20 MEASUREMENT OF HIGH TEMPERATURES. 
stages of low percentage alloying is independent of the ingredients of 
the alloy, except in so far as they modify its electrical conductivity. 
In Chapter IV I develop a method for the direct and expeditious 
comparison of the ther mo couple with the air thermometer. A compar- 
ison of the data of Chapters II and IV gives me a criterion of the ac- 
curacy with which the data in the region of high temperature are known. 
This indirect method of arriving at such data is not apparently as rig- 
orous as their direct evaluation by means of the air thermometer; but 
the indirect method requires much smaller quantities of substance and 
may be conveniently extended to much higher temperatures. Taking 
all liabilities to error into consideration, its inferior accuracy is only 
apparent. 
The results recorded in these chapters will lead directly to the com- 
parison of the data to be obtained with porcelain air thermometers 
containing different gases, or one and the same gas in all states of 
tenuity. When methods to be indicated in the text are pursued these 
comparisons can be made with great accuracy, since the stem errors 
and the expansion errors practically vanish. It is upon such results 
that the rigorous validity of known high-temperature data must ulti- 
mately depend. From my results, moreover, it does not seem absolutely 
essential to glaze the bulbs within. It thus appears probable that 
bulbs can be made of lire-clay body by which the upper limit of direct 
temperature measurement (1,500°) may be materially increased. 
Finally, I propose in Chapter V a new method of pyrometry based 
on the viscous behavior of gases. Using the results of the earlier 
chapters, I endeavor to investigate the law of variation of gaseous 
viscosity and temperature. Having found that the said variation 
takes place nearly as the two-thirds power of absolute temperature, I 
proceed to indicate divers methods of utilizing this principle for prac- 
tical high temperature measurement. The results show, I think, that 
when the law of thermal variation of gaseous viscosity is rigorously 
known, Poiseuille-Meyer's equation applied to transpiration data will 
enable us to measure temperature absolutely, over a wider thermal 
range, and with a degree of precision and convenience unapproached by 
any known method. 
With regard to the contents of the present volume it is but just to 
remark that the work in all its essential parts was done either by Dr. 
Hallock and myself, or by myself alone without other assistance. The 
original practical construction of nearly all new apparatus, the design- 
ing and drawing of instruments, the extremely laborious computations 
which a task like the present involves, are our own work. If, there- 
fore, in the writing of the present memoir I have apparently placed 
superfluous stress on mechanical details, I offer in explanation the fact 
that the result of personal experiences is my subject. It is much to be 
regretted that the valuable researches of Messrs. Deville and Troost were 
not given to the world in more explicit form, for I have spent much 
(074) 
