120 BRIDGMAN. 



which are purer than my former one, have a higher pressure coefficient. 

 It is evident that the sign of the effect of impurity on the pressure 

 coefficient cannot be predicted with probability, as can the effect 

 of impurity on the temperature coefficient. At the same time the 

 fact is to be emphasized, previously already found to hold in a number 

 of cases, that the effect of impurity on the pressure coefficient is usu- 

 ally much less than on the temperature coefficient. Thus in the pres- 

 ent case, a change in the temperature coefficient of 31% (present 

 compared with former work) is accompanied by a change of pressure 

 coefficient of only 9%. 



Lanthanum. This material I owe to the kindness of Professor 

 Charles Baskerville, who had prepared it from the fused salts by 

 electrolysis. No chemical analysis was available, but a spectroscopic 

 analysis by Professor F. A. Saunders showed a large amount of INIg 

 (possibly 10-20%) and a considerable amount of Si. There was a 

 trace of Ca, no Ba, and none of the other rare earth metals were 

 detected. The rare earths tested for were Ce, Pr, Nd, Er, Y, Yt, Dy, 

 Lu. A nodule about one gram in amount was available for the 

 measurements. A small homogeneous piece was cut from the nodule, 

 and extruded to wire in a small die of special construction. It is 

 necessary to heat to abovit 450° to extrude, and even then the extru- 

 sion is a matter of some difficulty. The wire so formed is exceedingly 

 stiff; it is evident that its elastic constants and its elastic limit are 

 both high. It is quite brittle, and can be bent only into a circle of 

 large radius. I prepared two pieces of wire, one at a somewhat higher 

 extrusion temperature than the other; the mechanical properties 

 seemed unaffected by the temperature of extrusion. The wire on 

 which measurements were made was only 1.7 cm. long, and was that 

 prepared at the lower extrusion temperature. In order to attach the 

 four terminals, spring clamps of special design had to be used; it is not 

 possible to solder this metal. The clamps gave some difficulty with 

 shifting of position, and the results were not so regular as usual. 



The temperature coefficient of this material between 0° and 100° 

 was only 0.001476. This is ^'ery low, and indicates that the material 

 was not very pure. For this reason it did not seem worth while to 

 spend a great deal of effort on the pressure measurements, although 

 these could have been improved by repeating the measurements 

 with a longer specimen, which was obtained after the easiest extrusion 

 temperature was discovered. 



Two runs were made for the pressure coefficient, at 0° and 50°. 

 The ascending points of the run at 0° were entirely regular, but the 



