The Thermal Expansion of Pure Nickel and Cobalt. 



311 



The agreement of the individual values is not quite so good as in the 

 ease of nickel, owing to the impossibility of obtaining such absolute 

 perfection of the surfaces of the cobalt block as was obtained in the 

 case of the nickel block. In the case of the constant a the differences 

 only amount to 3 per cent., and the whole amount of /; is so minute that 

 one is fortunate in finding the agreement so good. These differences, 

 however, from the nature of their cause, are bound to be on both sides 

 of the truth, and the mean of so large a number as nine is sure to be 

 very near the true value. 



It will now be interesting to compare these results with those of 

 Fizeau. The latter were published in very brief form in the £ Comptes 

 Eendus,' for 1869* and also in ' Poggendorff 's Annalen,' for the same 

 year.f In neither of these publications are any further details given 

 beyond the values of the coefficient of expansion for 40° and the incre- 

 ment per degree, (= 2b), which occur in a table of similar- 

 quantities for various metals ; together with the information that the 

 specimens of nickel and cobalt employed had been reduced by hydrogen 

 and compressed, and that the range of temperature of the observations 

 was from 10° to 80°. The values in question are — 



a 40°. Aa/Atf. 



Nickel 0-000 012 79 071 



Cobalt 0-000 012 36 0'80 



It will be observed that the values of the coefficient for 40° now 

 presented are higher than those of Fizeau ; in the case of nickel the 

 difference is 1307 - 1279 = 0028, and in the case of cobalt 1259 - 1236 

 = 0023. The author's increments are likewise higher, 148 and 128 

 against 71 and 80 respectively. The fact that the author's increment 

 for nickel is twice as great as Fizeau's might suggest the possibility of 

 a mistake between b and 2b. The author has certainly not made any 

 such mistake, for the mode of calculation employed yields b directly, 

 and the values afforded were 74 and 64 respectively. The increment 

 Aa/A0 (Fizeau's 6 being the author's t) is equally certainly 2b. More- 

 over, the author's value of the increment for aluminium, 2*12, calcu- 

 lated in precisely the same manner, agrees fairly with the value given 

 by Fizeau, 2*29, in the same table in which the values for nickel and 

 cobalt are published. It may be that Fizeau inadvertently gave the 

 value of b instead of 2b in the particular cases of nickel and cobalt, but 

 it is much more likely that the numbers are correctly given, and that his 

 results were not very concordant with those now given. For Fizeau could 

 certainly not have possessed specimens of nickel and cobalt of the same 

 degree of purity as those supplied to the author by Professor Tilde n. The 

 recent discovery of nickel carbonyl has afforded an incomparable means 



* Vol. 68, p. 1125. 

 f Vol. 138, p. 30. 



