DETERMINATION OP ALTITUDES BY BAROMETER. 



115 



of abnormal error is not balanced. Should one of the hourly observations bo wanting on any 

 day, a value, interpolated as correctly as possible by comparing the character of the curve 

 between the hours preceding and following it on other days, should be substituted. A moment's 

 consideration will show the necessity of this interpolation when there is any abnormal chango 

 from day to day. Still the great mistake of omitting it has often been made. 



The observations having been prepared as explained above, a mean of all the observations at 

 each hour is taken, and a curve plotted to represent these mean results. It should be a smootb 

 curve, generally with two maximum and two minimum points in the twenty-four hours, the exact 

 times of which vary somewhat. Should this curve not be smootb, some error of observation or 

 calculation has been made. It now only remains to find the mean reading for this mean day, 

 and to take the difference between it and each mean hourly reading, affecting the result with 

 the positive sign when the hourly reading is the less, and with the negative when it is the 

 greater. The correction from this table, applied with its sign to an observation taken at any 

 hour, eliminates the error due to horary oscillation. 



It may be well to remark, that it is a very good test of the value of a table of horary correc- 

 tions to apply it to the curve representing observations taken for a few days at a depot camp. 

 If a more sweeping line is produced, without a daily recurrence of any peculiar form, the table 

 may be considered good for observations taken in the vicinity, where the mean temperature is 

 about the same. 



From the observations taken on our survey, the following tables of horary corrections were 

 deduced. They proved to be well adapted to the peculiar characteristics of the different tracts 

 of country through which we passed. The manner in which they were computed is fully shown 

 in Appendix E. 



Corrections for Horary Oscillation. 



The curves on Plate XIII illustrate these tables. They represent the oscillation of the barometer as twenty times greater 

 than it actually is, in order to clearly show its character. 



Table No. 1 was deduced from six days' observations, taken in the latter part of July, at 

 Fort Reading, at an elevation between five and six hundred feet above the level of the sea, 

 and a mean temperature of 83° Fahr. The condition above stated was very well satisfied, and 

 the table was applied to the observations taken in the Sacramento valley, where the mean tem- 

 perature was very high. 



Table No. 2 was deduced from five days' observations, taken in the latter part of August, near 



