Madeline H. Caviness


Madeline H. Caviness © 2000


“Tucks and Darts: Adjusting patterns to fit figures for stained glass windows around 1200”

Web Supplement

Print version published in Medieval Fabrications ed. Jane Burns, London: Palgrave, 2004.


This web site was designed with the help of Matthew Lacey to complement the published article.  It should be read in conjunction with the print version, which however contains only a few illustrations in black and white. That text is abbreviated here, without footnotes, but explanations are added for more illustrations (figs. 1-7, 9-11, 15-16, & 18 are not in the published article).

If you cite this work, please do so in full, as above, and include the url:



The standard way to design and execute medieval stained glass was to draw the full-size cartoon on a sized tabletop.  This process was described by a monastic author who dubbed himself « Theophilus » in the twelfth century, and the only extant tabletop with a window design is in the Cathedral of Gerona, where it was used more than once in the fourteenth century.  Such designs showed very clearly the matrix of lead cames that were to join the pieces of colored glass, so that the glasses could be marked for cutting, or even cut, on the rigid working surface.  They also showed sufficient detail – drapery folds, facial features, leaf veins etc. – to guide the draftsmen who were to paint these features on the glass.  The Gerona table demonstrates the versatility of this kind of pattern, in that the architectural canopy was repeated in at least two lancets, whereas the figures under it were changed; this was easily done by sizing over part of the design and drawing new elements.  Colors were noted by letters, and these too could be changed.  When the glaziers had finished with this tabletop, they abandoned it under the eaves of the cathedral. 


The question raised in this paper is what might they have done if they had wished to make replicas of this window at another site?  Transporting large panels is not impossible, but it would be costly. A portable intermediary was needed, to generate the new setting-table design.


In my first book, The Early Stained Glass of Canterbury Cathedral, I published photographs of six ornamental window borders, in each case pairing a design in Canterbury with one in the ambulatory of Sens Cathedral. My observation then, that they are exact replicas one of the other and that therefore the glass painters made full-size patterns of some sort, did not depend on simple measurements, but on rubbings of the lead cames, and notations on the colors used. Now, with the use of a computer, the colors can be added to the scanned rubbings with Adobe Photoshop, and the pairs of rubbings can be superimposed. 


The window borders were typical for the period, consisting of symmetrical multi-colored sprays of foliage on a blue or red ground, interwoven with a white trellis or stems to give the appearance of continuous growth up the sides of the window.  My assumption all along had been that these designs were transmitted through one repeat, just as for modern block prints. The unit size, about 20 x 15 cm., does not preclude the use of parchment, from which the design could be transferred by pricking through to the new setting table on which the glass would be cut, a technique known in manuscript production. 

Even so, the computer revealed slight adaptations, whether to the width of the whole border, or to the length of the repeat, dimensions that would depend on the overall size of the window opening.  Such adjustments could have been made by hand after transfer of the design to the sized table, once the panel size was known.  For the most part the leafy fronds are very similar, indicating a degree of detail in the pattern, but close scrutiny of photographs indicates differences in style between the Canterbury and Sens painters:


Fig. 1a. Canterbury, Christ Church Cathedral, Trinity Chapel, n II: Border to a window with the miracles of Thomas Becket (photo: Caviness).


Fig. 1b. Border of the  Thomas Becket window,  Cathedral of St. Stephen of Sens, ambulatory.

Fig. 1c.  Rubbings of leads in 1a and 1b, colored and superimposed in photoshop.

In another case, one leafy element in the Canterbury border did not appear in the comparable border in Sens. This was not an adaptation for reasons of size, but a simplification that had an impact on the visual richness and on the expenditure of time and materials in cutting and leading more small pieces of glass. The adaptation was in line with a broad trend toward simplification in the ornamental designs used within Canterbury.  I had established this on the basis of the number of shapes that had to be cut for each repeat, thinking as I did so that the glass designers were moving from Vogue to Simplicity patterns:


Fig. 2a.  Border of the Good Samaritan window, Sens ambulatory.


Fig. 2b.  Superimposed rubbings of 2a and of a border made by George Austin Jr. ca. 1860, for a window in the « Corona » of Canterbury, copied from a lost original.


Another pair of Canterbury-Sens designs shows clearly how the repeat pattern could be shortened:


                                                                                          Fig 3.  Sens ambulatory, and a matching 3.  Superimposed rubbings of the border
                                                                                                        of the Saint Eustace window, panel misplaced in Canterbury.

So what was the material used for these portable patterns that the glass painters or their patrons carried with them to use at a different site?  Parchment remains a viable choice for any ornamental details less than about 40cm. in dimension, and it has the advantage of being pliable, strong, and durable.  But parchment was costly and precious. 


An invitation to give a paper on the re-use of cartoons (full-size patterns) in monumental stained glass programs at a conference in Auxerre in 2000 inspired me to see what could be added to existing knowledge by scanning photographs into a computer.  For a series of more-or-less life size figures made around 1200 for the clerestory windows of three widely dispersed buildings, I have been able to settle the question of the material used as a carrier.  Paper of that size was not yet available, so the candidates were wood panels, parchment pieced together, or cloth.  In this paper I explore the ways in which patterns were adapted to different window sizes and proportions, and conclude that cloth was being used in ways that are very well known to seamstresses.



The stained glass made for the clerestory windows of the Abbey Church of Saint-Remi of Reims, for the Premonstratensian Church of Braine, and for the Abbey and Cathedral Church of Canterbury were the subject of my book The Sumptuous Arts at the Royal Abbeys in Reims and Braine, published in 1990.  I proposed on the basis of visual similarities that a single atelier supplied glass for all three sites.  Each time, the members of this atelier had to make large seated figures – almost life-size – to install in pairs, one above the other, in the clerestory lancets:


Fig. 4.  Saint-Remi Abbey Church, Reims, choir clerestory, second bay from the east on the north side.


Fig. 5. Canterbury Cathedral, reconstruction in photo-montage of the Trinity Chapel clerestory window with Amminadab and Naashon (N.X).


Until now, the precise relationship between the figures of the Reims-Braine-Canterbury group has remained elusive, though my definition of an atelier is a group of artisans who shared patterns, including some full-size models. I had argued that such a group worked in the following sequence:  From my study of the choir clerestory of Saint-Remi, constructed and glazed about 1180-82, I concluded that several cartoons had been successively adapted, beginning from the first bay to have been constructed, the westernmost on the north side. Groups of figures are almost identical in their lead matrix (which corresponds to the cut-lines on the tabletop), but sometimes the glass painters had adapted their big drawings, to redesign a head, or to add an attribute such as a crown.  They treated these triple-light window compositions as a unit, so that figures in the smaller lateral lights were paired parenthetically, their heads turned toward the center. The colors in each such pair are identical, as also the outlines of the rest of the body and legs; only the heads were redesigned:


Fig. 6a. Saint-Remi, three archbishops in the choir clerestory made from the same pattern, with changes to the head.


Fig. 6b.  The cutlines for these figures; successive adjustments are in color.


The figures in the larger central light of each triplet are repeated in the opposite window across the choir, but with changed colors.  For the lower figures in all these lights – the archbishops of Reims – I was able to check the accuracy of the photographs taken for the Monuments Historiques on a scale of one tenth against measurements and rubbings I made from the outer sill of the windows. This allowed me to trace a series of adaptations to one cartoon in some detail. For the upper figures, of prophets and apostles, I had limited access from a pair of steps placed on the sill. The need to make rubbings in order to trace design changes placed a limit on archaeological methods before the advent of computer scanning:


Fig. 7.  Saint-Remi, eastern part of the chevet with the author on the clerestory sill to make rubbings (photo: Meredith P. Lillich)


Study of the ancestors of Christ made for the clerestory of Saint-Yved of Braine, of which the provenance of four figures installed in the nineteenth century in the choir clerestory of the Cathedral of Soissons is the most certain, presented the same problems of access as the choir windows of Saint-Remi.  Even with the hydraulic ladder of the Soissons fire department, I could only reach the feet of the second figure from the bottom.  Panels preserved elsewhere, in various collections in the U.S.A., were easier to rub, but these dismembered figures are more heavily restored – only one remains intact.   


Beyond the closeness in style (system of folds, facial types, bodily proportions, and hand gestures) I was not able to discern the exact relationship between Reims and Braine designs.  According to my dating for the completion of construction at Braine, before 1208 (or even 1204) based on documents, and the more recent opinion of Anne Prache that the work had begun about 1190-95, the stained glass of the clerestory should date about 1195-1205/8.


Either in parallel with the glazing campaigns at Braine and at Canterbury, or slightly later, the atelier designed a double tier of seated figures for the clerestory of the nave of Saint-Remi in Reims. For these kings and abbots or archbishops I originally used the same methods (rubbings, measurements and photographs) to demonstrate that certain cartoons had been reused up to seven times within the series, in a rather monotonous manner.  Taken with the style and the building chronology, this indicates a later date than the choir glass, perhaps during the abbacy of Simon (1181-1198) who is known for having decorated this part of the church, or even a bit later. In that case too, losses, distorting repairs and restorations to the glass, and the almost complete reconstruction of the walls following World War I, necessitated the use photo-montage to envisage the original appearance of these windows. Although I could easily make rubbings of the glass from the aisle roof on the north side, the roof on the south side was steeper and less secure.  The task was possible, but it was rudely interrupted by the town police, called out by a nervous parish priest, and I very much regretted not to have profited from a masons’ scaffold that had been in place a decade earlier! 


To that point, it had become clear that a Rémois atelier had been in the habit of re-using cartoons several times within each of four series of large figures (Reims choir, Canterbury, Braine, Reims nave), but in each site the procedure could have been to adapt the design on the setting table, as we know was done in Gerona. But once I re-examined the rather random assortment of rubbings from all four series it began to appear to me that the same designs had been adapted at different sites. The computer study confirmed these suspicions, and supports an evolution of design from the choir of Saint-Remi.


These results were obtained, with the assistance of Heidi Gearhardt, on a personal computer with scanner and Adobe photoshop.  I began by scanning each photograph with a ruler for scale, so that figures of different sizes could be adjusted accurately on the same scale.  The black and white photographs are too detailed to be read clearly if they are superimposed in photoshop, even if they are given transparency, so instead I traced selected lead lines in color onto a separate layer that could then be moved over another photograph – much as one might by making a tracing on transparent paper.  This method demonstrated three kinds of re-use of a pattern, and examples of each type will be illustrated here:

1.  the exact repetition of the main lines of the composition

2.  adaptation to a different size

3.  selection of certain elements only.


As an example of the first type, the main contours of the panel with the head and shoulders of King David, in the fourth bay from the east on the north side of the retro-choir of Saint-Remi, correspond very closely with those of the upper part of Abiud from Braine:


Fig. 8a.  Saint-Remi Abbey Church, Reims, choir clerestory, central lancet in the fourth bay from the east (N.IVb) : upper part of King David.



Fig. 8b.  Braine, upper part of’Abiud, with contours from David superimposed (actually in the Metropolitan museum of Art, New York, Medieval Department #14.47).


Not only do the outlines of their haloes, shoulders, collars, and faces coincide, but also the edging filets to the pointed panels, as shown on the illustration of Abiud. The system of rectangular panes aligned horizontally in the blue grounds is not modular, but the top of the throne back of Abiud does coincide with the bottom of the inscription for David, and the bands with lettering have the same vertical dimension. In fact, the head of the lancet in Braine is a replica of those of the central openings in the choir of Saint-Remi, in the position of the iron armature that defines the panel, and even in the template that served the stonemasons to resolve the pointed arch.


Click here for next page