PRESENT STATUS OF TEMPERATURE MEASUREMENTS 145 



are just published.^ As a result of these^ temperatures in the vicinity of 

 1000 degrees centigrade can now be determined within 1 degree, and the 

 probable error of temperature measurements does not exceed 2 degrees 

 until 1600 degrees is passed. It is possible to continue such measure- 

 ments beyond this point, but the uncertainty undoubtedly reaches 5 de- 

 grees at 1750 degrees, and perhaps 20 degrees or more beyond that. Such 

 temperatures (above 1600 degrees) are rarely encountered in the opera- 

 tions more immediately concerned with rock formation, although the 

 refractory oxides (CaO, MgO), when pure, melt far beyond 2000 degrees, 

 and the determination of their properties, both alone and in certain com- 

 binations, has required occasional measurements in the region above 2000 

 degrees, with which we have encountered no serious difficulty. In respect 

 of this absolutely indispensable factor in any quantitative study of min- 

 eral or rock formation, we are now assured of a scale of temperatures of 

 sufficient range and accuracy to meet the most exacting requirement likely 

 to arise. 



Interpretation of Melting-point Data 



On the other hand, the interpretation of melting temperatures when 

 obtained is a matter about which still hangs a cloud of obscurity. On 

 this subject the laboratory viewpoint has changed somewhat with in- 

 creasing experience, and the changes affect the whole question of the ex- 

 perimental study of rock formation as at present understood. 



In its first inception the original plan was to make a collection of typi- 

 cal minerals, as pure as possible, and to observe, on an appropriate ther- 

 mometer, the temperature at which they appeared to melt. In so far as 

 this procedure commended itself to investigators as simple, direct, and 

 free from any probability of misinterpretation, it has proved disappoint- 

 ing and misleading, for the data gathered in this way have differed so 

 widely in the hands of different observers, and even in the hands of the 

 same observer at different times, as to lead us to consider most of the 

 early observations uncertain and therefore untrustworthy. 



The experience of the laboratory student is perhaps better calculated to 

 provide an explanation of these differences than that of the mineralogist. 

 To make the case as concrete as possible, I will therefore take the liberty 

 to introduce some observations of these phenomena as they have been 

 gathered in the Geophysical Laboratory from time to time. 



s Arthur L. Day and J. K. Clement: American Journal of Science (4) 26, 1908, pp. 

 405-463 ; Arthur L. Day and Robert B. Sosman : American Journal of Science (4) 29, 

 1910, pp. 93-161. 



