tl2 M. R. Clausius on the Moving Force of Heaf, 



it decidedly decreases with tlie temperature. Between 35° and 

 90° this decrease is very uniform. Before 35°, particularly in 

 the neip^hbourhood of , considerable irregularities take place ; 

 which, however, are simply explained by the fact, that here the 



pressure p and its differential quotient -—; are very small, and 



hence the trifling inaccui-acies which might attach themselves to 

 the observations can become comparatively impoiiant. It may 

 be added, further, that the cune by means of which, as men- 

 tioned above, the single values of jo have been obtained, was not 

 drawn continuously from —33° to 100°, but to save room was 

 broken off at 0°, so that the route of the curve at this point 

 cannot be so accurately determined as within the separate por- 

 tions above and below 0°. From the manner in which the di- 

 vergences show themselves in the above table, it would appear 

 that the value assumed for p at 0° is a little too great, as this 



would cause the values of Ap(s'^a-) to be too small for the 



a "T~ I 



temperatures immediately under 0°, and too large for those above 

 it. From 100° upwards the values of this expression do not 

 decrease with the same regularity as between 35° and 95°. They 

 show, how ever, a general coiTCspondence ; and particularly when 

 a diagram is made, it is found that the curve, which almost 

 exactly connects the points within these limits, as determined 

 from the numbers contained in the foregoing table, may be car- 

 ried forward to 230°, the points being at the same time equally 

 distributed on both sides of it. 



Taking the entire table into account, the route of this curve 

 may be expressed with tolerable accuracy by the equation 



Ap{s~-(T)—-=m—ne'^^; . . . (26.) 



in which e denotes the base of the Napierian logarithms, and m, 

 n, and k are constants. When the latter are determined from 

 the values given by the curve for 45°, 125° and 205°, we obtain 



m=31-549; «= 1-0486; ^=0007138; . (26^.) 



and when for the sake of convenience we introduce the loga- 

 rithms of Briggs, we have 



log[31-549-Ajo(5-o-) _^J=00206 + 0-003100^. (27.) 



From this equation the numbers contained in the third column 

 are calculated, and the fourth column contains the differences 

 between these numbers and those contained in the second. 

 From the data before us we can readily deduce a formula 



