the third Persepolitan column. Meanwhile, he devoted his attention exclusively to the simplest form of writing, which is found in the first column; and he speedily recognised that the diagonal wedge which occurs so frequently was evidently intended to separate one word from the other. He compared it to the cypress tree that divides the groups in the procession on the sculptured staircase seen at Persepolis; and adds that in one of the old Hindu alphabets the words are similarly separated by a small oval.[261] This discovery, now announced for the first time, had till then escaped the observation of Tychsen, who, it will be remembered, fancied he found three different words enclosed within the same diagonals. In order to find values for the cuneiform letters, he had recourse to a twofold method. He sought out the signs that recurred the most frequently and that were the most uniformly repeated in the same word, for he concluded that these would naturally turn out to be vowels. He soon identified three in particular (𐎠, 𐎹, 𐏂) that were constantly recurring: the first in almost every word, and occasionally several times in the same word.[262] In the inscriptions analysed, he found that the first was repeated 183 times, the second 146, and the third 107 times.[263] He then proceeded to compare the forms of these letters with those of the vowels in other kindred languages; and he thought he discerned a strong resemblance between the first and third in the Zend character for _a_ and in the Armenian for _o_.[264] He could not find a letter anywhere that resembled the second. However, he observed another cuneiform sign that also recurred with great frequency (𐎶), and which might easily bear comparison with the Zend letter for _a_ long.[265] A likeness between a defective cuneiform sign ([Persian]) and the Zend letter for _i_ gave him a fourth vowel. Similar considerations led him to assign the values of _ou_ _w_ _ii_ _y_ to another sign (𐏁), a conjecture that turned out to be less happy than that of Tychsen, who accidentally hit upon its correct value, _s_. Münter had now pointed out six signs he thought expressed vowels (viz. 𐎠 _e_ or _a_, 𐎹?, 𐎡 _o_, 𐎶 _a_, [Persian] _i_, 𐏁 _ou_, &c.). The second he dropped out of his alphabet, for after careful search he could find no letter in any other alphabet to give him a clue to its value. The fifth was not a genuine letter; and of the four that remained two were, as he surmised, vowels (Nos. 1 and 3). The other two were both consonants. Only one of the former—the first, _a_—was found to be correct; but it had already been recognised by Tychsen. The other—the _o_—was afterwards found to be _i_. He also identified seven other signs with the consonants _p_, _kh_ (two), _r_, _r_ strong, _s_ and _b_, which he obtained by the simple process of comparing them with other letters found in Zend, Armenian and Georgian, to which they had ‘no small similarity.’[266] Three of the signs he