 

 necessary to employ the ratio of each annual to the mean amount in order to ren- 

 der the results in different localities strictly comparable. 



The following table contains these ratios for a number of stations selected on 

 account of their extent of record. 



