With regard to the _t_, it will be recollected that Grotefend gave the value of _n_ to a sign that completed the word ‘bun,’ to which he gave the meaning ‘stirps.’ This word had long been a stumbling-block to Zend scholars, and Lassen determined to get rid of it. He showed in the first place that the _b_ or _p_ at the beginning could not be interchangeable, and the word must at all events be treated as ‘pun’; but he proposed to alter it still further by reading ‘put.’ By this means he came nearer to its obvious meaning, ‘son’—that is, to the Zend ‘putra.’ He found this innovation supported by another word, _k_ _s_ _t_ _m_, to which he thought he could attach a Zend meaning.[460] The nine incorrect values he admitted into his alphabet[461] show little or no improvement on those suggested by Grotefend or Burnouf; and unfortunately the decipherer himself can rarely distinguish the incorrect values from the correct. A glance over a page of Lassen’s transliteration will show the havoc these nine incorrect letters have made in his work. But, as we have said, he introduced errors peculiar to himself that were even more fatal than his failure to identify all the signs correctly. For example, he remarked that the sign he took for a short _a_ (𐏃) seemed composed of the sign for _n_ (𐎴) and an angular wedge which might be an abbreviation of the sign itself. He was led to this hypothesis by comparison with the Zend, where _ă_ is clearly a combination of _a_ and _n_.[462] He goes farther and gives the short _a_ and _n_ the guttural sound of _ang_, when it is found before the letter he thought was _h_ (𐎹, 27; really _y_), and he cites several instances which he thinks will justify this opinion. He recognises, however, that the rule even thus limited is not always applicable.[463]