158 THE THEORY OF GRAVITATION. 



CONSTITUTION OF HEAVY BODIES. 



First. Their indivisible particles are cages ; for example, hollow cubes 

 or octohedra. (They are, in other words, skeletons of solids of which 

 there is nothing material except the edges.) 



Second. The diameters of the bars of these cages, even if supposed 

 increased by the diameter of the gravitational corpuscles (as they must 

 be in order to conveniently evaluate the portion of the atoms inter- 

 cepted), are so small, relative to the distances between the parallel bars 

 of the same cage, that all the particles included in the terrestrial globe 

 intercept not the ten-thousandth part of the corpuscles which present 

 themselves to traverse it. 



Third. These diameters are all equal, or if they are unequal their 

 inequalities sensibly compensate each other. If, for instance, in the 

 smallest portions of matter separately ponderable (which, it has been 

 stated, may weigh one thirty-second part of a grain) the mean diam- 

 eter of the bars of the one portion does not differ a tenth part from the 

 mean diameter of the bars of the other, then it would follow that in 

 the greatest ponderable masses the mean diameters do not differ by a 

 ten -thousandth part, for every such great ponderable mass is com- 

 posed of so large a number of indivisible particles that simple chance 

 suffices to almost perfectly effect a compensation of diameters. 



CONSTITUTION OF GRAVITATIONAL CORPUSCLES. 



First. Conformably to the second of the preceding suppositions, the 

 diameter of the gravitational corpuscle added even to that of the bars 

 of the indivisible particles is so small relatively to the mutual distance 

 of the parallel bars of a single cage that the weight of celestial bodies 

 does not sensibly vary from the ratio between their masses. 



Second. The gravitational corpuscles are isolated, so that their 

 progressive movements are necessarily rectilinear. 



Third. They are so thinly scattered — that is to say, their diameters are 

 so small relative to their mutual mean distance — that there are no more 

 than a few hundreds which encounter one another in the course of a 

 thousand years. Hence the uniformity of their movements is never 

 sensibly disturbed. 



Fourth. They move in several thousand of thousands of different 

 directions, even counting as one all those which are parallel to the same 

 line. The distribution of these directions may be conceived as follows : 

 First, imagine all the points conceived to lie in different directions 

 strewn upon a sphere as uniformly as is possible, and consequently 

 separated from one another by less than a second of arc; then imagine 

 a corpuscular path radiating from each of these points. 



Fifth. Parallel to each of these directions there moves a stream or 

 torrent of corpuscles. Now, in order to give it no more than the neces- 

 sary size, the transverse section of this current has the same contour as 



