Two joining fragments preserving the ends of three lines of a letter. Of line 1 only bottoms survive. Lines 2-3 read: ].ndie · cum / ].aseris. Neither pridie nor meridie looks likely (and cot(t)idie is impossible); interpunct after this looks clear. There may be traces of a line or two below line 3. On the back a line in address script: .a...[; Flauio [ may be possible if we suppose that almost all trace of l has disappeared. Below this a line written on the slant, therefore the name of the sender: .pro...o[; a Proculo is likely (cf. 219).