POTASSIUM,, RUBIDIUM, AND CESIUM. 17 



metallic sodium. For the later more exact work a more limpid oil selected 

 from the higher boiling fractions (150 to 250) of ordinary illuminating 

 oil (kerosene) was found to be more convenient. After several distilla- 

 tions over chips of sodium this oil was without action on sodium and po- 

 tassium when cold. We were never able to secure a sample wholly without 

 action on caesium ; but with care a sample could be prepared which had no 

 significant effect during the time of the compression experiments. Enough 

 oil was prepared at once to supply all that was needed in all the experi- 

 ments, so that its constants could be found once for all. 



MERCURY. 



It is very important that this substance should be pure, for obvious rea- 

 sons; therefore pains were taken to purify it carefully for the present 

 purpose. 



PRELIMINARY EXPERIMENTS ON THE COMPRESSIBILITY OF LITHIUM, 



SODIUM, AND POTASSIUM * 



At first the compression of sodium which had been solidified in a bulb- 

 tube or in the glass jacket itself was attempted, in the hope that the metal 

 was plastic enough to transmit the pressure to all parts. The internal 

 friction was nevertheless found to be too great for thoroughly satisfactory 

 compression, although it was not too great to prevent the invariable shat- 

 tering of the jacket or tube by the expansion of the compressed metal upon 

 relieving the pressure. Not many trials were needed to assure us that the 

 metal must be cast separately into a bar fitting snugly but not too tightly 

 into the glass jacket, and surrounded on all sides by oil. Of course, great 

 care was needed to prevent the access to the metal of the mercury in the 

 bulb-tube. 



It is not necessary to describe these experiments in detail, because their 

 numerical results are supplanted by the more accurate subsequent work to 

 be recounted. Two glass jackets were used one with a capacity of only 

 2 milliliters. In this latter tube the determinations of both lithium and 

 potassium were made, and considering the small volume under investiga- 

 tion, it is surprising that the results were as accurate as they afterwards 

 proved to be. The compressibility of lithium was found to be 0.000011, 

 instead of the true value 0.000009 ; that of sodium was found to be 

 0.000018, instead of the true value 0.000015; and that of potassium 

 0.000032, instead of the true value 0.000031. These errors are not great 



*These experiments were carried cut by F. Bonnet, jr., and constitute his con- 

 tribution to the present paper. 



