I maintain that of the two theories the second is just: that a gradual experimental growth in its roads, a method coincident with local caprice, burdens with imperfect communication the society adopting it; that conscious design is essential to efficiency. And this I propose to illustrate by a single example. Take two points A and B, such that a line joining them must lead across a marsh, a river, and a range of hills. Let some primitive wanderer make his way from A to B, knowing, when he is at A, the direction of B by, let us say, a distant peak overtopping the range between. That primitive wanderer would first of all skirt about the marsh and, finding its narrowest place at C, would set to work and make his causeway there. Having crossed it, he would come to the river. He must either swim or ford it. Supposing him to prefer, through the necessity of a pack or what not, to ford it, he casts about for a ford. He finds one at D, and perhaps he also, if he takes time to look about him, finds another deeper one at E and another at F, but as his causeway is near D he takes that ford. [Illustration: Sketch I.] Then he has to make for the hills. We will suppose that the peak directing him from beyond B is still visible. He takes his new direction from it and looks towards the base of the hills at G. There, in the direct line to the peak, the contours are so steep that the trouble of getting up would more than counterbalance the shortness of the cut. He casts about for a better chance, and at last finds a gradient just worth his while at H. He climbs up that; but though the gradient is easy on the A side at H on the far side it is very difficult, so he turns along the ridge to K, where he finds an easier down gradient: a spur leads him on by its gentle slope, and from the bottom of the spur he makes straight for B, which is now right in front of him and plain sailing.