Vincent van Gogh

A sourced technical report. Every claim cites a source key; the sources are listed at the end. "Uncertain" marks claims that are unverified, disputed or taken from a single object. It covers his oils from Paris (1886) to Auvers (1890) and notes where the Dutch practice differs. The main evidence is the Van Gogh Museum's technical study of 93 Antwerp and Paris paintings (read in full), studies of his Paris supports and his canvas weaves, open-access analyses of later paintings, and his letters, mostly read through the online edition's search excerpts and page summaries. The yellow decay studies, one Auvers analysis and a palette record were read through press releases or search summaries.

Summary

From Paris on, Van Gogh painted on commercially primed canvas: in Paris on ready-stretched standard sizes bought one at a time, from Arles on almost always on canvas cut from 10-metre rolls of a Paris colorman's ordinary grade [AM2 pp.42, 47–50; WM pp.114–116]. The grounds are lead white in oil, often over chalk; in 1887 he also mixed his own toned grounds to paint over failed pictures [AM2 pp.50–61]. He used bright tube colors, several known to fade or darken, and asked for coarsely ground paint holding less oil [VGM2 pp.129–144; L668]. Paint went on thin in early 1887; otherwise it is mostly direct and thick, often straight from the tube, worked fast out of doors and retouched later [AM2 p.58; VGM2 pp.152–153; ALT; L635]. He did not varnish [L853]. Red lakes have faded and some chrome yellows have darkened [GEL18; VGM2 pp.133–138]. The documented pigments are listed in §4.

Terms. Ground/priming: the preparatory layers between canvas and paint. Carton: primed cardboard sold as a support. Toile ordinaire: a colorman's ordinary grade of primed canvas. Absorbent ground: a priming of chalk or a similar white in glue, which draws oil from the paint and leaves it matt. À l'essence: painting with paint thinned heavily with turpentine. Perspective frame: a frame with threads strung across it, looked through to fix a view. Lake: a dye fixed on a white base, usually transparent. Extender: a cheap white filler in tube paint. Impasto: thick, raised paint. Wet in wet: paint laid into paint still wet. IRR (infrared reflectography): imaging that shows carbon drawing under paint.

1. Supports

• Paris. Of 93 Antwerp and Paris paintings examined, 71 are on primed canvas and 17 on ready-primed carton [AM2 p.42]. He bought canvases ready primed and stretched, one at a time: 54 of 67 are standard French formats, most often figure 8, 10 and 6 [AM2 pp.47–49, 58]. Cheap 2 mm cartons served his first weeks in Paris [AM2 p.47].
• Weave. Nearly all plain weave, three twill. Thirty are a skeletal weave of 11.5–13.5 threads per cm, probably the cheapest study grade, used even for an exhibited picture; six of 1887 are very fine (19–30 per cm). Linen with one exception, apart from a fine cotton used in late 1887 [AM2 p.50].
• Arles to Auvers. After ready-made canvases and a local roll, from August 1888 he worked almost only on rolls of his main colorman's ordinary grade, chosen in July 1888 after trying a finer one [WM p.114]. It averages 11.5 × 18, 11.5 × 17 or 12 × 15.5 threads per cm [WM p.116 n.18]. He ordered ten metres at a time, "at 2.50 francs" a metre [L710; L800]. Size 30 was his favourite in Arles [WM p.117]. He found that "this thin canvas perishes after a while and can't take a lot of impasto" [L800].
• Jute. In late 1888 he and a painter friend sharing his Arles house divided a roll of jute and primed it in turn with chalk, barytes and lead white grounds [AM2 p.60]; two chair pictures are "on sailcloth in thick impasto" [L721].
• Auvers. In June–July 1890 he painted landscapes on canvases of about 50 × 100 cm [NGL].
• Dutch period. In Nuenen he cut canvas from rolls of one type (13–14 × 14–16 threads per cm, commercial cream ground) and put it on non-standard stretchers made by a local carpenter [AM2 pp.45–46; WM p.115].

2. Grounds

• Paris: lead white in oil. Most Paris canvases have commercial lead-white oil primings: a single lead-white ground in one or two coats (the commonest), or lead white over a thin, probably glue-bound chalk layer [AM2 pp.50–53]. They hold barium sulphate, pipe clay, chalk or gypsum, with traces of ochre, carbon black and umber that tint them pale [AM2 pp.50–53]. The cartons have smooth pale grey or white grounds [AM2 p.47].
• His own absorbent grounds (1887). From spring 1887 he tried thin, pale preparations on fine, often unsized fabric: pure lead white about 0.05 mm thick; barytes (barium sulphate) on paper laid on cotton or on cotton alone; chalk in glue, once greyed with bone black [AM2 pp.53–54]. He seems to have primed the barytes ones himself [AM2 p.54].
• His own toned grounds (1887). To reuse failed canvases he brushed one mixed batch onto several at once: a warm pinkish brown (lead white, red ochre, vermilion, ultramarine, barium sulphate, emerald green, zinc white, carbon black, an organic red) or a cool pinkish grey (lead white, emerald green, barium sulphate, red ochre, ultramarine) [AM2 pp.57, 61]. A light bluish layer of lead white, zinc white and a little ultramarine covers abandoned designs on three spring landscapes [AM2 p.60]. In late 1887 he used twill with a warm pinkish-grey ground, probably made to order [AM2 pp.57–58].
• Arles to Saint-Rémy. The February 1888 snow landscape has 100–120 µm of chalk and lead white under about 20 µm of lead white [ALT]. His main colorman's canvas of 1888–90 has six kinds of lead-white ground, single or double, some with zinc white and barium sulphate [SALV] (read through a search summary); an 1889 mountain view has a thin, warm-toned lead-white ground [ALT].
• Ground left showing. In early 1887 the light ground shows through thin paint and between touches [AM2 p.58]; in the 1888 snow landscape and the 1889 mountain view bare ground shows between strokes [ALT].

3. Drawing and lay-in

• Perspective frame. He traced the frame itself, a cross with diagonals, onto the canvas. Traced lines were found in 11 or 12 of the 93 paintings examined, all of 1887, usually centred so the horizon splits the picture in half [VGM2 pp.119–124]. In Arles he "made my last three studies with the help of the perspective frame" [L585]; by June 1888 he said he could finish a drawing in an hour without it [VGM2 p.127, citing letter 620].
• Underdrawing. Early 1887 pictures, even thin ones, have detailed graphite or charcoal underdrawing, its outlines often gone over in paint; later 1887 landscapes have only a sketchy frame, the drawing done with the brush in one go [VGM2 pp.126–127]. Once the frame was traced in red paint from the tube [VGM2 p.127]. A Paris landscape is laid out in dilute red lake lines; in the 1888 snow landscape long white strokes set the perspective; a Saint-Rémy view was drawn with a brush in dilute blue paint [ALT].
• Lay-in. In 1886 he usually covered the light ground early with opaque warm underpaint [AM2 p.58]. From 1887 the lay-in is thin veils of color, or there is none [AM2 p.58; VGM2 p.127].
• Dutch period. A Nuenen head has freely sketched lines in IRR; forms were blocked in with thin, fluid paint, then worked from dark to light [NAMA].

4. Pigments

One color often held up to 12 pigments and extenders, some of them the paint maker's [VGM2 p.129]. From Arles on his letters carry paint orders; "Chrome yellow 1, 2 and 3" mean lemon, yellow and orange [L593]. The French "vert émeraude" ("emerald green") was viridian; true emerald green (copper aceto-arsenite) had another French trade name, and he ordered both [VGL-T; L593]. "Silver white" was a lead white [VGM2 p.143 n.183].

Pigment	Evidence	Notes
Lead white ("silver white")	Throughout [AM2 p.50; VGM2 p.143; GEL18]; ordered [L593; L779]	In poppy oil [GEL18]
Zinc white	A separate color [VGM2 p.143; GEL18; MET17]; ordered [L593; L877]	"This zinc white that I'm using now doesn't dry" [L591]
Chrome yellow ("chrome 2")	[VGM2 p.133; GEL18; MON15]; ordered [L593]	Nearly pure lead chromate [GEL18]
Lemon chrome ("chrome 1")	[GEL18; MON15]; ordered [L593]	Half sulphate; light-sensitive [GEL18; MON15]
Chrome orange ("chrome 3")	[VGM2 p.133; NAMA]; ordered [L674]
Zinc yellow	[VGM2 p.133; CLX-ZY]	From mid-1887; "a more transparent lemon yellow" [VGM2 p.133]
Cadmium yellow	[VGM2 p.134; ESRF12; ALT]	Mostly in mixtures [VGM2 p.134]
Cadmium orange	His self-portrait as a painter [VGM2 p.134]	Single object
Yellow ochre	Ordered [L674; L779]; a Saint-Rémy view [ALT]	Earths minor in Paris [VGM2 p.142]
Red ochre, red iron oxide	1887 toned grounds [AM2 p.57]; 1888 [ALT]; ordered [L779]
Raw sienna	One tube ordered [L779]	Single purchase
Umber	Early Paris [VGM2 p.142]; 1888 [ALT]
Vermilion	Most pictures [VGM2 p.135; GEL18]; ordered [L593]
Red lead ("orange lead")	From 1887 [VGM2 p.135; GEL18]; ordered [L593]
Geranium lake (eosin)	Arles to Auvers [GEL18; ALT; MET17; CIN]; ordered often [L593; L877]	On alumina, later also lead sulphate; paraffin wax added [GEL18; MET17; CIN]
Cochineal lake ("carmine")	Paris [VGM2 pp.137–138]; Arles [GEL18]; ordered [L593; L800]	A scarlet tin-and-starch type from 1887 covers better [VGM2 p.137]
Madder and purpurin lakes	Paris [VGM2 p.137]; "the madder lakes" [L595]
Redwood (brazilwood) lake	Paris [VGM2 p.137]; 1889–90 [MET17]	Very fugitive [MET17]
Cobalt violet	Six paintings, early 1887 [VGM2 p.139]
Emerald green	[VGM2 p.140; GEL18; ALT]; ordered [L593; L779]
Viridian	[VGM2 p.140; GEL18]; ordered as "emerald" [L593; L800]	Transparent [VGM2 p.140]
Chrome green ("cinnabar green")	One hidden picture [VGM2 p.141]; ordered once [L593]	A ready mixture
Cobalt blue	[VGM2 p.131; GEL18; ALT; MET17]; ordered [L674; L779]	For bright accents [VGM2 p.131]
Synthetic ultramarine	[VGM2 p.130; GEL18; ALT]; ordered [L674; L779]
Prussian blue	[VGM2 p.130; GEL18; ALT; MET17]; ordered [L593]	Main blue in Nuenen [NAMA]
Cerulean blue	Paris [VGM2 p.130]; 1888 [ALT]
Bone or ivory black	Minor [VGM2 p.141; GEL18]; ordered [L779]
Naples yellow	Nuenen [NAMA]; Paris to summer 1886 [VGM2 p.131]	Dutch palette
Bitumen and bistre	Nuenen [L450]	Dutch palette

• Extenders. In one 1887 picture he added coarse barytes to his tube colors on the palette [VGM2 p.143].
• Unstable on purpose. "All the colours that Impressionism has made fashionable are unstable, all the more reason boldly to use them too raw, time will only soften them too much" [L595].
• Dutch palette. In 1882: "ochre (red, yellow, brown), cobalt and Prussian blue, Naples yellow, terra sienna, black and white" with smaller tubes of carmine, sepia, vermilion, ultramarine and gamboge [L253]. Brighter colors replaced the Prussian blue, Naples yellow and earths at Antwerp in 1885 [VGM2 p.94].

5. Media and varnish

• Tube paints. He used commercial tube paints from Paris colormen [VGM2 pp.94–96]. His whites were ground in poppy oil, his cochineal and eosin in linseed oil, the eosin with some paraffin wax; the eosin and chrome yellow paints seem to have held excess oil [GEL18].
• Less oil. "The more finely a colour is ground, the more it is saturated by oil"; grinding "just long enough to make it workable" would give "colours that were fresher, perhaps darkening less" [L668]. He then ordered "coarsely ground colours, in large tubes" [L674], and wanted lakes "freshly ground, if they're greasy I'll send them back" [L593]. Absorbent grounds also drew oil out [VGM2 p.152].
• Thinner. Early 1887 pictures are painted à l'essence [AM2 p.58]. Later he still thinned: in one study "as the thinner evaporated immediately, the painting doesn't adhere at all firmly to the canvas" [L800]. He had dry canvases washed "to remove the oil and the thinner from the impasto" [L777], then retouched them [L721].
• Varnish. He did not varnish. A study "dry right through, also in the impasto, which will not be the case for a year" could then take "a good coat of varnish" [L853; L863]. Restorers varnished many after wax-resin lining in 1926–33 [VGM2 pp.31–34; GEL18].
• Dutch period. In Nuenen he wrote that thick paint "a year old" has "sweated out" its oil and can then be scraped "quickly with a razor blade" for more lasting color [L538].

6. Layers and application

• 1886: thick, layered. The 1886 Paris pictures were often built in several sessions over opaque warm underpaint, with impasto touches layered in the flower pieces [AM2 p.58].
• Early 1887: thin. Veils of color often under 0.01 mm, rubbed down so color stays in the hollows of the weave; scraping through wet paint with a hard bristle brush to get back the ground; paint dabbed thin with the fingertips; ground left between open touches [AM2 p.58]. He used "pointed, springy brushes" [VGM2 p.74].
• Brushes. Strokes run from under 5 mm to "one of his widest brushes (1.5 cm)" [VGM2 pp.151, 235]. His self-portrait as a painter shows flat hog's-bristle brushes for dragging stiff paint over the weave and springy pointed ones for fine touches [VGM2 p.156]. From Saint-Rémy he ordered "6 fitch brushes, black hair" [L800].
• Palette. A wooden palette of about 35 × 27 cm, said to have been used in Auvers in 1890, keeps traces of dark chrome yellow, Prussian blue and geranium lake [ORS]. Uncertain.
• Straight from the tube. Outdoor studies of 1887 repeat a few tube colors across the picture; elsewhere colors were slurried together within single blobs lifted from the palette [VGM2 pp.152–153].
• Wet in wet, then over dry. In a Saint-Rémy view pure colors were mixed on the canvas over a thin underpainting, the sky painted before the dark mountain outlines, and dots of impasto added over dry paint [ALT]. In 1890 still lifes red lake and white were laid wet on wet [MET17].
• Impasto. He thought the lights should be "painted on heavily" to keep their color [L538]. By December 1889: "I'll do hardly any more things in impasto" [L829]. Read through a search excerpt.
• Dutch period. Thin paint with low impasto, wet into wet with round brushes up to 5/8 inch wide, dark to light [NAMA].

7. Working sequence and reworking

• On the spot. Landscapes were painted out of doors. In wind: "You shove the feet of the easel in and then you push a 50-centimetre-long iron peg in beside them. You tie everything together with ropes" [L628].
• Fast but planned. "Canvases done quickly one after another but calculated long beforehand" [L635]. He planned campaigns of similar subjects and reckoned paint and canvas ahead [WM p.113].
• Retouching later. In the 1888 snow landscape the figure was reworked after the sketch had partly dried; the 1889 mountain view was reworked over partly dried paint [ALT].
• Reuse. 25 of the 93 paintings examined are painted over earlier ones [VGM2 p.113]. He scraped the old picture roughly or covered it with a dark or mid-toned layer, sometimes before it was dry, which roughened the new paint [VGM2 p.115].
• Repetitions. In Saint-Rémy he repainted earlier compositions: "I've redone the canvas of the Bedroom" [L800].

8. Condition and aging

• Faded red lakes. Eosin and cochineal have faded: pinks look white, violets blue [GEL18; ALT; MET17]. In an Arles field most irises now look more blue than purple [GEL18]. In an Auvers landscape 37.7% of 387 now-white flowers hold bromine from eosin and were once pink [CIN]. Fading is worst where lake is mixed with white [MET17]. Thin glazes of tin-based cochineal have almost vanished; thick strokes are cracked and blanched [VGM2 p.138].
• Darkened chrome yellow. Chrome yellow has darkened at the surface to greenish brown or blackish where exposed [VGM2 p.133]; the sulphate-rich lemon type darkens in light [ESRF11; MON15].
• Cadmium yellow. In an 1887 flower piece a varnish added after his death reacted with cadmium yellow, turning flowers orange-grey [ESRF12].
• Red lead. It has darkened, gone ochre or translucent, and formed lead-soap lumps [VGM2 pp.135–136].
• Linings and varnish. Wax-resin lining soaked into porous grounds, darkening grey ones to black and white ones to beige [AM2 p.57]; later varnishes yellowed [VGM2 pp.31–34].

9. The tubes here

The historical names in this note are mapped to these tubes: lead white ("silver white") → lead white; zinc white → zinc white; chrome yellow ("chrome 2") → chrome yellow; lemon chrome ("chrome 1") → lemon chrome; chrome orange ("chrome 3") → orange chrome; zinc yellow → zinc yellow; cadmium yellow → cadmium yellow; cadmium orange → deep cadmium (a proxy, a single object); yellow ochre → yellow ochre; red ochre and red iron oxide → red earth; raw sienna → raw sienna (a single purchase); umber → raw umber; vermilion → vermilion; red lead ("orange lead") → red lead; eosin ("geranium lake") → geranium lake; cochineal ("carmine"), and the "ordinary lake" and "crimson lake" of his orders → carmine (the last two inferred); madder and purpurin lakes → rose madder; cobalt violet → cobalt violet; emerald green → emerald green; viridian (his "emerald") → viridian; cobalt blue → cobalt blue; synthetic ultramarine → ultramarine blue; Prussian blue → Prussian blue; cerulean blue → cerulean blue; bone or ivory black → bone black. Redwood lake has no tube; carmine is the nearest. Chrome green is mixed from chrome yellow and Prussian blue. The Dutch palette's lead antimonate yellow and asphaltum are not here.

10. At this easel

• Support. Plain-weave linen with ground layers laid from the tubes.
• Paint. Pigment from the tubes with an oil-medium share.
• Taking paint off. A brush picks up wet paint as it passes; that is how paint comes off the canvas.
• Everything else the easel does is in the easel guide.

Sources

• [VGM2] Hendriks, E. and van Tilborgh, L., Vincent van Gogh Paintings, Volume 2: Antwerp & Paris 1885–1888 (Van Gogh Museum, 2011): Hendriks's technical essays (pp.91–157, analysis by M. Geldof), treatment history (pp.29–37), catalogue entries; p.74 is from van Tilborgh's essay. Read in the Internet Archive full-text OCR. https://archive.org/details/Vincent-van-Gogh-paintings-Volume-2-Antwerp-Paris-1885-1888
• [AM2] Hendriks, E. and Geldof, M., "Van Gogh's Antwerp and Paris picture supports (1885–1888): reconstructing choices," ArtMatters 2 (2005), pp.39–74. https://www.amjournal.org/publ/11
• [WM] van Tilborgh, L. et al., "Weave matching and dating of Van Gogh's paintings," The Burlington Magazine 154 (2012), pp.112–122. https://people.ece.cornell.edu/johnson/tilborgh_2012.pdf
• [GEL18] Geldof, M. et al., "Reconstructing Van Gogh's palette to determine the optical characteristics of his paints," Heritage Science 6:17 (2018); an Arles painting of May 1888. https://doi.org/10.1186/s40494-018-0181-6
• [ALT] Pozzi, F. et al., "Altered identity: fleeting colors and obscured surfaces in Van Gogh's landscapes in Paris, Arles, and Saint-Rémy," Heritage Science 9:15 (2021). https://doi.org/10.1186/s40494-021-00489-1
• [MET17] Centeno, S. A. et al., "Van Gogh's Irises and Roses: the contribution of chemical analyses and imaging to the assessment of color changes in the red lake pigments," Heritage Science 5:18 (2017); Saint-Rémy, May 1890. https://doi.org/10.1186/s40494-017-0131-8
• [CIN] Fieberg, J. E. et al., "'Paintings fade like flowers': pigment analysis and digital reconstruction of a faded pink lake pigment in Vincent van Gogh's Undergrowth with Two Figures," Applied Spectroscopy 71:5 (2017); Auvers, 1890. Read through a search summary. https://doi.org/10.1177/0003702816685097
• [NAMA] Schafer, M. and Twilley, J., "Vincent van Gogh, Head of a Man, 1885," technical entry, French Paintings and Pastels, 1600–1945, Nelson-Atkins Museum of Art (2023). https://nelson-atkins.org/fpc/post-impressionism/735/
• [SALV] Salvant, J. et al., "Investigation of the grounds of [a Paris colorman's] commercially primed canvas used by Vincent Van Gogh in the period 1888 to 1890," in M. Vellekoop et al. (eds), Van Gogh's Studio Practice (2013), pp.182–201. Read through a search summary. https://www.researchgate.net/publication/265251185
• [ESRF11] ESRF press release, "X-rays show why van Gogh paintings lose their shine" (14 February 2011), on Monico, L. et al., Analytical Chemistry 83 (2011). Read through search summaries. https://www.esrf.fr/news/general/van-gogh/van-gogh-paintings-lose-shine
• [MON15] Monico, L. et al., "Evidence for degradation of the chrome yellows in Van Gogh's Sunflowers," Angewandte Chemie Int. Ed. 54 (2015). Read through search summaries. https://onlinelibrary.wiley.com/doi/abs/10.1002/anie.201505840
• [ESRF12] ESRF press release, "X-rays unravel mysterious degradation of a Van Gogh painting" (September 2012). Read through search summaries. https://www.esrf.fr/news/general/van-gogh-cds/index_html
• [NGL] National Gallery, London, Farms near Auvers (1890), 50.2 × 100.3 cm. https://www.nationalgallery.org.uk/paintings/vincent-van-gogh-farms-near-auvers
• [ORS] Musée d'Orsay, "Palette de Vincent Van Gogh" (c.1890). Read through search summaries. https://www.musee-orsay.fr/en/artworks/palette-de-vincent-van-gogh-98555
• [CLX-ZY] ColourLex, "Zinc yellow." https://colourlex.com/project/zinc-yellow/
• [VGL-T] Vincent van Gogh: The Letters, ed. Jansen, Luijten and Bakker (2009), online glossary entries for the French green names. https://vangoghletters.org/vg/terminology.html

Letters, from the same online edition (English translation), to his brother unless stated; URL pattern https://vangoghletters.org/vg/letters/letNNN/letter.html with NNN the letter number.

• [L253] The Hague, 5 August 1882.
• [L450] Nuenen, mid-June 1884. Read through a search excerpt.
• [L538] Nuenen, 3–4 November 1885.
• [L585] Arles, 16 March 1888. Read through a search excerpt.
• [L591] Arles, c. 1 April 1888. Read through a search excerpt.
• [L593] Arles, c. 5 April 1888: paint and canvas order, with the editors' note on the chrome numbers.
• [L595] Arles, 11 April 1888.
• [L628] To a painter friend, Arles, c. 19 June 1888.
• [L635] Arles, c. 1 July 1888.
• [L668] Arles, 23–24 August 1888, as quoted in [GEL18] note 12.
• [L674] Arles, 4 September 1888: paint order.
• [L710] Arles, 22 October 1888: paint and canvas order.
• [L721] Arles, 19 November 1888.
• [L777] Saint-Rémy, 31 May–6 June 1889.
• [L779] Saint-Rémy, 9 June 1889: paint order.
• [L800] Saint-Rémy, 5–6 September 1889: canvas, paint and brush order.
• [L829] Saint-Rémy, c. 19 December 1889. Read through a search excerpt.
• [L853] To a critic, Saint-Rémy, 9 or 10 February 1890.
• [L863] Saint-Rémy, 29 April 1890: canvas, paint and brush order.
• [L877] Auvers-sur-Oise, 3 June 1890.