What is carrageenan and why does its concentration matter so much for paper marbling size?

Carrageenan is a family of linear sulfated polysaccharides extracted from red algae, particularly Chondrus crispus (Irish moss) and Kappaphycus alvarezii. The kappa-carrageenan fraction used for marbling size consists of an alternating disaccharide repeat: D-galactose-4-sulfate (with a sulfate ester at the C4 position) alternating with 3,6-anhydro-D-galactose. This repeat unit polymerizes into linear chains with molecular weights of 400–600 kDa. The sulfate ester groups on the galactose residues carry a fixed negative charge at ambient pH, giving kappa-carrageenan its polyelectrolyte character. In hot water (above ~60°C), kappa-carrageenan chains adopt a random-coil conformation and the solution is freely fluid. On cooling below a critical temperature (approximately 40°C in distilled water, somewhat lower in the presence of K⁺ or Ca²⁺ ions), the chains undergo a conformational transition from random coil to double helix: two chain segments wind together in a right-handed helix, stabilized by hydrogen bonding between the hydroxyl groups of opposing galactose and anhydrogalactose residues. Critically, K⁺ ions bind specifically within the cavities between adjacent double helices, screening the electrostatic repulsion between the negatively charged sulfate groups that would otherwise prevent helix-helix association. This allows double helices to pack laterally into larger aggregates called junction zones — the physical cross-links that give the gel its mechanical integrity. The gel network is therefore a combination of double-helix formation (physical bonds within the polymer chain) and helix-helix aggregation into junction zones. The working concentration for marbling size, 2–4 g/L (0.2–0.4% by weight), sits in a narrow optimal window for two physical reasons. Below 2 g/L the gel network is too sparse and weak: the junction-zone density is insufficient to support a stable gel at room temperature, the viscosity is too low to support floating paint drops on the surface without them sinking, and stylus-dragged patterns close back together too quickly before the composition is transferred to paper. Above 4 g/L the gel is too stiff: stylus drag creates turbulence and surface rupture rather than smooth distortion, and the increased surface rigidity prevents paint from spreading in the large, even circles that sharp marbled patterns require. Temperature matters for similar reasons. At the optimal working temperature of 18–22°C, the kappa-carrageenan has maximum gel structure and viscosity: the double helices are fully formed and the junction zones well-developed. The size holds pigment patterns in position for minutes, allowing careful stylus work and comb passes without the pattern diffusing or collapsing. Below 15°C the gel is stiffer still and patterns are very persistent but spreading of freshly dropped paint is more limited — a useful property if very fine patterns are desired but not for normal work. Above 25°C the gel structure weakens as helices partially melt back to coil, reducing viscosity, and patterns diffuse more quickly, limiting the time available for stylus work after each color is laid. Practical preparation requires mixing the carrageenan powder into cold water first (to prevent surface hydration forming a dry skin that traps powder inside), then heating to 85–90°C while stirring constantly until the powder is fully dissolved. The size is then cooled and rested for 12–24 hours to allow complete polymer hydration and full gel network development. Surface foam — produced by the polysaccharide acting as a foaming agent at the air-water interface during preparation — must be skimmed before use, as foam disrupts the smooth size surface on which painting floats.

How does the Marangoni effect explain ox gall spreading in paper marbling, and why do successive colors require more ox gall?

The Marangoni effect is the flow of a fluid driven by a gradient in surface tension across a liquid surface. In paper marbling, ox gall is the agent that creates this gradient, and understanding the mechanism explains both why ox gall works and why its required amount scales geometrically across successive colors. Ox gall is an aqueous solution of bile acids produced by the gallbladder of cattle, primarily sodium taurocholate, sodium deoxycholate, and sodium chenodeoxycholate. These are steroidal surfactants — molecules built on the four-ring A-B-C-D sterane backbone (cyclopentanoperhydrophenanthrene), with hydroxyl groups on the concave alpha face of the steroid backbone providing hydrophilicity, and methyl groups and aliphatic side chains on the convex beta face providing hydrophobicity. This asymmetric amphiphilic structure allows bile salts to adsorb at the air-water interface and dramatically reduce surface tension. The critical micelle concentration (CMC) of sodium taurodeoxycholate is approximately 2 mM; at this concentration and above, bile salt molecules aggregate into micellar structures in the bulk solution. Below the CMC, individual bile salt molecules adsorb at the air-water interface, with their hydrophilic alpha face oriented toward the aqueous phase and their hydrophobic beta face toward air. The surface tension of a carrageenan size solution (without surfactant) is approximately 68 mN/m — similar to that of pure water (72 mN/m) since the carrageenan polysaccharide at low concentration does not significantly alter the air-water interfacial energy. When ox gall is added to marbling paint and a droplet of that paint is placed on the size surface, the bile salt molecules in the droplet rapidly migrate to the liquid-air interface around the droplet perimeter. Within milliseconds, the interface immediately surrounding the droplet perimeter is loaded with bile salt, reducing local surface tension from ~68 mN/m (size alone) to approximately 30 mN/m at the ox-gall-rich interface. This creates a sharp surface tension gradient: ~68 mN/m in the surrounding size surface (no surfactant) and ~30 mN/m at the droplet perimeter (bile-salt-loaded interface). Surface tension is a force per unit length tangential to the interface: the high-tension size surface pulls inward, and the low-tension ox-gall-loaded interface at the droplet perimeter is pulled outward by this gradient. The net result is a Marangoni flow — fluid is swept outward from the high-tension zone toward the low-tension zone, carrying the paint droplet outward and expanding it rapidly across the size surface. The expansion continues radially outward as the bile salt monolayer spreads with the expanding paint disc, until the bile salt is diluted below its effective surface-tension-reducing concentration by the expanding area of the monolayer, at which point the Marangoni driving force dissipates and spreading stops. The practical result is a large, thin, flat circle of paint on the size surface — the characteristic open ring of a freshly dropped color in paper marbling. Why successive colors require more ox gall follows from the physics of competing monolayers. The first color deposited on the size surface leaves behind a bile salt monolayer — a roughly monomolecular layer of bile salt molecules covering the size surface in the area where the paint was placed. This monolayer has already reduced the local surface tension of the covered area from ~68 to ~30–40 mN/m. When a second color is dropped onto the size surface that already carries the first color's bile salt monolayer, the Marangoni driving force for the second color depends on the surface tension differential between the pre-existing bile-salt-covered area (low tension, ~35 mN/m) and the area around the second droplet (also low tension if it lands on the existing monolayer, ~35 mN/m). There is essentially no gradient — and therefore no Marangoni flow — unless the second color carries a higher bile salt concentration sufficient to reduce surface tension further below the existing monolayer. To achieve adequate spreading of color 2 against the pre-existing color 1 monolayer, color 2 must carry enough ox gall to lower surface tension from ~35 mN/m (existing monolayer) to something significantly lower, perhaps ~20–25 mN/m, to regenerate a usable gradient. This requires roughly double or triple the ox gall concentration of color 1. Color 3, landing on an even more bile-salt-saturated surface (both color 1 and color 2 monolayers), needs still more. The empirical ratio observed by experienced marbling artists follows an approximately geometric progression: if color 1 requires 1 drop of ox gall per 10 mL of paint, color 2 requires 2–3 drops, color 3 requires 4–6 drops, color 4 requires 8–12 drops, and so on roughly doubling at each step. This geometric progression is a direct consequence of the Gibbs adsorption isotherm (discussed further in the multi-color compression section): each additional color saturates the size surface monolayer further, and diminishing marginal returns mean that dramatically more surfactant is needed to achieve each incremental reduction in surface tension. This thermodynamic limit is why practical marbling palettes are limited to 5–7 colors before the ox gall requirements become counterproductive — at high enough concentrations, the ox gall itself causes paint to sink below the size surface rather than float on it.

Explainers · 2026-07-09 · ~3,900 words

Patreon for paper marbling creators: carrageenan gel surface chemistry, Marangoni ox gall spreading, alum mordanting electrostatics, stylus vortex dynamics, ebru vs suminagashi technique physics, multi-color compression, and the Apple Tax

Q: How does the Apple Tax affect paper marbling Patreon creator income?

Paper marbling creators build audience primarily through Pinterest (dominant for marbled paper aesthetics — endpaper images, stationery, bookbinding — 72–82% iOS) and Instagram (marbling process photography and Reels — 70–78% iOS). TikTok marbling reveals (the comb-drawn peacock or gel-git pattern filmed overhead and revealed in time-lapse) track at 74–84% iOS. YouTube marbling tutorials sit lower at 58–68% iOS as some viewers watch on desktop while mixing size or preparing materials. Beginning November 1, 2026, Apple charges Patreon 30% on every subscription payment processed through the iOS app. In dollar terms: at $200/month with 74% iOS, approximately $44.40/month ($532.80/year) in Apple fees. At $350/month with 78% iOS, approximately $81.90/month ($982.80/year). At $500/month with 80% iOS, approximately $120/month ($1,440/year). Enable Patreon's web-only billing toggle in Creator Settings before October 31, 2026, and update all social media bio links — Pinterest profile, Instagram bio, TikTok bio — to the Patreon web URL. Verify the subscription flow from an iPhone browser (not the app) before November 1 to confirm the web-only path is live.

Paper marbling Patreons retain subscribers when they deliver the physics and chemistry layer that process videos structurally compress out: what makes carrageenan gel the ideal floating medium at the molecular level, how ox gall exploits the Marangoni effect to expand paint across the size surface, why alum-treated paper electrostatically attracts a floating pigment layer that plain paper repels, and the Stokes-flow vortex mathematics behind every comb and stylus pattern. The marbling audience is heavily iOS across Pinterest, Instagram, and TikTok — the November 1, 2026 Apple Tax warrants action before October 31.

Carrageenan gel surface chemistry: polymer structure, gelling, and viscosity control

The carrageenan size that makes paper marbling possible is not merely a thickened water: it is a physically cross-linked polymer gel whose molecular architecture is precisely why paint floats on its surface, stays where stylus and comb place it, and transfers cleanly to paper on contact. Understanding the gel chemistry makes every troubleshooting decision comprehensible.

Polymer structure: Kappa-carrageenan is extracted from red algae — principally Chondrus crispus (Irish moss, from North Atlantic seaweeds) and Kappaphycus alvarezii (from tropical Indo-Pacific aquaculture). The polymer chain is built from a strictly alternating disaccharide repeat unit: D-galactose-4-sulfate (a galactose residue bearing a sulfate ester group at its C4 carbon, connected via a 1→3 glycosidic bond to the next sugar) alternating with 3,6-anhydro-D-galactose (a galactose residue in which the C3 and C6 oxygens are bridged to form a bicyclic anhydro sugar, connected back via a 1→4 glycosidic bond). The sulfate ester at C4 of each galactose residue carries a fixed negative charge at ambient pH. The linear chain has a molecular weight of 400–600 kDa, corresponding to approximately 700–1,100 disaccharide repeat units per chain. Iota-carrageenan and lambda-carrageenan are related polysaccharides with different degrees and positions of sulfation; kappa-carrageenan, with one sulfate per disaccharide, gels most firmly of the three and is the fraction used for marbling size.

Gelling mechanism: In hot water above approximately 60°C, kappa-carrageenan chains adopt a random-coil conformation — the chain backbone rotates freely around glycosidic bonds, entropy drives a disordered, extended conformation, and the solution is a freely flowing liquid. On cooling below the coil-to-helix transition temperature (approximately 40°C in distilled water; lower in the presence of K&spplus; or other monovalent cations that stabilize the helix), the chain segments adopt a right-handed double-helix conformation: two chain segments of kappa-carrageenan wind around each other in a coaxial helix, stabilized by hydrogen bonds between the hydroxyl groups of opposing saccharide residues in the two strands. The 3,6-anhydro-D-galactose residue is particularly important here — its bicyclic anhydro ring enforces a rigid, fixed conformation on that part of the chain, providing the geometric template around which the double helix can form. The sulfate groups at C4 of the galactose residues are positioned on the exterior of the helix, facing the surrounding water.

K&spplus; ions and helix aggregation: The sulfate groups on the exterior of each double helix are negatively charged, and two adjacent helices would normally repel each other electrostatically. K&spplus; ions (potassium cations) bind specifically within the grooves between adjacent double helices — the size of K&spplus; (ionic radius 1.38 Å) fits precisely into these interhelix cavities, whereas the larger Cs&spplus; and Rb&spplus; bind less effectively and the smaller Na&spplus; and Li&spplus; are too small to bridge effectively. The bound K&spplus; ions screen the sulfate-sulfate electrostatic repulsion between neighboring helices and allow the helices to aggregate laterally into larger bundles, called junction zones. Junction zones are the physical cross-links of the gel network: domains where multiple double helices are packed together in register, held by the screened electrostatic and van der Waals interactions between the helix surfaces. Between junction zones, individual double-helix segments act as flexible links of the gel network. Mg²&spplus; and Ca²&spplus; also promote gelling by non-specific charge screening but are less selective than K&spplus;; hard water that naturally contains dissolved Ca²&spplus; and Mg²&spplus; will gel kappa-carrageenan at somewhat lower polymer concentration than distilled water.

Working concentration for marbling size: The optimal concentration range is 2–4 g/L (0.2–0.4% by weight). At this concentration the gel network is just stiff enough to support floating paint drops and hold stylus-drawn patterns in position, but soft enough for the smooth, large-radius Marangoni spreading of paint drops and for stylus drag to create cleanly defined lines rather than tearing the surface. Below 2 g/L, junction-zone density is insufficient at room temperature and the viscosity too low: freshly dropped paint sinks instead of floating, and comb-drawn patterns close together or diffuse before paper transfer. Above 4 g/L, the gel surface is too rigid: paint spreads in smaller circles (the stiffer gel resists the Marangoni-driven flow), stylus drag produces turbulent surface disruption rather than smooth pattern, and the high surface modulus causes the transfer to paper to be incomplete in fine-detail regions. The concentration-to-viscosity relationship is steep in this range — a small error in weighing (say, 5 g/L instead of 3 g/L) produces a noticeably stiffer, less workable size.

Temperature and preparation: At the optimal working temperature of 18–22°C, kappa-carrageenan gel has its maximum junction-zone density and highest viscosity for a given concentration; the gel structure is fully developed and patterns hold for minutes before diffusing, giving ample time for multi-step comb and stylus work. Below 15°C the gel is firmer still — patterns are very persistent but spreading of dropped paint is reduced and stylus drag feels stiffer. Above 25°C, helices begin to melt back to coil, viscosity drops, and patterns start to diffuse and close together within 30–60 seconds, limiting complex stylus work. For preparation, the powder must be added to cold water (not hot) to prevent surface hydration before the interior of each particle wets through; the slurry is then heated to 85–90°C with constant stirring until no particles remain visible (typically 10–20 minutes). The hot size is cooled to room temperature and rested 12–24 hours to allow full polymer hydration and gel network equilibration. Foam that develops on the surface during heating and cooling must be skimmed before use; the carrageenan polysaccharide is an excellent foaming agent at the air-water interface, and residual foam interferes with the uniform, clean size surface on which marbling depends.

Ox gall Marangoni spreading: bile salt surfactant mechanism and geometric ox gall scaling

The Marangoni effect — fluid flow driven by a surface tension gradient across a liquid surface — is the fundamental physical mechanism by which marbling paint spreads into large, thin, flat circles rather than staying as small, dense droplets. Ox gall is the tool that creates this gradient, and its chemistry determines every practical spreading behavior.

Ox gall composition and steroidal surfactant structure: Ox gall is an aqueous solution of bile acids from the gallbladder of cattle. The principal components are sodium taurocholate (the taurine conjugate of cholic acid), sodium deoxycholate (the taurine conjugate of deoxycholic acid), and sodium chenodeoxycholate. These bile salts are steroidal surfactants — structurally distinct from the more familiar linear alkyl surfactants (such as sodium dodecyl sulfate, SDS) in that their hydrophobic and hydrophilic domains are built into opposite faces of a rigid bicyclic steroid backbone. The sterane backbone consists of four fused rings (labeled A, B, C, and D in steroid nomenclature): three six-membered cyclohexane rings (A, B, C) and one five-membered cyclopentane ring (D). In bile salts, the hydroxyl groups (two or three, depending on the specific bile acid) are located on the concave alpha face of this ring system, creating a hydrophilic face, while the methyl substituents and the aliphatic side chain (with a carboxylate or sulfonate group conjugated to glycine or taurine) protrude from the convex beta face, which is predominantly hydrophobic. This facial amphiphilicity — one face hydrophilic, the other hydrophobic — is fundamentally different from head-tail surfactants and gives bile salts unusual aggregation behavior (forming non-spherical, back-to-back dimers and small oligomers rather than classical spherical micelles). The critical micelle concentration of sodium taurodeoxycholate is approximately 2 mM — considerably higher than that of most synthetic surfactants. At concentrations above the CMC, bile salt molecules form micellar aggregates in the bulk aqueous phase; at lower concentrations, monomeric bile salt molecules adsorb at the air-water interface. The surface tension of bile salt solution at typical working concentrations is approximately 30 mN/m, compared to approximately 72 mN/m for pure water and approximately 68 mN/m for a carrageenan size solution (the polysaccharide at low concentration does not significantly alter the air-water interfacial energy).

Marangoni spreading mechanism: When a paint droplet containing ox gall at a concentration above the CMC is placed on the size surface, bile salt molecules at the droplet periphery rapidly migrate from the droplet bulk to the liquid-air interface in the surrounding region. Within milliseconds of contact, the local surface tension at the droplet perimeter drops from approximately 68 mN/m (size alone) to approximately 30 mN/m (bile-salt-loaded interface). This generates a surface tension gradient: high tension (68 mN/m) in the surrounding size surface, low tension (30 mN/m) at the droplet perimeter. Surface tension acts as a force per unit length tangential to the interface, directed from low toward high tension. The gradient drives a Marangoni flow: a rapid outward spreading of the interface at the droplet perimeter, dragging both the size surface layer and the overlying paint outward from the droplet center. The paint droplet expands radially as its perimeter is continuously swept outward. Expansion continues until the bile salt monolayer has spread over a sufficiently large area that the local bile salt surface concentration drops below the level needed to maintain the surface tension gradient — at which point the Marangoni driving force dissipates and the paint circle stops expanding. The result is a large, thin, even disk of paint on the size surface. The correct amount of ox gall produces a circle that spreads to 10–25 cm diameter (depending on the tray size, size viscosity, and paint consistency) and then holds its position stably. Too little ox gall: the paint droplet spreads only slightly (3–5 cm) and remains as a dense, opaque globule. Too much ox gall: the paint spreads extremely aggressively, becomes very thin, and the pigment particles lose the interfacial support of the size surface and sink below it — the paint “dives” and is lost.

Why successive colors require geometrically more ox gall: Each color deposited on the size surface leaves a residual bile salt monolayer covering the area where it spread. This pre-existing monolayer has already reduced the local surface tension of that area from 68 mN/m to perhaps 35–40 mN/m. When the second color is dropped, it lands on (or adjacent to) this pre-existing low-surface-tension area. The Marangoni driving force for the second color equals the surface tension differential between the surrounding pre-existing monolayer area (~35–40 mN/m) and the bile-salt-rich perimeter of the new droplet. For the second color to generate sufficient spreading force, it must carry enough ox gall to reduce surface tension below ~35 mN/m — a smaller absolute reduction than the first color needed, but against a lower baseline, requiring more surfactant to achieve the necessary marginal reduction. The empirical consequence, well-known among marbling artists, is a roughly geometric doubling: if color 1 requires 1 drop of ox gall per 10 mL of paint, color 2 requires 2–3 drops, color 3 requires 4–6 drops, color 4 requires 8–12 drops. This geometric progression reflects the Gibbs adsorption isotherm behavior of the bile salt monolayer (discussed in detail in the multi-color compression section below) and is the reason practical marbling palettes are limited to 5–7 colors: beyond that, the ox gall loading required for adequate spreading crosses the threshold where excess bile salt causes paint to sink rather than float.

Alum mordanting electrostatics: the mechanism of pigment transfer to paper

The transfer of a marbled pattern from the floating size surface to paper appears instantaneous — press the paper down, lift it up, and the entire floating design is now adhered to the paper surface. The mechanism is electrostatic, and understanding it explains why alum pretreatment is essential and what goes wrong when it is done incorrectly.

Why plain paper fails: The surface of cotton or wood-pulp paper fibers carries a net negative charge at ambient pH (5–7). This charge arises from carboxylate groups (–COO⊃;−) formed by oxidation of cellulose hydroxyl groups during pulp processing, phosphate ester groups from residual pulp chemicals, and anionic sizing agents (rosin, alkylketene dimer) applied to the paper surface during manufacture. The carrageenan-pigment complex floating on the size surface also carries a fixed negative charge from the sulfate ester groups (–OSO₃⊃;−) on the kappa-carrageenan polymer. When negatively charged paper fiber approaches the negatively charged carrageenan-pigment surface, electrostatic repulsion prevents adhesion: the pigment-laden size surface is pushed away from the paper rather than transferring to it. The result with untreated paper is that the marbled pattern does not transfer: the paper lifts away carrying no or almost no pigment, and the floating pattern remains on the size surface nearly intact.

Alum pretreatment: Aluminum potassium sulfate KAl(SO₄)₂·12H₂O (potassium alum) or aluminum sulfate Al₂(SO₄)₃ is dissolved in water at a concentration of 10–15 g/L. This solution is applied to the paper surface by sponge or brush and allowed to dry completely before use. The chemistry of the alum mordant involves aluminum ion hydrolysis. In aqueous solution at ambient pH (4–6), the Al³&spplus; ion undergoes stepwise hydrolysis: the first step, Al³&spplus; + H₂O → Al(OH)²&spplus; + H&spplus;, produces the mononuclear aluminum monohydroxide species Al(OH)²&spplus;, which carries a net charge of +2. At slightly higher pH, a second hydrolysis step, Al(OH)²&spplus; + H₂O → Al(OH)₂&spplus; + H&spplus;, produces Al(OH)₂&spplus; with a net charge of +1. These positively charged aluminum hydroxide species adsorb strongly onto the negatively charged cellulose fiber surface through electrostatic attraction and ligand exchange (the aluminum coordinates to the cellulose oxygen atoms), forming a monolayer coating of fixed positive charge on the paper surface. This is mechanistically identical to aluminum mordanting in textile dyeing (where Al³&spplus; bridges the negative dye molecule to the negative fiber surface), here adapted to reverse the electrostatic charge of the paper surface rather than bridge two negatives.

Transfer mechanism: When alum-treated paper (carrying fixed positive cationic charge from adsorbed Al(OH)⊃x+;(3–x) species) is pressed down onto the size surface (carrying fixed negative anionic charge from carrageenan sulfate esters), the opposite charges attract each other electrostatically across the contact interface. The carrageenan-pigment complex adheres to the alum-treated paper surface essentially instantaneously on contact — the electrostatic attraction is strong enough at the ionic concentrations present to overcome any remaining repulsive interactions and the small kinetic barrier to molecular contact. A press-and-lift time of 15–30 seconds is sufficient for complete pattern transfer; longer contact times do not significantly improve transfer but can cause slight lateral blurring as carrageenan continues to wick into the paper. The carrageenan itself acts as a binder, carrying the embedded pigment particles physically to the paper surface along with the electrostatic adhesion. After transfer and rinsing (to remove excess carrageenan and non-transferred pigment), the pigment is mechanically encapsulated by the carrageenan binder on the paper fiber surface and further held by the dried polymer film.

pH effects and common errors: The alum mordanting chemistry is sensitive to pH. The optimal alum solution pH is 3.5–5: in this range, the Al(OH)²&spplus; and Al(OH)₂&spplus; hydrolysis products are the dominant species, coating the fiber surface with positive charge. Below pH 3.5, unhydrolyzed Al³&spplus; is dominant; while Al³&spplus; is positively charged, it is more strongly solvated by water (hydration shell) and adsorbs less effectively to fiber surfaces than the partially hydrolyzed species, resulting in weaker mordanting. Above pH 6, Al³&spplus; undergoes further hydrolysis to neutral Al(OH)₃, which precipitates as a white flocculent (the familiar “alum floc” used in water treatment) rather than adsorbing to fiber, entirely losing the positive-charge coating function. A common practical error is using tap water with high Ca²&spplus; and Mg²&spplus; hardness to prepare the alum solution: these divalent cations compete with Al(OH)⊃x+; species for adsorption sites on the negatively charged cellulose, partially blocking the mordant deposition and producing weaker, less uniform transfer. Using distilled or deionized water to prepare the alum mordant solution avoids this competition. A second common error is insufficient drying of the alum-treated paper before marbling: wet alum-paper still carries free Al³&spplus; in solution at the surface, which can cross-contaminate the size and precipitate carrageenan sulfate groups, locally disrupting the gel structure and causing pattern defects near areas where wet paper contacted the size.

Stylus vortex dynamics: irrotational flow and pattern mathematics

The stylus — a pointed metal rod, a hair comb, or a single wooden toothpick — is the instrument that transforms the concentric rings of ox-gall-dropped paint into the complex geometric patterns that define the Turkish ebru tradition. The physics governing stylus-induced pattern distortion is well-described by viscous flow mechanics at very low Reynolds number.

Stokes flow and the dipole field: Pulling a stylus through the carrageenan size at a typical working velocity of 3–10 cm/s creates a flow disturbance in the gel. The key physical parameter is the Reynolds number: Re = ρUL/μ, where ρ is the density of the size (~1,000 kg/m³), U is the stylus speed (say 0.05 m/s), L is the stylus diameter (say 0.002 m), and μ is the dynamic viscosity of the gel (which at 3 g/L kappa-carrageenan at 20°C is approximately 0.1–1 Pa·s, orders of magnitude higher than water’s 0.001 Pa·s). Re = (1,000 × 0.05 × 0.002) / 0.5 ≈ 0.2 — well below 1. At Re << 1 (the Stokes flow or creeping flow regime), viscous forces completely dominate inertial forces. There is no turbulent wake behind the stylus; instead, the flow field is a Stokes dipole (or Stokeslet), a smooth, irrotational flow pattern in which fluid displaced from in front of the stylus flows around both sides and fills in behind it. The paint pattern embedded in the size surface is carried along by this dipole flow: the paint on the sides of the stylus path is swept outward and backward, while the paint directly in the stylus path is displaced to either side. The result is the characteristic sharp “parted” line in the paint pattern at the stylus path, flanked by outward-curved bulges where paint was pushed aside. Faster stylus motion sweeps a wider zone and produces more aggressive pattern deformation; slower motion produces a narrower, more precise line with less lateral disturbance.

Pattern vocabulary: The standard ebru pattern vocabulary builds complexity through successive operations on the size surface, each exploiting the Stokes-flow distortion in a controlled way. The stone (Turkish: taš) pattern — the simplest — consists of ox-galled paint drops applied without any stylus work: the concentric ring of each color sitting inside the previous color’s ring produces a bull’s-eye pattern. The base comb (gel-git, “come and go”) pattern is produced by pulling a wide, coarsely-toothed comb through the stone pattern in a series of parallel passes alternating in opposite directions (left-right-left or front-back-front across the tray): each pass distorts the rings into a series of parallel waves. The alternating direction of successive passes produces the wavy, horizontal-stripe character of the gel-git. The chevron or nonpareil pattern is produced by following the gel-git base comb pattern with a second, finer-toothed comb dragged at 90° to the first: the perpendicular fine comb distorts the wavy horizontal stripes into the herringbone chevron motif with fine vertical undulations. The peacock feather (Turkish: tavus tüyü) is produced by starting from the stone pattern and dragging a single stylus point through the center of each ring set: the point distorts each bull’s-eye into a teardrop shape, producing the characteristic curved spine with lateral fronds of the peacock feather. The wing or bouquet pattern is produced by multiple crossing stylus paths at varied angles, creating asymmetric curved interlocking shapes across the entire tray surface.

Comb geometry and pattern spatial frequency: The spatial frequency of the pattern — how many stripes, feathers, or chevrons fit across the paper — is controlled directly by comb tooth spacing. A comb with tooth spacing of 5 mm produces stripes with a spatial frequency of approximately 2 stripes per centimeter; a comb with 15 mm tooth spacing produces approximately 0.67 stripes per centimeter. The tooth diameter affects line sharpness: a fine-pointed tooth (0.5 mm diameter) creates a narrow, sharp-edged distortion line; a blunt tooth (2 mm diameter) creates a wider, softer-edged distortion. The depth to which comb teeth penetrate the size also matters: teeth that barely skim the surface create a surface-only distortion; teeth that penetrate 1–2 cm below the size surface create a deeper, three-dimensional distortion. Because kappa-carrageenan at working concentrations is a soft gel rather than a simple liquid, the distortion is largely permanent: the gel’s elastic character partially “sets” the pattern in place after comb withdrawal, which is why marbled patterns hold for minutes before diffusing. At warmer temperatures where gel elasticity is reduced, patterns diffuse more quickly after stylus withdrawal.

Ebru vs suminagashi: two marbling physics traditions

Paper marbling independently developed in two distinct cultural traditions — the Turkish ebru tradition in Ottoman art and the Japanese suminagashi tradition — that arrived at superficially similar visual results (floating-ink patterns transferred to paper) through fundamentally different physical mechanisms. The physics comparison illuminates both traditions.

Turkish ebru: Ebru (from Persian abru, “water surface” or “cloud-like”) is the Ottoman Turkish marbling tradition, documented in manuscripts and book endpapers from at least the 15th century, and practiced as a high art in Istanbul through the present day. The technical elements are precisely those described in this guide: a carrageenan (or gum tragacanth in historical practice) size prepared in a shallow rectangular tray; water-based gouache, watercolor, or finely ground mineral pigments (earth pigments, malachite, lapis lazuli in historical work) with ox gall bile salt added as the surfactant to control spreading; a stylus, needle, and comb as pattern-making tools; and alum-pretreated paper for transfer. The carrageenan size provides a stable, high-viscosity surface on which paint floats indefinitely without sinking or diffusing if the ox gall level is correctly calibrated. The Marangoni-driven spreading allows precise control of circle size for each color. The stylus and comb, operating in the Stokes-flow regime described above, permit complex deterministic geometric patterns. The electrostatic transfer to alum paper ensures complete, sharp-edged pattern transfer in a single contact. Ebru was used primarily for book endpapers, official Ottoman document backgrounds (fermans), calligraphy practice paper, and decorated letter paper in the Ottoman chancery tradition.

Japanese suminagashi: Suminagashi (墨流し, “ink floating”) is the Japanese marbling tradition, documented in the Heian period (794–1185 CE) and associated with the washi decorated paper tradition. It differs from ebru in a fundamental physical respect: suminagashi uses plain water — no carrageenan, no thickener, no gel — as the floating medium. The ink used is traditional Japanese sumi ink, prepared by rubbing an ink stick (sumi) against an ink stone (suzuri) with water: the ink contains ultra-fine carbon particles (derived from pine wood soot or lamp soot) dispersed in a binder of animal hide glue (nikawa, prepared by boiling and gelling collagen from animal hides — essentially a Japanese hide glue similar to rabbit skin glue used in Western oil painting grounds). The nikawa in the sumi ink is the key to its surface behavior: at the air-water interface, the nikawa forms a very thin film — a viscoelastic protein monolayer — that gives the ink drop a slight surface tension reduction relative to the plain water around it. In suminagashi practice, drops of diluted sumi ink and drops of plain water (or very dilute sake, which contains dissolved ethanol and organic esters that slightly reduce surface tension) are alternated on the plain water surface from fine brushes. Each fresh water drop, when placed adjacent to an existing ink drop, has a slightly lower surface tension than the ink film edge — the differential surface tension between the clean water drop and the older, partially evaporated or oxidized ink surface causes the ink at the contact boundary to be swept outward. The pattern is built by patiently alternating ink and water drops at the same or different positions, building up a system of concentric floating rings as each addition slightly compresses or expands the previous rings. Paper is laid on the surface and the floating ink transfers by the same electrostatic mechanism as in ebru (alum-treated washi, or the natural negatively charged washi surface against positively charged components in the nikawa binder).

Key physics comparison: The fundamental physical difference between ebru and suminagashi is the agent and magnitude of the Marangoni driving force. Ebru relies on exogenously added surfactant (ox gall bile salts) producing a large surface tension reduction (from ~68 mN/m to ~30 mN/m, a Δγ of ~38 mN/m) and correspondingly strong Marangoni flow. This produces bold, well-defined, controllable spreading: large circles, sharp edges, and strong response to comb and stylus. Suminagashi relies on a differential surface tension between materials already present (nikawa-containing ink vs plain water or dilute sake), producing a much smaller surface tension gradient (perhaps 2–5 mN/m between fresh and aged ink drops). This produces delicate, slow, subtle spreading: small, softly-edged, flowing rings that build through patient accumulation of many drops. The absence of a thickening agent in suminagashi means the floating ink is on plain water with essentially no viscosity above water’s own; patterns are far more sensitive to air currents, breath, and vibration, giving suminagashi its characteristic soft, organic, less precisely controllable aesthetic. Ebru patterns can be held on carrageenan size for minutes without diffusing; suminagashi patterns on plain water must be transferred to paper within seconds to minutes before diffusion and surface currents blur them. Both traditions use alum-mediated electrostatic transfer to paper, though the mechanism is identical even though the historical discovery was independent.

Multi-color compression and the Gibbs adsorption isotherm

Adding successive colors to a marbling tray does not simply accumulate independent floating layers: each successive color interacts with the existing monolayer coverage to produce the compression behavior that limits palette size and determines optimal color layering order.

Surface pressure and monolayer compression: When the first paint drop is placed and spreads via Marangoni flow across the size surface, the ox gall bile salt molecules from the drop populate the liquid-air interface over the entire area that the paint spread to cover. The resulting interface has a lower surface tension than the surrounding bare size surface. Surface pressure is defined as Π = γ₀ − γ, where γ₀ is the surface tension of the bare size surface (~68 mN/m) and γ is the surface tension of the monolayer-covered area (~30 mN/m after the first color), giving Π ≈ 38 mN/m for the first color. As successive colors are added, each new color carries additional ox gall, and the entire size surface gradually becomes more uniformly covered by bile salt monolayer. As the monolayer approaches maximum packing density — the point where each bile salt molecule at the interface occupies its minimum area per molecule (approximately 0.5–1.0 nm² for bile salts) — the surface pressure approaches a plateau, analogous to the collapse pressure of a Langmuir monolayer at the air-water interface in a Langmuir trough experiment. Beyond this surface density, additional bile salt added by new colors cannot reduce surface tension further; instead, molecules that cannot fit in the monolayer aggregate into micelles at the sub-surface, eventually forming a visible white foam at the size surface if too much ox gall is applied — a clear diagnostic of overloading.

Gibbs adsorption isotherm and diminishing returns: The quantitative relationship between bulk surfactant concentration and surface tension reduction is described by the Gibbs adsorption isotherm: dγ/d(ln c) = −RTΓ, where γ is surface tension, c is bulk surfactant concentration, R is the gas constant, T is temperature, and Γ is the surface excess concentration (mol/m², the amount of surfactant per unit area adsorbed at the interface above bulk concentration). At concentrations well below the CMC, increasing bulk concentration c increases surface excess Γ and produces significant Δγ per increment of added surfactant: the surface coverage is sparse and each additional molecule finds available sites at the interface. At concentrations near or above the CMC, the interface is nearly saturated (Γ is at its maximum value Γ_max), and additional surfactant in the bulk does not further decrease γ (the isotherm flattens). This is the molecular basis for the diminishing returns observed in multi-color ox gall scaling: successive colors encounter a surface approaching monolayer saturation, so each additional drop of ox gall in the new color produces less surface tension reduction than the same amount did for the first color. The practical consequence is that later colors spread less per unit of ox gall, and larger amounts of ox gall per unit paint are needed to achieve the same spreading area as the first color.

Optimal color layering order: The physics of multi-color compression dictates the color layering strategy used by experienced marbling artists. The first color laid on the size surface will be compressed inward to the center of each subsequent color’s Marangoni expansion: as each new color spreads outward from its drop point, it pushes the existing paint radially inward. Colors laid first end up as the innermost rings of each circular set; colors laid last end up as the outermost, largest-diameter rings. Within each color ring, the center (highest local ox gall concentration, highest initial spreading force) has the most opaque, concentrated pigment; the outer edge (diluted by Marangoni spreading) is more translucent. This creates the characteristic gradient from opaque center ring to translucent outer edge visible in high-quality marbled paper. Practical consequence: lay lighter, more transparent colors first (they will be compressed into small, opaque center rings, where their transparency is less critical) and darker, more opaque colors last (they will form the largest outermost rings, where their opacity creates the bold border that frames the pattern). The total palette size — typically 5–7 colors in traditional ebru — is directly determined by the Gibbs isotherm saturation of the monolayer: beyond 7 colors, the size surface is uniformly covered by a nearly maximally packed bile salt monolayer, and the last colors either fail to spread (insufficient surface tension gradient) or cause all paint to dive (surface oversaturation). Both failure modes are visible within seconds of dropping the problematic color.

iOS rates and Apple Tax for paper marbling creators

Paper marbling creators build audience across platforms where the visual drama of the floating paint pattern — the Marangoni spreading of each ox-gall drop, the comb drag revealing the chevron pattern, the paper lift revealing the transferred design — performs compellingly in both still photography and short-form video.

Pinterest is the dominant discovery platform for marbled paper aesthetics: endpaper images, stationery designs, bookbinding materials, and decorative paper pattern inspiration drive very high re-pin activity, and marbled paper aesthetics have been among the highest-performing visual categories on Pinterest since the platform’s early years. Pinterest marbling content: 72–82% iOS. Instagram marbling process photography and Reels (overhead time-lapse of Marangoni spreading, the comb pass across the stone pattern, the paper lift, the rinse reveal): 70–78% iOS. TikTok marbling reveals (the gel-git comb pattern transformation filmed in real time, or the paper-lift reveal showing the transferred pattern for the first time): 74–84% iOS. YouTube marbling tutorials (carrageenan size preparation, full session demonstrations showing multiple color drops and comb patterns, troubleshooting videos for sinking paint or transfer failure): 58–68% iOS, somewhat lower as some viewers watch tutorials on desktop while working at a nearby marbling tray.

Beginning November 1, 2026, Apple charges Patreon 30% on every subscription payment processed through the iOS app. In dollar terms: at $200/month with 74% iOS: approximately $44.40/month ($532.80/year) in Apple fees. At $350/month with 78% iOS: approximately $81.90/month ($982.80/year). At $500/month with 80% iOS: approximately $120/month ($1,440/year). Enable Patreon’s web-only billing toggle in Creator Settings before October 31, 2026, and update all social media bio links — Pinterest profile link, Instagram bio, TikTok bio — to the Patreon web URL. Verify the subscription flow works correctly from an iPhone browser (not the Patreon iOS app) before November 1 to confirm subscribers can pay without triggering Apple’s 30% commission.

KeepTier is a self-hosted membership page for creators who want 100% of their tier revenue and zero Apple tax. Plans start at $9/month.