Infrared Reflectography: Invention or Transfer

What infrared reaches, and what it is blind to

Artists ordinarily drew their composition on the prepared panel or canvas before painting, and that drawing was meant to disappear. Infrared radiation penetrates most pigment layers, but it is strongly absorbed by carbon, so a drawing made in charcoal, black chalk or carbon-based ink registers as a dark line beneath paint that has become transparent to the camera. That is the entire physical basis of the method.

The technique was developed by the Dutch physicist J. R. J. van Asperen de Boer, using a camera fitted with an infrared vidicon that transmitted the image of a small area of a panel to a television screen. Its advantage over ordinary infrared photography was a clearer rendition of the different handwritings of panel preparation. Modern practice uses indium gallium arsenide arrays. The Bosch Research and Conservation Project, to take a fully documented example, imaged with a camera whose array responds from 900 to 1700 nanometres, documenting 400 by 400 millimetres of paint surface at 4,096 by 4,096 pixels, roughly 260 pixels per inch, with a macro setting covering 36 by 36 millimetres at about 2,884 pixels per inch, and then stitched and coregistered the tiles so that infrared and visible images could be overlaid exactly.

The blindness matters as much as the reach. As Maryan Ainsworth put it in 1993, the problem of underdrawings and workshop practice in Lucas Cranach and his circle is not easily resolved by underdrawing analysis, because in many cases the drawing was made in a non-carbon substance such as iron-gall brown ink, which infrared does not penetrate and which is therefore invisible on the reflectogram, even where traces can be seen on the surface with the naked eye. A report that says no underdrawing was found has not established that there is none.

The distinction it makes cleanly: searching line against transferred line

What reflectography reads better than any other instrument is the difference between a drawing made while a composition was being decided and a drawing made to reproduce a composition already decided. The first hesitates, doubles its contours, redraws a hand, moves a hem, and diverges from the painted surface above it. The second is continuous, confident and identical to the finished picture, because it was traced, pounced or copied from a pattern.

The Metropolitan Museum's Christ Appearing to His Mother is the standard demonstration. Acquired in 1921 and long regarded as one of the museum's greatest early Netherlandish paintings, it was held to be Rogier van der Weyden's original, with the Berlin version dismissed as a workshop copy. In 1991 and 1992 Rainald Grosshans published the underdrawing of the Berlin panel, which showed Rogier's characteristic hook-ended drawing style with several changes of composition. Maryan Ainsworth then found that the underdrawing of the New York panel was mechanical, executed without changes, and probably taken from a pattern or a tracing. Grosshans further observed that the perspective of the New York picture had been subtly altered towards a one-point scheme belonging to the later fifteenth century, and the dendrochronological report placed the panel at around 1485, twenty years after Rogier died. The reflectogram did not close the case by itself. It reframed the picture as a reproduction, and the other evidence then fell into place.

The same reading works in reverse. Beltracchi's forged Campendonks were compared with genuine Campendonks whose reflectograms show real deviations between the drawn design and the executed painting, mapped section by section. A copyist has no such deviations to show unless he manufactures them, and manufacturing them convincingly at the drawing stage, under a paint film, is a considerably harder problem than manufacturing a surface.

Raphael's Madonna of the Pinks: a drawing a copyist had no reason to make

The Madonna of the Pinks was famous in the middle of the nineteenth century and then demoted, treated as one of many copies of a popular composition, and it was rediscovered as Raphael's original in 1991. Infrared examination supplied the decisive technical argument.

The reflectogram shows an underdrawing in metalpoint, a delicate medium in which the figures and drapery are freely outlined, with Raphael's characteristic loops for toes and arcs for knuckles and careful hatching to indicate shadow. Raphael followed the principal outlines closely when he painted, but he also made several changes, notably to the bodice and neckline of the Virgin's dress. The National Gallery's argument is precisely calibrated: the technique and style of the underdrawing are comparable in particular with the Garvagh Madonna, and the free hand and inventiveness shown in it are not compatible with the painting being a copy.

That is the strongest form the argument takes, and it is worth stating exactly what it claims. It does not claim that only Raphael could have drawn those lines. It claims that nobody copying a finished Raphael would have produced a drawing that hesitates, invents and then departs from its own decisions. The evidence is about the process, and the name follows from comparison with a securely attributed work in which the same process is visible.

Bosch at the Prado: comparable evidence, opposite verdicts

The clearest demonstration of the method's ceiling arrived in 2016, when the quincentenary of Hieronymus Bosch's death produced two rival retrospectives. The Bosch Research and Conservation Project, a nine-year Dutch undertaking that documented the paintings with infrared reflectography, ultra-high-resolution macro photography in both visible and infrared light, and digitised X-radiographs, all stitched and coregistered, concluded that the autograph oeuvre numbered twenty-four works. The Museo Nacional del Prado, whose own catalogue expressed conspicuously little gratitude to the Dutch team, put the number at twenty-seven.

The disagreements are specific and they turn on underdrawing. The Dutch team reassigned Extracting the Stone of Madness to the workshop or a follower, dating it 1500 to 1520; the Prado shows it as Bosch, dated 1501 to 1505, and explains that although the underdrawing technique differs slightly from other Bosch paintings, this reflects the way the artist approached a commission for Philip of Burgundy. The Dutch team gave the Temptation of Saint Anthony to a follower and dated it 1530 to 1540; the Prado's curator Pilar Silva rebutted this, arguing that technical examination and infrared images confirm the picture is by Bosch, citing among other things the modelling of a pig's snout, and dating the painting 1510 to 1515. On the Table of the Seven Deadly Sins, rejected by the Dutch team as workshop or follower, the Prado published infrared images showing considerable changes in the underdrawing, particularly in the figures of Pride, Greed and Wrath, and offered those changes as evidence of Bosch's own hand.

Nobody in that dispute questions the images. Both sides accept that the Table's underdrawing was altered during work; they disagree about whether alteration indicates the master or a competent follower inventing within his idiom. Infrared told them what happened on the panel. It could not tell them who was standing in front of it.

How to read a reflectography claim in a report

Five questions separate a usable technical statement from a decorative one.

First, what was the instrument and the waveband, and at what resolution. An InGaAs array working from roughly 900 to 1700 nanometres sees far more than an older vidicon, and a claim that a picture shows no underdrawing carries a different weight at 260 pixels per inch than at photographic resolution. Second, was the whole surface imaged or only sampled areas, and were the tiles registered to the visible image so that a drawn line can be located precisely relative to a painted one. An unregistered reflectogram cannot support a statement about divergence between drawing and paint.

Third, what was the drawing medium likely to be. Carbon media are legible; iron-gall inks and metalpoint may be faint or invisible; leadpoint and metalpoint on a chalk ground behave differently again. Fourth, is the comparison a real one. Underdrawing style is only meaningful against a corpus, which is why Molly Faries's archive of more than ten thousand infrared reflectograms, covering artists from the Byzantine period to the nineteenth century, was assembled in the first place: to allow scholars to establish underdrawing styles for individual painters rather than judging one drawing in isolation.

Fifth, what is actually being inferred. Reflectography establishes whether a design was worked out on this support or brought to it. That is a statement about process, and it maps onto attribution only through comparison. A workshop can transfer the master's cartoon; a master can trace his own earlier composition for a repetition; an assistant can paint over a drawing the master made. Multispectral infrared, in which several infrared images are combined into a false-colour reflectogram, separates pigments more clearly than the conventional image and has become standard in museum practice, but no increase in image quality changes the shape of the inference.