Spatial Dimensions in Roman Wall Painting and the Interplay of Enclosing and Enclosed Space: A New Perspective on Second Style

: This article engages with the interplay of two-dimensional and three-dimensional wall decoration in Roman wall decoration of the so-called four Pompeian styles. Instead of describing the rapid changes in the use (or non-use) of techniques for creating perspectival depth in August Mau’s four styles within an autonomous development of decorative principles, either favoring surface over depth, or vice versa, this article will discuss the imaginary space / surface on the walls in relation to the ‘real’ space enclosed by the decorated walls and—foremost—their inhabitants as the actual referent of the decoration. The discussion will focus on second-style wall decoration, with glimpses on the earlier ﬁrst and later third and fourth styles in a ﬁnal section.


Introduction
The development of Roman wall painting is often described as a succession of styles that attempts either at closing or at opening up the surface of the wall. While first style decorations are themselves imitating a wall, this wall is more and more breached through by perspectival vistas in the second style. Towards the third style, however, there is again a preference for closed walls, while this trend is once more inverted in the fourth style, when perspectival vistas enter the decorative repertoire again. 1 1 Any textbook on Roman wall painting may exemplify this. See, e.g., (Mielsch 2001), p. 29, on the beginning of second style: "Die wichtigste Neuerung ist die Öffnung der Wand", and p. 67, on the transition from second to third style: "An einer Reihe von Wänden der frühaugusteischen Zeit [ . . . ] war die Tendenz bemerkbar, die "gebauten" Architekturen zu reduzieren und die räumliche Illusion fast aufzuheben". As it combines elements of the third and second styles, statements on the fourth style found in textbooks become more nuanced, emphasizing only the re-appropriation of perspectival architectural illusion as are found in second style, while elements of surface decoration typical for the third style are still in use. See, e.g., (Croisille 2005, p. 81), on the fourth style in general: "Sur le plan formel, ce "style" se caractérise par son hétérogénéité: il exploite librement les tendances antérieures, insistant tantôt sur l´aspect "architectonique", tantôt sur l´aspect ornemental, dont on a vu les manifestations dans le II e et III e styles". See also (Ling 1991, pp. 71-72). That the differences between the styles with regard to their spatiality are not clear-cut but a matter of relative importance of either surface decoration or spatial illusion has not passed unnoticed, of course. Third-style walls, for example, maintain some elements of spatial illusion. Accordingly, description of them as surface decoration is made only in comparison to second-style walls, see (Zanker 1987, pp. 281-83;Ling 1991, pp. 52-53, 57;Mielsch 2001, pp. 70-73;Croisille 2005, pp. 68-71;Barbet 2009, p. 104). The interplay of two-and three-dimensionality in Roman wall painting is a recurrent topic in the recent volume on the frame edited by V. Platt and M. Squire: (,]B63-arts-480974), on spatial ambiguities in third style wall painting, see also (Platt 2009). On the interplay of two-and three-dimensionality in Roman wall painting, see also (Dietrich 2017, especially pp. 13-21). The differentiation of the styles according to the main criterion of the three-dimensionality or two-dimensionality of the painted wall decoration goes back to the groundbreaking (Mau 1882). For the study of the complexities of 'inside' and 'outside' (another way of tackling the matter of two-and three-dimensionality) created in Roman illusionist wall painting, (Elsner 1995, pp. 49-87), has been groundbreaking. See also (Bergmann 2002).
Concerning the mythological pictures which often occupy the center of these decorated walls, we witness quite the opposite development, as I will soon show. These opposing developments confer to the widely accepted textbook model of alternating the closing and opening-up of the wall from the first to the fourth style an unexpected inner tension. It is on this inner tension that my paper will be based, although I will not focus my discussion on mythological pictures. The phenomenon can be described in short like this: While, generally speaking, there is a strong increase in the use of the effects of spatial depth between third-and the fourth-style walls, the opposite holds true for central mythological pictures. The images of Perseus and Andromeda provide particularly obvious examples of this. While the rescue of Andromeda in the villa of Boscotrecase appears as a mythological landscape (Figure 1), 2 the picture from the Casa dei Dioscuri is almost exclusively focused on the story's two main protagonists and features only very few landscape elements ( Figure 2). 3 There is no quest for spatial depth any more. Katharina Lorenz had clearly carved out this tendency towards a concentration on few main protagonists in fourth-style mythological pictures in her comprehensive study of mythological imagery in Pompeian houses (Lorenz 2008, pp. 38-39). We need, thus, to further differentiate the linear model of increasing spatial depth from third-to fourth style-wall decorations. To the general increase in spatial depth concerning the overall decoration of the wall, there corresponds a decrease in spatial depth concerning the central mythological pictures. 4 From the third-style mythological landscapes in which the stories' main protagonists occupy an almost subordinate position, we pass to fourth-style mythological pictures which renounce spatial surroundings more or less completely.
This opposite development of spatial depth concerning the central mythological pictures shows that we are not only dealing with a 'history of perspective', but rather seeing how the way in which space and landscape is depicted also strongly depends on the figures acting in that space. The relation between the acting figures and their backdrop appears to be essential for the depiction of the space itself. However, acting figures play an important role not only for the central mythological pictures, but also for the decorated walls as a whole. While the former appear on the painted walls themselves, the latter appear in front of these walls. By the latter, I mean the real people acting within the real space enclosed by the walls, for whose proper appearance the whole effort of the splendid painted decoration was done. 5 In this paper, I will thus discuss space depiction and landscape 6 -or indeed their absences-in Roman wall painting not within any autonomous 'history of perspective', 7 but 2 New York, Metropolitan Museum 20.192.16 (central mythological landscape painting from cubiculum 19); for a detailed description and commentary, see (Von Blanckenhagen and Alexander 1990, pp. 33-40); most recently: (Plantzos 2018, pp. 324-25). On Perseus and Andromeda in Roman wall painting general, see (Schmaltz 1989;Hodske 2007, pp. 180-84;Lorenz 2008, pp. 124-49). 3 Naples, Museo Archeologico Nazionale 8998 (from the peristyle of the Casa dei Dioscuri in Pompei); PPP II, 223; PPM IV, 975, Figure 224; (Lorenz 2008, pp. 131-35, Figure 38, pp. 562-64, cat. no. K 31 [on the house and its mythological pictures in general]). 4 On this phenomenon, see already (Dietrich 2017, pp. 18-21). 5 Replacing wall decorations within the Roman house as a specific social context, and especially as the stage of the dominus' representation, instead of treating it as an autonomous type of art, has been one of the major trends in scholarship on Roman wall painting since the 90s of the twentieth century. For this re-orientation, the works of A. Wallace-Hadrill were groundbreaking (see Wallace-Hadrill 1988and Wallace-Hadrill 2009. See also (Zanker 1979(Zanker , 1995(Zanker , 2000Leach 1988Leach , 2004Bergmann 1994). For the interpretation of the mythological imagery in Roman wall painting, the most comprehensive and far-reaching attempt to replace the paintings in the (social and 'atmospheric') context of the Roman house is (Lorenz 2008), with a short introduction on the respective history of scholarship on pp. 3-11, with more bibliography. 6 There is an extensive literature on Roman landscape painting, especially on the so-called sacro-idyllic genre. See, e.g., (Peters 1963;Silberberg 1980;Leach 1988, pp. 197-260;Kotsidu 1998Kotsidu , 2007Mielsch 2001, pp. 179-92;Hinterhöller 2007aHinterhöller , 2007bHinterhöller-Klein 2015;Croisille 2010;Colpo 2010, especially pp. 167-79, andDietrich 2017;La Rocca et al. 2009, pp. 53-54), provides an extensive further bibliography on Roman landscape painting. 7 Indeed, whereas scholarship on Roman wall painting in general has long begun to replace the decorated walls in the social context of the house and its owners (see note 6 here above), this does not necessarily hold true for the more specific topic of spatial depiction. Here, attempts to trace a general 'history of perspective' in the wake of the groundbreaking (Panofsky 1927, English translation: Panofsky 1991) through the (more or less 'geometric') analysis of perspectival vistas on the walls taken on their own are still produced quite regularly, and not necessarily with much attention paid to the social context and finalities of wall decoration. For attempts to reconstruct methods of perspective construction in Roman painting, see most recently the monograph (Hinterhöller-Klein 2015), or (Stinson 2011, pp. 406-8, with earlier bibliography on p. 403, note 2). within their dialectical relation to the acting figures in the respective decorated rooms, be it the painted figures on the space-enclosing walls or the real figures in the enclosed space. Whereas the latter are invisible to us, they were all the more crucial for the viewing context and functional context originally intended. In a first step, I will show from the examples of some second-style walls how strongly the spatial conception of the wall decoration indeed hinges on, and is designed for, these real actors in the room. Then I will pass more quickly to the earlier first style and to the subsequent developments in the third and fourth styles. their absences-in Roman wall painting not within any autonomous 'history of perspective', 7 but within their dialectical relation to the acting figures in the respective decorated rooms, be it the painted figures on the space-enclosing walls or the real figures in the enclosed space. Whereas the latter are invisible to us, they were all the more crucial for the viewing context and functional context originally intended. In a first step, I will show from the examples of some second-style walls how strongly the spatial conception of the wall decoration indeed hinges on, and is designed for, these real actors in the room. Then I will pass more quickly to the earlier first style and to the subsequent developments in the third and fourth styles.  Rocca et al. 2009, pp. 53-54), provides an extensive further bibliography on Roman landscape painting. 7 Indeed, whereas scholarship on Roman wall painting in general has long begun to replace the decorated walls in the social context of the house and its owners (see note 6 here above), this does not necessarily hold true for the more specific topic of spatial depiction. Here, attempts to trace a general 'history of perspective' in the wake of the groundbreaking (Panofsky 1927, English translation: Panofsky 1991) through the (more or less 'geometric') analysis of perspectival vistas on the walls taken on their own are still produced quite regularly, and not necessarily with much attention paid to the social context and finalities of wall decoration. For attempts to reconstruct methods of perspective construction in Roman painting, see most recently the monograph (Hinterhöller-Klein 2015), or (Stinson 2011, pp. 406-8, with earlier bibliography on p. 403, note 2).

Second-Style Wall Decoration: Framing the Actors in the Room
Once one has engaged in a more differentiated view between the transformations of two-and three-dimensionality on Pompeian walls, one may discover many phenomena which seem to contradict the common model. This holds true especially for the idea that the illusionistic breaking of the wall's surface would have been the general characteristic of the second style. But as Ernst Heinrich's study from 2002 has shown, the prime textbook examples from Boscoreale or Oplontis with their perspectival colonnades, opening views on spectacular architecture, are not at all representative for this style in general, but constitute rare exceptions at the summit of decorative splendor. 8 Instead, the standard second-style wall decoration is still the closed monumental masonry wall as we know it from the first style, even though the solely painted decorations now do without the use of plastered imitations of blocks as in the first style. 9 The exceptional, rather than regular, use of grand perspectival illusion is apparent even in the most paradigmatic example of the second style: the villa of Boscoreale itself (Figure 3). 10 Spectacular architectural vistas are to be found only in three 8 (Heinrich 2002, pp. 9-11). However, see already (Mau 1882, pp. 128-30), who noted the existence of such simple second-style decorations. Not having been illustrated in his groundbreaking book, these walls have largely missed out in subsequent studies on second-style wall decoration. 9 The wall decoration of the Casa di Cerere as depicted in (Heinrich 2002, pp. 78-83, cat. no. 10-17, Figures 15-38), provides an example of such standard second-style walls. Only the more lavishly decorated rooms make some use of the motif of projecting columns in the foreground, deemed paradigmatic of second-style wall decorations as a whole. 10 The scholarly literature on the Boscoreale Villa is far too large to be cited here. The most comprehensive and up-to-date presentation of the frescoes in their architectural context is currently (Barbet and Verbanck-Piérard 2013), vol. 1 with a thorough and well-illustrated room-by-room description of the Villa's wallpaintings and with extremely useful reconstructions by F. Ory, incorporating the extant fragments of the

Second-Style Wall Decoration: Framing the Actors in the Room
Once one has engaged in a more differentiated view between the transformations of two-and three-dimensionality on Pompeian walls, one may discover many phenomena which seem to contradict the common model. This holds true especially for the idea that the illusionistic breaking of the wall's surface would have been the general characteristic of the second style. But as Ernst Heinrich's study from 2002 has shown, the prime textbook examples from Boscoreale or Oplontis with their perspectival colonnades, opening views on spectacular architecture, are not at all representative for this style in general, but constitute rare exceptions at the summit of decorative splendor. 8 Instead, the standard second-style wall decoration is still the closed monumental masonry wall as we know it from the first style, even though the solely painted decorations now do without the use of plastered imitations of blocks as in the first style. 9 The exceptional, rather than regular, use of grand perspectival illusion is apparent even in the most paradigmatic example of the second style: the villa of Boscoreale itself ( Figure 3). 10 Spectacular architectural vistas are to be found only in three rooms in the northern wing: 8 (Heinrich 2002, pp. 9-11). However, see already (Mau 1882, pp. 128-30), who noted the existence of such simple second-style decorations. Not having been illustrated in his groundbreaking book, these walls have largely missed out in subsequent studies on second-style wall decoration. 9 The wall decoration of the Casa di Cerere as depicted in (Heinrich 2002, pp. 78-83, cat. no. 10-17, Figures 15-38), provides an example of such standard second-style walls. Only the more lavishly decorated rooms make some use of the motif of projecting columns in the foreground, deemed paradigmatic of second-style wall decorations as a whole. 10 The scholarly literature on the Boscoreale Villa is far too large to be cited here. The most comprehensive and up-to-date presentation of the frescoes in their architectural context is currently (Barbet and Verbanck-Piérard 2013), vol. 1 with a thorough and well-illustrated room-by-room description of the Villa's wall-paintings and with extremely useful reconstructions by in the large reception-room H opening directly to the peristyle (Figure 4a-c), 11 and in the two smaller and more withdrawn rooms G ( Figure 5) 12 and M (Figure 6a,b), 13 which may be entered only through an anteroom. In all other sectors of the villa-the fauces (Figure 7), 14 the peristyle, or in the exedra-like rooms D, N or L-we find 'only' painted masonry walls with projecting columns in the foreground, painted in the highest quality indeed, but without any perspectival vistas that are commonly thought of being so typical of the second style. The respective positioning of the rooms with perspectival vistas and those with closed walls shows a clear dependence on the social hierarchical structure of the Roman house as analyzed by  The architectural vista or, in the case of Cubiculum M, the sacro-idyllic-Dionysiac landscape, is a feature reserved for the rooms at the highest rank of the domestic space's social hierarchy. Significantly, the most spectacular perspectival colonnades are not found in the biggest room H, which was designed for the reception of larger groups of people, but in the small Cubiculum M, which was designed for selected and for therefore all the more high-ranking guests. The perspectival vistas breaching through the surface of the wall are thus not a general characteristic of the second style but appear to be a means of distinction for elevating particular rooms within the house's social hierarchy. However, this social hierarchy is, of course, first and foremost a hierarchy between people and groups of people. The design and decoration of the domestic space should only fulfil the function of facilitating the performative reproduction and experience of this hierarchy in the house's ritualized everyday life. Therefore, we may infer that the perspectival breaking of the wall's surface was intended to have its actual effect only through the interplay with the acting figures in the room. In the analysis of the few most celebrated architectural vistas of the second style, which will now follow, we thus have to bear in mind this actual point of reference of the wall decoration.
F. Ory, incorporating the extant fragments of the frescoes in line drawings of the walls' overall decorative designs (except Cubiculum M), vol. 2 with a collection of articles on the villa and its frescoes, making extensive use of digital reconstructions. The most recent overview is (Plantzos 2018, pp. 313-17). Issues of spatiality and framing relevant to the present argument are tackled by V. Platt in (Squire and Platt 2017, pp. 102-8). For the present aim of linking wall decoration, the architecture of the respective rooms, and their inhabitants, (Bergmann et al. 2010), which presents the villa through a digital model that projects the frescoes back on the walls, is key. A more traditional account of perspectival painting in this villa is found in (Ehrhardt 1991, pp. 42-46). An extensive (though not complete) bibliography (especially on the frescoes of Cubiculum M) is to be found on the website of the Metropolitan Museum (https://www.metmuseum.org/art/collection/search/247017, last viewed on 21 January 2019) For a useful virtual tour realized by the King's Visualization Lab (King's College London) in cooperation with the Metropolitan Museum and the eight other museums owning paintings from the Boscoreale Villa, see https://www.metmuseum.org/metmedia/video/collections/gr/boscoreale-model (last viewed on 21 January 2019); on this digital visualization, see (Beacham et al. 2013).       On asking concretely how the perspectival vistas through the enclosing walls may have functioned in the interplay with the acting figures in the enclosed space, there is an important observation to be made. Even in the case of the few outstanding examples of walls with architectural vistas that illustrate every textbook, emphasis lies not at all on the perspectival opening of the wall's surface exclusively. Equally noteworthy is how the viewer is systematically refused full view of the splendid architectures occupying the center of these walls, by some strategically positioned shear wall or parapetasma. Indeed, in most cases, the viewer of the architectural vista is given an undisturbed view into the spatial depth only in the wall's upper zone, while the middle zone is closed by a shear wall, in such a way that he/she never properly sees the architectures peeping over this shear wall. 16 This phenomenon of an only partially opened view is particularly noteworthy when a prominent motif like a round temple or a monumental tripod on which the whole architectural vista is oriented occupies the center of the wall. One may expect that the viewer would at least be allowed full view of such a compositional center-but even in these cases the view tends to be disturbed. In the often illustrated double-alcove room 16 of the Villa dei Misteri, we get a glimpse only of a round temple's splendid roof. 17 On the lavish second-style wall of oecus 15 of the Oplontis villa (Figure 8), 18 16 The prominence of devices closing the view in the middle zone of second-style walls has, of course, already been noted in scholarship, see, e.g., (Hiller 1988, especially p. 200). However, this structural (as I claim) characteristic of second-style decorations has been discussed as a transitional characteristic of earlier second style's breaking through the wall not having yet fully reached its (teleological) end. See, e.g., (Ehrhardt 1991), who stresses the fact that earlier second-style walls (as those from the Villa dei Misteri) tend to be relatively more closed than later ones (as those from the Villa of Oplontis). He is perfectly right in noting that those deep spaces behind the shear walls are presented as being-in principle-accessible, see p. 47 (on Oplontis). However, it is precisely this accessibility that makes the closure all the more noteworthy. On shear walls in second style, see also (Borbein 1975). 17 For splendid photographs of the villa's frescoes, see (Mazzoleni and Pappalardo 2005, pp. 102-24), for cubiculum 16, see pp. 104, 109-11. On the second-style perspectival vistas in Cubiculum 16 of the Villa dei Misteri, see (Ehrhardt 1991, pp. 37-42, with useful line drawings on pp. 36-37, Figures 1-2). On the effects of spatial illusion created by painted wall decoration in the Villa dei Misteri, see (Elsner 1995, pp. 62-71). 18 For splendid photographs of the villa's frescoes, see (Mazzoleni and Pappalardo 2005, pp. 126-64), for oecus 15, see pp. 136-37. For a comprehensive study of the villa in its various aspects, see most recently (Gazda On asking concretely how the perspectival vistas through the enclosing walls may have functioned in the interplay with the acting figures in the enclosed space, there is an important observation to be made. Even in the case of the few outstanding examples of walls with architectural vistas that illustrate every textbook, emphasis lies not at all on the perspectival opening of the wall's surface exclusively. Equally noteworthy is how the viewer is systematically refused full view of the splendid architectures occupying the center of these walls, by some strategically positioned shear wall or parapetasma. Indeed, in most cases, the viewer of the architectural vista is given an undisturbed view into the spatial depth only in the wall's upper zone, while the middle zone is closed by a shear wall, in such a way that he/she never properly sees the architectures peeping over this shear wall. 16 This phenomenon of an only partially opened view is particularly noteworthy when a prominent motif like a round temple or a monumental tripod on which the whole architectural vista is oriented occupies the center of the wall. One may expect that the viewer would at least be allowed full view of such a compositional center-but even in these cases the view tends to be disturbed. In the often illustrated double-alcove room 16 of the Villa dei Misteri, we get a glimpse only of a round temple's splendid roof. 17 On the lavish second-style wall of oecus 15 of the Oplontis villa (Figure 8), 18 the view of the central tripod is disturbed by a door lintel. Doors or other passageways are often found in the center of such walls. Other than in this case where it is open, such passages have a strong tendency to be closed, as in room G of the Boscoreale-Villa ( Figure 5). The double-wing door on the sidewall is closed. The central propylon on the main wall has no door, but access is nevertheless denied, be it only to the viewer, by a wooden fence. The splendor of the architecture hidden behind the shear wall is suggested to the view only by what he/she glimpses over this shear wall. Yet the desire created for a fuller view is consciously denied. Closed doors or otherwise obstructed pathways are surprisingly often found on second-style walls. Where the central opening seems to provide a way into the space hidden behind the shear wall, we find doors shut, parapetasmata hung, or a wooden fence built in the way, or some object placed directly behind the opening which disturbs the view. The painted architectural vistas of those ever cited splendid second-style walls thus make a promise that is not fulfilled on the wall's most crucial spot.
Is there more allowance or more disturbance of the vista breaking the wall's surface in such second-style decorations? In comparison with the earlier first-style wall decoration, the dominant impression is that of an allowed vista. Viewed on its own and compared to what would have been possible, the dominant impression on such second-style walls may nevertheless be that of a denied vista. From the perspective of modern-age examples of illusionistic wall painting, it would in any case seem a rather funny idea to cover the lower two-thirds of the picture by a shear wall. the view of the central tripod is disturbed by a door lintel. Doors or other passageways are often found in the center of such walls. Other than in this case where it is open, such passages have a strong tendency to be closed, as in room G of the Boscoreale-Villa ( Figure 5). The double-wing door on the sidewall is closed. The central propylon on the main wall has no door, but access is nevertheless denied, be it only to the viewer, by a wooden fence. The splendor of the architecture hidden behind the shear wall is suggested to the view only by what he/she glimpses over this shear wall. Yet the desire created for a fuller view is consciously denied. Closed doors or otherwise obstructed pathways are surprisingly often found on second-style walls. Where the central opening seems to provide a way into the space hidden behind the shear wall, we find doors shut, parapetasmata hung, or a wooden fence built in the way, or some object placed directly behind the opening which disturbs the view. The painted architectural vistas of those ever cited splendid second-style walls thus make a promise that is not fulfilled on the wall's most crucial spot. Is there more allowance or more disturbance of the vista breaking the wall's surface in such second-style decorations? In comparison with the earlier first-style wall decoration, the dominant impression is that of an allowed vista. Viewed on its own and compared to what would have been possible, the dominant impression on such second-style walls may nevertheless be that of a denied vista. From the perspective of modern-age examples of illusionistic wall painting, it would in any case seem a rather funny idea to cover the lower two-thirds of the picture by a shear wall.  As I have already said, it is often precisely the center of the wall where the perspectival vista is hidden from the viewer by a shear wall, a closed door or some other device preventing an undisturbed view into the illusionary space behind the wall's surface. This is certainly the reason why from the most celebrated of all second-style walls-those from cubiculum M of the Villa of Boscoreale-on the side wall from cubiculum, most textbooks illustrate not the central panel, but one of the side panels, where the view into the depths of the formidable architectural perspective is less obstructed (Figure 6b). However, it is obvious that the two side panels, with their symmetrical disposition, did not constitute themselves the main focus of attention, but had the function to direct the gaze toward the wall's center-and, in this center, on which the perspectival architecture are running from either side, a free view into spatial depth is denied by a shear wall covering two-thirds of the wall's height. Hence, the general rule in second-style walls, that the view into the distance is granted only in the wall's upper zone, holds true in this case only for the central panel. In this example, we clearly have an exception that confirms the principle according to which perspectival opening of the wall's surface happens on the wall's margins and is precisely not intended in its most central and important zone. One may also describe the phenomenon within the dialectics of framing, a topic which has recently received increased scholarly attention. 19 In this instance, the emphasis, in the use of the means of perspectival painting, lies on the framing margins, not on the framed center of the wall. This is all the more significant because the perspectival architectural vistas on the wall's periphery have their vanishing point in the wall's center, and therefore focus the attention on this center even more. This center can be emphasized by other means too, as for example by positioning a round temple or some other tall motif there. Furthermore, all those doors, entrances or wall recesses that we find in this zone help to mark the wall's center. How is it, then, that the view in the distance tends to be denied precisely in this focal point? I would like to explain this surprising fact in the following way: the center of the room so strongly emphasized by the wall decoration is reserved not for an element of the pictorial illusion, but for the real people in the room. After all, the actual function of wall decoration is to provide a proper frame for the appearance and experience of these real people. In the same way as architectural vistas help to distinguish certain rooms within the social hierarchy of domestic space in the domus as a whole, they also help to constitute a spatial hierarchy within the single room itself, enabling it to host people of different ranks and to mark, in their interaction, those subtle differences to which the decoration of domestic space is striving so much.
The central zone which a second-style architectural prospect is both emphasizing and leaving out may, in some cases, coincide concretely with the positioning of important figures in the room. This is the case in a double alcove room in the Villa dei Misteri (a textbook example of early second style), as clearly visualized in its digital reconstruction (Figure 9). 20 The top of a round temple, on which the vanishing lines are running from both sides, is 'floating' exactly over the niche intended for the setting of a klinai (typical couch on high legs for reclining at dinner and other occasions). 21 Another alcove marks the position of another klinai. However, the painted backdrop of this niche has both a simpler architectural prospect and a less 'centralizing' one. The room thus singles out two positions for klinai, but by means of a more complex and more emphatically centralized architectural vista in the wall's upper zone, one of them is set a little higher in the room's spatial hierarchy. But of course, the architectural prospect is only framing the people lying on these klinai, for whom the architectural prospect provides the decorum proper to their respective rank in their appearance in front of the decorated wall. With its vanishing lines leading into the spatial distance, the architectural 19 See (Squire and Platt 2017). 20 On the perspectival vistas in this room, see (Ehrhardt 1991, pp. 37-42, Figures 1 and 2). On rooms with alcoves and their function within an 'archaeology of intimacy' amidst the houses' social hierarchy, see (Anguissola 2010). 21 A very good analysis of the typical situations of social interaction of the dominus with his guests in the context of Pompeian domestic architecture and its mobile furniture (klinais, but also cathedrae) is provided in (Dickmann 1999, pp. 281-87, without special attention paid to wall painting). On alcove rooms as a setting for the more intimate interactions between people, see (Anguissola 2010).
prospect is not meant to drive the viewer's attention away from the acting figures in the real space, but inversely it is meant to focus, with subtly differentiated intensity, attention on them. In perfect consistency with this aim, the vanishing lines of the architectural illusion are not extended into the wall's middle zone, where they would compete with the appearance of the real figure lying on his klinai, but are withheld within the periphery of the wall-and the periphery of the viewer's field of vision. 22 The shear wall relegating the architectural prospect to the wall's upper margins has the function of a backdrop for the real figures' appearance on the stage of domestic space.
Arts 2019, 8, x FOR PEER REVIEW 11 of 26 but inversely it is meant to focus, with subtly differentiated intensity, attention on them. In perfect consistency with this aim, the vanishing lines of the architectural illusion are not extended into the wall's middle zone, where they would compete with the appearance of the real figure lying on his klinai, but are withheld within the periphery of the wall-and the periphery of the viewer's field of vision. 22 The shear wall relegating the architectural prospect to the wall's upper margins has the function of a backdrop for the real figures' appearance on the stage of domestic space.

A Glimpse at Other Media: Figure and Architectural Ground on Marble Reliefs and on an Ancient Theatre's Stage
The same kind of mise en scène of acting figures in an architectonic frame, where a shear wall separates them from the architecture set in a deep space, is found in other pictorial media of this epoch too, as on the well-known relief types of the Apolline Triad ( Figure 10) 23 or on the so-called Ikarios reliefs ( Figure 11). 24 Just as the real figures in the second-style rooms remain separated from the architectural prospect restricted to the upper wall zone, the acting figures on these reliefs do not mingle with the architecture in the background. The spatial illusion created by the architectural prospects does not have the function of projecting the figures into that distant pictorial space, but rather of further focusing the attention on them and thus of underscoring their presence in the real space. This distinguishes these kinds of architectonic backgrounds sharply from what may seem to be a close modern equivalent, namely the built scenery of a motion picture. Despite the total artificiality of such built scenery, the acting figures would be filmed in such a way as to make them merge completely into this (illusionary) space, to make the viewer believe that the figures effectively act in this remote space, and to neatly separate them from the viewers' (real) space. 22 (Muth 2007), to which this article owes much, makes a similar point concerning the astonishingly small role played by sculpture in the otherwise over-densely decorated late antique Villa of Piazza Armerina. In order not to overshadow the dominus' appearance in front of his guests, the focal points of the Villa's many apsidial rooms-where the dominus or other important people would have been seated-are precisely not taken by statues, even though such apses would otherwise constitute the perfect location for the setting of statuary. 23 Berlin, Antikensammlung Sk 921. On the type, see (Cain 1989;Di Franco 2017, pp. 23-37). 24 Paris, Louvre MA 1606. The first discussion of the type is (Hauser 1889, pp. 189-99). Extensive further bibliography is provided on the website of the British Museum (concerning the London replica of the same type [British Museum 1805,0703.123]). The same kind of mise en scène of acting figures in an architectonic frame, where a shear wall separates them from the architecture set in a deep space, is found in other pictorial media of this epoch too, as on the well-known relief types of the Apolline Triad ( Figure 10) 23 or on the so-called Ikarios reliefs (Figure 11). 24 Just as the real figures in the second-style rooms remain separated from the architectural prospect restricted to the upper wall zone, the acting figures on these reliefs do not mingle with the architecture in the background. The spatial illusion created by the architectural prospects does not have the function of projecting the figures into that distant pictorial space, but rather of further focusing the attention on them and thus of underscoring their presence in the real space. This distinguishes these kinds of architectonic backgrounds sharply from what may seem to be a close modern equivalent, namely the built scenery of a motion picture. Despite the total artificiality of such built scenery, the acting figures would be filmed in such a way as to make them merge completely into this (illusionary) space, to make the viewer believe that the figures effectively act in this remote space, and to neatly separate them from the viewers' (real) space.
The second-style wall with architectural prospect framing the man lying on his klinai, but nevertheless emphasizing his real presence in front of his interaction partners, calls for another comparison, namely the scenae frons of an ancient theatre. Indeed, the scenae frons both provides a powerful frame for the actors' (or, in the case of rhetorical performances, for the orators') appearance in front of the spectators and leaves no doubt at any time about their concrete standing on the theatre's material stage, sharing the same real space with the spectators, whatever illusionary spaces the theatrical performance may allude to. Therefore, the endless scholarly discussions on the relation between Roman wall painting and theatre painting, 25 despite their failure to provide indisputable evidence of wall painting's genetically depending on the latter, may nevertheless be grounded in some other way: both are similar in essence, in so far as they serve the similar end of providing a frame to the 'actors' on the stage/in the room-a frame, however, that is not meant to draw a clear line of separation between them and their 'spectators'. The second-style wall with architectural prospect framing the man lying on his klinai, but nevertheless emphasizing his real presence in front of his interaction partners, calls for another comparison, namely the scenae frons of an ancient theatre. Indeed, the scenae frons both provides a powerful frame for the actors' (or, in the case of rhetorical performances, for the orators') appearance in front of the spectators and leaves no doubt at any time about their concrete standing on the theatre's material stage, sharing the same real space with the spectators, whatever illusionary spaces the theatrical performance may allude to. Therefore, the endless scholarly discussions on the relation between Roman wall painting and theatre painting, 25 despite their failure to provide indisputable evidence of wall painting's genetically depending on the latter, may nevertheless be grounded in some other way: both are similar in essence, in so far as they serve the similar end of providing a frame to the 'actors' on the stage/in the room-a frame, however, that is not meant to draw a clear line of separation between them and their 'spectators'. Based on a notice in Vitruvius, the idea was first discussed in relation to extant Roman wall painting in (Beyen 1957, specifically on the Boscoreale Villa). Recent discussions, no more necessarily interested in the question of historic origin, include (Leach 2004, pp. 93-122;Gros 2008;Lorenz 2013, pp. 368-71). The second-style wall with architectural prospect framing the man lying on his klinai, but nevertheless emphasizing his real presence in front of his interaction partners, calls for another comparison, namely the scenae frons of an ancient theatre. Indeed, the scenae frons both provides a powerful frame for the actors' (or, in the case of rhetorical performances, for the orators') appearance in front of the spectators and leaves no doubt at any time about their concrete standing on the theatre's material stage, sharing the same real space with the spectators, whatever illusionary spaces the theatrical performance may allude to. Therefore, the endless scholarly discussions on the relation between Roman wall painting and theatre painting, 25 despite their failure to provide indisputable evidence of wall painting's genetically depending on the latter, may nevertheless be grounded in some other way: both are similar in essence, in so far as they serve the similar end of providing a frame to the 'actors' on the stage/in the room-a frame, however, that is not meant to draw a clear line of separation between them and their 'spectators'. Based on a notice in Vitruvius, the idea was first discussed in relation to extant Roman wall painting in (Beyen 1957, specifically on the Boscoreale Villa). Recent discussions, no more necessarily interested in the question of historic origin, include (Leach 2004, pp. 93-122;Gros 2008;Lorenz 2013, pp. 368-71). 25 Based on a notice in Vitruvius, the idea was first discussed in relation to extant Roman wall painting in (Beyen 1957, specifically on the Boscoreale Villa). Recent discussions, no more necessarily interested in the question of historic origin, include (Leach 2004, pp. 93-122;Gros 2008;Lorenz 2013, pp. 368-71).

'Mistakes' in Perspectival Construction and How to Make Sense of Them
As the person in front of the painted wall is not at all meant to be projected into the illusionary space behind him, there is no need for the created illusion of deep space to be perfect and flawless. This might explain many of the presumed 'mistakes' in the perspectival construction on Roman walls noticed by archaeologists of the past. A typical 'mistake' can be detected on the rows of consoles from the Cubiculum M from Boscoreale. The consoles sustaining the upper profile of the socle in the wall's middle zone are painted in a strictly parallel left half profile on the left side of the wall, as well as in an equally strictly parallel right half profile on the right side, resulting in a sharp angle in the center of the wall, where both rows of consoles meet (Figure 12). This 'problem' of perspectival construction is most common in Roman decorative wall painting or mosaic. 26 The sudden change in orientation of the vanishing lines is definitely not helpful to the spatial illusion. However, as soon as one understands the use of perspectival depiction as a means of structuring and centralizing the space in front of the wall, this incorrect central perspective does the job even better. Indeed, the left row of oblique consoles 'draws' the gaze to the right, and the right row 'draws' the gaze to the left, while the sudden change in orientation where both rows meet emphasizes the wall's center even more.

'Mistakes' in Perspectival Construction and How to Make Sense of Them
As the person in front of the painted wall is not at all meant to be projected into the illusionary space behind him, there is no need for the created illusion of deep space to be perfect and flawless. This might explain many of the presumed 'mistakes' in the perspectival construction on Roman walls noticed by archaeologists of the past. A typical 'mistake' can be detected on the rows of consoles from the Cubiculum M from Boscoreale. The consoles sustaining the upper profile of the socle in the wall's middle zone are painted in a strictly parallel left half profile on the left side of the wall, as well as in an equally strictly parallel right half profile on the right side, resulting in a sharp angle in the center of the wall, where both rows of consoles meet (Figure 12). This 'problem' of perspectival construction is most common in Roman decorative wall painting or mosaic. 26 The sudden change in orientation of the vanishing lines is definitely not helpful to the spatial illusion. However, as soon as one understands the use of perspectival depiction as a means of structuring and centralizing the space in front of the wall, this incorrect central perspective does the job even better. Indeed, the left row of oblique consoles 'draws' the gaze to the right, and the right row 'draws' the gaze to the left, while the sudden change in orientation where both rows meet emphasizes the wall's center even more. As a general rule, the space hierarchizing effect of architectural perspective on second-style walls reveals itself even more clearly on the 'wrong' perspectives than on the 'correct' perspectives. The architectural prospect in room 15 of Oplontis is a good example of this. While the architectural vista seen from a central position gives a coherent impression and provides a splendid frame to what happens in front of this wall, the same architectural vista seen from a more peripheral position 26 The earliest occurrence of this phenomenon known to me is the dentil frieze framing the Alexander mosaic from the late second century BC Casa del Fauno. The illusionistic rows of small blocks depicted in oblique view suddenly change their left or right orientation in the center of each side of the frame (see, e.g., Andreae 1977, p. 76, Figure 25). As in later Roman perspectival wall painting, the mosaicist did not make any attempt to conceal this 'mistake'. The same 'mistake' would probably have embarrassed an Italian Renaissance painter eager to master central perspective. Indeed, analogous problems with changing orientations in the center of a picture also occur in Late Medieval/Proto-Renaissance perspectival painting (typically with painted paved floors). Accordingly, the artists' coping with such problems constitutes one of the crucial aspects in Panofsky's famous account of the development of Renaissance perspective (Panofsky 1927   As a general rule, the space hierarchizing effect of architectural perspective on second-style walls reveals itself even more clearly on the 'wrong' perspectives than on the 'correct' perspectives. The architectural prospect in room 15 of Oplontis is a good example of this. While the architectural vista seen from a central position gives a coherent impression and provides a splendid frame to 26 The earliest occurrence of this phenomenon known to me is the dentil frieze framing the Alexander mosaic from the late second century BC Casa del Fauno. The illusionistic rows of small blocks depicted in oblique view suddenly change their left or right orientation in the center of each side of the frame (see, e.g., Andreae 1977, p. 76, Figure 25). As in later Roman perspectival wall painting, the mosaicist did not make any attempt to conceal this 'mistake'. The same 'mistake' would probably have embarrassed an Italian Renaissance painter eager to master central perspective. Indeed, analogous problems with changing orientations in the center of a picture also occur in Late Medieval/Proto-Renaissance perspectival painting (typically with painted paved floors). Accordingly, the artists' coping with such problems constitutes one of the crucial aspects in Panofsky's famous account of the development of Renaissance perspective (Panofsky 1927  what happens in front of this wall, the same architectural vista seen from a more peripheral position appears to be completely distorted ( Figure 13). The architectural prospect thus functions like a magnifying glass that makes the viewer experience his relative distance from the room's center in an over-accentuated manner. More generally, one may say that the architectural prospects are not intended for the single 'perfect' viewpoint of an implicit beholder, but for acting and moving people within the room-people both viewing the decorated walls from whatever centralized or marginal spot they happen to be 'seated' within the hierarchically structured space, and being seen (by other people) in front of whatever centralized or marginal section of the decorated wall, making them appear in a more or less advantageous setting. The centralizing effect of architectural prospects on the walls, both through the formidable views and settings they produce for those well positioned, and through the equally formidable distortions they produce for those less well positioned, transposes social hierarchy to a somehow corporeal level. Everyone's-including one's own-position within the social microcosm of a morning reception or a dinner party may be experienced visually. appears to be completely distorted ( Figure 13). The architectural prospect thus functions like a magnifying glass that makes the viewer experience his relative distance from the room's center in an over-accentuated manner. More generally, one may say that the architectural prospects are not intended for the single 'perfect' viewpoint of an implicit beholder, but for acting and moving people within the room-people both viewing the decorated walls from whatever centralized or marginal spot they happen to be 'seated' within the hierarchically structured space, and being seen (by other people) in front of whatever centralized or marginal section of the decorated wall, making them appear in a more or less advantageous setting. The centralizing effect of architectural prospects on the walls, both through the formidable views and settings they produce for those well positioned, and through the equally formidable distortions they produce for those less well positioned, transposes social hierarchy to a somehow corporeal level. Everyone's-including one's own-position within the social microcosm of a morning reception or a dinner party may be experienced visually. Rooms whose painted architectural prospects have more than one vanishing point-another clear 'mistake' in perspective!-are of particular interest in this respect. Again, Cubiculum M of Boscoreale provides a perfect example (Figure 6a,b). The rear end of the room designed for the positioning of a klinai is granted its own prospect, with a kind of a 'private' vanishing point. This cubiculum thus disposes of two centers-one reserved for the sole dominus on his klinai, and one for all his guests.

Vistas on Natural Landscapes: More of the Same? The Rear Wall of Cubiculum M and the Case of the Odyssey Frescoes from the Esquiline
What has been said on the architectural landscapes of the second style is paralleled in interesting ways by the rarer natural landscapes. 27 While the case of the bucolic dionysiac on the rear wall of Cubiculum M from Boscoreale ( Figure 6a) seems to invite a similar interpretation, the well-known Odyssey frescoes from the Esquiline (Figure 14) might have made use of centralizing effects for rather different purposes. Cncerning the rear wall of Cubiculum M with its bucolic dionysiac landscape, in 27 For bibliography on Roman landscape painting, see above note 7. Rooms whose painted architectural prospects have more than one vanishing point-another clear 'mistake' in perspective!-are of particular interest in this respect. Again, Cubiculum M of Boscoreale provides a perfect example (Figure 6a,b). The rear end of the room designed for the positioning of a klinai is granted its own prospect, with a kind of a 'private' vanishing point. This cubiculum thus disposes of two centers-one reserved for the sole dominus on his klinai, and one for all his guests. Cubiculum M from Boscoreale ( Figure 6a) seems to invite a similar interpretation, the well-known Odyssey frescoes from the Esquiline (Figure 14) might have made use of centralizing effects for rather different purposes. Cncerning the rear wall of Cubiculum M with its bucolic dionysiac landscape, in front of which the dominus would receive his most distinguished guests, the structural parallels to what has been said on the second-style architectural prospects are obvious. Only the upper part of the wall opens up a view into a deep space, whereas the middle zone of the wall, which constitutes the backdrop to the dominus' appearance on his klinai, is covered by rocks. Finally, in the center of this zone, there is a kind of balustrade of shimmering gold with a monochrome architectural landscape. This golden backdrop both provides a proper frame for the dominus and prevents him from merging too much with the rocky landscape on the wall. Where the view into deep space is allowed in the upper zone, symmetric perspectival pergolas prevent the viewer's gaze from wandering off onto the distance of the landscape and orient it firmly to the wall's center, where the dominus appears as the actual subject of the room's prestigious decoration.
Structural parallels with what has been said on architectural landscapes are apparent with the Odyssey frescoes from the Esquiline too, at least concerning the use of similar decorative techniques, but not necessarily concerning the pursued goal. This long row of mythological landscapes with scenes from the Odyssey was arranged in the upper zone of a long corridor, in such a way as if these were views between pilasters reaching down to the socle, but opening the view only in their upper third ( Figure 14). 28 The wall thus keeps a closed surface on the viewers' eye-level, and provides a plain background to those wandering up and down this corridor. But the wall with the Odyssey-landscapes bears similarities to the architectural prospects discussed until now on not only this level. Also, the underscoring of a center and the orientation of all parts of the wall towards this center can be found again. Just as in a normal architectural wall decoration, all pilasters on the left side of the wall are shown in strictly parallel left half profile, and those on the right side respectively in strictly parallel right half profile, in such a way that it makes all the perspectival vistas in the upper wall zone 'run' towards the center of this long corridor. It is no coincidence that, in the central panel, where the flanking pilasters are oriented toward one another, we see the palace of Kirke. This distinguishes itself as the most 'high-ranking' and decorous motif from the raw rocky landscapes to the left and to the right. The single mythological scenes, arranged in strict narrative order, lead according to the epic poem toward the palace, and then away from it, thereby making any position in this long corridor correspond in its relative spatial distance from the corridor's center to the temporal distance from the central palace-episode within the mythical narrative. 29 The parallel with the techniques of centralization analyzed above is nonetheless not yet complete. Indeed, while the decorated rooms discussed previously are designed for rather static social interactions-as, e.g., a reception with the dominus and his guests lying on klinai-the long corridor with its Odyssey frescoes is a space designed for movement, with no fixed spot assigned to specific people. Here, the described centralizing effects of the painted decoration might have less to do with emphasizing social hierarchy than with inviting people to perambulate to and fro in this corridor, instead of using it as a mere functional passage, as corridors tended to be used in private architecture since the nineteenth century. The decoration 'draws' gazes ever back to its center.

Framing the Actors in the Room in Second Style: Summary
What has been said on the uses of the means of perspectival painting and spatial illusion in Roman second-style wall painting and their function in relation to the decorated rooms as a social context may now be summarized in the following way. (1) The perspectival opening of the wall's surface, which is commonly thought of as the general characteristic of the second style, has proved to be a means of distinguishing and highlighting particular rooms within the socio-hierarchical structure of the house. The famous second-style architectural landscapes set in a deep space thus stand in a general context in which the 'first style-like' closed wall continues to be the rule.
(2) But also among the relatively few outstanding walls with architectural prospects, the closed wall surface continues to play an essential role. Closed surfaces typically cover the wall's middle zone, roughly on the eye-level of viewers standing or lying on their klinai, while perspectival deep space opens up the wall's margins, especially in its upper zone. Those closed surfaces created by shear walls, hanging textiles (parapetasmata) or closed doors can be interpreted as a backdrop for the appearance of the real actors in the space enclosed by the decorated walls, providing a proper frame and decorum for their social interaction.
(3) For the decorated wall's function as a stage for the dominus and his distinguished guests, the perspectival effects of recessing deep space in the second style have their particular aptitude in creating a strong orientation towards the center of the wall. This decorative strategy on the walls results in a differentiation, within the enclosed space, between more central and more marginal sectors, and thus actively supports the socio-hierarchical differentiation of the real actors in the room and their interactions over subtle differences of rank. In clear contrast to modern-age central perspective, the actual reference-point of spatial illusion on the walls is the real space enclosed by the decorated walls, and not the pictorial space. Likewise, these decorations are not primarily intended for viewers, but for actors in space who are moving, interacting with one another, viewing and being viewed.
(4) Accordingly, the architectural vistas on the walls, although they all have a very clear center, are not oriented towards a single and consistent vanishing point, calling for an implicit viewer, but instead tolerate and even invite many 'mistakes' in the perspectival construction. (5) Finally, close parallels could be drawn between architectural prospects and the rarer natural prospects of second-style walls, at least concerning the decorative techniques used therein. My two case studies, the bucolic dionysiac landscape in Cubiculum M and the Odyssey frescoes from the Esquiline, both function within a similar structure of the wall's opening in its upper margins and closure in its middle zone, together with the perspectival vistas' orientation towards the center of the long corridor from either side. However, only in Cubiculum M does this serve the same purpose of emphasizing social hierarchy, while the same decorative techniques in the case of the Esquiline frescoes might be understood in relation to the function of the corresponding corridor as a space of movement.

Towards a Larger Picture: Framing the Actors in the Room and the Succession of Styles in Roman Wall Painting
Since the results of my main argument have already been summarized above, I would like to present in a final section some preliminary thoughts about how the phenomena observed here on second-style wall decorations might prompt us to rethink more globally our explanatory model for the succession of styles in Roman wall painting. For the characteristics of second-style wall decorations in their handling of opened and closed wall surfaces, which are designed to provide a stage to the social interactions taking place in these rooms, we find close structural parallels in wall decoration systems that precede the second style and those which were developed later. On the walls decorated in the so-called first style, which imitate monumental masonry walls through stucco and painting, two peculiarities should attract our attention. First, the socle on which the actual masonry wall is based is usually absurdly high, when compared to real built masonry walls. This high socle lifts the subsequent orthostate layer up to the wall's middle zone. Secondly, these orthostates raised to about eye-level seem to be strongly oversized when compared with the subsequent lying slabs. The consequence of these two clear departures from the model of built monumental masonry-walls is that the orthostate layer ends up covering almost the entire middle-zone of the decorated wall, leaving space only for one or two layers of lying slabs-although these constitute, of course, the main part of a built masonry wall. A peristyle wall decorated in the first style in a house later transformed into a shop ('Officina of Ubonius': Pompei, regio IX, insula 3) provides a random example of this effect of raising the orthostate layer (height: approximately 1.30 m!) to eye level by 'lifting' it over an oversized socle of almost one meter height (Figure 15). 30 How significant these peculiarities of Roman first-style walls are appears not only through a comparison with real built masonry walls but also through a comparison with Greek forerunners of this wall decoration system. 31 Indeed, although the orthostates of similar wall decorations from the Greek East also tend to be oversized, we do not find any examples of high socles typical of the Pompeian first style. leaving space only for one or two layers of lying slabs-although these constitute, of course, the main part of a built masonry wall. A peristyle wall decorated in the first style in a house later transformed into a shop ('Officina of Ubonius': Pompei, regio IX, insula 3) provides a random example of this effect of raising the orthostate layer (height: approximately 1.30 m!) to eye level by 'lifting' it over an oversized socle of almost one meter height (Figure 15). 30 How significant these peculiarities of Roman first-style walls are appears not only through a comparison with real built masonry walls but also through a comparison with Greek forerunners of this wall decoration system. 31 Indeed, although the orthostates of similar wall decorations from the Greek East also tend to be oversized, we do not find any examples of high socles typical of the Pompeian first style. The effect of these general characteristics of Roman first-style walls may be expressed in the following way. The wall's middle zone structured in large monochrome surfaces by the orthostates strongly differs from the wall's upper zone, where lying slabs, each typically painted in a different colour, various profiles, and sometimes even plastic columns, 32 result in a much greater ornamental density. The central zone of the wall, which constitutes the direct backdrop for the people lying on 30 On the house's decoration, see PPP III, pp. 428-29; PPM IX.3.2, 128-39 (with bibliography until 1999 on p. 129); for photographs of the wall in its current state of preservation, see . For a comprehensive overview of first-style walls in Pompei in their overall structure, see (Laidlaw 1985), on the 'Officina of Ubonius': pp. 285-89, with a drawing of the wall depicted here on p. 286 and pl. 85. 31 That the Pompeian first style has much older forerunners in the Greek East has long become consensual (Mielsch 2001, p. 22). On the differences between Roman first-style walls and their Greek forerunners (especially concerning the former's high socles), see, e.g., (Ling 1991, p. 13): "The low plinth of the East is replaced by a high plain socle which pushes the orthostates up nearer to the middle of the wall". See also (Barbet 2009, p. 25). A good example for a Greek plastered and painted wall comes from the 'Maison des Comédiens' in Delos from late second century BC (i.e., roughly contemporaneous with Pompeian first style walls): (Bruneau et al. 1970, p. 154, Figure 110), for an overview of the house's architecture, see (Trümper 1998, pp. 202-5, cat. no  The effect of these general characteristics of Roman first-style walls may be expressed in the following way. The wall's middle zone structured in large monochrome surfaces by the orthostates strongly differs from the wall's upper zone, where lying slabs, each typically painted in a different colour, various profiles, and sometimes even plastic columns, 32 result in a much greater ornamental 30 On the house's decoration, see PPP III, pp. 428-29;PPM IX.3.2, 128-39 (with bibliography until 1999 on p. 129); for photographs of the wall in its current state of preservation, see . For a comprehensive overview of first-style walls in Pompei in their overall structure, see (Laidlaw 1985), on the 'Officina of Ubonius': pp. 285-89, with a drawing of the wall depicted here on p. 286 and pl. 85. 31 That the Pompeian first style has much older forerunners in the Greek East has long become consensual (Mielsch 2001, p. 22). On the differences between Roman first-style walls and their Greek forerunners (especially concerning the former's high socles), see, e.g., (Ling 1991, p. 13): "The low plinth of the East is replaced by a high plain socle which pushes the orthostates up nearer to the middle of the wall". See also (Barbet 2009, p. 25). A good example for a Greek plastered and painted wall comes from the 'Maison des Comédiens' in Delos from late second century BC (i.e., roughly contemporaneous with Pompeian first style walls): (Bruneau et al. 1970, p. 154, Figure 110), for an overview of the house's architecture, see (Trümper 1998, pp. 202-5, cat. no. 18, with bibliography). 32 See, e.g., the fauces of the Casa del Fauno: (Mazzoleni and Pappalardo 2005, p. 56, with photograph showing plastic columns in the upper zone).
density. The central zone of the wall, which constitutes the direct backdrop for the people lying on their klinai or sitting on their cathedrae, is thus treated in a similar way to what has been observed for second-style walls with architectural prospects. By its reduced ornamental density, this zone is ceded to the dominus and his guests, while the full range of decorative splendor is displayed only in the wall's upper zone, where there is less danger that this splendor might make the dominus' own appearance fade away. The peculiarities of Roman first-style wall decoration in comparison with real built masonry or with their Greek forerunners give an idea of the general problem that the decoration of representative domestic spaces had to cope with. On the one hand, enriching the wall's decoration with polychromy and complex ornaments, as well as later with various iconographies and spatial illusion, is a suitable way of increasing the representative and decorous effect of the wall. On the other hand, this decorative splendor threatens to overshadow the appearance of those whom the decoration of domestic space is all about: the real actors in the room. The decorated walls should not gain too much aesthetic autonomy, but rather they should keep serving their main goal, which is to constitute a proper frame and stage for the appearance and interaction of socio-hierarchical layered people. We are thus dealing with two potentially conflicting endeavors: to increase the decorative splendor of the wall on the one hand, and not to overshadow the actors on the stage of domestic space on the other hand. To solve this conflict, the consecutive four decorative styles of Roman wall painting found ever new solutions. I would even recognize in this search for better solutions to this conflict an important driving force in the very fast-paced development of decorative styles from the first century BC to the first century AD.
The solution already found in the first style-which consisted in creating a relatively calm backdrop in the wall's middle zone for the appearance of the people in the room by raising large orthostates on a high socle-would prove seminal for all future developments. 33 The frescoes of the Casa dei Grifi on the Palatine marking the beginning of the second style may illustrate a further development of the same basic principles of decoration ( Figure 16). 34 An increase in ornamental density can be observed not only towards the upper margins of the wall (with a row of laying slabs, friezes and the coffered ceiling supported by columns and seen from underneath), but also towards its lateral margins. The colorfully marbled orthostates on the side panels focus, by their symmetrical display, on the center of the wall-but there, the orthostate painted in plain red cedes the stage to the person appearing in front of it. Furthermore, the columns set on the illusionistic protruding socle provide an additional focus on the wall's center by the socle's symmetrical oblique perspectival orientation, without however overloading ornamentally and iconographically this central area of the wall.
To what spectacular effect this new decorative strategy would eventually lead has already been discussed. But even with all the attempts to reduce the aesthetic impact of the architectural prospects towards the wall's middle zone by closing this area with shear walls, hanging textiles or closed doors, in order not to overshadow the appearance of the dominus and his guests, there is an intrinsic limitation to moving continually further in this path. Indeed, the wall decoration should be the frame, and not turn into the main object of aesthetic pleasure. Towards the end of the second style, though, a way was found to further increase the splendor and extravagance of the painted architecture without increasing its visual dominance at the same time: namely, by taking away all tectonic substance from the painted buildings by means of absurdly slender architecture. In the course of this new development, 33 The most fundamental element in which this solution found in the first style shaped all future developments is perhaps the tripartite structure of the wall (high socle-orthostate/middle zone-upper zone), which is basically conserved in all subsequent styles. On this tripartite structure as a leading compositional principle in Roman wall painting of all subsequent styles, see (Ling 1991, p. 15): " [ . . . ] in moving higher on the wall, they [the orthostates] begin to assume more prominence in the design; the way is set for the threefold division of dado (socle), main zone (orthostates) and upper zone which is fundamental to the later styles". See also, e.g., (Mielsch 2001, p. 9). 34 Casa die Grifi, Room II. For splendid colour photographs, see (Mazzoleni and Pappalardo 2005, pp. 65-76). For a short overview on the wall decoration, see, e.g., (Mielsch 2001, pp. 29-32).
columns could eventually be replaced by plant-candelabra. 35 By the completely surreal quality of such painted architecture, these make less direct concurrence to the real world and their actors in the space enclosed by the decorated walls. The famous passage in Vitruvius criticizing this new trend in wall painting shows that this new decorative strategy is most surprising and extravagant not only in the eyes of the modern beholder, but could be felt even as too extravagant in the eyes of some conservative ancient beholders. 36 In the long run, however, this new trend would prove to be the winning strategy, in so far as Roman wall decoration never came back, in its further developments, to the tectonically consistent architecture of the early second style.
development, columns could eventually be replaced by plant-candelabra. 35 By the completely surreal quality of such painted architecture, these make less direct concurrence to the real world and their actors in the space enclosed by the decorated walls. The famous passage in Vitruvius criticizing this new trend in wall painting shows that this new decorative strategy is most surprising and extravagant not only in the eyes of the modern beholder, but could be felt even as too extravagant in the eyes of some conservative ancient beholders. 36 In the long run, however, this new trend would prove to be the winning strategy, in so far as Roman wall decoration never came back, in its further developments, to the tectonically consistent architecture of the early second style. While the painted architecture in the subsequent third style became so exceedingly slender as to blur the difference between three-dimensional architecture and spatial illusion on the one hand, and surface ornamentation on the other hand, large monochrome surfaces gain greater importance on the decorated walls. 37 Sections of monochrome and closed wall surfaces were already an ingredient of earlier styles, as a way of obtaining a relatively neutral ground for the appearance of the dominus and his guests. In the third style, however, monochrome and closed wall surfaces become a very dominant element, and architectonic elements in their new lofty form now fulfil mere framing and structuring functions. But despite August Mau's appraisal of such qualities for the purposes of wall decoration, the early third style with its far-reaching restraint concerning architectural spatial 35 For a general account of this second phase of the second style (usually dated to 40-20 BC), as defined by H. G. Beyen and since then largely accepted (Beyen 1938(Beyen , 1960, see, e.g., (Ling 1991, pp. 31-42;Mielsch 2001, pp. 53-66;Barbet 2009, pp. 40-44). 36 Vitruvius 7.5.3-4. On this much discussed passage, see the analysis in (Grüner 2004, pp. 186-211, 233-63), and the remarks in (Platt 2009, pp. 53-56). The detailed archaeological commentary in (Tybout 1989, pp. 55-107) deals with the passage immediately preceding Vitruvius' outburst against the 'irrational' new trends in wall painting of his own present time, namely an account of what used to be done in the past (and ought still to be done): Vitruvius 7.5.1-2. 37 For a general description of the third style, see, e.g., (Ling 1991  While the painted architecture in the subsequent third style became so exceedingly slender as to blur the difference between three-dimensional architecture and spatial illusion on the one hand, and surface ornamentation on the other hand, large monochrome surfaces gain greater importance on the decorated walls. 37 Sections of monochrome and closed wall surfaces were already an ingredient of earlier styles, as a way of obtaining a relatively neutral ground for the appearance of the dominus and his guests. In the third style, however, monochrome and closed wall surfaces become a very dominant element, and architectonic elements in their new lofty form now fulfil mere framing and structuring functions. But despite August Mau's appraisal of such qualities for the purposes of wall decoration, the early third style with its far-reaching restraint concerning architectural spatial illusion, as in the 35 For a general account of this second phase of the second style (usually dated to 40-20 BC), as defined by H. G. Beyen and since then largely accepted (Beyen 1938(Beyen , 1960, see, e.g., (Ling 1991, pp. 31-42;Mielsch 2001, pp. 53-66;Barbet 2009, pp. 40-44). 36 Vitruvius 7.5.3-4. On this much discussed passage, see the analysis in (Grüner 2004, pp. 186-211, 233-63), and the remarks in (Platt 2009, pp. 53-56). The detailed archaeological commentary in (Tybout 1989, pp. 55-107) deals with the passage immediately preceding Vitruvius' outburst against the 'irrational' new trends in wall painting of his own present time, namely an account of what used to be done in the past (and ought still to be done): Vitruvius 7.5.1-2. 37 For a general description of the third style, see, e.g., (Ling 1991, pp. 52-70;Mielsch 2001, pp. 67-78;Croisille 2005, pp. 68-80;Barbet 2009, pp. 96-178). often-cited textbook example of the Villa of Boscotrecase (Figure 17), 38 eventually turns out to be a rather short episode. Indeed, elements of architectural spatial illusion regain more importance in the later third style in the course of the first half of the first century AD, and seem to be completely rehabilitated by the mid-first century with what is called the fourth style-the final stage of the (pre-79 AD) development of decorative wall painting, by far best represented in Pompei and Herculaneum, but which was almost completely dismissed by A. Mau for its eclecticism in not following any single principle of decoration. 39 illusion, as in the often-cited textbook example of the Villa of Boscotrecase (Figure 17), 38 eventually turns out to be a rather short episode. Indeed, elements of architectural spatial illusion regain more importance in the later third style in the course of the first half of the first century AD, and seem to be completely rehabilitated by the mid-first century with what is called the fourth style-the final stage of the (pre-79 AD) development of decorative wall painting, by far best represented in Pompei and Herculaneum, but which was almost completely dismissed by A. Mau for its eclecticism in not following any single principle of decoration. 39 Figure 17. Cubiculum 16 of the Villa of Boscotrecase with its early third-style wall decoration, as reassembled in Naples, Museo Archeologico Nazionale (inv. 147501); © photograph by author.
These architectural elements that massively re-entered the repertoire of Roman wall painting by the mid-first century AD are now assigned well-defined positions within the wall decoration schemes. Illusionistic architecture, with its effects of spatial depth and its oblique perspectival lines orienting the gaze towards the wall's center, are found almost exclusively in the upper zone of the wall and on pilaster-like vertical stripes structuring the wall's middle zone in separated panels. Room f of the House of Loreius Tiburtinus in Pompey (Regio II, insula 2), which is decorated in the fourth style, may serve here as a random example (Figure 18). 40 The resulting decoration scheme combines 38 On the Villa of Boscotrecase and its wall painting, see (Della Corte 1922;Blanckenhagen and Alexander 1990;Anderson 1987Anderson -1988. For short overviews, see, e.g., (Mielsch 2001, pp. 70-73;Croisille 2005, pp. 77-80;Barbet 2009, pp. 109-10). See also (Dietrich 2017). 39 See (Mau 1882, p. 451): "Den zweiten und dritten Stil mögen wir als zwei aus ganz verschiedenen Geschmacksrichtungen hervorgegangene Decorationssyteme betrachten, welche, jedes in seiner Art hoch entwickelt, sich gleichberechtigt gegenüberstehen; die letzten pompejanischen Malereien sind der Beginn des Verfalls". For a (less opinionated…) general description of fourth style, see, e.g., (Ling 1991, pp. 71-100;Mielsch 2001, pp. 79-92;Croisille 2005, pp. 81-102;Barbet 2009, pp. 179-273 ;Platt 2002;Lorenz 2008, pp. 538-40, cat. no. K12;Knox 2015). We have to assume that a statuette standing in the wall's central niche took the place of the mythological picture which we would otherwise have expected here. Those scholars who interpret this room as an Isis sacellum tend to reconstruct a statuette of Isis, see, e.g., (Tran Tam Tinh 1964, pp. 43-45). See (Platt 2002, pp. 104-7), for comments on the statuette's three-dimensionality in relation to the house's two-dimensional painted images. In any case, an These architectural elements that massively re-entered the repertoire of Roman wall painting by the mid-first century AD are now assigned well-defined positions within the wall decoration schemes. Illusionistic architecture, with its effects of spatial depth and its oblique perspectival lines orienting the gaze towards the wall's center, are found almost exclusively in the upper zone of the wall and on pilaster-like vertical stripes structuring the wall's middle zone in separated panels. Room f of the House of Loreius Tiburtinus in Pompey (Regio II, insula 2), which is decorated in the fourth style, may serve here as a random example (Figure 18). 40 The resulting decoration scheme combines the simple 38 On the Villa of Boscotrecase and its wall painting, see (Della Corte 1922;Von Blanckenhagen and Alexander 1990;Anderson 1987Anderson -1988. For short overviews, see, e.g., (Mielsch 2001, pp. 70-73;Croisille 2005, pp. 77-80;Barbet 2009, pp. 109-10). See also (Dietrich 2017). 39 See (Mau 1882, p. 451): "Den zweiten und dritten Stil mögen wir als zwei aus ganz verschiedenen Geschmacksrichtungen hervorgegangene Decorationssyteme betrachten, welche, jedes in seiner Art hoch entwickelt, sich gleichberechtigt gegenüberstehen; die letzten pompejanischen Malereien sind der Beginn des Verfalls". For a (less opinionated . . . ) general description of fourth style, see, e.g., (Ling 1991, pp. 71-100;Mielsch 2001, pp. 79-92;Croisille 2005, pp. 81-102;Barbet 2009, pp. 179-273 Platt 2002;Lorenz 2008, pp. 538-40, cat. no. K12;Knox 2015). We have to assume that a statuette standing in the wall's central niche took the place of the mythological picture which we would otherwise have expected here. Those scholars who interpret this room as an Isis sacellum tend to reconstruct a statuette of Isis, see, e.g., (Tran Tam Tinh 1964, pp. 43-45). See (Platt 2002, pp. 104-7), for comments on the statuette's three-dimensionality in relation to the house's two-dimensional monochrome surface as an ideal backdrop for the appearance of the actors in the real space enclosed by the walls, the breath-taking perspectival architectural illusion with its oblique lines leading from any marginal spot of the wall to its center, and its decorous effect, which gives to the decoration the much desired panegyric tone. In contrast with second-style architectural vistas, the perspectival architecture of the fourth style have the tendency not to lead into distant space, but to project forward into the enclosed space, 41 so as to keep the viewer's attention within the real space enclosed by the decorated walls and their inhabitants acting in front of these walls, instead of letting the attention get lost in the depths of illusionary space. The mythological pictures that now regularly decorate the central panel of a wall integrate much less landscape and spatial depth than in the former second and third styles, in line with the closed surface of the panels on which they appear. But these mythological pictures on fourth-style walls prevent the viewer from escaping too far into distant mythical worlds by another means too. In a great number of cases, they show couples of hero and heroine with little narrative content. But for this very reason, these pictures of couples in heroic costume invite all the more the parallel and the dialogue with the staged reality in front of the walls, a staged reality presided by the couple of the dominus and his wife. 42 In this final section, I tried to expand the argument that I developed through an analysis of second-style walls in the two previous sections on the preceding and following styles of Roman wall decoration. In this brief outline of how the structures observed on second-style walls might help understanding characteristics and peculiarities of first, third and fourth style walls, I did not try to give a full account of the complexities of those decorative styles. Many topics would have deserved a much more detailed analysis that might sometimes have further detailed my argument. In other cases, such an in-depth analysis might also cause problems to arise that I bypass here for the sake of clarity and might thereby weaken my argument. 43 Further research would thus certainly result in a more nuanced picture of the general development outlined in this last section. My overriding point, however, would stay the same. The developments observed between the first and fourth styles concerning the closing-up and opening-up of the wall's surface do not reflect directly an ever-changing attitude towards perspectival painting or surface decoration, within a most fast-paced (and quite chaotic) history of perspective and decorative taste. They rather reflect the ever-changing solutions found to the same general problem of two potentially conflicting endeavors, which the decorators of Roman private space had to reconcile with each other: to increase the decorative splendor of the wall on the one hand, and not to overshadow the actors on the stage of domestic space on the other hand. painted images. In any case, an interpretation of room f as a sacellum does not preclude its use for more intimate social gatherings too, see (Zanker 1995, pp. 155-56, comparing this sacellum with the Amaltheion in a villa of Cicero's friend Atticus, with more similar examples in note 52). 41 This important remark has been made in (Lorenz 2013, p. 369). 42 For the complex relations between the mythological couples on the walls and their real counterparts in the couple of the dominus and his wife, see the detailed and brilliant analysis by Lorenz (2008). 43 What to do, for example, with those large scale mythological or sacro-idyllic landscape typical of the transitional phase between second and third style, as we find them in the Houses of Augustus and of Livia on the Palatine? Being well centred both horizontally and vertically on the walls, and being large in size, they seem to concur strongly with the dominus' own appearance. I currently do not have a good answer to this feature that does not fit well into the general model outlined here.