340 MM. H. Sainte-Claire Deville and Troost on the 



also easy to understand why the temperatures ascertained by 

 M. Ed. Becquerel, and which differ from those which we have 

 ourselves published, are lower by about 100° than all those 

 which have been obtained either by ourselves or by others. 



This porosity of the platinum perhaps prevents the gases from 

 being maintained at a high temperature and at a high pressure ; 

 but the most important effect is that endosmose causes the gases 

 of the furnace to enter to the air of the pyrometer in spite of a 

 contrary pressure. Water is thus formed, and there is a dimi- 

 nution of volume due to the disappearance of oxygen. It will 

 be hence understood why, in M. Ed. BecquereFs experiments, it 

 was necessary to keep pieces of chloride of calcium constantly in 

 the pyrometer, and even to change them between the different 

 experiments (see p. 88 of his Memoire). 



It is thus evident how the mass of gas, which in the platinum 

 pyrometer ought to remain the same (constant C), has, on the 

 contrary, varied from 23*8622 to 19*9164 in the series of expe- 

 riments in which M. Ed. Becquerel appears to place most con- 

 fidence. This variation of the constant attains the proportion of 

 1 6*5 per cent, of its maximum value, which approaches the pro- 

 portion of oxygen which the air contained at the beginning of 

 the experiment. M. Ed. Becquerel explains (page 90) this consi- 

 derable variation, on the supposition that the mercury in his 

 manometer could combine in the cold with the oxygen of the air 

 contained in the apparatus. This hypothesis is inadmissible: 

 it would take away all precision from experiments made with his 

 or similar apparatus ; it would throw doubt upon all coefficients 



of expansion, &c and in fact upon numbers which have 



received every possible experimental confirmation. 



The true explanation is plainly to be derived from our experi- 

 ments; it is that, at a temperature not necessarily very high, 

 platinum becomes capable of producing the phenomena of endos- 

 mose, and perhaps even of the losses by pressure which we find 

 in porous substances. 



The differences between the temperatures observed by M. Ed. 

 Becquerel and ourselves are the following : — 



Deville and 

 E. Becquerel. Troost. Difference. 



Boiling-point of cadmium . 746*3 860 113*7 



Boiling-point of zinc . . . 932*0 1040 108*0 



We have made use, for our determinations, of a direct process, 

 independent of any hypothesis, and free from any capital source 

 of error ; it is the measurement, under the same pressure, of a 

 volume of air taken successively at 0°, then at the temperature 



