van gogh's cobalt blue

12
256 Van GoGh's studio practice MurieL GeLdoF and Lise steYn We are indebted to the conservators ella hendriks and Kathrin pilz of the Van Gogh Museum and to Luuk struick van der Loeff and Margje Leeuwenstein of the Kröller-Müller Museum in otterlo for their help in taking paint samples and examining the paintings. We thank Maartje stols- Witlox of the university of amsterdam for her advice and help with the identification of sources for this article. Furthermore, we wish to thank Katrien Keune for analysing a number of the paint samples and johanna salvant (c2rMF) for putting her results at our disposal. it was Katrien who drew our attention to the fact that silicon, phosphorus and calcium are present as trace elements in some cobalt blue paints. the seM-edX analyses were largely carried out in collaboration with Kees Mensch of the shell technology centre in amsterdam, to whom we offer our heartfelt thanks. Finally, we are indebted to Luc Megens and Birgit reissland for their help in the writing of this essay. 1 | see chapter 13 in this volume. 2 | hendriks and Van tilborgh 2011, p. 128. 3 | ibid., p. 93. 4 | selected excerpts from Van Gogh’s letters, as well as all supporting documentation for many of the statements made in this article, are available from the university of amsterdam in the form of a master’s thesis on painting restoration by Lise steyn: steyn 2011. 5 | Van Gogh mentioned that he had, at any rate, bought canvases from a grocer in arles shortly after his arrival there (letters 578 and 593), and both paint and canvas while staying in saint-rémy (letter 808). 6 | orders from tasset: letters 593, 613, 629, 644, 668, 674, 687, 689, 694, 710, 719, 752, 758, 776, 779, 800, 801, 806, 816, 829, 863, 877, 887, 893 (the underlined letter numbers contain a specific order for cobalt blue). order placed with tanguy: letter 863. 7 | Van Gogh was satisfied with most of tasset’s paints, about which he wrote in letter 597: ‘i find his prussian blue poor, and his cinnabar. the rest is good.’ 8 | roy 2008. On 5 August 1882, Vincent wrote a letter to his brother Theo in which he described several paints he had recently bought. He also added a sketch of his palette, complete with colour notes. One of the paints that appears in both the description and the sketch is cobalt blue; however, analysis of paint samples from works executed in his Dutch period (November 1881-November 1885) show that Van Gogh seldom used cobalt blue at the beginning of his career. 1 Only in Antwerp (December 1885-February 1886) did he begin truly to appreciate its properties, as emerges from his correspondence of that period. On 19 December 1885, for example, he wrote: ‘The most expensive is still sometimes the cheapest. Cobalt especially – it can’t be compared with any other blue as regards the delicate tones that one can get with it’ [549]. And later that same year, on 28 December, he wrote: ‘Cobalt – is a divine colour, and there’s nothing so fine as that for putting space around things’ [550]. Cobalt blue was identified in two of the six paintings examined from this brief period. 2 It would be impossible to imagine Van Gogh’s subsequent work without this pigment. In fact, it has been found so far in cross-sections of fifty-six paintings that Van Gogh made in Paris, Arles, Saint-Rémy and Auvers-sur-Oise. During his time in Paris, Van Gogh bought his painting materials from various suppliers, the best known being the artists’ colourman Julien (Père) Tanguy. 3 From Van Gogh’s letters it is known that in the period after Paris, when Van Gogh lived successively in Arles, Saint-Rémy and Auvers-sur-Oise, he preferred to order paints via his brother Theo from the Paris firm of Tasset et L’Hôte. 4 Van Gogh also bought at least some tubes from a local supplier in Arles, 5 but because cobalt blue was a particularly expensive pigment, he probably preferred to order it through Theo. Vincent wrote very detailed letters and often thanked Theo for specific orders: ‘In my last letter I forgot to tell you that I received – a fortnight ago now – the consignment of colours from Tasset’ [629]. Between April 1888 and his death in July 1890, Van Gogh placed – through Theo – twenty-four orders from Tasset and three orders from Tanguy. Of the order lists that have survived, seven specifically ask for cobalt blue from Tasset, whereas only one VAN GOGH S COBALT BLUE Muriel Geldof and Lise steyn 15

Upload: independent

Post on 30-Apr-2023

1 views

Category:

Documents


0 download

TRANSCRIPT

256 �  Van GoGh's studio practice – MurieL GeLdoF and Lise steYn

We are indebted to the conservators ella hendriks and Kathrin pilz of the Van Gogh Museum and to Luuk struick van der Loeff and Margje Leeuwenstein of the Kröller-Müller Museum in otterlo for their help in taking paint samples and examining the paintings. We thank Maartje stols-Witlox of the university of amsterdam for her advice and help with the identification of sources for this article. Furthermore, we wish to thank Katrien Keune for analysing a number of the paint samples and johanna salvant (c2rMF) for putting her results at our disposal. it was Katrien who drew our attention to the fact that silicon, phosphorus and calcium are present as trace elements in some cobalt blue paints. the seM-edX analyses were largely carried out in collaboration with Kees Mensch of the shell technology centre in amsterdam, to whom we offer our heartfelt thanks. Finally, we are indebted to Luc Megens and Birgit reissland for their help in the writing of this essay.

1 | see chapter 13 in this volume.

2 | hendriks and Van tilborgh 2011, p. 128.

3 | ibid., p. 93.

4 | selected excerpts from Van Gogh’s letters, as well as all supporting documentation for many of the statements made in this article, are available from the university of amsterdam in the form of a master’s thesis on painting restoration by Lise steyn: steyn 2011.

5 | Van Gogh mentioned that he had, at any rate, bought canvases from a grocer in arles shortly after his arrival there (letters 578 and 593), and both paint and canvas while staying in saint-rémy (letter 808).

6 | orders from tasset: letters 593, 613, 629, 644, 668, 674, 687, 689, 694, 710, 719, 752, 758, 776, 779, 800, 801, 806, 816, 829, 863, 877, 887, 893 (the underlined letter numbers contain a specific order for cobalt blue). order placed with tanguy: letter 863.

7 | Van Gogh was satisfied with most of tasset’s paints, about which he wrote in letter 597: ‘i find his prussian blue poor, and his cinnabar. the rest is good.’

8 | roy 2008.

On 5 August 1882, Vincent wrote a letter to his brother Theo in which he

described several paints he had recently bought. He also added a sketch of his

palette, complete with colour notes. One of the paints that appears in both the

description and the sketch is cobalt blue; however, analysis of paint samples

from works executed in his Dutch period (November 1881-November 1885) show

that Van Gogh seldom used cobalt blue at the beginning of his career.1 Only in

Antwerp (December 1885-February 1886) did he begin truly to appreciate its

properties, as emerges from his correspondence of that period. On 19 December

1885, for example, he wrote: ‘The most expensive is still sometimes the cheapest.

Cobalt especially – it can’t be compared with any other blue as regards the

delicate tones that one can get with it’ [549]. And later that same year, on 28

December, he wrote: ‘Cobalt – is a divine colour, and there’s nothing so fine as

that for putting space around things’ [550]. Cobalt blue was identified in two

of the six paintings examined from this brief period.2 It would be impossible to

imagine Van Gogh’s subsequent work without this pigment. In fact, it has been

found so far in cross-sections of fifty-six paintings that Van Gogh made in Paris,

Arles, Saint-Rémy and Auvers-sur-Oise.

During his time in Paris, Van Gogh bought his painting materials from various

suppliers, the best known being the artists’ colourman Julien (Père) Tanguy.3

From Van Gogh’s letters it is known that in the period after Paris, when Van

Gogh lived successively in Arles, Saint-Rémy and Auvers-sur-Oise, he preferred

to order paints via his brother Theo from the Paris firm of Tasset et L’Hôte.4 Van

Gogh also bought at least some tubes from a local supplier in Arles,5 but because

cobalt blue was a particularly expensive pigment, he probably preferred to order

it through Theo. Vincent wrote very detailed letters and often thanked Theo for

specific orders: ‘In my last letter I forgot to tell you that I received – a fortnight

ago now – the consignment of colours from Tasset’ [629]. Between April 1888

and his death in July 1890, Van Gogh placed – through Theo – twenty-four

orders from Tasset and three orders from Tanguy. Of the order lists that have

survived, seven specifically ask for cobalt blue from Tasset, whereas only one

van gogh’s cobalt blue

Muriel Geldof and Lise steyn

15

257257257Van GoGh’s cobalt blue

mentions ordering it from Tanguy.6 Van Gogh probably preferred Tasset et L’Hôte

because he was generally more satisfied with their materials than with Tanguy’s.7

Comparing the two merchants, he explained that ‘the reason why we went to

Tasset’s is that the latter’s colours are in general less insipid’ [889]. Moreover, he

explicitly said that Tanguy’s cobalt blue was ‘too poor to order more from him’

[634], though his letters do not reveal his reasons for thinking it inferior.

Examination of this pigment in a number of Van Gogh’s paintings has

revealed cobalt blue pigments of varying chemical compositions. Cobalt blue is

a compound of cobalt, oxygen and aluminium with the chemical name cobalt(II)

oxide-alumina or cobalt(II)aluminate.8 In addition to cobalt, aluminium and

oxygen, a number of other elements are present in his cobalt blue. One of these is

nickel, such as found in the Self-portrait with straw hat (fig. 1). In the cobalt blue

in Van Gogh’s The bedroom (fig. 2), nickel is largely lacking, whereas the elements

silicon and phosphorus are indeed present.

Fig. 1a Vincent van Gogh, Self-portrait with

straw hat (F469)

b Cross-section of a paint sample taken from the blue paint in the background. On top of the ground layer the blue paint consists of lead white, cobalt blue and viridian.

c SEM-EDX spectrum of the cobalt blue pigment particle indicated with a circle in the paint layer. The cobalt blue has a relatively high nickel content.

Fig. 2a Vincent van Gogh, The bedroom (F482)

b Cross-section of a paint sample taken from the wall. On top of the ground layer a blue (originally purple) paint layer is present which contains zinc white, cobalt blue and cochineal.

c SEM-EDX spectrum of the cobalt blue pigment particle indicated with a circle in the paint layer. The cobalt blue has a relatively low nickel content and contains traces of silicon and phosphorus.

Fig. 1a

Fig. 1c

Fig. 1b

Fig. 2b

Fig. 2a

Fig. 2c

258 �  Van GoGh's studio practice – MurieL GeLdoF and Lise steYn

9 | eastaugh et al. 2004, p. 112.

10 | Wehlte 1982, p. 148.

11 | thénard 1803: ‘sa présence nuirait singulièrement à la purieté de la couleur.’

12 | ibid: ‘très beau, très vif et très pur.’

13 | ibid.

14 | harley 1982, p. 57.

This essay relates these small differences in composition to the methods

generally used to produce cobalt blue in the nineteenth century, and presents an

overview of the elements identified in this pigment in paints used by Van Gogh.

The variants thus distinguished are then linked to his paint suppliers. Finally,

an attempt is made to understand why Van Gogh was dissatisfied with the cobalt

blue supplied by Tanguy.

The producTion of cobalT blue

Production methodsThe basic method of producing cobalt blue was discovered by Josef Leithner

as early as 1775 in Vienna,9 but it is the French chemist Louis Jacques Thénard

who is generally credited with its discovery in 1803.10 Thénard produced a blue

pigment by calcining at cherry-red heat (c. 1000-1300˚C) alumina mixed with

either cobalt phosphate or cobalt arsenate – compounds obtained from cobalt

ores. In his recipes, Thénard stated that he used cobalt ore from Tunaberg and

that this ore contains sulphur, arsenic and iron, in addition to cobalt. He pointed

out that the preparation of the cobalt salts is very important for the quality

of the final product, and stressed that iron should be separated from cobalt,

since ‘its presence would be particularly damaging to the purity of the colour’.11

Indeed, his recipes include steps to remove these impurities. To obtain cobalt

phosphate, he first roasted the cobalt ore (to release arsenical gases) and then

dissolved it in nitric acid. He subsequently filtered this solution to remove the

red iron oxide thus formed, and mixed it with a solution of sodium phosphate

in water, which gives a violet precipitate of cobalt phosphate. To produce cobalt

arsenate, he dissolved the cobalt ore in nitric acid to remove the sulphur. He then

added a weak solution of potassium carbonate, which was filtered to remove the

precipitated iron arsenate. Adding more of the potassium carbonate solution

resulted in a pink precipitate of cobalt arsenate.

Thénard found out that if the right amounts of cobalt phosphate and alumina

were mixed, a blue could be obtained that was ‘very beautiful, very bright and

very pure’.12 Such a beautiful blue could not be created by starting with cobalt

arsenate.13 Thénard’s cobalt blue was being produced by 1807, and the production

method changed little over the course of the century. No major alterations were

made to the original recipe, and the choice remained between using cobalt

phosphate or cobalt arsenate as the salt needed to synthesise the pigment.14

By the late nineteenth century, both methods of making cobalt aluminate were

259259259Van GoGh’s cobalt blue

15 | Brühlmann 2003, p. 25.

16 | carlyle 2001, p. 471.

17 | thénard 1803; chaptal 1807, pp. 373-75; Watin 1823: pp. 44-47; Mérimée 1839, p. 153.

18 | Mérimée 1839, pp. 153-55.

19 | nowadays the French pigment referred to as ‘thénard’s blue’ is used for the cobalt phosphate variant, and the cobalt arsenate variant is referred to as ‘Leithner’s blue’. harley 1982, p. 57.

20 | roy 2008, p. 155.

21 | Brühlmann 2003, p. 110.

22 | overman 1852, p. 628.

23 | ure 1854, p. 445.

24 | ibid., pp. 447-48.

25 | church 1890, p. 187.

well known,15 but the cobalt phosphate method appeared to be the method

of choice, since this was the one usually described in recipes, often with no

mention of cobalt arsenate.16 Although Thénard’s paper also mentioned the

cobalt arsenate variant, he advocated the cobalt phosphate variant, as did other

French sources from the early nineteenth century onwards.17 The painter and

chemist Jean-François-Léonor Mérimée, for example, stated in 1830 that ‘the

arsenate, the borate or the phosphate of cobalt may be employed, but the latter

in preference, for it produces the purest colour. The arsenate always has a violet

tinge, more visibly by lamp light than by day’.18 This distinct preference may be

the reason why the pigment obtained from cobalt phosphate was referred to as

the ‘French’ pigment.19 The presence of phosphorous (either unchanged as cobalt

phosphate20 or converted into aluminium phosphate) or arsenic in the final

product is indicative of the method used.21

Cobalt occurs in nature in combination with sulphur, arsenic, iron and

nickel. Cobalt and nickel appear next to each other on the periodic table and

have similar chemical characteristics. It is difficult to separate them, which is

why cobalt blue sometimes contains nickel. The nineteenth-century mining

engineer Frederick Overman mentions the difficulty of separating the two in

his treatise on metallurgy (1852): ‘nickel and cobalt, the separation of which is

somewhat laborious.’22 The Scottish chemist Andrew Ure mentions the purity of

the Tunaberg ores and describes the purification of cobalt ore, which is pounded

into a powder, passed through a sieve, and sometimes washed.23 It is then placed

in a reverberatory furnace and, as in Thénard’s recipe, calcined for several hours

to expel the arsenic content. This method cannot remove all the impurities, but

Uwe declares that what remains is ‘not injurious to subsequent operations’ and

goes on to say: ‘The cobalt ores, rich in nickel, are exposed to slow oxydizement

in the air, whereby the iron, cobalt, arsenic, and sulphur get oxygenated by

the atmospheric moisture, but the nickel continues in the metallic state. This

action of the weather must not be extended beyond a year, otherwise the nickel

becomes affected, and injures the cobalt blue. […] The freer the cobalt ore is from

foreign metals, the finer its colour, and the deeper is the shade […] The presence

of nickel gives [glass] a violet tone.’24 The chemist Arthur Church also underlines

the importance of using ‘pure’ materials. In The chemistry of paints and painting,

he describes the production of cobalt blue and states that ‘the complete freedom

of the materials used from iron and nickel is essential to the purity and beauty

of the blue pigment formed’.25 Church does not explain the exact effect of nickel,

260 �  Van GoGh's studio practice – MurieL GeLdoF and Lise steYn

26 | From the end of the fifteenth century to the nineteenth century, cobalt mines in the schneeberg district in saxony, Germany, were exploited for the production of smalt. this ended with the arrival on the market of synthetic ultramarine. hammer 2004.

27 | hamilton 1994, p. 10.

28 | it has been shown that all three mines traded with France at that time. tunaberg is referred to specifically in several French sources and is therefore known to have supplied the parisian market. new caledonia was a French colony and a major commercial supplier of cobalt in the nineteenth century, and chalanches was the most active known cobalt mine in France itself. these should not be interpreted as the only viable sources, however.

29 | Zakrzewski et al. 1980, p. 165.

30 | although these ores were known for their purity (Zakrzewski 1980, p. 165), they were not necessarily nickel-free, depending on the other surrounding – so-called siderophile – ores, which may have been mined simultaneously. cobalt and nickel ores (including cobalt safflorite) are of the siderophile kind and occur in melted magmas in the earth’s crust. as these magmas cool, the nickel and cobalt tend to settle into layers. due to their similar characteristics and atomic weight, cobalt and nickel will associate. Various factors, such as the cooling temperature and the base composition of the melt, influence the composition of the final ore. ores in the same ore field may therefore contain different proportions of cobalt and nickel.

or why its absence is ‘essential to the purity and beauty’. All of this goes to

show, however, that in the nineteenth century it was known that the separation

of cobalt and nickel was very important but difficult to accomplish, and it was

therefore better, when manufacturing fine pigments, to use a cobalt ore that

contained little or no nickel.

Sources of cobalt in the nineteenth centuryIn the late nineteenth century, cobalt was no longer mined commercially,26 but

was usually a by-product of mining nickel, copper or silver.27 Most countries were

profiting from international trade, so the ores needed in France to produce cobalt

blue – or any of the other pigments already in production – could have found

their way to Parisian colour merchants from anywhere in the world. It emerges

from the literature, however, that in the late nineteenth century the mines of

Tunaberg, New Caledonia and Chalanches were the most likely sources of the

cobalt ores used to produce cobalt blue.28

Tunaberg

Thénard’s recipe and other French sources stipulate Tunaberg as the source of

the cobalt ore required in the production of cobalt blue. Cobalt was mined at

Tunaberg (in the Hällefors area, Bergslagen, Sweden) as a by-product of copper

mining. Interestingly, of the three main mines in the Hällefors area, Tunaberg

produced only 1.3% of the cobalt (89.6% came from the Gladhammar mine and

8.7% from Vena). It therefore seems odd that French recipes for cobalt blue

stipulate Tunaberg as the source of the cobalt. However, research into the ores

mined in the Hällefors area has shown that the Tunaberg ore field contained a

specific variety of cobalt ore. Although cobaltite – the most widespread cobalt

ore found in nature – was the main source of cobalt in all three Hällefors mines,

a cobalt ore named safflorite, which was not common at the other mines in the

Hällefors area, was mined at Tunaberg.29 This rare mineral is a cobalt-iron-

arsenide (CoFe)As2 which contains no nickel in its pure form, unlike the nickel-

containing cobaltite mineral.30 Thénard’s recipes give explicit instructions on

how to remove iron and arsenic, but do not mention nickel. It is likely, therefore,

that he used safflorite, the nickel-free cobalt ore from Tunaberg, to conduct his

experiments. When cobalt blue is prepared according to Thénard’s recipe with

a cobalt ore other than safflorite, some nickel might be detected in the pigment,

because the recipe does not provide for its extraction.

261261261Van GoGh’s cobalt blue

31 | sudol 2005, p. 2.

32 | there are two types of asbolane: cobalt-rich asbolane (co>ni) and nickel-rich asbolane (ni>co).

33 | Llorca and Monchoux 1991, p. 154.

34 | slags are the by-product or waste of smelting ore or extraction/mining operations in which many minerals and metals are left after extraction of the intended ore.

35 | rickard 1894, p. 694.

36 | ibid., pp. 694-95.

New Caledonia

New Caledonia, located east of Australia, was a colony of France. Nickel deposits

were discovered in this archipelago by Jules Garnier in 1864, and commercial

exploitation began in 1875. New Caledonia was the world’s largest source of

nickel until 1905.31 The ore found in these mines is asbolane – a type of ultramafic

rock containing variable proportions of nickel to cobalt32 – which was the main

source of cobalt from this area.33 The extensive commercial mining of New

Caledonian cobalt led to a decline in European cobalt mining at the end of the

nineteenth century. It is possible that nickel-rich cobalt blue is representative of

this asbolane type, which – if used to make cobalt blue using Thénard’s method –

would presumably have left some nickel residue in the resulting pigment.

Chalanches

The main mining activity at Chalanches in the Vosges area near Dijon

concentrated on silver deposits that were discovered in 1767. A cobalt ore field,

the only one then known in France, was also exploited in Chalanches. This

deposit was large enough to supply the needs of the colour manufacturers, yet

it seems an unlikely source, because even though these ores were known, they

do not appear to have been exploited for this purpose in France. A report by

T.A. Rickard, director of the Chalanches mine in the late nineteenth century,

states that its slags34 were rich in cobalt and nickel, but that the directors of

mining operations took little interest in their extraction.35 Though an unlikely

source, it is mentioned here because Rickard specifically mentions the use of

their ores in attempts to produce cobalt pigments. In a report of 1894 he states:

‘More recently, a German chemist was employed at Allemont in an experiment to

manufacture cobalt pigments for the arts. He was not successful, and the attempt

was abandoned.’36 This proves that mining operators and chemists experimented

with – and possibly used in the production of pigments – ores from mines where

cobalt was merely a by-product.

The composiTion of cobalT blue

Cobalt blue in Van Gogh’s paintingsTo corroborate the assumption that Van Gogh used various kinds of cobalt blue,

forty-one cross-sections taken from thirty-two paintings of his Paris and post-

Paris periods were systematically analysed by means of SEM-EDX. After examining

each cross-section to determine the percentage of weight (wt%) of nickel and

262 �  Van GoGh's studio practice – MurieL GeLdoF and Lise steYn

cobalt in the particles of cobalt blue pigment or in the cobalt blue paint layers, the

proportion of wt%(Ni) to wt%(Co) was calculated.37 The presence of the elements

phosphorus, silicon, calcium and/or potassium was also determined.38

The graph in fig. 3 gives the average proportion of nickel to cobalt and the

date of origin of the paintings in this study that were shown to contain cobalt

blue.39 Each point represents the average of the results of one to six analyses of

37 | the seM-edX analyses of most samples were carried out in collaboration with Kees Mensch at the shell technology centre, amsterdam, using a jeoL jsM 6490 LV scanning electron microscope and a noran system six eds-system. the primary electron beam energy used was 25 keV. the seM-edX analyses of the samples taken from F462, F792, F754 and F783 were performed by johanna salvant at the c2rMF, using a philips XL30cp scanning electron microscope and a Link isis 300 eds-system. the primary electron beam energy used was 20 keV. the cross-sections were coated with carbon and examined in high vacuum mode. the weight percentage (wt%) of the elements was determined by means of the Proza method.

38 | the element arsenic, which also occurs as an impurity in cobalt blue, was not identified in the cobalt blue.

39 | the date of a painting as given in fig. 3 is the middle of the period to which the painting is dated. For example, if a painting is dated to september 1889, the point is given as 15 september 1889, whereas with a work dated to september-october 1889, the point given would be 1 october.

Fig. 3 Average Ni/Co ratio and presence of phosphorus for each of Van Gogh’s paintings examined.

0

5

10

15

20

Apr 86 Oct 86 Oct 87Apr 87 Oct 88Apr 88 Oct 89Apr 89 Apr 90

(wt%

Ni/w

t%Co

) x10

2

Date of painting

Paris Post-Paris

Ni/Co < 0.05; trace of phosphorusNi/Co ~ 0.15; no phosphorusNi/Co ~ 0,08; no phosphorusNi/Co < 0.05; no phosphorus

ni/co < 0.05; trace of phosphorus · Sunflowers gone to seed (F377), mid-august- mid-september 1887

· The Zouave (F423), june 1888 · Eugène Boch (The poet) (F462), early september 1888 · The bedroom (F482), mid-october 1888 · Orchard in blossom (F511), april 1889 · Undergrowth (F746), first half of july 1889 · Entrance to a quarry (F744), mid-july 1889 · Tree trunks with ivy (F747), july 1889 · Pièta (after delacroix) (F630), mid-september 1889 · The sheaf-binder (after Millet) (F693), mid-september 1889

· Peasant woman binding sheaves (after Millet) (F700), mid-september 1889

· The sheep-shearer (after Millet) (F634), c. 20 september 1889

· The woodcutter (after Millet) (F670), probably between 20 september-october 1889

· Night (after Millet) (F647), late october-early november 1889

· Garden with butterflies (F402), spring 1890 · Thatched cottages (F792), first half of june 1890 · Landscape at twilight (F770), late june 1890 · Wheatfield under thunderclouds (F778), first half of july 1890

· Tree roots (F816), late july 1890

ni/co ~ 0.15; no phosphorus · Flame nettle in a flowerpot (F281), late june-mid-july 1886 · View of Paris (F261), june-july 1886 · Vase with gladioli and Chinese asters (F248a), august-mid-september 1886

· The Seine bridge at Asnières (F240), summer 1887 · Self-portrait with straw hat (F469), august-september 1887

· Self-portrait as a painter (F522), december 1887- February 1888

· The green vineyard (F475), september 1888 · Portrait of Doctor Gachet (F754), june 1890 · Two children (F783), june 1890

ni/co < 0.05; no phosphorus · Wheatfield with partridge (F310), mid-june-mid-july 1887 · Portrait of Theo van Gogh (F294), summer 1887 · Red cabbages and onions (F374), october-november 1887

ni/co ~ 0.08; no phosphorus · Garden with courting couples: Square Saint-Pierre (F314), mid-late May 1887

263263263Van GoGh’s cobalt blue

40 | the average standard deviation was 1.2 × 102.

41 | in nearly all the cross-sections with cobalt blue that contain traces of phosphorus, the cobalt blue was found to contain small amounts of silicon and calcium and/or potassium. however, the cross-sections with cobalt blue in which phosphorus is absent sometimes also contain small quantities of some of these elements, because they regularly occur in other pigments in the paint layer.

42 | in Four sunflowers gone to seed (F452) of august-october 1887 and in the Courtesan (after eisen) (F373), of october-november 1887, the firm’s stamp was found on the works’ original stretchers. such a stamp was also found on the back of the canvas of Grapes, lemons, pears and apples (F382), dated to the autumn of 1887. hendriks and Van tilborgh 2011, p. 528.

various particles of cobalt blue pigment or paint layers in cross-sections.40 The

presence or absence of phosphorus in the cobalt blue was also determined.41 In

this way it was possible to distinguish four kinds of cobalt blue: a nickel-poor

cobalt blue (Ni/Co < 0.05) that contained some phosphorus, and cobalt blue

with a high (Ni/Co ~ 0.15), moderate (Ni/Co ~ 0.08) or low (Ni/Co < 0.05) nickel

content and no phosphorus.

The cobalt blue that contains a trace of phosphorus and relatively little

nickel (Ni/Co < 0.05) was identified in nineteen paintings. The presence

of phosphorous in the pigment indicates that the cobalt blue was prepared

according to Thénard’s cobalt phosphate method. The almost complete absence

of nickel indicates that, also with regard to the synthesis of cobalt phosphate

from cobalt ore, the instructions in Thénard’s original recipe may have been

followed closely, since Thénard stipulates the use of cobalt ore from Tunaberg,

where the nickel-poor ore safflorite was mined.

This type of pigment occurs most frequently in paintings made in the post-

Paris period, when Van Gogh was ordering his paints (through Theo in Paris)

primarily from Tasset et L’Hôte, as evidenced by the letters that specify this firm

as the desired supplier. It is therefore likely that this variant of cobalt blue came

from Tasset et L’Hôte. Van Gogh also used this kind of cobalt blue in one of his

Paris works, Sunflowers gone to seed (F377), made in August or September 1887.

Earlier research has shown that Van Gogh bought canvas and stretchers from

Tasset et L’Hôte in the autumn of 1887,42 so perhaps he also purchased the blue

paint used in this painting at that time.

The variant of cobalt blue that contains no phosphorus but a relatively high

concentration of nickel (Ni/Co ~ 0.15) was found in nine paintings. Since neither

phosphorus nor arsenic was detected in this variant, it is not possible to say

which method – cobalt phosphate or cobalt arsenate – was used to produce the

pigment. It is clear, however, that the nickel-poor cobalt ore from Tunaberg was

not used but an ore richer in nickel, such as cobaltite (which occurs worldwide)

or asbolane (which was mined in the French colony of New Caledonia): ores that

were probably less expensive because they were widely available.

Six paintings containing this type of cobalt blue can be dated to Van Gogh’s

Paris period and three works to his post-Paris period. One of these later works

is The green vineyard (F475), which Van Gogh painted in September 1888. In a

letter of 3 October 1888, Van Gogh wrote: ‘I’ve painted the vineyards entirely

with Tanguy’s colours’ [694], and among the paintings he referred to here was

264 �  Van GoGh's studio practice – MurieL GeLdoF and Lise steYn

43 | this order contained paints from both tanguy and tasset (letter 889).

44 | hendriks and Van tilborgh 2011, p. 93.

45 | the samples were examined only for inorganic pigments and fillers; no research was carried out into possible organic additives. a sample taken from the sky of The green vineyard (F475) proved to contain both cobalt blue and prussian blue. it is unclear whether this was added by the manufacturer or by Van Gogh himself.

46 | Red cabbages and onions (F374) was painted on the same kind of primed canvas as Self-portrait as a painter (F522), for which Van Gogh used another type of cobalt blue. hendriks and Van tilborgh 2011, p. 546.

47 | townsend et al. 1995, p. 69. the cross-section of Red cabbages and onions (F374) contains only one particle of cobalt blue, so it is impossible to say whether a similar mixture from a tube of paint was used here too.

The green vineyard (F475). Also the other two post-Paris works containing

this variant of cobalt blue – Portrait of Doctor Gachet (F754) and Two children

(F783) – both dated to June 1890, were probably painted with cobalt blue paint

procured from Tanguy. In this period, in fact, Van Gogh ordered paints from both

Tasset et L’Hôte and Tanguy, one such order containing ‘cobalt’ appearing in

letter 863 of 29 April 1890, which he received on 11 May of that year [870].43 That

cobalt blue from Tanguy was among the paints in this delivery can be deduced

from a letter of 17 June 1890, in which Van Gogh referred to the tubes of paint

from Tanguy: ‘the tubes of cobalt, for example like the one I have in my hands’

[889]. He probably used this paint for the Portrait of Doctor Gachet (F754) and

Two children (F783). On the other hand, two works from July 1890, Wheatfield

under thunderclouds (F778) and Tree roots (F816), were found to contain the

cobalt blue that matches Tasset’s variant of the pigment. In these paintings Van

Gogh probably used the cobalt blue he had ordered on 14 June [887] and received

on 17 June of that year [889].

It is likely that the six Paris works containing relatively nickel-rich cobalt

blue were also executed in paint bought from Père Tanguy, since this colourman

was one of Van Gogh’s principal suppliers in Paris.44 One Paris painting that

presumably contains Tanguy’s cobalt blue is the Self-portrait as a painter (F522),

made in the winter of 1887-88, in which cobalt blue appears on the artist’s

palette. Tanguy also supplied other artists in Van Gogh’s Paris circle, and they,

too, prove to have used cobalt blue with the same characteristics (see below).

In general Tanguy’s paints were favourably priced. He probably bought

pigment made of cheap cobalt ore to cut costs on his cobalt blue paint, since

it was not found to contain other fillers or additives that could have made it

less expensive.45

Three paintings from the Paris period proved to contain a nickel-poor

cobalt blue (Ni/Co < 0.05) and no phosphorus. Van Gogh used this variant in

Red cabbages and onions (F374),46 dating from the autumn of 1887, as well as in

his Portrait of Theo van Gogh (F294) and Wheatfield with partridge (F310), both

made in the summer of that year. In the Portrait of Theo van Gogh and Wheatfield

with partridge, the cobalt blue is mixed with ultramarine, Prussian blue, barium

sulphate and a little white pigment. Cobalt blue was frequently extended with

other blue pigments, such as Prussian blue and ultramarine,47 and barium

sulphate was presumably added as a filler. This was done because cobalt blue

was very expensive, as evidenced by one of Tanguy’s price lists from that period:

265265265Van GoGh’s cobalt blue

48 | hendriks and Van tilborgh 2011, p. 93.

49 | ibid., pp. 94-95.

50 | ibid., p. 46.

51 | Poudre de riz is painted on a canvas bearing the stamp of the firm rey & perrot. research carried out by devi ormond, Van Gogh Museum, amsterdam.

52 | callen 2000, p. 103.

a tube of cobalt blue costs twice as much as ultramarine blue and four times as

much as Prussian blue.48 It is not known where Van Gogh bought this adulterated

cobalt blue paint, but it probably came from one of the many paint dealers in

the Montmartre quarter,49 where he lived, though it is also possible that Tasset

et L’Hôte or Tanguy sold another kind of blue paint with a different variant of

cobalt blue.

The last type of cobalt blue that could be distinguished contains a moderate

amount of nickel (Ni/Co ~ 0.08) and no phosphorus. This variant was found

in only one painting, Garden with courting couples: Square Saint-Pierre (F314),

which is dated to the second half of May 1887. The source of this cobalt blue is

also unclear.

The five Paris paintings that Van Gogh probably did not paint with cobalt

blue from Tanguy were all produced around the summer of 1887. A letter Vincent

wrote to Theo in mid-July suggests that in the spring of 1887 he had received

supplies from Tanguy, which at least included canvases, but that his wife – much

to Van Gogh’s displeasure – had put a stop to this in the summer [571]. This might

explain why Van Gogh temporarily bought his paint, starting in the summer of

1887, from other colourmen as well. Even before this run-in with Tanguy’s wife,

however, Van Gogh had sometimes bought his blue paint elsewhere. It is likely

that he had started work on Garden with courting couples: Square Saint-Pierre

(F314) before Tanguy’s deliveries had begun, probably around mid-May 1887,50

since he painted this work over previous compositions, which he probably would

not have done if he had had an adequate supply of new canvas from Tanguy.

Cobalt blue in paintings by Van Gogh’s contemporariesCross-sections taken from seven works by Van Gogh’s contemporaries – three by

Emile Bernard, two by Paul Signac and one each by Henri de Toulouse-Lautrec

and Armand Guillaumin – have also been examined to determine the type of

cobalt blue used (Fig. 4). The paintings are all dated to between 1885 and 1887:

in or shortly before the period in which the artists would have come into contact

with Van Gogh in Paris. Four of these works – Toulouse-Lautrec’s Young woman

at a table (Poudre de riz),51 Guillaumin’s Portrait of a young woman, Bernard’s

Street in Pont-Aven and Signac’s Railway junction near Bois-Colombes – contain

cobalt blue with a relatively high nickel content (Ni/Co ratio ~ 0.15) and no

phosphorus, the variant that matches the cobalt blue sold by Tanguy. Indeed, it

is known that Guillaumin and Signac also patronised Tanguy’s shop.52 The cobalt

266 �  Van GoGh's studio practice – MurieL GeLdoF and Lise steYn

blue used by Signac to paint Breakfast is the type sold by Tasset et L’Hôte, which

is nickel-poor and contains traces of phosphorus. The stretcher and the back of

the canvas bear the stamp of Tasset et L’Hôte, so Signac evidently bought from

this supplier not only the paint but also the canvas used for Breakfast.

The cobalt blue used by Bernard for Ragpicker fishing contains a moderate

amount of nickel (Ni/Co ~ 0.06) and no phosphorus. This variant therefore

resembles the cobalt blue used by Van Gogh in Garden with courting couples:

Square Saint-Pierre (F314). In Bernard’s grandmother, Bernard used a cobalt blue

that contains neither nickel nor phosphorus, a type not found in any of the works

by Van Gogh examined for this study.

conclusion

In Paris, Van Gogh used at least four different kinds of cobalt blue. The type

occurring most frequently in this period is characterised by a relatively high

nickel content and the absence of phosphorus. This variant of cobalt blue, which

was the kind sold by Tanguy, was also used by Guillaumin, Bernard, Toulouse-

Lautrec and Signac.

Because arsenic is also lacking, it is unclear whether the cobalt blue sold by

Tanguy was produced according to Thénard’s cobalt phosphate method or the

cobalt arsenate method. It is certain, however, that a nickel-containing cobalt

ore was used from which the nickel had not been completely extracted. Such

Fig. 4 Average Ni/Co ratio and presence of phosphorus for the paintings by Van Gogh’s contemporaries

Jun 85 Dec 85 Jun 86 Dec 86 Jun 87 Dec 87

(wt%

Ni/w

t%Co

) x10

2

Date of painting

0

5

10

15

20

Emile Bernard, Bernard's grandmother (s205); no phosphorusEmile Bernard, Ragpicker fishing (s367); no phosphorusEmile Bernard, Street in Pont Aven (s390); no phosphorusArmand Guillaumin, Portrait of a young woman (s227); no phosphorusPaul Signac, Breakfast (KM106.849); trace of phosphorusPaul Signac, Railway junction near Bois-Colombes (s381); no phosphorusHenri de Toulouse-Lautrec, Young woman at a table (Poudre de riz) (s274); no phosphorus

267267267Van GoGh’s cobalt blue

ores were widely available and therefore probably less expensive than the nickel-

poor ore from Tunaberg prescribed by Thénard. This accords with the fact that

Tanguy’s paints were generally lower in price.

In addition to the cobalt blue he bought from Tanguy, Van Gogh also

purchased this paint in Paris from at least two unknown suppliers, and one

painting was shown to contain cobalt blue that matches the paint sold by Tasset

et L’Hôte. Evidently Van Gogh benefited from the large number of artists’

colourmen in the Montmartre area.

Only two types of cobalt blue were found in the works dating from Van

Gogh’s post-Paris period (Arles, Saint-Remy and Auvers-sur-Oise). The variant

that occurs most frequently contains relatively little nickel but traces of

phosphorus, which means that this cobalt blue was probably produced according

to Thénard’s cobalt phosphate method, making use, moreover, of the nickel-poor

cobalt ore from Tunaberg that he prescribed. In this period Van Gogh ordered

his paints mainly from Tasset et L’Hôte through his brother Theo in Paris, so it is

highly likely that this type matches the cobalt blue sold by that firm. Signac, too,

used this kind of cobalt blue paint. Van Gogh also ordered paint from Tanguy in

his post-Paris period, which explains the second variant.

In his letters to Theo, Vincent frequently complained about the quality of

Tanguy’s paint, and he once specifically mentioned cobalt blue in this respect.

Exactly why Van Gogh preferred Tasset’s cobalt blue to Tanguy’s cannot be

gleaned from the correspondence, but comparison of the chemical composition

of the pigments used in the two paints provides some clues. Tasset’s cobalt

blue was made according to Thénard’s cobalt phosphate method, which he and

other French authors recommended, since it would produce the most beautiful

colour. On the other hand, the high nickel content of Tanguy’s cobalt blue must

have adversely affected its colour, as nineteenth-century chemists already

knew. In conclusion, the results indicate that Tasset et L’Hôte’s cobalt blue was

more appealing in colour, which was presumably why Van Gogh preferred it

to Tanguy’s.