138 SCIENCE PROGRESS. 



I have reason to hope that such an investigation is 

 about to be undertaken in Cambridge, and it will, I trust, 

 lead us on to firmer ground. 



A few remarks on one other point. I have already- 

 indicated the importance of changes in the conditions under 

 which the experiments are performed. 



Rowland's determinations are, I consider, deficient in 

 this respect. It is difficult to ascertain, from his tables, 

 to what extent he varied the mass of the water and the 

 rate of the work done. He appears to have changed his 

 mass of water from about 5 to 8 kilos., and the work per 

 kilo, seems to have varied but slightly. 



The method adopted by Miculescu enabled him to 

 employ large masses of water — his variation ranging- from 

 5 to 18 kilos. Small errors are of much less importance 

 when weights such as these are used, especially when the 

 method employed, as in the case of Rowland, involves 

 estimation of the water equivalent, losses by radiation, etc. 

 In this respect, their conditions are distinctly superior to 

 mine. This applies more particularly to the work of Micu- 

 lescu, for the method he adopted reduced errors of the 

 above description to a minimum. In my paper I have 

 given my reasons as to why we were compelled to work 

 with comparatively small masses and how the resulting evils 

 were counterbalanced by other advantages. 



During our experiments we were, however, able to 

 vary the conditions to an extent not attempted by either 

 Rowland or Miculescu. The rate of production of heat 

 was continually altered (the maximum rate as compared 

 with the minimum was as nine to one) and the mass of 

 water was in some experiments nearly three times as 

 great as in others. The close agreement between results 

 obtained under such varied conditions greatly strengthens 

 the conclusions arrived at. 



The following table gives a summary of the results 

 obtained by the observers I have referred to, expressed 

 in kilogrammetres in latitude of Greenwich (g-^gSi'ij). 1 



1 My values as given in this table differ slightly from those pub- 

 lished in the Transactions. The reasons for the alterations are given in 



