AND THE CUBIC COMPRESSIBILITY OF MERCURY. 49 



because the pressure of 500 atmospheres is insignificant compared with 

 the probable internal pressure of the mercury ; but nevertheless one reason 

 for the use of the two metals was the fear that a slight change in the sur- 

 face tension of mercury might change the shape of the meniscus under 

 pressure. Obviously this would cause the opposite effect in the concave 

 meniscus in the copper cup from that caused in the convex meniscus in 

 the iron cup. As the total contraction of the iron rod over the same 

 range was 0.13 mm., the presence of the mercury involved a correction 

 cf 2.4 per cent. This is most conveniently applied by making another 

 addition to the value representing the length of the rod, which now 

 becomes 2.74 meters, instead of 2.68 meters. In the case of the copper 

 rod the actual length of the rod was 2.69 meters, and the corrected length 

 2.74 meters. 



It is conceivable that yet another very slight error might be caused by 

 the settling or lowering of the screw-diaphragm on which the rod rested, 

 because of the widening of the steel tube under pressure. The small inte- 

 rior diameter of the tube must make the error from this source very slight, 

 however ; and for this reason no correction was applied for it. The diffi- 

 culty could be wholly eliminated by cutting the thread in such a way that 

 the bearing surfaces would be horizontal radially. 



The micrometer by means of which the readings of changing length 

 were made was very carefully standardized by means of an accurate milli- 

 meter scale, whose corrections had been found by the National Bureau of 

 Standards at Washington. The scale was of course placed at exactly the 

 same distance from the telescope as the points to be observed during this 

 standardization. The successive measurements show a maximum devia- 

 tion from the mean of about 0.02 mm., in the case of iron, and somewhat 

 more in the case of copper. A smaller range of error would have been 

 desirable, but because we were unable to do better than this, it was neces- 

 sary to make many determinations. The deviations from exact constancy 

 were probably due partly to temperature, for in spite of much care it was 

 found very difficult to keep the temperature perfectly constant during an 

 experiment. 



From the linear compressibilities as thus determined, the cubic com- 

 pressibilities were found simply by multiplication by three. This simple 

 calculation involves the assumption that the rods were equally compressi- 

 ble in all directions an assumption which is not unreasonable in the case 

 of these pure elementary substances. Nevertheless, if a very high order 

 of accuracy is sought, especial pains should be taken to be certain that 

 the structure of the metal is identical in all directions. Probably heating 

 almost to fusion would accomplish this. In the present case this pre- 

 caution was deemed supererogatory. 



^ 



