Some of the letters added to the alphabet by the Greeks (xi, phi, chi, psi) have caused much perplexity. In the Western Greek alphabet xi, phi, chi, with the forms X 4) Y, come at the end of the alphabet. In Eastern Greek xi, with the form I, the Semitic samekh, comes in the same place as the Semitic letter, after nu; phi, chi, psi, with the forms ) X 'Y, come in that order at the end. The confusion has given rise to all sorts of explanations. By some attention to the names of the letters, a little more light can be obtained. First of all, however, it must be pointed out that there must be a definite relation between the Eastern and Western alphabets as regards these added letters. The similarities, partial or complete, of forms, function, position, and name make that certain. The fatal weakness in Nilsson's treatment of these letters in his otherwise admirable article' is the implicit assumption that these letters developed independently in the two main branches of the Greek alphabet. It is obvious that each of the four letters is named in the same way, by attaching one and the same vowel2 to the consonant sound which the letter represents. This is true of no other letter of the alphabet except pi, and here the name is a natural development of the Semitic name pe. It seems, therefore, that pi is the prototype of the others. But why was pi picked out rather than nu, rho, or any other letter? Was the first of the new letters the xi, for example? If so, why was it named on the analogy of pi, with which it has nothing in common? Scholars have noted this fact and have tried to explain it by the nearness of xi to pi in the alphabetic order. But mere nearness, without actual juxtaposition, is a conclusive argument against rather than for this explanation. It will be noted that the four added letters consist of two aspirated mutes (chi, phi) and these same mutes followed by s (xi, psi), or, putting it another way, of the palatal surd k followed by 1 Tbernahme u. Entwickelung d. Alphabets durch d. Griechen, Kgl. Danske Videnskabernes Selskab, Hist.-fil. Meddelelser, I (1918), 6. 2 For our purpose it is immaterial whether the spelling 41?, etc., or 41s, etc., is used.