( 343 ) 



the cooling. Indeed this former mean value is 17 /< larger than the latter. 



Now this difference of 17 p, which refers to a bar of platinum 

 of 840 mm. (for a bar of 1 M. it would be 20 u) exceeds the 

 errors which may be ascribed to the inaccuracy of the observation 

 by about half the difference which exists between Scheel's formula 

 and our formula of June 1906. 



As basis for the calculation of our formula the mean \ of the 

 two lengths has been taken. We arrive at values for the expansion 

 nearer lo those of Scheel when for the length at the ordinary tem- 

 perature we take that which was found immediately after cooling. 

 instead of the mean of this length and the length which was found 

 long before and after the cooling, as was done in the calculation 

 of our formula of June 1907. If we now make use of the first- 

 mentioned length, that which was found immediately after cooling, 

 in order to find the coefficients now distinguished by (a) and (b) 

 from the former a and b in the formula : 



""(ifo)'! 10 -') 



Platinum 



(— 183° to + 16°) 



(a) 877.7. Kamerlingh Onnes 



(b) 35.7* and Clay (1905) 



whereas 

 (b) ?7.o! SCHEEL (19 ° 6) 



It is true that the now remaining difference of 34 jx per M. with 

 an expansion of — 183'' to -f- 16° remains considerably larger than 

 the accuracy of the observations would lead us to expect, but it is 

 considerably smaller than that found originally, and taking into 

 consideration the different sources of uncertainty whether we observe 

 really what we think we observe, the small number of measurements, 

 and the difference of the methods applied at low temperatures for 

 the first time, it is not great. 



We had hoped to obtain further information on the difference in 

 length of our bar at ordinary temperature immediately after the 

 cooling and long after it, but have not yet been able to do so. 



Differences as the one discussed now have more occurred in our 

 measurements. We have pointed this out in Comm. N°. 95' J and 



') In the calculations for the glass the values of the length immediately after 

 the cooling, Dec. 23 in Table I, and April 15 and l(i in Table 111, have lien lefl 

 out of account in connection with the further observations. 



