300 PROCEEDINGS OF THE AMERICAN ACADEMY 



seems tolerably conclusive that bars of iron and zinc are liable to suffer 

 a change in their coefficients of expansion, either by an actual change 

 in their molecular structure or by the action of external causes. 



The continuous measurements of the Conservatory and of the Inter- 

 national Bureau will, in the course of the next decade, furnish data 

 which will add much to our knowledge of this subject. 



(h.) The laio which governs the expansion of bars of different materi- 

 als, and having different masses, under a varying temperature. 



It is well known that Mr. Sheepshanks, in the comparison of 

 " Bronze bar 28 with Cast-steel bar D," near the close of his labors on 

 the National Standards, found deviations which he could not explain. 

 The importance of this matter justifies me in quoting in full the state- 

 ment of Sir George Airy in his account of the construction of the 

 national standards. 



" I proceed now to allude to a discordance which was a source of 

 great anxiety to Mr. Sheepshanks. 



"In April, 1855, Mr. Sheepshanks was engaged in measuring the 

 bar Cast-steel D. By comparisons with four iron bars, (as stated in 

 the table above,) whose results agreed very closely, the excess of Cast- 

 steel D above Bronze 28 was found to be — 3 d . 61. But a direct com- 

 parison of Cast-steel D with Bronze 28 immediately preceding had 

 given — d .46. This comparison was made at the temperature 45°. 54, or 

 16°. 46 below the standard temperature. A trifling error of expansion 

 might account for part of the discordance, and the ordinary errors of ob- 

 servation might account for part. But in the opinion of Mr. Sheep- 

 shanks, though the whole discordance scarcely exceeded the effect of the 

 thermometric expausion of Bronze 28 for 0°.3 Fahrenheit, it was impos- 

 sible so to explain away the whole or a large part of it ; and he was 

 convinced that Bronze 28 had sensibly shortened. And so deeply and 

 so painfully was this impression fixed in his mind, that he actually con- 

 templated the rejection of all the results which had cost so many years 

 of labor, and the commencing the work de novo." 



Mr. Sheepshanks first disproved, by observation, the first conjecture 

 on the possible cause of the apparent change ; viz. " that Bronze 28, 

 still covered with gold-beater's skin and cement (as in the earlier com- 

 parisons), might have been so constrained by that covering that it 

 could not shrink down to its natural length; but that in the last com- 

 parisons with Cast-steel D, when that covering had been removed, it 

 had contracted itself." 



lie then compared Bronze 28 with twenty-seven different bronze 

 bars, and by comparing the old and the new measures found, with only 



