Updated May 2019
Steven Dick has written an article on this topic, focusing on the classification itself, for the Encyclopedia of Knowledge Organization: Astronomy’s Three Kingdom System: a comprehensive classification system of celestial objects (2019).
It has always interested me to see how the development of ideas of classification and categorisation in the information sciences has been intertwined with analogous developments in the natural sciences. This is most obviously the case for botany, where Linnaeus’s stipulation that “classification and name-giving will be the foundation of our science” could equally well apply to the information disciplines. But it is also true elsewhere, not least in astronomy, as is shown by Steven J. Dick’s magisterial work Discovery and Classification in Astronomy. Dick, an astronomer, astrobiologist and historian of science, is perhaps uniquely qualified to write such a book.
Dick argues that classification is fundamentally enmeshed with the process of discovery in astronomy, and many (if not all) major astronomical discoveries amount to the recognition of a new class of objects. Therefore, what is counted as a ‘class’, what criteria are used for classification, and by what means and processes new classes and classification systems become accepted, are vital issues. The wide-ranging historical and conceptual discourse in this book, which includes a novel classification of astronomical objects into 82 classes in three ‘kingdoms’ (planets, stars, and galaxies), gives the fullest account of these debates that we have yet had. Remarkably, despite the many classifications of particular kinds of astronomical objects presented in this book, this is first modern comprehensive classification of the astronomy domain.
He proposes a new idea of astronomical discovery, typically drawn out over a long time period, and comprising phases of detection, interpretation, and multiple stages of understanding. This is complicated by periods of ‘pre-discovery’: it has often been the case that astronomical objects have been seen and noted, but without their true nature being realised until much later.
The origins of systematic classification in astronomy are dated by Dick to William Herschel’s classification of nebulae, which first raised the possibility that some were dense groups of stars, which might be resolved with a powerful enough telescope, while others were different; we know today that the latter may be gaseous objects within the Milky Way galaxy, or remote galaxies in their own right. He used eight main classes of nebulae based on appearance: bright nebulae, faint nebulae, very faint nebulae, planetary nebulae, very large nebulae, very condensed and rich clusters of stars, compressed clusters of small and large stars, and coarsely scattered clusters of stars. Dick mentions the intriguing possibility that Herschel may have been inspired to produce this first astronomical taxonomy under the influence of his brother Dietrich, a keen natural historian and butterfly collector. Herschel himself recognised an analogy with library classification and book arrangement, modestly describing his classification as “little more than an arrangement of the objects for the convenience of the observer compared to the disposition of the books in a library, where the different sizes of the volumes is often more considered than their contents”.
The continuing relation between classification in astronomy and in natural history is emphasised by a quotation from the historian David DeVorkin about twentieth century astronomy: “Akin to the naturalist, the typical American professional astronomer was collector and classifier. Instead of museum shelves and cases, astronomers stored their systematic observations in plate vaults and letterpress log books, and displayed them in catalogues sponsored by universities and observatories”.
Classifications of nebulae were revised as astrophotography and spectroscopy revealed more of their detailed nature. There was a continuing debate as to whether such classifications were true reflections of reality, or, as Edwin Hubble said of Max Wolf’s classification of 1909, merely “temporary filing systems”. The debates continued into the second half of the twentieth century, with the recognition of new classes such as quasars and blazars, now known to be types of active galaxies. New discoveries have raised the question as to what counts as an ‘object’ to be classified. Is a black hole an object? A galactic filament, a huge and extended structure? A galactic void, which is defined by the absence of any objects?
Dick goes on to consider the classification of stars, initially by colour, later shown to be related to temperature and composition; an illustration of the way in which an increasing sophisticated understanding is aided by, and in turn influences, classifications.
Classifications of stellar systems led to debates as to whether ‘double stars’ were an example of ‘multiple stars’ or were a separate category, and at what arbitrary point multiple star systems end, and sparse open clusters begin. The same question may be asked asked of galaxies: at what point does a group of galaxies become a cluster, and a cluster a super-cluster, if indeed there is any real difference between the concepts.
Naturally, Dick gives a lot of attention to the case of Pluto, which has been mentioned on this blog. Classed as one of the nine planets of the solar system on its discovery in 1930, its status was changed in 2006 to that of a dwarf planet, alongside many other such with orbits outside that of Neptune. He notes the oddity that dwarf planets are not considered a sub-set of planets, but as a parallel category, criticising it as incorporating neither neither consistent principles, nor precise language, and suggesting that astronomy have much to learn from biological taxonomy in this respect.
The Pluto debate raised a number of issues familiar in bibliographic classification. Was dual classification allowed, i.e. could Pluto be both a planet and a trans-Neptunian object? Was the classification to be made on purely scientific grounds: Dick quotes David Levy, the biographer of Pluto’s discoverer, Clyde Tombaugh, as arguing that “science wasn’t just for scientists or taxonomists, but for people, and normal people considered Pluto a planet”. Indeed, Pluto’s classification became a political issue, and not just in the sense of academic politics; the House of Representatives of New Mexico, where Tombaugh was a long-time resident, passed a resolution to the effect that when Pluto was visible in the New Mexico night sky it was a planet. Should the classification be dependent on its users: planetary scientists consider Pluto a planet, as it has an atmosphere, geological changes, and weather, while astronomers, more interested in its size, shape, and orbit, do not.
The Pluto case clearly shows up the realist/constructivist dichotomy: Pluto’s existence is a reality, but what class it is assigned to is socially determined. Dick suggests that the less that is known conceptually about an object, the more social construction plays a central role. Had Pluto’s true size been known at the time of its discovery, plus the fact that there were many trans-Uranian objects of similar size, then it would not have been declared a planet in the first place. The discovery of astronomical objects is an unveiling of nature, while the creation of classification systems is a wholly human invention; creation of classes within a system, and the assignment of objects to them, falls somewhere in the middle.
“In the end”, Dick writes, “despite being grounded in nature, the declaration of a new class of astronomical object is a socially determined exercise, and the construction of any classification scheme doubly so, as class is piled upon class in the attempt to order nature”. This viewpoint, fully justified by the detailed analyses in the book, is a useful corrective to claims of ‘ontological realism’ on the part of those who claim that scientific ontologies, or taxonomies for that matter, but necessarily be ‘true’. The same is true of biological ontologies, as shown by the studies of Charlie Mayor and Lyn Robinson on the Genome Ontology.
One general conclusion that Dick draws from this lengthy and detailed story is that astronomy classification schemes must have a “Goldilocks” quality – not too simple, and not too complex – if they are to be useful to the practitioner. This is likely to be true for all forms of classification. Another is that classification schemes can evolve over time, as new information and understanding is attained, but they must not evolve too much, or they will lose their original usefulness. These points are true for bibliographic classifications, and other systematic vocabularies in the information sciences. Scientific taxonomy and document classification still have much in common, and much to learn from each other.