Patterns of Reconstruction at Pompeii

The preceding chapter identified several areas which provide evidence of chronology of construction with respect to the 62 earthquake. The following discussion seeks to bring the various observations together to develop a more encompassing description of the building's varying states over time.

The centerpiece of the discussion is a three dimensional computer model which proposes an approximate state of the masonry walls immediately following the earthquake. The discussion begins with a brief description of the model and its development, and then moves to a discussion of the various building areas.

7.1 The 62+ model

Understanding the nature of the post-earthquake reconstruction of the Macellum requires understanding the damage which the earthquake induced. To depict this damage and understand its patterns, a three dimensional model has been developed, called "62+" because it describes a possible state of the walls immediately following the earthquake. Figure 7.1 below shows how parts of the geometry have been derived from the photogrammetric models.

Figure 7.1: Area A and the northwest corner. The profile of the damage at Area A is derived from the photogrammetric model. Other areas are based on supposition consistent with the patterns. The image is linked to a brief video clip dissolving between the photogrammetric and derived models.

Figures 7.2 and 7.3 provide overviews of the model, viewed from the south and northwest respectively. These figures are linked to video clips which dissolve from a simplified model of the walls immediately preceding the 62 earthquake (called the 62- model), to the model immediately afterward.

Figure 7.2: The 62+ model viewed from the northwest. The image is linked to a video clip which dissolves from the 62- model to the 62+ model.

Figure 7.3: The 62+ model viewed from the northwest. The image is linked to a video clip which dissolves from the 62- model to the 62+ model.

The profile of damage at Areas A through E are based on the photogrammetric models. Damage profiles of walls adjoining these areas area estimated based on patterns observed in other walls at Pompeii. The damage pattern of the western wall is based on the assumption that there were no western shops prior to 62; this assumption is discussed in more detail below, following a discussion of each of the five damage areas.

Note that the 62+ model does not show floor framing in the north shops, nor does it show any roof framing; this is because there is not sufficient evidence to develop a reasonable reconstruction at all locations. Similarly, the model shows many of the walls, such as the south shop walls, extending to their full six meter height. It is likely that the upper portions of these walls sustained damage in the 62 earthquake, but since there is not sufficient evidence to speculate about what that damage might have been, the model simply shows the complete wall.

7.2 Areas that sustained limited damage

Figures 7.2 and 7.3 show that the east wall and western portion of the south wall sustain relatively little damage. Concerning the south wall, as discussed in chapter 5, the floor framing of the south shops helps to brace the shop walls since the floors do not have openings for stairs, making the shop walls effective buttresses for the south perimeter wall.

On the eastern end of the building, the configuration of the shrine building gives it good seismic resistance; the walls are thick to accommodate sculpture niches, and are well buttressed by intersecting walls. The other portions of the eastern wall were buttressed by intersecting walls on the exterior face. The portion of the wall north of the shrine, had intersecting walls and floor framing buttressing it, as shown in figure 7.4 below.

Figure 7.4: The northern end of the east perimeter wall, viewed from the exterior. Note the joist pockets and bond scars indicating the locations of walls and floor framing that intersected with the wall, providing buttressing against out-of-plane failure.

Considering the southern end of the wall, figure 7.5 shows the exterior face extending from the south-east corner to the shrine; the figure highlights a bond scar which is also shown in detail.

Figure 7.5: The southern end of the east perimeter wall, viewed from the exterior. The upper photo shows an overall view, with a bond scar highlighted; a detail of the bond scar is shown in the lower photo.

The bond scar corresponds to an intersecting wall dividing the approximately 14 meter length of the wall stretching from the corner to the shrine into two structural panels: a roughly 9 meter panel to the south and a roughly 5 meter panel to the north. It is possible that these panels sustained damage in 62 in the upper portions of the ancient fabric which are no longer preserved. In any case, the bond scar provides an explanation for the lack of the deep scooping failure found in the other areas.

7.3 Area A

As discussed in chapter 6, the repair at Area A is probably modern repair of seismic damage associated with the eruption of 79 AD. This conclusion does not preclude the possibility that the area was also damaged in 62. As discussed in chapter 6, the evidence suggests that shaking in 62 was more severe than that of 79, so a wall that failed in 79 would also have been prone to failure in 62. The 62+ model assumes that the perimeter wall at Area A failed along the currently visible profile of failure in 62, and then failed again along the same line in 79, assuming the seam of the repair following the 62 quake created a line of weakness enabling the failure of 79. figure 7.6 shows the model at Area A, viewed from the north side.

Figure 7.6: Area A in the 62+ model.

The failure in 62 was probably the progressive mode of collapse described in chapter 5, resulting from the stair openings at the back of the shops. This mode of failure explains why the scooping profile is so close to the ground at Area A plus the near complete failure of W7.8 flanking the north gate. This mode of failure was probably different in 79, however, since much of the ancient fabric of W7.8 was preserved. One of the key questions is why Area A and W7.8 both failed in 62, but in 79 Area A failed while W7.8 did not, at least not in the currently visible portions.

One potential explanation for this difference in behavior is that pumice fall in 79 collapsed the floor framing in the shop 7, and possibly other shops. The loss of the framing in shop 7 would have prevented the battering ram action that apparently collapsed W7.8 in 62, while remaining framing in shops 3 through 6, which have much shorter spans, remained to induce the progressive collapse of the shop walls west of W7.8, leading to the collapse of the perimeter wall.

7.4 Area B

Relative to Area A, the failure region at Area B is much higher above the ground, as shown in figure 7.7 below.

Figure 7.7: Area B in the 62+ model.

The higher profile of the failure region implies that the shop walls adjoining area B did not experience the progressive collapse mechanism suffered by those adjoining area A. As discussed in section 6.4, the key difference at area B is the presence of internal walls perpendicular to the shop walls. These internal walls would have been able to resist the inertia load of the floor framing, even if there had been stair openings separating the floor framing from the perimeter wall¹. The internal walls also serve to buttress the shop walls, allowing them to maintain a greater height after damage.

Unlike the other shop walls adjoining Area B, W8.9 flanking the gate clearly sustained significant damage. This wall is very similar in its repair to W7.8 on the other side of the gate, with a rough seam at the juncture with the perimeter wall, indicating that nearly the entire wall had been replaced; the large plaster patches on the wall indicating that it is ancient repair. Such severe damage to a wall is expected for the progressive collapse mechanism, but the conditions are Area B indicate that this mechanism did not occur there. A possible explanation is that the progressive collapse mechanism at Area A sent debris across the gate passage where it knocked down W8.9; this explains why the failure of W8.9 is more similar to the shop walls at Area A than to the other shop walls at Area B.

The plaster on the south face of area B indicates that the currently visible damage area was repaired in ancient times, but was not damaged in 79, providing another indication that the seismic shaking in 79 was not as severe at that in 62.

7.5 Area C

Area C is the only failure region not near a gate, and this appears to explain its relatively symmetric profile in contrast to the other areas where the profile is lower at the end near the gate. The low profile in these areas probably resulted from the failure of the shop walls flanking the gate, resulting in a lower buttress at that end. At Area C, the buttressing walls at both ends are part of the perimeter wall, as shown in figure 7.8 below.

Figure 7.8: Area C in the 62+ model

It is difficult to discern whether the repair material visible today at Area C is ancient or modern. If this area did fail due to shaking in 79, then it probably also failed in 62, because of the indications that shaking in 62 was stronger. If it did fail in 62, it is possible that the progressive collapse mechanism ocurred, since the arrangement of beam pockets indicates that there were stair openings at the back of shops 17 and 18, however this mechanism clearly did not occur in 79, since W17.18 is preserved up to the level of the pockets.

Another uncertainty at Area C involves the horizontal seam in the repair area itself, clearly visible in the south face shown in figure 7.9 below. Areas A and E show similar horizontal seams.

Figure 7.9: The south face of Area C, note the horizontal seam in the middle of the repair material.

Figure 7.10 below shows a portion of the north face at shop 17, with the seam of the repair and the horizontal seam within the repair highlighted for clarity.

The horizontal seams in Areas A, D, and E can be explained as a natural result of masonry construction, where the horizontal strata reflect stages of construction, where the wall is repaired to a horizontal level which serves as a starting point for the next level. At Area D, however, there is a curious characteristic concerning the bond scar of W16.17, which changes character at the seam, as shown in figure 7.11 (a detail from figure 7.10.)

Figure 7.11: Detail of figure 7.10, showing the change in character of the bond scar at the horizontal seam.

Below the horizontal seam, the scar shows many tightly-packed, fist-sized stones which recess slightly into the wall. Above the seam, the scar shows only two, much larger stones which protrude from the wall. The significant difference in character suggests that the seam may mean the seam marks a significant chronological difference, rather than stages in the same construction operation. If so, then the question arises whether the two parts of the repair are both ancient, both modern, or one is ancient and one is modern.

Given the conclusion that the other four areas were damaged in 62, it is also very likely that Area C was damaged in 62 as well. A scenario where Area C survives the 62 quake, but is damaged in 79 is exactly opposite what the evidence indicates for Areas B, D, and E, which were damaged in 62 but largely survived events in 79. The only way that Area C would counter that trend is if there were some change in the wall's condition which made it relatively stronger in 62 than in 79. It's not clear what such a change in condition would be, and the study has found no evidence suggestion what such a change in condition would be, so this possibility is considered highly unlikely. The conclusion is that Area C was damaged in 62, it may have been damaged again in 79, as Area A was. The multiple damage events may help explain the apparent multi-stage repair process.

7.6 Area D

As discussed in chapter 6, the bonding of W49.50 with the repair at Area D strongly indicates that this repair is ancient. Recall that the seam of the repair passes through the beam pocket in W49.50, as shown in figure 7.12 below:

Figure 7.12: The juncture of W49.50 with the south wall, with the repair area noted

Figure 7.13 shows Area D in the 62+ model, including the second floor framing and gallery for clarity.

Figure 7.13: Area D in the 62+ model, viewed from the southeast.

The key at Area D is that the configuration of the second floor framing allows the perimeter wall, shop walls and floor framing to act together as an effective seismic resisting system. Damage is limited to the area near the gate because the shop adjacent the gate lacked floor framing, which is evident from the lack of beam pockets in W50.51, as discussed in section 5.5.1.

In addition to the damage shown in the 62+ model, there may well have been additional failure of the framing for the second floor and gallery framing, as well as additional damage to W49.50, but it is difficult to ascertain the extent of this damage.

7.7 Area E

The faliure of Area E is similar to that of Area D in that it was initiated by the wing wall flanking the gate, however since Area E lacks floor framing, the damage to the perimeter wall was much more extensive. The failure of the wing wall gave the perimeter a much lower buttress height at the west end, resulting in the asymmetric profile shown in figure 7.14 below.

Figure 7.14: Area E in the 62+ model, viewed from the southeast. The failure of the wing wall adjacent the gate resulted in the asymmetric failure of the perimeter wall.

Areas D and E make an important point about the earthquake reconstruction of the building. It appears that the reticulatum facade on the south wall never had a plaster coating; patches of plaster can be found on the reticulatum faces facing the gate, but there is not a trace of plaster to be found on the south faced. This combined with the conclusion that the repair at Areas D and E is ancient leads to the conclusion that the wall appeared in ancient times much as they appear today. They were certainly taller, but also clearly showed the scars of seismic repair. In particular, the repair did not attempt to restore the reticulatum facade to its original condition. This approach indicates that the repair placed functionality over architectural appearance.

7.8 The Western Wall

As discussed in section 6.5, the construction history of the western wall involves a great deal of uncertainty. This study makes the assumption that the western shops visible today are a post-62 addition, and that there were no shops prior to the earthquake. This assumption means that the wall was long and straight, with little buttressing, making it highly vulnerable to seismic damage. Therefore, the wall would have been severely damaged by the 62 earthquake, as shown in the 62+ model below.

Figure 7.15: The western wall in the 62+ model, viewed from the northwest.

This near total damage of the wall led to the reconstruction sequence described in section 6.5, where the inner wall was reconstructed with the priority of restoring the function of the market, and the western shops were added later as part of the enhanced architectural engagement with the Forum.

Supporting, though not conclusive, evidence for the extensive destruction of the wall in 62 is provided by the frescos in the northwest corner. These frescos are in the fourth style, dating them at some time after 50 AD, which is consistent with an extensive reconstruction of the wall in 62.

7.9 Conclusions

Post-62 earthquake reconstruction scenarios for the Macellum can be cast between one of extreme of minimal reconstruction, incomplete at the time of the 79 eruption, to another extreme of a reconstruction that not only restored the building to its original function, but also enhanced its architectural presence on the Forum. The conclusions of this study indicate that there were elements of both these extremes.

The reconstruction of Areas D and E reflect a utilitarian response, leaving visibly prominent scars in the strongly patterned reticulatum of the south facade. Like areas D and E, the ancient repair at Area B also includes large blocks and a less refined construction technique than the surrounding fabric, again reflecting a priority of restoring building function over construction quality. Given the Macellum's function as a food market, the scenario of a rapid patching of the building to restore function is reasonable. Food distribution is typically a significant problem following any urban earthquake, and it's likely that there was significant pressure to restore the building to operation as quickly as possible.

The architectural enhancement of the building, probably happened after it was restored to operation, with the addition of the western shops. The double wall construction of the shops imply that the shops were built as a separate project from the main building. Architecturally, the tapering plan of these shops accommodates the difference in angle between the axis of the Macellum and the axis of the Forum, with the effect of making the Macellum's facade more squarely engage the Forum. It's credible that these shops were built as part of a larger renovation of the Forum, performed well after the Macellum had been be restored to operation by the rebuilding of the inner wall.

There very likely has been post-excavation damage and rebuilding of the Macellum, as reveled by comparison with the model at the Naples museum and examination of historic earthquakes. The evidence of such damage and reconstruction is most evident at the Macellum's north wall and northwest corner. Corroborating evidence from other archival sources is needed to make draw these conclusions more firmly.

In closing, the story of the Macellum's reconstruction appears to be a common one for a vital building following in an urban earthquake. There was a pressing priority to restore operation as soon as possible, reflected in the rough ancient patching of Areas B, D, and E. Once basic human services were restored, there was then a longer term priority to take advantage of the opportunity the rebuild bigger and better. This opportunity appears to have been taken in the western shops, as an independent addition to the building, plugging it into a larger renovation and rebuilding of the Forum.

7. Reconstruction Scenarios