86 
It was on days when the Thermometer in the sun varied its height rapidly 
and much — as on bright days with cold winds, or when clouds were driven 
over the sun— that the descent was greatest. So remarkably, indeed, was 
this the case, that every cloud which shut off the sun for a time from the lead, 
and every gust of wind which blew upon it in the sunshine, seemed to bring 
it down a step. On such days it would descend from | to | an inch. On the 
contrary, when the sky was open and clear, and "the heat advanced and 
receded uniformly, the descent was less, although the difference of the 
extreme temperatures of day and night might be greater. It was least of all 
on days of continuous rain. The sun was the obvious cause of the descent 
of the lead— a dilatation and contraction of it was caused by the passage into 
it, and the withdrawal, of the sun's radiant heat, and this dilatation and 
contraction of the lead caused it to descend. Why it should do so may be 
easily explained. 
Let A B (fig. 1) be an elementary plate of the solid, and conceive it to be 
divided into an infinite number of equal elements by planes perpendicular 
to its length. Let X be a point so taken in it that, if it were divided in X, 
the thrust necessary to push the part X A up the plane would equal that 
necessary to push X B down it. Let the element at X be imagined to have 
its temperature so raised as just to equal this thrust ; and let the tempera- 
tures of all the elements in X A, beginning at X, be equally raised in 
succession. Each will thus be dilated more than the one before it, because its 
dilatation will be opposed by a less resistance ; and the displacement A Ai 
of the extremity upwards will equal the sum of these several dilatations. 
In like manner,, 
Fig. 1. 
if the same temperature be added to the elements of X B in succession! 
beginning from X, each will be dilated more than the one before it, and th 
displacement B Bi of the extremity B downwards will equal the sum of thes 
several dilatations. The point X will obviously be nearer to A than to Tj| 
