Appendix A
Artemidorus, Oneir . 1.56

Only a year after Roger Pack published his Teubner edition of the Oneirocritica of Artemidorus in 1963, Toufic Fahd produced the text of an Arabic translation of this work from the Sultan's library in Istanbul. Although the translator did not always fully comprehend the Greek in front of him, it is usually possible to deduce what his Greek text looked like, thus providing a witness considerably earlier than the two manuscripts (V and L) on which Pack had been obliged to rely. For the Arabic version dates, in the opinion of virtually everyone, to the great days of translation at Baghdad in the ninth century.

In responding to this new witness, as well as to reviews of both his own edition of the Greek and of Fahd's of the Arabic, Pack acknowledged the dramatic impact upon the study of Artemidorus: "On Artemidorus and His Arabic Translator," TAPA 98 (1967), 313-26; and "Artemidoriana Graeco-Arabica," TAPA 106 (1976), 307-12. Since the Arabic version now clears up several difficulties in a passage that is relevant to the argument in the first chapter of the present book, it would seem desirable to lay out the evidence here. It has not been assembled before.

Oneir . 1.56 as printed by Pack in 1963, p. 63, l.15 - p. 64, l.1:

[Full Size]

This passage appears in Arabic in Toufic Fahd, Artémidore d'Éphèse —Le livre des Songes , Institut Français de Damas (1964), p. 117, l. 12 - p. 118, l.8. By a regrettable slip Fahd omitted a substantial chunk of the Arabic text in this section, as Franz Rosenthal noted (JAOS 85 [1965], 142, quoting the missing words from the manuscript). Fortunately the words that Rosenthal has restored do no more than reassure us that the translator did not actually omit anything. The rendering (of lines 3-4, from to , inclusive) is unproblematic.

The Arabic can serve to correct Pack's Greek text at four points.

1. Line 1, . Contrary to his normal (and reasonable) instinct in preferring V to L, Pack adopted the reading of L here in printing . He evidently did so because V and L agree in ll. 7-8 in , although it is difficult to fathom what Pack thought the word could mean in either place. In a paraphrase in TAPA 106 (1976), 309, he wrote of "scenes," which seems inadmissible philologically as well as hard to construe with in the obvious sense of "possess." The Arabic renders the whole phrase by "possess a book of recitation pieces or a book of fabrication [sanaca^[*] of poetry." Apart from the equation, made repeatedly,

of tragedy with recitation and the assumption that verse is in question, it is absolutely clear that the translator saw and rendered it literally. When he saw , as he did in the later line (where both extant independent manuscripts have it), he translated dalil , another literal rendering.

2. Line 4, . Both V and L have , and Pack had opted for Hercher's conecture. But the Arabic gives ("troubles, misery") at this point and proves Hercher wrong. Pack, in TAPA 106 (1976), 309, proposed , despite objections that he candidly acknowledged. But if we remember the frequent interchangeability of o and w in later Greek and the pertinence of this fact for orthography, we can see at once that V and L have the right word after all. It is simply (or ).

3. Lines 7-8, . Here the Arabic translator certainly had , but apparently not . His version is "possess a [comic] representation [dalil ] in a book [fi kitabi ]." Now, as we have already observed, makes little sense here, and the parallel with the lines on tragedy would suggest the disjunction . The words , if that is what the Arabic translator saw, must have been a gloss to explain . Reiske had already proposed deleting in V and L for that reason. The gloss would have come when had fallen out, perhaps creating the prefix ANA in the process: <<D PAMATAH>PL AS MATA.

4. Line 9, . This is the reading of L, whereas V has . The Arabic has clear correspondences for the first three nouns in V, and only for these (al-madhmuma wa-l-idtarab wa-l-khusuma ). The word must represent a gloss on these three terms that found its way into the text between the translator's manuscript and V.