I love to be outdoors during solar eclipses, enjoying the universe appearing to darken around me while my research observations get under way. Long ago I used to suggest that people make a pinhole projector or even use cheese graters from their kitchens to watch these events. But in recent years the availability of partial-eclipse filters for only a dollar or so has made such advice obsolete. Now anyone can glance up at the sun through such a filter starting more than an hour prior to totality and see an apparent bite being taken out of the solar disk. During the last few minutes before totality, you will notice that the ambient light changes in quality, becoming eerie. Shadows sharpen because they result from a thin crescent of sunlight rather than the full disk of the sun. The air cools, and a wind stirs. Shadow bands may sweep rapidly over the ground.

With seconds to go, as the moon moves completely in front of the sun, just a few shafts of sunlight leak through valleys on the moon's edge, reducing the sun to an arc of bright beads. These fade out until only one is left—so bright that it looks like the diamond on a ring, perhaps with a narrow, reddish rim to its sides and a whitish band all around the lunar silhouette. Then the diamond, too, disappears. You can and should drop your filters and look straight at what is left of the sun, a region of its atmosphere that had been hidden by the blue sky.

This is the inner and middle solar corona, the plume of plasma that flies out from the sun's surface. It is about as bright as the full moon—a million times as faint as the everyday sun—and equally safe to look at with the naked eye. You first glimpse the corona as the band of the diamond ring, and then you see it in all its glory: a pearly white halo of gas that extends outward to several times the sun's radius. If you are lucky, you might see a mighty eruption of plasma into interplanetary space.

But what, really, is the point of my trying to describe a total solar eclipse in words? It is so astonishingly moving and beautiful that nobody has ever described it adequately. People routinely come up to me after eclipses to say that they know how I had tried to convey the excitement but that I had nonetheless fallen short. Television and computer screens do not do it justice. Photographs flatten the dynamic range and lose the dazzling contrast. To be outdoors as the universe apparently darkens, gradually at first and then by an additional factor of 10,000 within seconds, is completely discombobulating. It conjures up primal fears of the sun being taken away.

I saw my first eclipse as a first-year college student, and I was hooked. Starting then, I have been all over the world to see 65 solar eclipses (including 33 total eclipses). I look forward to my 66th on August 21, when the path of totality traverses from the U.S. West Coast to the East Coast for the first time since 1918.

And I do not catch these events just for the fun of it—eclipses offer scientists viewing conditions that routine observations cannot replicate. Although terrestrial telescopes can be equipped with a small metal cone or disk—making a so-called coronagraph—to blot out the sun on demand, their artificial eclipses are not as good as the real ones. The ambient air molecules leave the sky too blue and bright, even from pristine and high mountain sites. And space coronagraphs have to blot out not only the everyday solar disk but also a wide band around it, or else too much light would scatter inside the instrument. Furthermore, any telescope has a limited resolution and smears out incoming light a bit. Natural eclipses do not have this problem, because the “telescope” is, in effect, the entire Earth-moon system, with an exceptionally high resolution. We link our ground-based observations with spacecraft views to get a complete picture of the sun. Only in the crisp shadow of the moon are we able to see the inner and middle part of the corona in visible light.

It is in those inner expanses that we seek an answer to one of the most nagging puzzles in astrophysics: Why does the sun's temperature increase as you move away from its surface? Usually things cool down as you retreat from a hot object, such as a campfire or a steam radiator. Within the sun, the temperature starts at 15 million degrees Celsius at the center and steadily falls as you move outward, dropping to 5,500 degrees C at the solar photosphere, the surface that emits sunlight into space. But then the trend reverses. The tenuous gas just above the visible surface climbs back up to over 10,000 degrees C and abruptly leaps to millions of degrees. Scientists still debate the details of how that occurs.

We have made tremendous observational and theoretical advances since I first described the science of the solar corona in Scientific American in 1973. A flotilla of spacecraft now monitor the sun in ultraviolet light and x-rays, which we cannot view from the ground, and researchers have developed sophisticated tools to link all our observations together. We know the outline of the solution of the coronal-heating problem—that it involves the sun's magnetic field—but the details remain murky. And this is hardly the only problem that the corona presents to us. Observations during the upcoming eclipse should help tackle these questions.

The Solar Landscape

Scientists already understand much about the solar corona. For one thing, it looks like a giant porcupine. It is drawn into fine streamers, some of which are wider at their base and come to a peak at higher altitudes, like pointy helmets. The shape they form varies with the sunspot cycle.

When spots proliferate, as in the years 2012 through 2014, streamers burst out even from latitudes as high as 30 degrees north and south so that the corona appears round overall. During sunspot minimum periods, such as the one we are in, the corona is squat, and the streamers we see are limited to regions nearer the sun's equator, and thin, straight coronal plumes appear at the poles. From the open regions between streamers, a flow of charged particles called the solar wind escapes outward into the solar system at hundreds of kilometers per second, perhaps twice the rate of the solar wind from other regions. At the base of the corona, anchored to the solar photosphere, are small loops of gas, perhaps made up of multiple threads too fine for our current observations to discern. These coronal loops may pulse as waves bounce back and forth along or through them.

Credit: Don Dixon

All this delicate intricacy is the product of the solar magnetic field, which arises from churning gas deep within the sun. What researchers do not know, however, is exactly how the dynamics of the magnetic field are responsible for the bizarrely high temperature of the corona. We know the magnetic field is involved because magnetic processes are not subject to the same thermodynamic restrictions that prevent energy from flowing by heat conduction from the hot surface to the even hotter corona.

Explosions or Waves?

Scientists have two main ideas for how the sun's magnetic field could transfer some of its energy into the corona to heat it up. One way is through extremely tiny solar flares. These explosions occur when the magnetic field undergoes an abrupt change in its configuration, within seconds. When you map out the field at the sun's surface, you occasionally see the north and south polarities in sunspot regions become jumbled. This condition puts the magnetic field under enormous stress, and to relieve it, the two polarities suddenly connect in a new way, emitting tremendous amounts of stored energy. Such a reconnection heats the corona locally to 10 million degrees C or higher, gives off a bright flash, and sometimes ejects plasma into space. The flare can zap spacecraft orbiting Earth and could pose a serious risk to astronauts journeying to Mars.

The flares we observe are too intermittent to explain the baseline temperature of the solar atmosphere, but might explosions too small to detect individually also wrack the corona? James Klimchuk of NASA's Goddard Space Flight Center has especially championed the idea of such nanoflares. Millions of small explosions going off in the corona every second, each with a billionth as much energy as a large flare, would keep it broiling hot.

The main competing set of theories is that oscillations in the magnetic field heat the corona. Vibrating loops in the lower corona could shake the surrounding gas, thereby raising its temperature. These waves could take multiple forms. Scientists have ruled out sound waves, driven by gas pressure, but Alfvén waves, driven by magnetism or by a hybrid of the two, called magnetoacoustic waves, are still viable. Could magnetic waves of some kind be enough to raise the coronal temperature to millions of degrees?

In principle, researchers should be able to distinguish between the nanoflare and wave mechanisms by measuring oscillations of coronal gas. Fluctuations with periods from about 10 seconds to minutes would betray the passage of standard Alfvén waves along coronal loops. Observations of vibrations of the sun's surface using a technique known as helioseismology suggest that the sun is capable of triggering such waves. Although its strongest oscillations occur with a comparatively languid period of about five minutes, those are only one type among many undulations that the surface undergoes.

Eclipse observations could be crucial to measuring fluctuations in coronal loops. The logistic advantages of observing from Earth allow us to use equipment that has higher temporal resolution than exists on any current spacecraft. My team uses rapid-readout charge-coupled devices (CCDs) that capture images numerous times a second. By comparison, the Atmospheric Imaging Assembly cameras on NASA's Solar Dynamics Observatory (SDO) have been taking observations through several of their range of 10 filters every 12 seconds, and the Solar Ultraviolet Imager on the National Oceanic and Atmospheric Administration's new Geostationary Operational Environmental Satellite (GOES-16) has a 10-second cadence at best for its six filters.

SEQUENCE from the annular solar eclipse of February 2017, as photographed by the author in the Patagonia region in Argentina. Credit: Jay M. Pasachoff (photographs); Christian Lockwood Williams College (compositing)

What we have found so far extends the realm of possibilities. Some oscillations may have periods shorter than one second, matching a theoretical prediction of a special mode of Alfvén waves that travels along the surfaces of loops rather than through their interiors. But our data are scanty: only a few minutes of such high-cadence observations from a pair of prior total solar eclipses. This year we will be using our complicated CCD apparatus, with filters of astonishingly pure color, to isolate the hot coronal gas to search for the time spectrum of waves again. We hope that our results will help researchers choose between the different theories of coronal heating or even lead them to the conclusion that several mechanisms are at work simultaneously. In the active regions above sunspots, the conditions for flaring are auspicious, and waves are comparatively weak. In quiet regions, however, we may have either waves on small loops or trillions of nanoflares all the time.

Eclipse Tactics

Scientists have devised some tricks for making the most of the exceptional opportunities eclipses offer. Eclipse observations enable us to scrutinize the shape of the corona in high spatial and temporal resolution. Our ground-based eclipse images show detail about eight times finer in each dimension than the best space coronagraph. Eclipses do have the downside of being brief and intermittent, but we compensate by melding data sets from separate eclipses and from different sites during a single eclipse.

For instance, by observing eclipses over the full 11-year solar-activity cycle, we follow changes in the degree of roundness of the corona, which reflects the distribution of streamers at various latitudes, and compare them with other measures of solar activity. I work with astronomer Vojtech Rušin of the Slovak Academy of Sciences on such studies. Although the corona is visible during an eclipse for only a few minutes from any given site, we can combine observations from multiple sites to ascertain changes in coronal streamers and plumes over the hours it takes the moon's shadow to traverse Earth. During the August 21 eclipse, we may even get continuous coast-to-coast observations with viewings from citizen scientists.

A further reason to combine multiple eclipse images is to capture the huge range of brightness in the corona. From individual images taken over many exposures, we can pick out the properly exposed pieces and merge a dozen of them at a time. The widely acknowledged expert in this computer-imaging work is Miloslav Druckmüller, a computer scientist at the Brno University of Technology in the Czech Republic. Given that the corona is about 1,000 times brighter just outside the edge of the sun than it is only one solar radius farther out, we must select the best-exposed parts from dozens of different images and assemble them. Using such composite images from past total eclipses—seen most recently in Indonesia, Svalbard, Gabon, Australia, and elsewhere—my team has measured velocities in coronal streamers, polar plumes, and mass ejections. We hope to add significantly to these observations in August.

Another trick is to exploit the gradual encroachment of the lunar silhouette during an eclipse. As sunspot regions are covered or uncovered by the edge of the moon, telescopes might see abrupt changes in the sun's brightness, allowing us to pinpoint details. To get the very highest spatial resolution this year, my team is collaborating with Dale Gary of the New Jersey Institute of Technology, Tim Bastian of the National Radio Astronomy Observatory and Tom Kuiper of NASA's Jet Propulsion Laboratory to use radio telescopes to measure changes in the radio radiation from whatever active sunspot regions may be visible at various frequencies as the moon covers the sun. Even though these telescopes are outside the path of totality, about 70 percent of the solar disk will still be covered by each of them. We will get the highest-resolution radio observations with the Expanded Owens Valley Solar Array in California, with its 13 linked radio telescopes that will take continuous observations of the sun at hundreds of frequencies from 2.5 to 18 gigahertz. Lower-resolution images from the Goldstone Apple Valley Radio Telescope, also in California, will improve image quality by filling in the background. We should be able to match the exact positions of brightening in coronal loops, as seen at these radio wavelengths, with the spots that glow in the ultraviolet or in x-rays from spacecraft, and thereby learn how the loops are heated.

The magnetic field of the photosphere is well studied, but that of the corona is much less so. To rectify that issue, Ed DeLuca of the Harvard-Smithsonian Center for Astrophysics and Harvard University graduate student Jenna Samra, working with solar scientists Leon Golub of the Harvard-Smithsonian Center and Philip Judge of the High Altitude Observatory at the National Center for Atmospheric Research (NCAR) in Boulder, Colo., plan to follow the eclipse from an NCAR Gulfstream V aircraft. From their perch above the bulk of the infrared-absorbing atmosphere, they will be able to measure the strength of infrared spectral lines, hoping to find ones that are magnetically sensitive.

If successful, they plan to fly again during a later eclipse with polarization filters added to measure the coronal magnetic field. By separating out light waves with different orientations, polarization measurements help us to identify the different components of the corona. The inner middle part of the corona that we see with our eyes during a total eclipse comes from highly ionized gas scattering ordinary sunlight toward us. This scattering polarizes the light, and the motion of electrons caused by this process smears out the dark lines that otherwise intrude in the sun's rainbow spectrum. Farther out in the corona, nearer the orbit of Mercury, dust in interplanetary space bounces light toward us but does not polarize it or wipe out the ordinary solar spectrum. Others preparing to study polarization at this year's eclipse include Nat Gopalswamy of NASA's Goddard Center, Judge and Steven Tomczyk, both at the High Altitude Observatory, and Padma Yanamandra-Fisher of the Space Science Institute. After the Daniel K. Inouye Solar Telescope being built on Maui starts observing in 2018, one of its instruments should eventually be able to measure the coronal magnetic field directly by studying the polarization of infrared spectral lines. And when NASA's Parker Solar Probe launches in 2018, it will fly through the solar corona and help unravel the uncertainties in coronal heating.

Worldwide Effort

All in all, the observing effort during this eclipse will be truly enormous, and I have only scratched the surface here. NASA has funded 11 proposals, six for coronal studies and five related to the response of Earth's atmosphere to the dramatic eclipse cooling, a topic on which I have been working with Marcos Peñaloza-Murillo of the University of the Andes in Venezuela. Another major U.S. research group using eclipses to study the corona is led by Shadia Habbal of the University of Hawaii's Institute for Astronomy. Her team, which she calls the Solar Wind Sherpas, will image the corona through filters chosen to map plasmas of different temperatures. Habbal's new NASA grant supports the enhancement of the group's recently designed dual-channel imaging spectrograph, which was successfully tested in 2015. A variety of observations from the ground and from space will provide the most comprehensive study of the infrared corona, its spectrum and its polarization acquired to date.

My group has had the benefit of international collaborations during the 33 total solar eclipses that I have observed from sites around the world. Now it is time for us to repay the hospitality. We expect the high-quality imaging and analysis of Serge Koutchmy of the Institute of Astrophysics of Paris and his colleagues to contribute to the study of the August eclipse. At my own team's sites, we will be joined by our colleagues from Australia, Slovakia, Greece, Japan, China, Iran, and elsewhere.

Composite of dozens of images taken during a total solar eclipse in Libya shows the chromosphere (pink) and corona. Credit: Miloslav Druckmüller and Peter Aniol Getty Images

Citizen scientists will also have plenty of opportunities to contribute to researchers' eclipse efforts. I am involved in the Eclipse Megamovie Project, which is based at the Space Sciences Laboratory at the University of California, Berkeley, and headed by Laura Peticolas. People will be able to send in images through a Google interface for archiving and assembly into continent-spanning movies, which will be available to citizen scientists for viewing and analysis. In a similar vein, Matt Penn of the National Solar Observatory has organized the Citizen Continental-America Telescopic Eclipse (Citizen CATE) Experiment, a continent-spanning collaboration of about 70 sites with identical small telescopes and CCD detectors.

One unusual experiment this August has nothing to do with the corona; indeed, the corona will get in our way. Arthur Eddington famously tested Einstein's general theory of relativity at an eclipse in 1919. He looked for signs that the sun's mass was bending the light of distant stars behind it, an effect that is actually caused by the relativistic warping of spacetime. I have spent decades telling people that we have better things to do at a total eclipse than repeat this experiment. After all, physicists have more precise ways to test relativity theory nowadays. But it turns out that new observing capabilities may make the investigation at this year's eclipse a useful one—or at least interesting.

Retired California physicist Don Bruns will carry out such observations. He has intricate plans for calibrating his telescope by measuring many nighttime star images. An earlier attempt to use observations taken with digital single-lens-reflex (DSLR) cameras at the 2006 eclipse by Jean-Luc Dighaye of Belgium—which Carlton Pennypacker of U.C. Berkeley and Lawrence Berkeley National Laboratory and I tried to help analyze—failed, but that was with the large pixels of commercial DSLRs. We hope that the smaller pixels and precise calibration of an astronomical CCD detector will succeed. Bradley Schaefer of Louisiana State University has argued that modern imagers have enough resolution and sensitivity to exceed the accuracy of past tests, and he will also try to observe the effect. Because of a just released catalog created by the European Space Agency's Gaia spacecraft, we now know the positions of stars with extremely high precision, so we could look for any deflection caused by the sun with fewer calibrations at the telescope.

Nearly Here

The August 21 eclipse will begin at sunrise in the Pacific Ocean. Totality hits the U.S. mainland in Oregon, with partial phases visible throughout the U.S., Canada and Mexico and farther south into South America. After leaving South Carolina near Charleston some 90 minutes later, the total eclipse will end at sunset over the Atlantic, with partial phases visible from northwestern Africa and western Europe.

Assuming the weather cooperates, scientists and the general public should be impressed and even overwhelmed. Combining ground-based eclipse results with the observations from satellites in the visible, ultraviolet, x-ray and radio parts of the spectrum will provide the most complete view of the solar atmosphere ever seen.

Whatever we conclude for the sun will also apply to the billions and trillions of stars like the sun that we cannot see in the same detail. Some might find it disconcerting that the sun, arguably the best studied of all celestial objects, is so incompletely understood. But I see the lingering questions as a wonderful excuse to share one of the greatest experiences in nature.

As for me, decades ago I was so busy photographing during totality that I barely had time to look up to see it. But now, with computer automation, I can enjoy a few seconds to savor the eclipse while cameras click away and electronic sensors upload their data to computers. I look forward to the view of my 66th solar eclipse from Oregon. Those who are as dazzled as I am can think ahead to the 2019 and 2020 eclipses in Chile and Argentina and the 2024 total solar eclipse that will sweep across the eastern U.S. from Texas to Maine. And a 2023 annular eclipse will show partial phases over North and South America.