JAIC 1996, Volume 35, Number 3, Article 1 (pp. 169 to 183)
JAIC online
Journal of the American Institute for Conservation
JAIC 1996, Volume 35, Number 3, Article 1 (pp. 169 to 183)



ABSTRACT—In the course of treating Mountain Beloved of Spring, a painting on canvas by Arthur Bowen Davies, it became apparent that the sky was completely overpainted. No history of previous treatment existed, and evidence gleaned from examining the painting indicated that another paint layer had been applied some time after the painting had been completed. The possibility existed that the artist had repainted the sky, although in its present state the sky did not carry the same characteristics as the rest of the image. Historical records and analysis of the painting were used to develop a probable history, including an early photograph that appeared to show the painting prior to the application of the second layer. The evidence that led to the decision to remove the second paint layer and the subsequent treatment are discussed.

TITRE—Mountain Beloved of Spring d'Arthur B. Davies: Détermination du traitement d'un tableau par reconstitution de son histoire. RÉSUMÉ—Au cours du traitement de Mountain Beloved of Spring, une peinture sur toile d'Arthur Bowen Davies, il apparut que le ciel avait été totalement repeint. Cependant, il n'existe aucun document indiquant une intervention ultérieure, et les données, récoltées au cours de l'examen de l'oeuvre indiquèrent qu'une couche additionnelle de peinture avait été appliquée après que le tableau eÛt été terminé. Il est possible que l'artiste ait lui-même repeint le ciel, bien que ce dernier ne comportasse pas les mêmes caractéristiques que le reste du tableau. Des documents historiques, incluant une photographie montrant le tableau avant l'application de la seconde couche, ainsi que l'analyse des couches de peinture ont servi à retracer l'histoire probable de cette oeuvre. Les données qui menèrent à l'enlèvement de cette seconde couche et au traitement qui suivit sont examinées dans cet article.

TÍTULO—La Amada montaña de primavera de Arthur B. Davis: determinación del tratamiento al reconstruir la historia del cuadro. RESUMEN—En el proceso del tratamiento de Amada montaña de primavera, una pintura sobre tela de Arthur Bowen Davies, se descubrió que el cielo estaba completamente repintado. Ninguna historia de tratamientos previos existía, y evidencias obtenidas al examinar la pintura indicaban que otra capa de pintura había sido aplicada algún tiempo después de que la pintura fue terminada. Existía la posibilidad que el artista hubiera repintado el cielo, aunque en su estado actual el cielo no tenia las mismas características que el resto de la imagen. Registros históricos y análisis de la pintura fueron usados para desarrollar una historia probable de la pintura, incluyendo una fotografía que parecía mostrar la pintura antes de la aplicación de la segunda capa. La evidencia que condujo a la decisión de remover la segunda capa de pintura y el tratamiento subsecuente son discutidos.


In June 1993, the Addison Gallery of American Art, a member of the Williamstown Art Conservation Center, requested an examination and treatment proposal for a painting in its collection. Mountain Beloved of Spring (ca. 1906–7), by Arthur Bowen Davies, was the first painting to be included in the Addison's collection (fig. 1). Influenced by Davies's trip west in 1905, the imagery is probably based in the Rocky Mountains. Davies employed a characteristic horizontal format (18 × 40⅛ in.) to depict a lush green landscape dominated by a middle distant mountain. The dark branches of a large pine tree constitute the foreground, while a snow-capped mountain range is seen in the distance, backed by a blue sky.

Fig. 1. Arthur B. Davies, Mountain Beloved of Spring, ca. 1906-7, oil on canvas, 18 × 40⅛ in., acc.no 1928.1. Collection of the Addison Gallery of American Art, Phillips Academy, Andover, Mass., gift of Miss Lizzie P. Bliss, all rights reserved. Before treatment in 1993

The painting was scheduled for an upcoming exhibition, but cosmetic problems, including what appeared to be a thick grime layer and discolored varnish, disfigured the image. While the painting appeared to be structurally sound, a bulge in the lower left corner indicated some problems with the stretching tension in the canvas. The treatment of the painting was assigned to the author, and the standard examination was carried out.

Techniques used in the examination included study in specular, raking, and ultraviolet light and under low-power magnification, as well as microcleaning tests performed in representative areas of the painting. Materials and methods of construction were estimated, condition was summarized, and a course of treatment was proposed.


The painting was executed on a fine-weight, balanced-weave linen canvas with a thread count of 54 threads/in. in both the horizontal and vertical directions. No selvages were present to distinguish warp and weft. The canvas was secured by tacks to a five-membered stretcher with wooden panel inserts. A second set of tack holes was present in both the tacking edge and the stretcher. A thin, gauzelike material was visible beneath damages in the tacking edge and appeared to be attached to the original canvas with paste or glue. The panels in the stretcher prevented identification of this addition as a full or strip lining. Although the composition extended slightly over the top tacking edge and covered the bottom tacking edge, there was no evidence of other fold lines or tack holes to suggest that the dimensions of the painting had been altered after it had left the artist's hands.

Structurally, the painting was in very good condition. A slight bulge in the lower left corner was caused by an accumulation of debris between the canvas and stretcher and by improper stretching tension. The canvas in the area of the bulge did not seem to be stretched as tightly as the rest of the canvas. However, if the stretcher was sufficiently keyed out to maintain proper canvas tension in the corners, the panels were not large enough to sit adequately in the grooves of the stretcher bars and slipped out of their channels. Perhaps the past need to key out the painting also caused the panels to slip, precipitating the restretching of the painting.

The canvas appeared to be commercially preprimed, and the priming extended evenly to the cut edges. The thinly applied ground did not fill the canvas weave, which remained visible at the unpainted tacking edges and through thinner paint passages. The paint layers appeared to be composed of oil-bound paint varying from multiple thin layers, some of which were rubbed into the canvas, to thicker paint strokes that had low but appreciable body and texture. Microscopic observation and cleaning tests showed no evidence of solvent-sensitive paint layers, and interlayered varnish or resin-bound glazes were not observed. Examination of the paint surface under the microscope indicated some surface abrasion on high points of the canvas and darker pools of varnish in weave interstices, suggesting that the painting had undergone varnish removal or reduction in the past, possibly in conjunction with the structural treatment.

The paint surface was covered with a very thick, yellowed varnish coating composed primarily of natural resin, with a topmost synthetic resin layer. The varnish imparted a dull yellow color to the image, and its surface was crizzled in areas with irregular gloss. In ultraviolet light, the painting autofluoresced yellow-green overall, with some duller areas mottling the surface. These mottled areas were mostly in the sky, corresponding to dull areas seen in natural light, and were presumed to be retouching intermixed with varnish layers. Only a few discrete, nonfluorescent areas, indicative of recent retouching, were apparent. The varnish surface bore multiple layers of grime ranging from a dusty gray grime layer to a yellow-brown layer suggestive of tobacco smoke residue.


The proposed treatment included restretching the canvas in the lower left corner to locally address the planar deformation, removing surface grime and reducing or removing discolored varnish. Cleaning tests indicated that the varnish dissolved readily, without affecting the underlying paint. In areas tested, the removal of grime and reduction of varnish produced a significant improvement in the color and clarity of the underlying paint. It was presumed that retouchings on the surface and intermixed retouchings would be removed with the varnish. Lower pooled residues and more tenacious localized overpaint would then be reduced as possible, with remaining residues cosmetically corrected by retouching.


The treatment, accepted as proposed, was begun following photographic documentation. Overall grime was removed from the varnish surface with naphtha soap (5 parts naphtha emulsified with 2 parts triethanolamine and 5 parts oleic acid in 5 parts water) cleared with xylene. Although not usually used to clear naphtha soap, xylene was chosen in this instance because it also removed the uppermost synthetic varnish, exposing the discolored natural resin varnish. This natural resin varnish was very yellow, thick, and irregularly glossy, as observed in the initial examination. The composition appeared flat, with little distinction among compositional elements, especially in the sky and distant mountains.

At this time, the varnish still appeared to be a final layer applied somewhat evenly over the entire paint surface. The goal of the varnish removal was to slowly reduce the varnish overall, stopping when the disturbing effects of the varnish were satisfactorily diminished. A free solvent of 2-propanol was used, but as the varnish began to be reduced, two distinct zones appeared in the painting: one comprising the blue sky and the distant background mountains and the other consisting of the middle and foreground mountains, trees, and ground. In the sky zone, the natural resin varnish could be dissolved with 2-propanol and removed cleanly, revealing a green-blue paint layer below. In the middle and foregrounds, the varnish could also be dissolved with 2-propanol but appeared to be twice as thick as the varnish in the sky zone. While no paint sensitivities had been found during the cleaning tests, the possibility remained that the artist had used glazes in the middle and foregrounds. Therefore, at this point, the varnish in these areas was only reduced by an amount equal to the varnish that was removed in the sky, as originally planned. On the cleaning swabs, the varnish removed from both areas was a dark, transparent amber color, and it was felt that this represented a final overall varnish layer. This phase of the cleaning did not disrupt any passages in the paint layers.


After the varnish had been reduced over both zones of the painting, some material remained on the sky and background mountains (fig. 2). The surface of the sky at this stage was dirty, discolored, and uniformly flat. It was discordant with the brushwork and paint application in the rest of the painting, and it visually flattened the top third of the image. While the middle and foreground showed great color variety, a range of brushwork, and strong depth and perspective, the sky was lacking all of these. Microscopic examination of the edges of the painting, cracks in the paint surface, and the interface between mountains and sky strongly suggested that the entire sky and the background mountains had been overpainted after the rest of the composition was completed. The uppermost green-blue paint layer overlapped the top edge of the painting, extending very slightly onto the tacking edge but not completely covering a lower, bright blue paint layer. The uppermost paint also formed a hard edge that did not completely match the contours of the middle ground mountains and tree, again revealing a brighter blue paint beneath. In other areas, the green-blue paint overlapped and covered the edges of elements such as the foreground tree. The discolored surface did not respond to cleaning tests with aqueous systems and did not appear to be an embedded grime layer. Solvent tests with acetone and 2-propanol did not directly affect the uppermost layer but did appear to undercut the green-blue paint layer by dissolving a lower resin layer. These observations suggested that the discolored layer was overpaint, possibly applied at the time the structural treatment was undertaken.

Fig. 2. Mountain Beloved of Spring, during treatment. The upper varnish layers have been reduced, exposing the second paint layer in the sky. Some varnish remains on the mountains and foreground. Lighter areas along the top edge and at the top of left edge are cleaning tests revealing the lower blue paint layer.

The preliminary physical evidence and the poor harmony between the sky and the rest of the composition suggested that the present condition of the sky was the result of a previous restoration rather than an artist's change. Unfortunately, no treatment history for the painting could be found. Further examination of the painting was then undertaken in an attempt to find additional evidence of previous treatment. The added gauzelike fabric seemed to be a strong indication that previous restoration had been undertaken. To further examine the structure of the additions and the condition of the original canvas, the painting was removed from the stretcher. The original canvas was in very good condition, with no staining or previous repairs visible. The added fabric proved to be a neatly applied strip lining, which extended very slightly beyond the fold of the painting. Visual examination of the fabric indicated that it was composed of cotton fiber, and the adhesive resembled a dry paste or animal glue. The stretcher face bore a slight ridge in the center of each member, suggesting that it may have been resurfaced. Thin wooden strips had been added to one edge of both stretcher panels with animal glue, presumably to prevent them from slipping out of their grooves when the stretcher was keyed out. While the neatness of the strip lining and the stretcher modifications suggested a restoration effort, no conclusive proof of such an effort was found. The presence of a second set of tack holes and the fact that the strip lining was intact beneath damages in the original canvas supported the theory that the modifications were not done by the artist at the initial creation of the painting. At the top edge, the tack holes from the first stretching were cleaner than the surrounding canvas. The painting most likely remained in the first stretching configuration for some time before it was removed from its stretcher. This observation supported the idea that the structural changes were not immediately done by the artist.


Since the painting as a whole could not offer more information, attention was turned to analysis of the pigments and layer structure of the paint in the sky. Pigment samples were taken from the upper, green-blue paint layer and the lower, bluer paint layer. The samples were taken from cleaned areas near the edge of the painting and mounted on double-sided carbon tape for elemental analysis or in Cargille Meltmount 1.662 for optical analysis. The medium was not removed from the samples prior to mounting. Optical properties of the pigments were observed by polarized light microscopy, and elemental analysis was performed using scanning electron microscopy with energy-dispersive x-ray analysis. The aim of the pigment analysis was to compare the pigments used in both paint layers and look for pigments in the upper paint layer that had been manufactured after Davies's death in 1928, indicating that the paint was not artist-applied (Martin 1995). Specifically, manganese blue and phthalocynanine blue and green would indicate an application date after Davies's death. None of these pigments were found in any of the samples. The lower blue layer appeared to be a rather simple mixture of white (lead, zinc, and barium pigments) and blue (smalt, possibly with cobalt and viridian), while the upper paint layer appeared to be composed of a wider variety of pigments, including zinc and barium pigments, cerulean blue, and possibly some black, yellow, red, and green pigments. Although experience suggests that paint used in retouching often contains a greater variety of pigments than the original paint mixture, the evidence produced was again only circumstantial, and no conclusions could be drawn.

Analysis of a cross section taken from the sky was more fruitful. A sample taken from a crack near the fold edge contained all layers from the ground to remnants of varnish remaining on top of the suspected overpaint layer (fig. 3). In handling, the sample broke into four fragments, which were mounted on a polyester capsule using Norland Optical adhesive #65. The mounted sample fragments were cut with a microtome to expose the complete layer structure and examined in visible and ultraviolet light at 50–500 x.

Fig. 3. Mountain Beloved of Spring, cross section taken from the sky, using simultaneous oblique visible and epifluorescence illumination with an ultraviolet filter. 320 x. Layer (a): ground layer. Layer (b): blue paint layer with a large blue pigment particle (represented by the clear, rounded shape at the left edge.) Layer (c): interlayer varnish. Layer (d): green-blue paint layer. Layer (e): upper varnish layers intermixed with grime

The cross section confirmed the observations made during cleaning tests and microscopic examination. The lower blue paint layer was applied directly over the ground in a smooth single layer, which was slightly thinner than the upper green-blue paint layer. Slight variations in thickness were probably due to the texture of ground and canvas. As observed in the analysis of the crushed pigment samples, this paint layer was composed primarily of white pigment intermixed with a small proportion of large blue pigment particles. In ultraviolet illumination, the paint layer appeared darker than the more autofluorescent ground and adjacent varnish layers. No significant grime accumulation was present on top of the blue paint layer, and the varnish appeared to sit directly on top of the paint with no great degree of intermixing.

The varnish layer directly above the lower blue paint layer appeared very dark brown in visible light and autofluoresced a uniform, bright white in ultraviolet light, indicating it was a single, homogeneous layer. While particulate matter was visible on the upper surface of this layer and intermixed slightly into its surface, there was no evidence that pigments were deliberately mixed into the resin layer to produce a glaze or scumble. Rather, the large, irregular character of the particulate and its uneven distribution (more was located in recessed areas) strongly suggested a grime layer, indicating that the lower blue paint and varnish were exposed for some time before the application of the green-blue paint layer. The upper surface of the interlayer varnish was turbulent, and the interface between the varnish and upper green-blue paint layer appeared indistinct. The varnish surface may have been disrupted by partial cleaning, and the two layers may have intermixed after application of the green-blue paint layer. This particular sample also showed that the resin layer between the two paint layers was not even and appeared pooled in lower areas of the paint. This uneveness may also indicate that the varnish had been reduced (or an unsuccessful attempt had been made to remove it) prior to the application of the second paint layer.

The upper green-blue paint layer presented characteristics very different from those of the lower blue paint layer. In visible light the upper layer was decidedly more yellow, and in ultraviolet light the layer was less autofluorescent than the lower blue layer. The upper paint layer also appeared to be composed of a greater variety of pigments, and large red, blue, white, and yellow particles were mixed with the smaller, yellowish particles that predominated the matrix. Varnish layers on top of the second paint layer included two natural resin layers with interlayer grime and remnants of a topmost synthetic varnish.

The extended examination of the painting and analysis of samples led to the following conclusions: The painting had been removed from its present stretcher in the past; structural work in the form of a strip lining and stretcher modification had occurred; and a second paint layer was added to the sky over an existing varnish layer. No added paint layers were apparent in the other areas of the painting. There was no evidence, such as modern pigments, to date these changes after Davies's death in 1928. Therefore, while it appeared that the second paint layer in the sky was a restoration that occurred at the time of the structural modifications, there was still no proof that it was not done by the artist himself.


To gain a better understanding of Davies's technique and to place Mountain within the context of his work, other paintings by Davies were examined. In addition to many of Davies's watercolors and graphic works, the Addison owns several paintings on canvas. The earliest dated painting in the collection was completed by 1905, shortly before Mountain Beloved of Spring was painted. Three landscapes from 1927 served as excellent examples of Davies's work at the end of his career.

The three paintings from 1927, depicting very different landscape images, show many similarities in technique, including a paint surface in which the canvas texture remains visible, a strong sense of compositional depth, rich coloring, and sure execution, with little obvious evidence of reworking. Apuan Mountains, Lucca (26 × 39 in.) is a somewhat subdued landscape of cattle among green hills with a gray, overcast sky. The paint application is very thin, with the ground or a lower white paint layer visible below brush strokes and thinner color washes. The landscape is primarily composed of multiple thin layers of color in a rich variety of blues, greens, and earth tones. Davies appears to have used a medium-rich paint, and brush strokes retain a distinctly fluid character, especially apparent in thicker accent strokes in the mountains and foreground. The sky appears to have been painted in a single layer of very pale bluish white paint, with a few crisp-edged strokes of blue depicting a break in the cloud cover. In the sky, there is no evidence of reworking, and the sky meets the mountains with clean edges. The painting recently underwent conservation treatment, and no evidence of extensive overpaint or major areas of reworking was recorded. The paint surface presently has a clear, uniform varnish layer.

Umbrian Hills, Near Urbino (18 × 30 in.) is painted in a manner similar to that of Apuan Mountains, but the landscape is brighter, consisting primarily of rolling hills dotted with buildings and livestock, backed with a narrow band of sky. A strong canvas texture is very visible throughout the image. Livestock and buildings are painted primarily with thin color layers, accented by much thicker strokes that at times exhibit a low relief. The blue of the sky is painted in one layer, with the ground or a lower white paint layer used to give definition to the sky at the horizon. The paint surface presently bears a slight layer of grime and discolored varnish, and the painting appears to have undergone some conservation treatment, although no treatment records were found.

In the third painting from 1927, Matese Mountains (26 × 39 in.), the sky is the primary focus of the composition, constituting the majority of the image and stunning for its strong, vibrant blue color. The paint application is very fluid, and Davies's brush strokes remain visible, particularly as he moved the paint around the contours of the mountains. In this painting, it is very clear that the sky is painted in a single layer with no reworking. The remainder of the landscape is painted with techniques similar to the other two paintings. The paint surface is very fresh with a clear varnish coating and appears to have undergone recent conservation treatment.

The 1905 painting, Hills and Valleys (18 × 30 in.), was examined as an example of Davies's technique around the time Mountain was painted. In Hills and Valleys, the landscape is dominated by a figure in the foreground and cattle in the middle ground. While the imagery differs from the other pure landscapes, Davies's technique of using multiple thin paint layers with heavier accent strokes is apparent. However, in this instance, the layering also seems to include areas that have been reworked, and throughout the image, darkened pools of a resinous material can be seen between paint layers. This is particularly evident in the sky, where the edges of clouds are indistinct, and paint and lower varnish layers overlap compositional elements in a haphazard manner. These indistinct edges appeared very similar to the overlapped edges of the trees and mountains in Mountain Beloved of Spring. Also like Mountain, the sky in Hills and Valleys has a dull opaque nature that is not in harmony with the rest of the painting. Records for Hills and Valleys indicate that the painting underwent conservation treatment in 1966, but a treatment report was not found. It is possible that the sky bears a layer of overpaint added during the 1966 treatment and the sky in Mountains underwent a similar but unrecorded treatment. Conversely, it is possible that Davies reworked his earlier paintings more and the skies in both paintings represent his own work rather than that of a later restorer.

The 1927 landscapes and the two earlier paintings showed similar techniques, particularly in the use of a mixture of multiple thin paint layers and thicker accent strokes. However, whether the sure execution and strong coloring seen in the 1927 paintings were also present in Davies's earlier work remained unknown. In the mountains and landscape of Mountain Beloved of Spring, it did seem to be present, particularly in the single feathery brush strokes that defined the pine trees, the delicate color layers in the mountains, and the sense of depth defined by carefully placed, thicker brush strokes. An observation by an art critic of Davies's time may support the idea that the artist's landscape technique remained unchanged throughout his career: “It is difficult to say which is the more beguiling, the obviously accurate portrait of a place or the suggestion of a mood. Late in his career this fidelity to the phenomena of the things seen was almost as pronounced as in his earlier years” (Cortissoz 1931, 11).

Looking at Davies's other work made it seem more unlikely that he had repainted the sky of Mountain. That portion of the composition was least harmonious with the rest of the image, and it bore no relationship to the skies of the later 1927 paintings. The lower blue paint layer, revealed during cleaning tests, did appear to be more like the coloration used in the 1927 skies and more in harmony with the rest of the image. However, while removal of the upper sky layer would probably produce a painting that was more harmonious, the idea that Davies had reworked the sky (with questionable results) remained a possibility that had to be carefully considered before treatment could proceed. The investigation returned to historical research in an effort to create a timeline for the life of the painting.


Although a general chronology of Arthur Bowen Davies is available and his professional activities such as his involvement with the Eight and the Armory Show are well documented, little else is known about this private man. Davies led an incredibly complex double life that included his acknowledged wife and children living on a farm in Congers, New York, and his mistress and their child, with whom Davies lived in New York City under an assumed name. Apart from these lives, Davies maintained his studio and career as well. Few knew about this double life, and those who did closely guarded the secret, to the point of destroying all correspondence with Davies. It is unknown what insights about Davies's working methods and thoughts about his work may have been lost in protecting his private life. Although Davies further muddled his history by dating only a few of his paintings, this point does not greatly affect the history of Mountain Beloved of Spring. The subject can be tied directly to his 1905 trip to the American West, and the painting was probably executed between 1906 and 1907. Its first documented exhibition was at a show of landscapes by American artists held at New York's Union League Club, March 14–16, 1907. In 1910 it was exhibited at the American Arts Exhibition at the Royal Academy of Arts, Berlin, and the Royal Art Society, Munich. Exhibition labels on the stretcher confirm these venues and list the painting as owned by the art dealer William Macbeth. Davies had a close relationship with Macbeth beginning in 1888 and probably had consigned the painting to him shortly after its completion. The painting was subsequently purchased by Lizzie P. Bliss on December 17, 1913, and remained in her collection until she gave it to the Addison in 1928.

It can be argued that this history abounds with opportunities for Davies to have altered Mountain himself, but each point has a strong counterpoint as well. Davies was a personal friend of both Macbeth and Bliss. Around 1895, he began living above the Macbeth galleries and remained close to the dealer for the rest of his life. If Macbeth had requested restoration of Mountain, it could be possible that Davies, being so close, undertook the project himself. However, Davies was financially secure, and it would seem unlikely that he would have accepted such projects for the income. Davies's prolific output during his career would not seem to allow time for restoration work. How Davies felt about restoration of his own work is unknown. He is known to have reworked some unfinished paintings from his early career at a much later date, but such reworkings generally involved figurative works, not pure landscapes. Davies exhibited Mountain shortly after it was painted, an indication that he felt it to be a finished work.

Davies also enjoyed a close relationship with Bliss, the only known private owner of the painting. He advised her on her art collection and painted the murals for her music room. Their friendship raises the slight possibility that Davies modified Mountain to fit Bliss's environment or directed some modification to be done. This question is not easily resolved due to the loss of correspondence between Bliss and Davies. After Bliss gave the painting to the Addison Gallery in 1928, it was shown in several exhibitions, including a memorial exhibition in 1930 and traveling shows in 1960–61 and 1962–63. It is possible that the painting underwent treatment before Bliss donated it or prior to any of these shows. An exhibition condition report from 1962 noted that the surface coating was “irregular in coverage and gloss, discolored, and grimy” (Clapp 1962). This was also the state of the painting in 1994, suggesting that the second paint layer had been applied some time before 1962.


The most direct historical evidence of the previous state of the painting was a Peter Juley photograph now in the Addison Gallery files (fig. 4). Juley photographed the Addison's collection around 1930, and although it is undated, this photograph may be of that set. Macbeth may have also had Juley photograph Mountain while it was in his possession, which means the photograph was taken no earlier than 1906 (or when Davies completed the painting) and no later than 1930 (when the Addison Gallery's collection was photographed). Four other Juley photographs of Davies's paintings in the Addison's files are printed with a process different from that used for Mountain. These other photographs are decidedly yellow in tone, and the images appear to have undergone some fading. The four paintings were all in the Addison's collection at the time Juley photographed it, and their photographs probably represent his work done for the Addison. Because the photograph of Mountain is so different, it was most likely taken at a different time, possibly earlier, when it was in Macbeth's possession.

Fig. 4. Mountain Beloved of Spring, photograph taken by Peter Juley, ca. 1930. Courtesy of the Addison Gallery of American Art

Fig. 5. Mountain Beloved of Spring, after treatment in 1995

The photograph of Mountain appears to be a silver gelatin process made to mimic a platinum print. The paper support is in excellent condition, and the silver image has not faded or discolored. The corners of the photograph, particularly at the bottom, are slightly lighter, but this problem is probably in the original negative rather than the result of later fading of the print (Norris 1995). Compared to the painting, the photograph registers blues, greens, and reds throughout the image, indicating that a panchromatic film was used and the painting was accurately recorded (Evans 1995). The photographic image emphasizes the canvas texture of the painting, probably due to the lighting conditions used during photography.

The most notable aspect of the Juley photograph of Mountain is the range of values seen in the sky. In the photograph, a distinct tonal shift can be seen from left to right in the sky, with the right side of the sky being much lighter. The dark tone of the foreground tree indicates that the lightness in this area is not the result of fading but rather records value differences in the painting. Other distinct differences between the photograph and the painting in its partially cleaned state were observed. In the middle and foregrounds, the photograph shows great depth and tonal variety to the image and the strong texture of the canvas beneath the paint layer.

Even if this texture were heightened in the painting by mimicking the lighting seen in the Juley photograph, the varnish remaining on the painting was thick enough to obscure it. Tonal and spatial relationships in these areas were also slightly distorted by the varnish. Cleaning tests to remove the remaining varnish in the middle and foregrounds indicated that the lower paint layers were intact and the varnish could be removed evenly to the paint surface. The varnish that was removed continued to be a dark, translucent amber in color, as noted in the initial varnish reduction; no sensitive glazes were noted, with the exception of a very few small dark blue accents in a middle-ground tree.

The Juley photograph not only shows a tonal shift across the sky but also indicates the presence of brush strokes and small lighter areas that give the sky definition. In addition, the background mountains appear lighter in relationship to the middle mountains. These differences could not be observed in the painting during treatment because the green-blue upper paint layer was uniformly applied across the sky and distant mountains, producing a flat, even appearance.

While it was felt that the Juley photograph represented an earlier state of the painting before the second paint layer was added, the date of the second paint layer still could not be determined. The other early reproduction of the painting that was found was in the catalog of the 1930 Memorial Exhibition of the Works of Arthur B. Davies held by the Metropolitan Museum of Art (Metropolitan Museum of Art 1930). In the catalog reproduction, the sky appears flat and opaque, similar to the present appearance of the overpainted sky. While at first it would appear that the Metropolitan reproduction and the Juley photograph show two different states of the painting, the many variables of the photographic and printing processes could mean that the Metropolitan reproduction was taken from the Juley photograph and the subtleties of contrast were lost in printing. There was no documentation of the photograph given to the Metropolitan for use in the catalog, and the Juley photograph is the only existing one from that time period that has been found. Once again, circumstantial evidence produced no firm conclusions.


The evidence gleaned from examination, analysis, and research, while offering facets of the history of Mountain Beloved of Spring, did not produce the desired incontrovertible clue for dating the application of the second green-blue paint layer in the sky. The painting had been removed from its stretcher at one time, and some structural work was undertaken. A varnish between the two paint layers in the sky suggested that the upper paint layer was added some time after the image was initially painted. The Juley photograph seems to confirm this. Davies was known to have reworked some of his paintings, and the many interrelationships between principal characters in the handling of the painting provided several opportunities for Davies to have reworked this one himself. However, other evidence can be interpreted to discredit this conclusion. With no conclusive proof, the decision as to how to continue the treatment rested with the curators of the Addison Gallery. After weighing the information, the curatorial committee decided that the Juley photograph documented the painting as it existed at one time without the overpaint. Noting that the overpaint overlapped other compositional elements and was dull, thick, and grimy, they felt “there was little reason to believe that Davies would have chosen to add a heavy paint layer in only one portion of the painting, that he would have chosen to obscure elements of the composition, and that he would have wanted to distort the texture and overall balance of the work.” The committee concluded that the overpaint was probably done by another hand and authorized its removal (Faxon 1995).

The treatment continued by restretching the painting with an added strip lining of light-weight plain-weave polyester multifilament screening fabric attached to the previous strip lining with an EVA-based film adhesive. Because the earlier strip lining had been trimmed to the edge of the original canvas, it was not useful in restretching. The previous modifications to the stretcher panels were left in place, and the painting was restretched with enough tension so that excessive keying of the stretcher (and possible panel slippage) would be avoided. The remaining varnish in the middle and foreground was removed with 2-propanol and acetone used separately as free solvents. No soluble glaze passages were found during this phase of varnish reduction. In the isolated areas of sensitive paint found during cleaning tests, the varnish was minimally reduced to prevent paint disruption. The overpaint and lower varnish and grime residues in the sky and distant mountains were removed with acetone as a free solvent and in a poultice made with a nonwoven poly(propylene) sorbant pad (manufactured by 3M). This procedure served to dissolve the lower varnish, undercutting the overpaint and removing most of the varnish. As overpaint was removed in the sky, some small areas of abrasion became apparent on high points of the canvas weave. The lower varnish appeared to be considerably yellowed with intermixed grime. In the weave interstices, the grimy varnish was pooled, as seen in the cross section. The most tenacious residues of pooled varnish and overpaint were reduced with 1-methyl-2-pyrrolidinone, cleared with Phillips 66 Soltrol 130 (petroleum distillate). Thus the residues could be reduced quickly without exposing the paint to much abrasive rolling from the cotton swabs.

The lower blue paint did not exhibit any significant damage, except for some minor abrasion and isolated dots of intractable overpaint. Following the brush application of an isolating varnish of Acryloid B-72 (approximately 10% in xylene), these damages were inpainted with Magna Colors (pigments ground in n-butyl methacrylate polymer resin). The painting received a final spray application of B-72 (10% in xylene) to saturate the inpainting and adjust final gloss.


As the investigation and treatment of Mountain unfolded, the most telling piece of evidence appeared to be the Juley photograph. While visual examination of the photograph served in the decision-making process, a more quantitative comparison of the value differences in the sky was attempted by measuring the image density across the sky. Using a reflected light densitometer, the image density was measured in each half of the sky, with the tallest mountain serving as the dividing point. The two readings were subtracted, with the resultant number representing the value difference between the two halves of the sky. In theory, a result of zero would indicate no value difference in the sky, while increasingly larger integers would indicate a greater value difference. Thus it was suspected that the before- and during-treatment photographs would have a value near zero, while the Juley and after-treatment photographs would have a larger value difference number.

The image density measurements of the individual photographs did appear to corroborate visual observations. The Juley photograph showed the greatest value difference, and the before- and during-treatment photographs showed almost no value difference. The after-treatment photograph showed a small value difference. Because the conditions and processes used in the Juley photograph were not duplicated for the treatment photographs, no comparisons could be made among the various photographs. Even if all the photographs were taken under the same conditions, unknown parameters, such as the presence of unaged varnish on the painting at the time of the Juley photograph, precluded comparisons.

Looking at the treated painting and the Juley photograph, faint variations in the sky can be seen in the photograph that appear to correspond to variations in the treated painting, but they are much more pronounced in the actual painting. It is possible that the film used for the Juley photograph was not as sensitive to blues as first suspected, and the image was not as accurate as originally thought. A varnish layer on the painting also may have affected the reproduction of values in the sky in the Juley photograph. Even though the treated painting and the Juley photograph do not correspond exactly, the painting overall appears to be more in keeping with the photograph after the removal of the green-blue paint layer, and it does appear likely that the photograph was taken before the application of the second paint layer. On the other hand, the possibility remains that the second paint layer had changed greatly with time, possibly becoming more opaque, and was indeed present in the photograph. Although the photograph appeared to offer a great deal of information, in the final analysis it did not offer much that was conclusive.

After removal of the second paint layer, the spatial unity and overall facile brushwork of the painting were restored. The blue paint layer beneath the overpaint and varnish had been applied with bold brush strokes, some of which reveal the cream-colored ground, creating cloud patterns and depth in the sky. These brushstrokes carefully delineate other compositional elements, particularly the distant mountains.

The pooled varnish residues and areas of abrasion found during overpaint removal do reinforce the suspicion that an attempt was made to clean the painting prior to the application of the green-blue paint layer. It is possible that the painting had undergone restoration that had included the structural work and an attempt to remove varnish in the sky, where darkened varnish would have been most noticeable. Cleaning may have stopped in the sky when it became apparent that the paint layer was being abraded. The middle and foregrounds may not have been cleaned due to fear of disrupting glazes or because a yellowed varnish was not as noticeable there. After partially cleaning the painting, the restorer probably overpainted the sky to cover abrasion and varnish residues, then revarnished the painting. This scenario would account for the double thickness of varnish in the middle and foregrounds and the layering of varnish and overpaint present in the sky area. It should be noted that while the overpaint was very disfiguring at the time of this treatment, it is possible that the overpaint had discolored with age and was not as visually disturbing when first applied.

Taken individually, the pieces of information gathered during analysis and treatment could not confirm that the second paint layer was later restoration. However, the entire body of evidence seems to confirm the hypothetical history constructed for the painting. Future research may uncover more pieces of the history of Mountain, but it seems unlikely that a piece of conclusive evidence will ever be found. The present treatment of the painting has now become part of its history. It is hoped that Davies would have approved of the chosen course.


The author would like to thank Susan Faxon, associate director and curator of paintings, prints, and drawings, Addison Gallery of American Art, and the staff of the Williamstown Art Conservation Center for their assistance and support during the treatment of the painting and in the writing of this article; and Nora Kennedy, conservator of photographs, Metropolitan Museum of Art, for use of her densitometer and her assistance in the analysis of the photographs.


Clapp, A.1962. Inspection condition report of Mountain Beloved of Spring. July 30. Intermuseum Conservation Association, Oberlin, Ohio.

Cortissoz, R.1931. A.B. Davies by Royal Cortissoz. American Artists Series. New York: Whitney Museum of American Art.

Evans, A.1995. Personal communication. Chief Photographer, Clark Art Institute, Williamstown, Mass. 01267.

Faxon, S.C.1995. Personal communication. Associate Director and Curator of Paintings, Prints, and Drawings, Addison Gallery of American Art, Phillips Academy, Andover, Mass. 01810.

Martin, J. S.1995. Analytical report, lab no. 94-TS-187B. March 7. Williamstown Art Conservation Center, Williamstown, Mass.

Metropolitan Museum of Art. 1930. Catalog of a memorial exhibition of the works of Arthur B. Davies. New York: Southworth Press.

Norris, D. H.1995. Personal communication. Associate Professor, University of Delaware, 303 Old College, University of Delaware, Newark, Del. 19716.


Czestochowski, J. S.1979. The works of Arthur B. Davies. Chicago: University of Chicago Press.

Wright, B.1978. The artist and the unicorn: The lives of Arthur B. Davies.New City, N. Y.: Historical Society of Rockland County.


NANCY R. POLLAK graduated from the University of Delaware/Winterthur Museum Art Conservation Program in 1991 with a major in paintings and a special interest in painted textiles. She then joined the Williamstown Art Conservation Center as a postgraduate fellow and continued her work there as an assistant paintings conservator. The treatment of Mountain Beloved of Spring was undertaken at Williamstown in 1993–95. She is currently in private practice as a conservator of paintings and painted textiles. Address: P. O. Box 4141, Frederick, Md. 21705.

Section Index

Copyright © 1996 American Institute for Conservation of Historic and Artistic Works