418 On the Volume of Water at various Temperatures, 



of the heat would be communicated from the bottom, for the 

 top of the fluid would condense the stratum of air in contact 

 with it : but we do not suppose it would continue to obtain heat 

 from the air, at least it would be at a very slow rate. Also it is 

 evident, that the lower part of tlie water must require a consi- 

 derable degree of expansion to enable it to overcome the co- 

 hesion and inertia of the part above ; and the expansion at \ow 

 temperatures being very small, it is probable that it would re- 

 ejuire several degrees of heat to put it in motion. The motion 

 once begun, a less quantity of heat would continue it ; also the 

 expansion would be more as the temperature increased. In this 

 manner the results of the first cx))erimcnt may be accounted for, 

 without supposing the water to contract with the addition of heat. 



In the second experiment we are not certain tliat any motion 

 took place till the upper thermometer sunk to 36°; and what- 

 ever care might be taken to keep the surrounding fluid at an 

 equal temperatiue, near the surface, that part of it under the 

 jar would acquire a temperature suflficient to cause the lower 

 tliermometer to become stationary, till the upper part of the water 

 jiecame dense enough to overcome the resistance and descend. 



In the third experiment, motion would speedily take place : 

 but it appears that when the diflfcrence between the top and bot- 

 tom temperatures was only one degree, the motion ceased ; and 

 here we meet with a fact which in some measure proves the cor- 

 rectness of our explanation ; that is, the fluid in the axis at the 

 top was at 34°, at the same time that part of the fluid, we suppose 

 at about half the height' of the jar, was freezing. Why did it 

 not ascend, and let the warmer portion take its place ? 



Without some difference in specific gravity, we must say that 

 motion could not take place. Now it appears clearly that there 

 was no motion in tlic water during the fourth experiment ; for, 

 when both the thermometers were at 36% a part of the water 

 must have been at 32° : had there been any motion, either the 

 upper or lower thermometer must have been at 32°, or nearly 

 so. The jar was placed in a pan, we .suppose without any thing 

 between it and the bottom of the pan: — the pan would be sup- 

 jiorted by something that would conduct heat ; and in conse- 

 fpiencc of this arrangement, that part of the water round the 

 lower thermometer would be receiving heat from the bottom of 

 tlic jar ; which will account for its not being cooled at a quicker 

 rate than the top one. 



The fifth experiment was made with a taller jar, and the 

 heat applied at llie middle : it would hjive been interesting to 

 have known the temperature in the axis of the jar, at the height 

 where the heat was apj)]ied. 



This experiment, in common with the others, tends to prove 



that; 



