POTASSIUM, RUBIDIUM, AND CESIUM. 



19 



In order to find the values corresponding to every hundred atmospheres, 

 the observed data were plotted with great care on a large sheet of accu- 

 rate coordinate paper by means of a spline, heavily weighted at each point 

 (see fig. 4). A much reduced copy of these curves is given in fig. 4 on 

 page 21. From these curves the following figures were found : 



Derived -values concerning compressibility of mercury and oil. 



Another less complete trial with the same amount of oil gave the final 

 tesult 3.100 instead of 3.096 (a difference of only 0.13 per cent), and yet 

 another with a different amount of oil gave upon calculation over to the 

 same basis by simple proportion the value 3.090. There can be no doubt, 

 therefore, that the trial tabulated above represents very nearly the com- 

 pressibility of the oil in question. 



The data needed for proceeding with the metals having thus 

 been found, the first two of these were prepared in the follow- 

 ing manner : The bars of lithium and sodium were hammered, 

 cut under oil into the shape needed to fit into the jacket, washed 

 with several successive portions of pure naphtha, warmed in a 

 current of hydrogen, and transferred to a weighed tube con- 

 taining paraffin oil. The tube was then reweighed, the differ- 

 ence being the weight of the metal. 



Potassium, being in smaller pieces, was first fused for one 

 determination under oil, and for another in hydrogen and 

 then cast into suitable bars. Subsequently, it was treated 

 exactly as the lithium and sodium. 



On the other hand, the rubidium alloy, being liquid, needed 

 different treatment. By alternately exhausting and admitting 

 gas pressure into the apparatus indicated in fig. 3, the small 

 inside weighed tube was filled with the alloy and oil. Another 

 weighing gave the weight of oil and alloy together ; and after the com- 

 pression experiments the alloy was for the most part separated by a hydro- 

 gen-filled weighing-pipette, the small remainder being dissolved and deter- 

 mined by titration. The somewhat numerous figures need not be given; 

 the essential outcome is recorded in the table on the following page. 



Fig. 3. 



