Acknowledgements. I am deeply grateful to my friend Mr. Sinclair Hood for correcting my English text and to Mrs. Voula Zerva, architect, for making the plans of the tank and the pool.

¹ ‘Sub-marine Exploration in Crete, 1955’, BSA liii–iv (1958–9) 264 ff.

² Hood and Leatham, loc. cit. 270, 14 on the plan, fig. 2 and figs. 4, 4a.

³ Hood and Leatham, loc. cit. 275 fig. 5, pl. 64e.

⁴ R. W. Hutchinson, Prehistorie Crete 241 n. I, had tentatively suggested an EM date.

⁵ For a preliminary report see Davaras, C., Amaltheia ii (1972) 46Google Scholar; AD xxvi (1971) B2 (in press).

⁶ I owe this information to Mr. Nicholas Karantonis, former Custodian of Antiquities at Zakros, who saw the tank many times when the water was clear. To the west he has observed some remains of another fish tank, today completely obliterated.

⁷ In fact they face rather to the south-east. A map of the coastline about Siteia has been published by me in EA 1967, 84 fig. 1.

⁸ Cf. Hood and Leatham, loc. cit. 264.

⁹ BSA xxxiii (1932–3) 97.

¹⁰ Inscriptiones Creticae iii. 164.

¹¹ KrCh xiii (1959) 195 n. 32.

¹² Cf. Bürchner, , ‘Eteia’, in RE vi. I, 706Google Scholar; Bursian, , Geographie von Griechenland ii. 575.Google Scholar

¹³ References in Davaras, C., EA 1967, 90 n. 8Google Scholar; cf. also Platon, N., Praktika 1953, 291 f.Google Scholar

¹⁴ Now in the Hagios Nikolaos Museum. Miss A. Lembessi prepares their publication.

¹⁵ Cf. the recommendations of Columella (viii. 16, 6) to owners of islands or of poor estates by the sea; another point of view is represented by Varro (iii. 17, 2), who disapproves of them as being expensive in building, stocking, and upkeep.

¹⁶ iii. 17, 3 (plscinae loculatae): ‘Who, however, goes in for fish ponds and does not have a row of them? For just as painters have large boxes with compartments for keeplng their plgments of different colours, so they have ponds with compartments for keeplng the varieties of fish separate.’ For the Roman plscina in general see Gundel, W., RE xx. 2 (1950) 1783 ff.Google Scholar

¹⁷ As we see them in Japan today. Two similar fish tanks but at some distance from the sea and not connected with it are in use near Halmyros in the Mani, Laconia. I owe this information to my colleague Mr. George Papathanasopoulos.

¹⁸ On this problem see Pendlebury, J. D. S., The Archaeology of Crete ² (1963) 3Google Scholar; Hood and Leatham, op. cit. 265fr.; Creutzburg, N., KrCh xv–xvi (1961–1962) 342Google Scholar; Hafemann, D., ‘Die Niveauänderungen an den Küsten Kretas seit dem Altertum’, Abh Mainz 1965, 12Google Scholar; Hood, Sinclair, The Minoans (1971) 17 f.Google Scholar, 151 n. a. For oscillations in the rise and fall of the water at Mallia since Roman times see Études Crétoises xiii (1963) 29–31

¹⁹ Davaras, C., EA 1967, 84–90.Google Scholar Cf. S. Hood, The Minoans 151 n. 2.

²⁰ This must have been the original depth of the other Cretan tanks: cf. Hood and Leatham, loc. cit. 265.

²¹ Ibid.

²² Four lines at the south side, two or three at the west, and one at the east side.

²³ The lowest row of holes was probably just above the ancient sea-level, a fact leading to the idea that this row did not support any part of the roof but merely a light wooden grid above the spacious tank enabling the personnel to reach easily every part of the tank in order to catch a fish or clean the place.

²⁴ Hood and Leatham, loc. cit. 270.

²⁵ Colder water is more healthy for the fish: cf. Columella viii. 17, 3–4; hence the necessity of a sufficient depth, as we have seen.

²⁶ Swimmers in the Mediterranean know from experience that sea-water enclosed during the summer in pools in the rocks by the shore can be as hot as the water in a bathtub.

²⁷ Cf. Hood and Leatham, loc. cit. 250.

²⁸ Hood and Leatham, Ioc. cit. 275 pl. 63a.

²⁹ Columella (viii. 17, 6) recommends bronze gratings with small holes.