can be attached empty and then filled either by 

 suction or in the manner described below). 

 The mount with attached atmometer is then 

 turned right side up, the mercury resuming its 

 position in the crook of the J, the feed-tube is 

 lowered into the reservoir, and the stopper is 

 forced home. The filling of atmometer and 

 tube is completed by turning the whole appara- 

 tus upside down and then alternately back and 

 forth, in a manner which a little experimenta- 

 tion will readily suggest, due precautions 

 being taken during this operation that the 

 mercury retains its position in the side-arm. 

 This same simple expedient is adopted at any 

 time when it is desired to ascertain whether air 

 bubbles are present in the atmometer and for 

 getting rid of them. Any difficulty in pei*- 

 suading air bubbles to .emerge from the atmo- 

 meter into the feed-tube can be overcome by 

 tapping the base of the instrument or by vari- 

 ously shaking the whole apparatus. 



The Musch atmometer mounting has been 

 thoroughly tested under various conditions, 

 both by the writer and by others. It has 

 already been adopted by several investigators 

 for field work during the past season and 

 appears to have given uniformly satisfactory 

 results. The chief objection to which it may be 

 open seems to be this: that during a heavy or 

 protracted rain (or when left under a tap) 

 there may be an initial absorption of water 

 from without amounting to as much as 0.6 cc. 4 

 This absorption is due to the fact that the mer- 

 cury valve does not remain stationary before 

 the downward-pressing water column but re- 

 treats a short distance up into the short arm 

 of the J. During a period when the weather 

 at relatively brief intervals is alternately wet 

 and dry it is conceivable that an error of sev- 



4 The amount of absorption may be greater if 

 too short a column of mercury is used. 



