In a letter to a young friend, the artist Burne Jones once drew a cartoon of himself “attempting to join the world of art” by climbing into a picture – the effort ends up disastrously, of course, when he falls right through the canvas! Spending much of my time struggling to get ‘inside’ of paintings I have a lot of sympathy with his feelings – when I analyse (or a restorer restores for that matter) I am probably as close to the physical creation of the picture as anyone since the artist himself, so I know that there can be an enormous gulf between knowledge of the object as an object and an understanding of it.

So, if it can be so difficult to move from a knowledge of materials and techniques to an understanding of art, why analyse paintings? What can we hope to gain by doing so? In the following article I want to explore this area, looking at what we can and cannot do, point out possibilities and limitations and introduce some of the basic methods used for the scientific analysis of paintings.

We can in practice approach the subject from two (equally valid) directions: the issues that can be resolved using scientific techniques and the analytical methods actually used. These views are interrelated of course – there is little point in studying trivial issues with the most sophisticated instruments just as it is ridiculous to attempt problems for which there are no practical solutions. So, from the aspect of what kinds of question we tackle with scientific tools, there are currently certain specific areas where these techniques are regularly applied; some broad examples might be:

• To assist in revealing information ‘buried’ in the painting that is relevant to making proper art historical or connoisseurship judgements, such as in identifying artists alterations (pentimenti) from X-radiographs or underdrawing from infrared images;
• Addressing issues of scholarship such as those concerning the economic and social structures of art production through a detailed knowledge of the materials and techniques and how and when these were applied chronologically and geographically;
• Identifying the likely origins of a painting from an examination of the materials and techniques;
• To help in assessing the condition of a painting and then with aiding the conservator/restorer to make judgements about (for example) the presence of later overpaint and the most appropriate – safe! – treatment techniques for removing it.

Likewise, a wide range of modern analytical instruments are applied in resolving these kinds of question though some are basic and recurrent – such as X-radiography and the use of various microscopic techniques.

Hidden meaning

Major advancement has taken place in areas of art history over the last few years specifically because of the new information available from scientific studies of paintings. It is becoming increasingly common to find radiographs and/or infrared images accompanying papers discussing the attribution of paintings, particularly where there are a number of versions, say, or if the extent of workshop involvement is at question. It is key with this type of information that the methods used are understood and their results properly interpreted, otherwise there are many pitfalls for the unwary.

The most familiar of the investigatory techniques are X-radiography (what I will call ‘conventional’ X-radiography) and infrared imaging; there are others – to be discussed later – but these two methods are those most frequently encountered. Both are capable of penetrating through a painting to reveal what is beneath, but are sufficiently different in their sensitivities to be complementary rather than equivalent.

An X-ray of a painting is not dissimilar in principle to a medical X-ray such as would be taken if you broke a bone. The image is formed by passing X-rays through the object of interest (arm, leg, painting) and exposing a photographic film placed on the other side. The image formed then depends on factors such as the thickness and X-ray density of the object in the middle and localised variations within it. Broken bones have gaps and let more X-rays through at these points to expose the film; therefore the breaks appear darker than the surrounding bone, which blocks the X-rays more. Likewise, if a painting contains materials which reduce the passage of the X-rays, then they will appear lighter than surrounding material – paint – which does not. Consequently we can see that, depending how the artist has arranged his materials, parts painted in X-ray dense compounds (such as lead based pigments), and/or more thickly, will show up lighter in contrast to those parts painted more thinly in relatively X-ray transparent materials. If we look at an X-radiograph of a portrait for example, where higher levels of lead white tend to be used in flesh tones – especially highlights – then the whiter, more X-ray dense highlights appear lighter than the shadows giving an image looking not unlike a black and white photograph.

It is worth giving a couple of illustrative examples showing the kinds of information we can obtain. The first of these is a Portrait of Sir Richard Willis by William Dobson (Newark County Council, in part to show the appearance of a radiograph of a portrait, but also as a good example of the kind of alteration that can be detected. Comparing the surface image and the radiograph immediately shows a number of differences, the main one perhaps being the appearance of a second – profile – head; the nose is visible projecting from the right ear and a wide triangular collar at the neck line. This in fact represents Dobson’s first essay at the portrait – as far as we can tell it is the same sitter and must therefore be a change of mind rather than reuse of an old canvas. We also know from X-radiographs of other paintings by Dobson that he was given to such major changes of composition; in a double portrait of two unknown men (now in the Courtauld Institute of Art) he both enlarged the canvas and moved the right-hand sitter substantially across. It has also been suggested however that at the time Dobson painted this portrait of Sir Richard Willis, during the Civil War, painting materials were very limited and he might have had to reuse his canvases.

Some other changes were found from the x-radiography. For a start, comparing the red sash currently visible in the picture and that in the radiograph shows that it was originally painted more vividly; the white highlights to be seen in the radiograph are crisper and more jagged in appearance. In fact the sash has been substantially repainted – exploratory cleaning has shown that before the current dull red colour there was a rich crimson sash. It is also interesting that the upper version of the sash does not lie under the baton the sitter is holding, whereas the earlier one does. This is further evidence that the upper sash is not original and almost certainly not by Dobson. There have been a number of changes to the positions of the hands too, this though also relating to a helmet originally planned as part of the composition. The helmet must have formed an important part of the composition, with both the sitter’s hands placed on it; a three-quarter length portrait of Charles II when Prince of Wales by Dobson (in the Scottish National Portrait Gallery) has a similar arrangement. Afterwards however the helmet was painted out, the right hand holding the baton added on top and the angle of the left forearm shifted slightly. Whether the change to the baton was a part of Dobson’s conception is still slightly uncertain; while it might have reflected Sir Richard Willis’s sudden elevation in rank, there is at the moment no clear technical evidence as to whether it was a late alteration by Dobson or an early modification by another hand.

Identification of pentimenti is not the only application of X-radiography of course, and I use it (among other things) for making estimates of damage, measuring canvas weaves, looking at how a canvas or panel was prepared for painting or if the picture has been cut down. This last question is resolved by studying the edges in canvas painting – during preparation the canvas will be stretched and a ground (the preparatory layer used prior to painting) applied; at the points of attachment of the canvas to the restraining stretcher, the fabric weave will become distorted, a feature which becomes locked in place when the ground dries. If these distortions (‘cusping’) are still fully present, then the painting has not been cut down.

It is also worth emphasising at this point that because such a X-radiograph is capable of showing details from throughout the structure from support to surface, interpretation can thereby be made difficult. This arises for several reasons. Firstly, there is no specific ‘depth’ information in a conventional X-radiograph – it does not, of itself, tell you where between the support and the surface the feature of interest you can see actually lies: it might be on the surface or the support. A classic example would be something actually on the back of a canvas or panel! Secondly, parts of the structure may physically obscure details of interest; a not uncommon problem is the influence of canvas weave where the regular mesh of threads (actually, where the ground applied to it fills the weave and is itself the X-ray dense component, textiles being relatively non-dense) interferes with the image of the paint layers. There are solutions to these problems though, which I shall discuss later.

There is another X-ray technique which I commonly use but which is not otherwise widely available. Known as ‘electron emission radiography’ (‘ee-radiography’; it also goes under several other titles such as beta or beta backscatter radiography) this method was originally conceived as a way of producing radiographs of paintings executed on metal plates such as copper sheet. If you take a conventional X-radiograph of a painting on copper, all you will probably see is an image of the plate itself. This has its uses – it is possible to study the manufacturing processes such as whether the sheet metal was hammered or rolled – but the support dominates and prevents us seeing details of the paint layers. ee-radiography functions in an apparently peculiar way in that the photographic film comes before the painting – X-rays pass through the film, strike the painting, knocking out electrons from the paint – but significantly, these electrons cannot travel far before being reabsorbed so that if they leave the paint and interact with the photographic film then they contribute to forming the image. Moreover, this means that only the upper layers of the painting give rise to the image, those formed deeper in the structure being absorbed before they can reach the surface, hence giving a radiograph of the upper strata of paint rather than ground and support layers. One example will show this: a painting by Rubens depicting Mercury; I have include three images here – a colour photograph and conventional and ee-radiographs. The conventional radiograph is dominated by the image of the support, notably two lighter squares due to an X-ray dense material applied to the reverse. However, the ee-radiograph is clear, showing the alternate first draft (Mercury in 3/4 view), as well as part of another (a broad sweep of drapery apparently unrelated to either).

The application of ee-radiography run much wider than paintings on copper though. For a start, there are times when panel and canvas supports interfere too much with the image of the paint layers – perhaps there is extensive woodworm or a ‘cradle’ applied to the reverse of a panel or the weave of the canvas is too dominant; in each of these cases ee-radiography can permit us to see details of the artist’s painting technique. However, I also use ee-radiography to help determine where within the depth of the structure a feature might lie; at the simplest level we can see that by comparing the ee-radiograph, which shows near-surface features, with a conventional radiograph, which shows features from throughout the structure, we can make deductions based on which radiograph shows the item of interest and how strongly. For example, I have used this approach to work out the sequence in which an artist developed a composition; the Spanish still-life artist Luis Melendez (1716 to 1780) seems to have worked on compositions quite slowly and was prepared to make radical changes; using ee- and conventional radiography allowed me to determine the order.

Finally, because of ee-radiography’s sensitivity to surface composition, it is ideally suited to resolving questions of condition and the presence of overpaint. However, I will leave this until a later discussion of how we assess condition generally.

Infrared (IR) imaging relies on the fact that the opacity of pigments varies according to the wavelength of light; notably, beyond the red end of the spectrum of visible light, into the ‘infrared’, various pigments are less opaque and so what is a covering paint film to our eye can, with suitable technology, be revealed as more transparent. Earliest examples of this used infrared sensitive photographic film to look at paintings in the infrared. From this it was shown that, under certain circumstances, features such as an artist’s preliminary drawing could be visualised. The ideal conditions are for a black, infrared absorbing material contrasting with a white, infrared reflective material, both overlain by the paint layer(s); a close approximation to this occurs in charcoal or similar drawings on white grounds such as those of chalk in early Netherlandish painting or on gesso grounds of early Italian panel painting technique and these are the fields in which the technique has been most widely applied.

An example will show the kind of information that we can get from IR imaging – this is from a picture in the Museum of London, the Panorama of London from Southwark. Probably painted in the earlier 1630s, this picture has extensive underdrawing and the example I have included here shows a detail from the southern end of London Bridge, an alteration revealing not only how the artist ultimately moved the bridge (!), but also something of the quality of the underdrawn line. This latter point can be of particular interest in that it is also possible to differentiate media – pen and ink from brush from charcoal for example.

However, this does not mean that in other times and places with different grounds and sketching techniques no useful information is to be gained. I have had excellent results with, say, Victorian paintings where there is (for example) underdrawing on a lead white/oil ground. Equally, less than perfect conditions do not necessarily preclude revealing hidden details, and I have found it possible to detect both preliminary sketches and also alterations (pentimenti) in paintings with dark grounds such as one might find in works from the Italian Baroque. The adjacent example is from a painting by Valentin (a contemporary of Caravaggio) where a figure who was originally depicted with an elaborate ‘slashed’ sleeve had it changed by the artist to a plainer cloak: the wrist is at the top, the elbow at the bottom and the later cloak is visible as sweeping lines across the very clear slashing in the sleeve.

In some circumstances I also use what we might (awkwardly) call ‘transmitography’ – rather than illuminating a painting from the front and looking at the reflected light (‘reflectography’), the illumination is placed behind and we image the transmitted light. This can be very useful with situations such as where paint overlying some feature of interest is particularly thick – lead white usefully increases in transparency in the IR – or in the case of inscriptions on the backs of canvases that have been subsequently relined where IR imaging invariably seems to be more useful than X-radiography.

IR photography is still done to some extent, but it has been largely superseded by what is known as IR reflectography. Instead of a photographic film, reflectography uses an IR sensitive video camera with the image being viewed on a television-like screen. The technical superiority lies in the fact that such camera systems are sensitive further into the IR than film, reaching a wavelength considered optimum for the technique, maximising the transparency of pigments. Therefore, much IR imaging is now done by this method and the examples accompanying this article were all produced using reflectography. At the same time, there is one major disadvantage to reflectography – the resolution (the ability to discriminate fine detail) is much worse than for photographic film; you only have to think of television pictures as compared to a good photograph to appreciate the difference in quality. To circumvent this researchers began to photograph the video screen and then assemble large arrays of detail images into huge composite mosaics which, while an improvement, are very time consuming and difficult to do (and the results not always justified…). Now, with computer technology, preparation of these mosaic images is much simplified, allowing us to produce high resolution IR images that correct for problems such as uneven illumination of the painting and so forth. They are still somewhat complicated to set up and produce, but for situations where there is detailed underdrawing can be invaluable. In practice I tend to use a combination of individual IR reflectograph images and mosaics where a fine and delicate underdrawing (for example) demands it. The Dutch artist Saenredam, famous for his church interiors, would extensively underdraw the architectural detail; IR reflectography mosaics are a useful method of revealing this.

A further example may help illuminate the issues and applications. When we look at how Raphael and his workshop executed a particular commission there are various types of evidence on which we can base our study of working practice. There are for example both preparatory drawings (sketches and so forth) and cartoons (drawings prepared for transfer of the design) as well as the underdrawing present on the final painting itself. Consequently it is useful to identify characteristic features of the preparatory methods applied. Evidence that we should look at is, for example, the range of technique used (form and number and materials of sketches extant, evidence of cartoon transfer methodology, type of underdrawing and so forth).

A number of paintings by Raphael also show lines incised into the gesso; examples are the Garvagh Madonna and the Saint Catherine of Alexandria (both National Gallery, London), the Small Cowper Madonna in the National Gallery, Washington and the Holy Family of Francis I (Louvre, Paris). The two London pictures in particular show that the artist used both horizontal and vertical construction lines – it was also possible to find inscribed arcs from the use of compasses, which were used to fix the perpendicular. It has also been remarked that there are “many slight but significant refinements to the outlines of forms characteristic of Raphael’s paintings”.

Another significant point relates to Raphael’s use of cartoons and drawings and the methods he employed to transfer these designs onto panel. A number of his paintings show pounced dots (spolveri) deriving from a pricked cartoon, while in other cases it has been convincingly argued that tracing was employed. At the same time we should keep in mind that spolveri are not necessarily readily identified – the Vision of a Knight in the National Gallery, London, for example was only found to have such dots after very careful study following restoration. Again, in some instances Raphael may simply have developed the basic design as a drawing and then carefully copied this over while there are some instances of him developing the design in detail on the gesso and we should also note the example of the Holy Family of Francis I (Louvre, Paris), where there are construction lines drawn in red chalk. Hence, while some have pointed out that, say, during the artist’s pre-Roman period Raphael normally used cartoons to prepare his panels and that “the typical underdrawing for this time consists of fairly direct and unelaborated contours” we should note Shearman’s remark that with Raphael “it is a mistake to assume a consistency in preparatory technique”.

As this example clearly shows, the ability to study underdrawing opens whole new areas to study.

The production of art – what materials and techniques can reveal

Scientific examination of paintings provides some of its most fascinating insights when we look at questions of economic and social history. Where did the materials a fourteenth century Florentine artist used come from? Why do we find a wholesale changeover from panel to canvas painting in northern Europe in the seventeenth century? How did commercial manufacture and supply of paint change in the nineteenth century and what has this got to do with Impressionism ?

The materials an artist chooses are as intimately bound up in issues such as cost and availability as they are in Êsthetics – Leonardo could not use chrome yellow or cobalt blue because they had not been discovered; Monet did because they had. If you have to grind your own paints (or take on an apprentice) because they have to be made fresh then you are less likely to venture into the countryside to paint en plein air. To develop these arguments a little, let us take a slightly more extended look at the kind of information we can learn from what pictures are painted on, concentrating on what we can learn from tree-ring dating (dendrochronology) of oak panels, support sizes and how they were prepared for painting in the sixteenth and seventeenth centuries in northern Europe.

An artist in the seventeenth century had a number of options for a painting support, but the main choices were of course canvas and panel. In fact the selection of panel for a painting in the seventeenth century necessarily raises a range of important issues for us since such choice must reflect not only a range of Êsthetic questions, but also practical and economic issues that the artist faced. We shall touch upon a number of these: why for example should all the paintings in this study be on oak rather than some other wood? Also, why was the oak apparently imported from remote sources rather than locally grown, and what is the significance of the apparent change in source for the later paintings in the group? Then, was the choice of panel significant when canvas was both widely available and frequently used at this time? There is no satisfactory published review of the structure of trade in panels for painting covering the sixteenth and seventeenth centuries and if and how supply changes affected artists’ choice of panel and the painting techniques employed, but we may briefly summarise some of the various issues as follows.

To begin with, where does our information come from? Obviously the starting point is the observation and identification of wood (and particular species of wood) used along with the manner, time and place of use. In this instance however much additional data can be derived from the analytical technique known as dendrochronology. Dendrochronology is the dating of wood through a study of the patterns of growth rings; in essence, by looking at sequences of growth ring widths – which relate to how much the tree grew each year and thereby factors such as the weather – and comparing these to known sequences, a specific year can be put to each ring up to the last present in the wood you have. The growth ring widths can in practice generally be measured, sequences determined and matched to give date ranges – earliest and latest rings present. Making allowances for factors such as seasoning and storage we can then relate this in turn to dates at which a painting was likely to have been executed. However, the possibilities are greater. Firstly, the ‘chronologies’ – dendrochronology parlance for the standard reference sequences – vary geographically. With oak, there are several covering the Baltic, central Europe and Germany, England and Ireland. As now, weather varied across continental Europe and this, among other things, affected the growth patterns; therefore we can obtain not only a date range for a particular panel but also an idea of where it probably came from.

It has been known for a long time that much Netherlandish, Flemish and Dutch painting up until around the middle of the seventeenth century used oak as a support. Consequently, perhaps the first question must be why oak? In part the answer has to be one of technical appropriateness – the qualities we would expect an artist to be looking for in a panel are the obvious ones of suitable sizes, strong yet relatively light in weight, minimal warping, and so on which are well satisfied by good oak boards. Additionally however there is the economic component of availability; thus we find that there was a tendency to exploit locally available supplies of wood rather than import ‘exotic’ types. This would broadly explain why there was such a divide between the use of (for example) oak in the north of Europe and poplar in Italy. However, we do know that high quality oak boards were exported by the Hansa from Baltic ports throughout Europe. Clearly this changed during the seventeenth century so that boards were sourced from different forests; almost certainly this was as a result of the decline of the Hanseatic League and the collapse of trade which took place probably due to the Thirty Years War of 1618-48. Whatever, the actual pattern of change will not become known until incidences of panel source and use has been mapped in detail through the sixteenth and seventeenth centuries.

We have therefore high quality oak being transported large distances across Europe to the markets. Was this timber exported as boards or was it made up into panels in situ? The evidence would seem to suggest that the trade was in boards, although again this has not apparently been studied in detail. We may however briefly present three arguments. The first is the variation in boards assembled into individual panels; if panels were manufactured at source then we would expect relatively frequent use of boards from the same tree to occur (because as the length of the supply chain increases so boards would become separated). Secondly, we know that there were well established and powerful joiners’ and cabinet-makers’ guilds in the Netherlands who protected their monopolies by restrictive charter and whose products we may identify from their makers marks (for example panel paintings manufactured in Antwerp would carry a quality mark from two guilds – one for the panel itself and another from the St. Luke’s Guild for the painting). Thirdly, there are examples in contemporary treatises on painting practice referring to aspects of panel production.

The overall pattern is thus of timber sawn into boards at source and exported to the European markets in this form. At those various centres the boards would be manufactured into panels by specialists for sale to the end-users – for furniture, panelling in houses, painters and so forth. Some finishing for particular purposes might be required prior to, for example, the artist (or a specialist imprimeur – see below) applying preparatory layers for painting. Additionally however we should note that the artist would select his panel with the painting in mind. Size was of concern because of issues such as weight and strength, so the availability of canvas, suited to larger works, becomes an issue here and we shall return to the subject later.

Shape was also important – that is the consideration of whether the panel might be destined for a full-length or shorter portrait or a landscape. There have been several tentative studies of this for paintings from the seventeenth century, with the suggestion that there were standard sizes from about 15cm x 12cm to 120cm x 90cm; it would seem that these were often denoted by cost (‘guilder-size’) and may also reflect local measurement units. However in other cases it has only been possible to discern a standard format; with Frans Hals’ paintings this was 1.25:1, and it has been suggested that Esaias van den Velde (1587-1630) specifically chose a 1:2 height to width format for his landscape paintings. In a recent study I made of Antwerp paintings on copper plate I again found a consistent format, but not in fact a range of fixed sizes. What does this mean ?

While this question of standard sizes and formats may appear academic, there are in fact significant points underlying the answers. The importance for us here is two-fold: firstly this links into the issues of trade and supply of ‘finished’ products – if it can be shown that there were standard sizes it would imply the extent of the market in ready-prepared panels; secondly, if panels were being prepared to particular sizes then there are implications for the dendrochronology in that some growth rings might have been removed to achieve such standardisation (i.e., a ‘standardisation allowance’ might be needed like the sap-wood allowance). Moreover, that standardisation might have existed in this period should lead us to take care in recording whether there is any evidence of original size or alteration to the support such as the presence or absence of bevelling around the back edges of panels so that future research can be conducted with this knowledge available.

Larger panels as a matter of course required the joining of several boards. There were several ways of joining boards into panels, generally by planing the edges true and butt-gluing them, sometimes (though not invariably) with the aid of dowels for strength. (Some artists also apparently had their panels modified according to inspiration – the bizarre assembly of certain of Rubens’ panels is legendary for example.) The largest size that could be reasonably produced would have depended on certain engineering factors – essentially how thickness has to increase in proportion to size, and the weight to size ratio (i.e., the density of the wood) compared to the actual strength of the wood – as well as economic ones. Again, a full discussion of this is outside the scope of this report but we may comment that there would be a tendency to move to canvas as size increased, with contemporary authors pointing to weight and transportability rather than cost as the issues which decide whether to use canvas or panel; there is apparently a distinct size cut-off point between using panel and using canvas in the works of van Goyen and van Ruysdael – for van Goyen this was about 0.75m x 1.00m, while with van Ruysdael there is an overlap in the middle size range where he used both panel and canvas.

The issue of joining boards to form a panel also leads to another significant point – an artist would not in general have wanted joins running under important parts of the painting; the clearest example of this exercise of choice is of portraiture where any slight unevenness in the surface where the face was to be placed would spoil the result. Interestingly therefore we often find in paintings that the main join lies to the left of the face and substantially off-centre. While it can be argued that the 1/3 plus 2/3 width arrangement of boards within a panel results from split radial-cut boards, such an arrangement of boards has been observed in many other portrait panels thus suggesting a degree of active choice by the artist.

A final comment must be made here regarding the transition of panel to canvas. Studies have shown that from the sixteenth century Netherlands prior and subsequent to 1575, and for Dutch seventeenth century painting, an enormous shift took place in the latter part of the sixteenth century from panel to canvas, but that none-the-less panel was an important support (at virtually half of the total) in the seventeenth century. If we look in more detail however we find that there are certain subtle shifts in relative use of panel and canvas within the general trend towards fabric supports. Rembrandt for example used a greater proportion of panels after he moved to Amsterdam where the supply was presumably more plentiful and cheaper; therefore we may expect a major port like London to correspond to a higher use of panel. Frans Hals, working in one place, shows a different aspect; there is a gradual decline in the proportion of panels he used from the outset of his career (1611-19) when they represented about 50% of his output, to the 1640’s and 1650’s when panels made up only about 14%. Interestingly however in the last years of his life (1660-64) this proportion rose to 27%.

Once the panel itself had been selected and any defects treated, a preparatory layer or ground would be applied prior to any preliminary drawing and the painting itself. The basic changes that have taken place in the form and composition of grounds for painting are reasonably well known and we can state that the northern European approach to grounds in the seventeenth century was firmly rooted in Netherlandish techniques of the immediately preceding era. Traditionally several layers of a chalk and glue mixture would be applied with time allowed after each coat for drying and smoothing of the surface prior to the next; the overall effect of this was to achieve a final result which was very flat and uniform. However, there were a number of specific modifications to this style which began in the sixteenth century and evolved in various ways during the seventeenth; in particular we may note features such as coloured imprimatura layers, semi-transparent grounds and textured surfaces.

It is apparent from seventeenth century documentary sources that the most important artistic conceptual development with grounds was the introduction of coloured imprimatura layers; essentially this was an isolating layer of pigmented or unpigmented oil and could vary from a transparent yellow-brown film to thick opaque coats one on top of another. Research has suggested that the imprimatura was being used in the early sixteenth century at least, though it is clear that it developed from the oil applied to the chalk grounds in the traditional Netherlandish technique preceding.

The significance of the imprimatura from an artistic point of view was that it could be used to provide a toned ground ‘in the Italian manner’. However, northern European use of the imprimatura seems to have been distinct from Italian methods and we find the particular features appearing that were mentioned above. In support of this, a study on the introduction of coloured grounds showed that a number of stylistic changes took place in Netherlandish painting during the sixteenth century which are customarily attributed to an Italian influence. Significant technical developments took place in parallel and which apparently relate to those thought to have originated in Venice, notably the use of a relatively coarse canvas and strongly coloured ground or preparatory layers. Moreover, it is possible to argue that the changes from panel to canvas supports and from white or pale coloured grounds to highly chromatic grounds were both contemporaneous and intimately related. In Netherlandish/Dutch paintings of the sixteenth/seventeenth century therefore the use of coloured grounds can be considered to reflect the adoption of Italian techniques (especially for canvas painting). We must also take into account at this point however that there were ground structures which appear to be peculiarly northern – an example of this is the red/grey grounds found on North European canvas paintings from the early seventeenth century on, a practice we know continued in France into the eighteenth century. The wide use of grey imprimatura layers is apparently a northern predilection as are the textured surfaces and wood-grain effects we shall discuss in a moment.

The coloured imprimatura of course had an enormous impact on the pictorial effect that artists could achieve – it could be used as a mid-tone, with the artist laying in the highlights and the deep shadows while leaving the imprimatura showing. A degree of standardisation also appears to have occurred since some of the seemingly idiosyncratic forms were surprisingly widely used – again we might mention the red/grey double grounds found on canvases, but also the textured creamy imprimatura. This might prove to be due to specialists (imprimeurs) who prepared panels and canvases commercially for others, much as artists’ colourmen do today; the important seventeenth century documenter of artists’ techniques, de Mayerne, records a recipe for a red/grey ground given to him by a Walloon imprimeur working in London in the early seventeenth century. Precise dates for the introduction and disappearance and the geographical usage of grey imprimatura layers are not available, but we might cite some examples such as a Massacre of the Innocents by Cornelis Cornelisz. van Haarlem in the Frans Hals Museum dating from the late sixteenth century and an Anne of Denmark in the National Portrait Gallery, London.

Another type of ground we encounter is one with creamy-white textured layers on the surface, composed of lead white and chalk – probably the pigment called ceruse in the seventeenth century. It is brushed broadly over the surface, the resulting texture being left in rather than smoothed. Over this imprimatura we may find another thin wash of a translucent brown pigmented oil; this gives a marked striated character to the surface. Examples of striated imprimatura abound in north European paintings of the period; early instances have been observed by the author in the work of Pieter Brueghel the Elder although the most famous example from the early seventeenth century must be the oil sketches of Rubens. The textured creamy white ceruse imprimatura does not appear to have been commented upon in the literature however; it appears to be a mid-seventeenth century characteristic.

From this extended discussion of sixteenth and seventeenth century technique in northern Europe it should be apparent that much can be learned from patterns of use, not only of individual materials, but also broader aspects of technique. Legitimate and fruitful areas of research cover not only specific artists and their workshops, but also the broader questions of how this fitted into societal structures as a whole. Again, given a particular technique – say, the use of coloured grounds or striated imprimatura layers – we can look at where it originated and how it spread; this is not dissimilar to the way a scientist such as an epidemiologist might look at a disease to discover its point of origin and mode of transmission.

Attribution and authenticity

Attribution and authenticity together form perhaps one of the most controversial areas in the whole field of the study of paintings. Unfortunately (for the author) discussion of it is something that cannot be avoided, the fundamental reason for this here being the emphasis in scientific analysis on the quality of the data used for research, a reliable reference point for comparisons being essential.

Clearly, the kind of information I have discussed above about imaging techniques can have considerable bearing on the attribution/authenticity of a painting. However, there is a whole range of other technical data that can be brought in to play when making these judgements. At the simplest level in scientific attribution/authenticity studies one is proposing a hypothesis and testing it: “This is a painting by Rembrandt/This painting contains the pigment titanium dioxide white/Titanium dioxide white was not introduced until the twentieth century, hence unavailable to Rembrandt/Therefore this is not a painting by Rembrandt”. Only rejection is certain and unfortunately, when applied without appreciating this subtlety, one can be led into some quite absurd judgements: if you are hypothetically committed to the painting being by Van Gogh, then the absence of contrary proof does not, sadly, make it a Van Gogh, merely not a disproved one. In practice such an approach proceeds by identifying the materials (and techniques) present in a painting and comparing this data to what we know of their use. I can explain this most clearly from examples where pigment analysis is involved and while this is not straightjacket might at first sight appear to severely constrain what we can say ‘scientifically’ about a painting, some examples might illustrate the range.

There are a number of pigments for which we know reasonably well their date of introduction, use and disuse. Examples are lead-tin and Naples yellows, Prussian blue, synthetic ultramarine, various chrome and cobalt colours, zinc and titanium whites. There is also a list of pigments used since antiquity – lead white, an assortment of earth colours such as ochres, umbers and siennas, vermilion, natural ultramarine and azurite blues among them. A couple of examples will illustrate:

Lead-tin and Naples Yellows: The historical use of both lead-tin and Naples yellows is reasonably well known. Lead tin yellow, as its name suggests, is a yellow compound (oxide) of lead and tin and it was generally known historically as massicot or giallolino, depending on the part of Europe. It is a manufactured pigment that was associated historically with the glass industry and was historically important in paintings between about 1300 and the late seventeenth/early eighteenth centuries; after this time lead tin yellow went into a period of disuse, during when much knowledge of it was lost. Scientific analysis rediscovered it at the beginning of the 1940s, when it acquired its present name. As a result of the effective loss of knowledge of the pigment it is a useful date marker since between the earlier eighteenth century and 1940 it would generally have been substituted in imitations by its effective replacement in the palette, the chemically similar Naples yellow.

If we examine the occurrences of lead tin and Naples yellows on paintings we can see that broadly speaking lead tin yellow first appeared in the artist’s palette around the beginning of the fourteenth century, followed by a brief decline before an escalation around the middle of the fifteenth century to the point where lead tin yellow was the major yellow pigment. Such widespread use continued until some time in the latter part of the seventeenth century (the period 1675-99) when it apparently fell completely from the range of pigments employed in painting. Naples yellow on the other hand is first encountered around 1630 in Neapolitan painting; from a gradual introduction and dissemination through Europe we find a steeper rise in occurrences from around 1700, probably because of its use as a replacement for lead tin yellow. K¸hn and Zocher found in a study of 386 paintings from 1600-1900 that overall 30% contained Naples yellow, 4% in those from the seventeenth century, 54% from the eighteenth and 46% from the nineteenth; this underlines the rarity of this pigment in paintings before 1700.

This crossover between the two pigments thus provides a useful boundary, basically because Naples yellow was perhaps the closest substitute for lead-tin yellow; consequently, where the latter might have been used the former subsequently took its place. Hence we tend to find this type of substitution in eighteenth and nineteenth century copies of seventeenth century and earlier works.

To take another example, the introduction and use of Prussian blue shows how a new material can rapidly achieve, if not ubiquity, then certainly widespread use on the back of strong economic incentive. There were a number of ‘traditional’ blue pigments – natural ultramarine, azurite, smalt, verditer, indigo – all of which had some fairly profound disadvantage: ultramarine and azurite were good but expensive, smalt, verditer and indigo had deficiencies in colour and as stable pigments. Historical factors in the seventeenth century also led to fluctuating availability of some of these. Therefore, when Prussian blue was discovered in the early eighteenth century it provided a much needed addition to the palette. Strongly coloured (though not all found its shade satisfactory), stable and, not least, cheap, this pigment very soon became firmly established in the pantheon of popular pigments. In consequence it provides another (strong) marker for date.

There is more than the provision of terminus ante and terminus post dates possible though. Geographical differences would be another application and we might cite the almost overwhelming division in the way panels were prepared in early Netherlandish and Italian paintings – the former used natural chalk while the latter employed calcium sulphate. A number of other examples of this kind of localised variation could be cited. However, there is another important but subtle point to be made here. Some judgements can be made completely – the use of a pigment of known discovery for example – while others are questions of probability. There is a strong historical tendency to exclude probability statements about paintings – the widespread, popular feeling that either somebody did or did not paint a picture, that there is no halfway house. In historical research of any kind where the record is partial (nobody was there to video Van Gogh painting the Sunflowers) judgements are made on likelihood; this applies equally to, say, palaeontology, as art history.

Clearly we can also look at the frequency with which materials are used (you would be surprised how often a pigment has to be the earliest example of it being employed in painting!). Put as a simple example, if we found only pigments from the range available since antiquity I mentioned earlier, what is the probability the painting comes from the nineteenth century? To do this we need a good idea of the probabilities of finding any particular material at any given time and place, for which an extensive database of information is required. No generally available examples of such technical databases exist although individual ones do (I have built up over the years a fairly large one with such information running to some thousands of paintings).

[A short aside is also appropriate at this point on the analytical techniques used to study materials and we need to mention methods for the following:

• pigment analysis;
• media analysis;
• study of layer structure.

Pigment analysis is usually carried out by one or more techniques. Of these the most common is probably what is called ‘polarised light microscopy’ (PLM). A method which derives from geology and the identification of minerals (‘optical mineralogy’), it is used to characterise the appearance of pigment particles and from their morphology determine what they are. Many common pigments can readily be differentiated this way. However, other techniques are also used to make certain; these can be roughly categorised into methods which tell you about what elements are present (such as the technique sometimes known as electron microprobe, but which goes by other names) or the crystalline structure (particularly that known as X-ray diffraction, ‘XRD’).

Media analysis is used to look at the material which binds the pigments in the paint. Most commonly this is either by the use of chemically selective coloured stains used microscopically or else by the technique of gas chromatography or gas chromatography-mass spectrometry (GC and GC-MS).

Finally, layer structure is studied by taking paint fragments which are embedded in plastic and then ground and polished from the side so as to reveal the strata present. This can give us information on not only the sequence, thickness and composition of paint layers the artist applied, but also subsequent changes.

Small samples are generally required to be taken for all these techniques, typically less than pinhead size.]

So far we have concentrated largely on materials issues – what we can tell from the presence of particular supports or pigments for example. We have also been working sequentially through the structure of a painting from panel and canvas, ground to paint. We might though complete this discussion by looking at how a specific artist combined his materials into a very characteristic technique: Sir Edwin Landseer, a somewhat reviled figure in more recent years while much admired in his lifetime, but an artist with a surprisingly wide range of expression in his painting methods.

Landseer produced drawings, sketches or studies and partially and wholly completed works throughout his career. From an examination of these we can gain some notion of his basic working methods. For many of the paintings there are known chalk or crayon drawings, and it would seem that Landseer would not infrequently sketch components of the overall composition in this manner. He might further produce a more ‘worked-up’ version of a composition in water-colour. However, our principle interest here probably begins with the oil sketches, since these provide clues as to his working methods in the lower strata of the finished paintings.

The oil studies vary widely in their degree of completion, ranging from lightly sketched works with fluid outlines of the elements of the composition with the broad colour washed thinly in to substantially finished pictures. This use of a simple brush sketch, locating the individual figures and their background, making any necessary adjustments and then blocking in the underlying colour probably underlies the finished works too; close examination often shows traces of the brush sketch in areas left largely bare in the final work, the wash of an underlayer frequently being revealed as well.

It is also clear that Landseer would transfer the designs he developed freehand to the final canvas. Apart from an early painting of a Bull Attacked by Dogs (1821, Private Collection) where the canvas has been ‘squared-up’, no evidence seems to have been otherwise found of such transfer techniques. Instead, under-sketching, where identified on finished paintings, is invariably free with no squaring-up lines. There are frequently minor pentimenti adjusting outlines and so forth, consistent with the artist modifying the details as he worked. Major alterations are less frequently encountered (although without benefit of X-radiography or IR reflectography on a large group of works we cannot be wholly certain); an example of these greater changes of mind being the Arab Tent in the Wallace Collection where the positions of the head and hoof of the mare have been shifted significantly.

At the same time as Landseer applied the brush sketch, he would also be laying in blocks of colour beneath the main elements of the design. This would serve as an underpainting which could modify the thinner scumbles and glazes to be applied on top and would probably be in a purer hue than the final appearance. There were many examples of this in both the sketches/studies examined and the final works where under-layers could be seen.

Some allusion has already been made to techniques Landseer used in the main paint layers, the most characteristic of these being his exploitation of the handling properties (rheology) of the paint itself. However, it is clear from the examination of Landseer’s paintings that there are other consistent features of technique as well that largely persist as a vocabulary of method throughout his career. We might list the following:

• Leaving areas largely without paint so that the ground, washes of underpainting and sketching lines are visible; normally these are to be seen as small regions at the junction of several elements of the composition.
• Use of relatively wet paint that is fluidly applied and ‘ploughed’ to leave distinct raised edges around the figures and other elements of the composition.
• Laying in of the background up to the objects of the composition late in the development of the painting.
• Extensive exploitation of bristle marks from brushes to reveal underlying colour that contrasts with the paint being applied, as well as inscribing with an implement such as the reverse end of a brush into wet paint to achieve a similar effect.

In addition there were frequent examples throughout the examined paintings of Landseer where the scraping or wiping out of passages was evident.

Condition and conservation – restauration

The condition of a painting would seem to be an absolutely fundamental consideration in any attempt to understand a painting on the physical level. It affects many areas – to what extent are we seeing the artist’s ‘original’ intention, for example ?

Most, if not all, of the investigatory techniques discussed above are capable of providing information about the condition of a painting. Before describing this application of them though there is at least one other analytical method to introduce – the ‘black light’, or ultraviolet (UV) fluorescence examination; this needs some comment as perhaps both the most used and abused.

The phenomenon of fluorescence is, simply, where a substance absorbs light at one wavelength and then re-emits it at another (longer) wavelength. Typically, light of UV (“beyond the violet”), blue-violet or blue wavelengths are absorbed and then re-emitted in the visible range and it is this emitted radiation which is observed as fluorescence. Importantly, as the resins used as varnishes age they acquire a fluorescence – under illumination with UV-blue light, they tend to glow with a dull greenish colour. Any disturbance to this layer will show; losses, perhaps from varnish removal using solvents, or additions, such as restoration overpaint, stand out. What cannot be forgotten though is that we are primarily seeing a surface effect and little of what lies beneath the fluorescent varnish will be revealed. A painting may be in an extremely damaged state, but if it is covered by an old undisturbed varnish you may not be able to tell by UV fluorescence.

As an example of condition let us look at a rather extreme case. In 1987 Madame Marie de Balkany bought a painting at auction: it appeared in the sale catalogue as by Egon Schiele – a Youth kneeling before God the Father – signed, with the initials E and S, and said to be an important early work of about 1907. The bid which secured the painting was £500,000 plus the buyers commission of 10%. A couple of years later a new catalogue on Schiele by the American art historian Jane Kallir appeared; in this, strong doubts were cast on the attribution of Madame de Balkany’s painting, so not surprisingly she then approached the auctioneers with a view to getting her money back. Unfortunately it transpired that the only way this could be achieved was to satisfy the auction house’s Terms and Conditions of Sale, which demanded that the painting be proved a forgery. Simplifying slightly, the legal case then consequently rested on these Terms and Conditions, forgery being defined there as ‘a lot created in wholeor in part with an intention to deceive’. When the trial came to the High Court in December 1994, a number of facts came out. Firstly, concerning the technical evidence, it was possible to establish that the painting was substantially damaged. The UV fluorescence image was, at the least, very confused; closer study showed the probable extent of damage – for example, by shining a light across the surface it was possible to see underlying paint losses with later paint (‘overpaint’) dipping down into them. Elsewhere it was clear that paint lay over cracks; this further underlined the fact that we were dealing with a heavily restored picture and I calculated that overpaint covered roughly 94% of the image.

Moreover, because the initials E and S were on top of the overpaint they must therefore be later additions. To confuse matters though, paint cross-sections taken through the initials had showed a paint layer beneath both that might be an ‘original’ signature; however, using the electron emission radiography technique described earlier it was also possible to show that there was an ‘ES’ monogram in the bottom right-hand corner. There was some art historical debate as to whether Schiele would have used this type of monogram as very few examples like this one are known on his paintings and some of these were even thought to be fake signatures. The final judgement found in favour of Madame de Balkany. Essentially this came down to a decision that there had been an intention to deceive by persons unknown, though the judge felt that the painting was a genuine Schiele (apparently because of the monogram rather than that the stylistic evidence was conclusive). Moreover, the auction house was explicitly found to have been negligent in not spotting the condition problems. Clearly the judgement raises a number of issues for the art world, a couple of which you might like to think about:

If 94% overpainting constitutes an ‘intent to deceive’, at what point does it not? 90%? 50%? 20%? 0%? To what extent are you participating in forgery when a painting is restored?

The negligence decision implies that auction houses (and probably other vendors) need to take great care when cataloguing works for sale. This may in future mean either more ‘in house’ advice, or else a clearer message to potential purchasers that they should have appropriate expert.

Dr Nicholas Eastaugh

It is probably apparent by this point that scientific techniques have an enormous and diverse role to play the examination of paintings from any age. What is now loosely being called ‘technical’ art history has reached a point of some maturity and the combination of a grasp of the analytical methods most appropriate and a sufficient body of information now allows us to tackle a wide range of fundamental questions in the field. The benefits are potentially enormous to our understanding of both detail (individual paintings or artists) and broad issues (times, places, movements). Why not take advantage?

Dr. Nicholas Eastaugh
Technical Studies in the Fine Arts
1 Park Street
Teddington
Middlesex TW11 0LT UK

Tel.: (Int. + 44) (0)181 943 4448
Fax.: (Int. + 44) (0)181 943 2228
Email: neastaugh@cix.co.uk

Dr. Nicholas Eastaugh: After graduating with an honours degree in physics from Durham University, Nicholas Eastaugh trained in the conservation of paintings at the Courtauld Institute of Art, from where he also gained his PHD with a thesis relating to the history and scientific study of artists’ pigments.

After four years as a lecturer in conservation science at the Textile Conservation Centre at Hampton Court and as a freelance consultant on the history and analysis of methods and materials in paintings and other artefacts he took up a Fellowship at the Canadian Conservation Institute before returning to this country nine years ago to develop a consultancy service specialising in technical studies in the fine arts.

He has gained a considerable reputation through his expertise in the scientific examination of paintings and other objects, working on projects for a wide range of public bodies and private clients from both the UK and overseas. Nicholas Eastaugh also frequently publishes scholarly articles dealing with issues in the scientific examination and conservation of artefacts and lectures widely.

Dr. Eastaugh also has a number of special research interests including imaging techniques (for example demonstrating a system he developed for infrared imaging on the BBC1 television programme Tomorrow’s World in 1992 and a feature on an X-ray technique he developed in New Scientist) and dating methods for paintings.