It was not till the following year that Mr. Loftus succeeded in excavating a building almost exactly similar to the Columnar Edifice at Persepolis. He determined the position of twenty-one bases of the central group; two bases in each row of the eastern colonnade, and three of the western. On the north he found three bases, all in the inner row, and it is still doubtful if there ever was a second row on this side.[214] He ascertained that the building consisted, as at Persepolis, of a central square of thirty-six columns, surrounded on three sides by a colonnade, and we are indebted to him for the measurements. His opinion is that the central group was roofed, but not enclosed by a wall, and the space between it and the colonnades was open.[215] He searched in vain for the traces of walls such as Fergusson suggested had existed at Persepolis; and was the more convinced that none had ever existed because he found distinct traces of foundations elsewhere. He brought to light a trilingual inscription, repeated four times on the bases of the columns, which were found to have been written by Artaxerxes Mnemon (Inscr. S). They are of more than usual interest, for the King traces his genealogy back to Hystaspes, and confirms thereby the statement of Herodotus. He states also that he built the hall, or, as he calls it, the Apadana, on the site of an earlier edifice erected by the great Darius, and afterwards destroyed by fire during the reign of Artaxerxes I. He likewise invokes Mithra and Anahita for the first time side by side with Ormuzd as tutelary deities. Another evidence of degeneracy is seen in the corruption of the language, which exhibits several grammatical solecisms. Another short trilingual legend of the same king was found round a column in a different part of the mound, and several detached bricks and vases with the names of Darius and Xerxes, but no other trilinguals of importance. On the other hand, the long inscription in thirty-three lines found by Rawlinson on the Citadel Hill, and the two inscriptions found by Layard at Malamir, gave rise to fresh difficulties. It was recognised that the writing was different from any yet known; and the perplexity was heightened when it was farther observed that they differed from one another. Here, then, were two new methods of writing, and possibly two new languages added to those already in hand; and there seemed to be no end to the task imposed on the cuneiform student. For some time little effort was made to grapple with these new problems. The script found on the Citadel Hill received the provisional name of ‘Old Susian,’ and many other specimens of it gradually accumulated. Subsequent investigations have shown that the writing and language found at Susa and Malamir are related to those in the second column of the trilingual inscriptions. It is now ascertained that the Old Susian is the most ancient form; and that the script and language of the second column descends from it,