If we could stretch an apparently impervious membrane, like the inner white skin just inside an eggshell, or a piece of parchment, and so form a wall through the middle of a glass box, and then pour into one of the compartments pure water and in the other a mixture of water and molasses, a very curious result would follow within a comparatively short period. We should find that presently there would be a gentle filtering of the water through the membrane toward the molasses water, and similar gentle current in the other direction. In other words, fluids of different density, if separated by a membrane, tend to equalize each other. This equalization may not be very rapid, and at first it will be more speedy from the less dense to the more dense, but eventually it will make the different fluids of a common density. This purely mechanical property of the equalization of fluids separated by a membrane is known as _osmosis_, and it is upon the possession of the equipment necessary for this that roots depend for getting food and water from the soil. In our discussion of roots in Chapter I, we found that they end in very fine subdivisions, which are themselves split up into practically invisible root hairs. These root hairs are the only way that roots can absorb the food and water in the soil, and they are able to do this because they are provided with a membrane which permits osmosis to act between the solution inside the root hair and the water in the soil. The solution in the root hair is mostly a sugary liquid, some of that surplus sugar made in the leaves, and it is denser than the soil water, so there is apparently nothing to prevent an equalization of the liquids on different sides of the membrane. If this actually happened, as it would in the case of the simple experiment noted above, then roots would exchange a fairly rich sugary liquid for a much more watery one, and we should find that plants did not get their food from the soil, but really have it drained away from them by osmosis. But nature has a cunning device for stopping such robbery, which is prevented by the membranes of root hairs being only permeable to the extent of letting water _in_, not permeable enough to allow sugar to escape. As we have seen, osmosis is a purely mechanical process which, if left to operate without interference, would not aid but injure the plant. Surely, nothing with which plants are provided is so important to them as this delicate membrane of the root hairs which, while allowing osmosis to act in a one-sided fashion, preserves to the plant the sugary liquid that alone makes the absorption of soil water possible.