of Air- and Mei'cury -Thermometers. 219 



the temperature, may extend to several degrees while the two 

 are being heated to 250° or 300°. Another adverse statement 

 advanced by M. Pierre, is that exactly the same coefficient of di- 

 latation for the same glass apparatus is not always found, although 

 the circumstances are apparently identical. If such uncertainty 

 is the normal state of thermometric measurement, how are we to 

 account for the results shown in fig. 2? In the face of them, it 

 seems impossible to accept the comparative thermometric obser- 

 vations of jMI\I. Pierre and Reguault as conclusive. 



§ 10. Besides the difference in the march of the air- and mer- 

 cury-thermometers, there must be a difference in the value of 

 the individual degrees marked on the two thermometers. Em- 

 ploying the formula [Appendix, III.] to obtain an approximate 

 value of the mercury degrees in terms of the degrees of the air- 

 thermometer, we find that at 0° the value of a degree, or y^yth 

 of the distance between the freezing and boiling temperatures, 

 is fyds the value at 100°. At the lower part of the scale, the 

 value exceeds the mean by :f gtli part, and at the higher point it 

 is ^'ijth part less than the mean. This inequality may at times 

 have a sensible effect on the quantitative determinations of physi- 

 cists. Thus the mechanical equivalent of heat is, according to 

 Mr. Joule, 772 for 1° P., at temperature about 60° F. If this 

 were reduced to the standard value of a degree of the air-ther- 

 mometer, we should have to reduce it by g^jth part, viz. to 759. 



§ 11. There is another point referred to by JI. Regnault as 

 a source of uncertainty in measurements by the air-thermometer. 

 The deviation from the primary laws of elastic fluids which he 

 discovered must have some effect in disturbing that ratio of 

 equality between the increments of volume and temperature, or 

 of pressure and temperature, upon which the value of the instru- 

 ment depends. 



The air-thermometer may indicate temperature by change of 

 volume under constant pressure, or by change of pressure with 

 volume constant. M. Ilegnault's observations make the pro- 

 portionate change of volume slightly larger than the change of 

 pressure between the standard points of the thermometer; the 

 difference being perhaps within the limits of error. 



AVith increased densities these proportionate changes sensibly 



1 

 augment at the rate of about ——r per atmosphere, and Messrs. 



Joule and Thomson's experiments with plugs, indicate a devia- 

 tion from the law of Mariotte of about jr-^j^ per atmosphere at 



<w7'W 



ordinary temperatures, and a probability that this amount in- 

 creases in descending the scale of temperature. [See paper on 

 Deviation from the Primary Laws of Elastic Fluids, Phil. Mag., 



