457 



point temperatures found on May IS^'^ and May 28*^ may be 

 attributed entirely to A^ having been taken too large on May 18^'", 

 all the other temperatures must be recalculated, this recalculation 

 yields the crosses. Although this approximation is theoretically not 

 quite correct, as T — Lt and not 7' ought to rise at every temperature 

 in the same ratio, yet the results are in favour of the suggested 

 assumption. 



aoio 



0,005 



0,000 



6.5 7.0 



9.0 1 1,0 



Fig. 2. 

 • P&— 1919— I, \ 



O Isotoi^e Pb— 1919— I, { 18 May 1920. 

 iqij PÖ— 1919-J5, ) 



VV; Pè— 1919— 5, 28 May 1920. 



X Reduced observations : § 6. 



