plished this work. Their experiments, commenced in 1870, have been the subject of various interesting communications to the Academy of Sciences. Their apparatus are deposited in the vaults of the Polytechnic School. They are much smaller than those of Cavendish and Baily, for, as Messrs. Cornu and Bailie have remarked, there is an advantage gained in these experiments by reducing the dimensions of the apparatus. The attracting mass, formed of mercury contained in two hollow spheres of bronze, 0·12 metre in diameter, weigh twelve kilogrammes. By transpiration the mercury can be made to pass from one sphere to the other, thus doubling the effect of the attraction, and this change is effected without shock or disturbance.[1] The lever of the torsion-balance is a little tube of aluminium, 0·50 metre in length, carrying at each end a ball of copper weighing one hundred and nine grammes. A flat mirror fixed in the middle reflects the divisions of a horizontal scale five or six metres distant, and the slightest movement of the lever is thus revealed by a displacement of the scale divisions. The time of a double oscillation of the lever is about seven minutes. The phases of these oscillations are registered by electricity. A great merit of these researches consists in the opportunity they afford for a thorough study of all the causes of perturbation that can introduce error into such experiments. The definite result can be accepted with confidence. The figure thus far obtained is 5·56. It may be added that Messrs. Cornu and Bailie have discovered the cause of the too large number given by Baily. In correcting the errors of system in his experiments, it is probable that a slightly different number will be obtained—5·55. To sum up, the earth's mean density thus appears to be five and a half times that of water, and the density at the surface is less than half that of the interior, or about 2·5. Consequently there must be in the interior heavy masses whose excess of density compensates for the lack thereof in the rocks at the surface. This need not be surprising, for the heavy pressure sustained by the deeper strata must naturally increase the density. But what is the law governing this increase of density from surface to center? Legendre formulated a simple law, adopted also by Laplace, according to which the surface density is 2·5, at the middle of the radius 8·5, and at the center 11·3, the mean being taken as 5·5. A different law, to which M. Edouard Roche arrived by theoretic considerations, gives a surface density of 2·1, a mid-radius density of 8·5, and 10·6 at the center. This agreement of results deduced from three different hypotheses shows that the decision of the question is narrowed to small dimensions. Adopting M. Roche's conclusions as the most probable, it can be said that the mean density of the earth is about double that of its surface, and that the density of the center is double that of its mid-radius. The central strata or masses have a density approximating to that of lead.
- ↑ In the later experiments the number of the spheres was doubled.
