The earliest attempt at this was to use a reflector of bright polished metal. In the most improved form these were made to that peculiar curve known as a parabola. This is a curve obtained by cutting a cone in a certain way, wherefore it is one of the "conic sections," and its particular appropriateness for this work resides in the fact that if a light be placed at a certain point known as the "focus" all the diverging rays which fall upon the reflector will be reflected in the same direction, parallel to each other. An ordinary spherical mirror would reflect them either back to the lamp or in diverging directions. At any distance the beam from the parabolic reflector will be more intense than that from the spherical one, since the rays will be closer together. But even with the parabolic one there is some diffusion, for the simple reason that whereas the focus is a mathematical point (position without magnitude) the most concentrated form of light known has a considerable magnitude. Hence the rays proceeding from the centre of the mantle are reflected as per the theory, but those from the outlying parts of it are somewhat diffused. This difficulty cannot possibly be overcome, and hence even in the finest examples of lighthouse architecture the flashes are not quite sharp and clear-cut. There is a central moment, so to speak wherein the flash is almost blinding in its intensity, but it is preceded by a period of growing brightness and succeeded by one of decreasing light. In the modern apparatus, however, metallic mirrors are entirely dispensed with, their place being taken by reflecting prisms of glass. The metallic ones had to be continually rubbed to keep them clean, and this soon dulled their brightness, while the glass prisms need only to be wiped carefully, which operation has little effect upon their surface. It may come as a surprise to some that reflecting prisms are possible. The idea of refraction through a prism is quite familiar. Such forms the essential principle of the spectroscope. Refraction is explained to every school child in order to account for the rainbow. But _reflection_ by a piece of the clearest glass seems a contradiction in terms almost. Yet it is only a question of shape. In some prisms the light is simply bent as it passes through. In others it is bent twice, so that it leaves the prism just as if it had been reflected off a mirror. Both devices are used in the lighthouse. Let us see how they are combined so as to perform the work to be done. Take first of all the case of a light upon an isolated rock where the warning is needed equally all round. All that is necessary here is to pick up those rays which, if left to themselves, would fall upon the water near the foot of the tower, and those which would waste themselves skywards, and then to gather all the rays into several bundles or beams. We will suppose a simple case in which the light is supposed to give flashes at regular intervals.