The estimation of the action of the liquid is a difficult part of this problem. La Place, Dr. Young, and other mathematicians, have considered the liquid within the tube to be of uniform density; but M. Poisson, in one of those masterly productions in which he elucidates the most abstruse subjects, has proved that the phenomena of capillary attraction depend upon a rapid decrease in the density of the liquid column throughout an extremely small space at its surface. Every indefinitely thin layer of a liquid is compressed by the liquid above it, and supported by that below. Its degree of condensation depends upon the magnitude of the compressive force; and, as this force decreases rapidly towards the surface, where it vanishes the density of the liquid decreases also. M. Poisson has shown that, when this force is omitted, the capillary surface becomes plane, and that the liquid in the tube will neither rise above nor sink below the level of that in the cup. In estimating the forces, it is also necessary to include the variation in the density of the capillary surface round the edges from the attraction of the tube. The direction of the resulting force determines the curvature of the surface of the capillary column. In order that a liquid may be in equilibrio, the force resulting from all the forces acting upon it must be perpendicular to the surface. Now it appears that, as glass is more dense than water or alcohol, the resulting force will be inclined towards the interior side of the tube; therefore the surface of the liquid must be more elevated at the sides of the tube than in the centre in order to be perpendicular to it, so that it will be concave as in the thermometer. But, as glass is less dense than mercury, the resulting force will be inclined from the interior side of the tube (N. 174), so that the surface of the capillary column must be more depressed at the sides of the tube than in the centre, in order to be perpendicular to the resulting force, and is consequently convex, as may be perceived in the mercury of the barometer when rising. The absorption of moisture by sponges, sugar, salt, &c., are familiar examples of capillary attraction. Indeed the pores of sugar are so minute, that there seems to be no limit to the ascent of the liquid. Wine is drawn up in a curve on the interior surface of a glass; tea rises above its level on the side of a cup; but, if the glass or cup be too full, the edges attract the liquid downwards, and give it a rounded form. A column of liquid will rise above or sink below its level between two plane parallel surfaces when near to one another, according to the relative densities of the plates and the liquid (N. 175); and the phenomena will be exactly the same as in a cylindrical tube whose diameter is double the distance of the plates from each other. If the two surfaces be very near to one another, and touch each other at one of their upright edges, the liquid