What the painters were given

Caspar David Friedrich: painting materials and technique

notes/research/friedrich_materials.md · 214 lines · Research

Given to 7 painters:

Caspar David Friedrich: painting materials and technique

Process and hand (sequence, studies and their transfer, edges, pace, what copies got wrong) are in friedrich_process.md.

Sourced technical report for a physical paint simulator. Every claim cites a source key; full URLs are in the "Sources" list at the end. "Uncertain" marks claims that are unverified, disputed or taken from a single object.

Bottom line

Friedrich painted in oil on fine, plain-weave linen that he bought already primed. The grounds had two to four thin layers: a warm lower layer (red or ocher) under a light, lead-white-rich top layer [KÖR p.284; CATS p.127]. Over a precise graphite or ink underdrawing he laid a very thin underpainting, then only one or two very thin paint layers [SMB-proj; CATS p.127]. He used few pigments. Before about 1820 his key blue was smalt, a semi-transparent cobalt glass. After that he used cobalt blue and chrome yellow [ALF pp.341, 348–349]. His smooth gradations come from stippling and from paint pooling in the fine texture of the ground, not from thick blending [CATS p.127; NG p.56]. No accessible source gives numeric thread counts or paint-film thicknesses in microns for his paintings. The simulator values below for those are labeled proxies or assumptions.

Terms used below. Ground/priming: the preparatory layers between canvas and paint. Cross-section: a microscopic sample cut through all layers. IRR (infrared reflectography): imaging that shows carbon-based drawing under paint. XRF (X-ray fluorescence): elemental analysis. Glaze: a transparent paint layer. Impasto: thick, raised paint. Lining: gluing a second canvas to the back. Craquelure: the network of cracks in paint or varnish.

1. Supports

2. Grounds (preparation layers)

3. Underdrawing

4. Pigments

Pigment Evidence Notes
Lead white Monk/Abbey [CATS p.127]; Winter Landscape snow [NG pp.51, 56] The main white; used in grounds too
Smalt (cobalt potash glass) Monk/Abbey [CATS p.127]; Winter Landscape [NG p.56]; Two Men by the Sea 1817 [ALF] Several grades from pale gray to deep blue, used alone or with lead white [NG p.56]. Average Co about 3 wt% [NPJ25]. Chosen "on account of its semi-transparent property" [CATS p.127]
Ochers and earths (goethite, hematite) [ALF p.348]; [CATS p.127] Yellow and red earths
Vermilion/cinnabar Abbey sky layer [NPJ25]; violet sky layer of The Cross in the Mountains (1807/08) [KÖR fig. 6] Small admixture in pinks and violets
Red iron oxide Pale mauve sky of Winter Landscape [NG p.56] A few particles
Bone black Grass in Winter Landscape [NG p.56] (with smalt, Naples yellow, ochre and possibly Prussian blue)
Naples yellow Grass in Winter Landscape [NG p.56] Disputed: XRF found no antimony in the areas it tested [ALF p.349]
Prussian blue "Possibly" in Winter Landscape grass [NG p.56]; cape in Man and Woman Contemplating the Moon (1818/24), inferred from K and Fe [ALF Table 2]
Cobalt blue Paintings from 1818/24 on [ALF]; "almost entirely replaced smalt… after about 1820" [NG p.56]
Chrome yellow (lead chromate) 1817 and 1818/24 [ALF] Cadmium yellow in the 1817 painting is probably a later retouching [ALF p.349]
Green pigments "a few… green" pigments [CATS p.127]; copper greens and green earth in Dresden paintings [MÄD p.102] Mostly mixed greens (Prussian blue, smalt or cobalt + ochre, Naples yellow or chrome yellow); Rinmann's green rare, c.1819–23. See §9
Copper blues No copper detected, which excludes azurite in the areas tested [ALF p.348] Natural or synthetic ultramarine was not reported

5. Binding media and additives

6. Layer structure and application

7. Varnish

8. Condition and aging

9. Greens

Same conventions as sections 1–8: source keys in brackets, full URLs in the source list below. "Unverified" means I could not see the primary text myself.

Bottom line. Friedrich's greens were mostly mixed on the palette from his blues and yellows. He also used two or three true green pigments sparingly. The Dresden team (14 paintings) writes, in my translation, that "the palette was very sparing" but his greens show a "lush variety" ("üppige Vielfalt"). They list the true greens as "copper-containing pigments and green earth." He "often mixed his green from blue and yellow," using Prussian blue "very often" and smalt and cobalt blue as well. The yellow part came from yellow ocher, Naples yellow or chrome yellow [MÄD p.102]. Green passages often have "several (up to four) layers of different greens" stacked on top of each other [MÄD p.102].

Available greens and dates

Green Date and availability Source
Green earth (Grüne Erde, terre verte) Ancient. Continental painters' manuals mention it through the 19th century (for example Field 1835). "Saxon Earth" appears among its trade names, and "Bohemian earth" was a grade [AP1 pp.143–144, 147]
Verdigris / copper resinate Old. "Rarely found in eighteenth- and nineteenth-century paintings" [AP2 p.132]. Around 1816 it was still made as green "varnishes" (verdigris heated into varnish) [AP2 p.149] [AP2]
Scheele's green (copper arsenite) "Introduced in 1778"; "use as an artist's pigment is little documented" [AP3 p.221]. Turner used it from c.1806–07 [AP3 p.256] [AP3]
Schweinfurt/emerald green Sattler and Russ in Schweinfurt produced it from 1814. It stayed a trade secret until Liebig published the method in 1822 [AP3 p.222]. Winsor & Newton sold it from 1832 [AP3 p.223]. The earliest easel finds in the AP3 table date from c.1832 [AP3 p.259] [AP3; WEB-em]
Rinmann's green (cobalt-zinc oxide) Rinmann, 1780; "poor tinting strength and high cost… kept it in limited use" [WEB-co]
Chromium oxide (opaque) Vergnaud (1831) says it was "rarely used because of its high price." Kühn's earliest finds are three paintings from 1845–50 [AP3 p.275]
Viridian Pannetier made it c.1838. Guignet patented it in 1859. The earliest finds are 1863 [AP3 pp.274–275; CL-vir]
"Green cinnabar" / Zinnobergrün / chrome green (chrome yellow + Prussian blue) Appeared "shortly after" chrome yellow; the name "green cinnabar" is recorded by 1842 [AP1 p.189; AP3 pp.203–204]. Chrome yellow was "only produced commercially in Germany from 1820" [MÄD p.102], so a German tube is plausible only after about 1820 (inferred). Field (1835) called these mixtures "unfit for fine art" [FIELD p.129]

Evidence in Friedrich's paintings

How he painted foliage and grass

Properties for simulation

Hex = oil masstone estimate. Hiding runs 0 (transparent) to 1 (opaque). Tint is tinting strength relative to an average of 1.0. Stiffness runs 0 to 1. Basis: S = sourced, P = proxy, A = assumption.

Pigment Masstone Hiding Tint Stiffness Basis and aging
Green earth #3A4843 0.2 0.3 0.35 Hex P: Munsell 7.5G/2.9/1.5 over white [AP1 Table 1, p.146], converted by me. "Relatively translucent"; tint and hiding "poor" (S) [AP1 p.146]; "not much body" [FIELD p.129]. Oil uptake is 80–100% (Bohemian vs. Verona) [AP1 p.147], so stiffness is A. Stable; can darken as oil penetrates [AP1 p.146]
Verdigris (in oil) #3F7F6A 0.25 1.0 0.4 Hex A. "Poor hiding power in oil" (S) [AP2 p.132]. Dries rapidly; blackens in "moist and impure air" [FIELD p.130]
Copper resinate glaze #2F5A32, aging to #5A4A28 0.05 0.8 0.1 A. Brown discoloration from "photocatalytic reaction" (S) [AP2 p.151]
Scheele's/emerald green #1FA57A 0.7 0.8 0.6 Measured color in parchment size is #00C285 (clipped) [AP3 p.226]; the oil value is darker (P) [AP3 p.227]. "Good hiding power" (S) [AP3 p.226]; "rather opaque"; works "difficultly in oil, and dries badly" [FIELD p.130]. Blackens with hydrogen sulfide [AP3 p.227]
Rinmann's green #5F8F76 0.35 0.4 0.5 Hex A. "Semi-transparent", "poor tinting strength" (S) [WEB-co]; "not very powerful" [FIELD p.129]. Permanent
Prussian blue #172440 0.35 4.0 0.45 Hex A. Refractive index 1.56, "transparent… useful for glazing" [AP3 p.197]. Tint "high" (S): "half an ounce will tint 20 pounds of lead white" [AP3 p.196]. Fades in light and recovers in the dark [AP3 p.199]
Chrome green mixture (Zinnobergrün) #3F6E35 0.75 1.5 0.5 A. Model it as a palette mix of chrome yellow and Prussian blue

Recommendation for the simulator palette

Early (pre-1820). Add Prussian blue (S: "sehr häufig" in greens [MÄD p.102]; "possibly" in grass [NG p.56]) and green earth (S [MÄD p.102], dates unassigned). By default, build greens by mixing: smalt or Prussian blue + ochre or Naples yellow + bone black [NG p.56]. Optional: a "copper green" tube, flagged rare.

Post-1820. Keep the early additions. Cobalt blue + chrome yellow mixes become possible with tubes you already have. Add Rinmann's green as a rare tube (three objects, c.1819–1823).

Leave out:

Rendering: allow up to four stacked green layers [MÄD p.102]. Paint foliage after the sky and grass last.

In the simulator (Palette::friedrich_early_greens, friedrich_1820_greens, copper_green): Prussian blue and green earth join both periods, Rinmann's green the post-1820 one, copper green as an optional rare tube. The base palettes are kept without them (the aimed search picked Prussian blue for skies).

Implications for a simulator

Values are sourced (S), proxy from contemporaries or generic studies (P) or assumption needing tuning (A).

Parameter Value Basis
Weave Plain (tabby) linen, handwoven S [KÖR p.283]; thread irregularity A
Thread density 10–16 threads/cm, so 0.63–1.0 mm spacing. Split early about 11–13/cm, late about 14–16/cm P [CATS-E]; split A; "finer late" is S [KÖR p.283]
Weave relief at surface Low: lower grounds level the weave S [KÖR p.284]
Ground stack (early, Berlin type) Bright red, then 2 × light brown; top layer with fine roller-like texture S [CATS p.127]
Ground stack (typical Dresden) 1–3 ocher/red-earth/chalk/barite layers plus a patchy lead-white top S [KÖR p.284]
Ground texture Top-layer brush striations, directional, visible through paint; occasional scratches S [KÖR p.284]
Ground thickness 150–300 µm total P [NPJ24 modeling range], unverified for Friedrich
Underdrawing Graphite/chalk lines, 2 passes, ruled straights. Keep it visible under thin paint (early) and covered (late) S [CATS; ART]
Paint layers 1–2 layers plus a very thin underpainting S [SMB-proj; NG p.56]
Paint film thickness 5–20 µm per layer in skies; thicker only in highlights and snow A (no measured value found)
Texture pooling Pigment fills ground valleys and thins on peaks, giving dotted, strokeless gradients S [CATS p.127]
Smalt layer optics Low hiding power; use several particle-size grades, with coarse particles visible as specks S [CATS; NG p.56; HH]
Opaque elements Lead-white snow and highlights S [NG pp.51, 56]
Transparent/semi-transparent Smalt skies and mist; early "thin transparent layers"; ground and underdrawing show through S [CATS p.127; NG p.49; HH; KÖR p.284]
Brush modes Stipple (skies/mist), short hatch (conifers), fine upward flick (grass, applied last), rare light impasto S [NG pp.50, 56]
Palette pre-1820 Lead white, smalt, yellow/red ocher, vermilion, bone black, umber (grounds); Naples yellow and Prussian blue uncertain S [NG; ALF; NPJ25]
Palette post-1820 Add cobalt blue and chrome yellow; smalt declines S [ALF; NG p.56]
Original varnish Very thin egg-white film, then mastic S [KÖR p.283]
Aged varnish Up to 7 yellowed layers; microcracked varnish scatters light S [SMB-blog]
Smalt aging Blue fades toward gray/brown; binder shows through S [NPJ25; HH]
Crack spacing / island size Center: 2–6 mm isotropic network (A). Corners: cracks perpendicular to the diagonal, 6 ± 3 mm spacing, within 5–10% of the diagonal (P) [NPJ24]
Abrasion Micro-losses exposing ground on peaks of texture S [SMB-blog]; peak location A

Sources

Source