An unfortunate misidentification hampered the early recognition of the northern sky’s brightest nearby galaxy, the Andromeda Nebula, also known as M31. In 1885 a bright star, previously invisible, appeared near the centre of M31, becoming almost bright enough to be seen without a telescope. As it slowly faded again, astronomers decided that it must be a nova, a “new star,” similar to the class of temporary stars found relatively frequently in populous parts of the Milky Way Galaxy. If this was the case, it was argued, then its extraordinary brightness must indicate that M31 cannot be very far away, certainly not outside the local system of stars. Designated S Andromeda in conformity with the pattern of terminology applied to stars of variable brightness, this supposed nova was a strong argument in favour of the hypothesis that nebulae are nearby objects in the Milky Way Galaxy.

By 1910, however, there was evidence that S Andromeda might have been wrongly identified. Deep photographs were being taken of M31 with the Mount Wilson Observatory’s newly completed 152-cm (60-inch) telescope, and the astronomers at the observatory, especially J.C. Duncan and George W. Ritchey, were finding faint objects, just resolved by the longest exposures, that also seemed to behave like novae. These objects, however, were about 10,000 times fainter than S Andromeda. If they were ordinary novae, then M31 must be millions of light-years away, but then the nature of S Andromeda became a difficult question. At this vast distance its total luminosity would have to be immense—an incomprehensible output of energy for a single star.

Completion of the 254-cm (100-inch) telescope on Mount Wilson in 1917 resulted in a new series of photographs that captured even fainter objects. More novae were found in M31, mainly by Milton L. Humason, who was an assistant at the time to Edwin P. Hubble, one of the truly outstanding astronomers of the day. Hubble eventually studied 63 of these stars, and his findings proved to be one of the final solutions to the controversy (see below The distance to the Andromeda Nebula).

The scale of the Milky Way Galaxy

At the same time that spiral nebulae were being studied and debated, the Milky Way Galaxy became the subject of contentious discussion. During the early years of the 20th century, most astronomers believed that the Milky Way Galaxy was a disk-shaped system of stars with the Sun near the centre and with the edge along a thick axis only about 15,000 light-years away. This view was based on statistical evidence involving star counts and the spatial distribution of a variety of cosmic objects—open star clusters, variable stars, binary systems, and clouds of interstellar gas. All these objects seemed to thin out at distances of several thousand light-years.

This conception of the Milky Way Galaxy was challenged by Shapley in 1917, when he released the findings of his study of globular clusters. He had found that these spherically symmetrical groups of densely packed stars, as compared with the much closer open clusters, were unusual in their distribution. While the known open clusters are concentrated heavily in the bright belt of the Milky Way Galaxy, the globular clusters are for the most part absent from those areas except in the general direction of the constellation Sagittarius, where there is a concentration of faint globular clusters. Shapley’s plot of the spatial distribution of these stellar groupings clarified this peculiar fact: the centre of the globular cluster system—a huge almost spherical cloud of clusters—lies in that direction, some 30,000 light-years from the Sun. Shapley assumed that this centre must also be the centre of the Milky Way Galaxy. The globular clusters, he argued, form a giant skeleton around the disk of the Milky Way Galaxy, and the system is thus immensely larger than was previously thought, its total extent measuring nearly 100,000 light-years.

Shapley succeeded in making the first reliable determination of the size of the Milky Way Galaxy largely by using Cepheids and RR Lyrae stars as distance indicators. His approach was based on the P-L relation discovered by Leavitt and on the assumption that all these variables have the same P-L relation. As he saw it, this assumption was most likely true in the case of the RR Lyrae stars, because all variables of this type in any given globular cluster have the same apparent brightness. If all RR Lyrae variables have the same intrinsic brightness, then it follows that differences in apparent brightness must be due to different distances from Earth. The final step in developing a procedure for determining the distances of variables was to calculate the distances of a handful of such stars by an independent method so as to enable calibration. Shapley could not make use of the trigonometric parallax method, since there are no variables close enough for direct distance measurement. However, he had recourse to a technique devised by the Danish astronomer Ejnar Hertzsprung that could determine distances to certain nearby field variables (i.e., those not associated with any particular cluster) by using measurements of their proper motions and the radial velocity of the Sun. Accurate measurements of the proper motions of the variables based on long-term observations were available, and the Sun’s radial velocity could be readily determined spectroscopically. Thus, by availing himself of this body of data and adopting Hertzsprung’s method, Shapley was able to obtain a distance scale for Cepheids in the solar neighbourhood.

Shapley applied the zero point of the Cepheid distance scale to the globular clusters he had studied with the 152-cm (60-inch) telescope at Mount Wilson. Some of these clusters contained RR Lyrae variables, and for these Shapley could calculate distances in a straightforward manner from the P-L relation. For other globular clusters he made distance determinations, using a relationship that he discovered between the brightnesses of the RR Lyrae stars and the brightness of the brightest red stars. For still others he made use of apparent diameters, which he found to be relatively uniform for clusters of known distance. The final result was a catalog of distances for 69 globular clusters, from which Shapley deduced his revolutionary model of the Milky Way Galaxy—one that not only significantly extended the limits of the galactic system but that also displaced the Sun from its centre to a location nearer its edge.

Shapley’s work caused astronomers to ask themselves certain questions: How could the existing stellar data be so wrong? Why couldn’t they see something in Sagittarius, the proposed galactic centre, 30,000 light-years away? The reason for the incorrectness of the star count methods was not learned until 1930, when Lick Observatory astronomer Robert J. Trumpler, while studying open clusters, discovered that interstellar dust pervades the plane of the Milky Way Galaxy and obscures objects beyond only a few thousand light-years. This dust thus renders the centre of the system invisible optically and makes it appear that globular clusters and spiral nebulae avoid the band of the Milky Way.

Shapley’s belief in the tremendous size of the local galactic system helped to put him on the wrong side of the argument about other galaxies. He thought that, if the Milky Way Galaxy was so immense, then the spiral nebulae must lie within it. His conviction was reinforced by two lines of evidence. One of these has already been mentioned—the nova S Andromeda was so bright as to suggest that the Andromeda Nebula most certainly was only a few hundred light-years away. The second came about because of a very curious error made by one of Shapley’s colleagues at Mount Wilson Observatory, Adrian van Maanen.

The van Maanen rotation

During the early 20th century, one of the most important branches of astronomy was astrometry, the precise measurements of stellar positions and motions. Van Maanen was one of the leading experts in this field. Most of his determinations of stellar positions were accurate and have stood the test of time, but he made one serious and still poorly understood error when he pursued a problem tangential to his main interests. In a series of papers published in the early 1920s, van Maanen reported on his discovery and measurement of the rotation of spiral nebulae. Using early plates taken by others at the 152-cm (60-inch) Mount Wilson telescope as well as more recent ones taken about 10 years later, van Maanen measured the positions of several knotlike, nearly stellar images in the spiral arms of some of the largest-known spiral nebulae (e.g., M33, M101, and M51). Comparing the positions, he found distinct changes indicative of a rotation of the spiral pattern against the background of surrounding field stars. In each case, the rotation occurs in the sense that the spiral arms trail. The periods of rotation were all approximately 100,000 years. Angular motions were about 0.02 second of arc per year.

Shapley seized the van Maanen results as evidence that the spirals had to be nearby; otherwise, their true space velocities of rotation would have to be impossibly large. For example, if M51 is rotating at an apparent rate of 0.02 second of arc per year, its true velocity would be immense if it is a distant galaxy. Assuming that a distance of 10,000,000 light-years would lead to an implausibly large rotation velocity of 12,000 km/sec, Shapley argued that, if a more reasonable velocity was adopted—say, 100 km/sec—then the distances would all be less than 100,000 light-years, which would put all the spirals well within the Milky Way Galaxy.

It is unclear just why such a crucial measurement went wrong. Van Maanen repeated the measures and obtained the same answer even after Hubble demonstrated the truth about the distances to the spirals. However, subsequent workers, using the same plates, failed to find any rotation. Among the various hypotheses that science historians have proposed as an explanation for the error are two particularly reasonable ideas: (1) possibly the fact that spiral nebulae look like they are rotating (i.e., they resemble familiar rotational patterns that are perceivable in nature) may have influenced the observer subconsciously, and this subtle effect manifested itself in prejudicing the delicate measurements, or (2) possibly the first set of plates was the problem. Many of these plates had been taken in an unconventional manner by Ritchey, who swung the plate holder out of the field whenever the quality of the images was temporarily poor because of atmospheric turbulence. The resulting plates appeared excellent, having been exposed only during times of very fine seeing; however, according to some interpretations, the images had a slight asymmetry that led to a very small displacement of star images compared with nonstellar images. Such an error could look like rotation if not recognized for what it really was. In any case, the van Maanen rotation was accepted by many astronomers, including Shapley, and temporarily sidetracked progress toward recognizing the truth about galaxies.

The Shapley-Curtis debate

The nature of galaxies and scale of the universe were the subject of the Great Debate, a public program arranged in 1920 by the National Academy of Sciences at the Smithsonian Institution in Washington, D.C. Featured were talks by Shapley and the aforementioned Heber Curtis, who were recognized as spokesmen for opposite views on the nature of spiral nebulae and the Milky Way Galaxy. This so-called debate has often been cited as an illustration of how revolutionary new concepts are assimilated by science. It is sometimes compared to the debate, centuries before, over the motions of the Earth (the Copernican revolution); however, though as a focal point the debate about Earth’s motion can be used to define the modern controversy, the Shapley-Curtis debate actually was much more complicated.

A careful reading of the documents involved suggests that, on the broader topic of the scale of the universe, both men were making incorrect conclusions but for the same reasons—namely, for being unable to accept and comprehend the incredibly large scale of things. Shapley correctly argued for an enormous Milky Way Galaxy on the basis of the P-L relation and the globular clusters, while Curtis incorrectly rejected these lines of evidence, advocating instead a small galactic system. Given a Milky Way Galaxy system of limited scale, Curtis could argue for and consider plausible the extragalactic nature of the spiral nebulae. Shapley, on the other hand, incorrectly rejected the island universe theory of the spirals (i.e., the hypothesis that there existed comparable galaxies beyond the boundaries of the Milky Way Galaxy) because he felt that such objects would surely be engulfed by the local galactic system. Furthermore, he put aside the apparent faint novae in M31, preferring to interpret S Andromeda as an ordinary nova, for otherwise that object would have been unbelievably luminous. Unfortunately for him, such phenomena—called supernovae—do in fact exist, as was realized a few years later. Curtis was willing to concede that there might be two classes of novae, yet, because he considered the Milky Way Galaxy to be small, he underestimated their differences. The van Maanen rotation also entered into Shapley’s arguments: if spiral nebulae were rotating so fast, they must be within the Milky Way Galaxy as he conceived it. For Curtis, however, the matter provided less of a problem: even if spiral nebulae did rotate as rapidly as claimed, the small scale of Curtis’s universe allowed them to have physically reasonable speeds.

The Shapley-Curtis debate took place near the end of the era of the single-galaxy universe. In just a few years the scientific world became convinced that Shapley’s grand scale of the Milky Way Galaxy was correct and at the same time that Curtis was right about the nature of spiral nebulae. Such objects indeed lie even outside Shapley’s enormous Milky Way Galaxy, and they range far beyond the distances that in 1920 seemed too vast for many astronomers to comprehend.