The Role of Images in Astronomical Discovery Read online

  Fig. 4.3 Schematic of the “sidereal system” derived by William Herschel from his stellar counts.

  From Herschel (1785), Philosophical Transactions of the Royal Society of London.

  To frame his endeavour soundly, Herschel made two simplifying assumptions: first, that

  his instruments allowed him to see into the farthest limits of the sidereal system; and sec-

  ondly, that the stars were distributed regularly throughout space. He then defined approxi-

  mately 300 sections having the same solid angle, spread across the sky that was visible from

  13:59:10, subject to the Cambridge Core

  .006

  4. Portraying “Nebulae” for the Mind

  97

  his observing location. Then he counted the stars in each of these conic sections. He counted

  and counted, using the brightness of the stars as an indicator of their distance, assuming that

  the fainter ones were further away. Herschel’s method was primitive and his assumptions

  were incorrect. But overall his statistical sampling procedure was valid. The Kapteyn pro-

  gram that I discussed in Chapter 3 was a rebirth of Herschel’s dream.

  Herschel’s counting method was adopted and improved by other astronomers during

  the nineteenth century, including his son John Herschel. It deeply influenced the work of

  Russian astronomer Friedrich Georg Wilhelm Struve (1793–1864), director of the Pulkovo

  Observatory in St. Petersburg.13 Applying a similar counting method, Struve demonstrated

  that the brighter stars tended to conglomerate in the plane of the Milky Way. In 1847, Struve

  even suggested from his counts the existence of an attenuation of stellar light by one magni-

  tude per c. 3,000 light-years of the line of sight. Although he did not explain the extinction

  mechanism, Struve had discovered an important component of the matter filling the space

  between the stars: fine dust particles distributed between the stars that absorb or diminish

  the intensity of stellar light as it travels through sidereal space, causing the so-called “zone

  of avoidance.” This dimming effect was ignored for the next 70 years.14

  In the 1910s, Heber Curtis used Herschel’s statistical counting approach to estimate the

  number of spiral galaxies observable with the 36-inch Crossley reflector. Always doubtful

  of the reality of the large recession velocities of galaxies, Edwin Hubble conducted a sys-

  tematic count of galaxies as a function of apparent brightness during the 1930s. He tried to

  use the number of galaxies counted to discriminate between models that explained galaxy

  spectral redshift as being due to the expansion of the universe, and models proposing instead

  some unknown phenomenon such as light “fatigue.”

  Confronting Bias

  Naturally there is a subjective aspect to synoptic representation. For example, decisions

  have to be made about what features are important or what quantities need to be illustrated.

  As large set of data points are averaged, and a consistent method for smoothing the data has

  to be adopted and described. Furthermore, all scientific measurements are subject to biases.

  The designer needs to state the appropriate warning and limitations. For example, the use

  of a particular telescope, instrument and detector biases the astronomical detection in ways

  that depend on the observing wavelength, and hence the filter used. Biases and systematic

  effects are introduced by optical effects (e.g. diffraction or limits of the optics), quantum

  physics effects (threshold of detection or sensitivity as a function of wavelength), the dura-

  tion of exposure (fainter objects remain undetected), and varying atmospheric conditions.

  As noted by James Keeler, the difference in the response of the human eye compared to that

  of the photographic emulsion explained some of the differences between objects recorded

  13 J. North, Cosmos: An Illustrated History of Astronomy and Cosmology, Chicago: University of Chicago Press, 2007, p. 448–

  449.

  14 Modern estimates of interstellar extinction in the Milky Way are 0.7–1.0 magnitudes per kiloparsec, or a factor of 2 to 2.5 of attenuation per 3,200 light-years.

  13:59:10, subject to the Cambridge Core

  .006

  98

  Part I – Images and the Cosmos

  by drawing and by photography.15 The use of different photographic emulsions has been

  driven both intentionally and by practical reasons. As we will see in Chapters 9 and 10,

  such choices affect the process of galaxy classification and influence interpretation.16

  Another biasing effect is brought about by sampling. For example, as we observe further

  and further away in the universe, intrinsically faint objects drop out. A sample of distant

  objects gets skewed towards the more luminous objects, as intrinsically brighter objects can

  be observed further away. This creates a false trend of increasing average luminosity as a

  function of distance. This is called the Malmquist bias, an important selection effect popu-

  larized by the Swedish astronomer Gunnar Malmquist (1893–1982).17 The effect must be

  understood and corrected for, which is complicated by the fact that there are also evolu-

  tionary effects that modify the underlying population. Distant objects are seen at a time the

  universe and these objects were younger. Furthermore, as objects evolve with time, their

  population representation changes over different periods of time.

  For extended objects, i.e. those that are not as point-like as stars, the surface brightness

  can introduce a biasing trend. The night sky has an intrinsic brightness, partly due to the flu-

  orescence of the Earth’s atmosphere, but also as a result of the general stellar background. It

  is easier to detect objects of high rather than low surface brightness. If you examine Plate 6.3

  closely, you will find dozens of faint galaxies, which are barely visible. This greatly affects

  the detection and measurement of extended faint objects, either regular nebulae in our own

  galaxy or galaxies of low surface brightness in the extragalactic world. The knowledge that

  these objects exist and the need to assess and count them galvanize efforts to optimize the

  transmission and control of scattered light in optical systems. Observations of the nearby

  universe and the modeling of galaxy evolution can correct for this. However, it remains

  extremely difficult to detect the large populations of faint galaxies.

  The Drive to Classify

  Synoptic representation triggers another epistemological stage: a classification process that

  tries to combine identical or similar features of several objects to create an encompassing

  class of objects or extract an archetypal object. In his 1811 paper, William Herschel pre-

  sented two plates to show the main nebular shapes (Fig. 2.4). It was an undifferentiated

  mix of what we now know to be galactic nebulae and external galaxies. Herschel saw all of

  them as one class of objects and he tried to sequence them.18 He suggested that the degree of

  condensation of “nebulae” was related to gravity, thus determining nebular shapes. Perhaps

  more importantly, in the light of later developments, Herschel inferred that the shapes of

  15 J. E. Keeler, Note on a Cause of Differences Between Drawings and Photographs of Nebulae, Publications of the Astronomical Society of the Pacific, 1895, Vol. 7, pp. 279–282.

  16 The impact of filters and wavebands with regard to galaxy morphology an
d classification is discussed by the authors in R. J.

  Buta, H. G. Corwin Jr. and S. C. Odewahn, The de Vaucouleurs Atlas of Galaxies, Cambridge: Cambridge University Press, 2007.

  17 G. Malmquist, On Some Relations in Stellar Statistics, Arkiv för Mathematik, Astronomi och Fysik, 1922, Vol. 16, pp. 1–52.

  18 M. J. Way, Dismantling Hubble’s Legacy? in Origins of the Expanding Universe: 1912–1932, Astronomical Society of the Pacific Conference Series, 2013, Vol. 471, pp. 102–103.

  13:59:10, subject to the Cambridge Core

  .006

  4. Portraying “Nebulae” for the Mind

  99

  “nebulae” were physically meaningful as criteria to be used by classifiers. Stephen Alexan-

  der also sought such an ordering (see Fig. 2.11 for a small collection of shapes assembled

  by Alexander).

  Almost a century after Herschel, the German astronomer Max Wolf (1863–1932) of Hei-

  delberg Observatory, also a pioneer of astrophotography and wide-field photography, intro-

  duced a classification scheme for “small nebulae” ( kleine Nebelflecken) (Fig. 4.4). Having

  commented on the limited power of spectroscopy to sort nebulae – this had been attempted

  by American astronomer Edward Charles Pickering (1846–1919)19 – Wolf used morphol-

  ogy as a criterion for classifying all “nebulae.” Avoiding any inference about the intrinsic

  nature of the “nebulae,” he wrote cautiously: “You will therefore need to be purely descrip-

  tive, without any consideration of a hypothesis about the origin of the mist, or the nature

  of the spectrum.”20 Wolf’s drawings, derived from photographs, appeared to be of poor

  quality compared to those produced at Birr Castle 50 years before. While William Parsons’

  team depicted specific objects as accurately as possible, Wolf purposely mixed and aver-

  aged features. What mattered was not the accuracy of the sketch but the common structural

  information that the morphology conveyed.

  In putting together his global scheme, Wolf extracted 23 “classes” that summarized the

  main shapes of the foggy nebular images. He rightly emphasized the dependence of the

  quality and shapes of the images delivered on the optical instruments used. Like Herschel’s

  1811 plates, Wolf’s set is a mix of objects that included Milky Way nebulae and galaxies.

  Although spiral shapes and edge-on systems were clearly recognized, Wolf did not add

  much to the work of Parsons. Wolf’s merit resided in his teleological purpose: to show that

  the natural sequence of shapes betrays deeper physical phenomena and processes. We will

  return to the topic of galaxy classification in Chapter 9.

  High-Level Representations

  As more quantitative physical measurements grew, the use of synoptic representations in

  nebular and galactic research increased. With the advent of photography, quantitative pho-

  tometry became possible and the technique was quickly applied to images of galaxies;

  images or charts in different wavebands were calibrated and constructed.

  Isophotal maps, connecting portions of equal brightness in the image of an object,

  turned out to be essential for comparing the same objects imaged at different wavelengths,

  especially for those obtained in the non-visible domains. It was particularly important to

  relate these findings with observations and images in the radio and the X-ray domains that

  astronomers had started working with during the second half of the twentieth century. Omar

  Nasim has given a fine description of the work of the young American astronomer Ebenezer

  Mason, who was introduced at the start of this chapter. Mason was probably the first to use

  19 Pickering led a large effort at Harvard College Observatory that resulted in an effective and successful spectral classification system of stars.

  20 M. Wolf, Die Klassifizierung der kleinen Nebelflecken, Publikationen des Astrophysikalischen Instituts Königstuhl-Heidelberg, 1908, Vol. 3, pp. 109–112 (translation by the author).

  13:59:10, subject to the Cambridge Core

  .006

  100

  Part I – Images and the Cosmos

  Fig. 4.4 Early classification of “nebulae” by Max Wolf. From Wolf (1908), Publikationen des Astro-

  physikalischen Instituts Königstuhl-Heidelberg. Courtesy of John G. Wolbach Library, Harvard

  College Observatory. 2016.

  13:59:10, subject to the Cambridge Core

  .006

  4. Portraying “Nebulae” for the Mind

  101

  the method of lines of equal brightness in astronomical observation.21 “It was first sug-

  gested by the method usually adopted for the representation of heights above the sea-level

  on geographic maps, by drawing curves which represent horizontal sections of a hill and

  valley at successive elevations.”22 In a parallel approach, John Herschel produced descrip-

  tive maps where a nebular drawing was started by making an accurate reference grid of

  stars, which served as a sort of triangulation basis; he then drew nebular intensity, carefully

  linking areas of equal brightness.

  Isocontour maps of physical properties became the standard means of non-homomorphic

  representation. Nasim refers to the “isoline craze” that embraced the natural sciences dur-

  ing the nineteenth century.23 Astronomy was one of these “infected” sciences. Isophotal

  maps as Mason invented them became extensively used in astronomy and continue to be a

  universal way to present imaging data (see Plate 4.1; Plate 7.2).

  Digital imaging enables an additional step, that of creating intensity ratios of images

  of different wavelengths. Another dazzling example has been the use of isovelocity maps:

  the joining up of locations of equal radial velocity points across an object (Plate 4.2). Such

  maps have been used to establish the behavior of the orbital velocities of stars or gas as a

  function of the distance to a galaxy center. These rotational curves and maps have become

  essential tools for deriving the mass distribution within individual galaxies and their total

  mass. The use of “false colours” is most effective in highlighting the information to be

  conveyed.

  Probably one of the most stunning non-homomorphic maps in modern astronomy is that

  of the cosmic microwave background (CMB), a relic of the distribution of matter in the uni-

  verse as it was about 380,000 years after the Big Bang (Chapter 5). Since the serendipitous

  discovery by Arno Penzias and Robert Wilson in 1964, we know that the universe is filled

  with microwave radiation. The intensity distribution across wavelengths corresponds to a

  perfect black body at 2.725 K. The fossil glow peaks in the millimeter–centimeter wave-

  length range of the spectrum, the microwave region. The present-day CMB is the remnant

  light from the time when the mean temperature of the universe had cooled enough, due

  to expansion, for the free electrons to be captured by the protons. The universe became

  transparent and the photons were left to cool as the universe continued its expansion.

  There were no stars or galaxies this far back in time. Intriguingly, the microwave back-

  ground is not absolutely uniform. It is spotted with tiny fluctuations in temperature (Plate

  4.3). These are of the order of only 1/100,000 K but have been precisely measured with

  very sensitive radio telescopes in space and on the ground. When mapped over the surface

  of the sky, the fluctuations in temper
ature show an orderly angular structure; the mottling

  has a pattern. The most straightforward interpretation of the pattern is that these structures

  correspond to regions of the young universe where matter had started to assemble under the

  21 J. Shears, et al., In the Footsteps of Ebenezer Porter Mason and His Nebulae, 2014arXiv:1401.7960. See also O. W. Nasim, Observing by Hand: Sketching the Nebulae in the Nineteenth Century, Chicago: University of Chicago Press, 2013, pp. 131–

  137.

  22 Cited in J. Shears et al., In the Footsteps of Ebenezer Porter Mason and His Nebulae, 2014arXiv:1401.7960, p. 3.

  23 O. W. Nasim, Observing by Hand: Sketching the Nebulae in the Nineteenth Century, Chicago: University of Chicago Press, 2013, p. 131.

  13:59:10, subject to the Cambridge Core

  .006

  102

  Part I – Images and the Cosmos

  action of gravity. As these enhanced structures grew, they led to large-scale structures and

  to the superclusters of galaxies we observe in the universe of today. The map of the spatial

  temperature fluctuations of the CMB is a powerful non-homomorphic representation of the

  distribution of matter in the young universe a few hundred thousand years after the Big

  Bang.

  These few examples show that images as representations of nature do not need to be an

  exact replica of what an omnipotent human eye would see, also assisted by an instrument

  like a telescope or a microscope. Non-homomorphic images can convey information of a

  higher level.

  This chapter concludes the overview of the different approaches and techniques that

  astronomers have developed over the centuries for viewing, sketching and photograph-

  ing “nebulae,” in brief, to make images of the “nebulae.” In the next four chapters, the

  style changes. I present the world of galaxies using a more thematic approach, highlighting

  key individuals and techniques, as well as the specificities of imaging in each wavelength

  domain (radio, infrared, visible, ultraviolet, X-rays and gamma-rays) that have all helped

  to elucidate the surprising objects and phenomena of the extragalactic universe.

  13:59:10, subject to the Cambridge Core

  .006

  Part II

  Images as Galaxy Discovery Engines

  14:07:19, subject to the Cambridge Core