¹ I should like to thank Manchester University and the British School at Athens for generous financial and practical support while this research was being carried out, and the British School at Rome for a travel grant which allowed me to visit Morgantina. I am also grateful to Roger Ling for commenting on a draft version of this article, and to Lisa Nevett for supplying me with a copy of her forthcoming paper on housing in Roman Greece.

Abbreviations:

Bulard = Bulard, M., Peintures murales et mosaïques de Délos (Monuments et Mémoires Piot, 14; Paris, 1908)Google Scholar

Délos VIII = Chamonard, J., Exploration archéologique de Délos VIII: Le, Quartier du Théâtre (Paris, 1922)Google Scholar

Délos XXVII = Bruneau, Ph., Vatin, Cl. et al. , Exploration archéologique de Délos XXVII: L'îlot de la Maison des comédiens (Paris, 1970)Google Scholar

Délos XXIX = Bruneau, Ph., Exploration archéologique de Délos XXIX: Les mosaïques (Paris, 1972)Google Scholar

Morgantina = B. Tsakirgis, The Domestic Architecture of Morgantina in the Hellenistic and Roman Periods (unpublished PhD dissertation, Princeton, 1984)

Tsakirgis I: Tsakirgis, B., ‘The decorated pavements of Morgantina I: the mosaics’, AJA 93 (1989), 395–416CrossRefGoogle Scholar

Tsakirgis II: Tsakirgis, B., ‘The decorated pavements of Morgantina II: the opus signinum’, AJA 94 (1990), 425–43CrossRefGoogle Scholar

Trümper : Trümper, M., Wohnen in Delos: Eine baugeschichtliche Untersuchung zum Wandel der Wohnkultur in hellenistischer Zeit (Internationale Archäologie 46; Rahden, 1998)Google Scholar

Mosaics on Delos are referred to by the catalogue numbers used in Délos XXIX, in the form ‘Delos xxx’ I have followed the standard terminology for mosaics suggested by Balmelle, C. et al. , Le Décor géométrique de la mosaïque romaine: Répertoire graphiqu et descriptif des compositions linéaires et isotropes (Paris, 1985)Google Scholar.

² Westgate, R. C., ‘Greek mosaics in their architectural and social context’, BICS 42 (1997–1998), 93–115Google Scholar.

³ Ibid., 108–11; fig. 1; table 2.

⁴ Ibid., 111–15.

⁵ The sculptures are catalogued by Kreeb, M., Untersuchungen zur figürlichen Ausstattung delischer Privathäuser (Chicago, 1988)Google Scholar; their relationship to their setting is explored in §2. 3, pp. 33–51. With three exceptions, only terracotta statuettes were found m the houses at Olynthos: ibid., 92.

⁶ For example, rooms N and J of ihe Maison des comédiens: Délos XXVII, 36–7; and room D of the Maison de l'hermès: Delorme, J., BCH 77 (1953), 452Google Scholar.

⁷ Finds of mosaics from the royal quarters of Pergamon and Alexandria presumably represent the ultimate in quality: Kawerau, G. and Wiegand, T., Altertümer von Pergamon V.1: Die Paläste der Hochburg (Berlin, 1930), text pls 26–39; pls 7–19Google Scholar; Guimier-Sorbets, A.- M., ‘Alexandrie: les mosaïques hellénistiques découvertes sur le terrain de la nouvelle Bibliotheca Alcxandrina’, RA 1998, 263–90Google Scholar.

⁸ Previous attempts at understanding the function of the paintings have had a more limited scope: de Meneses, U. T. Bezerra, ‘Essai de lecture sociologique de la décoration murale des maisons d'habitation hellénistiques a Délos’, Dialoghi, 3rd ser. 2.1 (1984), 77 88Google Scholar, focuses on the figured paintings found in a minority of houses: and Alabe, F., ‘Technique, décor et espace à Délos’, in Moormann, E. M. (ed.), Functional and Spatial Analysis of Wall Painting: Proceedings of the Fifth International Congress on Ancient Wall Painting. Amsterdam, 8–12 September 1992 (BABesch supp. 3; Leiden, 1993), 141–4Google Scholar, comments on the universal application of the same basic Masonry Style scheme to rooms of all types and sizes.

⁹ The relationship of the Masonry Style to real masonry construction is discussed by Bruno, V. J., ‘Antecedents of the Pompeian First Style’, AJA 73 (1969), 305–17CrossRefGoogle Scholar. More recent studies have separated the surviving examples into two types, those with no divisions in the zone above the frieze, which arc presumed to be earlier and to imitate walls with stone socles and mud-brick above, and those with divisions to ‘blocks’ above the frieze, which appear later, imitating monumental masonry: Bilde, P. Guldager, ‘The International Style: aspects of Pompeian First Style and its eastern equivalents’, in Bilde, P. Guldager, Nielsen, I., and Nielsen, M. (eds), Aspects of Hellenism in Italy (Acta Hyperborea, 5; Copenhagen, 1995), 151–77Google Scholar. The two types coexisted in the late Hellenistic period, and the distinction between them does not affect the conclusions presented here: either system could be represented in paint, incision or stucco relief.

¹⁰ It is clear from the remarks of Vitruvius (De arch. vii. 7–14) and Pliny (NH xxxv. 12–31) about the cost of pigments that a contemporary viewer would have been aware of the distinction between standard colours and more expensive pigments, but unfortunately no information is available about the chemical composition of the colours used at Delos.

¹¹ e.g. by Wallace-Hadrill, A.. Houses and Society in Pompeii and Herculaneum (Princeton, 1994) 45Google Scholar; fig; 3. 12.

¹² Délos VIII, 109; fig. 49.

¹³ Bulard, 99; iig. 34 d; Délos VIII, 360; fig. 221.

¹⁴ In fact, Trümper, 37–8, suggests that the lack of decoration in the vestibules of the Delian houses was actually intended lo heighten the impact ol the lavish decor in the peristyle—and thus, presumably, to emphasize the privilege of being invited inside.

¹⁵ Bulard, 110, b: Délos VIII, 141, fig. 62; fig. 229.

¹⁶ Bulard, 114–15, 154, 157, 160 1: figs. 44, 52 e: pls 6A b; 8A i,j, l, m; Délos VIII fig. 232; pl. 49 b.

¹⁷ Bulard, 108–9, 140–1; pl. 6 a; Délos VIII, pl. 49 a.

¹⁸ Wallace-Hadrill (n. 11), 57–9.

¹⁹ No information is available about the wall decoration.

²⁰ Bulard, 103–4; fig. 39.

²¹ E and H: plain white: Délos VIII, 390. L: white with double incisions marking blocks: Bulard, 93 fig. 31; Délos VIII, 368 fig. 225. F: white with incised divisions and red frieze: ibid., 366; fig. 224.

²² The wall decoration in the house is described in Délos XXVII, 151–60.

²³ Ibid., 152–5 fig. 108.

²⁴ Ibid., 155–7; fig. 110; 168–83; Pls 21; 22–1, 3–8; 23; 24; 25–1 and 4.

²⁵ Ibid., 155; fig. 109.

²⁶ Ibid., 157; fig. 114.

²⁷ Ibid., 157–60 figs. 115, 116; pl. 22. 2.

²⁸ Busing-Kolbe, A., ‘Ein neureiches Haus auf Delos’, in Busing, H. and Hiller, F. (eds), Bathron: Beiträge zur Architektur und verwandten Künsten für Heinrich Drerup zu seinen 80. Geburstag von seinen Schülern und Freunden (Saarbrücken, 1988), 99–106Google Scholar.

²⁹ Löhr, C., ‘Grierhische Häuser: Hof, Fenster, Türen nach 348 v. Chr.’, in Heilraeyer, W.-D. and Hoepfher, YV. (eds), Licht und Architektur (Tübingen, 1990), 10–19Google Scholar.

³⁰ Rooms D and E are a later extension of the house. D seems to have housed a dyeing workshop; Trümper. 204–5, suggests that G, with water from the peristyle channelled through it, was the latrine. Room W was a shop, entered From the street.

³¹ Délos XXVII, 191–3; pl. 26. 6–7.

³² Délos XXIX, 64, 67–8.

³³ Ibid., 105–6.

³⁴ Délos XXVII, 157; figs. 111–13.

³⁵ The wall decoration is described Ibid., 161–7.

³⁶ Ibid., 166. As area had been extended by appropriating part of the street to the south and east, Trümper, 205–7, points out that for part of its existence room AN was accessible through AH; she identifies the eastern complex, with a sweat-bath in AM.

³⁷ Délos XXVII, 161–3 fig. 118.

³⁸ Ibid., 163, 166; fig. 121.

³⁹ Ibid., 161. Trümper, 205, suggests that the difference in the paintings is intended to mark off the west end of the north portico as a separate space, imitating an exedra.

⁴⁰ Délos XXVII, 161 fig. 117.

⁴¹ Ibid., 163; figs. 119, 120.

⁴² Ibid., 166; fig. 123.

⁴³ Ibid., 165; fig. 122; pls 25. 9, 26. 1–5.

⁴⁴ Plan, Délos VIII, pl. 28. It is possible, however, that the rooms above House B belonged to the larger House A next door, although the decoration of House A is otherwise undistinguished, with no decorated mosaics.

⁴⁵ The decoration of the house is described by Chamonard, J., ‘Fouilles de Délos exécutées aux frais de M. le Due de Loubat (1904)’, BCH 30 (1906), 483–672Google Scholar, esp. 530–54.

⁴⁶ Ibid., 546–7 fig. 16; Bulard, 111–2; fig. 46; Délos VIII, 374 fig. 230.

⁴⁷ Chamonard (n. 45), 548–9; Bulard, 97 8; fig. 32 f The fragments were not found in situ, and Chamonard suggested that they were remains of an earlier decoration.

⁴⁸ Compare the examples cited by R. J. Ling, Roman Painting (Cambridge, 1991), 16; pl. I.

⁴⁹ Chamouard (n. 45), 536.

⁵⁰ Ibid. 536; Délos VIII, 372.

⁵¹ Chamonard (n. 45), 531–6 fig. 12; Bulard, 143–5; 153–4, a; 155 6 fig. 55; pls 8 b; 8A g–h, 9; 9A a–d.

⁵² Bulard, 110–11; pl. 6A c.

⁵³ Bulard, 113–14; 152; pls 6A a; 8A k; Délos VIII, fig. 83.

⁵⁴ Chamonard (n. 45), 543–4.

⁵⁵ Kreeb (n. 5). 30, 233–4 nos. S 38. 1–2. Parts of several other sculptures were found in the house (ibid., 29–33, 252–7), but their original positions are less easy to reconstruct: pieces of one head (ibid., no. S 38. 4) were found in different rooms, and the lime-kilns installed in the house after it was abandoned may account for the presence of some of the fragments.

⁵⁶ The final publication of the houses, in the Morgantina Studies series, was not yet available when this article went to press.

⁵⁷ Balmelle et al. (n. 1), pls. 102 c, d; 103 a, c, h; 104 a, d; 106 a–f; 109 a–b: III b–c. This type of decoration is identified by some modern authors (including Balmelle et al.) with what ancient writers call seutulatum. and by others with lithostroton, the inset slabs, at this date, are usually limestone rather than marble.

⁵⁸ On the plans (FIGS. 17–21), the patterns on the opus signinum pavements are not drawn to scale; instead, the stippling is intended to give an impression of the relative density of the decoration.

⁵⁹ The house is described in Morgantina 125–51. It appears eventually to have been divided into two houses.

⁶⁰ Tsakirgis II, 430, no. 26.

⁶¹ Room 1: Tsakirgis I, 401–2. no.5, fig. 16; room 12: ibid., 403, no. 8, fig. 19.

⁶² Ibid., 402, no. 7; fig. 18.

⁶³ Ibid., no. 6, fig. 17.

⁶⁴ e.g. Vergina, Palace, rooms E-E-G; Pella, House of the Rape of Helen, rooms Θ-I-K.

⁶⁵ Room 15: Tsakirgis II, 430; rooms 8, 13, ibid., 429–30, nos. 16, 19.

⁶⁶ Morgantina, 135.

⁶⁷ Tsakirgis II, 430, no. 24.

⁶⁸ Tsakirgis I, 403, no. 9; figs. 20–21.

⁶⁹ Tsakirgis II, 430, no. 25.

⁷⁰ Room 24: Tsakirgis II, 430, no. 22, fig. 8; room 25, ibid., 430, no. 23.

⁷¹ There is a dividing strip in the pavement of room 9 which looks like a scendiletto, but in fact it was added to fill the gap where a wall was removed to enlarge the room, and it is not clear whether the room was intended to be used as a bedroom: Tsakirgis II, 437, n. 70.

⁷² Description of house: Morgantina, 186–206.

⁷³ Pavements: room 20: Tsakirgis II, 434, no. 54; fig. 15; room 21, ibid., 434, and Tsakirgis I, 404, no. 11: room 22, ibid., 404–5, no. 12; figs. 24–6. Wall plaster: Morgantina, 194–5. The remaining rooms around the west peristyle have all disappeared down the hillside.

⁷⁴ Tsakirgis II, 434, nos. 55, 63–66.

⁷⁵ Ibid., 432, no. 43.

⁷⁶ Morgantina, 191.

⁷⁷ Tsakirgis II, 433, no. 47, fig. 14. The position of the rosette, in one corner, recalls bedrooms at Pompeii which were designed lor two beds, but the remainder of the floor and wall decoration appears to indicate only one bed-space.

⁷⁸ Tsakirgis II, 437.

⁷⁹ Tsakirgis I, 403–4, no. 10; figs. 22–23; Tsakirgis II, 433–4, no. 51.

⁸⁰ East: Tsakirgis II, 435, nos. 57–9; west, ibid. nos. 60–62. The design in the north court is loo ruined to identify.

⁸¹ Tsakirgis II, 434, no. 56.

⁸² There were no mosaics from upstairs rooms at Morgantina, and in fact Tsakirgis concluded that most of the houses had no upper floor: Morgantina, 394–6.

⁸³ The rooms to the south-west are on a lower terrace, entered from the street, and the south-west portico has collapsed.

⁸⁴ Tsakirgis II, 428, nos. 7–9. The decoration of the house is discussed in Morgantina, 46–70.

⁸⁵ Room 8: Tsakirgis II, 428, no. 4; room 9, ibid., 428; room 7, Morgantina, 59.

⁸⁶ Tsakirgis II, 427–8, no. 2, figs. 2–3.

⁸⁷ Rooms 5 and 6: Tsakirgis II, 428; room 4, ibid., no. 3.

⁸⁸ Room 1: Tsakirgis II, 427, no. 1; room 2, ibid., 428.

⁸⁹ Room 12: ibid., 428, no. 5; fig. 4; room 15, ibid., no. 6 and fig. 5.

⁹⁰ Description of house: Morgantina, 70–84.

⁹¹ Tsakirgis II, 429, nos. 12–15.

⁹² Tsakirgis I, 399–400, no. 3, figs. 10–13.

⁹³ Ibid., 397–8, no. 1; figs. 1–4.

⁹⁴ Ibid., 398–9, no. 2, figs. 5–8.

⁹⁵ For example, in the peristyle of the Roman villa at Rabat on Malta (now incorporated into the Museum of Roman Antiquities), probably dating from the first half of the 1st c. BC: Ashby, T., ‘Roman Malta’, JRS 5 (1915), 34 fig. 5Google Scholar.

⁹⁶ The house is described in a series of reports by Bloesch, H. and Isler, H. P. in Ant. K. 15 (1972)–33 (1990)Google Scholar; the final plan is Ant. K. 33 (1990)Google Scholar, fig. 3 on p. 57. Rooms 3 and 4 housed a fullonica, probably accessible only from the street.

⁹⁷ Ant. K. 29 (1986), 72–3Google Scholar, fig. 3.

⁹⁸ Xenophon, Oec. ix. 4; Mem. in. 8. 9–10.

⁹⁹ Lysias i. 9.

¹⁰⁰ See the discussion by Nevett, L. C., ‘Continuity and change in Greek households under Roman rule: the role of women in the domestic context’, in Blomqvist, K. and Ostenfeld, E. (eds), Greek Romans or Roman Greeks? (Aarhus Studies in Mediterranean Antiquity 3; Aarhus, in press)Google Scholar. Certainly the suggestion of W. Hoepfner and E.-L. Sehwandner, IIaus und Stadt im klassischen Gnechenland upper floors of the Delian houses were occupied by women's quarters, seems unlikely in view of the fine decoration from upstairs rooms. However, at both Delos and Morgantina the question is likely to be further complicated by the presence of non-Greek elements in the population.

¹⁰¹ For example, according lo the figures published by Trümper. 166–8, the largest house, the Maison du diadumène, covered less than 900 m²; only a few houses, even in the more spacious quarters of the city, were in the range 300–650 m², and in the crowded Theatre Quarter most were smaller still. In contrast, the few published figures and some rough calculations from plans indicate that the grandest houses at Morgantina had areas in the range 600–1250 m².

¹⁰² It should also be pointed out that the houses on Delos are built of stone and could therefore support mosaics on the upper lloor. whereas the weight of mosaics may have been a problem in houses with upper parts in mud brick,

¹⁰³ Délos VIII, 295–303.

¹⁰⁴ Compare the discussion by S. I. Rotroff the changing proportions of different vessel types in the ceramic assemblages from the Athenian Agora, which she attributes to the same phenomenon: The Missing Krater and the Hellenistic Symposium: Drinking in the Age of Alexander the Great (Broadhead Classical Lecture No. 7; Christchureh, 1996), 22–7Google Scholar.

¹⁰⁵ Wallace-Hadrill (n. 11), 28–61.

¹⁰⁶ As observed by Nevett, L. C., House and Society in the Ancient Greek World (Cambridge, 1999), 164–6Google Scholar.