If instead of a hill or building we erect a solid rod of metal, G H, then the field will be distorted as shown in Fig. 4. Now, it is quite evident that whatever be the relative distance of the cloud and earth, or whatever be the motion of the cloud, there must be a space, g g', along which the lines of force must be longer than a' a or H H'; and hence there must be a circle described around G as a center which is less subject to disruptive discharge than the space outside the circle; and hence this area may be said to be protected by the rod, G H. The same reasoning applies to each equipotential plane; and as each circle diminishes in radius as we ascend, it follows that the rod virtually protects a cone of space whose height is the rod, and whose base is the circle described by the radius, G a. It is important to find out what this radius is. [Illustration: Fig. 5] Let us assume that a thunder-cloud is approaching the rod, A B (Fig. 5), from above, and that it has reached a point, D', where the distance. D' B, is equal to the perpendicular height, D' C'. It is evident that, if the potential at D be increased until the striking-distance be attained, the line of discharge will be along D' C or D' B, and that the length, A C', is under protection. Now the nearer the point D' is to D the shorter will be the length A C' under protection; but the minimum length will be A C, since the cloud would never descend lower than the perpendicular distance D C. Supposing, however, that the cloud had actually descended to D when the discharge took place. Then the latter would strike to the nearest point; and any point within the circumference of the portion of the circle, B C (whose radius is D B), would be at a less distance from D than either the point B or the point C. _Hence a lightning-rod protects a conic space whose height is the length of the rod, whose base is a circle having its radius equal to the height of the rod, and whose side is the quadrant of a circle whose radius is equal to the height of the rod._ I have carefully examined every record of accident that was available, and I have not yet found one case where damage was inflicted inside this cone when the building was properly protected. There are many cases where the pinnacles of the same turret of a church have been struck where one has had a rod attached to it; but it is clear that the other pinnacles were outside the cone; and therefore, for protection, each pinnacle should have had its own rod. It is evident also that every prominent point of a building should have its rod, and that the higher the rod the greater is the space protected. * * * * * PHOTO-ELECTRICITY OF FLUOR-SPAR CRYSTALS.