223 



to have the proper proportions of one metal to the other, and 

 besides this corrections are made from actual tests in different 

 degrees of temperature. 



The principal objection to this kind of compensation is that 

 metals expand and contract by infinitesimal waves or jumps, 

 probably owing to the molecular friction of the metals, and this 

 is most apparent in zinc, owing to its crystalline formation and 

 this metal is useless unless carefully drawn and prepared before 

 using for the purpose in question. 



THE MERCURIAL COMPENSATION. 



This pendulum is constructed in the following manner. A 

 steel rod of the calculated length and diameter is selected, and 

 at its lower end is firmly attached a brass stirrup into which is 

 placed and secured, from one to four glass jars containing mer- 

 cury. If one jar is used, the volume must be sufficient to allow 

 its cubical dilatation to raise the center of oscillation just as much 

 as the longitudinal dilatation of the rod has lowered this pointy and if 

 four jars are used, their diameters .shall be reduced to the point, 

 that the four will contain the volume of the one jar, and be filled 

 each to the same level as it rose in the single jar. This repre- 

 sents more exposed surface to the changing temperatures, and 

 improves the conductibility of the tnercury, causing the compensa- 

 tion to respond more promptly to sudden changes. The four jar 

 compensation is the most difficult to construct, but when well 

 made and carefully adjusted is exceedingly satisfactory, and has 

 the preference in seconds pendulums when greater accuracy is 

 required. 



THE SELSMIC ERROR. 



This uncontrollable error is caused by earth waves, and may 

 occur at any time. One peculiarity is, that many hours elapse 

 before this error shows in the time of the instrument. This 

 error may not be suspected until compared by transit observations. 



The time it takes to develop this error is probably due to 

 the molecular disturbances and re-arranging of particles that is 

 taking place in the mercury used for compensation. The most 

 accurately compensated pendulums have been known to vary 

 several seconds in a day. I remember while in Geneva in 1872 

 that twice in one summer the standard pendulum of the Cantonal 



