Peripatus Home Page  pix1Black.gif (807 bytes)  Paleontology Page >> Replaying Gould Updated: 26 Apr 2004 

Replaying Gould


In 1989, Stephen Jay Gould published probably the most successful book on a paleontological subject since Darwin’s Origin: Wonderful Life. Although ostensibly an historical account of the discovery and interpretation of the Burgess Shale fauna, Gould uses the book to advance a number of bold claims about the very nature of evolution, and the science of paleontology. Thus, despite being a popular account nominally aimed at a lay audience, the ideas embodied in Wonderful Life have initiated many professional discussions also.

In subsequent publications, Gould retreated somewhat from a few of the less tenable positions he held in 1989, though he has always presented his reframed arguments as simply clarifying his earlier views; not abandoning them. However, even the blunted versions (contributions from 1991 and 1993 are considered here) are worth critical scrutiny, and our conclusion is that a good dose of scepticism would not be misplaced.

Keywords: evolution, Burgess Shale, Gould, disparity, phylogeny, Wonderful Life, Cambrian, arthropods, contingency, ‘tree of life’


Any popular discussion of Cambrian evolution in general, and of the Burgess Shale in particular, conducted since about 1990 has been doomed to fall under the long shadow of Stephen Jay Gould and his immensely successful book, Wonderful Life (Gould 1989).

Besides historical and taxonomic accounts of the Burgess fauna, this book promotes three principal messages:

  • That Walcott committed a ‘cardinal error’ by ‘shoehorning’ the Burgess problematica into existing phyla and classes.
  • That an ‘iconography’ depicting evolutionary lineages as a ladder or a cone of increasing diversity has developed,  persists to the present day, and, moreover, that a slavish obeisance to this paradigm of evolutionary progress has blinded everyone but Gould to the true nature of evolution.
  • That if it were possible to ‘replay the tape’ of evolution, the outcome would almost certainly be very different, both in detail and in general. Specifically, nothing like human consciousness would be likely to emerge.

Increasingly, these ideas are being rejected by professional evolutionists ("deeply muddled" in the words of Richard Dawkins, Dawkins 2003, p. 203) though the commercial success of the book – and Gould’s failure to issue any sort of retraction before his death in 2002 – guarantees that, to this day, generations of lay readers continue to soak this stuff up as gospel.


Related Topics

Further Reading

Related Pages

Other Web Sites

  • xxx

The Shoe Horn

Gould pillories Walcott for his pragmatic inclusion of the Burgess Shale taxa into broadly defined phyla – ‘shoehorning’ in Gould’s terminology. But, here, Gould (also Mayr 2001, see p. 209) has forgotten a principle which is so fundamental it has become hackneyed: man classifies; nature does not. The only taxonomic rank possessing any degree of objective ‘reality’ is the species. To approximate disparity by counting phyla is to embark upon an extremely suspect enterprise.

First, what is a phylum? What are the characteristics we use to define phyla? (Here we are really using the word phylum as a proxy for potentially several of the highest-level taxa.)

  • Phyla are not ‘naturally’ defined entities; they are man-made categories of convenience, and necessarily arbitrary.
Although it would be unfair to say that the concept of a phylum is meaningless [® sidebar], it is certainly reasonable to remember that any debate over how many phyla are present at a certain time probably carries some heavy semantic baggage.

The key point here, though, is that phyla are phyletic concepts, and it is phylogeny which Wonderful Life focuses upon – both in respect of the shoehorn argument and the tree of life argument which we discuss next. Gould attempts to rewrite history when he later (1991, 1993) claims to have been talking about morphology all along.

Even more bizarre is McShea’s (1993, p. 399) apologist manifesto, in which he claims "[Gould argues] that decreasing developmental flexibility, or increasing developmental entrenchment, may be a basic feature of the evolutionary process." These ideas were initially developed by people such as Riedl and Wimsatt (Riedl 1978; Wimsatt 1986; for more recent views, also see Arthur 1997; Nielsen 2001), and their case is very compelling indeed. However, neither of these worthies are named among the references in WL. Nor is the word ‘development’ to be found in the index (or conspicuously in the text, for that matter!) It is not that McShea makes a poor case – on the contrary! – it is just that these arguments were never articulated by Gould in 1989, and McShea’s claim that he did is simply untrue. Developmental entrenchment (= internal coadaptation) is the position that Gould retreated to in 1991 (p. 417) as it began to become obvious that the original phyletic arguments put forward in Wonderful Life are nonsense.

For example, Mark Ridley's (1990, p.12) comment that in a million years’ time a crustacean may emerge with one or three pairs of pre-oral antennae is just silly. Crustacea have possessed exactly two pairs since the Cambrian, and are unlikely to change now. The pattern of head segmentation has been a stable characteristic, achieved independently, in each of the high level arthropod taxa for over 500 million years: that obviously means something.
Gould’s ‘shoehorn’ argument is focussed upon phylogeny, and specifically upon Walcott’s "cardinal error" of including the Burgess problematica in modern phyla. Graham Budd has summed all this up better than I could: "No one would dispute that these fossils are problematic, in the sense that they are difficult to understand. However, that methodological difficulty should not be confused with the possibility that these fossils have only remote affinities with all living groups" (Budd 1997, p. 125). This observation is echoed by Richard Fortey (1997, p. 111): "One could not have predicted Anomalocaris' existence, of course, without the wonderful insights provided by special fossil preservations, but no purpose is served by assigning it to a phylum unknown. One may still marvel at the fecundity of nature without making wild assertions about every fossil belonging to a different world."

The Tree of Life

The Iconography Problem

Here we see Gould at his least consistent. He speaks disparagingly of a "cone of increasing diversity" yet any fool can see that diversity has indeed increased over time [® sidebar].

On the other hand, he is careful to distinguish between diversity and disparity (and later, in 1991, he relies upon this distinction to support the notion that he had been pitching a phenetic argument all along). But, despite this, in 1989, both his commentary and his examples are clearly phylogenetic.

"… the number of living animal phyla … is about 33. Of these, approximatly 20 [are] large and abundant enough to leave fossils of the kind preserved in beds of the Burgess Shale type. The number of Cambrian phyla identified with confidence remains at 11. To the best of our knowledge, no phylum has yet gone extinct. … The greatest rise in general has occurred during the past 100 million years" (Wilson 2001, p. 182). Note that Wilson is not here 'counting phyla' in the manner we have criticised above; he is observing the disappearance of known modern groups as one looks back further in time.
Gould explains to the reader that people’s views of evolution are clouded by a conceptual model that represents evolutionary relationships between living things variously as a continuously branching tree (a "cone of increasing diversity" where diversity of the lineage always increases) or a ladder.

His point is that, during certain times, diversity might sometimes decrease, but that we have been blind to that possibility.

One example Gould chooses to illustrate his point is the evolution of horses, about which he notes:

"To be sure, an unbroken evolutionary connection does link Hyracotherium ... to modern Equus. And, yes again, modern horses are bigger, with fewer toes and higher crowned teeth. But HyracotheriumEquus is not a ladder, or even a central lineage. This sequence is but one labyrinthine pathway among thousands on a complex bush" (Gould 1989, p. 36).

By the way, here is further confirmation, if any is needed, that in 1989 Gould is concerned with phylogeny.

Gould implies that evolutionists fail to recognise this "labyrinth" and perceive the evolution of horses as the simple, unidirectional modification of a few characteristics in a "progression" from one single species to another.

Were he still alive, I’d direct his attention to my favourite childhood paleontology book, the inspirational though now sadly dated Time, Life and Man, written in 1959 by R. A. Stirton, specifically to p. 466, where the horse lineage is depicted graphically – not as a simple-minded ladder, but in all of its ‘labyrinthine’ complexity (fig. 1).

<expand...> Note how disparity (if it is implied at all) appears maximal in the Pliocene.

I suggest that this iconography of the cone or ladder is a straw-man which Gould has introduced, simply in order to shoot it down. It never existed outside of his own mind (well, not since the 1950s, anyway!)

cgmEquidae.gif (283780 bytes)

Fig. 1: ... Note how disparity (if it is implied at all) appears maximal in the Pliocene.

In fact, as Gould knew full well, though many of his lay-readers might not, evolutionary trees (phylogenies) are actually represented by cladograms (sometimes known as dendrograms), a very simple example of which is reproduced to the right. The only ‘iconography’ inherent in such a construct is a belief in descent with modification from common ancestry – in short: evolution. There is no hidden agenda in the final shape; there is no hijacked thinking.
cgmPrimer1.gif (2056 bytes)

Fig. 2: A simple cladogram; the illustrative device most commonly employed to represent the relationships between organisms.

How Representative are Gould’s Examples?

What is Gould’s thesis based on, taxonomically? How representative of ‘life’ are the examples he cites from the Burgess Shale? The bulk of WL is about animals that, today, are considered to be arthropods, plain and simple. Even the famous Hallucigenia appears fairly conventional when stood on his feet (WL depicts him upside-down) and placed alongside Microdictyon et al. (fig. 3).
<fig of Microdictyon – after Hou et al.>

Fig. 3: ...

But let us consider the arthropods in perspective. As early as the late 1970s, work by Carl Woese and George Fox had revealed that our old ideas about the diversity of life were quite incomplete. By studying ribosomal RNA common to all life, Woese and Fox were able to show that the living world was profoundly divided into three vast "super-kingdoms" now named the Bacteria, Archaea and Eukarya (fig. 4).

There are only about 6,000 described species of 'prokaryotes' (Bacteria plus Archaea) – we are definitely talking about disparity rather than diversity, here!

It is the Eukarya which contains the plants and animals with which we have daily familiarity ... but not as the major constituents. Among the microsporidia, ciliates, slime moulds and fungi, the animal kingdom is but one branch, and the arthropod clade – great as it is – but one twig on the end of that. Pretty small fry from which to be extrapolating to the whole of nature!

But there is a more fundamental question we can ask: Why is WL primarily about arthropods?

The answer is because the Cambrian "bush" metaphor is essentially an arthropod phenomenon. Gould could not have made the same case if he had chosen to discuss, for example, brachiopods, molluscs or chordates: These groups achieved their peaks of diversity respectively in the Devonian, Cretaceous and Carboniferous.

In writing about the Cambrian explosion, Gould makes a great fuss of something that began about 530 million years ago and lasted for maybe 10 or 20 million years.

(These ages are different from those given in WL: At the time Gould wrote, there were far fewer good radiometric dates available than is the case today, and also the Cambrian was generally considered to have begun with rocks correlating to a unit called the Tommotian. These errors are not Gould’s fault: any book written around 1989 would contain the same errors.)

Gould outright claims that the majority of disparity in the living world arose during this time:

"The Burgess Shale includes a range of disparity of anatomical design never again equalled, and not matched today by all the creatures in all the world’s oceans." — Gould 1989, p. 208

(And also Gould 1991, p. 419: "... and for life in general...")

Yet we have already seen that the most fundamental differences – the deepest splits in the tree of life – are those between the super-kingdoms which arose at least 1,200 million years ago.

The real disparity of life was established more than twice as long ago as the Cambrian explosion, a fact which Gould finesses.

cgmrRNA.gif (8205 bytes)

Fig 4: Woese and Fox were able to show that the living world was profoundly divided into three vast "super-kingdoms" now named the Bacteria, Archaea and Eukarya.

Even if we confine ourselves to considering Phanerozoic animals, Gould is wrong. If we play Gould at his own game and approximate disparity by counting orders and higher taxonomic groupings (order and greater rank) within phyla, we find that:
  • Brachiopoda achieve their maximum ‘disparity’ in the Devonian and Carboniferous periods;
  • Mollusca in the Cretaceous;
  • Echinodermata in the Ordovician;
  • Hemichordata (if this is in fact a monophyletic group) in the Ordovician; and
  • Chordata in the Carboniferous and Permian.

(Data after Rich et al. 1996.)

And, finally, even restricting ourselves to arthropods, Gould’s interpretations are by no means universal:

"Morphometric and phylogenetic studies have shown that the supposedly ‘bizarre’ Burgess Shale-type arthropods fall into a phylogenetic scheme that gives no support to the idea that they are outliers in morphospace awaiting the grim reaper of contingent extinction" (Conway Morris 2000, p. 4428).

"This work was done jointly with Derek Briggs, who was so familiar with the details of the Burgess arthropods [Briggs & Fortey 1989; Briggs et al. 1992]. Instead of emphasizing their peculiarities, we were looking for the significant similarities they shared with one another. … If some of the ‘oddball’ animals could be accommodated convincingly into the tree, then it would mean that their distinctiveness might have been over-emphasized by the explosionists. …

"What surprised us was how easily we could produce a tree relating nearly all of the Burgess Shale arthropods" (Fortey 2000, pp. 126-127).

A growing body of research supports the view that the "roots" of the modern groups lie very deep in the Precambrian. Far from resembling a bush, if we are to seek arboreal metaphors for our ancestry, we might do well to consider a poplar!
<cladogram from Heckman et al.>

Fig. 5: ...

Changing Tunes

I contend that Gould significantly altered his position between 1989 and 1991. Compare the following:

1989: "But HyracotheriumEquus is not a ladder, or even a central lineage. This sequence is but one labyrinthine pathway among thousands on a complex bush." — p. 36

1991: "[Rooting of a cladogram] does not speak to the phenetic issue of broader or diminished disparity." — p. 415

This is his first attempt to rewrite history. In 1989 Gould's arguments are overwhelmingly phylogenetic; by 1991 he is not only blowing a phenetic trumpet, he is claiming that had been his instrument all along. It simply isn't true.

An interesting by-play in the 1991 paper is Gould’s attempt to have his arguments both ways: strongly divorcing morphology and phylogeny with the statement "these are not the same, either logically or empirically" on page 420, while at the same time crucially dependent on an argument acknowledging that fundamental ("architecturally deep") morphological properties are more phylogenetically significant than "paint and panelling" which are "superficial and easy to alter piecemeal" (p. 418). Of course both statements are literally correct, but Gould employs them to make mutually incompatible arguments.
He seems to retreat even further in 1993, almost to the point of reducing his claim to an ecological argument, with "the two proper meanings of biological relationship [are] the genealogical, or branching order, [and] the legitimate component of biological role (based on homologous, if pleisiomorphic, characters" (Gould 1993, p. 522).
The legitimate question which remains unanswered – and mostly ignored – is why no new body plans have arisen since the Cambrian. For me, this is a far more interesting and ultimately illuminating question than anything about the initiation or timing of the Cambrian explosion. I believe that Riedl, Wimsatt, Arthur, and so on, have probably captured the essence of the answer in their various ‘entrenchment’ concepts. Organisms have to be developmentally simple in order for mutations which express themselves early in ontogeny to survive. Post-Cambrian (or, more probably, post-Vendian) metazoans are simply not that developmentally simple any more, regardless of their 'grade' of organisation.

Replaying the Tape

Gould’s replayed tape metaphor refers to a hypothetical re-run of Earthly evolution, and posits:

  • If it were possible to ‘replay the tape,’ the outcome would almost certainly be very different, both in detail and in general.
  • Specifically, nothing like human consciousness would be likely to emerge.
In this, Gould seems to have been influenced by the idea of "critical sensitivity to initial conditions" (borrowed from chaos theory) which is invoked to explain phenomena which are difficult to predict, such as the weather.

Some of the main sorts of things to consider might include:

  • Is the apparent ‘progress’ in the evolution of life on Earth real or illusory?
  • How much real morphological ‘freedom’ is there, or, conversely, how ‘full’ is morphospace?
  • … lots of others …
  • To what extent are organisms really the products of their genes?
Think of a pencil balanced on its point. It will fall over, though in exactly which direction will be unpredictable, because the slightest vibration or air current will disproportionately effect the outcome.

Is the Apparent ‘Progress’ of Evolution Real or Illusory?

The commonly observed evolutionary trends which might be considered ‘progressive’ are:

  • Size
  • Complexity
  • Diversity

"Nowhere to go but up" versus "driven" (Carroll 2001)

How Much Real Morphological ‘Freedom’ is There?

What kinds of constraints might there be?

  • Natural selection
  • Biomechanics
  • Genetics and development
An alternative angle from which to consider the question of morphological ‘freedom’ is to consider how fully exploited is the total range of morphological possibilities: How ‘full’ is morphospace?

"To simplify the analysis of [a] large number of cases, we compare the skeletons used by different groups of organisms in a matrix of all possible pairs of character states.

"Thomas and Reif have shown that more than half of these character pairings are abundantly used, generally by animals in several phyla; two-thirds of these pairings are common. Only logically or functionally implausible combinations are unrecognised among living and extinct animals" (Thomas et al. 2000).

This question could be rephrased in terms of convergence, which we know to be extremely common. for example:

"[M]ulticellularity evolved independently many times and in all three domains of life" (Carroll 2001, p. 1103).

"Photosensitive, eye-like organs have developed in the animal series independently at least 40 times, and all the steps from a light-sensitive spot to the elaborate eyes of vertebrates, cephalopods, and insects are still found in living species of various taxa" (Mayr 2001, p. 205).

Bacteria, 'protists' and plants all respond to light in some way. Ernst Mayr claims that separate animal clades independently evolved vision 40 times (Mayr 2001, pp. 113, 156 and 205). I wouldn't know about 40 times, but it has certainly been more than a few. Quite clearly, vagile animals were always "determined" to see, and nothing was going to stop them. Since eyes are expensive to maintain (note how quickly they are lost in those creatures which blunder into deep caves) their independent development in almost all vagile clades cannot be coincidence.

We all know about moths and their acute "sense of smell" (actually, I think they only are receptive to pheromones; not to smells in general) but even moths have eyes – presumably to avoid bumping into things. Sight provides more information, to moths and everything else, than smell ever can. There is no alternative to vision: the physical relationship between wavelength and spatial resolution is such that no other energetically viable mechanism is available to Earthly life forms.

In the immortal words of Mr. Scott, "Ye canna break the laws of physics."

Another example is very well described in Clack 2002 (although, ironically, various comments throughout the text suggest that Clack herself might actually be fairly sympathetic to Gould). The stapes is now a small bone which has been co-opted as a hearing ossicle in both diapsids (lizards etc.) and synapsids (mammals etc.) Originally, however, the stapes was an important structural bone, helping to tie the braincase and the skull roof together. This remained the case some time after the divergence of diapsids and synapsids and, in fact, quite different bones were co-opted to to take over the structural role of the stapes when the two groups independently became autostylic (Clack 2002, p. 288 and 298). Thus, in two quite independently (and differently) evolving groups, the stapes was separately freed to find its way into the middle ear and become a hearing ossicle: a remarkable convergence. In fact, it seems likely that hearing evolved multiple times among tetrapods (Clack 2002, p. 300).
Once again, though, it is important not to allow Gould to polarise one from common sense. Convergence is noteworthy mainly because its absence is not. If any portion of history or pre-history were to be re-run, of course it would be expected to run differently in some ways. This is a completely unremarkable thought and nobody believes any different. As before, what we are objecting to is the meal Gould makes of the commonplace, and the intellectual claim he makes on an understanding that everyone else had held for years.


Stephen Jay Gould gained the reputation as an iconoclast early in his career, upon the publication of the punctuated equilibrium paper (Eldridge & Gould 1972) - a sanctamonious and ultimately banal piece of work, if ever there was one. Subsequently, iconoclasm became a hallmark of his career, occasionally getting the better of prudence.

Wonderful Life, though in many respects a brilliant piece of writing, became perhaps the greatest casualty of this prediliction. He painted himself into a corner - when I'm feeling at my least charitable, I think made a complete ass of himself (not that much of the subsequent commentary was of a conspicuously higher standard) - and probably realised it.

I wouldn’t be the first to suspect Gould of hubris, and one supposes that the book’s enormous commercial success among the lay readership made it nearly impossible to abandon wholesale even his least tenable positions. Whatever his motives, instead of any form of retraction, he produced a series of articles (notably 1991 and 1993) reframing his arguments - and, at the same time, rewriting history. In this latter endeavour, he seems to have had at least one willing accomplice, and probably many.


Arthur, Wallace 1997: The Origin of Animal Body Plans. Cambridge, 338 pp.

Briggs, D.E.G.; Fortey, R.A. 1989: The Early Radiation and Relationships of the Major Arthropod Groups. Science 246: 241-243.

Briggs, D.E.G.; Fortey, R.A.; Wills, M.A. 1992: Morphological Disparity in the Cambrian. Science 256: 1670-1673.

Budd, G.E. 1997: Stem Group Arthropods from the Lower Cambrian Sirius Passet Fauna of North Greenland. In Fortey, R.A.; Thomas R.H. (eds.) 1997: Arthropod Relationships. Systematics Association Special Volume Series 55.

Carroll 2001

Clack 2002

Conway Morris, Simon 2000: The Cambrian "Explosion": Slow-Fuse or Megatonnage? PNAS 97 (9): 4426-4429.

Dawkins, Richard 2003: A Devil's Chaplain. Weidenfeld & Nicolson, 264 pp.

Eldredge, N.; Gould, S.J. 1972: Punctuated Equilibria: An Alternative to Phyletic Gradualism. In Schopf, T.J.M. (ed.) 1972: Models of Paleobiology, Freeman, Cooper and Co., pp. 82-115.

Fortey, Richard 1997: Life - An Unauthorised Biography. HarperCollins, 399 pp.

Fortey, Richard 2000: Trilobite! Eyewitness to Evolution. Harper Collins, 269 pp.

Heckman, Daniel S.; Geiser, David M.; Eidell, Brooke R.; Stauffer, Rebecca L.; Kardos, Natalie L.; Hedges, S. Blair 2001: Molecular Evidence for the Early Colonisation of Land by Fungi and Plants. Science 293: 1129-1133.

Gould, S.J. 1989: Wonderful Life

Gould, S.J. 1991: The Disparity of the Burgess Shale Arthropod Fauna and the Limits of Cladistic Analysis: Why We Must Strive to Quantify Morphospace. Paleobiology 17 (4): 411-423.

Gould, S.J. 1993: A Reply to Ridley

Mayr, Ernst 2001: What Evolution Is. Weidenfeld & Nicolson. 318 pp.

McShea, D.W. 1993: Arguments and Tests

Nielsen, Claus 2001: Animal Evolution. Second ed. Oxford University Press. 563 pp.

Rich, Patricia Vickers; Rich, Thomas Hewitt; Fenton, Mildred Adams; Fenton, Carroll Lane 1996: The Fossil Book. Dover.

Ridley, M. 1990: Dreadful Beasts. The London Review of Books, 28 June 1990, pp. 11-12.

Ridley, M. 1993: Analysis of the Burgess Shale

Riedl, R. 1978: Order in Living Organisms: A Systems Analysis of Evolution. Wiley.

Thomas, R.D.K.; Shearman, Rebecca M.; Stewart, Graham W. 2000: Evolutionary Exploitation of Design Options by the First Animals with Hard Skeletons. Science 288: 1239-1242.

Wilson, Edward O. 2001: The Diversity of Life. Penguin, 406 pp.

Wimsatt, W.C. 1986: Developmental Constraints, Generative Entrenchment, and the Innate-Acquired Distinction. In Bechtel, W. (ed.) 1986: Integrating Scientific Disciplines. Martinus-Nijhoff, Dordrecht.

 Peripatus Home Page  pix1Black.gif (807 bytes)  Paleontology Page >> Replaying Gould Contact me.