1 54 Experimental Investigations 



similar to the adjoining portion of the curve, of equal 

 length, which is convex toward that axis, as it ought to 



O * * o 



be according to our theory ; and even the part of the 

 curve which is convex toward the axis, A E, differs sen- 

 sibly from the logarithmic, particularly toward its ex- 

 tremity, P. 



It ought necessarily to differ from this curve as far 

 as the divisions of our thermometers are defective ; but the 

 deviation between the ordinates, A /and B g, indicated 

 by the results of the experiment in question, appears to 

 me much too considerable to be ascribed to the imper- 

 fection of our thermometers. 



To see how far the curve, P Q, differs from the loga- 

 rithmic, we have only to draw a logarithmic curve, R S, 

 through the points and /, and we shall find, that the 

 ordinates corresponding to the points 



A, B, C, D, E, 



o o o o o 



Instead of being 212.00 162 132! io6 32.00 

 Will be i99'55 *62 131 106^ 86.3$ 



Difference '2.45 o if o 



The very great difference that exists between the tem- 

 perature of cold water and that indicated by the results 

 of the experiment for the extremity of the cylinder 

 which was in contact with this water led me to suspect 

 that it was owing to the quality possessed by water in 

 common with other fluids, which renders it a very bad 

 conductor of heat. 



If it be true, as I believe I have elsewhere proved, 

 that there is no sensible communication of heat be- 

 tween the adjacent particles of a fluid, from one to 

 another, and that heat is propagated through fluids 

 only in consequence of a motion of their particles, re- 



