^{page 122 note 3} Budé, , de asse et partibus eius, 1514, v, p. 135Google Scholar. He, like Tarn, attributed this system to the ancient triremes.

^{page 122 note 4} My thanks are due to the Trustees of the National Maritime Museum, Greenwich, for allowing a photograph to be taken from their copy of Paris's book.

^{page 122 note 5} What Koester (Das antike Seewesen, Berlin, 1923) calls the ‘Urrojer’, p. 105Google Scholar.

^{page 124 note 1} Cambridge, 1930, p. 128 and Appendix IV.

^{page 124 note 2} B.M., B 436. Davison, J. A., C.Q. xli, p. 24Google Scholar, n. 3. This arrangement appears first in some probably Phoenician galleys depicted in the palace of the Assyrian King Sanherib which dates from the end of the eighth century B.C. Koester (p. 54) supposes that the Phoenicians copied the Cretan type of warship (which was also the ancestor of the Greek type), and the two-level system may have had a similar origin. Another two-level ship is shown on the Bf. neckamphora in the Bruschi Collection (Museum at Tarquinia:rep. Jahrbuch, 1912, pp. 76, 77). It has eight oars in each level and is either an inaccurately drawn triakontor or an intermediate type.

^{page 124 note 3} Torr's reproduction, which is copied by Koester and the British Museum handbook Greek and Roman Life, has been doctored so that it shows eleven oars on the top level.

^{page 124 note 4} The fullest description of this operation is given by Apollonius Rhodius i. 378 (the launching of the Argo):

ὓψτ δ' ἓνθα κα⋯ ἓνθα μεταοτρέψαντες ⋯ρετμ⋯

πήχυιον προήοντα περ⋯ σκαλμοîσιν ἒδησαν.

Later (ibid. 392), when it is time to row, σκαλμοîς δ' ⋯μφ⋯ ⋯ρετμ⋯ κατήρτυον. Cf. Ovid, , Metam. II. 475 f.Google Scholar

portibus exierant, et moverat aura rudentes,

obvertit lateri pendentes navita remos.

Also ibid. 3.676 (when Dionysus' wonder-working had made rowing impossible) ‘at Libys obstantes dum vult obvertere remos’. In Ap. Rhod. the oars were placed ready for use, in Ovid half-shipped after use. The characteristic action (obvertere, μεταοτρέψαι) is the same. Contemporary representations of a scaloccio galleys under sail show the oars fully out but ‘parked’ with the blade higher than the handle. There are several in the library of the National Maritime Museum, Greenwich.

^{page 125 note 1} Oars were worked through a side-fence or lattice in the long-ship, but that did not involve an oar-port, which is a port below the gunwale.

^{page 125 note 2} C.R. lv, 1941, p. 89Google Scholar.

^{page 125 note 3} Oar-ports are visible in many pictures of ships: e.g. Aristonothos vase 650, Nikosthenes cup 530, Rf. stamnos, in B.M. (C.A.V. 3, pl. 20. 1Google Scholar), 470, Talos vase 390. In the case of the first and third it is possible that two-level ships are shown with only one squad working, since human figures are always too big for the ships in which they are drawn. The trireme on the Talos vase shows oar-ports for the zygii also. This was no doubt designed to increase the free-board. (Cf. also the Delphi ‘Dioscuri’ relief and the Ficoronian chest.)

^{page 125 note 4} C.R. lvi, 1942, 55Google Scholar

^{page 125 note 5} e.g. Koester, p. 114.

^{page 125 note 6} Cf. Aesch. in Ctes. 146; Cicero, , In Verr. 2. 5. 51. 135.Google Scholar

^{page 126 note 1} When of course it was too rough to row.

^{page 126 note 2} The relief is dated at the end of the fifth century, the vase at the beginning of the fourth.

^{page 126 note 3} Frogs 364; Acharnians 549 and 97.

^{page 126 note 4} I.G. ii² (VII Tabulae Curatorum Navalium), 1604. 68 and 75.

^{page 126 note 5} 1607. 24–5. It may be argued that port-holes do not fall out with age. But we know nothing about this ship. She may have been καιν⋯ ⋯νεπικλήρωτος and unfinished to the extent of not having oar-ports made.

^{page 126 note 6} Second Scholiast on the Acharnians 97: ⋯ τ⋯ς κώπης ⋯φθαλμ⋯ς τ⋯ ἃσκωμα κώπης δ⋯ ⋯φθαλμ⋯ς τ⋯ τρ⋯μα: and. cf. Eustathius, 1931. 42 … ⋯φθαλμ⋯ς ⋯ποîι κα⋯ ο⋯ κα⋯ οἱ κατ⋯ τ⋯ς τριήρεις λέγονται γ⋯ρ ⋯φθαλμο⋯ ῥητοικ⋯ς ⋯ν ⋯κείναις αί ⋯παἰ ὦν αί κ⋯παι διείρονται.

^{page 127 note 1} εἰοφέρειν MSS.: εἰσφρεîν edd. recc.

^{page 127 note 2} Fittings of a very similar nature are visible in the Roman ship on the Vatican relief.

^{page 127 note 3} J.H.S. XXV. 141, note 10.

^{page 127 note 4} Article ‘Seewesen’ in Baumeister's Denk- mäler des klassischen Altertums, iii. 1593 ff.

^{page 127 note 5} A. M. A., Om betydelsen af ordet παρεξειεσία: Comm. Philol. in hon. Ioh. Paulson, 1905.

^{page 127 note 6} Chr. B., Det kgl.Danske Videnskab. Selskab.: Arch.-Kunst. Meddels. ii. 3. Triémiolia: étude sur une type de navire Rhodien: København, , 1938, p. 38. 1Google Scholar

^{page 127 note 7} ii. 338.

^{page 127 note 8} Corp. Script. Hist. Byz., Pars iii, p. 326, 12: Agathiae Hist. v. 21 ad fin.

^{page 127 note 9} The other passages where the word appears are: Thuc. 4. 12. 1; 7. 34. 5 and 40. 5; Plutarch, de glor. Ath. 347 B (an account of the incident in Thuc. 4. 12. 1); Arrian, Periplus Mar. Eux. 3 (the waves come in, not only through the oar-ports, κατ⋯ κώπας, but over the παρεξειρεσία as well) and fr. 160 J.; Polyaenus 3. 11. 13 and 14. The most informative is Arrian, Peripl. M. Eux. 3.

^{page 127 note 10} Breusing (Die Nautik der Alten, Bremen, 1886, and Die Lösung der Trierenrätsels, Bremen, 1889) asserted that it would be impossible to keep time with oars of different lengths, and that therefore there could never have been ships with banks of oars at different levels. Tarn appears to adopt this argument. Koester (p. 106), who was a practical sailor, approves Assmann's contradiction of this assertion (Seewesen, p. 1610, B. ph. W. 1888, p. 26), and Mr. R. C. Anderson's views (Mariner's Mirror, vol. xix, p. 237 and vol. xxvii, pp. 314 ff.) are similar. As Anderson said, time-keeping would not be affected by a mere difference in length between the oars, only by a difference in the proportion of inboard to outboard parts. My approach is a more positive one. If, as the experts hold, there is a right length of oar for a man rowing in a given set of conditions, and if, as I have demonstrated, it is possible for all the oars in each oar-group in a three-level trireme to have been the same length, then I think they must have been so, particularly as this conclusion is supported by the evidence.

^{page 128 note 1} My thanks are due to the Society of Nautical Research for permission to reproduce the plan here.

^{page 128 note 2} C.R. lv, 1941, p. 89Google Scholar.

^{page 128 note 3} Thuc. vi. 31. 3; Aristoph, . Acharn. 162Google Scholar; I. G. ii². 1604 et al.

^{page 128 note 4} I.G. ii². 1604 passim et al.; Polyaenus 5. 22. 4; Pausanias ap. Eustathium 629. 41.

^{page 129 note 1} Thuc. 4. 32. 2; Soph. fr. 1052 Pearson.

^{page 129 note 2} e.g. Scholiast on Frogs 1072 in Codex Venetus Marcianus 474, Eustathius 640. 10.

^{page 129 note 3} Thranite: in the 9 cases where the total is given, 8 give 62, one 64. Zygian: in the 12 cases where the total is given, 11 give 54, one 52. Thalamian: in the 13 cases where the total is given, 12 give 54, one 53. περίνεω: in the 16 cases where the total is given, all give 30.

It is plain that in two of the three cases of variation, confusion between 62 and 54 has produced 52 and 64.

^{page 129 note 4} The Aquila fragment: see Rumpf, , Römische Fragmente, Winckelmannsprogramm 95, Berlin, 1935Google Scholar.

^{page 129 note 5} See particularly the Talos vase ship (which I claim as a trireme), and Aeneid 6. 4 f. curvae puppes.

^{page 129 note 6} C.R. ix. 166.

^{page 130 note 1} The two platforms appear in an Attic Bf. fragment in Graef, Die antiken Vasen von der Akropolis zu Athen, 2414a, Taf. 98.

^{page 130 note 2} 629. 41.

^{page 130 note 3} Zosimus 5. 20. 3–4. See my ‘Greek Trireme’, M.M. Jan. 1941, pp. 15–21.

^{page 130 note 4} Preface to his edition of the scholia.

^{page 130 note 5} So Koester, p. 103, though his description of the source as ‘antiken Autoren’ is rather overdoing it.

^{page 130 note 6} C.R. xix, 1905, p. 375Google Scholar.

^{page 131 note 1} When he speaks of the ‘docked’ oars of the thalamii and says that they got less pay on that account, he may be thinking of a contemporary a zenzile galley in which the oarsman nearest the gunwale did use a considerably shorter oar (see Fig. I) and may have got less pay on that account. Or he may have been thinking of Thucydides’ statement (6. 31. 3) that the thranites got a bonus on the Sicilian expedition.

^{page 131 note 2} Eustathius 640. 10. His use of the word θαλάμακες suggests that he is thinking of Frogs 1072.

^{page 131 note 3} C.R. xix. 371 ff.

^{page 131 note 4} Le Triremi, ed. 2, Rome, 1881; trans. Serre, Les Marines de guerre de l'antiquité et du moyen âge, Paris, 1885.

^{page 131 note 5} Hell. Nov. and Mil. Dev., p. 129, n. 1.

^{page 131 note 6} This is a constant because it depends on the average size of the human body.

^{page 131 note 7} Fincati (Serre, p. 161 f.) quotes these measurements from a letter of the Venetian admiral, Messire Cristoforo da Canale.

^{page 131 note 8} British naval cutters show the same difference in oar-lengths. The shorter ones are worked where the narrowing of the hull makes it necessary, and the smaller men are given them. Aristotle (de part. anim. 4. 10. 687^b18) says that the oars in the middle of the ship were longer than those at each end.

^{page 132 note 1} Torr and Koester suppose these oars to have been used in an emergency from the upper deck, but they are not long enough for that. Koester suggests also that they might have been used on such occasions in the bow or stern. But the only emergency in which every available ounce of energy was required at once was in a sea-battle, and then the extra men would have been otherwise engaged, e.g. the marines.

^{page 132 note 2} It is, I imagine, a coincidence that there were also 30 men (officers, marines, etc.) who with the 170 oarsmen made up the complement of 200, but I am not entirely happy in dismissing it as such.

^{page 132 note 3} S. Dragatsis, Πρακτικά, 1885 with Dörpfeld's plan and reconstruction, which is also given by Judeich, Topographie v. Athen ², p. 438. The plan shows a maximum length of 38 m. (= 124.26 ft.) from the back wall to the sea.

^{page 132 note 4} Koester (p. 97, n. 1) speaks of Selim II's admiral Uechiati who built galleys of 30 oars a side and 164 ft. long. Since these galleys must have been rowed on the a scaloccio (or multiple-handed) system, we can see that this system too required a greater length than the ancient trireme admits.

^{page 132 note 5} ‘Oars must be nearly horizontal, so in the model the gunwale is. 12 in. above the waterline.’

^{page 132 note 6} Koester, p. 138.

^{page 132 note 7} It is most unfortunate that the new Liddell and Scott should have accepted the former (see under τριήρης) and laid itself under grave suspicion of believing the latter (see δίκροτος, τρίκροτος in Add.).

^{page 133 note 1} C.R. lv, 1941, p. 89Google Scholar; cf. J.H.S. xxv. 149, n. 46.

^{page 133 note 2} Urkunden über das Seewesen des Attischen Staates, Berlin, 1840 (Beilage zur Staateshaushaltung der Athener).

^{page 133 note 3} C.R. lxi, 1947, pp. 114–16Google Scholar.

^{page 133 note 4} Based on a misinterpretation of the phrase κατάλυσις τριήρους in [Dem.] adv.Polyclem (50), 11.

^{page 133 note 5} e.g. I.G. ii². 1607.

^{page 133 note 6} e.g. 1632 (which includes the present passages).

^{page 134 note 1} e.g. 1623^b77–82.

^{page 134 note 2} e.g. 1631^c196 ff.

^{page 134 note 3} He may owe the gear or the money for it. In one list the entries are in this form: 1622^e170–8: ⋯π⋯ τ[⋯ν Σ]ω[τ]ηρίαν ‘Αγνοδέμου ἒργον ἓργον. Mενεσθεἓς ‘Pαμνο⋯(σιος) οὐδ⋯ν ⋯ποδέδωκ[ε]. είχεν δ' ⋯πἰ τἠν ν[α⋯ν] σκε⋯η κοντο[⋯ς] κτλ.

^{page 134 note 4} Which was one of those issued under the previous supervisors by the law of Diphilus (see 1632^a17, the general rubric for the list) and was at sea when the present supervisors took over.

^{page 134 note 5} The ‘handing over’ presumably consisted only of a book-entry.

^{page 134 note 6 a}42 τετρέρ]ης Παραγία Δ[ημοτέλους ἓργον]. τριή[ραρ(χος)…’Αμφ]ιτρόπ(ηθεν) [κα⋯ συντριήραρ(χοι)…] Kυ[δ]αθ[η(ναιε⋯ς).

^{page 134 note 7} From Professor Robertson's article above-mentioned.

^{page 135 note 1} 1629^b46–60; also 1627^b42–66 et al.

^{page 135 note 2} 1627^a26–38 ξ⋯λινα τετρήρων, 131–67 σκε⋯η τετρήρων ^c84–102 among quadrieme gear certain ropes are specified as τριηριτικά. 1628^b67–81 σκε⋯η ξ⋯λινα τετρήρων, 178–83 τετρήρων σκε⋯η κρεμα[στά] κτά] κτλ.