Cosmic Yardsticks, Spiral Nebulas, and Island Universes: The Early History of Measuring the Universe
To talk about stars, the ancients needed a way to qualify how bright they were. With this in mind, the Greeks developed the magnitude scale. Initially, their version implemented 6 levels with each subsequent level being 2.5 times brighter. 1 was considered the brightest star in the sky and 6 the dimmest. However, modern refinements to this system now means that the difference between levels is more like 2.512 times brighter. Additionally, the Greeks were unable to see every star out there and so we have stars that are brighter than magnitude 1 (and even go into the negative range) plus we have stars that are way dimmer than 6. But for the time, the magnitude scale brought order and a standard to star measurements (Johnson 14).
And so the decades, centuries, and millennia passed by with further and further refinements as better instruments (like telescopes) came into being. Many observatories sole operation was the cataloging of the night sky, and for that we needed position in terms of right ascension and declination as well as the color and magnitude of the star. It was with these tasks at hand that Edward Charles Pickering, the director at Harvard Observatory, set out in the late 1870’s to record every star in the night sky. He knew that many had recorded the place and motion of the stars but Pickering wanted to take star data to the next level by finding their distances, brightness, and chemical make-up. He didn’t care so much as for finding out any new science so much as he wanted to give others the best chance by compiling the best data available (15-6).
Now, how does one get a good fix on the magnitude of a star? Not easily, as we will come to find that difference in technique yield substantially different results. To add to the confusion is the human element that was present here. One might simply make a comparison mistake, for no software existed at the time to get a good read. That being said, tools did exist to try and level the playing field as much as possible. One such instrument was the Zollmer astrophotometer, which compared the brightness of a star to a kerosene lamp by shining a pinpoint amount of light via a mirror from the lamp onto a background in close proximity to the star being viewed. By adjusting the size of the pinhole, could get close to a math and then record that result (16).
This was not good enough for Pickering, for the aforementioned reasons. He wanted to use something universal, like a well-known star. He decided that instead of using a lamp, why not compare against the North Star, which at that time was recorded at magnitude 2.1. Not only is it faster but it removes the variable of inconsistent lamps. Also of consideration was the low-magnitude stars. They don’t emit as much light and take longer to see, so Pickering chose to us photographic plates to have a long exposure in which the star in question could then be compared (16-7).
But at the time, not every observatory had said equipment. Plus, one needed to be as high up as possible to remove atmospheric disturbances and back glow of outdoor lights. So Pickering had the Bruce Telescope, a 24 inch refractor sent in Peru to grab him plates to examine. He labeled the new location Mt. Harvard and had it begin immediately but problems arose right away. For starters, Pickering’s brother was left in charge but mismanaged the observatory. Instead of looking at stars, the brother gazed upon Mars, claiming to have seen lakes and mountains in his report to the New York Herald. Pickering sent his friend Bailey to clean up and get the project back on track. And soon enough, plates began pouring out. But how would they be analyzed? (17-8)
As it turns out, the size of a star on a photographic plate is related to the brightness of the star. And the correlation is as you expect, with a brighter star being larger and vice versa. Why? Because all that light just keeps getting absorbed by the plate as the exposure continues on. It is through the comparison of those dots the stars make on the plates to how a known star does in similar circumstances that the magnitude of the unknown star can be determined (28-9).
Naturally, Humans Are Computers Too
Back in the 19th century, a computer would have been someone Pickering would use to catalogue and find stars on his photographic plates. But this was considered a boring job and so most men did not apply for it, and with a 25 cents an hour minimum wage translating to $10.50 a week, the prospects were not appealing. So it should be no surprise that the only option available to Pickering was to hire women, who in that time period were willing to take any work they could get. Once the plate was backlit by reflected sunlight, the computers were tasked with logging each star in the plate and record the position, spectra, and magnitude. This was the job of Henrietta Leavitt, whose later efforts would help spark a revolution in cosmology (Johnson 18-9, Geiling).
She volunteer for the position in the hopes of learning some astronomy but this would prove to be difficult as she was deaf. However, this was seen as an advantage for a computer because it meant her eyesight was likely heightened to compensate. Therefore, she was seen as abnormally talented for such a position and Pickering brought her onboard right away, eventually hiring her full-time (Johnson 25).
Upon beginning her work, Pickering asked her to keep an eye out for variable stars, for their behavior was strange and was deemed to be worth a distinction. These strange stars, called variable, have a brightness that increases and decreases over a span as short as a few days but as long as months. By comparing photographic plates over a time span, computers would use a negative and overlap the plates to see the changes and notate the star as a variable for further follow-up. Initially, astronomers wondered if they might be binaries but the temperature would fluctuate as well, something that a set pair of stars should not do over such a sort span of time. But Leavitt was told to not be concerned about the theory but to just log a variable star when seen (29-30).
In spring 1904, Leavitt started to look at plates taken of the Small Magellanic Cloud, what was then regarded as a nebula-like feature. Sure enough, when she started to compare plates of the same region taken over different spans of time variables as dim as 15th magnitude were spotted. She would publish the list of 1777 variable she uncovered there from 1893 to 1906 in the Annals of the Astronomical Observatory of Harvard College over a span of 21 pages in 1908. Quite the feat. And as a brief footnote at the end of the paper, she mentioned that 16 of the variables stars known as Cepheid’s showed an interesting pattern: those brighter variables had a longer period (Johnson 36-8, Fernie 707-8, Clark 170-2).
This was so huge, because if you could use triangulation to find the distance to one of these variables and note the brightness then, by comparing the difference in brightness to a different star can lead to a calculation for its distance. That is because the inverse-square law applies to light beams, so if you go twice as far away the object seems four times dimmer. Clearly, more data was needed to show if the pattern of brightness and period held at all and a Cepheid needed to be close enough for triangulation to work, but Leavitt had a host of problems plague her after her paper was published. She got sick and once she recovered from that her father dies so she went home to help her mother. It wouldn’t be until the early 1910s that she would begin to look at more plates (Johnson 38-42).
Once she did, she began to plot them on a graph that examined the relationship between brightness and period. With the 25 stars she examined, she published another paper but under Pickering’s name in the Harvard Circular. Upon examining the graph one sees a very nice trendline and sure enough as brightness increased, the slower the blinking occurred. As for why, she (and for the matter no one) had a clue, but that didn’t deter people from using the relation. Distance measurements were about to enter a new playing field with the Cepheid Yardstick, as the relation became known (Johnson 43-4, Fernie 707)..
Now, parallax and similar techniques only got you so far with Cepheids. Using the diameter of the Earth’s orbit as a baseline meant that we could only get a grasp on some Cepheid’s with any degree of reasonable accuracy. With only Cepheid’s in the Small Magellan Cloud, the Yardstick only gave us a way to talk about how many distances away a star was in terms of the distance to the Cloud. But what if we had a bigger baseline? As it turns out, we can get that because we move with the Sun as it moves around the solar system and scientists notices over the years that stars seem to spread out in one direction and come closer together in another. This indicates movement in a certain direction, in our case away from the constellation Columbia and towards constellation Hercules. If we record the position of a star over the years and note it, we can use the time between observations and the fact that we move through the Milky Way at 12 miles a second to get a huge baseline (Johnson 53-4).
The first one to make use of this baseline technique along with the Yardstick was Ejnar Hertzspring, who found the Cloud to be 30,000 light-years away. Using just the baseline technique, Henry Morris Russel arrived at a value of 80,000 light-years. As we will see shortly, both would be a big problem. Henrietta wanted to try her own calculations but Pickering was determined to stick to the data collecting and so she continued on. In 1916, after years of data collecting, she publishes a 184 page report in the Annals of the Astronomical Observatory of Harvard College in Volume 71, Number 3. It was a result of 299 plates from 13 different telescopes cross referenced and she hoped it would improve her Yardstick’s capabilities (55-7)
Those Island Universes In The Sky
With the distance to one far away object was found, it sparked a related question: just how big is the Milky Way? At the time of Leavitt’s work, the Milky Way was held to be the entire Universe with all those thousands of blurry patches in the sky to be nebulas called island universes by Immanuel Kant. But others felt differently, such as Pierre-Simon Laplace, who considered them to be proto solar systems. No one felt they could contain stars because of the condensed nature of the object as well as the lack of resolving one inside it. But by looking as the spread of stars in the sky and the distances to the known ones plotted, the Milky way seemed to have a spiral shape to it. And when spectrographs were pointed at island universes, some had spectra similar to the Sun but not all of them did. With so much data conflicting with each interpretation, scientists hoped that by finding the size of the Milky Way we could accurately determine the feasibility of each model (59-60).
Which is why the distance to the Cloud was such a problem as well as the shape of the Milky Way. You see, at the time the Milky Way was considered to be 25,000 light-years based off the Kapteyn Universe model, which also said the Universe was a lens-shaped object. As we mentioned earlier, scientists had just found the shape of the galaxy to be a spiral and that the Cloud was 30,000 light years away and therefore outside the Universe. But Shapley felt he could resolve these problems if better data came about, so where else would one look for more star data than a globular cluster? (62-3)
He also happened to choose them because it was felt at the time that they were at the boundaries of the Milky Way and therefore a good gauge as to the boundary of it. By looking for Cehpeids in the cluster, Shapley hoped to use the Yardstick and get a reading on the distance. But the variables he observed were unlike Cepheid’s: they had a period of variability that lasted only hours, not days. If the behavior is different, can the Yardstick hold? Shapley thought so, though he decided to test this out by using another distance tool. He looked at how fast the stars in the cluster were moving towards/away from us (called the radial velocity) using the Doppler Effect (more on that later), then compared that value to how fast they appeared to move towards/away from us. This gave him a means of finding their distance from us by plotting the differences on a graph, he argued, along with the Cepheid variable data. Combining and interpreting all of this gave scientists the Shapley Curve, which was then extended toward the lower-period variables in an effort to use the Yardstick on them (63-5).
Hopefully we can see some problems with this methodology. It makes plenty assumptions about proper extensions of theory and mathematical analysis, especially without any means of verifying that any of them are sound. But trust me, it gets worst from here. You see, Shapley needed a cluster with both types of variable stars for a proper calibration but finding a cluster with both was challenging. On top of that, many proved to have stars too far away to properly resolve so instead Shapley took the light output of the cluster itself as the representative light of a star and then compared that cluster’s light to others for a distance calculation (65-6).
Using this Curve, the Milky Way had a 300,000 light-year diameter. If true, then the island universes would be an incredible distance away and the light from supernova would have to be way more powerful than what theory predicted. Shapley knew that was not reasonable and thus needed more information to make a reasonable Curve, and where better to go for that than Pickering and his computers. Starting on August 27, 1917 the two wrote back and forth but sparingly until Pickering’s death in 1918. Shapley still waited (66-8).
During this time frame, scientists discovered a new way to relate a stars temperature to its size and luminosity. Dubbed the H-R diagram, it developed new patterns for astronomers to use in their study of the sky. Once such relation was that B-type stars were on the average about 200 times brighter than our Sun. This means a new standard candle was on the scene, and Shapley took advantage of them to help resolve his Curve. After finding B stars in the Hercules cluster during the spring of 1920, he was able to calculate a distance of 35,000 light years to the cluster. Via extension, Shapley found a distance of 300,000 light-years for the size of the Milky Way. Again (Johnson 74-5, Clark 172).
So what about the findings of Adriann van Maanen at Mount Milson? While looking at the Whirlpool and Pinwheel nebulas, he compared plates of them taken months apart and found that they seemed to spin at a rate of 1/50 of a second of arc a year, meaning they completed a spin once every 100,000 years. If the spin could be seen, then the object couldn’t be far away for otherwise the spin would be immeasurable. But if it was as far as it seemed to be, then the object was rotating faster than the speed of light, a clear impossibility (Johnson 68).
And so Shapley offered to those who supported the Cepheid variable findings that those stars themselves may follow a different pattern than found, not taking into consideration that he himself might be doing the same thing. Heber Curtis, a Cepheid star follower, offered this for a counter and pointed out the rotating spirals needed a smaller Milky Way but again Shapley felt not enough was known about them to justify their use as evidence. Again, Curtis rebukes the use of the stars in Shapley’s Curve as well as the extension of the B stars from the H-R diagram (75-7).
Besides, wouldn’t it make more sense to use Sun-like stars, Curtis argued? After all, most of the stars one observes in the night sky are similar to ours. Using them as his standard candle gave Curtis a diameter of 30,000 light-years. That meant if Shapley was right about the cluster distances, they would have to be crazy bright to be seen from so far away, violating known physics (77).
Curtis also pointed out how the spiral nebula seemed to be distributed towards the galactic pols and yet were lacking in the galactic plane. Why would there be a bias in distribution? Shouldn’t they be spread out rather uniformingly? Yes, he argues, but the stars in the Milky Way block out the light from them because they exist outside our galaxy (78).
This little debate between Shapley and Curtis would be formally represented in separate works published in the Bulletin of the National Research Council in 1921. It is pretty amazing that the same data led to two different interpretations, but clearly only one could be right. Shapley needed evidence to deal a deadly blow to Curtis’ extra-galactic interpretation of the spiral nebula. And so he turned to an unlikely source for some information: Edwin Hubble (79).
Why is this bizarre? For one, Hubble felt Shapley was way off base in his extensions as well as inconsistent. While Hubble was an island universe fan and did not buy the spiral nebula theory, he had to go to Mt. Wilson to make use of its 100 inch telescope. And so in December of 1919 he goes there to collect data (84).
Meanwhile, Curtis had moved on to become the director of the Allegheny Observatory and so was replaced by Knut Lundmark who was also an island universe follower and continued Curtis’ research. While looking at M33, Lundmark claims to have spotted individual stars, which when look at with a spectroscope and interpreted with the H-R diagram led Lundmark to conclude M33 was millions of light-years away. Surely, this is too difficult to reconcile with the spiral nebulas but Shapley was ready with a counter. He simply ignored the data and asked why no one had addressed the spiral spin rate measured before (92).
But Lundmark was willing to play ball, and so he questions the legitimacy of the rotation measurements a few months later. And he had support, for James Jean, a British astronomer, found the data did indeed violate the laws of physics and was worthless. Shapley contended that Newton’s gravity theory simply needed to be modified and then the data was good. As a brief aside, these gentlemen were hinting at the existence of dark matter and MOND theory, but that is for another article, located in my profile page. Lundmark countered that the measurements were so small that error would have been easy to propagate. In fact, when he went to measure the rotation rates, none were to be found (93).
And so with this back and forth about the spirals and M33, Hubble decided to train the telescope at Mt. Wilson on it on October 4, 1923. Once he examines the plates he finds 3 novae on them and compares the plates with those Shapley had taken in the past to make sure they were indeed as they appeared to be. One was a definite nova but another was a Cepheid variable and so was labeled “VAR!” on the plate. And in February 1924, Hubble informs Shapley that according to the Yardstick, M33 (known at the time as the Triangulum Nebula) was a million light-years away (94).
Seemingly against all reason, Shapley had a counter which is completely ridiculous. He claimed that Hubble’s data was unreliable and that Cepheid’s with a period of one month or more are unreliable as standard candles. Hubble obviously puts much stock in what Shapley said, for he moved right on to studying NGC 6822 aka Barnard’s Cloud. From 1923 to 1924 Hubble was able to procure 50 plates of the Cloud and using a blink comparator was able to find 15 Cepheids. Once again using the Yardstick, Hubble found the Cloud to be 700,000 light-years away. Hubble also examined older plates of Andromeda and M33 and was able to find more Cepheid’s which confirmed Hubble’s findings. With these in mind, Hubble again writes to Shapley about them (95-6).
With the overwhelming evidence stacked against him, Shapley finally concedes, saying, “I do not know whether I am sorry or glad. Perhaps both.” With Shapley now in defeat, most astronomers begin to accept the island universe theory, which slowly began to be called galaxies because of their Milky Way-like characteristics. Yet Hubble would wait until May of 1925 to publish “Cepheids in Spiral Nebulae” because he wanted to be sure about the rotation rate problem by examining the spirals and being 100 percent sure that no rate was measurable (96-7). It was well worth it, as we now have a large body of evidence to support this theory.
Clark, Stewart. The Unknown Universe. Pegasus Books, New York: 2016. Print. 170-9.
Fernie, J.D. “The Period-Luminosity Relation: A Historical Review.” Publications of the astronomical Society of the Pacific Vol. 81, No. 483: 1969. Print. 707-8.
Geiling, Natasha. “The Women Who Mapped the Universe and Still Couldn’t Get any Respect.” Smithsonian.com. Smithsonian 18 Sept. 2013. Web. 08 Feb. 2017.
Johnson, George. Miss Leavitt’s Stars. Atlas Books, Norton Publishing Co., 2005. Print. 14-9, 25, 28-30, 36-44, 53-7, 59-68, 74-9, 84, 92-7.
Questions & Answers
© 2017 Leonard Kelley