302 Dr. S. Young on the Law of Cailletet and 



were not taken even to the boiling-point, the errors in that 

 case are still greater. 



There seems to be no doubt that, in order to find the critical 

 density with the greatest possible accuracy, it would be best 

 to make use of the whole of the available data and to employ 

 the formula with three constants, thus 



D e =D + a(T e -273) + /3(T c -273) 2 . 

 This has been done for the thirty substances, and the new and 



old critical densities, also the new values of jS , are given in 

 Table IV. 



For convenience of reference the critical temperatures and 

 pressures, the specific gravities at 0°, and the boiling-points 

 under normal pressure are also given in the table. The order 



is in ascending order of the recalculated values of j£ . 



It will be seen that the differences between the new and old 

 critical densities are quite small. They are greatest for 

 chlorobenzene, bromobenzene, and iodobenzene ; but the 

 critical temperatures of these compounds are very high, and 

 the density determinations have only been taken to 260° or 

 270°. The greatest difference, that for iodobenzene, amounts 

 to 0*55 per cent., and for this substance there is an extrapola- 

 tion through 180°. 



In the case of the alcohols the differences are not excessive 

 because the old values were not obtained from the simple 

 formula but graphically from curves drawn through the points 

 representing the mean densities. 



Of the remaining 26 substances benzene shows the greatest 

 difference, 0*26 per cent. ; in fourteen cases it is less than 0*1 

 per cent. 



The alterations in the order of the substances in the tables 



[depending on yr? ) are few in number; di-isopropyl is 



brought next to isopentane ; the four halogen derivatives of 

 benzene are brought together, whereas in the old table bromo- 

 benzene was separated from the others ; lastly, the new order 

 of the alcohols is propyl, ethyl, methyl in place of ethyl, 

 propyl, methyl. 



In both tables the iso- and di-isocompounds have lower 



values of -p4 than the corresponding normal compounds. 



It is perhaps worth remarking that if, in the formula 



«T 

 a—-^-, we take the value of a in the formula with three 



XJc 



