Tornado in N. E. part of Ohio, Feb. 4, 1842. 293 
at Kirtland had a tight ceiling of plaster, without any opening to 
the garret ; so that currents of air which might find their way 
into the building might strike upon this ceiling, but would be ef- 
fectually cut off from the roof. The ceiling was uninjured, but a 
large hole burst in the roof. 'The reason then assigned, though 
admitted to be a vera causa, is wholly inadequate to explain the 
phenomenon. I ascribe it to a rarefaction of the air, A current 
of air forcibly impelled over an obstacle like the roof of a build- 
ing, by friction drags along with it the air lying upon the leeward 
side of the roof, producing a partial rarefaction which might ea- 
sily be sufficient to lift the roof. The degree of rarefaction requi- 
site to produce the effect is much less than some might imagine. 
Suppose a barometer above the roof to stand only one tenth of an 
inch lower than within the building, we have an upward force 
of seven pounds per square foot, a force probably sufficient to 
throw off a considerable part of the barn roofs of Ohio. 
In the tornado which occurred at Stow in 1837, a circumstance 
was remarked which I had never seen noticed before, that several 
fowls were picked almost clean of their feathers. In the New 
Haven tornado of 1839, the same fact was noticed. I made par- 
ticular inquiry on this point at Mayfield. Four geese were found 
near Mr. Dean’s house lying dead among the rubbish, without any 
perceptible loss of feathers. I heard of one goose so far stripped 
that its remaining feathers were not considered worth saving: its 
legs and wings were both broken. A turkey also was found 
lodged in a tree near Dean’s, with the loss of about half of its 
feathers, and its bones broken. As electricity has been maintain- 
ed to be the efficient cause of tornadoes, I have inquired whether 
this loss of feathers could be due to electric repulsion. I 
think not—Because first, the effect is not uniform; many fowls 
near the centre of the Mayfield tornado were killed but did not 
lose their feathers. Secondly, although electricity was manifest- 
ed in the passage of the tornado, its intensity did not appear to 
be equal to that of a common summer thunder-shower. Thirdly, 
the effect may be explained by other causes known to be in ope- 
ration. 'The gun mentioned on p. 284, was charged with five 
ounces of powder, and a chicken just killed added for a ball. 
As the gun was small, it was necessary to press down the chick- 
en with considerable force, by which means it was probably some- 
what bruised. The gun was pointed vertically upwards and 
