H. A. Ilazen — A o'-j;/'r.«x >/,■<> "/ Ui<jh stations. 107 



In order to show that varying temperature does not appreci- 

 ably affect the relative results of such computation, the follow- 

 ing comparisons are given. 



At a recent meeting of the Washington Philosophical Society, 

 Professor G-. K. Gilbert, of the United States Geological Survey. 

 gave a method for determining differences of elevation where 

 air-pre>suiv> un observed at three or more neigl boring stations 

 i lie rent heights, which may be outlined as follows: 

 Let P, P' and P" represent the observed air-pressure at stations 

 A, B and C, then will the formula, 



H=C (log P-log P0/(0/(O)/(m) (1) 



express the difference in height between A and B, and 



H'=C (log-P-log P")/(0/(O)/<™) (2) 



the same for A and C ; dividing (1) by (2) we have 





the 



other, without referring to temperature, latitude or l 



During the month of June, 1873, hourly observations were 

 made at the base and summit of Mount Washington and at two 

 intermediate stations. Considering forty-three cases taken at 

 random, I find twenty-three of them in which the mean height 

 of the summit above the base is too small, by Gilbert's 

 and (in vol's formula-, fortv-six mid tiftv-five feet respectively; 

 while the remaining twenty cases give a value too great by 

 iiltv-one and fort y -eight feet. We may therefore, in comparing 

 relative heights, "neglect the effect of varying temperature as 

 introduced by computations with Guyot's formula. 



If the wind affects the pressure directly, we would expect 

 that the computed difference of level would be the same as the 

 when there was no wind, and would gradually 

 ; wind increased, unless there were some causes 

 e, temperature and wind affecting the computa- 

 irouped these computed differences in elevation 

 according to the force of the wind, as may be seen in Table I. 

 In the following table, for May, 1872, all winds under ten and 

 above forty miles per hour are "included, and in May, 1873, all 

 the cases except a few which were omitted because of serious 

 errors in the observations. The table shows this rcu 

 peculiarity, that, though with winds above sixtv-one miles per 

 hour, the mean computed difference in height is too great by 

 sixty-six feet : with winds under ten the mean difference is too 

 small by thirty-five feet. We conclude, then, that some other 



t: 



