Wave-length of Light. 261 



ments have tended to somewhat lessen this discrepancy, 

 it seems to be sufficient, considering the fact that " No. 11 " 

 and the Imperial Yard are of the same shape, material, and 

 mass, and were compared on the same apparatus as during 

 the original comparisons in 1857, and at nearly the standard 

 temperature, to establish the fact of a real change. While 3 

 or 4jx is absolutely a small quantity, its systematic appear- 

 ance under conditions almost identical with those of the 

 original measurement can hardly be ascribed to experimental 

 errors. The other cases cited in the above-mentioned paper 

 tend to confirm this conclusion. 



The gradual and sometimes very irregular changes that are 

 known to take place in both the bulbs and stems of thermo- 

 meters would lead one to expect that glass standards of length 

 would be liable to similar changes, though probably far less 

 in amount. It was, therefore, with special interest that I 

 examined glass Decimetres III. and IV. belonging to the 

 Coast Survey, and used by Peirce in his wave-length measure- 

 ments. These scales are on plate glass, of the same dimen- 

 sions, and having coefficients of expansion not widely different. 

 A series of comparisons made at a nearly constant temperature 

 of 16° # 5 0. gave the direct relation 



III. = IV. + 2-l At . 



While the same relation deduced from Peirce's direct com- 

 parison, by applying the coefficients of expansion assigned by 

 him, is 



III. = IV. + P3^. 



The defining-lines on both standards are fine and sharp; and 

 unless Peirce's coefficients are grossly in error, the evidence 

 of change between 1879 and 1887 is very strong indeed. 



Having now the exact present relation of S a 2 to the original 

 standard R 2 , it remained only to investigate the difference 

 between this result and the length of S a 2 as deduced from the 

 Berlin comparisons. I have been unable to obtain the details 

 concerning E 78 , the standard used in these comparisons, but 

 it was determined by comparison with the standard metre of 

 the International Bureau. The comparisons of S a 2 with R 78 

 were carefully made by two observers, and it is probable that 

 the result represents the relation between these standards with 

 considerable exactness. It should, however, be borne in mind 

 that the microscopes had each a power of only 50 diameters, 

 and that the bars in question are of very different material and 

 mass, thus giving a chance for small errors due to varying 

 temperature. 



Phil. Mag. S. 5. Vol. 25. No. 155. April 1888. T 



