CI

Objectives and excavation history

The small trench located in the north end of the presbytery of the church (Fig. 1) was, according to the records of the superintendency, a clandestine excavation of December 1998.¹ The trench, measuring approximately 2.0 x 1.5 m (dimensions irregular), was cut through the terracotta floor and arrived at the cosmatesque pavement some 35 cm below it. In order to recover the stratigraphy, the trench was reopened and its original limits were cleaned and extended to make it a regular 2.0 x 2.0 m square. The corners of the trench were excavated stratigraphically to the level of the cosmatesque pavement, preserving materials and documenting the levels.

Stratigraphy

Excavation of CI, in the northeast corner of the nave of S. Pietro, revealed 3017, an intact section of flat polychrome geometric ornament more commonly referred to as Cosmati or cosmatesque pavement in reference to the most active workshop that produced this type of flooring from the late eleventh to the fourteenth century (Figs. 2–3).² The poorly preserved pavement fragment is composed of two distinct checkerboard patterns, which are partially enclosed by thick slabs of white marble. Both motifs are dominated by white marble squares of various sizes. The interstices are filled with little red, green, and yellow marble triangles.³ The patterns resemble those found in a group of pavements dated by Dorothy F. Glass to the twelfth-century.⁴ They are, however, not a perfect match and could date to the second quarter of the thirteenth century when the Cosmati family was actively working on commissions in the area.⁵

There are indications of an attempt to consolidate the cosmatesque floor prior to the church’s repaving. The consolidation (3066) involved framing the preserved marble elements with tiles and cement. This was covered over eventually by a thick preparation of rubble and earth (3016), which includes marble slabs and a piece of early medieval liturgical sculpture with characteristic Kerbschnitten carving and interlace motif (subsequently stolen when the church was vandalised). The shallow depth of the cosmatesque floor suggests that it was raised and re-laid at a later date.⁶ (3016) is the bedding for a smooth, level surface of grey plaster 3015 mixed with ash, chalk, and gravel. This floor was later covered by a light brown sandy mortar 3–5 cm, thick mixed with brick fragments (3014), which acted as a preparation of the terracotta floor 3003. 3003 was made of bricks each measuring 26.50 x 13.50 cm, bonded with a hard, white mortar. The bricks were laid in a diagonal pattern, and framed at the edges of the walls [3012] and [3013], the north and east corner of the presbytery, respectively, with a row of straight bricks. Comparable floors have been identified in the church of S. Andrea and in the crypt of S. Vito, both in in Anagni, according to De Meo.⁷ The modern clandestine excavation 3001 cut this floor surface and the trench was backfilled with bricks and earth (3002).

C II

Objectives and excavation history

Trench CII, referred to in earlier reports as the ‘chapel,’ is a walled rectangular space situated in the northwest corner of the monastery’s church (i.e. the large apsidal structure that dominates the MON area; see Fig. 1). The external northern [3007] (and the earlier [3191] and [3190]) and western [3006] (and the earlier [3195] and [3193]) walls of the building defined the corresponding limits of the trench. The southern and eastern limits were, in turn, determined by [3004]/[3005] and [3008], interior walls added relatively late in the building’s history to delimit the space. The arrangement and central position of a number of the later graves supported our initial hypothesis that the room was built as a private funerary chapel. Contemporary graves uncovered on the ‘chapel’s’ periphery and evidence of their continuation beyond the foundation of the southern wall make it clear, however, that the graves pre-date the room. The intended function of this space remains unclear. The maximum limits of the trench were 4.45 x 3.03 m (dimensions slightly irregular). CII was excavated to the level of (3137), a thick layer of sticky, indurate, dark brown clay with characteristic orange-pink veining. This clay was integral to the construction of the brick building, which was part of the earliest phase uncovered.⁸

Trench CII was excavated during the 2006 and most of the 2007 season at which point it was hastily abandoned at signs of the building’s structural instability. Excavation of the space was taken up again briefly in 2010, at which time an effort was made to remove the few remaining contexts over (3137). Due to constraints of time, machinery and manpower, certain larger contexts such as [3029] and [3068] were cleaned and recorded, but left in situ. The trench was backfilled at the end of the 2010 season.⁹ In 2008, Francesca Candilio conducted a full osteological analysis on all articulated human remains.

Stratigraphy

The brick building

Trench CII provides us with a view of the interior of the free-standing, apsidal opus testaceum building. A maximum of nine courses of the northern [3190] and western [3192] exterior walls were visible. The joints are wide and the bricks vary slightly in width and length. The offsets [3189] (north) and [3188]=[3163] (west) are level and narrow (6–14 cm wide) and are made of a hard, characteristic purple-blue mortar with chunky red (brick and terracotta) and white (chalk) inclusions. A sondage opened outside the building along the northernmost corner of the façade in BI revealed the foundation of [3188].¹⁰ Here, the foundation trench 3976 for the western wall was clearly visible and excavated, yielding large sherds that can be dated to the fourth or fifth century. No signs of a corresponding trench were encountered within CII, where (3137), the construction clay mentioned above (and discussed in greater detail below), was packed tightly against the offset.¹¹ The northern wall [3190] is pierced by a square opening measuring approximately 15 x 15 cm, and located 6 cm above its offset. The reason for such an aperture is conjectural, but it may be related to the circulation of air in this area, perhaps under an elevated wooden floor.¹² That said, there is no sign that the clay was ever covered by any type of flooring. Indeed, there is a curious lack of evidence that the interior of the building was decorated at all: there are no revetments in the walls, nor was a significant quantity of plaster uncovered.

The foundation clay (3137) is present throughout CII; its color and texture are not, however, uniform.¹³ In the eastern half of the trench it is more yellow-brown and, when disturbed, it breaks into near perfect spheres, strikingly uniform in size (approximately 1.5–2.5 cm in diameter). The balls are extremely hard; their cross-section reveals a clay core encircled by a thin iron (or perhaps hematite) shell of a dark reddish-brown hue. Given what seemed to be clear differences of color and compaction, this clay was initially assigned its own context number, (3173). Microscopic examination of (3173) and comparison with (3137) revealed, however, the superficial nature of the differences between the two contexts. Magnified, their composition and color was nearly identical. The marked difference on the macro level must have arisen from the clay being worked or exposed in diverse ways. Before this was understood, an effort was made to define and separate (3173) from (3137), a task that proved impossible and was abandoned quickly. Both contexts were left in situ and are considered equal in the record.

A number of contexts were cut into or built directly onto the foundation clay (3137)=(3173). Establishing even a relative chronology of these actions has proven difficult since they were, for the most part, not contiguous and their excavation yielded little diagnostic material. Despite these uncertainties, the contexts have been roughly divided into two phases below: the first associated with the brick building, the second with the later opus vittatum building and narthex.¹⁴

Among the earliest features in the brick building (possibly contemporary with the building itself) is a channel for a Roman fistula [3178] embedded in concrete near identical to that of the offsets (Fig. 4). A deep L-shaped cut 3187 was made into the foundation clay (3137)=(3173) for an a cappucina covered channel comprised of terracotta tiles (maximum height of channel: 24 cm, width at base: 31 cm). Presumably, the channel once housed a fistula, which was either removed or robbed at some point. A fine, silty fill (3177) was emptied from the channel. A few slightly larger components were found near the rear of this fill, none of sufficient size to be of diagnostic value, which was excavated to a depth of 40 cm (as far as could be reached with a trowel), but which continued at least 99 cm beyond this point.¹⁵ The channel is preserved only beneath [3029], a massive rectangular block of white marble (0.95 x 0.60 x 0.28 cm; discussed below). There is no trace of the conduit’s continuation beyond the eastern limit of [3029] or on the other side of cut 3135 made for the bell-casting pit (discussed below), which truncated the fistula on its west side. (The vertical section created by 3135 enabled [3178] and its associated contexts to be visible without removing the large overlying block, [3029].) The channel runs at a slight angle (southeast–northwest) with respect to the walls of the building. Isolated as it is, the function of the fistula might have served is difficult to discern. Its relationship to the building cannot be determined with any certainty, since it does not intersect with the foundation or walls.

The cover tiles of the fistula were broken at their peak by 3194, a cut made for the insertion of a cement foundation that incorporates two thick slabs of fine white marble, highly finished on their broad surfaces and laid flat, one atop of the other [3186].¹⁶ The lower slab is set into and partially covered by cement the color and composition of which is very like the mortar used for the foundation of the north and west walls of the brick building ([3189] (north) and [3188]=[3090]=[3163] (west)) suggesting that the two were part of the same construction campaign. As noted above, the cement does not extend over the entire surface of the lower slab; its limits may have been determined by the original dimensions of the upper slab (now broken), which it held in place. It is possible that this upper slab, which is not covered by cement served as part of a pavement or as a plinth. In an effort to rationalize the presence of these slabs their heights were compared with other elements in the trench. The top of the lower slab aligns both vertically and horizontally with a column base [3068]—specifically, with the upper limit of [3068]‘s squared plinth—situated against the western wall of the building in the southwest corner of CII, but this feature seems to belong to a later phase and is discussed below.

The style of the brick building’s masonry and the sherds of pottery uncovered in the foundation trench suggest a fourth- or fifth-century date for the structure. The absence of signs that the building’s interior was ever decorated imply that the space may have served a practical function, possibly as a winery. Palladius in his Opus agriculturae describes a basilica-style building with an apse cella vinaria.¹⁷ The construction of such a space at Villa Magna would have been in keeping with the estate’s historical and continuing identification with viticulture, a use which may also explain the presence of the fistula though its exact function even in such a context remains unclear. The marble slab, almost certainly part of the original structure, likely served as a plinth for a colonnade, which was probably mirrored on the southern side, dividing the space into three large aisles.

As stated above, none of elements described above can be tied conclusively to the brick building or the later opus vittatum church. It is possible that a few of the contexts discussed below were, in fact, related to this earlier structure and merely preserved in the later building.

The opus vitattum church and narthex

Some time after their construction in the fourth or fifth century, the walls of the brick building were uniformly razed to a height of about 0.5 m and rebuilt in masonry of reused bricks, terracotta tiles and tufa blocks in the opus vittatum style (Fig. 5). Though not entirely regular, the basic pattern adopted was two rows of roughly rectangular tufa blocks alternating with a thin red-orange band of terracotta tiles and bricks. Cream-colored mortar with flecks of dark grey bound the layers (this mortar is a near match to the mortar covering the column base). This moment of reconstruction is not well preserved in the northwest corner of the church, which was largely rebuilt ([2047]=[2046]=[3193]=[3195], discussed below) in the modern period; the little that remains of the opus vitattum walls remained obscured by a thick layer of modern plaster. In the southwest corner of the trench a single course of opus vitattum is discernible just above the brick wall. This is the lower portion of [2057], a section of wall just north of the central portal of the western façade. The door is one of three that once pierced the façade. The contour of the northern door is partially preserved in the wall. From the interior, two of five large limestone blocks can be seen just north of [2057]. From outside the building, all five of blocks [3950] are visible. These formed the threshold of a door 1.45 m wide.

[3068], a chipped white marble Attic column base, sits on a thin bed of tawny-colored mortar (3044)=(3054)=(3067) (Fig. 6). The mortar is very pale with dark grey iron specks; its texture is powdery/sandy. Rather confoundingly, this same mortar also covered the base and bonded to the western wall [3006]. Though the base was left in situ at the end of the 2010 season, much of the mortar that surrounded (and obscured it) was removed with pick and trowel. Three corners (possibly all four, but only three are visible) of the plinth block (50 x 50 cm) were broken, leaving semi-circular voids that slightly undercut the base molding. The edges of the plinth are not aligned with the building, but are at a very slight angle (northwest–southeast). The column base and its thin mortar bed sit on the foundation clay (3137) and the offset [3188]=[3099]=[3163] of the western wall of the brick building.

The column base molding has a triple torus.¹⁸ The top torus suffered numerous breaks and chips (vaguely aligned with the broken corners of the plinth). The total height of the base is 25 cm. The southern wall [3004] (not firmly dated; discussed below) incorporated the (later robbed-out) column, column base and mortar into its construction; it preserves the impression of the lost column’s profile. From it we know that the base supported a column with maximum diameter of around 40.5 cm that tapered as it rose to a height of approximately 3.2 m. This is an ancient column base (of uncertain date, likely prior to third century), which was reused and repositioned at some point into its present situation. It may be that the brick building had an internal pair of columns or—if we interpret the upper slab of [3186] as part of a plinth—a colonnade. The cream-colored mortar, however, differs significantly from the pozzolanic cement that bonds to the other elements in the room (like the offsets) that are clearly associated with the brick building. The shift in materials suggests that this was a different moment of construction. The light and powdery mortar resembles instead that which was used in the tombs and later structures discussed below.

Packed against the column base [3068] was a hard yellow-brown clay (3162), not unlike (3137)=(3173) (the so-called foundation clay discussed above) in color and composition, but markedly different in texture. As with (3173)=(3137) it seems that this textural difference arose from the clay being worked in a particular manner. In this case it appears that (3162) was the clay displaced in the digging of 3170, a large slightly ovoid hole (diameter: 38 cm, depth: 33 cm) with steep vertical walls cut into (3137). The depth of the hole was increased by packing clay (3162) along its rim. 3170 was later filled by (3149), a large quantity of charred material and sherds of Roman pottery. Adjacent to 3170 were two small, shallow ancillary holes. One of these 3182/(3181) had a funnel-like shape and was joined to 3170 by a short canal. The other 3184/(3183) was (roughly) kidney shaped. The high density of burnt material within 3170 suggests that this was used for firing or cooking. (3149) included a wine amphora from southern Italy of a type that circulated from the late-fourth through the seventh centuries. It’s presence does not offer much insight as to the dating of features in this area of the trench since it may be residual.

A long, narrow trench, 3160, extending approximately .8 m along the west wall and bottoming on the wall’s offset, was cut into the worked clay surrounding 3170 and the foundation clay below. The trench’s fill (3161) comprised dark brown clay/earth and a considerable amount of construction/destruction debris including large fragments of worked tufa and white marble, terracotta tiles, mosaic tesserae, Roman pottery and traces of a tawny, straw-colored mortar. A little further west along the wall was 3176, a roughly semi-circular shaped cut filled by (3175). Cut and fill closely resemble 3160/(3161) and are considered equal in the record. They were, however, excavated as separate entities since elements of (3175) were loosely bonded to the wall and seemed to be distinct from the rest of the fill. The color, texture, and composition of the mortar matched that associated with the column base. It is thought that some structure (its precise nature unknown, but likely part of the same campaign that placed the column base in its current position) was partially, but not entirely, demolished during the digging of trench 3160=3176. A small section of this feature remained bonded to the brick wall.

Immediately adjacent to 3176 was another trench cut into the foundation clay along the offset of the north wall [3191]. Later, probably at some point during the twelfth century, the trench was severed by a cut made for a tomb (T241, see description below). The trench now exists in two non-continuous stretches that are equal in the record: 3172/(3171) in the northwest corner of the church and 3158/(3159), further east and extending beyond the eastern limits of CII. This east-west running trench is thought to be akin to the north-south running trench described above. These trenches, too shallow to be for construction, seem to have been exploratory in nature, dug in order to verify the stability of the brick walls and their foundation in anticipation of the building’s reconstruction. Their fills contained sixth and seventh-century ceramics, which support typological arguments for the date for the opus vitattum building.¹⁹ Fill (3171) contained fragments of brick, mortar, painted plaster and pieces of opus sectile marble. Suggesting that decoration was added at some point to brick building’s interior.

A fragment of white mortar pavement 3144 that covered (3137)=(3173) and (3159) and bonded to [3191] may be the only remaining section of a floor of the opus viattum building. The original limits of this context were diminished by a series of cuts that define its (irregular) western, southern and eastern edges: 3103, 3148 and 3169. The reason for the latter two cuts is unclear; 3103 is the cut for a constructed tomb (discussed below). No other fragments of this pavement were uncovered. A few pieces of colored opus sectile marble (AE1045, AE347, AE334, and AE335) were found in the fill of the aforementioned early tombs (discussed below).

Though sparse, evidence from the shallow trenches (3160=3176 and 3172) along the northern and western walls and from the foundation trench (3940=3979) for the threshold of the main door outside the building in trench BI suggests that the rebuilding of the brick building in the opus vittatum style happened some time in the sixth or seventh century. The near wholesale reconstruction of the building was later followed by the elaboration of the west façade with a narthex—an element that clearly signals the building’s use as church.²⁰ The narthex cannot be dated with certainty, but such structures are rare after the sixth century in Rome.

CII yielded little evidence for just how the interior of this early church might have looked. The handful of opus sectile marble found in the earliest tombs (discussed below) suggest that such decoration may have been applied to the walls or the floor of this early structure. In their study of Roman marble pavements from the fourth to the ninth centuries, Frederico and Alessandra Guidobaldi attributes the scarcity of sixth-century pavements at sites with cosmatesque floors to their near total absorption into the later pavements.²¹ The fragment of cosmatesque pavement 3017 in CI, and evidence of such a floor in CII (discussed below) implies that this may have been the case at S. Pietro.

The opus vittatum building housed a number of burials. The earliest human remains, found inside the church and outside in the narthex, were buried in constructed tombs. A deep, oblong (1.62 x .59 m), roughly rectangular cut 3103 with an even bottom was made for a tomb (T240, described in detail below) along the north wall of the church into the foundation clay (3137). A small kidney-shaped hole 3180 with straight, regular sides was dug into the floor of this cut, near its eastern limit, and filled by (3179).²² The composition of (3179) was a unique mix of coarse, sandy earth (dark—almost black—grey color) containing some teeth, the disarticulated bones of a hand and a number of cervical vertebrae. The lowest layer of the fill was a decomposing tufa block. The fill was roughly stratified and the arrangement of the contents seemed intentional, suggesting a formal, though extremely partial, interment of an individual, or perhaps the re-interment of a burial disturbed by the building of the tomb.²³

T240
The floor 3139 of the constructed tomb T240 comprised reused marble slabs over a preparation layer of tufa blocks and terracotta tiles. (Fig. 7). The central slab of marble was pierced by a small hole.²⁴ The floor and its preparation were pressed into an even layer of dark brown clay (3155). Re-used Roman brick and terracotta tile fragments formed the lower courses of the tomb’s north wall [3150]; its uppermost course was constructed of roughly hewn tufa blocks. The wall was loosely bonded and leaned against the north wall of the building [3190] for support. A large, upright terracotta tile marked the tomb’s western limit. This tile was mirrored on the east end of the grave by another upright tile, though this one was broken.

Inside T240 were the partial remains of adult 3102*. 3102* provided a radiocarbon date of 922 ± 26 BP or 1020–1180 cal. AD (95% probability; OxA-24965). An adult male, between 20 and 30 years old in primary deposition, supine position, on a west-east axis with legs extended. The skeleton was cut along its spine for the construction of the bell-casting pit (discussed below). Only the lower leg remains of the dx side. The skeleton lay on a bed of soft, sticky yellow clay flecked with carbon (3138), which covered the marble floor of the tomb. The grave had two fills: (3104) and (3143). The latter occupied the southeast end of the tomb and was the earlier of the two fills. It contained a near-complete skeletal reduction of a single individual, found in a jumble under the lower section of 3102*‘s right leg. Osteological analysis revealed that these bones came from a complete, if disarticulated, skeleton of an adult male. (3143) was finished with a layer of small bricks loosely arrayed in an opus spicatum pattern, more of which were found throughout the second fill for 3102*, (3104).

T241
The materials employed in T240 and their arrangement as well as the orientation and placement of the tomb were similar to those of an adjacent grave to the south, T241; far less of this second tomb remained intact and there were no human remains associated with it (Fig. 8). The cut 3103 for T240 was not continuous with the cut for the T241, 3154=3152. The same dark brown earth/clay mixed with small fragments of brick and cement (3155) and (3156)=(3157) was packed into the cuts of both tombs, however, and like T240, an upright terracotta tile defined T241‘s eastern limit. At the foot of this tile lay a second tile covered with straw-colored mortar. The mortar preserved the impression of something flat that had once laid on its surface, suggesting that here too the floor of the tomb was finished with some finer material. Indeed, fragments of polished white marble slabs were found in fill (3153) at the eastern end of the tomb. The fill at the western end of the tomb (3151) also contained traces of mortar, tufa and pieces of terracotta tile. Given their alignment and the homogeneity of materials and method of construction, T240 and T241 are understood to be the result of contemporary actions.

Dating of the tombs is difficult, due to a lack of securely dated materials as well as significant disturbance in this area caused by the construction of the bell-casting pit and numerous subsequent burials (discussed below).²⁵ Pottery recovered from T240‘s two fills dates to different periods: (3143) contained mid-fourth/mid-fifth-century material (including Hayes 67), (3104) later Roman and medieval material (not datable). Radiocarbon dating of 3102* places its interment no later than the third quarter of the twelfth century, but given the date of the diagnostic material in 3102*‘s earlier fill (3143) and the near-complete skeleton reduction, it is assumed that 3201* was not the tomb’s first occupant and that the tombs were constructed at some point after the sixth/seventh century (since T240 cuts the so-called exploration trench) and subsequently reused in the twelfth century and after. The fills (3175)=(3161) of the “exploration” trench along the western wall contained large fragments of many of the same materials that were used in the tomb (e.g., tufa blocks, white marble slabs, terracotta tiles). Likewise, the mortar used in the tombs was identical to that found in these contexts and attached to the column base. If we take these similarities as connoting contemporaneous or near-contemporaneous actions, then it is possible that the tombs were associated with the early church, possibly contemporary with the burials in the narthex. Outside the church, in BI, tombs typologically similar to those found in CII are clearly related to this addition to the façade.²⁶ Typology of construction is not, of course, a consistently reliable indicator of chronology. If, however, contemporaneity is assumed on the basis of technique, then we might state that once the building was transformed into a church, its interior and immediate surrounds began to be used as a place of burial.

The monastery (central medieval period)

A monastery was founded in the tenth century at Villamagna. Between c. 1000–c. 1200 a bell tower, cloister, and cistern were constructed in the immediate vicinity of S. Pietro. The basilica’s narthex was replaced with a porch, its apse rebuilt, and a cosmatesque floor was laid.

In the northwest corner of the trench a rectangular cut 3146 was made in the fill (3171)=(3159) of the “exploratory” trench 3172=3158 for a small ossuary pit (3112). A secondary burial, the pit did not contain entire skeletons, but only major bones, including four crania, neatly stacked against the wall, and numerous long bones. These bones were covered by fine, powdery earth of an even consistency with very few other elements. The context was partially uncovered in 2006; at which time it was thought to be associated with the later burials in the trench (T232, etc., discussed below) since, unlike the other early contexts discussed above, it was not covered by pavement 3127=3128. At this time, (3112) was visible through 3111, which appeared to be a large L-shaped cut in pavement 3127=3128 made for the ossuary pit. When an attempt was made to excavate (3112) at this early point, it was found that it passed under 3127.²⁷ 3111 was reassessed and is now thought to have been a break in the pavement rather than a cut. As the fill of (3122) settled, the pavement was weakened and gave way.

At some point, the uneven surface of the foundation clay (3137)=(3173) was leveled with a layer (3141)=(3142) of dirt and debris (i.e. bricks, roof tiles, worked marble, painted plaster, and shards of glass and pottery). This was covered by a bluish-grey mortar flecked with white and red. Fragments of this pavement preparation were found throughout CII: 3127=3128=3131 (Fig. 9). The pavement must have once covered the entire area of the trench and extended beyond its eastern and southern limits. It was cut along these edges by the foundation trenches of the southern and eastern walls, 3033=3197 and 3174=3196, respectively.

The pavement preparation may be associated with the cosmatesque pavement, which, on stylistic grounds, might have been installed as early as the first quarter of the twelfth century.²⁸ Laying this floor seems to have entailed the dismantling the ninth-century liturgical furniture. Fragments of white marble and pale-colored limestone carved in low relief on one or two sides were found throughout the site of Villamagna, with relatively greater frequency in the immediate vicinity of the church of S. Pietro. Remnants of liturgical furniture, these fragments bear witness to an extensive decorative campaign possibly undertaken in the late eighth century, but more probably during the first half of the ninth century, when this form of decoration had become widespread and the villa was inhabited by a wealthy occupant.²⁹ None of the fragments were found in situ, many were neatly trimmed and reused as construction material. One such fragment, AE1049, was incorporated into the wall of tomb T262. Inside this tomb were the partial remains of an adult 3688*/(3689). 3688* provided a radiocarbon date of 977 ± 27 or 990–1160 cal. AD (95% probability, OxA-24873), which helps us date the dismantling of at least some of the ninth-century liturgical furniture to the eleventh century/first half of the twelfth century. The later date would suggest that the dismantling did in fact correspond with the making of the cosmatesque floor.

The subsequent features relate to the use of the building during the middle ages, when it was the abbey church of S. Pietro in Villamagna. The first major feature was a large oblong cut 3135, a bell-casting pit, which cut through many of the earlier contexts discussed above (Fig. 10). The pit was for the casting of a fairly large bell with a diameter of approximately 70 cm. The excavation of the bell-casting pit is best understood in light of the 2004 article by Elisabetta Neri and a detailed description of the process in the early twelfth-century treatise De diversis artibus attributed to Theophilus ‘Presbyter’.³⁰ The material unearthed in 3135 and its articulation suggest that the bell-making technique employed at Villamagna was very like the method described by Theophilus, which is largely distinguished by its use of a wax rather than clay model.

According to this method, a clay negative of the bell was formed and then covered with strips of wax, which was covered, in turn, by more clay. The whole was bound with metal girds. A metal hook, by which the clay and wax preparation for the mould would be maneuvered, was inserted at the top. At this point the proto-mould was put in the praefurnium, a deep pit (the southernmost area of 3135), the contours of which echoed the rounded shape of the bell (Fig. 11). At the bottom of the pit a low platform was constructed, equal in diameter to the base of the mould, which would rest on it (Fig. 12).³¹ The platform was divided into two low, hemispherical walls, [3124] (west) and [3132] (east), made of reused brick fragments and bonded with a mortar darkened and made brittle through exposure to heat. These low walls were each about 30 cm high, a little less than 1 m long, and 32 cm wide. They were oriented on a north–south axis. Their uppermost course was finished with a layer of soft, gummy yellow clay, into which terracotta tile fragments had been pressed (3134) .

A 48 cm-wide canal, also running north–south, separated the platform walls. To the north, this channel opened onto an oblong extension (also (3135)) of the firing pit used to feed and ventilate the fire, which stretched nearly to the north wall of the building (cutting T240 and T241 as discussed above). Presumably, this set-up would have been mirrored on the south side of the praefurnium, which falls outside CII’s southern limit of excavation. According to Neri, two small dividing walls would have enclosed the praefurnium; in the case of Villamagna these would have run east-west along the north and south sides of the platform.³² Survival of these walls (or even evidence of them) is rare. However, a concentrated deposit of building debris (e.g. bricks, mortar) was found at the northern mouth of the canal, which may be the remains of this wall or the dismantled praefurnium wall (discussed below).³³ In order to insert the proto-mould without damaging it, the praefurnium was first filled with dirt, the proto-mould set on top and the dirt removed (first from one side and then from another) until only a thin layer separated it from the platform.³⁴ A fire was lit in the channel. Thus heated, the wax would have melted, flowing out from between the layers of clay and leaving a void in the form of the bell to be cast. In the praefurnium the melted wax was completely incinerated; a thin layer of carbon would be its only trace.³⁵

Higher walls were then constructed along the west and east sides of the platform. These fully enclosed the mould for its firing. At the time of excavation, no trace of the praefurnium‘s eastern wall remained, but the western wall [3106] was still intact.³⁶ [3106] stood 67 cm high and was made of re-used brick and rough-cut tufa blocks weakly bonded by a pale grey mortar. The narrow wall was curved slightly so that it hugged the convex western face of [3124]. Stiff and sticky greenish-grey clay (3130) was packed around the base of the mould, around the platform and along the bottom of 3135 (Fig. 11). This layer acted as an armature supporting the various elements involved in making the bell.

To fire the mould, wood and carbon were packed into the space between it, [3106] and the (lost) eastern praefurnium wall.³⁷ A cover was used to control the temperature. When the firing was nearly complete, bronze for the bell was heated nearby in a crucible (not found).³⁸ The fire of the praefurnium was then doused, its east wall dismantled (to permit access) and the bronze poured into the hot mould to about half the height of the bell. Earth was then tightly packed around the entire mould to its very top, leaving only a small opening for pouring the remaining bronze. The bronze was poured slowly and with caution. Sounds, a ‘murmur as of thunder’, would have alerted the workers to any unwanted airpockets. The bell was allowed to cool slowly, the earth packed around it acted as insulation.³⁹ Once cooled, the bell was freed by breaking the mould. Numerous charred fragments of the mould, (3133), arrayed in an ovoid pattern, were found embedded in (3130) (Fig. 13). Over this was a stratified deposit of burnt material (3122).

(3122) was replete with bronze balls and pieces of slag. On its east side the consistency of the context was variegated: most of the largest inclusions—terracotta fragments, opus spicatum bricks, pieces of worked marble and pottery (medieval; not dated)—were found here. Nearer to wall [3106] the context had a consistent, sandy texture. Its color, however, varied dramatically. It was black close to the wall and a deep reddish-black moving eastward. The upper-most layer was yellow clay covered in a thick layer of ash. This context and all the structures associated with the praefurnium ([3124], [3122] and [3106]) were easily removed using a hand pick and trowel. A brown earth fill (3136) acted as a lining between these elements and cut 3135.

A document from the archives of the monastery of Villamagna, now in the Archivio Capitolare di Anagni, is a contract between the monastery and Magister Jacobo de Ponsulino, whereby the monastery dispenses of its debt for the casting of a bell by selling an olive grove near S. Pietro in Vineis.⁴⁰ The document dates to 28 January 1274, and it seems likely that the bell cast by Master Jacob, was the bell cast in trench CII.

The remains of the bell-casting process and all the earlier contexts were covered by a layer of dirt (3125) that acted as a leveling agent for mortar pavement 3058=3035=3019=3100 and its numerous preparation layers. A bright orange layer of burnt clay 3118, a thin layer of fine mortar 3126 and a layer of yellow clay 3126 were all part of the pavement preparation system in the southwest area of the trench; which were (respectively) equal to 3076 and 3117 in the east area and 3056, 3119 and 3120 in the northwest. A handful of cosmati fragments were found in the fills below this pavement; a far greater number was found above. This evidence suggests that the twelfth-century cosmatesque pavement was preserved and relaid after the bell was cast.

Castrum/late medieval burials

Pavement 3058=3035=3019=3100 was cut by a series of predominantly west–east oriented intercutting rectangular earthen graves containing the articulated remains of four adult males and females, as well as a number of infant burials (Fig. 14). The area immediately above T241, along the northern wall, was used for two burials. The earlier grave (T240; 3098/(3096)), was made for an adult (30–40 years old) of indeterminate sex 3097*. The remains of 3097* were much disturbed, pushed to the north for the subsequent burial of 3069* in grave T232 (3064/(3065)) immediately above. 3069* is a 20­- to 40-year-old adult male laid out on a west–east orientation, in a supine position with arms crossed and legs extended. The remains show signs of severe rheumatoid arthritis and spondyloarthropathies (inflammatory joint diseases of the vertebral column), as well as various traumas, including healed fractures on a rib and finger-bone. The fills from these graves yielded no datable materials; 3069* provided a radiocarbon date of 647 ± 25 BP or 1280–1400 cal. AD (95% probability; OxA-24876). A number of small grave goods were found in (3065), the fill of 3069*‘s grave, including a barrel-shaped bead made of bone (O13), a small ornament made out of a shell with a drill hole (O538) and a bronze finger-ring fragment (O323). This last object will be discussed in conjunction with similar objects below.

T227 (3047/(3046)) contained an adult female 3048* of 30–50 years. The cut made for the grave, 3047, was large, deep and in the very centre of the trench, around which many of the numerous infant and child burials seem to have been arrayed. A bronze ring (O5) with an oval bevel, but missing its (presumably glass) inlay, was placed on the fourth finger of 3048*‘s right hand; a second ring (O7) and a piece of worked bone (O536) were found in the grave fill (3046).

A series of poorly preserved infant burials surround T227. Immediately north of the tomb were the graves of two infants between 1–3 years old. These have been counted as a single tomb, T229 (3049/(3050)), since they share a cut and fill (the individual cuts were indistinguishable). The earlier, 3060*, was cut by 3059* who lay directly on top. To the west of T227 is T228 (3051/(3052)) the grave of another infant 3053*, some 2–4 years old. Above this grave and to the east lay several intercutting infant burials, only some of which were intact. The earliest of these graves appears to be T239 (3095/(3093)) that of an infant of 1–2 years (3094*). T239 was cut by T238 (3090/(3092)) of a 2- to 4-year-old child (3091*), which in turn was covered by T237 (3087/(3089)) of a 2- to 4-year-old child (3088*) and T236 (3084/(3086)) for an infant of 0­–2 years (3085*).

South of T227 were more infant burials. T234 (3070/(3078)) was made for a 3- to 6-year-old child 3074* and, next to this, T233 (3071/(3072)) for an infant of 0–1 years 3073*. These tombs were cut by T231. Partially covered by the (later) south wall, lies the west–east T231 (3063/(3062)) of a 20- to 30-year-old female 3061* buried, like 3048*, with a ring with a circular bevel on her right hand (O6). The disarticulated bones of a neonate were found above her pelvis, suggesting that she may have died during pregnancy or childbirth. Two infant tombs seem to have been oriented with respect to 3061*: T235 (3080/(3081)) for a child of 3–5 years 3079*, who was buried at the head of 3061*, and T242 (3113/(3115)) an infant of 1–3 years 3114*, buried at her feet. Unlike its contemporaries, 3114* was laid out on a south–north orientation parallel to the short side of the rectangular stone [3029] in the southeast corner of the trench.

Both sexes are present among the adults (three females, two males in this area), but these burials are striking for the high number of neonate, infant and child burials under the age of six—some 73% of the individuals in this phase. The individuals were uniformly laid out in a supine position with arms flexed and legs extended. The majority was buried on a west–east orientation, aside from an infant 3114* in the southeast corner. The lack of evidence for nails and the oval shape of grave cuts, when visible, suggests that wooden coffins were not used. Given the absence of shroud pins, it seems likely that the normal procedure was for bodies to be dressed for burial. Datable fragments of pottery in the grave fills of 3048*, 3061* and 3022* place these burials somewhere in the mid-thirteenth to the mid-fourteenth centuries, which is consistent with the radiocarbon date for 3069*.

Although only sixteen articulated partial or complete skeletons were recovered during excavations, osteological analysis revealed a minimum of twenty-three individuals were buried here (nine adults and fourteen sub-adults). The location of these burials inside the church suggest that these individuals were of a higher social status than those buried outside in the cemetery. This interpretation is supported by the presence of a greater number of costume elements and bone and bronze jewelry (three of the nine adults had costume elements associated with their graves) in CII than in the burials outside the church.

Early modern and modern use of the church

Subsequent to these burials, the church roof collapsed, an event attested by a layer of burnt debris (3027) covered by roof tiles (3040) and mortar (3036). These layers may represent a destruction of the church by burning. (3040) was cut by a series of foundation trenches along the main walls of the church, which seem to reflect the rebuilding of the church at some point after the fortifications of the castrum were no longer important.⁴¹ The opus listatum of the exterior walls was taken down to a very few courses above the brick walls and replaced with coursed rubble walls, [3006] in the west and [3007] in the east. These were built into construction trenches: that to the north, 3025/(3026) is very clear along most of its length, while that for the western wall was to some extent confused with the burial of two individuals along it (see below). The rubble shows a very regular use of large ashlar blocks, perhaps deriving from the bell tower or other monastic structures.

Subsequently, further construction created the room in which the excavation took place. The east wall, [3008], forms part of a transverse arch, the westernmost of a series of three that crosses the church, abutting the outer walls and supporting the roof. Their construction technique, in rubble without the use of ashlars, is rather different from that of the new outer walls, but it is hard to see how the roof would have been supported without them. The south wall [3004] was built over a continuous foundation, [3031], which bonds with that of [3008], [3032]. This suggests that the room was an integral part of the planning that included the transverse arches. A large door in the south wall opened onto the vestibule of the church. This seems to have been roughly in the middle of the wall. To the east of it two small walls were built to surround the old marble plinth [3029]: [3027] on its north side, [3028] on its west side. This may represent a cupboard for church furniture, opening onto the vestibule to the east of the door. These walls were subsequently removed, and the earlier door blocked by [3005] creating a new door over the plinth.

As noted above, [3004] preserved the contour of a column associated with the column base [3068] at its westernmost limit. This column must have been in place when the wall was built and then subsequently removed, possibly when the room was reconfigured and the door over [3029] was opened. After the removal of the column, two individuals were buried in the southwest corner of the room. Their burials are of a much less certain date. The grave T225 is on a south–north orientation. The earlier of the two burials was an adult female 3042*/(3041)/3021=3038, 20–40 years old. She was buried with a bone rosary (O572). The grave was re-opened to inter an adult male 3022*/(3020)=(3039) of 20–30 years. Both individuals lack crania. It was initially assumed that this loss was related to the removal of the column, but the diameter of the impression left by the column in the wall and the position of the column base suggest that the column would have had to have been removed before their burial (the cut for the column’s removal, which was never identified, may have been lost with the digging of these graves and the foundation trenches discussed above). How then, they lost their heads remains unclear. The anomalous position of the graves may suggest that whoever buried them was aware of the position of the earlier burials.

The construction trenches were covered by a layer of earth (3018) which may represent a floor or, perhaps, the makeup for the cement paving 3009 which was laid over it, forming the final floor of the space and just covering the foundation offset. The cement included many fragments of cosmatesque pavement in its bedding, and we might suggest that any remains of the cosmatesque floor were destroyed at the point of its creation. This would tend to suggest that (3018) was in fact the preparation for the floor rather than an earlier version.

Our interpretation of this space remains uncertain. Although often referred to as the “chapel” during the excavation, there was no evidence for any decoration, except for some plaster with red frescoed paint in the destruction layer cut by its foundations. A more likely explanation for this dark space, covered by a loft on wooden beams, is as a sacristy. This whole construction remains undated: it is certainly later than the burials, which radiocarbon dates to the early fourteenth century. Insofar as it is probably related to the reconstruction of the door of the church, with its reused Roman inscription, it is earlier than c. 1889, when this was recorded by Stephenson. Visitation documents do not refer to either chapel or sacristy in the church.

Bibliography

De Meo, M. (1998). S. Pietro di Villamagna presso Anagni: una villa romana si trasforma in Abbazia. Quaderni di architettura e restauro. Rome.

Flascassovitti, C. (2007). Le Pergamene del Monastero di S. Pietro di Villamagna. II: 1238–97. Lecce.

Glass, D. F. (1980). Studies on Cosmatesque Pavements. BAR international series. Oxford.

Guidobaldi, F. and A. G. Guidobaldi. (1983). Pavimenti marmorei di Roma dal IV al IX secolo, Studi di antichità cristiana. Pontificio Istituto di archeologia cristiana. Vatican City.

Neri, E. (2004). “Tra fonte scritti ed evidenze archeologiche: Un modello per interpretare i resti materiali della produzione di campane.” Archeologia Medievale 31: 53–98.

Theophilus (1961). De diversis artibus. Trans. by C. R. Dodwell. London.

---

1^ De Meo 1998: 29