Her discovery of the relation of period to brightness is destined to be one of the most significant results of stellar astronomy, I believe. I am quite anxious to have her opinion as to the periods because of its bearing on some statistical work I am now bringing to a close.
—Harlow Shapley, writing to Edward Pickering
about Henrietta Leavitt
Back in the eighteenth century, William Herschel had theorized that nebulae, like Andromeda, might be distant galaxies. The philosopher Immanuel Kant, called them “island universes,” arguing, “It is much more natural and reasonable to assume that a nebula is not a unique and solitary sun, but a system of numerous suns.” The universe, he wrote, may be filled with Milky Ways.
Other astronomers, however, were swayed by a different philosopher, Pierre-Simon Laplace, who proposed that the spiral-shaped clouds like Andromeda were no more than “proto solar systems”—a new sun and its orbiting planets in the process of congealing from a whirling cloud of gas. The theory seemed all the more plausible when in 1885 a new star, or “nova,” flared within the center of the hazy disk of Andromeda, as though a solar system was being born.
Time exposures of the sky soon revealed some 100,000 of the luminous swirls and more were being found all the time. With spiral nebulae appearing everywhere, it seemed absurd to suppose that each could be a galaxy filled with millions of stars. “No competent thinker, with the whole of the available evidence before him, can now, it is safe to say, maintain any single nebula to be a star system of coordinate rank with the Milky Way,” Agnes Clerke, an astronomer and historian of science, wrote in 1890. “A practical certainty has been attained that the entire contents, stellar and nebular, of the sphere belong to one mighty aggregation.” The universe was just another name for the Milky Way.
But the issue was far from settled. For one thing, the Milky Way itself seemed to be shaped like a spiral. Viewed from afar it might appear little different than Andromeda or any other nebula. An even stronger argument for island universes came from analyzing a nebula’s light with a prism, breaking it into its component colors. These spectroscopic patterns were believed to reveal which chemicals a celestial object was made from. Andromeda’s rainbow looked very much like the one cast by the sun. Both seemed to consist of star stuff. The tests were less than conclusive. Other nebulae produced dull, simple spectra—what one might expect from a homogeneous cloud of luminous gas. People tended to find in the data what they were already disposed to believe.
The impasse was broken in 1914 by Vesto Melvin Slipher at Lowell Observatory in northern Arizona, who had figured out how to estimate the speed at which a nebula was traveling through space. His technique was based on the Doppler effect. If a star is moving toward you, its light waves will be compressed. Thus the frequency—the number of waves that strike the eye every second—will increase. Brains interpret frequency as color, and so the starlight will shift toward the higher, bluer end of the spectrum. Conversely, if the star is moving away, its light will be stretched toward the lower-frequency reds.
Measuring the red and blue shifts of fifteen spiral nebulae, Slipher found them to be traveling at incredible velocities. Two of them appeared to be flying off at a dizzying 1,100 kilometers per second. That hardly seemed possible if they were simply small objects within the gravitational grip of the Milky Way. For many astronomers that settled the matter, in favor of island universes. (Slipher himself initially clung to the prevailing notion that a nebula was but a single star “enveloped and beclouded by fragmentary and disintegrated matter.”)
More evidence arrived three years later when another nova suddenly appeared inside a spiral nebula called NGC 6946 (after its designation in the New General Catalog of Nebulae and Star Clusters). Heber Curtis of Lick Observatory, a California stronghold of the island universe theory, found novae inside other nebulae, and when astronomers reexamined old photographic plates they found still more.
Curtis believed the novae might serve as standard candles. Astronomers had estimated that those in the Milky Way surged to an intrinsic magnitude of about –8. (Remember that the lower the magnitude, the brighter the star, making one with a negative value very bright indeed.) Curtis assumed that the novae in the distant nebulae were probably peaking at about the same intensity. Compare that number with how bright a nova appeared from Earth and you could use the inverse square law to get its distance. Measured this way, the nebulae appeared to be huge spinning galaxies many millions of light-years away.
By 1917 the consensus had shifted toward island universes. In addition to Curtis, and by now Slipher, supporters included such prominent astronomers as Arthur Eddington, James Jeans, Ejnar Hertzsprung, and a young researcher named Harlow Shapley, who had recently moved to Mount Wilson Observatory, the astronomical powerhouse perched high above Pasadena, California. Shapley however was about to change his mind. Using Leavitt’s variable stars, he would spend the next few years calibrating the Cepheid yardstick and measuring the size and shape of the Milky Way. He was ultimately forced to conclude that it was far larger than anyone had dared imagine—so large, he believed, that it must constitute the entire universe, nebulae and all.
2
Shapley had a peculiar fixation with ants. When he wasn’t looking upward at the stars, he liked to watch a colony of medium-size brownish black ants—Liometopum apiculatum—as they streamed along a concrete wall by Mount Wilson’s maintenance shop. Shapley noticed that the ants slowed down when they reached the shade of some manzanita bushes and sped up again in the sun. Armed with various instruments, he studied the ants under different atmospheric conditions, even watching them with a flashlight at night. The correlation between running speed and temperature was so tight that he could use the ants as a thermometer, reading off the temperature within one degree. Returning to Mount Wilson thirty years later, he was annoyed to find that an assistant engineer was making a practice of burning off the ant trail with a blow torch—“genocide,” Shapley called it. (“Formicide” would have been a better word.) But the resilient ants always came back.
He drew lessons from these tiny creatures. Asked to deliver a commencement address at the University of Pennsylvania, he chose the topic “On Running in Trails,” warning the students against the comfortable allure of conformity, of following the same narrow paths as their ancestors, afraid to break from the pack.
When Shapley arrived in Pasadena in 1914, the common wisdom held that the Milky Way was a lens-shaped disk some 25,000 light-years long and about a fourth that wide, with the sun at almost dead center. This picture of the heavens was sometimes called the Kapteyn universe, after the Dutch astronomer Jacobus Kapteyn, who had estimated its size. The methods he had used were far from exact. With the Cepheid variables and Mount Wilson’s powerful 60-inch telescope at his command, Shapley decided to measure the galaxy for himself.
Spread throughout the Milky Way were a hundred or so “globular clusters,” each consisting of hundreds of thousands, even millions, of stars. Shapley suspected that these huge concentrations formed a kind of framework or “skeleton” marking the extent and shape of the galaxy. By using Cepheids to determine the distances to these mileposts, he could map out the whole thing.
The Milky Way
Shapley figured he knew something about variable stars. His Ph.D. dissertation at Princeton, under Henry Norris Russell, had focused on a type of variable called eclipsing binaries— two stars orbiting around a common point and periodically blocking each other’s light. One of Shapley’s first papers at Mount Wilson showed that the Cepheids did not belong to this class. Rather, they were single stars that expanded and contracted with a regular beat. For now, however, these details were unimportant. He knew from Henrietta Leavitt’s research that Cepheids would serve as standard candles.
He also knew that most of the variables in the globular clusters were a little different from the ones she had discovered in the Magellanic Clouds. Shapley’s stars—called cluster variables—blinked much faster, with cycles measured in hours, not days. Hertzsprung, in fact, thought his eager colleague was mixing apples and oranges. How could he be so sure that both kinds of stars showed the same connection between period and brightness?
Shapley was insistent. “[T]his proposition scarcely needs proof,” he wrote in a paper in the Astrophysical Journal. “Practically all writers on the subject are more or less inclined to accept this view.”
Undeterred, he proceeded with his plan. For his early measurements, he relied on the fact that the farther something is from an observer, the more slowly it appears to move—think of a tiny jet plane inching across a windowpane. The speed at which a star is heading directly toward or away from Earth, its “radial velocity,” can be clocked using redshift and blueshift. But it is the “transverse velocity”—how fast the star is moving across the sky—that hints at how far away it is. It seemed sensible that, on average, stars in a cluster would move at the same velocity, whatever the direction. Drawing on a method called statistical parallax, Shapley used Doppler shifts to estimate the average velocity of a sampling of stars and compared that figure with how fast the stars appeared to be moving. That revealed their distance. Eleven of these were Cepheids, forming the anchor of what came to be called “Shapley’s curve.”
Taking a second leap, he extended the curve to include the far more common shorter-period variables. First he would find a cluster that had both types. The slow-paced Cepheids gave him an estimate of the cluster’s distance, which he could then correlate with the period of the faster variables. Now the distance of clusters with only fast variables could be measured— provided that both kinds of stars really obeyed the same law.
New yardstick in hand, he gauged the distances to several of the nearest globular clusters. Then he ran into a wall. In most of the clusters, not a single blinking star could be found. He would have to extrapolate further, and that meant finding another kind of standard candle. It seemed sensible, he reasoned, that each cluster would consist of stars spanning the same range of magnitude. The brightest stars in cluster A, whose distance had been measured with his yardstick, would be about as intense as the brightest stars in cluster B, whose distance was unknown. If they appeared dimmer, it would be because they were farther. The inverse square law would reveal by just how much, extending the map a little more.
Many clusters, however, were so distant and so blurry that not even Mount Wilson’s telescope could pick out a single star. And so came the final leap: one could take a cluster whose distance had been established by these other indirect methods and use the whole thing as a standard candle. The farthest clusters, Shapley reasoned, were probably as intrinsically big and brilliant as the nearest ones. By measuring how much smaller and fainter they appeared, he could judge their distances, and reach to the farthest edges of the galaxy.
As he followed this artful chain of assumptions, Henrietta Leavitt was living alone in a Cambridge rooming house, where she had moved after the death of her uncle Erasmus in 1916. She was still working for the observatory and occasionally Shapley wrote to Edward Pickering inquiring about the latest developments with her variable stars.
Shapley had noticed some very faint variables in the Magellanic Clouds and wondered if these might be similar to the ones he was using to plot out the Milky Way. “Does Miss Leavitt know if they have shorter periods, that is, are their periods shorter than one day, similar to cluster variables?” he wrote on August 27, 1917. “It may be her work has not progressed far enough to give a definite answer.” He was hoping for some ammunition against those, like Hertzsprung, who continued to argue that the faster variables did not necessarily obey Leavitt’s rule. He considered the matter “of much importance. … In fact, the Magellanic clouds and their variables seem to me one of the most important outstanding problems of stellar photometry.”
Pickering replied about three weeks later: “Miss Leavitt is now absent on her vacation.” (She was on Nantucket, visiting with Margaret Harwood, a fellow Harvard computer and astronomical assistant who had become director of the Maria Mitchell Observatory.) “When she returns, she will investigate the matter of the Magellanic Clouds.”
Of the two men, Shapley was the quicker and more loquacious correspondent. Within a week he had fired off another letter, praising Leavitt’s work and emphasizing how much he needed the information. “Her discovery of the relation of period to brightness is destined to be one of the most significant results of stellar astronomy, I believe. I am quite anxious to have her opinion as to the periods because of its bearing on some statistical work I am now bringing to a close.”
Nine months later Shapley was still waiting. On July 20, 1918, he checked in again, still heaping on the praise:
I believe the most important photometric work that can be done on Cepheid variables at the present time is a study of the Harvard plates of the Magellanic clouds. Probably Miss Leavitt’s many other problems have interrupted and delayed her work on the variables of the clouds for the interval of six or seven years since her preliminary work was published.… The theory of stellar variation, the laws of stellar luminosities, the arrangement of objects throughout the whole galactic system, the structure of the clouds—all these problems will benefit directly or indirectly from a further knowledge of the Cepheid variables.
It took almost three weeks for Pickering to reply: “A few days ago I talked with Miss Leavitt.… She has the material for about a third of the brighter variables, and photographs are now being taken with the Bruce 24-inch, which I hope will provide the remainder.”
That is the last letter from Pickering in Shapley’s files. Less than five months later, he died of pneumonia at age seventy-two.
3
Shapley ultimately decided to run with his theory, using the Cepheids as the first step in his hopscotch across the galaxy. The results were astonishing. First of all the Milky Way, by Shapley’s measure, was gargantuan in size—300,000 light-years across. That was some ten times greater than Kapteyn’s estimate—so much larger that he felt he must abandon the notion of island universes. If one insisted on maintaining that the thousands upon thousands of spiral nebulae were galaxies each the size of the Milky Way, then Andromeda, judging from its apparent size, would have to lie at an enormous distance. That, in turn, would mean that its novae were absurdly bright. It must be a small gas cloud after all.
To Shapley there was now an even more damning argument against island universes. One of his colleagues, a Mount Wilson astronomer named Adriaan van Maanen, had recently announced that several of the great spirals, including the aptly named Whirlpool and Pinwheel nebulae, were gradually turning. Van Maanen made the measurements with an arrangement of lenses and mirrors called a blink comparator. With the device, an astronomer could mount two photographic plates taken months or years apart and, gazing through a binocular eyepiece, switch back and forth between them. Anything that had changed would appear to move or vary in size. Comparing plates of nebulae taken five years apart, van Maanen thought he saw a slight rotation.
Viewed from Earth the spin was minuscule—2/100 of a single second of arc each year. A complete cycle would take about 100,000 years. The surprise was that the movement was visible at all. Few could doubt that spirals spun. Why else would they have their pinwheel shapes? But for the motion to be perceptible at all, these nebulae would have to be small and nearby. If the spirals were truly distant galaxies, van Maanen’s data would mean they were spinning at impossibly high velocities—faster than the speed of light.
Just as unsettling as the Milky Way’s enormous size was Shapley’s conclusion about where we lived inside the galactic disk. Astronomers had noticed that the globular clusters are not distributed evenly through the sky but congregate in the direction of the constellation Sagittarius. There, according to Shapley’s measurements, they formed a roughly spherical shape, a cluster of clusters. That, he proposed, must be the center of the galaxy, the central bulge of the Milky Way. If we lived within this region, we would see the clusters spaced uniformly around us. The fact that we do not is because we lie in the galaxy’s outskirts, tens of thousands of light-years from the core. We were not New York City but Pensacola or North Platte.
“So the center has shifted: egocentric, lococentric, geocentric, heliocentric,” Shapley wrote to George Ellery Hale, Mount Wilson’s director. Or, as he later put it: “Man is not such a big chicken. If man had been found in the center, it would look sort of natural. We could say, ‘Naturally we are in the center because we are God’s children.’ But here was an indication that we were perhaps incidental. We did not amount to so much.” People were no more important than ants.
The center of the galaxy was in Sagittarius. And so the center of the universe must be there as well.