Chapter 10 Conclusions

In: The Dragon and the Rainbow
Robert Blust
Search for other papers by Robert Blust in
Current site
Google Scholar
Open Access

Once we begin to seriously investigate the ethnology of the rainbow, the truth about the dragon is seen to lie all around us, like a field of newly fallen snow, immaculate and clear. Yet, everywhere we see it trampled into an unrecognizable slush by the ceaseless traffic of baseless speculation. This has been the story of the dragon for well over a century. To state once more the only conclusion that can possibly be supported by the evidence considered in this book, let me end on the words with which I began:

Dragons evolved from rainbows

This pronouncement should not appear as outrageous now as when first seen by the reader. Once it is understood that the Western concept of the rainbow is not typical of traditional cultures, and that many otherwise obscure traits of the dragon have a natural explanation if the dragon is a relic of the Rainbow Serpent, many things that were puzzling become surprisingly clear.

First, it is now clear that the Rainbow Serpent—long believed, and still commonly treated, as a phenomenon sui generis to aboriginal Australians—is globally distributed, appearing in locally different but easily recognizable forms around the world, except where the serpent and rainbow have separated and become dragon and rainbow. This pattern provides another insight, namely that mythological creatures evolve over time as a result of culture change: the Rainbow Serpent in Australia may show many local variations, but no one could confuse it with the dragon in Europe, the ancient Near East, South Asia, the Central or East Asia, or North America. Rather, what must be described as cultural stasis in geographically isolated regions like Australia, has resulted in the preservation of the most ancient forms of the Rainbow Serpent complex, with only minor change over tens of thousands of years, while highly evolved cultures such as that of China, transformed the same original idea into a dragon tradition of extraordinary richness and complexity.

Why has it taken so long for anyone to come to this realization? This is undoubtedly the most vexing question about the history of scholarship on dragons, and it probably is best for me to leave the search for a definitive answer to historians of science. But it would be inappropriate for me to pass silently over this subject at the end of a book which has amassed more relevant data on why the dragon idea is universal than any previous work, and which has argued for a specific approach to the problem—an approach that for unknown reasons no one has previously chosen to follow.

One reason that can be offered for the failure of previous scholarship to come to grips with the problem of the dragon is the virtual abandonment of basic scientific method in all earlier proposals, even by seasoned scientists. In any branch of science, the first step in addressing a question of interest is to collect all observations that are relevant to answering it (in this case the widespread traits of dragons listed in Table 1 and discussed in later chapters). The second step is to propose a hypothesis which accounts in a straightforward manner for as many of these observations as possible, while making the fewest possible assumptions. The last step is to compare this hypothesis with its competitors to determine whether it is truly better able than them to explain the data.

Up to the time of this writing standard compendia of mythology and folklore treat the rainbow, Rainbow Serpent and dragon as separate entries with no cross-referencing, or any indication that the latter two are different evolutionary stages of the same mythic creature. Some critics have said in private that these relationships are obvious, and there is thus no need to state them plainly, but if that is the case, why has this never been presented explicitly before? To cite an example of this mentality, a reader of a paper on the origin of the dragon that I submitted for publication, wrote twenty years ago “Unfortunately this essay does not appeal to a broad international audience, and does not contribute to present-day scholarship.” Among reasons given for this bizarre conclusion was that “The author gleaned a rather large sample from their contexts to prove his claim that the dragon is really the rainbow (as both fact and fiction) that does not seem an earthshaking discovery.” Yet in the same set of remarks this reader suggested that it would be worthwhile to consider the psychoanalytic argument in Roheim (1940) as an alternative hypothesis, and it was stated that the radical diffusionist argument in Hentze (1966) “is a must.” For anyone seriously interested in the data, however, one must ask whether either of these papers even addresses the trait distributions that form the substance of this book, let alone offers an explanation for them.

Another reason that previous treatments have fallen short is due to an ‘either/or’ mentality in the sociology of science. When nineteenth century evolutionary anthropology was rejected by field-oriented scholars it was abandoned wholesale. This meant that, for whatever emotional reasons, earlier work was treated as having no value, a reaction that is obviously unjustified. As has been mentioned repeatedly, Tylor’s 1871 theory of animism remains as solidly grounded today as it was when he proposed it nearly 150 years ago. Later writers criticized Tylor for putting Europeans at the top of the world hierarchy of cultures, but in the late nineteenth century, by any measurable material or economic standard, Europeans were at the top of this hierarchy. Tylor never claimed that Europeans are inherently superior to others—on the contrary, he was at pains to show that many features of European cultures that no longer have a clear explanation are survivals from a time when Europeans lived more like the tribal peoples he compared them with. Tylor’s idea of a ‘survival’ (1958.1:70–111) was the key element in his argument that Europeans had once been more like ‘primitive’ people. To anyone with curiosity and an open mind, it should be apparent that many of the traits of dragons discussed here are arguably survivals from a time when the dragon was the Rainbow Serpent. Why are dragons controllers of rainfall? Why do they guard springs, wells, lakes and the like? Why have they been reported as residents of virtually all major waterfalls where ethnographic data is available? Why do they live in caves or fly? Why are they opposed to thunder and lightning in widely-separated cultures? To refuse to see these and other examples as survivals in Tylor’s sense because one believes that all of Tylor’s ideas are ‘old-fashioned’ is an inexcusable intellectual prejudice. The notion of cultural survivals has its utility, and there is no reason to reject it out of hand when it can provide insights that are otherwise unavailable.

Much the same can be said for the work of Friedrich Max Müller. Yes, his ‘solar theory’ went too far, and he made other mistakes, as any scholar does today. But how does that justify refusing to even consider his central thesis, that many myths, at least at their first formulation, are disguised attempts to explain the workings of nature? A basic thesis present here is that the universal dragon idea was not a whimsical creation, but satisfied a basic need to explain the workings of nature in relation to the critically important factor of rainfall. If the preliminary explorations of comparative anthropology in the nineteenth century had been refined, extended, and respected, instead of being largely rejected in their entirety, someone surely would have addressed the ethnology of the rainbow a century ago. And if that had happened, the true nature of the dragon would have been understood far earlier, and much spilt ink used to support baseless speculations about why the dragon idea is universal could have been saved.

Yet another reason that the dragon has so long eluded the net of scientific inquiry is simple confusion on the part of researchers who tried to generalize on the basis of an inadequate database. Smith’s radical diffusionist theory starts with the questionable assumption that humans are fundamentally uninventive, and the observation that the earliest historical records of dragons are from Egypt and Mesopotamia. Since dragons are connected in some way with water, and these arid regions depend critically on rainfall for life, the idea of dragons must have started in the ancient Near East—or so the thinking goes. However, there is no logical transition here: just because the earliest written records describing the dragon idea are from Egypt and Mesopotamia hardly means that the dragon idea does not have a longer history. Add to that the fact that control of rainfall is important in most parts of the world, and the whole fabric of Smith’s argument unravels. Similarly, Lœwenstein (1961) provided a valuable survey of the worldwide association of rainbows with water serpents, but he then reversed the basic logic of animism to argue that the Rainbow Serpent complex used the rainbow to represent large and psychologically impressive snakes, rather than using the latter to explain the rainbow. In all of these varied accounts the most basic questions that anyone should ask were simply ignored: how would preliterate humans explain the appearance of a rainbow? How would they represent the fact that it appears only when sun and rain are competing for control of the sky? How would they account for where it is when it is not visible in the heavens?

Finally, one other reason that scholars have failed to use available resources to draw the conclusion that dragons evolved from Rainbow Serpents, is the widespread adoption of the philosophical position of radical inductivism. Basically, what this means is that all comparison must begin ‘from the bottom up’, starting with understanding the interrelations of the smallest and most closely-related units before moving to the next larger ones, and so on. To such scholars the idea of comparing more distantly related data points before all intermediate ones have been thoroughly analyzed is anathema.

Standard handbooks of mythology, legend and folklore have entries for the dragon, Rainbow Serpent, and rainbow, but these are treated as distinct because it is possible to treat them separately, not because they are inherently unrelated. In what was perhaps the first attempt to understand the universality of the dragon idea, Charles Gould (1886:163–164) had the following to say:

The dragon appears in the mythological history and legendary poetry of almost every nation as the emblem of the destructive and anarchical principle .... No one has attempted to collate the vast bulk of material shrouded in the stories of all lands. If this were perfectly effected a diagnosis of the real nature of the dragon might perhaps be made, and the chapter of its characteristics, alliances and habits completed like that of any other well-established species.

This statement was made over 130 years ago, and until now the research program that Gould envisaged had not been undertaken. Some have come to the brink of recognition, but stopped short, apparently because they felt ill-equipped for the task. To choose one prominent example, A.R. Radcliffe-Brown, one of the leading cultural anthropologists of his generation, saw connections between the Rainbow Serpent of Australia and the dragon in other parts of the world that he did not dare to explore. In his words:

I do not propose to enter into the vast and difficult subject of the relation of the Australian myth to the similar myths of other cultures—the dragon of China, for example, with its positioning as representing the principle of moisture and its connection with the pearl. My belief is that such comparisons can only lead to valid conclusions after intensive study of the belief or custom in each culture-area separately. A good deal of further study of the Australian myth and its relations to other elements of Australian culture is necessary before we can really be sure of understanding it sufficiently to make it the subject of a comparative study (Radcliffe-Brown 1926:25).

Over half a century later a similar position was expressed by Mountford (1978:92–93) who surveyed the Rainbow Serpent myths in Australia, and concluded somewhat more insightfully:

The rainbow-serpent myths of Australia are an extension of the universal serpent myths which permeate the beliefs of all peoples, past and present. Mythical creatures having characteristics similar to those of the rainbow-serpents of Australia are recorded in many countries: in the myths of the dragons of China, of the nāga of India, of the taniwah of New Zealand; in the many half-submerged myths of Europe and the ancient world of Greece and Egypt; in those of the water serpents of the primitive Bushmen of the Kalahari desert; and in those of the Indonesian and Melanesian islands. But we can enter no further into this vast field of serpent beliefs.

Both writers recognized that many cultures have some kind of serpent myth with generic similarities, but were unwilling to venture further into examining why this should be the case. Mountford was basically a descriptivist focused exclusively on Australian aboriginal cultures, with little concern for general theory, but Radcliffe-Brown was a major figure in cultural anthropology in his time, and his position on this issue was highly influential. Some will insist that this is the only responsible approach: work gradually ‘from the bottom up’, comparing only the most similar cultures first, and once the researcher feels confident that the differences between these cultures have been understood, then move on to the next more inclusive level. However, not everyone shares this ‘bottom up’ mentality. Opponents will argue that there is no reason to ‘put on blinders’ and work in this way, ignoring everything else until the time is ripe for comparison between units that have themselves already been securely established from the bottom up. In evaluating the merits of these two points of view, perhaps the first question to ask is whether radical inductivism has always been followed in successful scientific work, and here comparison with the history of scholarship in other social sciences is instructive.

Nineteenth century science is generally credited with two great achievements: 1. the discovery of the Indo-European language family, and 2. Darwin’s theory of evolution by natural selection. The scholars in Denmark and Germany who led the way in the decades-long enterprise of establishing the field of historical linguistics, starting in 1816, quickly recognized that the Germanic languages and others of northern Europe, which had ‘no history’ compared with the classical languages of the Mediterranean and India, are related to them. This insight did not wait for, or require a ‘bottom up’ comparison of Germanic languages, Slavic languages, or others to determine their immediate common ancestors before comparison with Greek, Latin or Sanskrit. Rather, a ‘top down’ approach was used from the beginning, and with the remarkable and lasting success that is known by anyone familiar with the history of scholarship in this field. As Greenberg (1957:50) has put it with reference to problems of linguistic reconstruction that are best resolved by taking more distantly related languages into the comparison:

These and many other examples which could be cited are of interest because they refute the common belief that more remote relationships should be ignored, while each distinct subbranch is reconstructed separately and independently. In many cases we cannot choose between alternative reconstructions without taking the wider family into account. In fact, in Indo-European the reconstruction of the protolanguage of the family as a whole progressed far more rapidly than did that of the individual branches. Moreover, Proto-Germanic and other comparable intermediate unities have always been reconstructed with one eye backward to Proto-Indo-European and one eye forward to the contemporary Germanic languages.

Greenberg’s remarks in this passage refer to the reconstruction of Proto-Indo-European, but the matter that prompted these remarks was a problem in the reconstruction of Proto-Bantu, and it is easy to show that the same simultaneous ‘top down/bottom up’ approach has been successfully employed in the reconstruction of the ancestral stages of most language families; Dempwolff (1934–1938), for example, reconstructed his version of Proto-Austronesian with no concern about needing to first reconstruct Proto-Polynesian, Proto-Micronesian, etc.

If we take historical linguistics as a model for scientific inquiry in the social sciences, then, it does not support the attitude that the process of inference must begin on the most restricted level of inquiry before expanding to ever wider domains. In fact, most scholars in the physical and biological sciences probably would find this level of intellectual insecurity incomprehensible: why should relevant information be ignored out of fear this might lead to unjustified inferences? It should not have to be emphasized that the possibility of error exists on all levels of comparison, from the closest to the most distant, and the best control in coping with this possibility is the one already outlined above: 1. collect all data relevant to answering the research question at issue, 2. construct a hypothesis that explains the maximum number of relevant observations with the minimum number of assumptions, 3. compare the performance of a favored hypothesis with those of its competitors, and 4. select the winner.

This demonstration that a ‘bottom up’ approach to drawing inferences is not the only scientifically responsible or productive one is welcome, but is it even necessary? A moment’s reflection will show that a strict ‘bottom up’ approach is inherently problematic, as it leads to an infinite regress, since in order to study a complex problem A one must first study each subset of A (in Radcliffe-Brown’s case perhaps each culture-area within Australia). But in order to understand each subset of A one needs to first understand each sub-subset of A (in Radcliffe-Brown’s case perhaps each ethnolinguistic group within Australia). And to study each sub-sub-set of A one needs to first understand each sub-sub-sub-set of A (in Radcliffe-Brown’s case perhaps each local group). If conducting meaningful research requires the adoption of such methodological strictures where does one stop (or begin)?

Radcliffe-Brown’s reference to “the Australian myth” implied that there is just one Rainbow Serpent myth across the Australian continent, and that once this belief complex is understood in detail the time may be ripe to move on to a comparison with the dragon idea in China and perhaps elsewhere. However, half a century later Maddock (1978:19) concluded that rather than one Rainbow Serpent, there are many ‘rainbow serpents’ ---that the Rainbow Serpent concept as it has been presented in the literature is not a discrete category, and that therefore “taking ‘the rainbow serpent’ as an object of study is likely to be profitless.”

That there is variation in the form of the Rainbow Serpent complex within Australia (as there is globally), should hardly come as a surprise. This was noted some nine decades ago by Piddington (1930:352), who noted that the Karadjeri of western Australia

possess a very extensive mythology in which the bulaiŋ, a mythical water snake, plays an important part. But one of the many anomalies in Karadjeri culture is that the bulaiŋ which apparently corresponds to the Rainbow Serpent elsewhere in Australia, is in no way connected with the rainbow.

Piddington goes on to describe the bulaiŋ as “a gigantic snake which lives in waterholes”, and is physically a chimera, as in addition to its large eyes, it has long ears. Despite the absence of an explicit connection with the rainbow, several other traits give away the identity of the Karadjeri bulaiŋ as a local variant of the Rainbow Serpent. If it is cooked, for example, it will burst open and flood the countryside with the waters so released. In addition, although the men have a different belief that they keep to themselves, they tell the women that the bulaiŋ precipitates rain and storms, further traits of the Rainbow Serpent/dragon in many other regions.

Why should these variations stop anyone from believing that there is one Rainbow Serpent complex in Australia that arose from the same ultimate cause? Even a superficial consideration of the global distribution of this idea shows that it must have enormous antiquity, reaching back to the origins of fully modern brains, and the separation of ethnic groups over time invariably produces change in any aspect of culture. This is particularly obvious for language, and once the Rainbow Serpent existed as a cultural construct, it could be freely used in narrative contexts of all kinds, giving rise to elaborations of character, and the loss of original elements of the belief system or the acquisition of new ones.

Surely, the documentation of a worldwide and ancient belief that has been given in this book shows that what is profitless is the kind of reductionism Maddock espouses. What can be gained from treating a plethora of mythic figures, nearly all of which are connected with the rainbow, as though they have no connection with one another just because they are not identical? The Rainbow Serpent concept has been present in Australian aboriginal societies for countless generations. What may well have begun as a simple belief about the character of the natural environment became a mythic being with an imputed personality, allowing it to figure in a wealth of oral traditions over many millennia. To expect a belief system or mythic tradition of this kind to remain invariant as it is passed from one generation to the next is like expecting a language to show no change under the same conditions—something that is clearly counter to fact. At its worst, the reductionist approach to category membership would deny the existence of such obvious categories as birds. Should finches, penguins and cassowaries be denied membership in a common category just because they differ in size, coloration or ability to fly? Yet only birds have feathers, and few observers have any difficulty deciding what is or is not a bird. For mammals the problems become even more extreme. Biologists tell us that whales, bats, lions, and humans are members of a single order within the animal kingdom: the order Mammalia. This is far from obvious to the untrained observer, and in folk classifications cetaceans are commonly grouped with fish rather than with their proper kin. Yet only mammals have hair and mammary glands.

The truth is that a reductionist approach would contribute nothing toward understanding the evolution of the dragon idea. At bottom this approach assumes that variation may not coexist with relationship, an assumption that is contrary to everything we know about classification in science. A sparrow and a penguin are obviously very different, but that does not mean that we ignore their unifying features and treat them as members of completely different categories. If there is a valid category called ‘bird’ why would there be no valid category called ‘Rainbow Serpent’? Needless to say, the problems that reductionism creates carry over to many other cases, among them there is no human race, and most to the present point, there is no dragon.

But there IS a dragon, as many writers have recognized since at least Gould (1886), and although it varies in detail from one region of the Earth to another, it shares an overlapping set of traits that distinguish it from any other creature, real or imaginary. In this basic sense the concept of the dragon is as real as the concept of a bird or a human being. Borges (1978) was unduly pessimistic in his claim that “we are as ignorant of the meaning of the dragon as we are of the meaning of the universe.” The meaning of the universe may well elude us forever, but the story of the dragon has at last been told.

  • Collapse
  • Expand