306 
MINUTES OF PROCEEDINGS OF 
the lower edge, cooled in water, contracted ‘6 inch, and the circumference* 
at about one inch above the water-line, was reduced 5*5 inches; the internal 
surface had increased in depth *35 inch. 
The small cylinder diminished *7 inch on the upper edge, increased *3 inch 
on the lower edge, and contracted 5*25 inches at about 1 inch above the 
water-line; the internal surface had increased in depth *3 inch. 
Experiment 3. A cylinder of very thin wrought-iron, so thin that it 
could not be welded, and was therefore riveted, of the same external dimen¬ 
sions as the 9-inch one of the foregoing experiment, was heated to redness 
and cooled by half-immersion ten times, in order to test the effect when the 
thickness of the metal was reduced as much as possible. 
. The upper and lower edges were not altered materially, while the greatest 
contraction took place on the water-line, instead of 1 inch above it as in the 
last experiment, and amounted to 3*5 inches. The depth measured on the 
curve had increased *15 inch (see fig. 6). 
Fig. 6. 
One-eighth of full size. 
Experiment 4. Two wrought-iron cylinders, exactly similar to those 
used in Experiment 2, were heated and cooled by being immersed to two- 
thirds their depth in water twenty times. 
The upper edge of the large cylinder was reduced 2*1 inches, and the 
lower edge ‘9 inch; it contracted 5*9 inches at about an inch above the 
water-line, and the inside surface had increased in depth *35 inch (see fig. 7). 
The upper edge of the small cylinder was reduced in circumference 3*6 
inches and the lower edge *65 inch, while the greatest contraction at about 
one inch above the water-line was 4*6 inches; and the internal surface had 
increased *15 inch in height (see fig. 8). 
