All the devices which I have quoted for picturing to ourselves the architecture of a crystal I would regard as merely models representing something that may be really quite different. But I venture to suggest that the time has come when we should make use of moving and not stationary models. One need not go further than a spinning top for an illustration of stability due to movement, and there is nothing unreasonable in the suggestion that the rigidity of a crystal structure may be due to the motion of its parts. One curious observation which I have made is suggestive. I have many times noticed that when the appropriate crystal is introduced into a supersaturated solution, which is not strong enough to crystallize spontaneously, it may cause crystals to grow not only in actual contact with itself, but at some little distance in its neighbourhood. If this be so, then the crystallizing force, the power of propagating crystal growth, is not merely a frontier problem, but can be exercised through the liquid to a distance. If I try to picture to myself what is happening, I must again have recourse to analogy; I can only think of the manner in which a string or a tuning-fork is set in vibration and responds to a similar string or tuning-fork which is giving out its note at the other end of the room; and so is it not possible that the movements, whatever they are, vibrations or pulsations or regular oscillations of some sort, which constitute crystalline growth may be communicated through the almost crystallizing liquid, and culminate at some point where they set similar material vibrating, that is to say, crystallizing, in the same way?