Paul de Man once remarked that what most interested him in Peter Szondi's Traktat Uber philologische Erkenntnis-that methodological position paper that remains, in my opinion, his most eloquent and capacious statement-was not general argument about place of hermeneutics in contemporary German literary studies, but rather more specific question Szondi raised of whether a passage in Hl61derlin was metaphorical or not, suggesting that this question was most important one, in Holderlin or with any other literary text. Years after it was written, Szondi confirmed this interest when he cited Traktat while asking whether the ambiguity is one of matter itself (einer der Sache selbst).' I would like my essay to stand under sign of this question of metaphorics, of metaphorics in Szondi's work as well as in literary texts themselves, for I believe it will appear to be a crucial aspect; just as I would like to direct toward Szondi's studies of German romantic genre theory2 same relating of what we know from and of his work (Wissen) back to specific acts of knowing (Erkenntnisse) that he advocated, in Traktat, for literary studies in general.3 What we know of Szondi's interpretations of German romantic genre theory is certainly less here in United States than in Europe, for they are scarcely known at all. They can be summarized as arguing for a progressive development of poetics of genre, from their