Such was the state of the inquiry when Grotefend entered upon it. The Memoirs on the Antiquities of Persia, published by M. de Sacy in 1793, afforded a sort of text-book to the decipherer. De Sacy had succeeded in reading some inscriptions at Naksh-i-Rustam, written in Pehlevi. Like those at Persepolis, they were engraved above the sculptured representations of kings, and they were found to contain the royal name and title. Grotefend inferred that the cuneiform inscriptions had very probably served as models for these later legends. The simplest of these and, from its brevity, the one that afforded the most striking resemblance to the B and G of Niebuhr ran: ‘N N rex magnus rex regum [rex-um] Filius ... [regis] stirps Achaemenis[?]’[304] The first step Grotefend made in advance of his predecessor was to perceive that it required two words to make up the phrase ‘king of kings,’ and that these two words no doubt corresponded to the two in the cuneiform: the one with seven letters and the longer form of the same word that followed it. This apparently obvious necessity had, as we have seen, wholly escaped Münter. When it was once recognised that the word of seven letters was clearly ‘king,’ it became obvious that, according to the analogy of the Pehlevi model, the first word of the inscription was the name of the sovereign, the third a qualifying title corresponding to ‘magnus’ and the two following, where he found the word of seven letters again repeated, and on this occasion followed by the longer form, evidently corresponded to ‘rex’ and ‘rex-um’ (‘regum’).

Comparing the two inscriptions (B and G), he found they began with different words, which he now inferred were the names of two different kings; but he observed that the name in B, which was presumably in the nominative case, also occurred in the third line of G with a case-ending, followed by the word for ‘king.’ also with a case-ending. The termination differed from that already observed in the phrase ‘king of kings,’ and it marked, no doubt, the genitive singular, as the other denoted the genitive plural. Referring to his Pehlevi model, he inferred that the passage indicated the relationship of the two monarchs, and that the king of the second inscription (G) here declared himself to be the son of the king of the first inscription (B). This little bit of ingenuity solved the whole mystery. In the corresponding place in B he found another word in the genitive case, which was no doubt the name of the father of the king of that inscription; and he remarked that this name was not followed by the royal title. He had thus discovered the cuneiform signs that, with little doubt, expressed the names of three Achaemenian princes, and he had recognised that these personages stood to each other in the relation of father, son and grandson, and that the first was probably not of royal rank. That is to say, from ‘G’ he found that ‘King Z’ was son of ‘King Y’; and from ‘B’ he found that ‘King Y’ was son of ‘X’ without the addition of ‘King.’ It only remained to determine who these three princes were most likely to be: and as the Achaemenian dynasty was a short one and their names already known from history, the task was not a difficult one. The two kings at the head of the inscriptions could not be Cyrus and Cambyses, because their names did not begin with the same cuneiform letter; they could not be Cyrus and Artaxerxes, because there is no such discrepancy in the length of the cuneiform words. There thus only remained Darius and Xerxes for the names occurring first in the two inscriptions; and this result was confirmed by the absence of the royal title from the name of the father of the king in one of the inscriptions, for he recollected that Hystaspes is not called king by the Greek writers. He assumed, therefore, that the first word in B was Darius (and it must have been satisfactory to notice that the second letter, a, was precisely the a of Münter) and the other Hystaspes; while the first word in G he assumed was Xerxes.

From these three known words he now set himself to get at least the approximate values for the letters they contained. According to De Sacy, Grotefend first transliterated Darius and then Xerxes, from which two names he obtained the word for ‘king,’ and finally he transliterated Hystaspes. But according to Grotefend’s own account of the matter, he fastened in the first instance on the word that should read ‘Hystaspes.’ It consists of ten cuneiform signs, including the inflexion. He learned from Anquetil’s Zend-Avesta that the Zend form of the name was Goshtasp, Gustasp, Kistasp. Placing a letter of the name under each cuneiform sign, he arrived at the following result:

𐎻·𐎡·𐏁·𐎫·𐎠·𐎿·𐎱·
Goshtasp.

Here, then, were seven letters of the cuneiform alphabet for which values were provisionally assigned and three left over for the genitive termination. The word for Darius also consisted of seven letters, which he at first read thus:

𐎭·𐎠·𐎼·𐎹·𐎺·𐎢·𐏁·
Dar——ush.

The process so far was confirmed by the repetition of the same letters a and sh in both words in the position in which they were to be expected. There was more difficulty with Xerxes. The cuneiform word consisted of seven letters:

𐎧·𐏁·𐎷·𐎠·𐎲·𐏁·𐎠
—sh—er[305]she

of which he already knew, or guessed, five, and these known values occurred in the order he expected; the first and third letters remained to be determined. It happens that Herodotus mentions that the name of Xerxes corresponded in sound to that of the Persian for ‘warrior’ or ‘king’; and Grotefend noted that the first two letters in the words for ‘Xerxes’ and ‘king’ were the same in the inscriptions. He ascertained that the Greek letter ξ transliterates the Zend ‘kshe’; but he could find nothing in the Zend vocabulary under ‘kshe.’ There were, however, several forms under ‘kh,’ ‘sh,’ which left no doubt that the first letter required should be read ‘kh.’ This assumption also enabled him to read the word for ‘king’ which had so long attracted attention. Of the seven letters that composed it he now knew four, which occurred in the order