GEOPHYSICAL RE;SEARCH — DAY. 1 85 



The second (2) cause of inhomogeneity arises from superficial oxidation of 

 the wires. This can usually be prevented in various ways, or corrected by 

 scouring. 



The error, after these precautions have been observed, is less than 0.002° 

 up to 40° C. These thermometers are, therefore, as sensitive as the finest 

 mercury thermometers ; they have a much greater range, and are more cer- 

 tain in their behavior. Still greater sensitiveness could be attained by using 

 several elements in series, if sufficient galvanometer sensitiveness were 

 available. 



Platinum elements at high temperatures take up impurities which diffuse 

 into the wire. Two sources of contamination are also found here : ( i ) The 

 action of reducing agents (carbon, illuminating gas, etc.) upon the material 

 of the furnace which sets free substances which combine with platinum 

 (usually iron and silicon). Neither carbon itself nor the common gases 

 containing it affect the elements. The second cause of contamination (2) 

 appears in an oxidizing atmosphere, and has frequently been overlooked 

 hitherto. It is due to the presence in the furnace with the element of vapors 

 of the other platinum metals at temperatures above 900°. These (chiefly 

 iridium and rhodium), when alone or alloyed with platinum, will penetrate 

 both wires, and will cause low readings by diminishing the thermoelectric 

 potential difference after a very short exposure. Glazed porcelain is a suffi- 

 cient protection against these vapors, but not unglazed porcelain. When a 

 wire is once contaminated with iridium or rhodium, there is no way to restore 

 it except by cutting off the contaminated portion. The distribution of the 

 contamination, as well as its amount, can be readily determined by a thermo- 

 electric test against a piece of pure wire. A contaminated element can still 

 be used for accurate work, provided it can be compared with a perfect ele- 

 ment immediately afterward without changing the temperature gradient along 

 the wire. 



By observing the precautions above indicated, a platinum-platin-rhodium 

 thermoelement was found to show no variations greater than one-twentieth 

 of 1° at 1,000° — a remarkably high degree of perfection. 



