486 Prof. C. Barus on the Secular Softening 



the same result, but this does not give any assurance as to its 

 correctness. Only the investigation of the values 0{t g — t), 

 to "too, C", K, gives us information as to lioio far our results 

 are reliable. As we see from the above, the correct establish- 

 ment of the equilibrium is the most important side of the 

 question, since otherwise we run easily into several thousandths 

 or even hundredths of a degree experimental error, with 

 methods which at first sight seem careful and accurate, and 

 even where we do not work with very high, or very low, 

 temperatures. This quite accounts for the discrepancies 

 which we find between the observations of different inves- 

 tigators for the same substances. 

 June] 897. 



LX. The Secular Softening of Cold Hard Steel, with Remarks 

 on Electrical Standards. Second Report. By Carl Barus*. 



^pWELVE years ago I began a series of experiments on 

 -*- the gradual softening of the temper of glass-hard steel 

 in the lapse of time, at ordinary atmospheric temperatures. 

 The results which I published f in 1888 covered an interval 

 of but three years, so that with the present data I am able to 

 add materially to the inferences previously drawn. The work 

 was originally undertaken with a view to its bearing on the 

 construction of perm;inent magnets. Prof. V. Strouhal and 

 I had shown J that the breakdown of chemical configurations 

 during the tempering of hard steel is accompanied by an 

 accentuated breakdown of the magnetic configurations which 

 may be present. In spite of its high coercitive force, glass- 

 hard steel is therefore unsuitable for use when (as in all instru- 

 ments) the magnets are to preserve exceptional permanence 

 in the lapse of time. I was induced to resume the present 

 measurements in consequence of certain discouraging results 

 mentioned in a paper by Profs. J. Viriamu Jones and W. E. 

 Ayrton§, in which the discrepancies encountered from the 

 molecular changes of resistance-standards were serious enough 

 to vitiate the results of refined measurement. It appeared 

 from this that the laws subject to which the temper of any 

 metal (whether imparted by sudden cooling or mechanically) 

 changes with time and temperature are not generally known. 



* Presented to the U. S. National Academy of Sciences. Communi- 

 cated by the Author. 



t This Magazine, [5] xxvi. p. 397 (1888). 



% Bulletin U. S. Geological Survey, No. 14, p. 151 et seq. (1885) ; or 

 Wied. Ann. xx. p. 662 (1883). 



§ Paper read at the Toronto Meeting of the British Association, 1897, 

 on the determination of the ohm &c. 



