This article presents an overview of Liu Kang’s (1911–2004) canvas painting supports from his early artistic phases, Paris (1929–1932) and Shanghai (1932–1937). The research was conducted on 55 artworks from the collections of the National Gallery Singapore and Liu family. The technical examination of the paintings was supplemented with archival photographs of the artist at work to elucidate his practice of preparation of painting supports. The analyses conducted with light microscopy, SEM–EDS, and FTIR allowed us to characterise the structure of the canvases and identify the natural fibres and formulation of the grounds. In addition, references to contemporary colourmen catalogues, in relation to certain materials, were made. The obtained results suggest that, in both locations, Liu Kang employed commercially prepared canvases purchased by the roll or by the metre, together with bare strainers or stretchers of standard sizes. In Paris, the artist commonly used linen canvases, while in Shanghai he preferred cotton canvases, with linen used sporadically. The identified grounds from the Paris and Shanghai canvases are white and single-layered, but their formulations vary significantly between the two locations. Hence, grounds composed predominantly of lead white with extenders and drying oil as a binder are considered as exclusive to the Paris phase. However, semi-absorbent or absorbent grounds based on chalk are typical for Shanghai phase. This research contributes to the knowledge of Liu Kang’s painting materials and working practices during the pre-war period in Paris and Shanghai.
Liu Kang (1911–2004) was one of Singapore’s most prominent painters. He was born in Yongchun, Fujian province, China. In December 1928, after graduating from Xinhua Arts Academy in Shanghai, he moved to Paris, where immersed himself in the Western art. He returned to China in 1932 and in the following year he accepted the post of Professor at Shanghai Art Academy, the leading art training institution in China at the time. When the Second Sino-Japanese War (1937–1945) broke out, he moved to Malaya. In 1945, he came to Singapore, where he settled permanently and would contribute much to the development of art here.
The Paris period laid the foundation for Liu Kang’s Western mode of expression while in Shanghai he tried to implement some elements of the Chinese ink technique. However, in Singapore, he recognised the need for the birth of a new style of painting that not only synthesises Eastern and Western art, but also captures the spirit of the tropics. The style come to be known as the Nanyang style, which focused on local scenes and aspects of the regional way of life [1, 2].
Despite Liu Kang’s role in the development of modern art in Singapore, very little is known about his painting techniques and materials. Therefore, the ongoing research has involved an examination of the painting supports representative of his two artistic phases Paris (1929–1932) and Shanghai (1932–1937). The Paris phase also includes works that were painted during trips to Switzerland in 1929 and Belgium in 1930 .
Staying in two important cultural centres of the pre-war period exposed Liu Kang to an overwhelming variety of painting styles and techniques as well as a great range of painting materials from well-established artists’ colourmen and small retailers. The advertisement sections of Le Salon’s 1930 and 1932 exhibition catalogues give an insight into the manufacturers and retailers of art materials who were active in Paris at the time: Lefranc, Bourgeois Ainé, Robert Blanchet, Paris American Art Co., Paul Foinet Fils, J.M. Paillard, Morin & Janet, Merlin Denis, Tochon-Lepage, Sennelier, Tasset et L’Hôte, Hardy-Alan, Toiles A. Binant, Armand Drouant, and A. M. Duroziez (Fig. 1) [4, 5]. In Paris, artists could purchase a raw canvas and prepare it themselves or use commercially prepared canvases. The latter were either ready-stretched on wooden frames, usually of a standard size, or as a roll sold by the metre, which was a cheaper option.
Meanwhile, in China from around 1909, there was a high demand for commercial art, and art schools began to enjoy a boom , leading to an increased interest in art materials among students, and amateur and professional artists. According to T. Tsuruta, painting materials, including canvases from Winsor & Newton (W&N) and Reeves & Sons (R&S), were already available in Shanghai in the first decade of the twentieth century . Advertisements published in the Shanghai Art Academy Graduation Yearbook [8, 9] and two famous Shanghai pictorials, Liangyou  and Arts and Life , reveal that the pre-war Shanghai art materials market offered a selection of local brands like Marie’s and Eagle (Fig. 2a–c), as well as imported materials from Europe and America (Fig. 2d). In addition, local imitations were also available, creating a major challenge for R&S and others (Fig. 2e). Unfortunately, the authors have little information about the types of canvases that were available in pre-war Shanghai.
The study aims to characterise the canvases and grounds used by Liu Kang during his Paris and Shanghai phases. The obtained information can assist in the identification of the painting supports as commercially or artist prepared, giving insight into the artist’s working practice during the discussed periods. The present study is supported with the archival photographs to give a rare glimpse into the artist’s practice of preparation for painting. In addition, some references are made to Lefranc, Bourgeois Ainé, Toiles A. Binant, R&S and W&N colourmen catalogues from the turn of the 19th century and from the period under review, in relation to certain materials encountered in the studied painting supports. Although there were many other suppliers of art materials in France, Britain, and China who were active between 1928 and 1937, the references are restricted to the few aforementioned companies because of the scarcity of early 20th-century catalogues from other retailers.
Materials and methods
The discussion of Liu Kang’s painting supports and grounds is based on the examination of 20 paintings from the National Gallery Singapore (NGS) and 35 paintings from the Liu family collection. None of the examined paintings has the original auxiliary support. All artworks from NGS and seven from the Liu collection are stretched over non-original strainers, stretchers, or boards. The remaining 28 paintings from the Liu family are unstretched. In addition, 16 NGS paintings have undergone treatments, which impacted the examination of the canvases. The treatments included the artist’s commissioned lining of paintings onto plywood board, cardboard, and canvas. Some of them were strip and loose lined by conservators after the NGS accession. Seven NGS paintings and five from Liu family have their tacking margins cut off. The condition of the NGS paintings posed certain limitations for a proper examination of the canvases, while the inclusion of the paintings from the Liu collection significantly expanded the research base. Although the 55 paintings from both collections are not exhaustive, as a few hundred other paintings left in Shanghai perished during the Second Sino-Japanese War  and some others are with private collectors, they represent two artistic phases over a period of eight years and provide a sufficient material for the analyses.
Applied analytical methods
All paintings were photographed and technical data was recorded for each painting, including the dimensions, weave, and density of fabrics and the twist of threads. The density of canvases that provided no access to the reverse side was measured from the front, through the areas of thin paint layer that revealed a prominent texture of the fabric. The tacking margins were checked for the presence of a ground layer and nail holes. Unique features, such as partial ground coverage of the tacking margins or their absence, cusping, paintings on the reverse side of the canvases, underlying paint layers, and penetration of oil from the paint through to the back of the canvases, were documented. Paintings from the NGS collection with an unusual paint texture were also X-ray radiographed (XRR) to verify the presence of underlying paint layers. Then, samples of fibres were extracted from the canvases for the morphologic identification. A sampling of the grounds for the analyses was restricted to the tacking margins or edges of the paint layer from the areas of prior loss. With regard to four artworks painted on the reverse side of earlier compositions and seven stretched and framed paintings from the Liu collection, the standard features were recorded and fibre identification was conducted. The analytical results were studied to determine the presence of characteristic patterns that could be attributed to the examined artistic phases.
The paintings from the NGS collection were digitally X-ray radiographed using a Siemens Ysio Max digital X-ray system with a detector size of 35 × 43 cm and high pixel resolution of over 7 million pixels in the detector face. The X-ray tube operated at 40 kV and 0.5–2 mAs. The images were first processed with an X-ray medical imaging software, iQ-LITE, then exported to Adobe Photoshop CC for final alignment and merging.
Optical microscopy (OM) of samples was carried out in visible and ultraviolet reflected light on a Leica DMRX microscope at magnifications of × 40, × 100, and × 200 equipped with a Leica DFC295 digital camera.
High-resolution digital microscopy
The surface of the canvases was examined with a Keyence VHX-6000 digital microscope, using a zoom lens coupled with a high-speed camera. Observations were conducted at magnifications of × 20– × 200. For measurement analyses, a built-in Keyence software—VHX-H2M2 and VHX-H4M—was used.
Scanning electron microscope with energy dispersive spectroscopy
The cross-sections of the ground and paint samples containing a complete structure of layers were mounted on carbon tapes and examined with a Hitachi SU5000 field emission scanning electron microscope (FE-SEM) coupled with Bruker XFlash® 6/60 energy dispersive X-ray spectroscopy (EDS). The SEM, backscattered electron mode (BSE), was used in 60 Pa vacuum, with 20 kV beam acceleration, at 50–60 intensity spot and a working distance of 10 mm. Results were processed using the Bruker ESPIRIT 2.0 software.
Fourier transform infrared spectroscopy
Attenuated total reflectance-Fourier transform infrared spectroscopy (ATR-FTIR) was carried out using a Bruker Hyperion 3000 FTIR microscope with a mid-band MCT detector, coupled to a Vertex 80 FTIR spectrometer and Bruker ALPHA FTIR spectrometer. Measurements were carried out at room temperature in the spectral range of 4000–600 cm−1, at a resolution of 4 cm−1, averaging 64 scans. The interpretation of spectra relied on Bruker Opus 7.5 software.
Preparation of samples
A total of 104 micro samples of fibres were collected from the threads of weft and warp for the microscopic examination. The samples were first boiled in water to remove contaminants and then mounted on microscope slides with a drop of water under the cover glass. In addition, 45 micro-samples of the ground and paint for cross-section analyses were embedded in a fast-curing acrylic resin, ClaroCit (supplied by Struers), and fine polished. The samples from all NGS paintings and the most representative ones from the Liu family collection, in all totalling 28 samples, were selected for the FTIR measurements.
Results and discussion
Auxiliary supports and canvas formats
Although Liu Kang’s original auxiliary supports are not preserved, a photograph of the artist taken during the painting session in Saint-Gingolph, Switzerland, in 1929, shows his painting attached to a strainer with a cross-member (Fig. 3a, b). The strainer and canvas do not have a visible manufacturer or retailer stamp or label. Liu Kang’s Self-portrait in Paris and Self-portrait, both from 1931, also document the auxiliary supports with cross-members and without keys (Fig. 3c, d). As these findings are rare examples of the artist’s auxiliary supports, they cannot demonstrate his preference. However, Lefranc’s price list from 1924 suggests that stretchers cost approximately twice as much as strainers; therefore, the artist may have opted for strainers as they were cheaper.
The assessment of the format of the paintings was done with reference to contemporary French standards, which became common even beyond France. According to the Lefranc catalogue from 1930, three rectangular formats (portrait, landscape, and marine) were offered in 19 standard sizes, from 22 × 16 cm (number 1) to 195 × 130 cm (number 120). The standard sizes applied to both strainers (chassis ordinaire) and keyed stretchers (chassis à clés), but without the indication of the cross-members (Fig. 4).
The majority of paintings from Liu Kang’s Paris phase were created in plein air (18 paintings), revealing the influences of Impressionists, Post-Impressionists, and Fauvists advocating for plein air and modern life painting . Other subjects that he explored were portraits (seven paintings) and still lifes (four paintings). The dimensions of all examined paintings can be considered to be standard; however, about 1 cm margin of difference from the standard may have resulted from subsequent treatments. It could be speculated that small painting supports (number 10 and 8) were more convenient to carry around on outdoor painting sessions (Figs. 3a, b, 14a–c). Larger painting supports (number 12 and 20), were used sparingly. Financial constraints could also have had an impact on the artist’s choice of materials . Therefore, it is worth noting that 11 paintings were created over rejected compositions and four on the reverse side of earlier paintings. A comparison of the paintings’ genres with their formats revealed that a total of 16 landscapes and streetscapes were painted in the portrait format. This indicates that the artist did not follow common rules when choosing painting support formats (Table 1).
Regarding the type of auxiliary supports used by Liu Kang in Shanghai, two archival photographs from 1933 are the only evidence providing limited information (Fig. 3e, f). The said photographs captured the reverse sides of his paintings, which showed that they were probably stretched over unbranded strainers. Liu Kang’s passion for outdoor painting intensified in Shanghai—out of 26 examined artworks, 24 were in plein air. Their dimensions conformed to common standards. The range of recorded sizes was 10–20. The artist frequently used the portrait format number 10 for outdoor paintings, reflecting his preference for handy painting supports, which we know from the Paris practice. He also bent rules to his preferences and 18 landscapes were painted in the portrait format. It is noticeable that he also employed larger canvases of sizes 15 and 20 with more confidence. Like in Paris, Liu Kang occasionally reused unsuccessful compositions. Of the examined paintings, 13 were painted over earlier, rejected artworks (Table 2).
All the examined Paris paintings were executed on linen canvases made in plain weave with Z-twisted threads of weft and warp. Two types of canvases were identified (Table 3). Type 1 is a low-density canvas with a thread count in the range of 12–13 × 13–14 per cm (Fig. 5a); identified in 20 paintings. Type 2 is a denser canvas with a thread count in the range of 18–19 × 20–21 per cm (Fig. 5b); it was found in nine paintings. Based on the collected data, it appears that the artist’s choice of canvas densities was not directed by the size of the planned painting. That can be exemplified by Boat near the cliff (1931), measuring 53.7 × 72.4 cm, which was painted on a low-density canvas, while Autumn colours (1930), measuring 38.3 × 45.3 cm, was painted on a dense canvas. Although the density of the canvases may vary from one supplier to another, the naming convention for the canvases was consistent across the trade at the turn of the 19th century and reflected the quality and proposed function of the material. The comparison of the features of the examined canvases with those of Binant from 1889 to 1898 , Bourgeois Ainé from the post-1906 period , and Lefranc from 1927 to 1934 (Fig. 6a) assisted in the preliminary identification of canvas types chosen by Liu Kang. Hence, thin canvases could be equivalent to étude or pochade grades, whereas denser canvases could be comparable to demi-fine or fine. As reported by A. Callen, cheap étude or pochade canvases exhibit excessive weave distortions—primary cusping—during the initial stretching prior to the ground application . Such distortions were observed on four thin and two denser canvases.
The lack of a manufacturer’s or retailer’s stamp on the examined Paris canvases could mean that they were sold in lengths, by the metre, or purchased from small-scale retailers who did not brand the canvases. According to the 1930 Lefranc catalogue, bare and primed canvases were offered in rolls 10 × 2 m and 5 × 2 m as well as per square metre. The company also offered canvases prepared for decorative painting in rolls from 2 to 8 m wide and of unlimited length (Fig. 6a, b). As the art materials market in Paris was competitive, it is possible that other retailers offered similar items at more affordable prices.
As for the paintings executed in Shanghai, two types of painting supports were identified (Table 4). Type 1 comprises a dense, plain weave, cotton canvas, identified in 19 paintings; this type is characterised by a notable inconsistency of thread count in the range of 15–17 × 18–20 per cm (Fig. 5c), probably caused by uneven stretching forces before the ground application. Primary cusping, reflecting a susceptibility of the canvas to distortions, was observed in 10 paintings. Interestingly, the reverse side of Rustic landscape (1934) bears three similar stamps containing traditional Chinese characters. Although the poor print renders the top two characters almost illegible, it was possible to unravel the remaining six characters, as follows: 用 (yòng), 品 (pǐn), 商 (shāng), 店 (diàn), 經 (jīng), 售 (shòu) (Fig. 7a, b). Considering that Chinese words often consisted of characters appearing in pairs, a probable translation of the pairs of characters could be: “supplies (用品), shop (商店), sale (經售)”, which together mean “sold by supplies shop”. Hence, it is clear that the stamps correspond to a retailer of art materials. The first two characters might refer to the retailer, or they might be the family or place name or a generic word indicating the type of “supplies”.
Type 2 comprises a low density, plain weave, linen canvas with a thread count approximating 10 × 10 per cm (Fig. 5d); this canvas type was identified in five paintings. The threads of weft and warp are characterised by an S-twist, suggesting a different manufacturing process. Primary cusping, was observed in three paintings. Exceptionally, the canvases for Still life with green stool (1933) and Backyard (1934) were made of linen, corresponding to Paris type 2. The collected data does not show a correlation between the sizes and thickness of the canvases of both types, suggesting flexibility in the artist’s choices.
According to the advertisements of art suppliers, local and imported painting materials were readily available at dedicated shops or major stationery and bookstores in Shanghai (Fig. 2). However, our knowledge about the types of canvases available then remains limited. Western brands, such as W&N and R&S, were popular in pre-war Shanghai . Therefore, it can be assumed that products available in Britain were also offered in China. The R&S catalogue from 1926 listed both linen and cotton canvases, while the W&N catalogue from 1934 offered “artists’ prepared canvas”, such as linen, jute, cotton, mixture of hemp and cotton in rolls, stretched over wedged and bevelled stretchers of standard sizes (Fig. 8a, b) . Cotton canvases were preferred by Liu Kang. Considering that his teaching salary at Shanghai Art Academy was modest [6, 18], it can be deduced that Liu Kang preferred the more affordable local materials over the imported ones.
Characteristics of the grounds
Of the 29 examined Paris paintings, 19 have intact tacking margins with the ground layer on, suggesting a commercial preparation. The remaining 10 paintings have either cut-off tacking margins or were created on the unprepared reverse side of earlier compositions. Interestingly, five paintings created between 1929 and 1930, showed a partial ground coverage of tacking margins (Fig. 9, Table 5). It can be assumed that these canvases were part of long and wide pieces and the unprimed areas could have been used for mounting on a large frame for the commercial application of a ground.
Based on the acquired data, all grounds are white and single-layered, partially revealing the canvas weave. The variations recorded in the compositions allowed to determine five different types of grounds (Table 5).
The ground of type 1 was identified in 10 paintings, including—as mentioned earlier—five canvases with a partial ground coverage of the tacking margins. It is a mixture of clumps of lead white (lead carbonate) and fine particles of chalk (calcium carbonate) with minor or trace amounts of zinc white (zinc oxide) (Fig. 10a, b). Literature sources indicate that lead white was available in various quality grades, usually modified due to its yellow hue or to reduce the manufacturing cost. Barium white (barium sulphate), zinc white, gypsum (calcium sulphate), and chalk were commonly used to modify or adulterate lead white [16, 19, 20]. For the binding medium, a drying oil was confirmed with FTIR by peaks at 2920, 2850, 1735, 1456, 1168, 721 (± 5) cm−1. However, the ground sample from St Gingolph, Lac Leman, Switzerland (1929) also displayed peaks indicative of proteins at 1630 and 1539 cm−1. On one hand, this finding could suggest that the ground is a semi-absorbent emulsion of animal glue and oil binders and should be classified as a separate type. On the other hand, this could be an oil-based ground that absorbed proteins from the sized canvas ; however, this presumption requires FTIR analyses of more samples extracted from the painting. Additionally, a presence of lead soap formation was confirmed in two examined ground samples with FTIR by peaks at 2955, 2873, 1540 and 1515 cm−1 (Fig. 11a) [22,23,24].
The ground of type 2 was found in three paintings, and the elements identified are attributable to lead white extended with barium, zinc, and titanium (titanium dioxide) whites. The structure of these grounds is characterised by large, rectangular particles of barium white and clusters of lead white (Fig. 10c, d). The detection of Ti conforms to the introduction of titanium white by Bourgeois Ainé in 1925 and Lefranc in 1927 . FTIR confirmed oil by absorption peaks at around 2919, 2849, 1737, 1460, 1243, 1172 cm−1 and suggests a concomitant presence of proteins. However, while proteins were confirmed in one examined ground sample, their identification in the remaining two samples was inconclusive due to overlapping peaks of zinc soap at 1540 cm−1 (Fig. 11b) [26,27,28]. Nevertheless, it is difficult to determine if the examined type of grounds is semi-absorbent emulsion of animal glue and oil binders or absorbent ground bound with animal glue as the oil may have originated from the artist’s paint.
The ground of type 3 contains lead white extended with barium white and chalk. A concomitant presence of Sr and S in the analysed samples suggest strontium sulphate, a common impurity of barium white [29, 30]. This ground composition was identified in two paintings. The optical microscopy and SEM-BSE images of this type of grounds show a non-homogenous structure containing poorly ground particles of barium white and clumps of lead white (Fig. 10e, f). A drying oil was used as a binder, and lead soap was confirmed by peaks at 1515 and 1540 cm−1 in one sample.
The ground of type 4 was identified in two paintings and contains large and roughly ground chalk particles as the main component, mixed with lead and zinc whites (Fig. 10g, h). The grounds contain oil detected in Man in blue coat (1930) by peaks at 2917, 2849, 1734, 1456, 1233, 1160, 1103 cm−1 and in Zuo La Lu (1930) by peaks at 2952, 2918, 2849, 1728, 1151 cm−1. A concomitant presence of proteins was confirmed in the ground from Man in blue coat by peaks at 1636 and 1533 cm−1. The presence of proteins in the ground from Zuo La Lu remains inconclusive due to an overlapping band of zinc soap at 1538 cm−1.
The ground of type 5, also detected in two paintings, is composed of small particles of chalk well mixed with lithopone (mixture of zinc sulphide and barium sulphate) and/or barium white and zinc white, as well as lead and titanium whites (Fig. 10i, j). A presence of drying oil was detected by peaks at 2923, 2853, 1734, 1456, 1168, 1097, 721 cm−1 and the presence of proteins was detected by peaks at 1646 and 1527 cm−1 in one available ground sample. The formation of zinc soap was detected by absorption peak at 1539 cm−1 (Fig. 12c).
The obtained results are not conclusive about whether the grounds of types 4 and 5 are semi-absorbent or absorbent as it is possible that the oil might also have come from the paint, as clearly seen on the reverse side of My landlady, Madame Normand (1932). The figure outline in thinned paint penetrated through to the back of the canvas (Fig. 12a, b).
Commercially prepared canvases with varied degrees of ground absorbency were available for artists from Lefranc, among others, in 1930 (Fig. 6a). The main advantage of semi-absorbent and absorbent grounds is the ability to absorb the oil from the paint and to accelerate drying, a feature especially desired during plein air painting and rapid sketching. These types of ground were also considered for producing a bright, rough paint surface that scattered light and was less prone to darkening. Their use by Liu Kang would be reasonable as most of his paintings were created outdoors; however, a comparison of the grounds with the subject matter did not reveal any noticeable correlation. It is unclear whether the artist deliberately selected these grounds for his work to take advantage of their properties.
A comparison of the grounds’ formulation with the density of the canvases revealed a distinct correlation between them. The grounds of type 1 and 4 were identified on thin linen fabrics. However, the former makes up the most often used painting support. This suggests that the artist purchased them in larger quantities. In contrast, grounds of type 2 and 3 were found on thick linen fabrics. Only grounds of type 5 were found on fabrics of both densities.
The evidence collected from 15 paintings created in Paris confirmed that artist did not mind reusing his earlier compositions or utilising their reverse sides if he was short of painting supports. These paintings were created without the application of an intermediate ground layer . Hence, it could be possible that Liu Kang had seen a few, rare, examples of such an approach among Impressionist and Neo-impressionist painters , whose works influenced his own . Moreover, an investigation of the canvases and grounds of paintings created in Switzerland—St Gingolph, Lac Leman, Switzerland (1929), and Landscape in Switzerland (1930)—and Belgium—Portrait of a man with his hat (1930)—revealed identical features to the canvases and grounds used in Paris. This indicates that artist brought painting supports from Paris for his painting sessions in Switzerland and Belgium.
As for the paintings from Shanghai, 22 examined artworks out of the 26 show preserved tacking margins with a white ground indicating a commercial preparation. All examined grounds are single-layered and their composition is different from those prepared in Paris.
SEM–EDS and FTIR analyses allowed us to identify two types of grounds (Table 6). Type 1, detected in 23 paintings, contains a high concentration of chalk with admixtures of lithophone and/or barium white and zinc white, with trace amounts of titanium white and/or lead white. The structure of this type of ground is characterised by large and poorly ground particles of chalk (Fig. 13a, b). A concomitant presence of drying oil and proteins was detected in five examined ground samples. The proteins were either identified by bands at around 1635 cm−1 or 1536 cm−1. However, the latter band was challenging to interpret due to an overlapping band of zinc soap at 1540 cm−1 and the presence of a wide band of chalk at 1400 cm−1. In the remaining paintings, oil was prevalent while the identification of proteins was inconclusive due to absent or weak absorption peaks at 1636 cm−1 and an overlapping absorption band of zinc soap. This leads to the conclusion that the examined group of grounds could be semi-absorbent or absorbent, with oil originating from the artist’s paint. Dark outlines observed on the reverse side of three paintings seem to support this notion.
The ground of type 2 was identified in one painting. It is oil-based, composed of lithopone and/or barium white with zinc white and trace amounts of lead white. Its structure is homogenous with small and well-mixed ingredients (Fig. 13c, d). A distinctive absorption peak at 1539 cm−1 confirmed the formation of zinc soaps.
In Still life with green stool (1933) and Backyard (1934), the canvas structure and ground formulation are consistent with the Paris painting supports, suggesting that artist reused earlier artworks created in France.
Both R&S and W&N, which were active at that time in Shanghai, did not distinguish between absorbent and non-absorbent canvases. However, we know from the catalogues that R&S offered “single primed” linen and cotton canvases in 1926 (Fig. 8a). W&N listed “single” and “full primed” canvases in their 1934 London catalogue. The latter was usually composed of three coats of ground . A comparison of the composition of the grounds with the density of the canvases revealed that the ground of type 1 is present on both cotton and linen canvases. However, its predominance on cotton canvases suggests a bulk purchase by the artist. The ground of type 2 was found on one cotton canvas. The evidence collected from 13 Shanghai paintings suggests that Liu Kang continued his practice of reusing earlier canvases without the application of an intermediate ground layer .
Formation of metal soaps
As illustrated in the Tables 5 and 6, the pervasive formation of metal soaps has been detected in many of the analysed ground samples. Their presence is probably due to the reaction of metals in the lead- and zinc-containing pigments, with free fatty acids from the oil binder. However, their formation may be accelerated by the quality of the ingredient materials as well as the environmental condition . Coarse pigment particles used for the grounds enhanced their porosity, leading to an increased absorption of more oil from the paints. In addition, the exposure of the paintings to the hot and humid tropical climate of Singapore could also have played an important role in the development of the soap formations [32,33,34]. Future care and conservation treatments of the paintings should minimise the risk of their exposure to high relative humidity and take into consideration a negative impact of heat and moisture during the lining, consolidation and aqueous cleaning procedures [33, 35].
Stretching practice and role of auxiliary supports
The collected archival evidence seems to support the notion that in both locations, Liu Kang used commercially prepared canvas purchased by the roll or by the metre that he later cut to the required size. The examined works and archival photographs show that the artist fastened the canvas to the sides of the auxiliary supports (strainers or stretchers), probably with nails. The crudely cut edges and excess of the unfolded material in the corners reflect the transitional role of auxiliary supports (Fig. 14a–d). His unstretched paintings, seen on the walls of his rented accommodation in Paris (Fig. 14e, f) provide additional evidence that the auxiliary supports were reused continuously, suggesting that this practice could have been motivated by financial constraints.
The cross-referencing of archival and technical research provided insights into the type of painting supports used by Liu Kang in pre-war Paris and Shanghai. The archival photographs proved to be invaluable as they gave a rare glimpse into the artist’s practice. The contemporary colourmen catalogues and advertisements greatly assisted with understanding the industry trends and availability of materials, and provided a point of reference for an in-depth characterisation of the artist’s canvases. Unfortunately, mainly due to the language barrier, only a limited number of Chinese sources were accessible to support this study.
Based on the collected analytical data, it appears that, in both locations, the artist relied on small-scale retailers who supplied him with commercially prepared canvases. Liu Kang probably purchased the canvas by the roll or by the metre, together with bare strainers or stretchers, as these were affordable painting supports. Thus, his role in the preparation of the painting supports was limited only to cutting the canvas to the required size and mounting it on the auxiliary support, which played a transitional role. The archival photographs provide evidence that Liu Kang was equipped only with a few strainers or stretchers of different sizes, which he continuously reused. As for the auxiliary supports, the portrait format numbers 8 and 10 in Paris, as well as number 10 in Shanghai, were his most favoured, probably because they were more portable. However, it is clearly evident that the artist used larger canvases—of numbers 15 and 20—with greater confidence in Shanghai. In addition, the study shows that, in both locations, the artist ignored commonly accepted format-subject matter rules for his paintings. The investigation of the canvases did not reveal the names of the Paris and Shanghai retailers; however, the stamps on the reverse side of Rustic landscape (1934) correspond to a Chinese retailer.
Furthermore, the analyses revealed a difference in the characteristics of the fabrics and grounds of the Paris and Shanghai paintings. The artist’s choice of painting supports used in Paris consisted of linen canvases of various densities. Based on the comparative studies with other French canvases reported in the literature and listed in contemporary catalogues, the low-density linen used in Liu Kang’s canvases was preliminarily identified as étude or pochade grades, whereas the denser linen could be comparable to demi-fine or fine grades. A detection of variations in the composition of the grounds allowed at least five different types of grounds to be distinguished. The single-layered grounds composed predominantly of lead white with extenders and drying oil as a binder are considered as exclusive to the Paris phase. The absorbent grounds with natural glue or semi-absorbent grounds with natural glue and oil as a binder were chosen by the artist less frequently. It also appears that partial ground coverage of the tacking margins is a feature unique to the Paris phase. In Shanghai, Liu Kang displayed a preference for cotton canvases, probably for their affordability. Linen canvases were used sporadically. The majority of the examined grounds are single-layered and complex mixtures with chalk as a principal component. A concomitant presence of oil and proteins in the examined grounds suggests that the majority of the supports could be semi-absorbent or absorbent.
It was noticed that the artist painted over rejected compositions or on the reverse sides of his earlier artworks. None of these recycled supports were coated with an intermediate ground layer, confirming that the artist relied on commercially prepared canvases and did not prepare his own grounds.
The results obtained from this study may aid in preliminary dating of undated artworks or in determining the provenance of Liu Kang’s painting supports, as in the cases of Still life with green stool (1933) and Backyard (1934). The collected data can also be used in comparative studies with other commercially or artist prepared canvases of the same period and provenance. A notable presence of lead and zinc soaps detected during this study may be useful information for future conservation diagnostics and treatment. Further research on Liu Kang’s painting supports might include a longer period of time to determine whether the artist’s painting preparation practice underwent any fundamental evolution.
Availability of data and materials
The datasets used and/or analysed during the current study are available from the authors upon request.
National Gallery Singapore
Field emission scanning electron microscope
Attenuated total reflectance-Fourier transform infrared spectroscopy
Winsor & Newton
Reeves & Sons
Ong ZH, Wong S, Ong P, Tng P. Life of love, life of art. In: Yeo WW, editor. Liu Kang: colourful modernist. Singapore: National Art Gallery; 2011. p. 22–83.
Ravaud E, Eveno M, Calligaro T, Gonzales V, Wallez G, Nowik W, Aufort J. Study of a pre-primed canvas bundle from the Binant color merchant. In: Bridgland J, editor. Preprints ICOM-CC 17th Triennial conference, Building strong culture through conservation, 17–9 September 2014, Melbourne, Australia. Paris: International Council of Museums; art. 0102.
Callen A. The art of impressionism: painting technique & the making of modernity. Yale University Press, New Haven and London; 2000. p. 31–35, 67, 69–70.
Carlyle L. The artist’s assistant: oil painting instruction manuals and handbooks in Britain 1800–1900: with reference to selected eighteenth-century sources. London: Archetype; 2001. p. 515–517. https://doi.org/10.1179/sic.2002.47.2.139.
Otero V, Sanches D, Montagner C, Vilarigues M, Carlyle L, Lopes JA, et al. Characterisation of metal carboxylates by Raman and infrared spectroscopy in works of art. J Raman Spectrosc. 2014;45(11–12):1197–206. https://doi.org/10.1002/jrs.4520.
Salvadó N, Butí S, Pradell T, Beltran V, Cinque G, Juanhuix J. Identification and distribution of metal soaps and oxalates in oil and tempera paint layers in fifteenth-century altarpieces using synchrotron radiation techniques. In: Casadio F, Keune K, Noble P, Van Loon A, Hendriks E, Centeno SA, et al., editors. Metal soaps in art: conservation and research. Cham: Springer International Publishing; 2019. p. 200. https://doi.org/10.1007/978-3-319-90617-1_11.
Hermans JJ, Keune K, van Loon A, Iedema PD. An infrared spectroscopic study of the nature of zinc carboxylates in oil paintings. J Anal At Spectrom. 2015;30(7):1600–8. https://doi.org/10.1039/C5JA00120J.
Stulik DC, Kaplan A. Application of a handheld XRF spectrometer in research and identification of photographs. In: Shugar AN, Mass JL, editors. Handheld XRF for art and archaeology. Leuven: Leuven University Press; 2013. p. 111.
McGlinchey C. Handheld XRF for the examination of paintings: proper use and limitations. In: Shugar AN, Mass JL, editors. Handheld XRF for art and archaeology. Leuven: Leuven University Press; 2013. p. 155.
Duffy M, Martins A, Boon JJ. Metal soaps and visual changes in a painting by René Magritte—The Menaced Assassin, 1927. In: van den Berg KJ, Burnstock A, de Keijzer M, Krueger J, Learner T, Tagle dA, et al., editors. Issues in contemporary oil paint. Cham: Springer International Publishing; 2014. p. 197–203. https://doi.org/10.1007/978-3-319-10100-2_13.
Casadio F, Bellot-Gurlet L, Paris C. Factors affecting the reactivity of zinc oxide with different drying oils: a vibrational spectroscopy study. In: Casadio F, Keune K, Noble P, Van Loon A, Hendriks E, Centeno SA, et al., editors. Metal soaps in art: conservation and research. Cham: Springer International Publishing; 2019. p. 153–70. https://doi.org/10.1007/978-3-319-90617-1_9.
Burnstock A. Taking different forms: metal soaps in paintings, diagnosis of condition, and issues for treatment. In: Casadio F, Keune K, Noble P, Van Loon A, Hendriks E, Centeno SA, et al., editors. Metal soaps in art: conservation and research. Cham: Springer International Publishing; 2019. p. 254. https://doi.org/10.1007/978-3-319-90617-1_14.
Noble P. A brief history of metal soaps in paintings from a conservation perspective. In: Casadio F, Keune K, Noble P, Van Loon A, Hendriks E, Centeno SA, et al., editors. Metal soaps in art: conservation and research. Cham: Springer International Publishing; 2019. p. 16–17. https://doi.org/10.1007/978-3-319-90617-1_1.
Osmond G. Zinc soaps: an overview of zinc oxide reactivity and consequences of soap formation in oil-based paintings. In: Casadio F, Keune K, Noble P, Van Loon A, Hendriks E, Centeno SA, et al., editors. Metal soaps in art: conservation and research. Cham: Springer International Publishing; 2019. p. 42–43. https://doi.org/10.1007/978-3-319-90617-1_2.
The authors wish to thank the National Gallery Singapore for allowing us to analyse the paintings; the Heritage Conservation Centre for supporting this study; Gretchen Liu for sharing Liu Kang’s family painting collection, the artist’s archival photographs, and the Shanghai Art Academy Graduation Yearbook; Kenneth Yeo Chye Whatt (Principal Radiographer from the Division of Radiological Sciences at Singapore General Hospital) and Dr Steven Wong Bak Siew (Head and Senior Consultant from the Department of Radiology at Sengkang General Hospital) for facilitating the X-ray radiography; Jocelyn Lau and Lucien Low for the investigation of the Chinese characters from the stamp and their translation; Roger Lee (Assistant Painting Conservator from Heritage Conservation Centre) and Anthony Lau (Senior Painting Conservator from Heritage Conservation Centre) for the translation of the Chinese advertisements. The authors would like to thank Hanna Szczepanowska PhD at West Virginia University for helpful comments, which were crucial for making further improvements to the paper.
Authors and Affiliations
Heritage Conservation Centre, National Heritage Board, 32 Jurong Port Road, Singapore, 619104, Republic of Singapore
Department of Painting Technology and Techniques, Institute for Conservation, Restoration and Study of Cultural Heritage, Nicolaus Copernicus University, ul. Sienkiewicza 30/32, 87-100, Toruń, Poland
Department of Telecommunications and Teleinformatics, Wrocław University of Science and Technology, Wybrzeże Stanisława Wyspiańskiego 27, 50-370, Wrocław, Poland
DL carried out the examination of the paintings, sampling, SEM–EDS analysis; provided the interpretation of the datasets; and wrote the manuscript. TK carried out the FTIR analysis and provided the interpretation of data. BS carried out the FTIR analysis. All authors read and approved the final manuscript.
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.