o88 PROCEEDINGS OF THE AMERICAN ACADEMY 



Under circumstances similar to those described above, it will not be 

 safe to make the comparisons under about one hour after the change 

 of temperature. For a bronze bar having a cross section of one inch, 

 the writer has found that this limit of time is about five hours. This 

 form of the metres of the Society, therefore, seems to be well adapted 

 for ordinary use. But it should be stated that the small depth of the 

 bars W and Q^^ requires that they shall rest upon a flat surface during 

 comparison, — something not easy to obtain, 



111 the third test, the temperature of bar W was raised to about 

 28° C. by heating over a register. At 3h. 19m. p. m. it was removed 

 to the comparing room within which the temperature remained very 

 constant at about 19°.0 C. Comparisons were then made every three 

 or four minutes with the end-meter in melting ice, as follows. The 

 comparisons were begun when the thermometer placed upon the bar 

 ceased to rise. 



It will be noticed from these results, that the bar remained nearly 

 two minutes in the comparing-room before the change of length was 

 decidedly apparent. After that, the computed length was too great by 

 a maximum amount of 29 fi, and notwithstanding the fact that after 

 about fifteen minutes the thermometer readings decreased very slowly 

 and with considerable regularity, the indicated length of the meter was 

 at the expiration of 33 minutes still 1 2 /x too great. It is apparent, 

 therefore, that, under circumstances similar to those described above, it 

 will not be safe to make comparisons until the bar shall have remained 

 at a nearly constant temperature for at least one hour. 



The lines traced upon the silver surfiice are, in all the.se standards, 

 of the best quality. The edges are not rounded, and it is possible to 

 focus upon the lines with great sharpness. The writer has never 

 before seen heavy lines of as good quality as these. The width of the 

 lines in W, ^,. and ^., is about 20 /i, and in ^,j about 50 n- In Q^ 

 the width is slightly different at the two ends, the mean value being 

 about 2-') fi. 



The only serious criticism to be made upon these standards is, that 



