June i6, 1898] 



NATURE 



16! 



After pointing out that all the results of previous experiments, 

 except those of Lavoisier and Laplace, are extremely incorrect, 

 they describe their results obtained by the method of cooling, 

 conducted with many precautions to avoid error. 



Copy of Table by Dulong and Petit {Ann. Chim. 

 Phys., 1819, X. 403). 



The statement of the law is best given in the words of the 

 authors (p. 405) : 



" Les atomes de tous les corps .simples ont exactement la 

 meme capacite pour la chaleur." 



Here the question rested till resumed, many years later (1840), 

 by Regnault, who in his first memoir {Ann. Chim., 73, 5) 

 points out the difficulties which attended the acceptance of 

 he statement of Petit and Dulong in the form in which they 

 .;ave it. He then discusses the three principal experimental 

 methods, viz. (i) fusion of ice, (2) mixture, (3) cooling, and 

 decides in favour of the second, which he used throughout his 

 researches. The general form of the apparatus used by the 

 rreat physicist has been a model for the guidance of successive 

 xperimentalists since his time. 

 Another quarter of a century elapsed before the question of 

 the specific heat of the elements was resumed by Hermann 

 Kopp. His results were communicated to the Royal Society, 

 and are embodied in a paper printed in the Philosophical Trans- 

 actions for 1865. After reviewing the work of his predecessors, 

 he describes a process by which he has made a large number of 

 estimations of specific heat, not only of elements but of com. 

 pounds of all kinds in the .solid state. Concerning his own pro- 

 cess, however, he remarks that " the method as I have u.sed it 

 has by no means the accuracy of that of Regnault " (p. 84). 



In 1870 Bunsen introduced his well-known ice-calorimeter. 

 This is an instrument in which the amount of ice melted by the 

 heated body is not measured by collecting and weighing the 

 water formed, but by observing the contraction which ensues 

 when the ice melts, contained in a vessel of special form. The 

 results obtained by Bunsen himself are uniformly slightly lower 

 than those of Regnault for the same elements. 



Since that time experiments have been made by Weber, 

 Dewar, Humpidge, and others in connection especially with 

 the influence of temperature in particular cases. 



Setting aside the elements carbon, boron, silicon and beryl- 

 lium, as providing an entirely separate problem, the question is 

 whether the law of Dulong and Petit is strictly valid when 

 applied to the metals. Kopp, in his discussion of the subject, 

 came to the conclusion that it is not ; but the grounds for this 

 conclu.sion are unsatisfactory, since neither the atomic weights 

 nor the specific heats were at that time known with sufficient 

 accuracy. 



It has been customary to assume that the divergencies from 

 the constant value of the product. At. Wt. x Sp. Ht., are due 

 partly to the fact that at the temperature at which specific heats 

 are usually determined, the different elements stand in very 

 different relations to their point of fusion ; thus lead at the 

 tentperature of boiling water is much nearer to its melting- 

 point than iron. It has also been attributed to temporary or 

 allotropic conditions of the elements. As to the relation to 

 melting-point, the specific heats of atomic weight seem to be 

 practically the same in separate metals and alloys of the same 

 which melt at far lower temperatures. For example, the atomic | 

 heat of cadmium is 6*35, of bismuth 6-47, of tin 6-63, and of lead 



NO. 1494, VOL. 58] 



6 50 ; while the mean atomic heat in alloys of bismuth with tin» 

 and lead with tin, ranges from 6-40 to 6-66 (Regnault), which is 

 practically the same. 



Again, while the melting-point of platinum is at a white heat, 

 and it becomes plastic at a low red heat, the specific heat at this 

 lower temperature is very little less. Many other metals 

 change considerably in properties at temperatures far removed 

 from their melting-points, without substantial change in their 

 capacity for heat. 



As to allotropy it is a phenomenon which is comparatively 

 rare among metals, and in the marked cases in which it occurs 

 we have no information as to the value of the specific heats in 

 the several varieties (such as the two varieties ol^ antimony and 

 the .Mlver zinc alloy of Heycock and Neville), and they may be 

 left out of account. Bunsen compared the so-called allotropic 

 tin obtained by expo.sing the metal to cold for a long time, and 

 found it 0545 against "0559 for the ordinary kind {Pogg. Ann., 

 141, 27). In dimorphous substances, such as arragonite and 

 calcite. there is often no difference. Regnault found for these 

 two minerals "2086 and '2085 respectively. 



The differences between metals hammered and annealed, 

 hard and soft, were also found by Regnault to be very small 

 {Ann. Chim. [3] ix. ) :— 



Hard steel 

 Hard bronze ... 



•II 75 Same, softened 

 •0858 Same, softened 



•1165 

 •0862 



Kopp came to the conclusion, first, that each element in the 

 solid state and at a sufficient distance from its melting-point has 

 one specific or atomic heat which varies only slightly with 

 physical conditions ; and, secondly, that each element has 

 e.ssentially the same specific or atomic heat in compounds as it 

 has in the free state. This last is practically identical with the 

 statement which is known as Neumann's law. With Kopp's 

 conclusions I agree, but from some of Regnault's results, 

 coupled with my own, the effect of small quantities of carbon 

 and, perhaps, of sulphur upon the specific heats of metals is 

 greater than has been supposed. 



If we take the results of Regnault and of Kopp, and combine 

 them with the most accurately known atomic weights, the 

 products are still not constant. 



Ato.mic Weights most accurately known (1897), 



COMBINED WITH SPECIFIC HeATS. 



The law of Dulong and Petit is therefore only an approxim- 

 ation, but this may perhaps be due to impurity in the materials 

 used. That is the problem which I have endeavoured to solve. 



The introduction of a new method of calorimetry by Prof. J. 

 Joly, and the excellent results obtained by the author in the use 

 of the differential form of his instrument {Proc. K. S., 47, 241), 

 led me to think that with due attention to various precautions, 

 such as exact observation of the temperatures and practice in 

 determining the moment at which the increase of weight due to 

 condensation is completed, results of considerable accuracy 

 might be obtained. 



The problem is to find two elements very closely similar in 

 density and melting-point which can be obtained in a state of 

 purity, and then to determine with the utmost possible accuracy 

 the specific heat of each under the same conditions. The two 

 metals cobalt and nickel were selected for the purpose. They 

 were examined by Regnault, but the metals he used were very 

 impure. 



The cobalt employed in my experiments was prepared by 

 myself. For the nickel I am indebted to Dr. L. Mond. Both 

 were undoubtedly much more nearly pure than any metal 

 available in Regnault's time. The results obtained are as 

 follows : — 



