For the conductor of the ‘paratonnerre’ lengths of iron bars are principally used; formerly these were jointed together by means of a pyramidal bolt let into a notch of the same form, and connected by a simple iron pin. This method, however, was discovered to be very bad, as it failed to preserve the continuity of the conductor after it had been erected a little time. The following plan, as represented in fig. 13, is now used for the best work, as being more durable and affording a better contact. On each side of the bars to be joined, two flanges, about six inches long, and half the thickness of the bars, are filed out. A thin piece of carefully-prepared lead is then placed between them. The whole is then firmly fastened together by bolts at B and B and completely covered with pewter solder, and thus furnishes a solid, durable contact which possesses very small resistance. Formerly the conductors were, at regular intervals, rivetted to cramps let into the wall for the purpose of retaining the conductor in its place. As this plan left no room for the play of expansion and contraction caused by variations in the temperature, it was found that at times the conductor was very much strained and even bent by reason of this expansion and contraction. To avoid this evil an apparatus, which has been approved by the Paris Academy of Sciences, has been substituted for the cramps and rivets. This apparatus consists of a fork in which the conductor is held fast by a pin (see fig. 14). Being able to move backwards and forwards in the fork with great facility, the conductor is thereby permitted to expand or contract under the influence of temperature without threatening its supports with destruction. [Illustration: Fig. 14.] The question however arises, upon what part of the paratonnerre ought the effect of such contraction or expansion to be borne? The Paris Academy of Sciences has sanctioned and recommended the use of a compensator, which is designed to bear this strain. This compensator, which is now much used in France, may be seen in fig. 15. It is composed of an elastic plate F, made of well-annealed red copper, three-quarters of an inch wide, at least twenty-eight inches long, and about a quarter of an inch thick. The two extremities of this plate are firmly fastened to the two ends of two lengths of the conductor by the bolts and counterpieces B B´, and afterwards covered with a thick coating of pewter solder. When, in consequence of the heat, the conductor expands, the curve of the copper plate F will become greater, and in cold weather it will become less. As a rule, a single apparatus is supposed to compensate for the effects produced by long straight lengths, and it is therefore thought sufficient to place one at each bend.