Mr. Thornhill convinces himself by inspection that the first proposition is untenable; in fact its absurdity is at once apparent on a little thought, for the engraving of the original die is a laborious and costly piece of work, and that very fact, coupled with the comparative ease of exact reduplication by mechanical processes on the printing plate, furnishes the chief reason for the employment of this method of producing stamps. Since there is such a variety in the size of the stamps, therefore, the first theory would indicate many original dies, and this we know was not the case. Its refutation indeed is seen in the stamps themselves; for each original die, if differing in size from its fellows, meant a separate engraving, and it is humanly impossible to make these separate engravings exact duplicates, whereas, on the other hand, no appreciable variation in line or dot can be detected on the same stamp in its different sizes save the general expansion or contraction of the design, which is proportionate in all its parts. The different die or matrix theory is therefore thrown out on grounds of impracticability and absurdity. Accepting the one original die proposition, then, Mr. Thornhill agrees with Mr. Tapling in turning down the shrinkage of paper theory and favoring the second supposition, that the variation comes on the plates and is due to the process of transference. Let us glance at this a moment. The original die is engraved on a block of soft steel of very fine and even quality. When finished it is tempered to a very great degree of hardness. Next the engraving is transferred by tremendous pressure to a transferring roller of similar soft steel, which is in turn hardened. In this process there might be an opportunity for a slight variation in the size of the transferred impression, due to the expansion and contraction of the steel in the tempering process. Next, this hardened transfer roller is impressed upon the printing plate of soft steel as many times as there are copies desired. These naturally all agree among themselves and with the transfer roller impression in size. Now when the printing plate in turn receives its hardening, there may again be a chance for a slight difference between the transfer roller and the plate impressions; _but_ it is wholly unlikely that the plate impressions will vary much among themselves, otherwise the perfection of Mr. Jacob Perkins' invention, the chief merit of which was exact reduplication, would be impaired. As a matter of fact, the high grade and even quality of the steel necessarily employed, and the care naturally taken in hardening the plate, preclude any other than an even variation, if any, due to the tempering process. This means that such variations would be practically constant over the printing surface of the plate, and that therefore the impressions would still remain practically identical in size.