In trying to decipher texts of which we know neither the language nor the values of the written characters it is evident that the first steps must be taken by conjecture. It is only when the conjectures corroborate one another that progress can be made. I have tried to restrict my guesses to what is in itself probable, and had not intended to publish them without further investigation. But investigation takes time, and for special reasons it seemed best to print now such of my results as appear to be fairly well justified. It is unnecessary to criticize other attempts at decipherment, as I have tried to start entirely afresh, without a prejudice for or against any previous system. Several of the identifications of Professor Sayce, the pioneer in this as in many other lines of research, have, however, been adopted without discussion, because they appeared on consideration to be certainly right. Much else may be derived unconsciously from him, for we have often discussed the problem together, and but for his encouragement I should long ago have given it up altogether. The notes relate almost entirely to texts in Messerschmidt's Corpus, and only to the more legible of those. A valuable addition to the material has been made by the publication of excellent facsimiles of inscriptions at Carchemish (Jerabis) by Hogarth, Lawrence, & Woolley, with which I hope to deal on another occasion.