310 Messrs. Haldane and Pembrey on an Improved Method 



reading of the thermometer taken. It is convenient to 

 graduate the bottles roughly by pasting strips of paper at 

 intervals of half a litre up the uncovered strip of glass. 

 With the help of this graduation it is easy to ascertain if the 

 aspirator is running at about the proper rate. 



To graduate the aspirator, one of the bottles is allowed to 

 drain for a mriute, and is then weighed accurately. The two 

 bottles are now connected and the weighed bottle is filled 

 with distilled water to the mark, the tubing also being 

 filled with water. The cork of the weighed bottle is then 

 removed, care being taken to let no water escape from the 

 end of the long piece of glass tubing. The bottle filled with 

 water is then weighed again ; and by deducting this weight 

 from that of the empty bottle and allowing for air displaced 

 &c, as in the graduation of a measuring-flask, the amount of 

 water held by the bottle filled to the mark with the tubing in 

 it, is obtained. The difference between this amount and some 

 round number of cub. cent., such as 3000, is then measured 

 into the bottle, which has been previously emptied and 

 drained for a minute. The cork is then replaced, care being- 

 taken, as before, not to spill any of the water in the tubing. 

 The aspirator is now ready for use, and measures off with the 

 utmost exactness 3000 cub. cent, of air each time it is 

 reversed. 



We are now in a position to consider the advantages of the 

 method as thus modified. As regards, firstly, errors of weigh- 

 ing, we found, in accordance with previous observers, that 

 an absorption-tube, when left to itself, varies considerably in 

 weight from hour to hour. The same is true of a tube 

 through which dry air from another absorption-tube is passed*. 

 There is usually a gain in weight, but sometimes a loss. This 

 may be due partly to variations in temperature and barometric 

 pressure. It is also due to gradual penetration of moisture 

 into the tubes and to the varying amount of moisture de- 

 posited on the surface of the glass. The amount of this 

 variation in weight was often two milligrammes or more, 

 both in Shaw's experiments and our earlier ones with other 

 absorption-tubes. With the method just described the tubes 

 did not vary in apparent weight by more than three deci- 

 milligrammes in a day ; and these slight variations were in 

 part accounted for by small imperfections in our set of weights. 



The advantages of using a counterpoise are illustrated by 

 an experiment in which weighings with the counterpoise 

 were compared at intervals with ordinary weighings durino- 



* Cf. the data on this point given by Shaw, he. cit. p. 84. 



