This content downloaded from 160.111.254.17 on Thu, 09 Aug 2018 18:44:30 UTC
All use subject to https://about.jstor.org/terms
 Technical Notes on Michele da Firenze's Virgin and Child:
 Examination, Analysis, and Treatment
 CAROL SNOW, ELENA TOROK, ANIKÖ BEZUR, JENS STENGER,
 AND GWÉNAËLLE KAVICH
 After the Yale University Art Gallery com
 pleted its ambitious renovation in December
 2012—which required reinstallation of over
 four thousand artworks and conservation
 of many—there remained a considerable
 queue of curatorial requests for conserva
 tion, including Michele da Firenze's Virgin
 and Child (fig. i). Once the huge backlog of
 conservation treatments began to clear, con
 servators and conservation scientists could
 turn their attention to examination, analysis,
 and treatment of this painted and gilded
 high-relief terracotta sculpture. Acquired in
 1964, long before the Gallery hired its first
 full-time objects conservator in 2008, the
 sculpture was in such unstable condition that
 it could not be exhibited for decades. In 2013
 Paola DAgostino was hired as the Gallery's
 Nina and Lee Griggs Assistant Curator of
 European Art. With her full support and
 participation, the collaborative conservation
 project moved forward and became part of
 a postgraduate fellowship experience.1 It was
 impossible to know at that time what new
 discoveries would be made during the con
 servation treatment.
 Fig. I. Michele da Firenze, Virgin and Child (before
 conservation treatment), ca. 1435-45. Terracotta with
 traces of polychromy and gilding, 54 x 25Vi in.
 (137.2 x 65.4 cm). Yale University Art Gallery,
 Maitland F. Griggs, b.a. 1896, Fund, 1964.43
 EXAMINATION
 In order to ascertain the condition of the ter
 racotta relief, it was important to understand
 how it was fabricated and how it was used.
 The relief consists of two sections: the upper
 section portrays the Holy Father supported
 by two angels, and the lower section shows
 the Virgin and Child. Both sections contain
 ornate architectural details: columns, an
 arched niche above the Virgin with drapery
 behind her, and over the Holy Father a trian
 gular pediment edged with cherub heads and
 floral scrolls.
 To make the two sections, the artist
 wedged the earthenware clay to remove
 air bubbles and rolled it out to form thick
 slabs. The surfaces may have been carved
 down to create areas of background low
 relief, and then sculptural elements in high
 relief were added. It is quite likely that some
 of these high-relief elements were made in
 molds and were worked further once they
 were attached to the relief with wet clay.
 According to Aldo Galli (see "The Yale
 University Art Gallery's Virgin and Child"
 p. 24), the relief was made from one slab
 cut in two, but it is also possible it was
 made in two sections that were designed to
 fit together so that the whole piece rested
 securely against the wall in the niche or
 above the shelf where it was installed. The
 subject and relatively small size of this type
 35
ig. i. Michele da Firenze, Virgin and Child (before
onservation treatment), ca. 1435-45. Terracotta with
races of polychromy and gilding, 54 x 25Vi in.
(137.2 x 65.4 cm). Yale University Art Galler ,
aitland F. Griggs, b.a. 1896, Fund, 1964.43
This content downloaded from 160.111.254.17 on Thu, 09 Aug 2018 18:44:30 UTC
All use subject to https://about.jstor.org/terms
 Fig. 2. Archival photograph of fig. i, n.d.
 Kunsthistorisches Institut, Florence, 154407
 of relief indicate that it may have been used
 in a side altar or a private chapel.2
 After the two sections were fired in a
 kiln, the terracotta relief was elaborately
 decorated with polychromy and gilding.
 Due to its use as a devotional image, the
 relief would have been touched frequently,
 and thus it was repainted and regilded many
 times. Examination with ultraviolet illumina
 tion suggested that oil was used as a binder
 in the upper paint layers, as well as natural
 varnish and shellac, which may have been
 used for repairs. More information about the
 paint layers is provided in the description of
 analysis later in this essay.
 An undated black-and-white photograph
 of the Virgin and Child (fig. 2), taken before
 the relief was acquired by the Gallery, seems
 to show a better-preserved surface with flesh
 tones appearing distinctly lighter than the
 other colors as compared to what is preserved
 now. The current surface is overall a dull
 grayish brown. It was also apparent that res
 torations for missing areas, such as the lower
 corners, were made after the black-and-white
 photograph was taken. X-radiography of
 the relief was done to penetrate through the
 paint layers and the fired terracotta body. The
 X-ray images revealed repaired cracks in the
 relief as well as creases and irregularities in
 the original working of the clay.
 While the repairs seemed relatively
 stable, the extremely unstable nature of
 much of the surface on the front of the
 relief caused great concern. The layers of
 paint had suffered significant losses, and
 many areas of the terracotta substrate had
 been lost or were on the verge of flaking
 off. Other terracotta reliefs by Michele da
 Firenze show similar or even worse deterio
 ration of the surface.3 This kind of unstable
 condition is often caused by low firing tem
 peratures of the terracotta combined with
 centuries in uncontrolled environments,
 such as private chapels. However, in the
 case of the Gallery's relief, scientific analysis
 uncovered another major cause for the frag
 ile condition of the object: the introduction
 36
 Fig. 2. Ar ival photograph of fig. I, n.d.
 Kunsthist isches Institu , Florence, 154407
This content downloaded from 160.111.254.17 on Thu, 09 Aug 2018 18:44:30 UTC
All use subject to https://about.jstor.org/terms
 of reactive materials as a result of previous
 restoration campaigns.
 Preliminary chemical tests identified
 the presence of chloride and sulfate salts —
 that is, chemical compounds formed from
 the reaction of acids and bases. They are
 known to cause damage when exposed to an
 uncontrolled environment, where changes
 in humidity can cause the salt crystals to
 go in and out of solution, migrate toward
 the surface, and recrystallize at or below the
 surface. Before treatment could begin, it was
 necessary to identify what salts were present
 by performing more sophisticated analytical
 techniques on microsamples, such as X-ray
 powder diffraction and Raman microspec
 troscopy, in addition to optical microscopy.
 ANALYSIS
 Scientific analysis was undertaken by
 the Technical Studies Laboratory at Yale
 University's Institute for the Preservation
 of Cultural Heritage (IPCH) in order to
 characterize the salts, learn more about the
 artist's materials and techniques, and ulti
 mately understand more about why the
 object was in such poor condition.
 First, fourteen samples were taken from
 various parts of the object and prepared
 as cross sections, which make visible the
 layer structure of the surface decoration.
 It was clear through visual examination of
 areas of surface loss that the piece had been
 subjected to multiple campaigns of repaint
 ing in its nearly six-hundred-year history.
 Examination of the samples confirmed this;
 nearly all had a rich stratigraphy of differ
 ent paints, coatings, preparatory layers, and
 other materials. Significant color changes
 were usually not made from one repainting
 campaign to the next. For example, at least
 three different applications of blue paint lay
 ers were discovered in the Virgin's robe, and
 although the pigment composition and par
 ticle size dispersion differ between layers, the
 robe appears to have always been some shade
 of blue. This was also noted in the red back
 ground behind the central figures, where at
 Fig. 3. Cross section of fig. 1 from architectural decora
 tion on the upper proper-right side of the relief, dark
 field visible-light image (left) and ultraviolet-induced
 visible fluorescence (right). At least fourteen distinct
 layers of surface decoration are present in this sample.
 least four different campaigns of red layers
 were present.
 Other areas had undergone more signifi
 cant changes to style and color. For example,
in a sample taken from an area of yellow archi
tectural decoration on the upper proper-right
 side of the top section of the relief, it was noted
 that older layers once contained gilding (fig. 3).
 A sim lar stratigraphy was also observed in the
 yellow wings of an angel nearby. The substitu
 ti n of yellow paint in lieu of gilding may have
 been done because it was less costly.
 Not every area of the object had a mul
 tilayered decorative stratigraphy. A sample
 taken from the Virgin's hand (fig. 4) indicates
 that o ly one paint campaign is present; it
 consists of a single paint layer and a single
 clear coating layer. It is not clear if the layer is
 origin l or  result of a later campaign. Given
 the largely intact condition of these areas, it
 may also be possible that previous layers once
 xisted but were removed before the applica
 tion of this current paint.
 It is un lear how much of the surface
 decoration, if any, is original. Flowever,
 37
Fig. 3. Cross secti n of fig. 1 fr m architectural decora
tion o  the upper proper-right side of the elief, dark
field visible-light image (left) and ultraviolet-induced
vis ble fluorescence ( ight). At least fourteen distinct
layers of urface deco ation are pr sent in his sample.
This content downloaded from 160.111.254.17 on Thu, 09 Aug 2018 18:44:30 UTC
All use subject to https://about.jstor.org/terms
 100 pin
 Fig. 4. Cross section of fig. 1 from the Virgin's hand,
 dark-field visible-light image
 Fig. 5. Cross section of fig. 1 from an area of red back
 ground behind the Virgin's robe, dark-field visible
 light image. Challacolloite salts can be seen as white
 clusters growing in between surface decoration layers.
 pigment particle sizes in paint layers differ
 widely overall, indicating that repainting
 campaigns may have been applied over a
 long span of time and during different eras
 of paint-making technologies. In general,
 pigment particles in paint layers closer to
 the terracotta substrate are much larger and
 more irregularly shaped (consistent with pig
 ments that were hand-ground, as they were
 before the nineteenth century), and pigment
 particles in paint layers closer to the surface
 of the object are much finer and regularly
 shaped (consistent with pigments that were
 machine-ground and/or synthetically pro
 duced, as they most commonly have been in
 more recent history).
 Examination of cross-sectional samples
 indicated that restorations and alterations
 contributed to multiple changes in the over
 all visual appearance of the object over time.
 The large number of layers in some areas sug
 gests that the restoration history of the object
 is long. Differences in the overall number of
 layers from one area to another and changes
 in color in some of the pictorial elements
 suggest that the approaches taken to restoring
 the work were not uniform.
 Following analysis of the stratigraphy
 in the decoration of the object, examination
 of cross-sectional samples shed more light
 on the nature of the dull, gray efflorescence
 observed on the relief's surface. Under mag
 nification, it was clear that salts were not
 only present on the top layer but also form
 ing and crystallizing in situ between layers
 throughout. This phenomenon is particularly
 evident in a sample taken from an area of red
 background behind the Virgin's robe (fig. 5).
 It was evident that the serious deterioration
 and delamination overall was indeed related
 to this highly pervasive formation of salts.
 Further analytical work was undertaken
 to understand more about the decorative lay
 ers and, most importantly, about the nature
 of the problematic salts. Techniques included
 X-ray fluorescence (XRF) spectroscopy,
 Fourier-transform infrared (FTIR) spectro
 scopy, scanning-electron microscopy with
 38
ig. 4. Cross section of fig. 1 rom the Virgins hand,
ark-field visible-light mage
 100 pm
. . ross section of fig. 1 from an area of red back
 e ind the Virgin's robe, dark-field vis ble
i a e. Chal acol oite salts can be seen as white
 ro ing in betwe n surface decoration layers.
This content downloaded from 160.111.254.17 on Thu, 09 Aug 2018 18:44:30 UTC
All use subject to https://about.jstor.org/terms
 energy-dispersive X-ray spectroscopy (SEM
 EDS), Raman microspectroscopy, and X-ray
 powder diffraction (XRD). All work was con
 ducted at IPCH excluding XRD, which was
 performed in the facilities of Yale's Chemical
 and Biophysical Instrumentation Center.
 XRD analysis identified a very unusual
 potassium lead chloride compound, challa
 colloite (KPb Cl ), in addition to gypsum
 (CaSO^HC)), in the gray deposits present
 over the surface decoration. Raman micro
 spectroscopy and SEM-EDS were used to
 detect and map the distribution of challacol
 loite within and between decorative layers
 under the surface. This finding was signifi
 cant, because although mixed salts of potas
 sium and lead and lead chlorides have been
 reported in similar objects, challacolloite has
 never before been documented in or on a
 work of art.4 In nature, this compound has
 been found in Chilean silver mines and in
 volcanoes in Italy and Japan.s In the Virgin
 and Child, the form and distribution of the
 challacolloite grains suggest that this com
 pound formed in situ and was not deposited
 from its environment. The pervasive growth
 of this material is visible in the backscattered
 electron image of the sample in figure 3
 (fig. 6), as well as through the distribution
 of potassium (K), lead (Pb), and chlorine
 (CI) elements mapped with the technique
 of SEM-EDS.
 Although the sources of these ele
 ments and the mechanism of challacolloite
 formation in the Virgin and Child are not
 fully understood, the composition of the
 decorative layers and the terracotta, com
 bined with published literature on salts
 and historic treatment techniques, provides
 important clues. Lead ions appear to have
 come from lead-based pigments (such as lead
 white) in the paint layers, and possibly also
 from external environmental pollutants.6
 Potassium and chloride ions appear to have
 been contributed by external sources, likely
 through the course of previous restoration
 campaigns.7 The use of potash lye, a strong
 base prepared by soaking and boiling plant
 Fig. 6. SEM backscattered electron image of the cross
 section in fig. 3, showing the growth of challacolloite
 salts in between surface decoration layers
 ashes in water, is mentioned in many historic
 recipes for solutions to be used for the clean
 ing of painted surfaces, and such solutions
 would serve as a source of potassium ions.
 Chloride ions, on the other hand, could have
 been introduced through the use of hydro
 chloric acid (also known as muriatic acid), a
 strong acid that has been used historically to
 rid ceramics of disfiguring surface deposits.8
 Although there is no documented informa
 tion about the relief's treatment history
 predating its acquisition by the Gallery in
 1964, it is clear that it has endured multiple
 campaigns of restoration over time that
 have included both structural repairs and
 repainting. It would be logical to presume
 that cleaning occurred during one or more of
 these restoration campaigns.
 The discovery and documentation of
 challacolloite, such an uncommon material
 in a work of art, will undoubtedly serve in
 the future to influence the study of similar
 objects with comparable condition problems
 by conservators and conservation scientists.9
 For the Virgin and Child, the characterization
 of salts and overall technical findings helped
 the conservation team to understand why the
 surface of the object was in such an advanced
 state of deterioration and also contributed to
 treatment discussions.
 39
ig. 6. SEM backs attered electron image of the cr ss
ection in fig. 3, showing the growth of challacolloite
alts in between surface de oration l yers
This content downloaded from 160.111.254.17 on Thu, 09 Aug 2018 18:44:30 UTC
All use subject to https://about.jstor.org/terms
 TREATMENT
 The first and foremost goal for treatment
 of the relief was stabilization, to allow the
 Virgin and Child to be displayed in a verti
 cal orientation. There were no aspirations or
 expectations from the Gallery's curators that
 the general appearance of the work would
 change, and there was certainly no intent by
 the curators and conservators to remove lay
 ers of repainting and regilding. Rather the
 appearance was to remain unchanged, and
 even the damaged surfaces were considered
 an acceptable aesthetic.
 When a porous material such as terra
 cotta has been contaminated with salts, the
 conservator's first inclination is to perform
 a thorough desalination process involving
 rounds of soaking the object in deion
 ized water until all soluble salts have been
 removed. However, before such a treatment
 decision is made, many other factors and
 questions come into play: What is the struc
 tural integrity of the object? How porous is
 it? How soluble are the salts? How far have
 the salts penetrated into the object? What
 other components in the clay body may be
 leached out by soaking in water? How would
 layers of polychromy and gilding be affected
 by soaking in water? In sum, one must weigh
 the risks with the potential for successful
 desalination.
 Treatment options were limited for a
 polychromed terracotta in such a deterio
 rated condition. The surface was so unstable
 that the relief had to be kept in a horizontal
 position before and during treatment to pre
 vent further losses. "The surface could barely
 be touched with a soft brush without risk
 of causing damage. The low-fired terracotta
 had been repaired previously, but the extent
 of structural damage was not completely
 evident. It was also impossible to know how
 evenly fired the terracotta was when the
 object was fabricated. The layers were sepa
 rating at different levels among the gesso,
 paint, and gilding.
 Taking into account all of these factors,
 it was determined that desalination by soak
 ing was not an option. While it would have
 been preferable to remove water-soluble salts
 using this method, the need for desalination
 was thought to be not critical since the relief
 would be exhibited in a stable museum envi
 onment with a relative humidity of 45 to
 50 percent and a temperature of 70T +/- 5°.
 This environment is presumably far more sta
 ble than the small chapel and less-than-ideal
 s orage nvironments where the object previ
 ously resided, and it would slow or arrest the
 formation of additional water-soluble salts in
 the future. Therefore, the decision was made
to ca efully clean the fragile surfaces and then
 stabilize the front of the relief to halt the
 ngoing active loss of surface. The sides and
 reverse of the relief did not have polychromy
 and did not require consolidation.
 Surface cleaning began by using a micro
 attach e t on a high-efficiency particulate
 ar estance (HEPA) vacuum to carefully
 remove debris and dirt from low areas of the
 relief. D located flakes of gesso and paint
 ere collected and saved in plastic vials, with
 a note about the area from which they were
 collected. Further cleaning was done using
 cotton swabs with water and ethanol in
 are s stable enough to allow swabbing of the
 surface. For the stabilization part of the treat
 ment, two different but compatible synthetic
 polymers were chosen as consolidants: one
 for a eas of exposed, friable terracotta, and
the other for the layers of polychromy. The
 treatment achieved the goal of stabilizing the
 relief a d allowed it to be put on display in
 the Euro ean art galleries (see Galli, "The
 Ya e University Art Gallery's Virgin and
 and: fig. 1).
 Through this collaborative technical
 s udy and reatment, a better understand
 ing of past treatments and current condi
 tion allo ed for long-term preservation.
 Restoration to an as-new condition was
 neithe  pos ible nor desirable. Even in its
 current state, the Virgin and Child is an
 exquisite example of polychromed terracotta
 by Mi hele da Firenze.
 40
This content downloaded from 160.111.254.17 on Thu, 09 Aug 2018 18:44:30 UTC
All use subject to https://about.jstor.org/terms
 1. Elena Torok, one of the authors of this article,
 assisted with this project as a conservation fellow and
 has stayed on at the Gallery as a Project Conservator.
 2. A similar relief by Michele da Firenze, Virgin and
 Child Enthroned with Saint John the Baptist and
 Saint James (ca. 1440, Victoria and Albert Museum,
 London), measures 68Vs x 30% x 9% in. (173 x 78.4 x
 24 cm). See also John Charles Robinson, Italian
 Sculpture of the Middle Ages (London: Chapman and
 Hall, 1862), 6; John Pope-Hennessy, Catalogue of
 Italian Sculpture in the Victoria and Albert Museum
 (London: HMSO, 1964), no. 56; Giuseppe Fiocco,
 "Michele da Firenze," Dedalo 12 (1932): 542-63; Lidia
 Righi Guerzoni, "Una Madonna di Michele da Firenze
 nel Modenese," Musei Ferraresi Bolletino Annuale 9-10
 (1979-80): 67-75; Giancarlo Gentilini, "Michele di
 Firenze, San Leonardo," in Da Biduino adAlgardi:
 Pittura e scultura a confronto, ed. Giovanni Romano,
 exh. cat. (Turin, Italy: Allemandi, 1990), 25-38;
 "Inventory of Art Objects Acquired in the Year 1861,"
 in Inventory of the Objects in the Art Division of the
 Museum at South Kensington, Arranged According to the
 Dates of Their Acquisition (London: George E. Eyre
 and William Spottiswoode for HMSO, 1868), 1:19;
 and Eric Maclagan and Margaret Helen Longhurst,
 Catalogue of Italian Sculpture: Text (London: Victoria
 and Albert Museum, 1932), 17.
 3. Massimo Ferretti, "Nota su Michele da Firenze (e
 Nanni di Bartolo)," in Scrittiper I'lstituto Germanico
 di Storia dellArte di Firenze, ed. Cristina Acidini
 Luchinat et al. (Florence: Le Lettere, 1997), 104.
 4. Petria Noble and Annelies van Loon, "Rembrandt's
 Simeon's Song of Praise, 1631: Pictorial Devices in the
 Service of Spatial Illusion," Art Matters: Netherlands
 Technical Studies in Art 4, no. 19 (2007): 20—36;
 Eva Kotulanova et al., "Degradation of Lead-Based
 Pigments by Salt Solutions," Journal of Cultural
 Heritage 10, no. 3 (2009): 367; Annelies van Loon,
 Petria Noble, and Jaap J. Boon, "White Hazes and
 Surface Crusts in Rembrandt's Homer and Related
 Paintings," in ICOM-CC 16th Triennial Conference,
 Lisbon, 19-23 September 2011, preprints, ed. Catherine
 Antomarchi et al. (Lisbon: Critério, 2011), n.p.;
 Annelies van Loon, Petria Noble, and Jaap J. Boon,
 "The Formation of Complex Crusts in Oil Paints
 Containing Lead White and Smalt: Dissolution,
 Depletion, Diffusion, Deposition," in Historical
 Technology, Materials, and Conservation: SEM and
 Microanalysis, ed. Nigel Meeks et al. (London:
 Archetype, 2012), 205-7; Carolina Cardell-Fernândez
 and Carmen Navarrete-Aguilera, "Pigment and
 Plasterwork Analyses of Nasrid Polychromed Lacework
 Stucco in the Alhambra (Granada, Spain)," Studies in
 Conservation 51, no. 3 (2006): 161; David Hradil et al.,
 "Crocoite PbCrO and Mimetite Pb (AsO ) Cl: Rare
4 5 4 3
 Minerals in Highly Degraded Mediaeval Murals in
 Northern Bohemia," Journal of Raman Spectroscopy 45,
 no. 9 (2014): 848.
 5. Jochen Schlüter, Dieter Pohl, and Sergey Britvin,
 "The New Mineral Challacolloite, KPb CI, the Natural
 2 r
 Occurrence of a Technically Known Laser Material,"
 Neues Jahrbuch fur Mineralogie-Abhandlungen 182,
 no. i (2005): 95-101; S. I. Tkachenko et al., "Mineral
 and Ore-Forming Processes from High-Temperature
 Fumarolic Gases at Kudryavyi Volcano, Iturup Island,
 Kuril Archipelago," Geokhimiya 4 (1999): 410; Italo
 Campostrini et al., "Hephaistosite, TlPt^C^, a New
 Thallium Mineral Species from La Fossa Crater,
 Vulcano, Aeolian Islands, Italy," Canadian Mineralogist
 46, no. 3 (2008): 701.
 6. Celia Maqueda, José Luis Pérez Rodriguez,
 and Angel Justo Erbez, "Degradation of Ceramic
 Sculptures on the Cathedral of Seville," Mineralogical
 Petrogrography Acta 29-A (1985): 591; José Luis Pérez
 Rodriguez et al., "Characterization of Decayed
 Ceramic Sculptures Decorating the Pardon Portico
 of Seville Cathedral, Spain," Applied Clay Science 9,
 no. 3 (1994): 211; José Luis Pérez-Rodriguez et al.,
 "Effect of Pollution on Polychromed Ceramic Statues,"
 Atmospheric Environment 32, no. 6 (1998): 993.
 7. Although potassium is present in the terracotta
 substrate and certain pigments, these were deemed not
 to be major contributing sources because potassium
 containing compounds in these areas are either not in
 high enough concentration or not in a soluble form.
 8. Maartje Stols-Witlox, "Historical Restoration
 Recipes: The Cleaning of Oil Paintings 1600-1900," in
 Antomarchi et al., ICOM-CCAbstracts, 16th Triennial
 Conference, Lisbon, 19-23 September 2011, preprints, n.p.
 9. Anikö Bezur, Gwénaëlle Kavich, Jens Stenger, Elena
 Torok, and Carol Snow, "Discovery of Challacolloite,
 an Uncommon Chloride, on a Fifteenth-Century
 Polychrome Terracotta Relief by Michele da Firenze,"
 Applied Physics A 121, no. I (October 2015): 83-93.
 41
This content downloaded from 160.111.254.17 on Thu, 09 Aug 2018 18:44:30 UTC
All use subject to https://about.jstor.org/terms