observe how nearly he approached the correct value of Grotefend's h (𐎹, No. 27). He perceived that if it were changed into a y, it would yield yuna in the twelfth line, which there could be no doubt would indicate 'Ionia.' As it is, however, he retained the incorrect value; and he could find no satisfactory explanation of huna; for he, of course, rejected 'Huns' as an evident anachronism.[1] It would be tedious and unnecessary to go through the other signs to which he gave new values, for they unfortunately all turned out to be wrong. Indeed, if his services to decipherment were to be estimated by this test alone, they would not rank higher than those of St. Martin or Rask; for although he lays claim to have ascertained the value of twelve characters, eight of these are erroneous, one (the b) fairly belongs to Münter, another (the α) to Grotefend, and only two remain to be placed to his own credit: preci8ely the same number as were contributed by St. Martin and Rask. His alphabet gives definite values to thirty cuneiform signs and an uncertain value to three others.[2] Following the analogy of Zend, he allots a separate sign to the long and short values of each of the vowels α, i, u, and in this he considers he has reached 'a result that should satisfy criticism.' With respect to the consonants, however, he agrees with the maxim of Rask, and strives as far as possible to avoid according more than one sign to each. He has, however, found it difficult to avoid giving two signs to land h, and no less than four to gh. As regards l or h, he introduces the second signs apologetically, followed by a mark of interrogation, indicating that they may be variants or defective signs. We now know there is no well- authenticated l in the language, and his first sign
