Site map > Home page > CV > Misc > Life on earth

Life on earth: a cosmic origin?

This page contains the full text of the following article: Athel Cornish-Bowden and María Luz Cárdenas (2000) Life on earth: probability of a cosmic origin The Biochemist 22(2), 35–38, together with a brief précis of an article by Hoyle and Wickramasinghe and some additional notes added after publication.

Précis of the article by Hoyle and Wickramasinghe

Much of the discussion below should be intelligible without access to the article [1] that provoked it, especially as the views that Hoyle and Wickramasinghe express there are similar to ones they have written elsewhere. Nonetheless, it may be helpful to prefix this web page with a brief précis of their article.

They begin by contrasting real science, exemplified by the prediction of the solar eclipse of 11th August 1999, with technology-science (genetically modified crops) and paradigm-science (belief in speculation because others believe in it), and they argue that some aspects of modern biology, those that relate to the origin of life, belong in the third category, for which they also use the term bee-dancing.They point out that the earth only became inhabitable about 3.8 billion years ago, and they argue that the time available before the appearance of the oldest stromatolites about 3.6 billion years ago leaves too narrow a window for life to have originated on earth.

They then examine the probability that an enzyme consisting of 300 residues could be formed by random shuffling of residues, and calculate a value of 10–250, which becomes 10–500000 if one takes account of the need for 2000 different enzymes in a bacterial cell. Comparing this calculation with the total of 1079 atoms in the observable universe, they conclude that life must be a cosmological phenomenon. They argue that once an unlikely event has produced a viable cell somewhere in the universe the enormous multiplicative power of replication will produce enormous quantities of living material very fast (e.g. a doubling time of 2–3 h implies that one cell can generate the mass of a cluster of galaxies in 20 days).

The lengthy central portion of the article deals with the argument that interstellar clouds consist largely of dried bacteria, but as this part of the article is barely mentioned in the response below it will not be considered in detail here.

In the later parts of their article Hoyle and Wickramasinghe quote Lord Kelvin and Hermann von Helmholtz, and argue that there is currently a trend towards the views of these 19th century opponents of the idea that life originated in a warm little pond.

Life on earth: a cosmic origin?

In their recent article [1] Fred Hoyle and Chandra Wickramasinghe assess the current evidence in favour of their view that life cannot have originated on earth and must, therefore, have arrived from somewhere else. Unfortunately their argument is more sarcastic than helpful in many places, and they do not use their expertise as astronomers to illuminate several points that might be obscure to a biochemical audience.

They begin by contrasting real science with the ideas of biologists about the origin of life and its subsequent evolution. Using the prediction of the solar eclipse of August 1999 as an example of real science reveals rather a bleak view of the progress of science over the past several centuries: this prediction did not require understanding of gravitation or use of the apparatus of modern astronomy, and was made by Nostradamus [2] (and doubtless by other astrologers) about a century before Newton, and around four centuries before Einstein. Thus although it is a poor example of real science it is an excellent example of what they call technology-science, i.e. the use of rules, equations, etc., to make correct predictions without requiring understanding of the underlying causes.

Hoyle and Wickramasinghe disparage biological thinking by making numerous references to something they call bee-dancing. They define this rather loosely as believing in speculations that others believe in, but the term comes from von Frisch’s discovery that honey bees use body movements to communicate the location of nectar-bearing plants to other bees. Their claim that there is no evidence for it would be difficult to sustain. At one time many biologists were sceptical about bee-dancing, and a great deal of evidence was gathered to counter their objections. A dispassionate observer may certainly be sceptical after reading just a description of bee dancing, for example the account given by Richard Dawkins on pp. 84–87 of River out of Eden, [3] but to remain sceptical after reading about Gould’s study of sighted and blind bees in a lighted hive [4] on pp. 99–102 of the same book [3] leaves one with an obligation of providing an alternative explanation, something far more difficult and complicated than just accepting that bee-dancing is a reality.

The theory of the origin of life favoured by Hoyle and Wickramasinghe depends heavily on their calculation of the probability that an enzyme could be produced by shuffling amino acids is no better than one in 106900. There are many objections to this sort of calculation, but one that we have not seen mentioned previously is that it takes no account of actual observations of the catalytic properties of random co-polymers of amino acids. For example, random co-polymers of glutamate and phenylalanine imitate the bacteriolytic activity of lysozyme quite well, with about 3% of the activity of the enzyme from hen egg white on a weight for weight basis [5], and there are similar observations on other systems from other groups [6].

The probability calculation seems to be based instead on an assumption that the amount of variability tolerated in mutations away from existing protein structures provides a direct measure of the probability that other sequences of amino acids exist that define other conceivable enzymes that catalyse the same reaction. A bee living on a small island might well conclude from its forays of no more than a few kilometres in any direction that life was not possible anywhere else, but it would be wrong. The fact that all inhabitable locations that it could find are at coordinates very similar to those of its starting point just reflects how it got there and the fact that it cannot make flights long enough to find any of the others. Likewise the fact that in most cases all of the known enzymes that catalyse a given reaction have more or less the same structure does not mean that no other structures exist; it just means that once a solution to a problem has been found in evolution any subsequent solutions are likely to be derived from the first one, regardless of how many others may exist. It is certainly not the case that only a unique structure is capable of catalysing a given reaction and for some enzymes alternative structures are well known. For carbonic anhydrase, for example, there are three classes of protein with no discernable homology that catalyse the same reaction [7]; superoxide dismutase [8] offers another example that has been known for a long time. Given the large number of gene products of unidentified function that are appearing in genome sequencing there may well be many more examples than we are aware of at present.

Let us now ignore the experimental evidence and accept that 1 in 106900 is a reasonable estimate of the chance of assembling a catalyst by tinkering with amino acid sequences. We can accept that this is not a bet one would advise a friend to take, but we still know that the origin of life occurred somewhere, and why should that somewhere not be here? At least that would avoid the need to explain how life arrived here after originating somewhere else, and at least it would mean that it started in a place where it had a chance of surviving. The fact that the universe is larger than the earth is barely relevant, because the ratio of masses (or even of volumes) is trivial if we are trying to explain numbers like 106900. Even a small lake, say 1 km2 in area with an average depth of 10 m, contains around 1036 atoms of H and O, so the 1079 atoms in the observable universe quoted by Hoyle and Wickramasinghe only helps by a factor of about 1043, hardly worth bothering about in this context. They themselves ignore a factor of 256! in their calculation, i.e. about 10507, and would presumably ignore one of 1043.

A larger lake will contain many more atoms; Loch Ness, for example, contains around 1040, and if we consider all the water on the planet we arrive at more than 1045 atoms; if this is a negligible quantity then the number of 1079 atoms in the whole observable universe is scarcely less so. To get to 106900 we need not our friendly neighbourhood universe but something more than 106850 times as big.

As we are not accustomed to thinking in terms of of numbers as large as this, and as the universe appears so huge in relation to everyday objects, it is easy to exaggerate how large it is in relation to the earth. Actually the total of 1079 atoms, mostly H atoms, in the observable universe corresponds to a mass of the order of 1052 kg, about 1027 times larger than the earth mass of about 1025 kg. By contrast, an E. coli cell has a volume of the order of 10–15 litre, and hence a mass (taking it to have about the density of water) of the order of 10–15 kg. Thus the universe is actually far smaller on the scale of the earth (about 1027 earth masses) than the earth is on the scale of a bacterial cell (about 1040 E. coli masses). Put somewhat differently, the universe is about the same size in relation to the sun as the earth is in relation to a man (see Fig. 1). By contrast, the universe implied by Hoyle and Wickramasinghe’s calculation is not merely larger than the known universe, but vastly larger. Note, moreover, that this will just allow the chance appearance of one enzyme: to have a fair chance (according to their calculation) of forming a minimal set of enzymes the universe would have to be vastly larger even than this.

Sizes of various objects

Figure 1. The masses of various entities from a hydrogen atom to the observable universe, shown on a logarithmic scale. Notice that the earth is much larger in relation to a bacterial cell than the universe is in relation to the earth. The right-hand part of the figure shows how much the scale would have to be compressed if the universe were large enough for the chance appearance of an enzyme molecule were to be a likely event according to Hoyle and Wickramasinghe’s method of calculation.

To overcome the difficulty that the observable universe is not big enough to get around their probability arguments, Hoyle and Wickramasinghe introduce the concept of the

spatially infinite universe, a universe that ranges far beyond the largest telescopes.

Fair enough: if such a universe exists then it is certain that new origins of life are occurring with an infinite frequency. But does it exist? Here the astronomical expertise of the authors could have helped biochemical readers quite a lot, but instead of giving some indication of the experimental evidence for their idea they prefer to devote the rest of the paragraph to explaining exponential growth in terms that are surely superfluous for the audience for which their article was written.

If Hoyle and Wickramasinghe are right about this, they find themselves with the same problem as Aristarchos of Samos around 250 BC. He failed to convince his contemporaries of the merits of his heliocentric theory, but this was not, as often supposed, because they lacked the ability to draw the right conclusions from the data, but because the data available to them argued strongly the other way. [9] The total absence of a stellar parallax, which was not in fact observed until more than 2000 years after Aristarchos’s death, implied that the sphere of the fixed stars must be not merely large, but absurdly large. Scientists such as Archimedes could contemplate with equanimity the idea that the distance of the stars from the earth might be large enough for all observers from India to Gibraltar to see them from the same angle, but they could not accept that they were so far away that the earth’s entire orbit around the sun was just a point by comparison.

Suppose, however, we accept that the universe is more than 106850 times as big as we thought. How does it help? Even if it makes it certain that life has appeared many times in many places, it does not help us at all in explaining the particular manifestation of life that we know about, because it does not explain how life arrived on earth from wherever it originated. The speculation in the rest of the article about bacteria in space is irrelevant unless the authors explain how they passed from beyond the boundary of the observable universe to the regions of interstellar space where they are said to be observed. Again, the authors’ expertise as cosmologists would have been useful, as they certainly understand these matters better than the average reader of The Biochemist.

Unless this point is explained it would seem that what is required is not a spatially infinite universe but a temporally infinite universe, because if infinite time is available even the most improbable event will have occurred an infinite number of times (as long as its probability is great than zero). If Hoyle and Wickramasinghe believe in an infinitely old universe they should say so, and give some indication of why they think the big-bang model is wrong. Until they do, those of us who are not experts will probably prefer to stay with the big bang, and to believe that the universe is two or three times as old as the earth. [10, 11] It will take much more than a factor of two or three to convince most scientists studying the origin of life to search for extraterrestrial origins.

In their discussion of their Figure 5, Hoyle and Wickramasinghe lay great emphasis on their claim that the curve was a prediction. However, on their own description, the experiments were made in 1981, but they were predicted four years later, in a doctoral thesis of 1985. As Jay Bailey [12] has remarked in a different context,

Quotation marks are included... because some authors of models unfortunately carelessly substitute prediction for fit or simulation, e.g. the model successfully predicts the experimental data in a situation in which the experiments were done first.

In the present context, with its insistence that it was not something wheeled out after the event, the emphasis on prediction seems worse than just careless, unless there was a typographical error in the dates given.

According to Hoyle and Wickramasinghe the existence of extremophiles in various places on earth is in the course of convincing many scientists that the origin of life must have been external to the earth. They do not comment on the obvious non sequitur in this assertion, nor do they specify the scientists they have in mind or give any references to support it [13]. There is nothing to suggest that researchers who actively study extremophiles are among those being convinced. For example, a phylogenetic analysis of anaerobic thermophilic bacteria [14] led its authors to conclude that at present the data point towards an independent adaptation of mesophilic organisms to thermophilic environments, similarly, a review of Thiobacillus ferrooxidans [15], a strict acidophile, refers to it as a typical aerobic bacterium [that] is capable of anaerobic growth ... under certain conditions. This organism grows very slowly compared with more familiar bacteria, with a doubling time of around 10 hours, but its leisurely style of life is more easily explained in terms of very unfavourable equilibria in its redox chemistry and a lack of competitors for the same ecological niche than by regarding it as an extraterrestrial visitor. Accounts of the chemistry of the deep-ocean vents [16] likewise present a view of the origin of life that is a natural development from the work of Oparin [17], and of Darwin before him, and not from anything radically different.

Presumably Hoyle and Wickramasinghe are referring to physicists, philosophers, theologians and so forth, as they go on to claim there is currently a trend towards the view of Lord Kelvin (William Thomson) that

we all confidently believe that there are ... many worlds of life besides our own.

This is, of course, the same Lord Kelvin who was equally confident that the earth could not be more than 400 million years old, and might even be as young as 25 million years old [18]. Hoyle and Wickramasinghe’s mention of rocks in Greenland with an age of 4.1 billion years would seem to indicate that their own idea of the age of the earth is in line with modern thinking, but they still cling to the belief that life cannot have started on earth despite the fact that Kelvin’s reason for asserting this is now known to be erroneous.

Athel Cornish-Bowden is Director of Research and María Luz Cárdenas is Research Associate, both in the laboratory Bioénergétique et Ingéniérie des Protéines of the Institut Fédératif Biologie Structurale et Microbiologie, CNRS, Marseilles, France; email,

Notes added to this page after publication of the article

R. A. Fisher, one of the architects of statistical theory, and as competent a calculator of probabilities as there has ever been, was very well aware of the relevance of probability arguments to the theory of evolution by natural selection, but these did not prevent him from becoming one of the leading Darwinians of his time. A recent article by A. W. F. Edwards in Genetics 154, 1419–1426 (2000) entitled The Genetical Theory of Natural Selection is devoted to discussion of Fisher’s book of the same name published in 1958 (2nd edn.) by Dover, New York. In this article Edwards makes numerous interesting points, among which he says:

He emphasizes the importance of not being carried away by improbabilities viewed after the event anyway

and goes on to quote the following passage from Fisher (Retrospect of criticisms of the theory of natural selection, pp. 84–98 in Evolution as a Process (ed. J. Huxley, A. C. Hardy and E. B. Ford, Allen & Unwin, London, 1954):

Consideration of the conditions prevailing in bisexual organisms shows that ... the chance of an organism leaving at least one offspring of his own sex has a calculable value of about 5/8. Let the reader imagine that this simple condition were true of his own species, and attempt to calculate the prior probability that a hundred generations of his ancestry in the direct male line should each have left at least one son. The odds against such a contingency as it would have appeared to his hundredth ancestor (about the time of King Solomon) would require for their expression forty-four figures of the decimal notation; yet this improbable event has certainly happened.

Actually (5/8)100 is about 4 × 10–21 (though (3/8)100 is about 2.5 × 10–43, in better agreement with Fisher’s answer), so maybe I have misunderstood how to do the calculation, or the quotation is not exact, or he made an error, but no matter, the qualitative point that he is making is quite fair.

There is, of course, no strong reason to start at the time of King Solomon. Suppose that we date the beginning of the human species at 3.2 million years ago, and suppose, at about 30 years per generation, that this means that more than 100000 generations have passed since a single male proto-human living at the time of Lucy (though not necessarily a member of the same species, Australopithecus afarensis) might have contemplated the likelihood that he would eventually have a descendant through the direct male line who would become a well known cosmologist named Fred Hoyle. In this case we would have a probability of (5/8)100000, or much less than 10–20000. Yet, to quote Fisher again, this improbable event has certainly happened.