History of Astronomy Example Exam Question Hints and Suggestions

The basic skeleton of an answer to this question is as follows:

• 1860s, Fr Angelo Secchi: four (or five) numbered types
  • Type I, "Sirian", white stars with prominent hydrogen lines;
  • Type II, "solar", yellow stars with numerous fine lines;
  • Type III, "Antarean", orange and red stars with banded spectra, bands darkening towards the violet end;
  • Type IV, "small red stars", resembles Type III, but fewer bands, and darkening towards the red (Secchi correctly identified these as carbon);
  • stars with emission lines, such as γ Cas (sometimes called Type V).
• 1890, Pickering and Fleming, original Harvard system
  • finer subdivision of Secchi's system consisting of alphabetical list A-O in decreasing order of strength of hydrogen lines;
  • A-D correspond to Secchi's Type I, E-L to Type II, M to III, N to IV, and O to emission-line stars; additionally, P denotes planetary nebulae, and Q is "none of the above".
• 1890-1918, Maury and Cannon, further developments of the Harvard system
  • Pickering's group quickly recognised deficiencies in original system:
  • Antonia Maury replaced it with system of Roman numerals with qualifiers, which did not catch on;
  • Annie Jump Cannon deleted some classes which were spurious (e.g. C, doubled lines, which came from faulty plates), or insufficiently distinct (e.g. E and H absorbed into F), introduced subclasses 0-9, and reordered so that all line intensities changed smoothly through the sequence: this gave OBAFGKMN, which she knew was a temperature sequence (at least to M; the difference between M and N is composition rather than temperature).
• 1918-1930s, further refinement of Harvard scheme
  • 1920s, suffix letters introduced to denote particular features, e.g. e = emission lines, m = anomalous metallic lines;
  • subclass system extended to classes O and M (originally applied only to classes B-K);
  • classes R and S introduced: R and N are similar in temperature to GKM but carbon-rich (they were later combined into class C); S stars have ZrO instead of M class TiO (they are intermediate between C-rich C and O-rich M).
• 1943, Morgan, Keenan and Kellman, luminosity classes
  • suffix classes I-V introduced, based on line widths sensitive to density (or surface gravity, and hence to size of star): I = supergiant, V = "dwarf" (main sequence star).

The basic 19^th century argument is as follows.

1. Some nebulae are unquestionably gaseous, since they display emission line spectra (Huggins, 1860s on). These must be small objects within the Milky Way, since if they were extragalactic they would have to be unreasonably large.
2. Some of these gaseous nebulae have stars embedded within them.
3. Therefore, there is a continuum of nebular spectra, from pure gas through gas-plus-stars to purely stellar, and, since both the gas and gas-plus-stars classes are small objects within the Milky Way, the purely stellar class presumably is too.

Now, statements (1) and (2) are perfectly correct, but statement (3) does not follow (especially since various 19^th century astronomers, such as Lord Rosse, had resolved some of the nearer galaxies into stars, and had commented on how small those stars were - which should have indicated that these were much more distant objects than the gas-plus-stars nebulae which contained perfectly "normal" stars). The willingness to make this unjustified logical leap is very likely influenced by the fact that all of these objects were classified as "nebulae", which predisposed astronomers to view them as related objects.

A further argument in favour of "small, nearby" nebulae was S Andromedae, a "temporary star" observed in 1885 in M31. Interpreted as a nova, this was much too bright to allow M31 to be placed outside the Milky Way.

This is also a classification issue - S And was in fact a supernova - but since no local supernovae had occurred since 1604, 19^th century astronomers did not recognise the great difference between novae and supernovae, and therefore did not have the appropriate "box" available to put S And in.

• What was the evidence?
  Clearly, in order to establish that (some) nebulae are not within the Milky Way, you need to obtain convincing evidence of their large distance. Simply resolving local nebulae into stars was insufficient - several late 19^th century photographs of Local Group Galaxies, such as this 1899 image of M33 by J.E. Keeler, seem to be at least partially resolved, yet Clerke in 1902 was convinced that nebulae are all local. The key was to resolve stars of clearly identifiable type. In the 1920s, observers using the new silver-on-glass reflecting telescopes successfully resolved both classical novae (correcting the wrong impression given by S Andromedae!) and, crucially, Cepheid variables.
• Why was it available at this time?
  The principal factor is the availability of new technology - especially large silver-on-glass reflectors. After Keeler demonstrated the value of reflecting telescopes for photographic work at the turn of the 20^th century with the Crossley reflector, the first modern research reflecting telescope, the Mt Wilson 60-inch, was completed in 1908, followed by the 100-inch Hooker telescope in 1917. These telescopes, and improvements in photographic techniques, made the identification of Cepheids in nearby galaxies possible.
• Could it have been done earlier?
  Classical novae were observed in nearby galaxies using the 60-inch, and some astronomers (e.g. Heber Curtis, 1917) did indeed use these observations to argue that these systems were extragalactic. But the absolute magnitudes of classical novae were not well enough established to convince everyone - Shapley 1917 discusses much the same data as Curtis but comes to the opposite conclusion. [He appears to have been misled by S Andromedae, see above, and by Z Centauri, a "nova" seen in 1895 in NGC 5253; like S And, Z Cen is now accepted as a supernova.] On the whole, it seems possible with hindsight that the extragalactic nature of spiral nebulae should have been established with the data on novae, and could therefore have been done about 10 years earlier, but the definitive result needs Cepheids (which can be unambiguously identified by showing that your candidates have the correct period-luminosity relation), and this was first done by Hubble in 1925.