² On Hippodamos see Arist. Pol. ii. 5; Wycherley, R. E., ‘Hippodamos and Rhodes,’ Historia, 13 (1964), 135–9Google Scholar; McCredie, James R., ‘Hippodamos of Miletos,’ in Mitten, D. G., Pedley, J. G., and Scott, J. A. (eds), Studies Presented to George M. A. Hanfmann (Harvard University Monographs in Art and Archaeology, 2; Mainz, 1971), 95–100.Google Scholar Many have argued that the foundation of Rhodes came too late for Hippodamos to have been actively involved, but his connection with the design of the city is accepted by Wycherley, 135–9, and McCredie, 98–100.

³ Kondis 1958, 156, records that the very scanty early remains do seem to conform to the grid plan; cf. the annual excavation summary in BCH 82 (1958), 775: ‘Dans l'une des rues la couche inférieure a livré des tessons qui remontent à la première période de la fondation de la ville de Rhodes, ce qui preuve l'existence d'un plan poléodomique dès cette époque’. Kondis repeats this view in his supplement to Karousos, 119–20.

⁴ For a discussion of the ancient street plan see, inter alios, Kondis 1954; 1958; Bradford 1956; 1957, 277–86; Konstantinopoulos 1986, 201–4; 1994, 57; id., ‘Hippodamischer Stadtplan von Rhodos: Forschungsgeschichte’, in Dietz, S. and Papachristodoulou, I. (eds), Archaeology in the Dodecanese (Copenhagen, 1988), 88–95Google Scholar; W. Hoepfner, ‘Der Stadtplan von Rhodos,’, Ibid. 96–7.

⁵ The statue was erected in thanksgiving for Rhodes's deliverance from the great siege of 305/4 BC, but was toppled during the earthquake of c.228 BC. For the site of the sanctuary of Helios, see now Konstantinopoulos, G., ‘Ἀνασϰαφαὶ εἰς Ῥόδον’, PAE 1975, 238–48Google Scholar; Kontorini, V., Ἀνέϰδοτες ἐπιγραφὲς Ῥόδου, ii (Athens, 1989), 129–31Google Scholar (English summary p. 195); Rice, E. E., ‘The glorious dead: commemoration of the fallen and portrayal of victory in the late classical and hellenistic world’, in Rich, J. and Shipley, G. (eds), War and Society in the Greek World (Leicester–Nottingham Studies in Ancient Society, 5; London and New York, 1993), 224–57, at 235–9.Google Scholar

⁶ For a general discussion of the main acropolis area, see Konstantinopoulos 1968, 117–20.

⁷ See Birge, D. E., Sacred Groves in the Ancient Greek World, unpublished thesis, Berkeley, 1982Google Scholar; M. Carroll-Spillecke, Κῆπος: der antike griechische Garten (Wöhnen in der klassischen Polis, 3; Munich, 1989), passim; Gothein, M. L., Geschichte der Gartenkunst (2nd edn; Jena, 1926), i. 55–83Google Scholar (on Greek gardens in general). A particularly well-studied garden associated with a temple is that surrounding the temple of Hephaistos above the Agora in Athens, probably first planted in the early 3rd cent. BC: Thompson, D. B., ‘The garden of Hephaistos,’ Hesp. 6 (1937), 396–425CrossRefGoogle Scholar; Agora, xiv. 149, pl. 76; Camp, J. M., The Athenian Agora: Excavations in the Heart of Classical Athens (London, 1986), 87, fig. 64.Google Scholar For all (he gardens of the Agora and others connected with the famous schools of philosophy in Athens (all of which had religious connotations), see Carroll-Spillecke (above), 28–31; Thompson, D. B., ‘Ancient gardens in Greece and Italy’, Archaeology, 4 (1951), 41–4Google Scholar; cad. and Griswold, R. E., Garden hire of Ancient Athens (Agora Picture Books, 8; Princeton, 1963)Google Scholar; Ridgway, B. Sismondo, ‘Greek antecedents of garden sculpture’, in MacDougall, E. B. and Jashemski, W. F. (eds), Roman Gardens (Dumbarton Oaks Colloquium on the History of Landscape Architecture, 7; Washington, DC, 1981), 17–18.Google Scholar

⁸ Jones, C. P., ‘The Rhodian oration ascribed to Aelius Aristides’, CQ 84 [n.s. 40] (1990), 314–22Google Scholar, argues persuasively that this oration should indeed be attributed to the famous orator.

⁹ See also Carroll-Spillecke (n. 7), 69. It is known that trees were features of sanctuaries in other Rhodian cities. Pindar (O. 7. 48–9) refers to a grove on the acropolis at Lindos where the important temple of Athena Lindia was located, and an inscription from c. AD 50 records the planting of olive trees on the Linchan acropolis by a group of priests (ILindos 430). A series of poems in extravagantly heroic language commemorates a later plantation of olive trees there by one Aglochartos in the 3rd cent. AD. All of die Lindian texts are collected and discussed by Bousquet, J., ‘Les oliviers de Lindos’, Receuil Plassart: études sur l'antiquité grecque offertes à André Plassart par ses collègues de la Sorbonne (Paris, 1976), 9–13.Google Scholar

¹⁰ The fullest accounts of investigations in the area under discussion (Od. Pindou and Od. Voreiou Ipeirou) are found in Kondis 1951, 238–45; 1952, 547–63. A summary of the older Italian discoveries in the area (with bibliography) is given in the text accompanying the maps in Inglieri, 13–21 (on the city).

¹¹ A most welcome report appears, in Ergon, 1989, 136, no. 30 s.v. Ῥόδος, that Prof. G. Konstantinopoulos has recently cleared the N part of the acropolis in preparation for renewed investigation of the area in coming seasons; cf. AR 1989–90, 70.

¹² The question has been asked of me whether these structures were not merely quarry-pits or cisterns. In response to former suggestion, cf. n. 36; that some were turned into cisterns later is undeniable (cf. below). When I put these questions to colleagues in ancient architecture at a seminar in 1991, they were unanimous that the artificial features proved that these were indeed grottoes. I might add the opinion of Dr Susan Walker (British Museum). When early drafts of this paper were written I was unaware of her unpublished doctoral dissertation The Architectural Development of Roman Nymphaea in Greece (Univ. of London, 1979). She has since very kindly made available to me the section where she discusses these Rhodian grottoes in detail (pp. 56–81). Although our observations were made independently, I am gratified that we reached similar conclusions about their appearance, use, reuse, and general date; I have not, therefore, altered my text substantially. It is a pleasure to be able to record my gratitude to Dr Walker.

¹³ This road is examined in some detail by Kondis 1951, 238–45, where several photographs of the ancient remains underlying Od. Pindou are reproduced. These were much more visible some forty years ago than they are today owing to the building up of the street in the intervening period. For a brief account in English of Kondis's investigation, see Bradford 1956, 64; 1957, 284.

¹⁴ The N–S line of this structure was cleared by Kondis; cf. Kondis 1952, 553–9, with accompanying plates and drawings. It is marked as a stoa on the sketch of the area in Konstantinopoulos 1968, 117.

¹⁵ See e.g. Iacopi, G., ‘Recenti scavi nelle isole italiane dell'Egeo’, Historia (Milan), 5 (1931), 476Google Scholar; Laurenzi, L., ‘Attività del Servizio Archeologico nelle isole italiane dell'Egeo nel biennio 1934–1935’, B. d. A. 30 (1936–1937), 133Google Scholar; Inglieri, 16–17; Konstantinopoulos 1968, 118–19; Karousos, 92; Konstantinopoulos 1986, 218–19.

¹⁶ Among these shells are Ostrea, Murex, and Spondylus.

¹⁷ Laurenzi (n. 15)), 133: ‘Nell'età romana la grotta dev'essere stata trasformata in cisterna con una forte intonacatura idraulica di parte delle parete e del pavimento, che nascose una piscina rettangolare d'età anteriore’; cf. Inglieri, 16–17 (ad no. 17).

¹⁸ Kondis 1952, 553–9.

¹⁹ It is apparently this nymphaeum which Inglieri, 17, says preserves traces of transformation into a Roman cistern; I assume this comment is based on the added layers of cement in this chamber, which Laurenzi (n. 15), 133, mentions only in grotto system I. It is an unfortunate feature of these early reports, published without maps, sketches, or precise directions, that is often difficult if not impossible to ascertain exactly which area is being discussed.

²⁰ A third group of caves or grottoes also appears to be associated with an ascent to this N part of the acropolis. It lies further E and S of grotto systems I and II; the general direction is beyond the w end of Od. Cheimarras (ancient road P 13; Fig. 1, T). They are carved into a steep cliff-face, and can now best be found below the modern road branching s from Od. Voreiou Ipeirou (Fig. 4, C) where it bends after crossing Od. Pindou for the first time (this road leads directly to the top of Od. Diagoridon—the major ancient road, P 15; Fig. 1, V—and the temple of Apollo Pythios). Roughly in line with the rock-cut base w of grotto system II (Figs. 1, L; 4, F), which can be approached on their S end also from this modern road, a white cross marks the spot beneath which the cliff-face has been hollowed out into various caves and grottoes (fig. 28). A flight of rock-cut steps on the W connects the cliff-face to areas above and below. Along the cliff in both directions are cut niches, although no pattern can easily be discerned since the locality is badly overgrown and used by locals in connection with agricultural activities. One grotto has been remade into a rock-cut church (the modern white cross marks it). The niches continue along the natural contour of the cliff-face to the N, which appears to the eye to curve round to the line of the long, rock-cut terrace mentioned above (fig. 4, E) on Voreiou Ipeirou, itself due E of grotto system I (figs, 1, K; 4, D).

²¹ Information provided in the exhibition ‘Forty Years of Rescue Archaclogy in Rhodes’, seen in the Palace of the Grand Masters at Rhodes in 1989.

²² For a full description of this grotto, see Savage, R. J. G., ‘Natural history of the Goldney Garden grotto, Clifton, Bristol’, Garden History (The Journal of the Garden History Society), 17 (1989), 1–40.Google Scholar I am grateful to the warden and committee of Goldney House for granting me permission to visit and photograph the grotto. For a general discussion of garden grottoes from classical until modern times, see Miller, N., Heavenly Caves: Reflections on the Garden Grotto (London, 1982)Google Scholar; Jones, B., Follies and Grottoes (London, 1953).Google Scholar An article in the ‘Sunday Review’ section (pp. 49–51) of the Independent on Sunday for 24 July 1994 lists several easily accessible garden grottoes in England.

²³ See Savage (n. 22).

²⁴ For the Vari cave see infra n. 32. Archaiologia, 15 (1985), is entirely devoted to the subject of sacred caves in ancient Greece; for 7 caves in Attica sacred to Pan (including the Vari cave), see Ch. Deligiorgi-Alexopoulou, ‘Attic caves dedicated to the god Pan’ (Greek with English summary), Ibid. 45–54. Ridgway (n. 7), 21–5, considers various ancient Greek cave-shrines with rock-cut niches for votive reliefs.

²⁵ In connection with the investigation of this first nymphaeum, Iacopi (n. 15), 476–7. refers only to two male portrait heads of the later Roman period (inv. no. 13644 = Clara Rhodos, 5.1 (1931–2), 58, figs. 34–5 = Merker no. 125, figs. 72–3; inv. no. 13645 = Clara Rhodos, 5.1 (1931–2), 63, figs. 36–8 = Merker no. 122, figs. 68–9), a veiled female head (probably funerary) (inv. no. 13637 = Clara Rhodos, 5.1 (1931–2), 28, figs. 15–16), and a marble mask in the form of acroterion. He emphasizes that none was an original part of the nymphaeum, and laments ‘la mancanza assoluta di altri rinvenimenti atti ad identificare il complesso’.

²⁶ For recent surveys of Rhodian sculpture, see Gualandi, G., ‘Sculture di Rodi’, ASA n.s. 38 (1976), 7–259Google Scholar (pp. 21–36 give an account of all previous work on Rhodian sculpture with full bibliography).

²⁷ Merker, 7, 15.

²⁸ Rhodes Museum inv. no. 13614 = Merker no. 8, figs. 1, 7–8 (cf. 7, 15) = Gualandi (n. 21), 57–9, fig. 32.

²⁹ See Merker, 15.

³⁰ The well-known sleeping satyr in the Rhodes Museum (inv. no. 1160 = Merker no. 55, figs. 37–9), for example, has a water-channel and so must have been associated with a fountain; cf. Gualandi (n. 21), 202 n. 2; 223 n. 6.

³¹ Cf. Cole, S. G., ‘The uses of water in Greek sanctuaries’, in Hägg, R., Marinatos, N., and Nordquist, G. C. (eds), Early Greek Cult Practice (Proceedings of the 5th International Symposium at the Swedish Institute at Athens, 26–21) June 1986) (Stockholm, 1988).Google Scholar

³² On the Vari cave see Weller, C. H.et al., ‘The cave at Vari’, AJA 7 (1903), 263–349.CrossRefGoogle Scholar One dedicatory inscription from this cave refers to it as a ϰᾶπον Νύμφαις: IG i² 784a. See now Travlos, J., Bildlexikon zur Topographie des antiken Attika (Tübingen, 1988), 447–8.Google Scholar The Vari cave is also discussed in Arehaiologia, 15 (see n. 24), 51–3, as is the sacred Ellenokamaras cave (popularly known as the Hellinokamara) on the island of Kasos in the Dodecanese, which was under Rhodian control for much of antiquity (Ibid. 34–6 and AR 1993–4, 69–70).

³³ See, in general, Littlewood, A. R., ‘Ancient literary evidence for the pleasure gardens of Roman country villas’, in MacDougall, E. B. (ed.), Ancient Roman Villa Gardens (Dumbarton Oaks Colloquium on the History of Landscape Architecture, 10; Washington, DC, 1987), 7–30Google Scholar; E. Salza Prina Ricotti, ‘The importance of water in Roman garden triclinia’, Ibid. 135–84. The ultimate example of the use of water in a private villa must be the emperor Hadrian's villa at Tivoli near Rome: see Ibid. 173–81, and Ehrlich, T. L., ‘The waterworks of Hadrian's villa’, Journal of Garden History, 9 (1989), 161–76.CrossRefGoogle Scholar

³⁴ The grotto at Sperlonga and the famous sculptures (carved by Rhodian sculptors) that it housed have occasioned a huge and still growing bibliography. For a discussion of the use of water in the design of this complex see Salza Prina Ricotti (n. 33), 138–69. Against the view that this grotto was somehow inspired by Rhodian grottoes, see Rice, E. E., ‘Prosopographika Rhodiaka, part ii: the Rhodian sculptors of the Sperlonga and Laocoön statuary groups’, BSA 81 (1986), 245–50Google Scholar, arguing that although these sculptures were created by the famous trio of Rhodian sculptors who were also responsible for the Laocoön group (Athanodoros son of Hagesandros, Hagesandros son of Paionios, and Polydoros son of Polydoros), they were neither originally designed for Rhodian grottoes and later moved to Italy, nor based on Rhodian originals copied for the Italian grotto. The Sperlonga sculptures are part of a much grander type and scale of sculptural design than anything known from Rhodes itself, and the complex arrangement can surely only have been intended for the particular grotto at Sperlonga. It is the intention of the current paper to emphasize that the extant Rhodian grottoes were clearly designed as repositories for small votive offerings, not large sculptural compositions.

³⁵ Both Laurenzi (n. 15), 133, and Inglieri, 16, refer to the presence of a spring in this grotto, but I saw no sign of any. If they are correct, this is proof that there was a water source at ground level.

³⁶ The acropolis hill Monte Smith is cut in the Sgourou Formation attributed to the Upper Pliocene–Lower Pleistocene (Mutti, E., Orombelli, G., and Pozzi, R., ‘Geological studies of the Dodecanese islands (Aegean Sea), vol. 9: geological map of the island of Rhodes (Greece). Explanatory notes’, Annales géologiques des pays helléniques, 22 (1968 [1970]), 77–226Google Scholar plus map). This formation is up to 130 m thick. The uppermost 2 m on Monte Smith is made up of yellow, shelly, marine limestone, termed panichia by Mutti et al. This in turn is generally capped by what these authors term poros, a continental crust up to 2 m thick representing a late Pliocene early Pleistocene paleosol (Ibid. 149–52). Panichia and poros are easily cut and widely used in construction; it is into these that the grottoes arc cut, sometimes extending down into the relatively impermeable marine marl facies of the Sgourou Formation, the contact being a line of seepage (Ibid. 220, section in fig. 20). The large workings immediately N and S of the temple of Apollo Pythios may well have been a source of building material, subsequently modified for other purposes as indicated by the presence of niches, etc., in the southern working (cf. n. 40). Grotto systems I and II are by contrast smaller, without such obvious indication of quarrying as are apparent in the areas mentioned above and also in the extensive ancient quarries at Lindos.

³⁷ These aqueducts are discussed briefly and illustrated in part by Maiuri, A. in Rodi (2nd edn; Rome, 1921) 34–5Google Scholar; id., Clara Rhodos, 1 (1928), 50–2; Konstantinopoulos 1994, 59–62. Their lines are marked in Inglieri's map (Folio Città), and described in his text, 18, no. 23. Bradford 1956, 62–3, and 1957, 283–4, notes that stretches of them are aligned with the ancient grid plan, along the axes of the W–E roads descending from the acropolis to the town; they are so marked on his sketch maps. The position of some stretches of the aqueducts that are relevant to my discussion are approximately marked on Figs. 1, P and Q; 4, J. Remains of one branch of the aqueduct system which can be easily seen today lies within the walls of the moat of the medieval city (Fig. 35), in the area known as the Tongue of Provence (‘27’ on Inglieri's map; discussed on his p. 19). This section of the moat was cut through the presumed area of the ancient agora of the city, which was also served by the water system.

³⁸ Iacopi (n. 15), 476: ‘A metà d'uno dei lati lunghi, una galleria in discesa mette in comunicazione il ninfeo cogli acquedotti sotterranei. All'antro si accede da una scaletta intagliata nella roccia, che scende piegando ad angolo retto entro un pozzo appositamente scavato, fino ad una specie de vestibolo.’ It was here that he discovered the later marble heads (n. 25). This information is repeated directly in Karousos, 92, but no one else who discusses either the grottoes or the aqueduct system mentions that they are interconnecting. That they do connect was stated recently by G. Konstantinopoulos (pers. comm. with Prof. Dora P. Crouch), but he noted that the line was now impossible to follow owing to the collapse of the aqueducts after some 100 m.

³⁹ Segre, M., ‘Testi e documenti: l'oracolo de Apollo Pythaeus a Rodi’, PP 4 (1949), 79.Google Scholar

⁴⁰ Bradford 1956, 62, observes that a separate group of grottoes, still visible today S of the temple of Apollo Pythios, are also aligned to the ancient grid system; for these see Konstantinopoulos, G., ‘Ἀνασϰαφαὶ εἰς Ῥόδον’, PAE 1973, 127–36.Google Scholar The approximate location of these is located on fig. 1, Y.

⁴¹ Cf. Pollitt, J. J., Art in the Hellenistic Age (Cambridge, 1986), 230.Google Scholar

⁴² Morelli, D., ‘I culti in Rodi,’ Studi classici e orientali, 8 (1959), 62, 165 (nymphs)Google Scholar; pp. 63, 167 (Pan).

⁴³ IG xii. 1. 24.

⁴⁴ An intriguing rock-cut sanctuary dedicated at least partly to the Nymphs is located on the island of Kasos, incorporated into the Rhodian state probably c.250–200 BC. The site lies on the E side of the island (facing the sw tip of Karpathos) above the bay of Chramba, and in the last century at least was known as Ta (Ellinika) Grammata. It was first mentioned by Ross, L., Reise auf den griechischen Inseln des ägäischen Meeres, iii (Stuttgart and Tübingen, 1845), 43–5.Google Scholar The fullest discussion, with two photographs, is Susini, G., ‘Supplemento epigrafico di Caso, Scarpanto, Saro, Calchi, Alinnia c Tilo’, ASA n.s. 25–6 (1963–1964), 213–16.Google Scholar Two texts from the site, both said to be late hellenistic, were published as IG xii. 1. 1042a–b (both reproduced in Susini). When I visited Kasos in spring 1989 I was unable to find the sanctuary, although I found the cliff pictured in Susini's fig. 7; none of the inhabitants I spoke to had any knowledge of it. According to Susini the sanctuary consists of a rock-cut, crescent-shaped platform cut into a cliff-face, measuring 3.5 m wide by 12 m long. At the N edge were two short subterranean passageways, evidently ancient from the graffiti carved inside. Ross noted openings in the overhanging rock-face which he thought might have been a water source; Susini was not able to verify this. In one of the legible rock-cut inscriptions (IG 1042 a), Eudemos son of Demos greets unspecified gods (Χαίρετε | 0εοὶ | Εὔδημος | Δημοσ–); the other (1042 b) is an acclamation to the Nymphs ([...] | Χαίρετε | Νύμφαι | χαίρετε Νύμφαι | [...]). Susini suggests that the theoi in question may be the Samothracian gods, who had connections with the sea and were worshipped in Rhodian territory (cf. IG xii. 1. 8; 788; ILindos 134. 32; ASA n.s. 1–2 (1939–40), 184–5); particularly noteworthy is a list of priests of the Samothracian gods from the sanctuary of Poseidon Porthmios at Tristomo on the NW tip of Karpathos, very near to the Rhodian deme site of the Brykountioi (cf. IG xii. 1. 1034). The combination at Kasos of a rock-cut shrine open to the sky, niches, and a possible association with water bears certain analogies to the grottoes on the Rhodian mainland.

⁴⁵ See n. 40.

⁴⁶ Cf. H. Lauter, Die Architektur des Hellenismus (Darmstadt, 1986), 71. Here, and especially in ‘Kunst und Landschaft: ein Beitrag zum rhodischen Hellenismus’, AK 15 (1972), 53–8, Lauter discusses other, apparently ancient, grottoes in and near the modern park of Rodini SE of Rhodes town, outside the ancient city wall. He sees these as belonging to an ‘ancient park’ type of landscape system (using ‘park’ in the modern sense of a place for strolling and leisure activities), arguing that this area was comparable in its ornamental elements to the overall ‘love of nature’ he sees in many Rhodian monuments, not least the Rhodian acropolis. (These arguments unfortunately appear to have been accepted by Ridgway (n. 7), 13–14.) I hope to deal with the Rodini grottoes in the future, and do not accept Lauter's secular and anachronistic view of them. The connection of nature with sacred space is particularly evident in the case of Rodini a modern park, to be sure, but not on that account an ancient one – but, pace Lauter, its striking landscape features are surely to be explained by the fact that this was a major ancient necropolis area, near the so-called ‘Ptolemaion’, the largest monumental tomb so far known in the city. Garden areas were frequently part of cemeteries, as the Kerameikos in Athens attests.

⁴⁷ These objections to Lauter's interpretation have already been voiced briefly by Carroll-Spillecke (n. 7), 56, with whose views I entirely agree.