SOLID ELEMENTS, AND THEIR PERIODIC RELATIONS. 59 



the weights of added mercury reduced in each case to 100 grams of mer- 

 cury contents of each jacket, over the range of 400 kg. per square centi- 

 meter just mentioned. It will be noted that the jackets v and vi, made of 

 the same piece of glass tube, agree exactly. The respective extreme val- 

 ues correspond to compressibilities of glass equal to 0.0000021 in the case 

 of jacket in, and 0.0000023 in the case of jacket v. Similar differences 

 in the compressibility of different kinds of glass were noted by Amagat. 



The column headed IV contains the outcome of these measurements 

 needed for subsequent work with the respective jackets. 



In the mercury or in water above the mercury held by these jackets, the 

 metals to be studied were one by one immersed with all the precautions 

 discussed in previous papers. Amalgamable metals were never allowed 

 to touch the mercury ; and even when the mercury entirely surrounded the 

 cylinder of a metal not capable of amalgamation, a little supplementary 

 liquid was needed to fill accidental fine cavities from which mercury was 

 excluded by its surface tension. The weight of this water present was of 

 course always carefully found, as its compression was usually larger than 

 that of the metal to be studied. In a few cases, indicated by foot-notes in 

 the table on pp. 61 and 62, paraffin oil was used instead of water. 



CALCULATION OF RESULTS. 



The general treatment of the calculation of the results has been suffi- 

 ciently discussed in the preceding papers. The equation needed when 

 water is used as the supplementary liquid is described in the paper on the 

 non-metals,* and the similar one when oil is used in the paper on the alkali 

 metals.f 



In case paraffin oil is used, its appropriate coefficient for K (the weight 

 of the supplementary liquid) must be specially determined. Two sam- 

 ples of oil were used in the following work. One possessed a compressi- 

 bility corresponding to a coefficient for K of 0.4387, being the sample 

 which had previously been used for the alkali metals. The other was less 

 compressible, for 7.082 grams in jacket vin required only 0.760 gram of 

 added mercury at 148 units of pressure, 2.491 at 387 units, and 3.371 at 

 526 units, starting with no added mercury at 50 atmospheres. These fig- 

 ures correspond to a coefficient for K of 0.3830. The first sample of oil 

 was used with magnesium and calcium and the more dense specimen of 

 manganese ; the other sample was used with the less dense sample of the 

 last-named metal. 



Two cases may be given in detail, as typical examples of the method, 

 one in which paraffin oil was used to surround the metal, and another in 



*Loc. cit, p. 36. fLoc. cit, p. 22. 



