EXPERIMENTAL KNOWLEDGE OF THE PROPERTIES OF MATTER. 483 



in thig respect, in which it is comparable with the general rule in respect 

 to metals. Now this temperature, as is well seen in the diagram, 

 is reached by diamond at 600° : and therefore if we are not arbitrarily to 

 choose a temperature at which the specific heat will be such that the atomic 

 heat of carbon is 6'4, the range of temperature within which we should 

 determine the speciBc heat to test the conformity of carbon with the law 

 of Dulong and Petit should not much exceed 900°, at about which 

 temperature the specific heat is "45 : this would give atomic heat of 

 carbon 5'4. 



By experiments with crystallised boron and crystallised silicon Weber 

 found that the specific heats of both increased rapidly as the temperature 

 rose, the specially rapid variation ceasing for silicon before 100° ; the 

 atomic heat of silicon being 5'63 at 184°, and 574 at 300°. In the case 

 of boron the specific heat rose rapidly up to 230°, and there is no result 

 given to show at what temperature it becomes nearly constant. Moreover, 

 the crystallised boron is not pure, always containing aluminium, as 

 Hampe has shown.' 



It may be remarked that in respect of the relation of Dulong and 

 Petit's law to the molecular heat of compounds at ordinary temperatures, 

 Kopp finds that this relation is best expressed by taking the low atomic 

 heats of carbon 1*8 rather than the value which more nearly agrees with 

 Dulong and Petit's law. 



JBerylUum ; — Htimpidge ; Nilson and Pettersson. 



The specific heat of beryllium between 10° and 100° was found to be 

 •47, from which the atomic weight 13'6 was deduced by applying the law 

 of atomic heats. The impossibility of finding a place for a metal with 

 this atomic weight in the classification of elements in the periodic system 

 led chemists to doubt the accuracy of this specific heat, especially as the 

 metal from which the specific heat was got was known to be not free 

 from impurity. With the object of settling this point the late Professor 

 Humpidge devoted a great amount of labour to preparing beryllium in a 

 state as nearly pure as he could ; and he succeeded in preparing some 

 with not more than 0'8 per cent, of impurity.^ He proceeded to make 

 a great number of determinations of specific heats at various tempera- 

 tures, and found that beryllium, like carbon, silicon, and boron, has a 

 specific heat which rapidly increases with rise of temperature, and be- 

 comes nearly constant at temperatures above 400°, from that to 500° 

 increasing quite slowly. The number '5403 within those limits when 

 multiplied by 9'1 gives 5'64, which is not far from that required by the law 

 of Dulong and Petit. Now analyses of the chloride and bromide had 

 shown that, if their formulae are BcCIq, BeBrg the atomic weight is 91 ; 

 whereas the formulge BeClj, BeBrg correspond to atomic weight 13'6. 

 Humpidge published a further paper ^ giving details of his work on the 

 specific heat, and of his determinations of the vapour-densities of the 

 chloride and bromide, confirming that got by Nilson and Pettersson for 

 the chloride. 



The results of Humpidge were partly anticipated by Nilson and 

 Pettersson in respect of the rapid increase of specific heat of the element, 



' L. Ann. 183, 1876, p. 75. 



2 Proc. Boy. Soc. 38, 1884, p. 188 ; Abs. C. S. J. 1885. 



' Proc. Eoy. Soc. 39, 1885, pp. 1-19. 



I I 2 



