278 SCIENCE PROGRESS. 



made, the irregular errors cannot be kept much under 1 

 per cent. As was mentioned above, nearly all who have 

 used the method have assumed the gases are perfect, and 

 none but E. and L. Natanson and the writer have experi- 

 mentally determined the density of the gases as they were 

 used in the experiments, so that much of the work that has 

 been done is of little value. 



The table on p. 340 of the second edition of Landolt and 

 Bornstein's tables gives an almost complete list of all the de- 

 terminations that had been made up to 1893, with references. 

 The experiments of Jamin and Richard mentioned above 

 are omitted, as are also the writer's results, which have 

 been published since the table was drawn up {Phil. Trans., 

 185, 1, and Nature, p. 97, 7th March, 1895). 



In Ostwald's Zeitschrift (12, 116) Petrini gives a very 

 voluminous table containing not only the gases included in 

 Landolt and Bornstein's table, but many others, whose y's 

 he has calculated by an approximate formula from the 

 determinations of C ; „ by Regnault and Wiedemann. From 

 what has been said above, it is clear that it is little use 

 trying to construct theories from such a table, for the values 

 are so discordant that support could be found in it for 

 almost any theory. A little time spent in consideration of 

 the methods used and in consequent weeding out of results 

 would have been well spent. 



The following table contains what seem to the writer 

 to be the best values of y known at present. In most 

 cases only three figures are given, and in many of these 

 the third is probably only approximate. Four figures are 

 given where the observer has used accurate equations. 



