The decoration of walls and lloors was used to structure thr spare in Hellenistic houses, by means of hierarchies which would have been easily tradable by a contemporary observer, but which need to be elucidated before we can understand their workings. I shall propose possible structures for these hierarchies in the design of mosaics and wall paintings, and then analyse some surviving houses, to show how their schemes of decoration might have worked in practice. A more detailed understanding of decoration should also be a useful tool for appreciating differences in status between houses at the same site and comparing houses across different sites. Moreover, studying the decoration of these houses gives an insight into changing patterns in the use of domestic space in the Hellenistic period. Most of the examples arc taken from Delos and Morgantina, which were abandoned in the late Hellenistic period or soon after, and thus preserve a range of housing in morc-or-lcss its late Hellenistic state.’ The discussion will be dominated by mosaics and other types of pavement, as these tend to survive better than fragile wall plaster: the conclusions suggested are therefore incomplete and rather tentative. I shall mainly be concerned mainly with tessellated mosaics, which first appeared in the Greek world in the third century BC, replacing pebble mosaics as the standard type of decorated mosaic.
The standard composition was the same in tessellated mosaic as in pebble mosaic (FIGS. 2, 3, 10, 13): it consisted of a decorated ‘carpet’ framed by a series of concentric borders with geometric, vegetal or figured decoration, around a central field which might he left plain, filled with geometric patterns or plant motifs, or decorated with a figured scene; the carpet was enclosed by a plain surround, which was often interrupted by a small decorated ‘mat* in front of the door. This composition was shaped by the function of the rooms in which the earliest mosaics were used: the majority of pebble mosaics in houses were in the andron or dining room, and they were designed to be viewed by the diners reclining on couches around the sides of the room.* Unlike pebble mosaics, tessellated mosaics arc found in rooms of all shapes and sizes opening off the central courtyard or peristyle of the house.3 Very few are in traditional andrones with a raised border for couches, but it is still likely that many rooms with mosaics were used for dining, as they often have a broad, plain outer border of appropriate width for couches. Mosaics in ground-floor living-rooms arc roughly equally distributed between large and small rooms, although the small rooms are of two types: about a third are exedras with open fronts, and the rest have doors of normal width; it is not clear whether they were used for different purposes. The large rooms are of various shapes, but the most characteristic type is rectangular, with the entrance on one of the long sides (conventionally labelled the onus by modern authors).
The most striking difference between the distribution patterns of pebble and tessellated mosaics is that a large proportion of tessellated mosaics were on upper floors, whereas all the known pebble mosaics were on the ground floor: leaving aside those in courtyards and vestibules (which, obviously, are always on the ground floor), tessellated mosaics are almost equally divided between ground and upper floor rooms. I have argued elsewhere that there appears to have been a general inflation in the decoration of private houses by the late Hellenistic period: four times as many houses have mosaics on Delos as at Olynthos, and the Delian houses have wall plaster in every room, whereas at Olynthos plaster was restricted to a fcwr rooms (most frequently the andron), and about half of the houses had none at all.4 Delos was an unusually rich community, and perhaps not strictly typical, but there is enough evidence from elsewhere to support the picture of increased luxury and ostentation. Whereas in the fourth century a house with more than one decorated dining room was exceptional, it is not uncommon for late Hellenistic houses to have several rooms with mosaics and highly elaborate wall decoration. The finds from Delos illustrate another aspect of this decorative inflation: a large quantity of sculpture was found in the houses, ranging from small statuettes of stone or terracotta to life-size marble figures.5 Large-scale sculptures must have been especially prized: where bases survive in situ, it ran be observed that statues were carefully positioned to create attractive vistas, and it is likely that in some cases the placing of sculptures was planned when a house was built or remodelled; moreover, many of the niches in interior walls were probably intended to hold smaller figures. However, unlike mosaics and paintings, sculpture is portable, and it is rarely safe to assume that eventual findspots correspond to original locations: the frequency with which parts of the same statue are found in different rooms or even different houses is instructive. It is therefore rarely possible to give full weight to the part played by sculpture when considering the overall decorative programme of the house. As in the Classical period, the decoration is used to mark out hierarchies of space. The relative importance of the rooms is indicated by a hierarchy of pavement types and by the varying complexity of the designs on the pavements, complemented by wall painting, which was capable of expressing more subtly nuanccd distinctions.
The decoration serves to draw attention to certain areas of the house and to render other areas ‘invisible’. The hierarchy of pavement types seems to be based on the relative cost of the materials. It varies slightly between east and west Greek sites (the West is discussed below). The basic hierarchy in the East ranged from tessellated mosaic (FIG. i a), through pavements of stone chips (FIG. i b}, and then pavements of broken tile pieces set on edge (fig. i c), or occasionally sections of amphora-handles; the most economical flooring of all was beaten earth. Both stone chips and broken tiles must haw been relatively cheap, as they were either waste materials, or could be produced easily by breaking up tiles or blocks of stone, although the foundations of any type of pavement entailed a certain basic cost. The relative value accorded to each type can guessed by comparing the wall paintings associated with them: rooms with tessellated mosaics generally had elaborate wall paintings, whereas those with tile mosaics always had plain white wall plaster, suggesting that tile mosaic was a low-status, functional surface (it is, for example, the standard flooring in latrines). The wall decoration associated with stone chip pavements ranges from plain white plaster to complex polychrome paintings, which suggests dial chip mosaics were considered suitable for a wide variety of rooms, from service quarters to reception rooms, and for open courtyards, where chips had the practical advantages of durability and watertight ness.
There arc also more functional or specialized types of pavement: rough stone slabs are common in courtyards, presumably for entirely practical reasons; a few rooms are paved in terracotta tiles (Delos 20-2, 56), sometimes with incised grooves to make a non-slip surface, which suggests a water-related function (Maison dcs sccaux. room Jtl; and there are still a few rather crude and simply decorated pebble mosaics (Delos 51, 136). There may have been another prestigious type of floor-covering in addition to mosaic, which has left no trace in the archaeological record: the excavators of various houses on Delos observed that a few rooms with exceptionally elaborate wall decoration did not have any trace of a floor at all, not even the dark layer that indicates an earth floor, and they suggested that whatever was dvre had been totally removed—perhaps expensive carpets or fine parquet flooring.’1 Cut stone decoration (opus sectile) was very rarely used in the east at this date, and never for entire pavements. Within this broad hierarchy of pavement types, more subtle hierarchies were possible, depending on the complexity of the decoration and the regularity and size of the materials used.
This can be seen especially clearly on Delos, w here a wide range of housing is preserved, which can be assumed to represent almost the full spectrum of possibilities, from the most modest decoration to the most luxurious, although the very top end of the range is absent.7 Low-grade tile pavements are rarely elaborated, but chip mosaics could be decorated with one or two simple stripe borders; the most common addition, no doubt the cheapest possible, is a single frame in broken terracotta pieces (e.g. Delos 135,254, 260). Conversely, the cost of a tessellated mosaic could be reduced by using less regular materials in combination with tesserae. Many tessellated mosaics have the outer border paved in stone chips, as it would be hidden by the couches used for dining (FIG. 2); in some cases the border is wider than a couch, leaving only a small tessellated carpet in the centre (e.g. Delos 267, 276). Sometimes chips are also used for some of the white bands separating the decorated borders, further reducing the area of tessellation needed {FIG. 3); and in many eases the tesserae themselves arc not very regular (FIG. 13). Thus there arc various gradations between chip mosaic and tessellated mosaic, which probably reflect a range of prices. The time taken to lay the pavement, and thus the labour costs, could be reduced by increasing the size of the tesserae used: a few pavements on Delos arc made with double-sized tesserae, measuring 20 mm a side instead of the usual 8-10 mm, which presumably allowed the same area to be covered in a quarter of the time taken with normal-sized tesserae.
These arc found mainly in the less visible outer borders (Delos 171, 325) and in simple designs, most commonly a chequerboard of black and white tesserae (Delos 10-12, 40, 44-6, 84). In two cases (Delos 10 and 12) the chequerboard forms a carpet with a red frame made of short strips of tile, up to 4 cm long, which again were presumably quicker and easier to lay than individual tesserae. Savings could also be made in the decoration of the pavement, by reducing the complexity of the design: tessellated mosaics tend to have broad plain bands between the decorated borders, and the central field is often left plain or filled with a repeating pattern. Repetitive geometric borders and simple coloured stripes separated by large expanses of plain mosaic would be relatively quick to lay, and might be made by a less skilled mosaicist. Figured work, on the other hand, is rare and usually confined to a small area of the pavement. It must have been very expensive, as it was most effective when executed in tiny tesserae of many colours {opus vtrmiculatum-. FIG. 4 a), often in costly materials such as glass and faience, and it required the services of a highly skilled craftsman for a long period; the many holes where figured panels have been prised out of the pavements attest to their high value. In general, there is a tendency for the finest tesserae and the most colourful and elaborate motifs to be placed nearer the centre, which limited their area, while the size and irregularity of the materials increase towards the edges of the floor.
A customer commissioning a mosaic obviously had to weigh the complexity of the decoration against the size and regularity of the materials, and it is possible to guess at the compromises made by individual customers in order to get the desired area of mosaic for the price that they were able or willing to pay. The most striking example is the Maison des masques, which has four mosaics, three in very large rooms, and all with more than basic geometric decoration: one has two complex rosettes inhabited by tiny birds, and was later further elaborated by the addition of a still life of an amphora and a palm-branch (Delos 217); one has a border of comic masks linked by an ivy scroll (Delos 215); and two are figured, one showing Silcnos dancing to music played by a satyr (Delos 216), and the other Dionysos riding a leopard, flanked by centaurs (Delos 214). In terms of size, individualization and complexity, this is the most impressive mosaic ensemble on Delos. However, the ambitious designs arc not matched by the execution. With the exception of the Dionysos panel (FIG. 4 a), which was a prefabricated embUma, the mosaics arc entirely in standard-sized tesserae, mostly very irregular in shape, and laid carelessly and not very close together (FIG. 4 b). The overall effect is crude, and parts of the figured scenes arc difficult to make out: Silcnos appears to have only one arm. and the object held by one of the centaurs Is an unidentifiable blob. The mosaicists did not even manage to insert the Dionysos tmblana correctly, leaving roughly-patched gaps around the panel. It is hard to escape the conclusion that the owner of the house had aspirations beyond his means.
