54 Colorado College Studies. 



fore in the wide part of the fixed tube and more of it in the 

 cistern. Hence, a rise in temperature of the mercury tends to 

 make the instrument read too low. With the aid of somewhat 

 elaborate equations the writer has calculated that for a rise 

 of 45 deijjrees Fahr. in the mercury (at constant atmospheric 

 pressure) the cistern will fall about .174 inches. According 

 to the figures in the S. A. S. No. 209, the springs will stretch 

 about .03 inches, making an aggregate depression of .204 

 inches for 45 degrees, and thereby causing the atmospheric 

 pressure to appear .048 inches lower than it actually is. An 

 experiment with the barograph at Colorado College caused 

 the cistern to fall .211 inches for a rise of 45 degrees Fahr. 

 The experimental and the calculated figures are in fairly close 

 agreement. 



The conclusion then is that Draper's Barograph reads 

 too low for temperatures above the standard temperature and 

 too high for temperatures below the standard. It is not suf- 

 ficient to make corrections for the stretching of the springs, 

 for the errors arising from variations of temperature in the 

 mercury are over five times greater. 



It has been the practice to compare the barograph, from 

 time to time, with a standard barometer, but the results are 

 not satisfactory. Suppose comparisons are made at 12 m. It 

 is easy to adjust the barograph so that it will indicate very 

 nearly the same atmospheric pressure as does the barometer 

 at noon, except on days when the temperature varies widely 

 from the average noon temperature. But what about the 

 barograph readings in the morning before sunrise, when the 

 temperature is much lower? Will the barograph then not in- 

 dicate a pressure altogether too high? Suppose the tempera- 

 ture in the early morning be 10 degrees less in the barograph 

 case than it is at noon, then if the barograph reads correctly 

 at noon, it will read nearly .02 of an inch too high in the 

 morning. This is a serious error. If tables be constructed 

 for temperature corrections in Draper's Barograph, it will 

 not be difficult to use the barograph independently of a 

 standard barometer and to secure results of much greater 

 accuracy than those obtained at present. 



