132 EFFECTS OF WINDS AND OF 



A concrete idea of the possible errors inherent in the present determina- 

 tions of mean sea-level and their relation to the errors of precise leveling 

 may be secured as follows: Note that at the bottom of tables Nos. 26 and 

 27 the mean elevation as observed for a whole season agrees within 0.03 foot 

 with the corrected elevation for the season in every case except at Buffalo. 

 At Buffalo, however, the mean of the observed elevations for the three 

 months of 1909 differs by 0.11 foot from the mean of the corrected elevations, 

 and for the 5 months of 1910 the difference is 0.10 foot. In other words, 

 the omission of the wind and barometric corrections at Buffalo leaves an 

 avoidable error of 0.10 foot in the derived elevation of the water surface 

 from the 5 months of observation. 



The probable error of a difference of elevation of two points, determined 

 by the precise leveling of the Coast and Geodetic Survey, is about 0.7 

 millimeter into the square root of the distance between the two points in 

 kilometers measured along the line of leveling. In other words, an error of 

 0.03 foot =9.1 millimeters has an even chance of occurring in 169 kilometers 

 = 105 miles of leveling from a tide gage. The error of 0.10 foot referred to 

 above in connection with the 5 months observation at Buffalo has an even 

 chance of occurring in about 1,800 kilometers = 1,100 miles of precise level- 

 ing. This statement neglects the possible small systematic errors in the 

 leveling and also the strengthening of the level line by its connection with a 

 net of level lines. 



Judging from a study of the charts and his present knowledge of the laws 

 controlling wind effects, the writer estimates that the wind effects are as 

 persistently large and of one sign at Cedar Keys, Florida, as at Buffalo. 

 One of the gages at which mean sea-level was determined for use in con- 

 nection with precise leveling was at Cedar Keys. Similarly, the writer 

 estimates that at Sandy Hook, New Jersey, near New York City, the chance 

 of error due to the omission of corrections for wind effects and barometric 

 effects is not much less than that at Buffalo and is certainly greater than that 

 at the other four stations of tables Nos. 26 and 27. Hence, even though 

 mean sea-level is determined from an average of several years of observa- 

 tion at Sandy Hook, it is possibly in error by an amount exceeding the error 

 accumulated in the precise leveling over hundreds of miles. It should not 

 be overlooked in this connection that the prevailing winds and the prevail- 

 ing barometric gradients tend to be seasonal, to be repeated each year, and 

 that therefore the taking of a mean for several years is of only moderate 

 effectiveness in reducing the error in the mean. The monthly values of mean 

 sea-level at various tide gages support the statement by showing a seasonal 

 variation, as a rule, and thereby incidentally indicating that the wind 

 effects and barometric effects are certainly decidedly appreciable in the 

 monthly means. 



