22 PROCEEDINGS OP THE AMERICAN ACADEMY 



deduced by me from Obermajer's experiments at the three indicated 

 temperatures with the capiUary " D." This is in still closer agreement 

 with mine in numerical values of V( : Vt but has slightly less curva- 

 ture {B = 0.00000105). The value for A in my results is 0.003637, 

 with the same terms in the equation. When methods as different in 

 detail as those of Obermayer and myself are in so close agreement as 

 these appear to be when thus discussed, the chances of large constant 

 errors other than those inherent in the use of transpiration methods 

 seem to be much reduced. 



On the line JH are three points showing the results obtained by 

 Eilhard Wiedemann. The line is drawn from the equation 



^^ = ^. (1 _|_ 0.003727 t — 0.00000320 t""), 



which I have deduced from these three points. The curvature is 

 much greater thau for either of the other lines, the value of B being 

 more than twice as great as in the lines OA and OB, and three times 

 as great as in 0. A reference to my discussion of the results for 

 air will show that Wiedemann's results there exhibit similar relative 

 characteristics. Up to 100= for CO^ the results differ numerically by 

 less than 0.7 per cent from those of Obermayer and myself, but at 

 200" the difference is nearly 3 per cent. 



The results by Puluj, as shown by the line OF, obtained by oscil- 

 lating plates, are smaller than all others, a difference which character- 

 izes his results for air. The range of temperature used is too limited 

 to render the results available in the present consideration. 



Conclusion. — In the absence of any really satisfactory means of 

 assigning proper weights to the results of Obermayer, Wiedemann, 

 and myself, it seems best to allow the results to stand in the form 

 above given, without an attempt to deduce a mean. 



Air. 



In Plate II. are shown the residual curves for all available observa- 

 tions on air. The line IJ, representing Maxwell's results, doubtless 

 owes its steepness to some large constant error, for which several 

 explanations have been offered. The lines for O. E. Meyer's results 

 from capillary transpiration are 



AB from Eqn. rj = 0.000171 (1 + 0.0024 t), Eange 20° to 99° 

 CD " " ,, = 0.000170(1+0.0028 0, " 21° to 100° 

 £F " " v = 0.000174 (1 4- 0.0030 t), " 21° to 100° 



