At Harvard the “computers” who worked long days and nights calculating and tabulating observational data, were women, hired by Edward Pickering, the longtime director of the Harvard College Observatory, for twenty-five to thirty-five cents per hour. They worked with quill pens and black ink, writing long lists of figures in neat script, without corrections. Pickering had turned to women out of exasperation with a male assistant. Declaring that even his maid could do a better job, he hired Williammina P. Flemming, a twenty-four-year-old Scottish immigrant, to assist him. She stayed for a total of thirty years and before long was in charge of an entire staff of women assistants. Flemming, a divorced mother, supported herself on the meager wages Pickering paid. Other computers were college graduates interested in science. The prevailing social notion of “separate spheres” for men and women left them no room in the observatories and laboratories.
Henrietta Swan Leavitt came to the Harvard College Observatory from the Society for the Intercollegiate Instruction of Women, the forerunner of Radcliffe College, in 1895. Pickering stacked glass photographic plates from Harvard’s Southern Station observatory in Peru in front of her and told her to look for variable stars, stars that cycled in brightness.
It was tedious work. Star by star Leavitt compared glass photographic plates of the same area of sky until she found a speck that was darker or fainter than it had been on a plate taken earlier or later. With enough plates of the same area of sky, and records of when the plates were exposed, she could measure the period of the variable stars—how long it took to cycle from maximum to minimum brightness. After years of laborious measurements Leavitt had cataloged more than 2,400 Cepheid variable stars, named after the constellation Cepheus, where they were first discovered.
Not content just to catalog the data, Leavitt searched for a correlation between the intrinsic brightness of the Cepheid variables and the period of their cycle. In 1908 she tried plotting the logarithm of the period of variable stars in the Small Magellanic Cloud, and hence at the same distance, against their apparent brightness. The data on her graph fell on a straight line: Cepheids a thousand times as luminous as our sun completed their bright-dim-bright cycle in three days; Cepheids ten thousand times as luminous as our sun took thirty days to complete their cycle. By 1912 she had graphed enough data to publish an article, “Periods of Twenty-five Variable Stars in the Small Magellanic Cloud,” in the Harvard College Observatory Circular. After the article appeared Pickering ordered her not to pursue the subject further. His attitude mirrored what was then a widespread viewpoint: A lady of science’s place was in the back room, writing columns of numbers, not in the observatory or the scientific journals.
Another piece of the puzzle fell into place when the Danish astronomer Ejnar Hertzsprung used studies of proper motions (a form of triangulation using the long baseline of the sun’s motion through space over a period of decades) to determine the average magnitude of a typical Cepheid variable in the Milky Way. Independently, Henry Norris Russell determined the absolute magnitudes of thirteen Cepheids in the Milky Way. Shapley was still Russell’s graduate student at the time.
When Shapley saw Hertzsprung’s article in Astronomische Nachrichten, he realized that he might have his cosmic ruler. The Cepheids Leavitt had charted were all in the same star group, the Small Magellanic Cloud, which is visible only in the Southern Hemisphere. Shapley observed that if the stars on her charts were at roughly the same distance from earth, the differences in apparent brightness she had measured must indicate differences in the intrinsic luminosity of the stars. With the reasonable assumption that variable stars everywhere in the universe were the same, he took his extrapolation one step further: If all variables of comparable period, anywhere in the heavens, had the same intrinsic luminosity, then by measuring the apparent brightness of a variable star and comparing it to stars of comparable period on Leavitt’s chart, he could calculate the relative distance of the new star.
The mathematics, and the concept that “faintness means farness,” were invitingly simple. The apparent brightness of an object is inversely proportional to the square of its distance from the observer: When an object is twice as far away, it appears one-fourth as bright. If Shapley found a Cepheid variable of the same period as one of the sample on Henrietta Leavitt’s charts, with an apparent luminosity one-fourth that of her sample star, his star was twice as far away as hers. The question was where to look for these “yardstick” stars. Where could he find Cepheid variables far enough away to help him construct a theoretical model of the heavens?
Before he had gone off to his new job at Mount Wilson, Shapley had taken a trip to Yale, Brown, and Harvard. At Harvard, Solon I. Bailey, interim director of the Harvard College Observatory, told Shapley, “I have been wanting to ask you to do something. We hear that you are going to Mount Wilson. When you get there, why don’t you use the big telescope to make measures of stars in globular clusters?”
It was an intriguing suggestion. Even in a powerful telescope, globular clusters are faint, mysterious objects. On a photographic plate the centers of these clusters of thousands or millions of stars look like a stellar gridlock, as if the great mass of stars at the heart of the cluster were in physical contact with one another. Depending on the equipment used, the clarity of the atmosphere on a given night, and the observer’s mood, clusters sometimes seem as though they can be resolved into agglomerations of thousands of individual stars; other times the image, especially of faint clusters, is too nebulous to appear much more than a luminous blob.
Astronomers had cataloged globular clusters for centuries, wondering what secrets they held. How far away were they? Were they part of our local galaxy, the Milky Way? The difference for Shapley was that he was going to study these clusters with the largest working telescope in the world, the great sixty-inch reflector on Mount Wilson, which had been finished in 1908.
As a junior man Shapley’s principal assignment at Mount Wilson was to assist a more senior astronomer with traditional studies of star colors and magnitudes. On the nights when he was allowed to do his own research, Shapley used the big telescope to search for Cepheid variable stars in globular clusters. The nights were cold, the controls on the telescope were balky, and the long exposures put a premium on the astronomer’s skill of bladder control. Shapley spent so many hours examining plates that he discovered a new asteroid, which he named after his newborn daughter Mildred.
Even on the sixty-inch telescope, the most modern research instrument, astronomy was a physically demanding science. Astronomers work when the objects they need are “up,” the weather is clear, and the seeing is good—three conditions that rarely come together on a balmy summer evening. Metal can fatigue and crack from the cold, lubricants and even the ink used to write notes in the logbook freeze, and human efficiency falls.
The painstaking work paid off as Shapley began to identify Cepheid variable stars in the globular clusters. Of the approximately one hundred globular clusters visible in the Northern Hemisphere, he identified Cepheid variables in a dozen. By comparing the brightness and period of these stars to stars of comparable period on Henrietta Leavitt’s graph, he could extrapolate the relative distances to the globular clusters.
Each step in Shapley’s research required a leap of extrapolation. The variable stars on Henrietta Leavitt’s graphs cycled between bright and dim in a few days; the stars Shapley was studying took weeks to complete their cycle. There was no evidence to indicate that all variable stars he and she measured were not the same sort of star, so Shapley extrapolated the relationship Leavitt had plotted to include his own observations, even though the much longer periods of variation he measured put the intrinsic luminosities of his stars off the end of her chart. Leaps of reason and data are a necessity of astronomy. The paucity of available information forces astronomers to assumptions that might seem outrageously bold in sciences with a surfeit of experimental data.
He derived distances to a dozen globular clusters—an importan
t first. In July of Shapley’s first year at Mount Wilson, he showed his initial results to J. C. Kapteyn, perhaps the best known cosmologist of his day. Shapley’s distances were so enormous, compared to the scale of contemporary models of the universe, that Kapteyn suggested Shapley recheck his observations and calculations.
Shapley persisted. In January 1918 he reported a breakthrough to Arthur Eddington, announcing that the consequences of his studies extended not just to clusters but to the entire galactic system: “With startling suddenness and definiteness, they seem to have elucidated the whole sidereal structure.” Buoyed by his findings, Shapley began publishing his results. The articles followed each other so quickly that by the 1918–19 issues of the Mount Wilson Contributions, fully half the articles were by Shapley.
The results of these first efforts were so satisfying that Shapley went a step further. In each nearby globular cluster, he isolated the most luminous stars, stars that the astronomers identified as red giants and supergiants, and compared their apparent brightness to the brightness of the Cepheid variable stars. When he had a large enough sample to feel confident of his calibration of the absolute magnitude of the giant stars, he began using the giant stars as yardsticks to estimate the distance to faint, distant globular clusters where he could not resolve Cepheid variable stars but could resolve red giant and supergiant stars.
Shapley’s total sample for these extrapolations was only a few stars. But even a small sample was enough to begin measuring the universe, if he could assume that stars of similar spectra*—stars that reflected or absorbed the same colors of light—were in fact similar, whether they were relatively close to us or in some distant corner of the observable universe. Shapley also assumed that the relationship of brightness to period that Henrietta Leavitt had discovered in the Cepheid variable stars she had studied in the Magellanic Clouds would apply equally to variable stars throughout the observable universe. Finally, he assumed that space was essentially empty, that there was no absorption of light from distant objects by interstellar dust or gases. Skeptical critics shook their heads as they tallied up the assumptions, but Shapley’s results were too exciting to ignore.
Eager to have everyone understand his arguments, Shapley assumed little from his audience at the symposium. He did not define a light-year until the seventh page of his nineteen-page script, and he devoted the last three pages to descriptions of an intensifier he had developed to photograph faint stars, a subject that had little bearing on his argument, though it might impress those members of the audience who were associated with the Visiting Committee of the Harvard College Observatory, where he was a candidate for the directorship.
Shapley’s total evidence was meager—he had had only a few years to work sporadically on the big sixty-inch reflector at Mount Wilson, and had only observed for three months on the new one-hundred-inch reflector that had just come into service—but if you accepted his initial assumptions, the subtle arguments cascaded one upon the next with compelling logic, and with the elegant simplicity that so often characterizes good science. In Shapley’s model the globular clusters outline the extent of our galaxy, the Milky Way, and it is a big one. The diameter of Shapley’s universe was 300,000 light-years, or 19 × 1017 miles (19 with 17 zeros after it)—approximately ten times as large as the cosmological models that prevailed when he began his work.
His huge new universe made man seem very small indeed. But the consequences of Shapley’s argument went deeper. In his observations he found that the clusters were not scattered evenly around the observable heavens, as one would expect if the sun were at the center of the universe. Instead, the clusters appeared to be concentrated in the area of the constellation Sagittarius. To Shapley the implications were obvious: “One consequence of accepting the theory that clusters outline the form and extent of the galactic system, is that the sun is found to be very distant from the middle of the galaxy. It appears that we are not far from the center of a large local cluster or cloud, but that cloud is at least 50,000 light-years from the galactic center.”
In other words the center of our galaxy, the Milky Way, was tens of thousands of light years away, in the direction of Sagittarius. Copernicus and then Galileo had demonstrated that the sun, not the earth, was the center of our solar system. Shapley had taken Copernicus one step further to argue that our sun was not the center of the universe, but only a perfectly ordinary, second-class star, somewhere out toward the edge of the galaxy. His new universe made man seem very small indeed. If his evidence and calculations were correct, Shapley had revolutionized astronomy.
Not everyone was convinced he was right.
Heber Curtis was old enough to be Shapley’s father. For his official portraits he posed next to the telescopes at the Lick Observatory in a coat and tie, the usual observing attire in an age when science could still be a relaxed and gentlemanly pursuit. His trimmed mustache and stiff collar seemed appropriate for a man who had studied classical languages as an undergraduate at the University of Michigan and who was still as comfortable reading a Greek or Latin text as the daily newspaper. Curtis was also a classicist in his astronomy, a staunch believer in the tradition of incremental observation. He was a patient and careful observer who had put in his time on balky telescopes on cold nights. He had been trained in the old tradition, his astronomy studies devoted as much to practical optics as to the newer physics of particles and waves.
By temperament Curtis was a skeptic. He became the chief critic of Shapley’s articles, answering them with a vehemence that reflected a reaction to the perceived arrogance of an “upstart” going off half-cocked, and the rivalry between the Lick and Mount Wilson Observatories, as much as intellectual disagreement. To Curtis, Shapley’s theories weren’t necessarily wrong; he awarded the Scottish verdict: Not proved. Curtis had made the long trip to Washington to make sure those theories didn’t get more credit than they deserved.
Following Shapley to the podium, Curtis was in the position of a reviewer. He had no new cosmology of his own to present. His job was to challenge Shapley’s thesis by questioning the logic and the evidence, to convince the audience that while the arguments might be intriguing, the evidence was too thin, the hypothesis stretched too far. Curtis argued that Shapley’s sample of only eleven stars was too small to determine the average brightness of stars in clusters, that the dispersion of magnitudes of the stars in the sample made any calculation of an average from this data suspect, and that the techniques Shapley had used to “smooth” his data before plotting it were also suspect. Curtis’s own plot of Shapley’s data came out not a smooth curve but an essentially meaningless scatter plot. “It would seem,” he said, “that available observational data lend little support to the fact of a period-luminosity relation among galactic Cepheids.”
Curtis also attacked Shapley’s effort to use other stars as distance guides. Citing his own studies, Curtis showed that the average magnitude of stars in the neighborhood of the sun is less than the brightness of the sun. Since there is no evidence that giant stars predominate in the clusters, he argued, if we accept the proposition of uniformity throughout the universe, then the average stars in clusters must also be dwarfs, smaller than the sun. And if the stars in Shapley’s sample were dwarfs rather than giants, they could not be as far away as Shapley calculated. By Curtis’s reasoning the distance to the clusters Shapley had observed, and the diameter of our galaxy, were about one-tenth those determined by Shapley, or the same modest dimensions that had prevailed among astronomers before young Shapley came along.
Curtis was a deft speaker. He made his presentation without notes, after Shapley had read his long report. But he was in the unenviable position of having to debunk exciting, potentially path breaking work, and he was speaking not to astronomers who might share his respect for details or his skepticism about Shapley’s data, but to generalists. Curtis’s dry arguments weren’t the pinprick he needed to burst Shapley’s balloon.
As Curtis reached the conclusion of his re
marks, it seemed almost as though he were abandoning the subject of the debate. He turned to his own studies of the spiral nebulae, mysterious wispy structures, like pinwheels in the sky, the brightest of which are barely visible through binoculars or a small telescope. Little was known about the spirals. They looked so unlike any other class of celestial object that some believed they were entire galaxies, separate “island universes,” comparable in scale to our own Milky Way. In proposing the symposium for the annual meeting of the National Academy, George Hale had originally suggested “island universes” as a topic.
Less than a century before, Lord Rosse in Ireland had used a huge telescope with a seventy-two-inch mirror to study and sketch spiral nebulae. The size of the telescope, and the long hours Lord Rosse put into his observations, had given credence to his claim that the spiral nebulae could be resolved into individual stars. If there were individual stars in these spirals, they might be “island universes,” separate from but perhaps equal in scale to the Milky Way. The existence of other galaxies outside the realm of our own but as large as the Milky Way, would raise havoc with Shapley’s scale for the galaxy.
In 1898 James E. Keeler had begun a survey of spiral nebulae at the Lick Observatory. Although his telescope, the Crossley reflector, had a mirror just half the diameter of the one Lord Rosse had used, the carefully figured glass mirror provided higher resolution than the speculum (polished metal) mirror in Lord Rosse’s telescope. Keeler also had the advantage of using photographic plates to record images of the spirals. A photographic emulsion can accumulate faint light over a long exposure, building up an image over a period of hours from an object that might be invisible or barely visible to the human eye. A few of the nebulae Keeler photographed, like M31 in Andromeda and M33 in Triangulum, appeared incredibly large in plates taken on the Crossley. Unless they were fundamentally different from all other observable spirals—and there was no reason to believe they were—the tiny apparent size of the thousands of other spirals observed with the reflector suggested that they must be at great distances.
The Perfect Machine Page 3