Rochester Cathedral Masons’ marks

161 ROCHESTER CATHEDRAL MASONS’ MARKS jacob h. scott Symbols inscribed into dressed stonework, masons’ marks identify the work of the individual or team that quarried or dressed the stone, or else constructed the architectural feature on or near which it is found (Fig. 1). Marks for other purposes can also sometimes fall under this broad term, such as setting-out lines to instruct the dressing of stone or the construction of an architectural feature. Millions of these marks survive on the fabric of cathedrals, castles, churches, fine homes and palaces throughout Europe and beyond. How or when the use of masons’ marks was introduced to England is unknown. Marks survive on Anglo-Norman Period buildings, but it is not clear if the Normans introduced the use of masons’ marks or if it predates the Conquest (Alexander Fig. 1 Twelfth-century mason’s mark in the shape of a trefoil (type 3a282). JACOB H. SCOTT 162 2007). The most numerous masons’ marks found at medieval sites are those of the ‘banker’ mason who was responsible for the final preparation of masonry blocks. Banker marks were used in calculating pay and tracing issues back to source. At least 4,000 banker marks occur on visible surfaces of the twelfth-century fabric of the nave and west front, crypt, quire and cloistral buildings of Rochester Cathedral, then the Cathedral Priory of St Andrew. This is the last of three reports on a four-year graffiti survey at Rochester Cathedral (see also Scott 2018 and Graham and Scott 2019). This survey of Rochester Cathedral masons’ marks has strived to be the most comprehensive of a building of this size to date, and aimed to develop robust and practical recording and interpretation techniques for replication elsewhere. This study uses the new photographic record of these sequences of marks to establish a relative chronology for the construction of fabric. The feasibility of creating working biographies of these unnamed masons is assessed, in terms of time on site, features and designs worked. This data will be used to corroborate and supplement the absolute (albeit incomplete) chronology provided by architectural-historical models. Architectural history of Rochester Cathedral All studies of undated graffiti must work within the absolute chronology provided by architectural-historical models. It is immediately apparent that the standing fabric results from at least two major building campaigns, the largely Romanesque west end and the early Gothic east end. Although the estimated dates of these portions of the structure have long varied, most assessments agree that they were separated by less than a century and were both related to contemporary accounts of two medieval fires in 1139 and 1179. William St John Hope (1898) produced the first comprehensive architectural history of the cathedral. St John Hope’s study was based partly on first-hand observations during the late nineteenth-century restorations under Sir George Gilbert Scott, of James T. Irvine clerk of the works, and the cathedral precentor Revd Grevile M. Livett (Halsey 2006, 62). Livett was a notable architectural historian of West Kent and a contributor to Archaeologia Cantiana. St John Hope produced the first building phase plan of the cathedral, linking almost contemporary records of the patrons of work, typically the presiding bishops or priors, with the phases and styles observed in the architecture. The patron of the earliest form of the standing building is recorded as Bishop Gundulf (Textus Roffensis c.1123, f. 172). Gundulf’s episcopacy lasted from 1075 to 1108. Livett suggested the west end of the building including the nave arcades, partially constructed in tufa stone, had been cased and redecorated in Caen stone c.1139-1142, although St John Hope revised this to 1114-1124 (1898, 218). Studies of the architectural styles of the Caen stone west front generally agree that it immediately followed the redecoration of the nave arcades, in the 1150s or 1160s (Worssam 2005). The surviving portions of the chapter house and dorter within the east range of the cloisters have been dated to the episcopacy of Bishop Ernulf (1114-1124), inferred from the accumulation of several plots of land to the east of the cathedral during this time (St John Hope 1900, 7). Subsequent estimates place the architecture of the east range to around 1160 (McNeill 2006, 186), around the time of the ROCHESTER CATHEDRAL MASONS’ MARKS 163 major cathedral redecoration campaign after 1137 (Worssam 2005). Difference in sculptural decoration is interpreted as evidence that a different workshop produced the surviving east range of the cloister than the almost contemporary work on the nave arcades. The east end of the cathedral comprises the quire, presbytery, quire transept and crypt. Studies agree the tufa western portion of the crypt dates to the 1080s, with the transept and sanctuary crypt (now referred to as the Ithamar Chapel), generally dated to after the fire of 1179. St John Hope (1898, 233) suggested 1200 to 1215 as Bishop Gilbert de Glanville is recorded as having finished the quire from the proceeds of a new shrine to St William of Perth (d.1201), proposing that this referred to the entire east end of the building except the liturgical quire. St John Hope interprets a 1227 record of the monks’ entry into the quire as referring to the liturgical quire and marking the completion of the east end work. Caen stone and Reigate ashlar are used interchangeably in these parts of the building, with the contrast between the yellow Caen and the grey-green Reigate used for decoration. Recent revisions to St John Hope’s architectural-historical model tend to agree with this division of building phases, although attempt to more closely reconcile the dates for the Romanesque and Early Gothic fabric with contemporary references to two extensive twelfth-century fires. On the 3rd of June 1137, the entire building was said to have been burnt (Gervase of Canterbury c.1200). McAleer (1996, 158) concurred with Livett that the nave campaign was a result of the fire of 1137. McAleer (1985) proposed a building sequence for the nave arcades based on variations in architectural features and perceived changes in design between the piers and the arcades above. The bases of the westernmost piers of the north nave arcade differ from the easternmost pier and all those of the south nave arcade. There is Caen stone decoration on the large arches on the aisle side of the north nave arcade but not on the south. These architectural anomalies were taken as evidence that the south piers were constructed or cased first, then the north piers and the north arcade, then the south arcade, followed by the west front. On the 11th of April 1179, fire damaged the east end of the building (Cotton MS. Vespasian A. 22, f. 30.). Recent studies have tended to place the rebuilding of the east end more immediately after the fire than St John Hope (Draper 2006, McAleer 1996). Three forms of evidence are used to suggest that work on the east end began at the junction between the old presbytery and the new quire transept. There is a shift in use from Bethersden marble to Purbeck for the string course that runs throughout the crypt level of the quire transept and sanctuary, roughly in line with the entrance to the Ithamar chapel (Worssam 2000). Local Bethersden was the predominant marble in use in Kent in the twelfth century, being eclipsed by Purbeck marble from Dorset in the thirteenth century. This junction was also the logical place to begin work to maintain levels between the old and new work. Work would then be expected to have largely been concluded by the time of William of Perth’s death in 1201. His shrine is said to have resided in the north quire transept until the Reformation. The quire aisles, nave transept and the three easternmost bays of the nave are dated to the mid-thirteenth century (St John Hope 1898). A large annexe to the south nave transept Lady Chapel dates to the 1490s, concluding the major medieval building campaigns. Contemporary records exist for buildings in England from most periods in which JACOB H. SCOTT 164 masons’ marks survive. Ample contemporary documentary evidence details the rates of pay for masons and other artisans employed at some cathedral sites, as well as for the tools and techniques used, and the conditions of their employment (Alexander 2008, 22). No such documentation survives from the medieval works at Rochester Cathedral, although as will be discussed, the volume of marks closely resembles buildings with documented banker mark sequences. Methodology Academic interest in masons’ marks began in the nineteenth century. These earliest studies traced identical mark types between buildings in an attempt to identify the same masons working at separate sites. However, it became apparent that as the majority of twelfth and thirteenth-century marks are composed of just a few lines, they cannot have been unique to one mason. Instead, the vast majority must be coincidental uses of site-specific or even task-specific marks (Alexander 2007, 63). Recent studies of marks have found more success in focussed investigations Fig. 2 Three-dimensional virtual model of nave arcades and west front with masons’ mark types coloured at random. ROCHESTER CATHEDRAL MASONS’ MARKS 165 of marks within individual structures to identify building phases and significant pauses in work. The simplest of these surveys identify common marks on separate portions of fabric to suggest contemporary building phases. Distinct mark sequences are interpreted as resulting from pauses in construction. However, many of these surveys were conducted from ground-level and so are limited in the data they provide. Modern comprehensive surveys divide structures and sites into individual architectural features, recording all visible marks, and compare and contrast the quantities of mark types within sequences from each feature. Each mark at Rochester Cathedral has been photographed with a Canon EOS 6D. An index of records spreadsheet records the height from the floor of the bottom edge of each ashlar featuring a mason’s mark. A three-dimensional virtual model (Trimble Inc. Sketchup 2015) of the surviving portions of twelfth-century fabric records the location of each marked stone (Figs 2 and 3). The height of marks and proportions of architectural features out-of-reach of a 3m tape measure were estimated photogrammetrically, resulting in a lower level of accuracy than can be achieved by direct recording, although extending the survey to over 95% of the interior fabric. Plotting the 4,000 visible marks on two-dimensional plans or Fig. 3 Three-dimensional virtual model of east end with masons’ mark types coloured at random. JACOB H. SCOTT 166 elevations would be hugely impractical, with many marks occurring on the interior faces of arches or shafts. Comprehensive surveying and recording of masons’ marks provides more data than simple comparative analysis between forms and fabric. Banker marks can establish a course-by-course relative chronology for the construction of architectural features, impractical in buildings with thousands of marks without three-dimensional modelling. Distinctions can be confirmed between almost identical mark types by a close investigation of their relationship to one another. The many masons’ marks above head height were reached with a 10-metre Genie Aerial Work Platform 30S (Fig. 4). Marks higher than this were identified and recorded with a photographic zoom lens and without a scale, including those on the upmost portions of the quire transept clerestory and the interior and exterior triforium level of the west front. The limitations to using raking light from below the inscription to identify out-of-reach marks means it is easier to miss lightlyincised marks. The coding of masons’ mark types follows a methodology outlined by Alexander (2008, 34). In this system, mark types are coded according to how many lines are used within the symbol and assigned a letter according to their approximate likeness. The last number in each code denotes the number of examples recorded. Fig. 4 Using the vertical access platform and raking light to survey masons’ marks on the south arcade of the sanctuary. (Photo: Alex Pitcher.) ROCHESTER CATHEDRAL MASONS’ MARKS 167 The type of stone on which each masons’ mark occurs is recorded, with a description of sculptural decoration. It can then be ascertained if masons worked exclusively on one type of decoration, for example, or using only Caen or Reigate stone. These fields are recorded in the index of records spreadsheet. Typology This section investigates the types of masons’ marks at Rochester Cathedral, regarding both their purpose and form. This survey has recorded 3,912 examples of 73 distinct mark types. The sheer volume of marks suggests that the vast majority, if not all, are banker marks. Marks resulting from those constructing the finished ashlars into their respective architectural features would typically be far fewer in number, perhaps occurring only once on each feature or site. Marks occur on approximately 20% of visible twelfth-century ashlars, in both the east and west of the building (Fig. 5), indicating that the banker masons created them immediately after the dressing of each stone. These marks became redundant once the stones were counted, so the remaining 80% are believed to exist on the obscured faces of stones. Supporting this interpretation are several marks found on the faces of carved stone fragments originating from the cathedral that would not have been visible in situ (Fig. 6, A and B). Many truncated or partially obscured examples also supports their identification as banker marks, created before the final dressing to fit irregular ashlars. Fig. 5 Digital trace of masons’ marks on a pier in the crypt. Marks are visible on approximately 20% of twelfth-century ashlars. JACOB H. SCOTT 168 Fig. 6 (A and B) Worked stone fragments featuring masons’ marks recovered from the cloisters. (C to G) Fragments re-used in the north nave transept gable featuring masons’ marks. (H) Pencil number on a nineteenth-century ashlar in the south quire transept gable. A B C D E F G H ROCHESTER CATHEDRAL MASONS’ MARKS 169 A complete absence of visible marks in some portions of the building may have resulted from an evolution in the procedure of marking stones. The north nave transept gable was rebuilt in the 1890s under the architect Sir George Gilbert Scott. The rough stonework on the interior of the gable features many re-used worked stone fragments. Several of these feature masons’ marks, all of which would be obscured when the stones were in situ (Fig. 6, C to G). These marks are deeply incised to improve bonding between the stone and mortar. Masons yards today routinely leave marks only on the obscured faces of stones. Several of these marks are in the form of two-letter initials, which is unknown within the major sequences of marks in the nave and quire. The stones re-used in the north nave transept gable are apparently of a later phase than those in the main sequences on the in situ fabric, although a precise provenance has yet to be determined. An evolution in tradition to only leaving marks on obscured faces of ashlars may account for the apparent absence of visible marks in the quire aisles, nave transept and Lady Chapel, although these areas are heavily plastered. A dozen graphite numbers survive on apparently randomly dispersed ashlars of the nineteenth-century south quire transept repairs, and in the south nave transept gable (Fig. 6, H). These are perhaps notes created during transportation, sorting or storage, although they appear in so few a number that further investigation of their purpose is restricted. Seventy-seven masons’ mark types form the main sequences on in situ fabric (Fig. 7). Of the 3,912 examples recorded, 197 degraded or partially obscured marks (5%) are illegible. Fifty-four types occur on the surviving portions of the twelfthcentury nave and west front. Forty types recorded across the crypt, quire transept, presbytery and cloisters are almost entirely distinct from those in the west end. Fifteen mark types occur across the nave, crypt and the quire. As the earliest date for the construction of the east end provided by architectural-historical models is some twenty years after that of the nave, it seems unlikely that these marks were from the same masons. The fourteen mark-types on the surviving portions of the chapter house and dorter are also apparently of a distinct phase. Of these 77 types 63 are composed of six or fewer lines, indicating a high likelihood of coincidental use. Coincidentally similar mark types can be identified by differences in handstyle, epigraphic execution, or through observing a significant pause in its use. Fourteen mark types match or closely resemble twelfth-century Latin letters; 2r53, 2r414, 2x2, 2x191, 3h105, 3n215, 4m16, 4m89, 4w14, 5a2, 5a43, 9m13, 10w2 and 12a4. Several of these could conceivably be initials. Mason’s marks from the later Middle Ages more often resemble initials than earlier marks, thought to be a result of an increase in literacy over the period (Hislop 2012, 46). However, the twelfth-century marks at Rochester Cathedral that do resemble letters can often be reflected, or include ‘errors’, such as type 3n215. Many stones in-situ are also in a position that the banker would have known would leave a mark on its side, perhaps indicating at least some of these resemblances to actual letters may be coincidental, or else their recognition was unimportant. It is also of interest that seven mark types precisely match Anglo-Saxon runes; 2r414 (rad/r), 2x2 and 2x191 (gyfu/g/j or nyd/n), 3f74 (tiw/t), 3n215 (sigel/s), 3s280 (ior/ia/io), 4b386 (daeg/d) and 4m89 (eh/e). Like masons’ marks, runic is comprised of rectilinear characters best suited for inscribing into hard surfaces, JACOB H. SCOTT 170 so this could well be coincidental. However, all but ior/ia/io are the initials of the most common male forenames in Britain during the Middle Ages, many of these introduced by the Normans. The use of runes declined after the wider adoption of the Latin alphabet by the eleventh century, but it is conceivable that some masons were familiar with the signs and symbols of their grandparents and greatgrandparents. Runes do not constitute a true alphabet, instead seemingly most often used for magical purposes, basic signs, or to represent names. Six mark types resemble tools or weapons. Mark type 4a1 may represent a pair of tongs. 4p10 could be an axe and 8t1 an axe head. 5p57 and 6b20 may be knives and 6t1 possibly a mallet. 5t15 could represent a bow and arrow. 4s1 is a wellformed sword. Marks resembling tools are not uncommon at other medieval sites. The 42 other mark types are seemingly abstract designs, typically comprised of as Fig. 7 Typology of twelfth-century masons’ marks coded by a system outlined by Alexander (2008). ROCHESTER CATHEDRAL MASONS’ MARKS 171 few lines as possible to be distinct. The average number of lines in the mark types of the west end is five, and the average in the east end is four, perhaps indicating that more complex mark types were required to distinguish between the larger number of masons working on the west end. Marks 3r4, 5a2 and 5a43 are the only types in the building to feature serifs, a slight projection finishing off a stroke of a character. As mason apprenticeships are believed to be common on medieval building sites, this supports the revision of a traditional interpretation that seriffed marks were used by apprentices to distinguish between the mark of their master (Martin 2005, 379). It does seem that several mark types may have been modified from more commonly-used simple symbols to distinguish between similar marks; 2x191 and 6x95, for example, and 4a1, 5a43 and 9a2. Mark types 10w2 and 12a4, occurring on the west front c.1150, appear to be bold versions of 3n215 and 5a43 respectively. Superficially, it appears as though this could be to distinguish these marks from an identical mark of another mason. These latter marks, however, occur only in the east end of the building and thus antedate 10w2 and 12a4 by several decades. Of interest to the interpretation of an absence of masons’ marks on later portions of fabric are mark types 1l67 and 2t58. These very basic marks are only distinguishable from incidental marks by their recurring positions on a corner of each ashlar. Occasionally, two marks occur on a singular ashlar and even more occasionally they can be different types. As these stones do not appear re-used, the cause of this is not immediately apparent. In addition to the main sequences, several unique marks occur on other fabric. 5t1 occurs on the keystone above the bay window of the Chapter Library. When singular marks such as these occur in prominent positions on architectural features, they may be the mark of those constructing the feature. Without any recurrence of the mark, this is difficult to establish. Many could be setting-out lines, quarry marks, or partially legible graffiti. Biographies This section addresses the feasibility of establishing biographies of anonymous masons’ work from the recurrence of mark types. Is it possible to estimate the size of the banker mason workforce during each building phase from the number of mark types? Can it be discerned how many masons worked at the cathedral for a considerable duration and how many for a limited time only? Did some or all of the masons specialise in certain designs or types of stone? Tracing each mark type around the building may superficially appear to offer an opportunity to construct a biography of each mason’s work. However, the simplicity of most mark types suggests that several masons used identical designs. Only one or two examples of marks 2y2, 3r1, 4a1, 4h1, 4p3, 5t1, 6t1, 8n1, 8t1, 9x1 and 11t1 are recorded. Essential to the nature of these marks is an appreciation of their taphonomic circumstances; the extent of their decay, destruction and obscuration. The upper portions of the nave arcades and quire clerestory have seen much whitewashing in the post-medieval era, resulting in fewer marks surviving in these areas. Water ingress has obliterated marks in certain areas. Very few marks survive on the exterior of the building, although those that do suggest that more JACOB H. SCOTT 172 were once present. Such unfavourable taphonomic circumstances could eradicate a significant portion of the evidence of a mason’s work and limits biographical investigations. Several mark types routinely occur on the same corner or face of ashlar blocks. If a mason typically marked the same face of each stone, and only occasionally differed, this could result in the small quantity of marks recorded for many types. However, even given the destruction or obscuration of a high percentage of marks, it seems unlikely that so many masons would dress just a few pieces of stone before leaving the site. Instead, these could be types used for a short time, possibly for the construction of a single architectural feature, and discarded or modified for subsequent projects to avoid confusion between identical marks. Either interpretation means that a simple one-mark-type-equals-one-bankermason equation for estimating workforce size may not prove accurate and creating a biography of these masons’ work across features impossible. A more accurate estimation of workforce size may result from counting the masons dressing the stone for each feature at any one time. Between fifteen and twenty mark types occur on each of the piers in the nave and between ten and fifteen types are found on the nave arcades. The fewer types higher in the arcades may indicate the loss of more marks from water ingress or whitewashing campaigns. The masons’ marks of the nave suggest a banker mason workforce size of around twenty masons working at any one time, that team then moving to the next feature under construction. The marks of the quire indicate a team of around ten banker masons were at work at any one time. The occurrence of the same mark across almost all fabric within each building phase suggests that one team of banker masons were employed, working on features sequentially. Despite the evidence of single or limited-use marks, the similarity in hand style of many mark types throughout each major building campaign indicates that many were retained by individual masons, at least for the duration of each campaign. Mark types 2a50, 2b39, 2x191, 3a282, 3h105, 3r4, 3s91, 3s280, 4b386, 4m89, 4w14, 5a43, 5p57, 5s90, 5t15, 6x32 and 6x95 occur throughout the entire nave, across both arcades. These marks occur across a diverse set of designs, from plain square ashlars to the dog-tooth and chevron moulding voussoirs of the arcades and clerestory (Fig. 8), indicating that these masons at least did not specialise in sculptural detail. Work is indicated to have proceeded one feature at a time. One evolving team of banker masons worked on each of the nave and triforium piers and arches consecutively, rather than two teams working in parallel across each arcade or in another more complex arrangement. A contiguous sequence of marks occurs in the lower portions of the east end. Although the evidence for the upper portions and the clerestory is more fragmentary, here it also appears that one team of banker masons were at work. The nave arcades and west façade are faced entirely in Caen stone. The square ashlar and decorated arches in the east end comprise both Reigate and Caen stone. The surface of Reigate stone degrades faster than Caen, and so many more marks survive on the latter. All mark types that do survive on Reigate also occur on Caen. Although a later building phase for the more exclusively Reigate clerestory of the quire transept and sanctuary is proposed, it appears that masons did not specialise in either material while Reigate and Caen were being used interchangeably in the construction of the interior fabric of the east end. ROCHESTER CATHEDRAL MASONS’ MARKS 173 Estimations of the size of the banker workforce employed at any one time during the work on the east and west ends conclude that frequent reuse and discard of marks complicates the creation of biographies of most individual mason’s work. Fig. 8 Distribution of masons’ mark types by architectural detail, showing masons did not specialise in any one design or level of complexity. JACOB H. SCOTT 174 This reuse and discard further complicates the establishing of a relative chronology of building work, requiring the additional evidence provided by the absolute chronological framework based on architectural-historical models, construction logic and stratigraphic relationships between building features. Chronology There are many complexities to providing estimates for the time taken to dress and construct portions of the medieval fabric. If the masons were using a stone saw to reduce large blocks from the quarry into ashlars, instead of the older technique of splitting with chisels, wedges and hammers, it would significantly reduce preparation time. Diagonal linear tooling marks characteristic of the use of an axe occur on much of the twelfth-century ashlar, chisels being for finer work. It would be typical for an on-site blacksmith to prepare tools, but factors such as the quality of metal and smelting techniques used can affect the speed of use and wear, and subsequently the time taken to dress stone. A skilled mason can dress stones twice as fast as an apprentice. Documentary evidence indicates most works departments were significantly reduced in winter when frost could damage the partially complete structure (Alexander 2008, 22). Banker masons can, in theory, work year-round, with stone stockpiled in winter for the spring. However, they would only comprise a small portion of the entire workforce at the site. It is not known if construction occurred concurrently with the dressing of stone. Both of the two main building phases would have entailed laying foundations and demolition of significant portions of the standing structure, other unquantifiable pauses in laying ashlar. These variables make estimation of absolute chronologies using banker mark sequences impractical. A comparative analysis of the marks between each portion of the building has ascertained if they were worked on by the same team of masons. If so, they are presumed to have been working at different times, although within the same building phase. The portions were possibly constructed concurrently if they were worked on by different teams of masons but within a single building phase. In this manner, a relative chronology for the construction of the building can be produced, within the absolute chronology provided by architectural-historical models based on recorded patronage, fires and other events. An increase in the sculptural complexity of pier capitals and tympana decoration in the nave, as well as some rough work in the aisle-facing spandrels of the north nave arcade, have been interpreted as indicating the course of work in the 1140s (McAleer 1985). There is an intensification in the scalloping of pier capitals between the south and the north arcades of the nave. The bases of the four easternmost piers of the north arcade are floriated, whereas the westernmost pier of the north arcade and all the piers on the south arcade feature a simpler roll moulding. The capitals of the tympana shafts in each arcade are generally more complex to the west. These features have been interpreted by McAleer (1985) as evidence that the arcades were built three bays from the crossing tower, encompassing the fifthwest pier of the north nave arcade. The south arcade piers were then continued, followed by the remaining four piers of the north arcade with foliated bases. Then the north arcade triforium was completed, during which an intention to vault the ROCHESTER CATHEDRAL MASONS’ MARKS 175 nave aisles was abandoned. The south triforium was then finished, with more effort made to smooth the aisle-facing spandrels on this side now that it was known they would not be covered by vaulting (McAleer 1985, 157). Seventeen mark types are found almost exclusively on the north nave arcade. Five mark types occur almost exclusively on the south arcade (Fig. 9). The clear differentiation of many mark types between the arcades strongly suggests that one arcade was worked on and finished before the other, in support of the previous architectural-historical models. However, twelve marks occur on both the piers and the triforium: 2a50, 3b89, 3n215, 3s91, 4b386, 4m89, 5a43, 5p57, 5s90, 5t15, 6s49, 7t56, suggesting there was no significant pause in work during the construction of either. It is difficult to interpret this data as supportive of McAleer’s proposed model for the construction of the nave arcades. Instead, they appear to have been worked on consecutively as suggested by previous assessments, after St John Hope (1898). The roughness of the spandrels of the aisle-side of the north nave arcade, in contrast to those of the south, could have resulted from post-medieval repair rather than the change in design interpreted by McAleer. Ultimately, thick plaster and whitewash currently prevent the confirmation of either interpretation. Fig. 9 Seventeen mark types occur throughout the west end (yellow), indicating a single construction phase with no significant pause in work. Thirteen mark types occur almost exclusively on the north arcade (red) and nine types on the south arcade (blue). Three types occur exclusively on the west front (green). JACOB H. SCOTT 176 At least one altar and one shrine existed in the south nave aisle. St John Hope (1898, 273) proposes an altar at the east end on the architectural evidence for a screen here. Finishing work on the south arcade would have allowed the shrines to function while work continued on the north arcade. The eleventh-century west front is believed to have still been standing at this time, and so the demolition and reconstruction of one arcade at a time would place less stress on the west front while work progressed. Eight mark types occur only on one bay, or on two adjacent bays: 2a50, 2x191, 3n215, 3p4, 3y20, 4a19, 5t15, 6m28. These limited marks may indicate that around two bays were completed in each working season, March to November. These were presumably the marks of mason teams that were disbanded over the winter and did not return the next year. The eight mark types that occur throughout the entire nave arcades would seem to be a core group of masons retained over the winter months. Difficulties in estimating the time taken for such work notwithstanding, this would seem to be a slow pace for a group of ten to fifteen masons. Eight marks occur fewer than three times throughout the entire arcade: 1b1, 3p4, 3y20, 4a28, 4t10, 4w14, 6b20, 10w2. Remarkably, 30 marks survive on the greatly weathered and heavily restored west front. Most of the marks here are on the southern tower, in an area that was abutted by a building in the later medieval and early modern period, seemingly providing a degree of preservation. Several of these marks match those from the nave, supporting McAleer’s proposed date of c.1150 for the work on the west front, immediately following on from the reconstruction of the nave arcades beginning after the fire of 1139. The loss of almost all marks on the rest of the west front confounds further attempts to establish a relative chronology for this fabric, although similarities in mark types found in the lowest portions and a few surviving marks in the interior of the upper portions of the centre aisle towers suggests there was no significant pause in work. Very few marks survive in the cloisters, limited to small portions of ashlar protected from the weather. What survives does not resemble the marks of either the nave or the quire, supporting the architectural interpretation that work on these buildings was carried out by a different workshop than the main body of the cathedral. Of the six marks recorded within Rochester Castle keep, one type closely resembles mark type 4m35 from the cloisters. There is a clear distinction in mark types in the west end campaign of the 1140s to the 1160s and those of the rebuilt east end, now most often believed to date to after the fire of 1179. This distinction indicates a significant pause in work, of enough time for the banker mason workforce to be disbanded entirely or to have evolved to include few, if any, of the previous masons. The only complex mark type to occur across both campaigns is the trefoil 3a282. This type is one of the most common to occur in the nave campaign but occurs only in the earliest phases of the east end. Although it is tempting to propose a common mason, the trefoil is surely a symbol that every mason would have been aware of. Coincidental use by separate individuals seems more likely. Another possibility is that the mark belonged to an apprentice or family member of a mason working on the earlier campaign. A contiguous sequence comprised of mark types 2r53, 2r414, 3f74, 3n215, 4a19, ROCHESTER CATHEDRAL MASONS’ MARKS 177 4b386 and 4s43 occurs on the walls of the quire transept and sanctuary, from ground level to the string course running below each lower window approximately 6m above the floor of the quire (Fig. 10). There is then a distinction in mark types on the contiguous fabric above this height; comprising eight types. This later sequence also occurs on the east arcade of the quire transept and the eastern portions of the liturgical quire. Efflorescence damage and general wear has destroyed the majority of visible marks on the walls of the crypt transept and Ithamar Chapel. Not enough marks survive to form firm conclusions as to where work began and finished. Construction logic and the architectural model proposed by Worssam (2006) suggests work started at the junctions between Gundulf/Ernulf’s east end and the new work, moving east. Marks 2r414, 3a282 and 4b386 occur on the north wall of the crypt between the Fig. 10 Seven mark types occur in all parts of the east end (yellow). Three mark types occur only on the exterior walls, from ground level in the crypt to the arches of the lower windows in the quire above (red). Seventeen types occur across the crossing piers, the triforium and crypt (blue). JACOB H. SCOTT 178 two large buttress towers. Marks 2r414, 3f74, 3s280 and 4a28 occur on the walls of the south crypt transept. Although few, these marks broadly correspond with the mark types found on the respective walls of the quire transept above, suggesting the ashlars for each end of the crypt transept were dressed consecutively. Many more marks survive on the dressed ashlar of the wide piers of the east and west arcades of the crypt crossing; of types 2r414, 3f74, 4b386 and 4p10. Mark type 3s280 may be diagnostic in discerning which transept was started first, and in turn, whether the new work began at the north of the building or the south. It does not occur amongst any of the numerous examples on the wide piers in the crypt, nor on the crypt vaulting, yet is almost ubiquities throughout the upper portions of the building and is amongst the few surviving marks in the south crypt transept. Perhaps mason 3s280 joined after work had begun on the north transept exterior walls but before work began on the south transept shortly after. However, the exclusion of just one mark type in an area with so few surviving marks cannot be considered conclusive. A late medieval fire and modern plaster redecorations have destroyed or obscured all marks on the Ithamar Chapel walls except five examples of mark type 3n215 on the east arcade, a ubiquitous mark in all walls of the quire transept and sanctuary. Two marks of type 3s280 survive on the buttress on the exterior of the south wall of the Ithamar Chapel. Although based on the evidence of very few marks surviving on the walls of the crypt transept and Ithamar Chapel, the occurrence of mark types around the crypt can be seen to support previous interpretations that the rebuilding of the quire began at the west end of the new crossing and encompassed the sanctuary in one campaign. At other cathedral sites around this time, old work was left standing for as long as possible before being demolished, with the new work often constructed around the old before this occurs. Excavations from 2014 to 2016 confirmed the basic shape of St John Hope’s reconstructed plan of the late eleventh-century east end (1898, plate iii), with a previously unknown annexe on the north (Keevill and Ward 2019). The old east end was levelled before work could begin on the west arcade of the crypt, and the crypt could be vaulted (Fig. 11). There are burnt fragments of Caen stone re-used in these piers, apparently confirming that the old east end was demolished at this point in time and the material re-used in the ongoing construction. The completion of these piers allowed for the vaulting of the crypt, supporting the possibility that fragments of the demolished east end are reused as infill. The discovery of a painted tufa stone fragment re-used in a vault in the Ithamar Chapel during conservation work in 2012 supports this hypothesis (Perry and Lithgow 2014) (Fig. 12). Mark types 2r414, 4b386 and 4p10 survive on the vaulting in the Ithamar Chapel, matching the wide piers forming its west arcade. Marks survive in only one bay of the vaulting of the crypt crossing, all of them type 2r414. Although confounding attempts to establish a relative chronology for the vaulting of the two areas, they suggest one campaign extending throughout, following immediately on from the dressing of the ashlar for the wide piers. Mark types 4b386, 2r414, 3a282, 4a28 and 6b7 occur on this arcade, which includes the partial casing of the two broad tufa piers of the 1080s. Mark types 2x191 and 2r414 occur on the arches of the barrel vaults between these piers. The upper levels of the quire transept feature far fewer marks than the contiguous ROCHESTER CATHEDRAL MASONS’ MARKS 179 Fig. 12 Stone fragments found re-used in the vaulting in the Ithamar Chapel. Fig. 11 Portion of St John Hope’s (1898) conjectural plan of the old east end of the cathedral, confirmed by 2014-2016 excavations. JACOB H. SCOTT 180 sequence below. This work is more exclusively of Reigate stone which, as noted above, preserves its surface less effectively than Caen stone. Much of this work is in the arches of windows and is only one stone thick, suffering more from variations in temperature and subsequently from efflorescence damage. Despite this, a significant pause between the previous work and these upper portions is apparent. Mark types 2r414, 2d36, 4b386 and 6b7 occur on both the vaulting of the crypt, the work at the east of the blind arcades of the liturgical quire and the wide piers forming the east arcade of the quire transept. This pause may have resulted only from the time required to demolish the old east end and vault the crypt. Several masons’ mark types occur on both the upper portions of the exterior walls, the blind arcades of the liturgical quire and the triforium. Only at this point could the entire east end be vaulted. Much of the stonework for the ribs of the vaulting is far beyond the reach of the 10m vertical access platform used for this survey, although eight mark types occur on the lowest portions abutting the triforium. Though few, these types largely match those of the clerestory. It is possible to interpret this data within the framework provided by the current architectural-historical models for a single building phase construction of the east end. Bishop Gilbert de Glanville (1185-1214) is credited with ‘finishing the quire’ from the proceeds of the shrine of William of Perth, erected shortly after 1201. Further Study Thorough surveys at Rochester Castle, Canterbury Cathedral and churches in the Rochester Diocese considering hand-style, architectural detailing and epigraphic execution of marks may identify where the masons that built Rochester Cathedral had worked previously, where they went to, and if other structures were under construction at the same time. Definitive identification is likely to result from matching sequences of marks, rather than individual examples. Rochester Castle keep, just 80m from the west front of the cathedral, was constructed from 1127-1141, in a campaign that would have overlapped with the cathedral redecoration campaign. At this time Archbishop Corbeil possessed the castle. The architectural details of the keep are also in Caen stone. Just six surviving masons’ marks have been identified within the keep, which has lost its roof and has been greatly affected by weathering in the post-medieval era. Although some similarities in marks between the keep and the cathedral occurs, a comprehensive survey of any further surviving marks is still likely to remain inconclusive based on so few marks. It is of interest that at least a third of mark types at Rochester Cathedral identified by this survey match with those identified during a preliminary survey at Canterbury Cathedral. bibliography Cotton MS. Vespasian A. 22, c.1200, London, British Library, f. 30. Gervase of Canterbury, c.1200, Opera Historica, Rolls Series 73, i. 100. London, British Library. Textus Roffensis, c.1123, Rochester, Cathedral Chapter Library, MS A. 3. 5, f. 172. Alexander, J.S., 2001, ‘The Uses of Masons’ Marks as an Aid to Architectural History’, in J. Higgitt, K. Forsyth and D. Parsons (eds), Roman, Runes and Ogham: Medieval Inscriptions in the Insular World and on the Continent, Donington, pp. 211-222. ROCHESTER CATHEDRAL MASONS’ MARKS 181 Alexander, J.S., 2006, ‘A History of Stonemasons’ Marks and Stone Bonding Methods’, in Andrew Prescott (ed.), Marking Well: essays on the occasion of the 150th anniversary of the Grand Lodge of Mark Master Masons of England and Wales and its districts and lodges overseas, London, pp. 4-62. Alexander, J.S., 2007, ‘The Introduction and Use of Masons’ Marks in Romanesque Buildings in England’, Medieval Archaeology, 51, 63-81. Alexander, J.S., 2008, ‘2: Masons’ Marks and the Working Practices of Medieval Stone Masons’, in P.S. Barnwell and A. Pacey (eds), Who Built Beverley Minster?, Reading: Spire Books, pp. 21-40. Alexander, J., 2013, ‘Signed in Stone’, Country Life, 13 February 2013. Carden and Godfrey Architects, 2009, Rochester Cathedral – Metric Survey, Job No. 3081/21, unpubl. archive report. Devolder, M., 2018, ‘The functions of masons’ marks in the Bronze Age palace at Malia, Crete’, American Journal of Archaeology, vol. 122, no. 3, 343-366. Draper, P., 2006, ‘The Late Twelfth-Century East End of Rochester Cathedral’, in Ayers, T. and T. Tatton-Brown (eds), Medieval Art, Architecture and Archaeology at Rochester, pp. 97-113. Graham, D. and Scott, J.H., 2019, ‘Alphanumeric graffiti at Rochester Cathedral’, Archaeologia Cantiana, cxl, 181-199. Halsey, R., 2006, ‘The Twelfth-Century Nave of Rochester Cathedral’, in Ayers, T. and T. Tatton-Brown (eds), Medieval Art, Architecture and Archaeology at Rochester, pp. 61-84. Hislop, M., 2012, How to build a cathedral, London: Bloomsbury Publishing. Keevill, G. and Ward, A., 2019, ‘Hidden Treasures, Fresh Expressions; Archaeological Surveys, Excavations and Watching Briefs at Rochester Cathedral 2011-2017’, archive report. Martin, T., 2005, ‘Reading the walls: Masons’ marks and the archaeology of architecture at San Isidoro, Leon’, Essays on Medieval Spain in Honor of John Williams, T. Martin and J.A. Harris (eds), Leiden, pp. 373-412. McAleer, P., 1985, Some observations on the building sequence of the nave of Rochester Cathedral, Archaeologia Cantiana, cii, 149-170. McAleer, P., 1996, ‘The Medieval Fabric’, in Yates, N and Welsby, P. (eds), Faith and Fabric; A History of Rochester Cathedral 604-1994, Rochester: Boydell and Brewer, Friends of Rochester Cathedral. McNeill, J., 2006, ‘The East Cloister Range of Rochester Cathedral Priory.’, in Ayers, T. and T. Tatton-Brown (eds), Medieval Art, Architecture and Archaeology at Rochester, pp. 97-113. Moss, R.A., 2000, ‘Medieval jigsaw puzzle; the ancient stones of Christ Church’, Archaeology Ireland, vol. 14, no. 2, 20-23. Münchmeyer, A., 2013, ‘The masons’ marks in the western part of the cathedral of Santiago de Compostela: an approach to its construction history’, Construction History, vol. 28, no. 2, pp. 1-22. Perry, M. and Lithgow, R., 2014, ‘Rochester Cathedral Crypt Medieval Paintings and Plasters – Conservation Project’, The Perry Lithgow Partnership Ltd, archive report. Scott, J.H., 2018, ‘Pictorial and symbolic graffiti at Rochester Cathedral’, Archaeologia Cantiana, cxxxix, 47-74. St John Hope, W.H., 1898, ‘The Architectural History of the Cathedral Church and Monastery of St. Andrew at Rochester; 1. The Cathedral Church’, Archaeologia Cantiana, xxiii, 194-328. St John Hope, W.H., 1900, ‘The Architectural History of the Cathedral Church and Monastery of St. Andrew at Rochester; 2. The Monastery’, Archaeologia Cantiana, xxiv, 1-85. JACOB H. SCOTT 182 Worssam, B., 2000, ‘The building stones of Rochester Cathedral crypt’, Archaeologia Cantiana, cxx, 1-22.