358 THIHU RKJPORT 1833. 



greater. The coincidence is often less perfect. At 0'1775 Th. 

 the quantity calculated is 42'65, the quantity given by experi- 

 ment 43*03, a difference of 0'38 millionth. The experiment 

 mentioned above gave a recession = 41*51 at 0*20125 Th. 

 (mean of the temperatures of the beginning and end of the se- 

 ries). The calculation gives 41*63, or a difference of 0*12 mil- 

 lionth. At 0*005 Th. the change of volume produced by one 

 0*01 is 60*5 millionths, but inversely, as the water at low tem- 

 perature loses in volume by augmented heat ; thus an addition 



is to be made equal to ^ . 0*00025 = ] *5. Now 46*77 + 1 *5 



gives 48*27, experiment 48*02. At 0*019 the quantity calcu- 

 lated is 47*72, that given by experiment 47*97. I have not yet 

 finished the tedious discussion of all the experiments, but as 

 far as I have proceeded the agreement of the hypothesis with 

 facts is satisfactory. Messrs. CoUadon and Sturm have in the 

 calculation of their experiments introduced a correction founded 

 upon the supposition that the glass of the bottle in which the 

 water is compressed should suffer a compression so great as to 

 have an influence upon the results. Their supposition is, that 

 the diminution of volume produced by a pressure on all sides 

 can be calculated by the change of length which takes place in 

 a rod during longitudinal traction or pression. Thus, a rod of 

 glass, lengthened by a traction equal to the weight of the atmo- 

 sphere as much as 1*1 millionth, should by an equal pression on 

 all sides lose 3*3 millionths, or, according to a calculation by the 

 illustrious Poisson, 1 •05 millionth. As the mathematical cal- 

 culation here is founded upon physical suppositions, it is not 

 only allowable, but necessary, to try its results by experiment. 

 Were the hypothesis of this calculation just, the result would 

 be, that most of the solids were more compressible than mei*- 

 cury. For this purpose I have procured cylinders of glass, of 

 lead, and of tin, which filled the greater part of a cylinder, to 

 which a stopple of glass, perforated by a capillary tube, was 

 adjusted by grinding. I have not yet exactly discussed all the 

 experiments on this subject, but the numbers obtained are such 

 as to show that the results are widely different from those cal- 

 culated after the supposition above mentioned. The quantity 

 assigned by this calculation to the glass is very small indeed, 

 yet the experiment gives it much less. Lead, which extends, 

 according toTredgold, 20*45 milUonths by a weight equal to that 

 of the atmosphere, and thus much more by the pressure on all 

 sides, does not change one millionth. Tin is not more com- 

 pressible. The inverse experiment is, perhaps, still more 

 striking. I published it some years ago : however, as I have now 



