16 Prof. T. Carnelley on the Periodic Law, 



halogen compounds of the elements have . been calculated. 

 The results obtained in the case of the bromides are given in 

 Table VIII. p. 15. In the few cases where overlapping limits 

 occur those limits are taken as correct, in favour of which 

 there are the greater number of limits, the others being 

 rejected. 



In this way it has been found that the average differences 

 of the calculated limits from the experimental values were : — 



Per cent. 

 For the boiling-points of Groups iii.-vn.= + 4°'8 0., or 1*0 



„ „ „ I. & ii. = + 4°-0 or 0-6 



For the melting-points of Groups iii.-vii. = + 15 o, or 4*2 



„ „ „ i. An. =+15°-0 or 1*8 



The size of the average error is due chiefly to exceptionally 

 great differences in but a few cases. 



For some compounds the number of inferior or superior 

 limits, or both, is very small; and therefore the error in 

 these instances is chiefly due to this source. It is to be 

 remarked that an error in any one of the data used in the 

 construction of the tables employed in the calculations may 

 produce a corresponding or even greater error in the calcu- 

 lated melting-points or boiling-points of several other bodies. 

 Most of the errors in the above calculated values may be 

 ascribed therefore to errors in the experimental numbers of 

 but a very few substances. It must not be forgotten that all 

 the calculation really gives is the limits between which a sub- 

 stance melts or boils, and not the exact melting- or boiling- 

 point. 



Calculation of Melting- and Boiling-points not yet experi- 

 mentally determined. — By the method just described, a large 

 number of unknown melting- and boiling-points have been 

 calculated; and it is left to the future to decide whether these 

 predictions are correct or otherwise. 



In many cases the values have been calculated in two 

 ways : — (1) By using experimental data only ; (2) by using 

 both experimental and calculated data ; whilst for others the 

 experimental data are so few in number, that it has only been 

 possible to make the calculation by using both experimental 

 and calculated data. The results obtained are given in the 

 following table. 



