Technique and Materials

George Stubbs did not move through his career with one fixed material formula. He painted in oil on canvas, panel and copper; worked with wax-bearing mixtures; fired enamel on ceramic plaques; etched anatomical plates; and made or supplied designs for prints in several intaglio processes. Those activities belong to the same searching intelligence, but they are not the same technique. The clearest account begins by naming what physically carries an image, what binds its colour, how the image was made and who performed each stage.

That discipline matters because familiar shorthand can conceal the very experiments it is meant to celebrate. “Wax”, “encaustic” and “enamel” are not interchangeable. Nor is a print merely a paper copy of a painting. Chemical analysis, current collection records and inscriptions on printed sheets provide different kinds of information; each can answer some questions while leaving others open. Together they show George Stubbs varying materials in response to scale, permanence, tone and reproduction rather than following a single recipe.

Support, medium and process are different

A support is the object that carries the image: canvas, a wooden panel, a copper sheet or a Wedgwood ceramic plaque. Oil or a wax-bearing mixture may bind pigment spread over that carrier. In enamel, coloured glass-bearing material is fused to a prepared surface through firing. In an intaglio print, ink held below the surface of a metal matrix is transferred under pressure to paper. Paper is then the carrier of an impression; the plate is the matrix that made it. ^{[1] [3] [4]}

Copper demonstrates why the vocabulary cannot be guessed from a material name. Yale's Self-Portrait of about 1759 is oil on copper: the metal is the painted support. An etched copperplate has a different function, because its recessed lines hold ink for printing. A Wedgwood plaque is likewise not the medium in the same sense as oil. It is a ceramic body that survives further firings while enamel colours fuse on its surface. These distinctions also keep the drawings made before or around a print separate from the plate and from every sheet pulled from it.

Building a painting begins before the coloured image

A painted surface can seem continuous even when its construction is not. Whistlejacket is a useful reminder at the most physical level. Judy Egerton's National Gallery catalogue records a vertical seam in the canvas, sixty-four centimetres from the left edge, as well as small splashes of original grey-brown paint across the background. These are observations about how the object is made and survives. The catalogue's condition assessment belongs to its publication in 2000; it is not a newly issued report on the picture. ^[5]

The seam does not mean that the horse itself was assembled from separate painted fragments. It marks the construction of the support. Equally, a relatively plain background does not establish which future additions an artist intended. The carrier, the painted layer and the projected composition are related questions, but evidence for one does not answer all three. Looking for a single ingenious technique can obscure these ordinary yet consequential stages of making a large picture.

There is another kind of construction within the design. David Pullins has examined Hollyhock, in which Stubbs's horse was joined by landscape and figures associated with Claude-Joseph Vernet and François Boucher. He relates the work to the possibility of treating animal and setting as separable components. His account brings workshop practice and divided authorship into the history of the image. ^[6] It remains a scholarly interpretation of particular works, rather than proof that every Stubbs landscape was added by another hand.

This distinction helps make sense of the artist's apparent consistency. A recognisable horse silhouette might recur while the surroundings, support or means of reproduction changed. A composition could be adapted without a new binder being invented, and a new paint mixture could be used without abandoning an established composition. Technique includes both the handling of matter and the organisation of work. Neither can be reconstructed adequately by identifying a motif alone.

It also prevents anatomical knowledge from becoming an explanation for everything. Understanding the horse's body helped Stubbs represent it; it did not determine the width of a canvas, the preparation of a panel, the design of a plaque or the division of tasks among makers. Those decisions placed bodily knowledge into objects with different dimensions, surfaces and uses. Their practical differences remain visible even where the finished images share an exacting attention to animals.

What ten sampled pictures can prove

John Mills and Raymond White's National Gallery investigation considered ten pictures dated 1772–1800. Gas chromatography, supplemented in most cases by mass spectrometry, identified material classes rather than a complete recipe: drying oil, non-drying oil or fat, beeswax and pine resin. Their table groups three early panels without drying oil or a conventional ground; sampled works from the 1783 Reapers onward also contain drying oil. ^[1]

The authors warn that beeswax and other fatty material can be difficult to distinguish in mixtures. They also consider possible contamination from conservation treatment. Early heat fusion is a hypothesis, not a demonstrated procedure for the whole oeuvre. A suggested wax-resin coating on bare wood might instead have come from lost paint. Their discussion of eighteenth-century ‘encaustic’ literature shows competing methods rather than one recovered ancient formula. ^[1]

Those limits are productive. Ten sampled works can establish that a single standard label is insufficient, without supplying a replacement formula for every painting. A result from one passage need not describe every other passage, and a dated sequence of selected examples is not a complete diary of studio practice. What the enquiry makes possible is comparison: material choices can be connected to particular objects and then tested against further evidence.

A chemical component and an operation are also different findings. Detecting wax does not tell us how much heat was applied, at what moment, with which apparatus or for how long. A list of substances cannot recreate the order of the artist's actions on its own. The difference becomes clearer when set beside an inscription on a print: lettering can name publication, while a sample addresses composition. Neither kind of evidence is defective because it cannot answer the other's questions.

‘Encaustic’ therefore remains useful when it identifies a historical proposal or a carefully defined practice. It becomes misleading when applied as an attractive blanket term. The point is not to make Stubbs's work sound less adventurous, but to allow several adventures to remain distinguishable. Oil mixtures, wax-bearing paint and fired enamel asked different things of the artist and left different physical consequences for later examination.

Copper, panels and an object-specific wax mixture

Yale records the small Self-Portrait, B2006.16, as oil on copper, measuring 14 by 10.8 centimetres. Its date of about 1759 is approximate. The museum has cautiously related it to the period when he was teaching himself to etch plates for The Anatomy of the Horse. ^[2] That possible connection is illuminating because the same metal could support a painted likeness and serve, in a different object, as a printing matrix. It does not turn the portrait into a print or an enamel.

The high-resolution Yale image shows an oval bust emerging from a dark brown ground. The pale collar and coat interrupt the surrounding darkness, while a network of fine surface cracks is visible over the face. Such looking can describe the present surface, but it cannot identify a binder. The museum's material record, not the pixels, establishes oil and copper.

The same rule applies to Yale's Brown and White Norfolk or Water Spaniel of 1778. Its catalogue gives “beeswax and oil on panel”, dimensions of 80.6 by 97.2 centimetres and accession B2001.2.10. ^[7] The alternative breed title remains deliberately unresolved. More important here, the medium belongs to that object. It supports a focused case in the history of Stubbs's animal painting and a closer account of the spaniel; it does not authorise “oil and beeswax” as a label for every panel, canvas or copper.

A catalogue label and a laboratory sample answer different questions

The National Gallery now identifies NG3529 as A Gentleman driving a Lady in a Phaeton, dated 1787, with oil paint on a wooden support, probably oak. The occupants remain unidentified. ^[8] Earlier technical publications called the picture Lady and Gentleman in a Carriage. A change in title does not create a second physical object, just as a short current medium label does not necessarily reproduce every constituent found in an older sample.

This matters when comparing records. ‘Oil on wood’ tells a visitor the broad material class. A laboratory investigation can ask whether wax, fat or resin also occurs in a minute portion of the paint. These statements operate at different resolutions. They need not contradict each other merely because one is concise and the other chemically particular. A useful comparison keeps the inventory number attached to both.

Titles can also be deceptively similar across different works. The 1793 Royal Collection canvas called The Phaeton in the National Gallery's analytical table is a separate entry from NG3529. ^[1] Substituting one for the other would transfer not just a carriage subject but a sample result, date and support. Material history requires the same care over versions that a history of named racehorses requires over individual animals.

The practical benefit is a more exact description of experiment. Stubbs did not have to abandon oil whenever he introduced wax, and a wooden carrier did not automatically identify one particular wax-bearing preparation. A label identifies the surviving work; an analysis contributes a more local answer. Neither should be made to promise a complete account of all the artist's materials.

Fired enamel and Wedgwood's ceramic plaques

Enamel changes the operation. Museum process guides describe finely powdered, coloured glass mixed with a temporary oily vehicle, applied to a prepared surface and fused by heat; the vehicle burns away during firing. Multiple colours may require separate firings. ^[3] That general account explains why fired enamel is neither ordinary oil paint nor a wax-resin painting. It does not, by itself, reveal Stubbs's exact colour recipes or firing schedule.

Scale made the ceramic support difficult. The World of Wedgwood dates contact through Thomas Bentley to 1775 and describes years of work on large plaques. Wedgwood's warning that “large tablets are not the work of a day” expresses the manufacturing problem succinctly. ^[9] The image depended on producing a carrier before any enamel colour could be applied.

Three objects prevent “the Stubbs enamel self-portrait” from becoming another false singular. The National Portrait Gallery's NPG 4575 is a 1781 self-image in enamel on an oval Wedgwood plaque, 697 by 531 millimetres. It was exhibited at the Royal Academy in 1782. ^[10] The Lady Lever Art Gallery's LL 3684, Self Portrait on a White Hunter, is a different work: a mounted figure in a landscape, dated 1782 and made in enamel on Wedgwood biscuit earthenware. The NPG's historical catalogue notes that it was formerly called Josiah Wedgwood, but identifies the rider as Stubbs. ^[11] Neither object is Yale's earlier oil-on-copper portrait. Their relations and changing identifications belong in the fuller comparison of Stubbs's self-portraits.

The Yale Center for British Art's Sleeping Leopard, B2001.2.209, makes the contrast still more exact. This signed and dated 1777 object is enamel on Wedgwood biscuit earthenware and measures only 10.8 by 16.8 centimetres. ^[12] In the inspected image, the spotted animal curls tightly inside the oval, its head lowered near its forepaws. The confined ceramic format is a physical fact of this object, not a clue that every later leopard image shared its process.

The same institutional history lists a portrait of Richard Wedgwood, Josiah's father-in-law, among the plaque subjects. ^[9] That specific sitter should remain distinct from both the painter's self-images and the Wedgwood family group.

The manufacture of those supports also involved exchange. The Wedgwood account describes a family portrait and two modelled relief subjects as part of Stubbs's contribution in return for the plaques. One relief, The Frightened Horse, developed from his own lion-and-horse engraving; its companion was The Fall of Phaeton. The account traces the horse model through work at Etruria in August 1780. ^[9] Painting, print design and modelling were connected through the artist's activity, while remaining distinct operations.

A relief is especially revealing because it translates a drawn or printed contour into actual raised form. Its shadows belong partly to the physical object and the light falling on it. In a painted animal, shadows are represented by colour; in an intaglio impression, they arise from ink transferred to paper. The shared horse subject is therefore a starting point for a comparison of processes, not evidence that the finished objects were materially equivalent.

The relation between artist and manufacturer was likewise more reciprocal than a simple purchase of ready-made supplies. The painter asked for a difficult surface; the factory's experiments made that surface possible; the painter supplied images and models in return. The finished enamel can consequently be considered both as an authored picture and as the result of manufacture. Calling it merely an unusual painting support leaves much of that cooperation out of view.

Exhibition exposed the results to another kind of test. The NPG enamel self-portrait attracted a newspaper comment that its white flesh appeared ‘cadaverous’. The museum preserves that criticism alongside its record of the 1782 Royal Academy exhibition. ^[10] A contemporary response to colour is valuable, but it is not an analysis of pigments or proof that a firing failed. It shows that a technically ambitious object could still be judged as a portrait, through the appearance of the represented face.

Framing helped present the experiment as a substantial picture. The NPG's catalogue associates its neoclassical frame type with Wedgwood, Bentley and the frame-maker Thomas Vials, while explicitly noting that Stubbs's actual employment of Vials is undocumented. ^[10] That distinction preserves a useful connection without turning an inferred design relationship into a paid workshop transaction. The frame joined the ceramic image to familiar expectations of display, even though the object it surrounded had been made by an unusual process.

The Metropolitan Museum of Art's A Sleeping Leopard, 49.49.5, is a separate work published on 1 January 1791. It is a soft-ground etching with roulette work, the second of two states, measuring 18.1 by 23.7 centimetres. ^[13] Its rectangular paper impression uses dense granular and linear marks to set the curled body against a trunk and surrounding darkness. Shared subject and pose do not erase fourteen years, different dimensions, another holder and a transfer from plate to paper. The dedicated account of Sleeping Leopard can follow that design family while retaining two physical identities.

From a drawing to a plate and an impression

Traditional intaglio printing begins below the surface of a plate. In engraving, a burin cuts the metal directly. In etching, the printmaker draws through an acid-resistant ground and acid bites the exposed lines. These are ways of preparing an ink-bearing matrix. ^{[4] [14]} An original drawing may guide this work, but it is not the copperplate. The plate is not an impression, and an impression is not every later state taken from the same matrix.

The British Museum's copy of The Anatomy of the Horse records forty-seven pages of letterpress and twenty-four etched plates, published in London in 1766 by J. Purser for the author. It assigns the printmaking to Stubbs. Finished views of dissections are paired with numbered outline diagrams that act as keys, sometimes within one plate and sometimes on facing plates. ^[15] That compact structure is enough to establish that Stubbs did more than provide compositions; the full book, dissection method and sequence belong to the account of The Anatomy of the Horse.

Current catalogues can describe technique at different levels. The British Museum calls the book's plates etchings. Wellcome catalogues an impression of Table X, reference 570623i, as “engraving, with etching” and gives a 37.7 by 47.2 centimetre platemark. ^[16] The compound label should not be silently simplified. One institution is describing the bound publication broadly; another is identifying work visible in a particular impression. Exact terminology belongs to the exact object being discussed.

Soft ground, roulette, stipple and state

Soft-ground etching uses a yielding ground that can receive pressure through paper, preserving qualities closer to pencil or crayon before the exposed metal is bitten by acid. ^[14] Roulette introduces repeated texture with a toothed or patterned tool. Stipple builds tone from dot-like marks. Mezzotint begins with a systematically roughened plate capable of printing dark, then uses scraping and burnishing to recover lighter tones. ^[4] These definitions clarify collection labels; they do not prove which visible mark came from which tool unless close technical study does so.

The Met leopard securely combines soft-ground etching and roulette. The National Gallery of Art's Reapers, accession 2021.102.1, instead combines stipple and roulette on laid paper. This 1791 impression measures 48.26 by 68.58 centimetres, is trimmed to the platemark and is state i/iii. Its lower inscription assigns painting, engraving and publication to Stubbs himself. ^[17] It is the printed Reapers, not the separate painting. The object history of Reapers can compare versions; technique first requires that the NGA sheet remain an impression.

A state records a changed condition of a plate. Added lettering, reworked tone or another alteration can distinguish one state from the next. “Second state of two” does not mean that only two sheets were printed, just as state i/iii does not turn the first condition into a preparatory drawing. The surviving paper, its trimming, inscription and place in a sequence all help identify the particular impression.

Printing involved decisions after the plate was made

The etched lines were only part of a print's preparation. The Met's process account describes ink being spread across a plate and worked into its recesses, followed by wiping of the surface. Damp paper and the pressure of a rolling press bring the ink out onto the sheet. The design appears reversed, and the edge of the plate can leave a physical mark in the paper. ^[14] These operations distinguish a metal image capable of printing from an actual impression.

Their sequence explains why making a plate did not finish every visual decision. A plate supplied a durable arrangement of recesses; printing brought that arrangement into contact with ink and paper. An impression can be examined as the result of that encounter without assuming that the surviving plate, if it exists, would disclose everything about the sheet. The process joins a reusable tool to a particular material event.

Reversal is especially useful when comparing a painting with a print after it. It is not by itself proof that the printmaker misunderstood the composition or that a reproduction has been accidentally flipped. In intaglio, opposite orientation is an ordinary consequence of transfer. Establishing how the design was prepared requires more than noticing which way a horse faces. The exact relationship must be checked against the catalogued painting, plate or impression, rather than corrected to satisfy an expectation that all versions should look alike.

What survives on the sheet

The dimensions of a print need to be read as carefully as its technique. A plate measurement, the printed image and the sheet of paper are not necessarily the same size. A margin can contain lettering or simply preserve paper beyond the plate's edge. Trimming can remove that material while leaving the central composition apparently complete. The NGA's description of Reapers as trimmed to the platemark and the Met's separate leopard dimensions therefore describe physical surviving objects, not interchangeable sizes for the design. ^{[13] [17]}

The inscription on Reapers is particularly helpful. It names Stubbs as painter, engraver and publisher and gives 1 January 1791 with his Somerset Street address. ^[17] These words identify credited roles and publication circumstances. They should not be stretched to prove that no assistant handled ink, paper or the press. A printed statement of authorship is evidence; it is not a complete attendance list for a workshop.

In the inspected impression, the pale sky, darker trees, figures and sheaves form a broad range of tones, although the sheet does not reproduce the colours of an oil painting. The tiny marks are organised into bodies, foliage and distance. This makes a process such as stipple more than a technical name attached to a catalogue: its accumulation of marks helps create the image's continuous effects. The appearance can be discussed without assigning each dot to a particular tool stroke.

The leopard print makes a different use of tonal compression. Much of the rectangular field is dark; the spotted body remains legible through contrasts within that darkness. It is not simply the oval enamel with its colours removed. The surrounding trunk and ground, the distribution of the black ink and the proportions of the sheet make a fresh visual arrangement. Comparison across media is most useful when it asks what was changed to make the image work again.

Such differences explain why an impression is an artwork in its own right. It records decisions about tone, paper, scale and lettering as well as the pre-existing subject. The V&A's account of printmaking distinguishes the repeatable process from the particular print that results. ^[4] Multiplicity does not erase individuality: one surviving sheet can retain evidence of a specific state or trimming that another impression does not possess.

When another printmaker worked after Stubbs

Stubbs sometimes etched, engraved or published his own images, but many printed interpretations separate invention from execution. The phrase “after George Stubbs” credits the source composition; it does not say that Stubbs cut, rocked, bit, inked or printed the plate. The broader guide to his prints maps that production. Four examples show why the names beside an impression matter.

Benjamin Green's Brood Mares is a mezzotint after a Stubbs painting and reverses the painted design. The Government Art Collection documents a later issue published by Sayer and Bennett on 14 November 1776 after Green had sold the plate; the British Museum preserves a 1768 impression and identifies Green as printmaker. ^[18] Maker, plate owner, publisher and date can therefore change across the life of one matrix.

The National Gallery of Art's A Tigress, 2001.96.2, is a 1772 mezzotint on laid paper by John Dixon after Stubbs, state i/iv. ^[19] The British Museum's The Horse and Lion of 1770 gives another division: George Townley Stubbs made the mezzotint and etching after George Stubbs, while John Wesson published the scratched-letter state before the title; a later lettered state is also recorded. ^[20] It belongs to the reproductive history surrounding the lion-and-horse works, not to a claim that father and son were one printmaker.

Thomas Burke must remain separate too. The Albertina lists an Eclipse made by Burke after Stubbs with a date of 1772; a checked impression published by Robert Sayer in 1773 is a mezzotint on laid paper, third state of four. ^[21] The slight date difference describes catalogue and impression evidence that should stay visible. It is better to preserve that bounded variance than to manufacture a single effortless chronology.

Experiment and the later life of paint

Materials chosen for pictorial effects also affect what later custodians can safely do. A modern treatment account from Simon Gillespie Studio describes the difficulty of cleaning Two Bay Hunters Communing On The Bank Of A Lake. The conservator, Majo Prieto Pedregal, found the original paint unusually sensitive and tested different approaches before reducing discoloured varnish and overpaint. Some hardened earlier restoration could not safely be removed and was retained and retouched. ^[22]

This is testimony about one treatment, not a general cleaning method for all Stubbs pictures. Its importance for a reader lies in the decisions it makes visible. Removing something later does not automatically restore an untouched original: the added material may be entangled with vulnerable paint, and intervention may have to stop short of complete removal. The apparent unity of a clean surface can depend on carefully differentiated work across the picture.

It also complicates a simple celebration of permanence. An artist can seek durable colour while producing a surface that later proves difficult to treat. Survival, stability, ease of cleaning and an unchanged appearance are different achievements. A painting may remain structurally intact while its varnish discolours or some pigments alter. The word ‘permanent’ is most informative when attached to a particular material problem rather than offered as a final verdict on an experiment.

Digital reproductions introduce a further limit. A photograph can show a crack, but it cannot establish whether a colour difference comes from original pigment, old varnish, retouching, illumination or the capture itself. Nor can a brighter digital image demonstrate that a treatment was successful. Conservation records and direct examination answer questions that reproductions cannot. Close looking remains essential, but its strongest contribution is often to identify where a further question should be asked.

Hambletonian: technique meets the conditions of display

A study of Hambletonian, Rubbing Down at Mount Stewart connects those material questions with the room in which a painting is seen. Nigel Blades, Katy Lithgow, Stephen Cannon-Brookes and John Mardaljevic describe beeswax and pine resin with drying oil especially in its landscape, while the sky was painted mostly in oil. Their account also identifies light-sensitive colour, including possible red-lake tile outlines visible under ultraviolet illumination and vulnerable blue and yellow components. ^[23] One painting could therefore contain passages with different material histories and vulnerabilities.

The study followed earlier conservation by the Hamilton Kerr Institute in 1981–83. Its later concern was excessive daylight from the glazed dome above the picture. Darker surrounding paint and an external mesh screen were tested as ways of reducing exposure while retaining useful viewing conditions. ^[23] These are reported interventions from a particular research project, not a claim about the room's exact arrangements today.

The distinction between making and showing a picture becomes unusually concrete here. Stubbs controlled his painted surfaces; later custodians worked within a much larger architectural setting whose light could change throughout the day and year. Conserving the image consequently required attention beyond its frame. The surrounding room affected the amount of light reaching it, even though the room was not part of the painted object.

A present appearance also contains an uncertain span of earlier change. The researchers could not reconstruct all the exposure received in previous locations. ^[23] It would be misleading to assign every colour alteration to one move or one decade. A technical account can identify vulnerability and measure a current problem while leaving portions of the past unresolved. Acknowledging that limit gives the intervention a clearer purpose: to reduce further risk, rather than promise recovery of an unrecorded original appearance.

For anyone studying Stubbs, this changes what ‘technique’ includes. The subject is not exhausted by the moment the paint dried or the enamel cooled. The physical choices continued to matter when the work was varnished, moved, illuminated and treated. An account of materials connects the artist's decisions to the subsequent care that allows the picture to remain available for looking.

How to read Stubbs's materials

Stubbs's experiments become clearer when every object is asked the same questions in order. What carries the image? What binds colour, or what process fixes it? Was there a matrix, and which person worked it? Which impression or state survives? Finally, what kind of evidence supports each answer: holder metadata, an inscription, a technical sample, a period document or direct looking?

That sequence does not diminish experiment; it reveals its range. Oil on copper, beeswax and oil on one named panel, changing sampled mixtures, fired enamel on Wedgwood and varied intaglio processes can coexist without being collapsed into “the Stubbs technique”. The remaining uncertainties—mixture identification, heat, recipes, terminology, dates and divided labour—are part of the material history rather than gaps to be filled with certainty.