338 



two years from 1851 to 1853 it would have descended 20'38296 

 inches. But the daily variations of atmospheric temperature are less 

 in the other months of the year than in the month of August. For 

 this reason, therefore, the calculation is in excess. For the following 

 reasons it is in defect : 1st, the daily variations in the temperature 

 of the lead cannot but have been greater than those of the surround- 

 ing atmosphere. It must have been heated above the surrounding 

 atmosphere by radiation from the sun in the day-time, and cooled 

 below it by radiation into space at night. 2ndly. One variation of 

 temperature only has been assumed to take place every twenty-four 

 hours, viz. that from the extreme heat of the day to the extreme 

 cold of the night ; whereas such variations are notoriously of con- 

 stant occurrence during the twenty -four hours. Each cannot but 

 have caused a corresponding descent of the lead, and their aggregate 

 result cannot but have been greater than if the temperature had 

 passed uniformly (without oscillations backwards and forwards) from 

 one extreme to the other. 



These considerations show, I think, that the causes I have 

 assigned are sufficient to account for the fact observed. They sug- 

 gest, moreover, the possibility that results of importance in meteor- 

 ology may be obtained from observing with accuracy the descent of 

 a metallic rod thus placed upon an inclined plane. That descent 

 would be a measure of the aggregate of the changes of temperature 

 to which the metal was subjected during the time of observation. 

 As every such change of temperature is associated with a corre- 

 sponding development of mechanical action under the form of work*, 

 it would be a measure of the aggregate of such changes and of the 

 work so developed during that period ; and relations might be found 

 between measurements so taken in different equal periods of times, 

 successive years for instance, tending to the development of new 

 meteorological laws. 



The following are the results of recent experiments f on the 

 expansion of ice : 



* Mr. Joule has shown (Phil. Trans. 1850, Part I.) that the quantity of heat 

 capable of raising a pound of water by 1 Fahr. requires for its evolution 772 units 

 of work. 



t Vide Archiv f. Wissenchaftl. Kunde v. Russland, Bd. vii. S. 333. 



