For like reasons it is manifest that the process must repeat itself in each of the subordinate groups of units that are differentiated by the modifying forces. Each of these subordinate groups, like the original group, must gradually, in obedience to the influences acting upon it, lose its balance of parts—must pass from a uniform into a multiform state. And so on continuously. Whence indeed it is clear that not only must the homogeneous lapse into the non-homogeneous, but that the more homogeneous must tend ever to become less homogeneous. If any given whole, instead of being absolutely uniform throughout, consist of parts distinguishable from each other—if each of these parts, while somewhat unlike other parts, is uniform within itself; then, each of them being in unstable equilibrium, it follows that while the changes set up within it must render it multiform, they must at the same time render the whole more multiform than before. The general principle, now to be followed out in its applications, is thus somewhat more comprehensive than the title of the chapter implies. No demurrer to the conclusions drawn, can be based on the ground that perfect homogeneity nowhere exists; since, whether that state with which we commence be or be not one of perfect homogeneity, the process must equally be towards a relative heterogeneity. * * * * * § 110. The stars are distributed with a three-fold irregularity. There is first the marked contrast between the plane of the milky way and other parts of the heavens, in respect of the quantities of stars within given visual areas. There are secondary contrasts of like kind in the milky way itself, which has its thick and thin places; as well as throughout the celestial spaces in general, which are much more closely strewn in some regions than in others. And there is a third order of contrasts produced by the aggregation of stars into small clusters. Besides this heterogeneity of distribution of the stars in general, considered without distinction of kinds, a further such heterogeneity is disclosed when they are classified by their differences of colour, which doubtless answer to differences of physical constitution. While the yellow stars are found in all parts of the heavens, the red and blue stars are not so: there are wide regions in which both red and blue stars are rare; there are regions in which the blue occur in considerable numbers, and there are other regions in which the red are comparatively abundant. Yet one more irregularity of like significance is presented by the nebulæ,—aggregations of matter which, whatever be their nature, most certainly belong to our sidereal system. For the nebulæ are not dispersed with anything like uniformity; but are abundant around the poles of the galactic circle and rare in the neighbourhood of its plane. No one will expect that anything like a definite