9.2 Mr. J. J. Waterston on certain Thermomolecular 



projected to t in thirteen cases, which include the most accordant, 

 Nos. 1 to 8, and the least accordant, Nos. 12 to 13. The most 

 remarkable is sulphurous ether, No. 6. The first series of 

 five observations are in perfect accordance with the theory, and 

 prove the extreme accuracy of M. Pierre in this and several 

 others, which are almost as good. In other cases, as Nos. 11, 

 12, 13, where the law of continuity is not maintained, it appears 

 as if some chemical disturbance had taken place in the liquid 

 during the process of heating. 



M. Pierre has submitted forty-four chemically pure liquids to 

 thermometric trial from 0° up to their boiling-point, and several 

 of them from 0° down to —30°. I have graphically examined 

 and analyzed all but a very few. Some of them with high boil- 

 ing-points present a clear range of about 100° ; and the straight- 



ness of the line of -^-, in which the presumed law of expansion 



consists, is well developed (see Nos. 3, 4, 5, 6, 7, 8, 9). 



Those with boiling-point under 100°, when regular, afford 

 the means of computing the constants 7 and P of the equation 



— =P (ry— t), — although the range is too small to prove the law, 



CIV 



except through the value they give to 7 being in accordance with 

 the principle shown by Plate III., that causes the upper limiting 

 temperature, 7, of many of the liquids whose lines of vapours 

 trend to the ether-node to range in a line parallel to the axis 

 of temperature. 



§ 32. In most of these M. Pierre has given for each observa- 

 tion the value of a, the mean expansion from 0°. Thus, 1 being 



v — 1 

 the volume at 0° and v the same at t°, a= — — . The projec- 

 tion of these values of a as ordinates to t ought to present a 

 series of points in a curve convex to the axis of temperature : 

 chart Q, Plate IV., is an example of this projection in the case 

 of the three ethers — hydrochloric, hydrobromic, hydriodic. The 

 first has too small a range ; but it will be remarked that the 

 upper four points keep well in a line, and that three are exact. 

 This makes it probable that they are sufficiently accurate. Ac- 

 cordingly, the two observations at the points marked A, B are 

 selected to compute y and P , as follows : — Let / t l} v v x be the 

 respective temperatures and volumes at those points having re- 

 ference to the unit volume at 0°. We have 



-Po(Y-K) 



; — 1 civ 



the reciprocal of the absolute increment of volume at |/ , and 



