American Scientist

Sixteenth-century France was a society in which religious teachings coexisted uneasily with superstition. Unusual events in the natural world were probed for hidden meanings; an occurrence as remarkable as a total eclipse of the Sun must, it was thought, be an omen of some sort. It was in this context that Antoine Caron, a painter in the court of Catherine de Medici, depicted the scene of a solar eclipse shown on the facing page. The painting now hangs in the J. Paul Getty Museum in Los Angeles under the title Dionysius the Areopagite Converting the Pagan Philosophers. By contrast with Caron’s theological rendering, more recent attempts to understand solar eclipses have focused on the physical forces that cause them and the remarkable visual effects they produce.

Modern research into solar eclipses can be said to have begun around 1860, with a consensus among researchers that the appearance of a luminous area encircling the darkened body of the Sun—the corona—was in fact the Sun’s atmosphere. Normally invisible to earthbound observers, this atmosphere could be seen during solar eclipses only because the Moon was then blocking the much brighter, direct light of the Sun. The goal of early investigations was to determine how the brightness of the white light revealed during solar eclipses was distributed in space. Although the first investigations were carried out mainly by means of visual observations and photography, and later by quantitative measurements, ultimately it was a series of little-known paintings by the artist Howard Russell Butler (1856–1934) that provided the missing piece of the puzzle needed to fulfill this goal.

Visual Perception Captured by Art

When it comes to the scholarly investigation of visual phenomena, the adage “seeing is believing” turns out to be something of a trap. The image produced in the human brain by what we “see” can differ considerably from the image presented in a photograph, and nowhere is this truer than in efforts to capture accurately the visual perception of the Sun’s corona. Butler’s paintings captured what the eye saw, more credibly than any other depictions either before or since, thanks to his unusual talents. His extraordinary skill of being able to paint from memory what he had glimpsed for only a brief time—such as landscapes during sunrise and sunset—served him well in painting solar eclipses, which likewise last only a couple of minutes. He also painted portraits, his best-known subject being the eminent philanthropist and industrialist Andrew Carnegie. In this genre, he developed another skill that served him well: Wishing to shorten the sitting time of his subjects, he developed a system of shorthand sketching to record particular colors, shades, and features for later reference.

Butler’s most important qualification, however, was that he had studied physics and electrical engineering at the College of New Jersey (now Princeton University), receiving his science degree in 1876. This academic background put him in good stead when he was invited to join the U.S. Naval Observatory’s eclipse expedition in 1918. Although he had never yet seen an eclipse, he drew up meticulous scientific plans and executed them with the same artistic ability he used to paint other transient phenomena. Astronomers who saw both the eclipse and Butler’s painting looked upon the latter as a marvel of perfection, true to both form and color, a work of art that had the added advantage of being both accurate and realistic.

A comparison of Butler’s painting with the two photographs that appear below it (left) raises a number of questions. Why is the distribution of brightness in the photographs different from that portrayed by Butler? Why are the streamers of the paintings absent from the middle photograph displaying the globular corona? Between painting and photograph, which—if either—representation is true?

Imaging and Reality

To answer these questions, there must be a standard set of criteria that can be applied in each case, along with the premise that images are merely a display of quantitative measurements, or, in other words, numerical values of brightness distributed in space. Because quantitative measurements represent the true brightness, they are essential for understanding and validating images for scientific use. For instance, a useful quantitative way of representing how white light is distributed throughout an image is with isophotes, or contours along which brightness is constant, like the lines of constant elevation on a topographic map. An example appears on the right in the image of the globular corona of the 2006 solar eclipse and its corresponding isophotes. (That recent solar eclipse is chosen because by then isophotes were being more accurately measured than they had been in earlier eclipses.)

When the isophotes are superposed on the globular corona, the isophote closest to the boundary matches the shape of the corona. This is to be expected, because the perceived shape of the corona is determined by the threshold sensitivity of brightness, and the isophote closest to the shape identifies its level. When less light is captured in an image made with a shorter exposure time, the imaged corona shrinks, but its shape continues to match the relevant isophotes at lower altitude.

Representing brightness in terms of isophotes reveals another key quantitative property of the white-light corona that is not evident from the image itself. Its brightness falls steeply with radial distance from the disk of the Sun. In just one solar radius from the Sun’s limb, brightness drops by more than a factor of 100. With the limited dynamic range of photographic imaging, the brightness of the inner corona of an image showing the extended corona is saturated. Still, the coincidence between the shape of the outer corona (where it is not saturated) and isophotes at all altitudes validates the globular corona as representing the true distribution of white-light brightness—that is, the reality.

Illusion in Perception

One may wonder, then, why isophotes cannot be discerned in the shape of the corona painted by Butler. This apparent contradiction arises from the physical constraints underlying our sense of sight. The human eye and brain have an enormous dynamic range, which allows us to see on a sunny day and also in moonlight, but they can achieve this range only by changing their overall sensitivity, using a phenomenon known as brightness adaptation. Without being consciously aware of this phenomenon, we nevertheless take it into account when we walk from a well-lit lobby into a dark movie theater and wait for our eyes to adjust to the abrupt darkness.

Brightness adaptation comes at a cost. We can discriminate changes in daylight and among shades of darkness when we view them separately, but when they appear next to each other, our ability to discriminate changes simultaneously is much smaller than the adaptation range. In everyday life this limitation isn’t a problem; it is extremely rare for the eye and brain to encounter an object such as the corona with more than two orders of magnitude of simultaneous dynamic range of brightness. Confronted by such an extraordinary sight, the priority of the eye and brain is simply to identify objects in the visual field: No wonder it copes with this challenge by forming an illusion. The question then becomes, What is the nature of the illusion?

Artifacts of Processed Images

An answer to this question emerged in the 1960s, when the High Altitude Observatory in Boulder, Colorado, developed a specialized camera that apparently came close to functioning like the human visual system. The Newkirk camera is equipped with radially graded filters to compensate for the steep radial falloff in brightness. As in Butler’s paintings, the shape of the corona in the processed images produced by the Newkirk camera bears no resemblance to contours of the isophotes, and unless corrected for the filtering, do not represent the true brightness distribution of the corona. The apparent jagged shape of the corona is, therefore, an artifact. Moreover, the striking resemblance of coronal streamers in Butler’s painting of the 1932 eclipse to those in the Newkirk-photographed image of the 1994 eclipse tells us that streamers are an artifact of the processed image, as well as an illusion in visual perception.

The insight provided by the Newkirk camera into how the naked eye sees the corona is invaluable. By showing how the streamers are formed, it reveals the workings of the optical illusion. This outcome is not too surprising, because the purpose of the Newkirk camera is to capture the large dynamic range of brightness of the corona in a single exposure by suppressing the brighter inner corona—a goal apparently similar to that of visual perception.

Artifact But Not Illusion

Coronal streamers are not the only artifact of the processed image. Another is the so-called coronal holes that often appear at the poles as dark regions at the base of the corona, extending and diverging with increasing height. Such coronal holes are conspicuously absent from the illusions captured in Butler’s paintings, as seen on these pages and also in other paintings shown in my 2010 article in American Scientist. The Newkirk camera apparently suppresses the brightness of the inner corona to the point where it is not detected in the coronal holes, on account of inadequate brightness sensitivity of the image. When the brightness of the processed image is increased (as shown at below), the holes disappear and the processed image resembles Butler’s paintings.

Although the sight of polar coronal holes is an artifact, it is not an optical illusion like the sight of coronal streamers. The holes do not appear in quantitative measurements, nor are they evident in paintings or in unprocessed images of the globular corona. After decades of coronal imaging, it is surprising to learn that the diverging polar coronal holes commonly observed in most eclipse photos today do not in fact exist. Instead, they show up only in processed images—that is, those that have been processed to adjust the brightness—or else as a consequence of the selected radially graded filter. Thus, they are an artifact in two ways, being both human-produced and human-selected.

In the words of perceptual psychologist Claus-Christian Carbon, “It may be fun to perceive illusions, but the understanding of how they work is even more stimulating and sustainable.... Illusions in a scientific context are not mainly created to reveal the failures of our perception or the dysfunctions of our apparatus, but instead to point to the specific power of human perception.” Butler’s paintings show not only that human perception determines that the corona is brightest at the base of the corona but also that the brightness is unbroken all the way around the Sun. The ability to recognize this and to exclude coronal holes from its illusion attests to the power of human perception.

Video courtesy of NASA's Jet Propulsion Laboratory at the California Institute of Technology.