¹ The section used for this analysis does not include that portion of the original which is thought to have been written by a different author from the rest of the text. This article is expanded from a paper read at the meeting of the Linguistic Society of America in Chicago, December 1950.

² This paper does not treat problems that must be solved by purely historical evidence. In fact, one of the reasons for undertaking the paper is to see what kind of analysis can be made of an Old Bulgarian document by descriptive techniques alone, and to note the problems suggested by historical considerations on which there is no descriptive evidence. Thus, the combinations rǔ and lǔ between consonants may, according to the historical evidence, each represent two or three phonemically different entities; but since only one writing occurs in the ms., we cannot infer anything about such differences on a descriptive basis alone.

³ An example is wbače ~ obače.

⁴ An example is iskariotǐskaago. Except in foreign words, the symbol o does not occur after any other vowel.

⁵ A. Leskien, Handbuch der altbulgarischen Sprache⁶ 8 (Heidelberg, 1922); N. van Wijk, Geschichte der altkirchenslavischen Sprache 41 (Berlin and Leipzig, 1931).

⁶ Van Wijk 43 and 108.

⁷ André Vaillant, Manuel du vieux slave 1.59–60 (Paris, 1948).

⁸ Vaillant 1.30.

⁹ All the combinations of consonant and vowel symbols implied by this table occur in my data, except žü, ʒa, ʒě, ʒü, and ʒǐ. I assume that ž patterns like š and that ʒ (a symbol of very rare occurrence) patterns like c.

¹⁰ A similarly inconsistent arbitrary choice is found in several instances in Russian.

¹¹ This also is similar to the Russian writing system, where, after certain consonants, both o and e are written—the choice depending not on any phonemic difference but on morphology (although it is not quite as regular as in the Zographensis).

¹² In some cases where we would historically expect the writing with a hook, the hook is missing. Since certain forms occur both with and without the hook, the scribe may sometimes have been careless in his writing. This does not detract from the force of the pattern or from the analysis made here; but to determine the phonemic status of an individual form written only without the hook, historical evidence is needed.

¹³ In my material the only blank in the pattern is that no example of ʒǫ or ʒę occurs. Again I assume that ʒ patterns like c.

¹⁴ If ǫ and ę contrasted only after the hush-type sibilants, it might be reasonable to assume the phoneme /y/ here also, and to analyze šę as /šę/ and šo as /šyę/. This is not possible, however: the contrast between jǫ and ję after a vowel makes it necessary to assume /ǫ/ and /ę/ in contrast after /y/.

¹⁵ It is possible to devise a solution with [ǫ] and [ę] as allophones of the same phoneme, analyzing lǫ as /lę/, lę as /l_Ję/, l⁺ę as /l_Jyę/, and l⁺jǫ as /lyę/. There is, however, no other descriptive evidence to indicate that /y/ occurs after a non-palatal consonant, and there is descriptive evidence for the contrast between [ǫ] and [ę]. The historical evidence would also make this theoretically possible analysis an undesirable one.