line to the end, besides numerous occasional gaps elsewhere. But the second column is in a much worse condition. 'A fissure, varying in breadth, caused by the percolation of water, bisects it and destroys the continuity of the writing throughout its whole extent.'[1] The third is nearly perfect, except at the bottom, where several lines are wholly lost. The fourth column is worse than the second; 'a fissure transects the tablet longitudinally,' and in the lower half 'the rock is more or less broken by the trickling of the water.' But when we come to the fifth, we find 'a state of such deplorable mutilation that it would be waste of time and ingenuity to undertake an analysis of the text, or to attempt anything like a connected and intelligible translation.' In the face of these difficulties he was obliged to have recourse to very elaborate and ingenious restorations. At the end of the second column, for instance, he found the Susian copy perfect, and this enabled him to 'restore' the Persian text. It is one of the first instances of a long translation from the Susian, and his version of it turned out afterwards to be correct.[2] In the numerous repetitions that occur so frequently, he found a safe guide in other passages of the inscriptions. Sometimes, however, he had to work on much less solid foundation, as when he sought help from other sentences that were only 'of nearly similar construction'; or when his restoration was 'generally borne out by the context'; or merely by considerations of 'grammatical propriety.' In such cases he could never arrive at more than a high probability. He had frequently to measure the length of a blank and then tax his memory to supply one or more words with the required number of letters that would fit into the vacant place and at the same time make
