1 Le Périple de la Mer Érythrée, Göteborgs Högskalas Årsskrift xxxiii (Göteborg 1927). A translation based on this text has recently appeared: The Periplus of the Erythraean Sea, trans, with comm., Huntingford, G. W. B., Hakluyt Society, series 2, cli (London 1980).
2 E.g., Müller, C. in Geog. Gr. Min. i (Paris 1853) 257–305 (the standard text until the publication of Frisk's); Fabricius, B., Der Periplus des Erythraïschen Meeres (Leipzig 1883).
3 Mnemos. xxviii (1975) 293–5; JHS xcvi (1976) 154–7.
4 The species in question may well be Geochelone pardalis babcocki, the leopard tortoise. There are no tortoises on Socotra today and only two genera in the area of Africa nearest it. Of these one species is the leopard tortoise, which grows to more than two feet in length and so favors mountainous areas that some call it the mountain tortoise. Presumably it existed on Socotra in Roman times. For a full description see Loveridge, A. and Williams, E., Revision of the African Tortoises and Turtles of the Suborder Cryptodira, Bull. Museum of Comp. Zoology, Harvard College cxv. 6 (1957) 235–51. I have handled the shields of a specimen; they are distinctly thicker than those of the hawksbill turtle. I owe thanks to Dr Richard Zweifel of the American Museum of Natural History in New York for much generous help.
5 The chersinae of Pliny, , NH ix 38 and the cherson of Martial xiv 88 must be terrestrial types of some kind.
6 Plin., NH ix 39, xxxiii 146; Dig. xxxii 100.4; Mart. ix 59.9. For a useful summary with full citation of sources, see Blümner, H., Technologie und Terminologie der Gewerbe und Künste bei Griechen und Römern ii (Leipzig 1879) 375–8.
7 καὶ ὄντα troubled Frisk, who suggested in his apparatus either eliminating καὶ or assuming a lacuna. But the Periplus offers two other instances of καὶ used as a connective with a participle: 16, ἄνθρωποι ἀρόται κατοικοῦσιν κατὰ τὸν τόπον ἕκαστος ὀμοίως τιθέμενοι τυράννοις ‘men who are plowmen (cf. Mnemos. xxviii  294) live there and each acts like a king over his own land’; 22, ἔστιν δὲ τύραννος καὶ κατοικῶν αὐτὴν ‘there is a king and he inhabits it’ [sc. the capital city].
8 Frisk emended συντριβομένας to συντρίβεσθαι on the grounds that the word is correlative with τέμνεσθαι and could well have been abbreviated in the exemplar and wrongly expanded under the influence of the preceding participles (Frisk 115). Giangrande points out (JHS xcvi  155) that in the koine of this period infinitives and participles mingle as syntactical equivalents.
9 The word also appears in Eustathius', De Thessalonica urbe a Latinis capta 96, used more or less in the same sense as in Procopius: Eustathius describes how the Normans, by way of insulting their captives, bared their rears, bent over to evacuate, and tried ἐξ ἐναντίας ἡμῶν ἀποκοντοῦν τὰ περιττἀ τῆς γαστρός.
10 The citing of ‘ἀντέχουσιν at §46’ as a parallel is misleading. Its presence there is the result of restoration; see below.
11 Schmid in his review of Frisk's, edition (Philol. Wochenschr.  788–95) was the first to point out (792) the inadequacy of Frisk's rendering of ἀποκοντῶ. He suspected—rightly, as we shall see—a distinction between παρακειμένας and ἀποκοντουμένας but the only restoration he could offer had, on his own admission, serious drawbacks.
12 I owe thanks to my good friend and colleague, N. Lewis, for invaluable suggestions in connection with this passage.
13 Ancient craft carried many more anchors than their modern counterparts, which are generally content with three or four. The ship that carried St Paul to Malta had at least six (Acts xxvii 29–30). One ancient wreck had at least five, another eleven; cf. Casson, L., Ships and Seamanship in the Ancient World (Princeton 1971) 255–6.