THE CISTERN BAROMETER. 97 



tube. Do this repeatedly, if necessary , until the mercury appears perfectly 

 clear of bubbles. Then fill the tube with mercury, drawing- out with a straw 

 any bubbles that may then be near the top. Invert the tube in the case, put 

 on the glass ring- and the upper cone of the cistern, and screw them together. 

 Then fill the cistern with mercury, put on the lower cone, with the bag and 

 the brass cover, and the work is complete. The test of a satisfactory result 

 is the sound made by the column of mercury as it strikes the top of the tube. 

 If there is a sharp metallic click the vacuum is good, but if the sound is 

 muffled the vacuum must be improved. It is well to warm the mercury 

 before pouring it into the barometer, in order to drive out any moisture in 

 it. This is especially advisable if the atmosphere is damp at the time. 



It is by some thought advisable to boil the mercury in the tube during 

 the operation of filling. This is usually done over an alcohol lamp, two or 

 three inches of mercury being poured into the tube at a time and brought 

 to a boil until the tube is filled. The mercury which is to be poured into 

 the cistern is then also boiled. This is a very delicate and tedious operation, 

 and is attended with much risk to the tubes. Its utility is questionable, 

 inasmuch as the mercury in the barometer is exposed to the atmosphere and 

 soon contains as much moisture as before. 



It often becomes necessary to clean the surface of the mercury in the 

 cistern. To do this, take off the lower cone of the cistern ; then, placing 

 the finger, protected by a piece of kid, over the open end of the tube, 

 invert the barometer slowly and pour out the mercury from the cistern. 

 Strain it through chamois skin, replace it in the cistern, and put the latter 

 together again. 



Observations at the base stations should, whenever practicable, be 

 made hourly from 7 a. m. to 9 p. m., in order to insure having base obser- 

 vations coincident with those taken in the field. When not practicable to 

 do this, they should be made at 7 a. m., 2, 6, and 9 p. m. Each observation 

 should include the reading of the attached and detached thermometers. 

 Whenever the observations at a station of the U. S. Weather Bureau are 

 available, they may be used as base records. In most cases, howevei, these 

 observations are made with barometers reading only to one-hundredth of 

 an inch, but, upon proper application, the Weather Bureau has in all cases 

 MON xxii 7 



