A single macromutational leap from ground-dwelling ancestral shrew to flying, echolocating bat is ruled out just as safely as we can rule out luck when a conjuror successfully guesses the complete order of a shuffled pack of cards. Luck is not literally impossible in either case. But no good scientist would advance such prodigious luck as an explanation. The card-guessing feat has to be a trick we've all seen tricks that appear just as baffling to the uninitiated. Nature does not set out to fool us, as a conjuror does. But we can still rule out luck, and it was the genius of Darwin to rumble nature's sleight of hand. The echo-ranging bat is the result of an inching series of minor improvements, each adding c.u.mulatively to its predecessors as it propels the evolutionary trend on in the same direction. That is progress, by definition. The argument applies to all complex biological objects that project the illusion of design and are therefore statistically improbable in a specified direction. All must have evolved progressively.

The returning host, now unabashedly sensitive to major themes in evolution, notes progress as one of them. But progress of this kind is not a uniform, inexorable trend from the start of evolution all the way to the present. Rather, to take up the initial quotation from Mark Twain on history, it rhymes. We notice an episode of progress during the course of an arms race. But that particular arms race comes to an end. Perhaps one side is driven extinct by the other. Or both sides go extinct, maybe in the course of a ma.s.s catastrophe of the kind that did for the dinosaurs. Then the whole process starts again, not from scratch, but from some discernibly earlier part of the arms race. Progress in evolution is not a single upward climb but has a rhyming trajectory more like the teeth of a saw. A sawtooth plunged deeply at the end of the Cretaceous, when the last of the dinosaurs abruptly gave way to the mammals' new and spectacular climb of progressive evolution. But there had been lots of smaller sawteeth during the long reign of the dinosaurs. And since their immediate post-dinosaur rise, the mammals too have had smaller arms races followed by extinctions, followed by renewed arms races. Arms races rhyme with earlier arms races in periodic spurts of many-stepped progressive evolution.

Evolvability That's all I want to say about arms races as drivers of progress. What other messages from the past does the returning host carry back to the present? Well, I must mention the alleged distinction between macroevolution and microevolution. I say 'alleged' because my own view is that macroevolution (evolution on the grand scale of millions of years) is simply what you get when microevolution (evolution on the scale of individual lifetimes) is allowed to go on for millions of years. The contrary view is that macroevolution is something qualitatively different from microevolution. Neither view is self-evidently silly. Nor are they necessarily contradictory. As so often, it depends on what you mean.

Again we can use the parallel of the growth of a child. Imagine an argument about an alleged distinction between macrogrowth and microgrowth. To study macrogrowth, we weigh the child every few months. Every birthday we stand her up against a white doorpost and draw a pencil line to record her height. More scientifically, we could measure various parts of the body, for example the diameter of the head, the width of the shoulders, the length of the major limb bones, and plot them against each other, perhaps logarithmically transformed for the reasons given in the Handyman's Tale. We also note significant events in development such as the first appearance of pubic hair, or the first sign of b.r.e.a.s.t.s and menstruation in girls, and of facial hair in boys. These are the changes that const.i.tute macro-growth, and we measure them on a timescale of years or months. Our instruments are not sensitive enough to pick up the daily and hourly changes in the body microgrowth which, when summed over months, const.i.tute macrogrowth. Or, oddly, they may be too sensitive. A very accurate weighing machine could in theory pick up hourly growth, but the delicate signal is swamped by blundering increases in weight with every meal, and decreases with every act of elimination. The acts of microgrowth itself, which all consist of cell divisions, make no immediate impact on weight at all, and an undetectable impact on gross body measurements.

So, is macrogrowth the sum of lots of small episodes of micro-growth? Yes. But it is also true that the different timescales impose completely different methods of study and habits of thought. Microscopes looking at cells are not appropriate for the study of child development at the whole-body level. And weighing machines and measuring tapes are not suitable for the study of cell multiplication. The two timescales in practice demand radically different methods of study and habits of thought. The same could be said of macroevolution and microevolution. If the terms are used to signify differences in how best to study them, I have no quarrel with a working distinction between microevolution and macroevolution. I do have a quarrel with those people who elevate this rather mundane practical distinction into one of almost or more than almost mystical import. There are those who think Darwin's theory of evolution by natural selection explains microevolution, but is in principle impotent to explain macroevolution, which consequently needs an extra ingredient in extreme cases a divine divine extra ingredient! extra ingredient!

Unfortunately, this hankering after skyhooks has been given aid and comfort by real scientists whose intentions are innocent of any such thing. I have discussed the theory of 'punctuated equilibrium' before, too often and too thoroughly to repeat myself in this book,7 so I shall only add that its advocates usually go on to propose a fundamental 'decoupling' between microevolution and macroevolution. This is an unwarranted inference. No extra ingredient needs to be added at the micro level to explain the macro level. Rather, an extra level of explanation so I shall only add that its advocates usually go on to propose a fundamental 'decoupling' between microevolution and macroevolution. This is an unwarranted inference. No extra ingredient needs to be added at the micro level to explain the macro level. Rather, an extra level of explanation emerges emerges at the macro level as a at the macro level as a consequence consequence of events at the micro level, extrapolated over unimaginable time-spans. of events at the micro level, extrapolated over unimaginable time-spans.

The working distinction between micro- and macroevolution is similar to one that we meet in many other situations. The changes in the map of the world over geological time are due to the effects, summed over millions of years, of plate tectonic events occurring on a timescale of minutes, days and years. But, as with the growth of a child, there is practically no overlap between the methods of study that serve the two timescales. The language of voltage fluctuations is not useful for discussing how a large computer program, such as Microsoft Excel, works. No sensible person denies that computer programs, however complicated, are entirely executed by temporal and spatial patterns of changes between two voltages. But no sensible person attends to that fact while writing, debugging, or using a large computer program.

I have never seen any good reason to doubt the following proposition: macroevolution is lots of little bits of microevolution joined end to end over geological time, and detected by fossils instead of by genetic sampling. Nevertheless, there could be and I believe are major events in evolutionary history after which the very nature of evolution itself changes. Evolution itself might be said to evolve. So far in this chapter, progress has meant individual organisms becoming better over evolutionary time at doing what individuals do, which is survive and reproduce. But we can also countenance changes in the phenomenon of evolution itself. Might evolution itself become better at doing something what evolution does as history goes by? Is late evolution some kind of improvement on early evolution? Do creatures evolve to improve not just their capacity to survive and reproduce, but the lineage's capacity to evolve? Is there an evolution of evolvability?

I coined the phrase 'Evolution of Evolvability' in a paper published in the Proceedings of the 1987 Inaugural Conference on Artificial Life Proceedings of the 1987 Inaugural Conference on Artificial Life. Artificial life was a newly invented merger of other disciplines, notably biology, physics and computer science, founded by the visionary physicist Christopher Langton, who edited the Proceedings Proceedings. Since my paper, but probably not because of it, the evolution of evolvability has become a much discussed topic among students of both biology and artificial life. Long before I used the phrase, others had proposed the idea. For example, the American ichthyologist Karel F. Liem in 1973 used the phrase 'prospective adaptation' for the revolutionary jaw apparatus of cichlid fishes which enabled them, as their tale describes, so suddenly and explosively to evolve hundreds of species in all the great African lakes. Liem's suggestion goes beyond the idea of pre-adaptation, I should say. A pre-adaptation is something that originally evolves for one purpose and is co-opted to another. Liem's prospective adaptation and my evolution of evolvability carry the suggestion not just of co-option to a new function but of unshackling a new outburst of divergent evolution. I am suggesting a permanent and even progressive trend towards becoming better at evolving.

In 1987, the idea of the evolution of evolvability was somewhat heretical, especially for me as the alleged 'ultra-Darwinist'. I was placed in the odd situation of advocating an idea at the same time as apologising for it to people who couldn't see why it needed any apology. It is now a much discussed topic, and others have taken it further than I ever envisaged, for example, the cell biologists Marc Kirschner and John Gerhart, and the evolutionary entomologist Mary Jane West-Eberhard in her magisterial book Developmental Plasticity and Evolution Developmental Plasticity and Evolution.

What makes an organism good at evolving, over and above being good at surviving and reproducing? An example, first. We have already met the idea that island archipelagos are workshops of speciation. If the islands are near enough to each other to allow occasional immigrations, but far enough apart to allow time for evolutionary divergence between immigrations, we have a recipe for speciation, which is the first step towards evolutionary radiation. But how near is near enough? How far is far enough? That depends on the locomotive powers of the animals. For woodlice, a separation of a few yards is equivalent to a separation of many miles for a flying bird or bat. The Galapagos Islands are s.p.a.ced just right for the divergent evolution of small birds such as Darwin's finches, not necessarily for divergent evolution generally. For this purpose, separation of islands should be measured not in absolute units but in units of travelability calibrated to the kind of animal we are talking about as with the Irish boatman who, when my parents asked him the distance of the Great Blasket Island, replied, 'About three miles in fine weather.'

It follows that a Galapagos finch which either decreased or increased or increased its flight range in evolution might thereby its flight range in evolution might thereby decrease decrease its evolvability. Shortening the range lowers the chance of initiating a new race of descendants on another island. That way round, it is easily understood. Lengthening the range has a less obvious effect in the same direction. Descendants are seeded onto new islands so frequently that there is no time for separate evolution before the next immigrant arrives. To push to the extreme, birds whose flight range is large enough to render the distance between islands trivial, no longer see the islands as separate at all. As far as gene flow is concerned, the whole archipelago counts as one continent. So once again speciation is not fostered. High evolvability, if we choose to measure evolvability as speciation rate, is an inadvertent consequence of intermediate locomotor range, where what counts as intermediate, as opposed to too short or too long, depends upon the s.p.a.cing of the islands concerned. Of course 'island' in this sort of argument does not have to mean land surrounded by water. As we saw in the Cichlid's Tale, lakes are islands for aquatic animals, and reefs can be islands within lakes. Mountain tops are islands for land-bound animals that cannot easily tolerate low alt.i.tudes. A tree can be an island for an animal with a short range. For the AIDS virus, every man is an island. its evolvability. Shortening the range lowers the chance of initiating a new race of descendants on another island. That way round, it is easily understood. Lengthening the range has a less obvious effect in the same direction. Descendants are seeded onto new islands so frequently that there is no time for separate evolution before the next immigrant arrives. To push to the extreme, birds whose flight range is large enough to render the distance between islands trivial, no longer see the islands as separate at all. As far as gene flow is concerned, the whole archipelago counts as one continent. So once again speciation is not fostered. High evolvability, if we choose to measure evolvability as speciation rate, is an inadvertent consequence of intermediate locomotor range, where what counts as intermediate, as opposed to too short or too long, depends upon the s.p.a.cing of the islands concerned. Of course 'island' in this sort of argument does not have to mean land surrounded by water. As we saw in the Cichlid's Tale, lakes are islands for aquatic animals, and reefs can be islands within lakes. Mountain tops are islands for land-bound animals that cannot easily tolerate low alt.i.tudes. A tree can be an island for an animal with a short range. For the AIDS virus, every man is an island.

If an increase or a decrease in travelling range results in an increase in evolvability, would we want to call this an evolved 'improvement'? My ultra-Darwinist hackles start to quiver at this point. My heresy litmus starts to blush. It sounds uncomfortably like evolutionary foresight. Birds evolve an increase or a decrease to their flying range because of natural selection for individual survival. Future effects on evolution are an irrelevant consequence. Nevertheless, we might find with hindsight that the species that fill the world tend to be descended from ancestral species with a talent for evolution. You could say, therefore, that there is a kind of high-level, between-lineage selection in favour of evolvability an example of what the great American evolutionist George C. Williams called clade selection. Conventional Darwinian selection leads to individual organisms being finely tuned survival machines. Could it be that, as a consequence of clade selection, life itself has increasingly become a set of finely tuned evolving machines? If this is so, we might expect that in Kauffmanian reruns of evolution, the same progressive improvements in evolvability might be rediscovered.

When I first wrote about the evolution of evolvability, I proposed a number of 'watershed events' in evolution, after which evolvability suddenly improved. The most promising example of a watershed event I could think of was segmentation. Segmentation, you remember, is the train-like modularisation of the body, in which parts and systems are repeated serially down the body. It seems to have been independently invented, in its full form, in arthropods, vertebrates and annelid worms (although the universality of Hox genes argues for some sort of fore and aft serial organisation as a predecessor). The origin of segmentation is one of those evolutionary events that cannot have been gradual. Bony fish typically have about 50 vertebrae, but eels have as many as 200. Caecilians (worm-like amphibians) vary between 95 and 285 vertebrae. Snakes differ hugely in vertebral number: the record known to me is 565 for an extinct snake.

Every vertebra of a snake represents one segment with its own pair of ribs, its own muscle blocks, its own nerves sprouting from the spinal cord. You can't have fractional numbers of segments, and the evolution of variable segment numbers must include numerous instances in which a mutant snake differed from its parents in some whole number of segments: at least one, possibly more, in one fell swoop. Similarly, when segmentation originated, there must have been a mutational transition straight from unsegmented parents to a child with two (at least) segments. It is hard to imagine such a freak surviving, let alone finding a mate and reproducing, but it evidently happened because segmented animals are all around us. Very probably the mutation involved Hox genes, like those of the Fruit Fly's Tale. In my 1987 evolvability paper I guessed that ... the individual success, or otherwise, of the first segmented animal during its own lifetime is relatively unimportant. No doubt many other new mutants have been more successful as individuals. What is important about the first segmented animal is that its descendant lineages were champion evolvers evolvers. They radiated, speciated, gave rise to whole new phyla. Whether or not segmentation was a beneficial adaptation during the individual lifetime of the first segmented animal, segmentation represented a change in embryology that was pregnant with evolutionary potential.

The ease with which whole segments can be added or subtracted from the body is one thing that contributes to enhanced evolvability. So is differentiation among segments. In animals such as millipedes and earthworms, most of the segments are the same as each other. But there is a recurrent tendency, especially among arthropods and vertebrates, for particular segments to become specialised for particular purposes, and hence different from other segments (compare a lobster with a centipede). A lineage that manages to evolve a segmented body plan is immediately able to evolve a whole range of new animals by altering segmental modules, all along the body.

Segmentation is an example of modularity, and modularity in general is a main ingredient in the thinking of more recent writers on the evolution of evolvability. Of the many meanings of module listed in the Oxford English Dictionary Oxford English Dictionary, the relevant one is: One of a series of production units or component parts that are standardized to facilitate a.s.sembly or replacement and are usually prefabricated as selfcontained structures.

Modular is the adjective describing an a.s.semblage of modules, and modularity is the corresponding abstract noun, the property of being modular. Other examples of modular construction include many plants (leaves and flowers are modules). But perhaps the best examples of modularity are to be found at the cellular and biochemical level. Cells themselves are modules par excellence par excellence, and within cells so are protein molecules and, of course, DNA itself.

So the invention of multicellularity is another important water-shed event that almost certainly enhanced evolvability. It preceded segmentation by hundreds of millions of years, and segmentation is itself a kind of large-scale reenactment of it, another leap in modularity. What other watersheds have there been? The dedicatee of this book, John Maynard Smith, collaborated with his Hungarian colleague Eors Szathmary on The Major Transitions in Evolution The Major Transitions in Evolution. Most of their 'major transitions' would fit under my heading of 'watershed events' major improvements in evolvability. This obviously includes the origin of replicating molecules, for without them there could be no evolution at all. If, as Cairns-Smith and others have suggested, DNA usurped the key role of replicator from some less proficient predecessor, bridged by intermediate stages, each one of those stages would const.i.tute a leap forward in evolvability.

If we accept the RNA World theory, there would have been a major transition or watershed, when a world of RNA serving as both replicator and enzyme gave over to a separation between DNA in the replicator role and proteins in the enzyme role. Then there was the clubbing together of replicating ent.i.ties ('genes') in cells with walls, which prevented the gene products leaking away and kept them together with the products of other genes with which they could collaborate in cellular chemistry. A very major transition, and very probably a watershed of evolvability, was the birth of the eukaryotic cell by the commingling of several prokaryotic cells. So was the origin of s.e.xual reproduction, which coincided with the origin of the species itself, with its own gene pool, and all that that implied for future evolution. Maynard Smith and Szathmary go on to list the origin of multicellularity, the origin of colonies such as ant and termite nests, and the origin of human societies with language. There is a rhyming similarity between at least several of these major transitions: they often involve the coming together of previously independent units in a larger grouping at a higher level, with concomitant loss of independence at the lower level.

To their list I have already added segmentation, and I would stress another, which I call bottlenecking. Once again, to spell it out in full would be to repeat from previous books (especially the final chapter, 'Rediscovering the Organism', of The Extended Phenotype The Extended Phenotype). Bottlenecking refers to a type of life history in multicellular organisms. In bottlenecking, the life cycle regularly returns to a single cell, from which a multicellular body is grown anew. The alternative to a bottlenecked life cycle might be a hypothetical straggling water plant that reproduces by breaking off small, multicellular chunks of itself which drift off, grow and then break off more small chunks. Bottlenecking has three important consequences, all of which are certainly good candidates for improvements in evolvability.

First, evolutionary innovations can be reinvented from the bottom up, rather than as remouldings of existing structures the equivalent of beating swords into ploughshares. An improvement in, say, a heart, has a better chance of being a clean improvement if genetic changes can alter the whole course of development from a single cell. Imagine the alternative: take the existing heart and modify it by differential tissue growth within its continuously beating fabric. This on-the-trot remodelling would impair the working of the heart and compromise the would-be improvement.

Second, by continually resetting to a consistent starting point in a recurrent life cycle, bottlenecking provides a 'calendar' by which embryological events may be timed. Genes may be turned on or off at key points in the growth cycle. Our hypothetical straggling chunk-extruder lacks a recognisable timetable to regulate such switchings on and off.

Third, without bottlenecking, different mutations would acc.u.mulate in different parts of the straggling chunk-extruder. The incentive among cells to co-operate would be reduced. In effect, sub-populations of cells would be tempted to behave as cancers, to increase their chance of contributing genes to the extruded chunks. With bottlenecking, since every generation starts out as a single cell, the whole body has a good chance of being made of a uniform genetic population of co-operating cells, all descended from that single cell. Without bottlenecking, the cells of the body might have, from a genetic point of view, 'divided loyalties'.

Related to bottlenecking is another important landmark event in evolution, and one that may well have contributed to evolvability and might be rediscovered in Kauffman reruns. This is the separation of the germ line from the soma, first clearly understood by the great German biologist August Weismann. As we saw at Rendezvous 31 Rendezvous 31, what happens in the developing embryo is that a portion of the cells are set aside for reproduction (germ-line cells) while the rest are destined to make the body (somatic cells). Germ-line genes are potentially immortal, with a prospect of direct descendants millions of years into the future. Somatic genes are destined for a finite, if not always predictable, number of cell divisions to make the body tissues, after which their line will come to an end and the organism will die. Plants often violate the separation, most obviously when they practise vegetative reproduction. This could const.i.tute an important difference between the ways plants and animals evolve. Before the evolutionary invention of the separate soma, all living cells were potentially the ancestors of an indefinite line of descendants, as sponge cells still are.

The invention of s.e.x is a major watershed, which is superficially confusable both with bottlenecking and with the separation of the germ line, but is logically distinct from both. In its most general form, s.e.x is the partial mixing of genomes. We are familiar with a particular, highly regimented version of it in which every individual gets 50 per cent of its genome from each of two parents. We are used to the idea that there are two kinds of parent, female and male, but this is not a necessary part of s.e.xual reproduction. Isogamy is a system in which two individuals, not distinguished as male and female, combine half their genes to make a new individual. The male/female divide is best seen as a further watershed event, which came after the origin of s.e.x itself. Regimented s.e.x of this kind is accompanied in every generation by a 'reduction division', in which each individual donates 50 per cent of its genome to each offspring. Without this reduction, genomes would double in size with every generation.

Bacteria practise a haphazard form of s.e.xual donation that is sometimes described as s.e.x but which is really very different, having more in common with the cut-and-paste, or copy-and-paste, functions of a computer program. Fragments of a genome are copied or cut from one bacterium and pasted into another, which does not have to be a member of the same 'species' (though the very meaning of species is in doubt for bacteria). Because genes are software subroutines that perform cellular operations, a 'pasted' gene can immediately go to work in its new environment, doing the same task as it did before.8 What is in it for the donating bacterium? That may be the wrong question. The right question might be, what is in it for the donated gene? And the answer is that genes that successfully get themselves donated, and then successfully help the recipient bacterium to survive and pa.s.s them on, thereby increase the number of copies of themselves in the world. It is not clear whether our regimented eukaryotic s.e.x has evolved from bacterial 'cut-and-paste' s.e.x, or whether it was an entirely new watershed event. Both must have had a huge impact on subsequent evolution and are candidates for discussion under the heading of the evolution of evolvability. Regimented s.e.x, as we saw in the Rotifer's Tale, has a dramatic effect upon future evolution because it makes possible the very existence of species with their gene pools.

The positioning of the apostrophe in The Ancestor's Tale The Ancestor's Tale indicates a singular. I admit that the motive was partly stylistic. Nevertheless, through the millions probably billions of individual ancestors whose lives we touched along our Pilgrims' Way, one singular hero has recurred in the minor, like a Wagnerian leitmotiv: DNA. Eve's Tale showed that genes have ancestors, no less than individuals. The Neanderthal's Tale applied the lesson to the question of whether that maligned species perished without any legacy to soften the blow. The Gibbon's Tale warmed to the theme of 'majority votes' among genes clamouring to a.s.sert their different views of ancestral history. The Lamprey's Tale identified the a.n.a.logy between gene duplication and speciation, each at its own level an a.n.a.logy so close that separate family trees can be drawn up for genes, which parallel, but do not coincide with, the conventional family trees of phylogeny. The leitmotiv in the field of taxonomy echoes, but is distinct from, the major theme of the 'selfish gene' in the understanding of natural selection. indicates a singular. I admit that the motive was partly stylistic. Nevertheless, through the millions probably billions of individual ancestors whose lives we touched along our Pilgrims' Way, one singular hero has recurred in the minor, like a Wagnerian leitmotiv: DNA. Eve's Tale showed that genes have ancestors, no less than individuals. The Neanderthal's Tale applied the lesson to the question of whether that maligned species perished without any legacy to soften the blow. The Gibbon's Tale warmed to the theme of 'majority votes' among genes clamouring to a.s.sert their different views of ancestral history. The Lamprey's Tale identified the a.n.a.logy between gene duplication and speciation, each at its own level an a.n.a.logy so close that separate family trees can be drawn up for genes, which parallel, but do not coincide with, the conventional family trees of phylogeny. The leitmotiv in the field of taxonomy echoes, but is distinct from, the major theme of the 'selfish gene' in the understanding of natural selection.

The Host's Farewell If, as returning host, I reflect on the whole pilgrimage of which I have been a grateful part, my overwhelming reaction is one of amazement. Amazement not only at the extravaganza of details that we have seen; amazement, too, at the very fact that there are any such details to be had at all, on any planet. The universe could so easily have remained lifeless and simple just physics and chemistry, just the scattered dust of the cosmic explosion that gave birth to time and s.p.a.ce. The fact that it did not the fact that life evolved out of nearly nothing, some 10 billion years after the universe evolved out of literally nothing is a fact so staggering that I would be mad to attempt words to do it justice. And even that is not the end of the matter. Not only did evolution happen: it eventually led to beings capable of comprehending the process, and even of comprehending the process by which they comprehend it.

This pilgrimage has been a trip, not just in the literal sense but in the countercultural sense I met when a young man in California in the 1960s. The most potent hallucinogen on sale in Haight or Ashbury or Telegraph Avenue would be tame by comparison. If it's amazement you want, the real world has it all. Not to stray outside the covers of this book, think of Venus's girdle, migrating jellyfish and tiny harpoons; think of the platypus's radar and the electric fish; of the horsefly larva with the apparent foresight to pre-empt cracks in the mud; think redwood; think peac.o.c.k; think starfish with its piped hydraulic power; think cichlids of Lake Victoria, evolving how how many orders of magnitude faster than many orders of magnitude faster than Lingula, Limulus Lingula, Limulus or or Latimeria? Latimeria? It is not pride in my book but reverence for life itself that encourages me to say, if you want a justification for the latter, open the former anywhere, at random. And reflect on the fact that although this book has been written from a human point of view, another book could have been written in parallel for any of 10 million starting pilgrims. Not only is life on this planet amazing, and deeply satisfying, to all whose senses have not become dulled by familiarity: the very fact that we have evolved the brain power to understand our evolutionary genesis redoubles the amazement and compounds the satisfaction. It is not pride in my book but reverence for life itself that encourages me to say, if you want a justification for the latter, open the former anywhere, at random. And reflect on the fact that although this book has been written from a human point of view, another book could have been written in parallel for any of 10 million starting pilgrims. Not only is life on this planet amazing, and deeply satisfying, to all whose senses have not become dulled by familiarity: the very fact that we have evolved the brain power to understand our evolutionary genesis redoubles the amazement and compounds the satisfaction.

'Pilgrimage' implies piety and reverence. I have not had occasion here to mention my impatience with traditional piety, and my disdain for reverence where the object is anything supernatural. But I make no secret of them. It is not because I wish to limit or circ.u.mscribe reverence; not because I want to reduce or downgrade the true reverence with which we are moved to celebrate the universe, once we understand it properly. 'On the contrary' would be an understatement. My objection to supernatural beliefs is precisely that they miserably fail to do justice to the sublime grandeur of the real world. They represent a narrowing-down from reality, an impoverishment of what the real world has to offer.

I suspect that many who call themselves religious would find themselves agreeing with me. To them I would only quote a favourite remark that I overheard at a scientific conference. A distinguished elder statesman of my subject was having a long argument with a colleague. As the altercation came to an end, he twinkled and said, 'You know, we really do agree. It's just that you say say it wrong!' it wrong!'

I feel I have returned from a true pilgrimage.

1 The Blind Watchmaker The Blind Watchmaker in this case. in this case.

2 In the bees, wasps and ants, the sting is a modified egg-laying tube, and only females sting. In the bees, wasps and ants, the sting is a modified egg-laying tube, and only females sting.

3 The habit was first described by W. A. Lambourn [ The habit was first described by W. A. Lambourn [165].

4 Mutke's claim is disputed. Either way, the first to do it wasn't USAF Major Chuck Yeager in 1947, as patriotic Americans are taught. An American civilian, George Welch, did it two weeks before him. Mutke's claim is disputed. Either way, the first to do it wasn't USAF Major Chuck Yeager in 1947, as patriotic Americans are taught. An American civilian, George Welch, did it two weeks before him.

5 Tom Lehrer, probably the all-time wittiest composer of comic songs, included the following musical direction at the head of one of his piano scores: 'A little too fast.' Tom Lehrer, probably the all-time wittiest composer of comic songs, included the following musical direction at the head of one of his piano scores: 'A little too fast.'

6 Hume said: 'All these various machines, and even their most minute parts, are adjusted to each other with an accuracy which ravishes into admiration all men who have ever contemplated them.' Hume said: 'All these various machines, and even their most minute parts, are adjusted to each other with an accuracy which ravishes into admiration all men who have ever contemplated them.'

7 My view is that it is an interesting empirical question, which is likely to have a different answer in different particular cases, and which does not deserve its elevation to the status of major principle. My view is that it is an interesting empirical question, which is likely to have a different answer in different particular cases, and which does not deserve its elevation to the status of major principle.

8 This is why transgenic manipulation in modern agricultural breeding works, for example, the legendary importing of 'antifreeze' genes from Arctic fish into tomatoes. It works for the same reason as a computer subroutine, copied from one program into another, can be relied upon to deliver the same result. The case of GM crops isn't quite so straightforward. But the example serves to allay fears of the 'unnaturalness' of importing, say, fish genes into tomatoes, as though some kind of fishy 'flavour' goes too. A subroutine is a subroutine, and DNA's language of programming is identical in fish and tomatoes. This is why transgenic manipulation in modern agricultural breeding works, for example, the legendary importing of 'antifreeze' genes from Arctic fish into tomatoes. It works for the same reason as a computer subroutine, copied from one program into another, can be relied upon to deliver the same result. The case of GM crops isn't quite so straightforward. But the example serves to allay fears of the 'unnaturalness' of importing, say, fish genes into tomatoes, as though some kind of fishy 'flavour' goes too. A subroutine is a subroutine, and DNA's language of programming is identical in fish and tomatoes.

By Richard Dawkins

The Selfish Gene The Extended Phenotype The Blind Watchmaker River Out of Eden Climbing Mount Improbable Unweaving the Rainbow A Devil's Chaplain The Ancestor's Tale The G.o.d Delusion The Greatest Show on Earth

FURTHER READING.

Numbers in square brackets refer to sources listed in the Bibliography.

Barlow, George (2002) The Cichlid Fishes: Nature's Grand Experiment in Evolution The Cichlid Fishes: Nature's Grand Experiment in Evolution. Perseus Publishing, Cambridge, Ma.s.s.

Diamond, Jared (1997) Guns, Germs and Steel: A Short History of Everybody for the Last 13,000 Years Guns, Germs and Steel: A Short History of Everybody for the Last 13,000 Years. Chatto & Windus, London.

Fortey, Richard (1997) Life: An Unauthorised Biography Life: An Unauthorised Biography. HarperCollins, London.

Fortey, Richard (2004) The Earth: An Intimate History The Earth: An Intimate History. HarperCollins, London.

Leakey, Richard (1994) The Origin of Humankind: Unearthing Our Family Tree The Origin of Humankind: Unearthing Our Family Tree. Science Masters series, Basic Books, New York.

Maynard Smith, John & Szathmary, Eors (1999). The Origins of Life: From the Birth of Life to the Origin of Language The Origins of Life: From the Birth of Life to the Origin of Language. Oxford University Press, Oxford. (See also [ [189] for a more detailed treatment.) Quammen, David (1996) The Song of the Dodo: Island Biogeography in an Age of Extinctions The Song of the Dodo: Island Biogeography in an Age of Extinctions. Hutchinson, Oxford.

Ridley, Mark (2000) Mendel's Demon: Gene Justice and the Complexity of Life Mendel's Demon: Gene Justice and the Complexity of Life. Weidenfeld & Nicolson, London.

Ridley, Matt (1999) Genome: The Autobiography of a Species in 23 Chapters Genome: The Autobiography of a Species in 23 Chapters. Fourth Estate, London.

Southwood, Richard (2003) The Story of Life The Story of Life. Oxford University Press, Oxford.

Tudge, Colin (2000) The Variety of Life: A Survey and a Celebration of all the Creatures that Have Ever Lived The Variety of Life: A Survey and a Celebration of all the Creatures that Have Ever Lived. Oxford University Press, Oxford.

Weiner, Jonathan (1994) The Beak of the Finch: A Story of Evolution in Our Time The Beak of the Finch: A Story of Evolution in Our Time. Jonathan Cape, London.

Wilson, E. O. (1992) The Diversity of Life The Diversity of Life. Harvard University Press, Cambridge, Ma.s.s.

ADVANCED READING.

Brusca, Richard C. & Brusca, Gary J. (2002) Invertebrates Invertebrates. 2nd ed. Sinauer a.s.sociates Inc, Sunderland, Ma.s.s.

Carroll, Robert L. (1988) Vertebrate Paleontology and Evolution Vertebrate Paleontology and Evolution. W. H. Freeman, New York.

Macdonald, David (2001) The New Encyclopedia of Mammals The New Encyclopedia of Mammals. Oxford University Press, Oxford.

Ridley, Mark (2004) Evolution Evolution. 3rd ed. Blackwell, Oxford.

NOTES TO THE PHYLOGENIES AND RECONSTRUCTIONS.

Yan Wong Numbers in square brackets refer to sources listed in the Bibliography.

PHYLOGENY DIAGRAMS.

The following notes outline the scientific basis for the phylogenies in this book, particularly in areas of major recent taxonomic revision and current debate. A good, relatively recent phylogenetic survey is given in Colin Tudge's The Variety of Life The Variety of Life [ [289].

RENDEZVOUS 0 The Americas are omitted because evidence points to humans having arrived there recently from Asia. Concestor 0 must logically be at least as recent as any gene MRCA (such as Y-chromosome 'Adam'), and even low levels of interbreeding are enough to result in a very recent MRCA of all humans [45], hence the recent date used here.

RENDEZVOUS 1 & 2 Phylogeny (as for the rest of the trees, the majority 'vote' among genes see see the Gibbon's Tale) supported by morphology [ the Gibbon's Tale) supported by morphology [102] and molecules [20]. Divergence dates based on the molecular clock [105, 230 230].

RENDEZVOUS 3 Phylogeny and divergence dates based on morphological, fossil, and molecular data [102, 105 105, 273 273].

RENDEZVOUS 4 Gibbon phylogeny is unsure: this tree is based upon mtDNA data [246, fig 2c], supplemented by molecular clock dates for the Concestor and Symphalangus/Hylobates Symphalangus/Hylobates nodes [ nodes [105].

RENDEZVOUS 5 Conventional phylogeny. Divergence dates given by molecular and fossil data [105].

RENDEZVOUS 6 Phylogeny and dates taken directly or inferred from [105]. The position of the Aotinae is not very secure, and may change in the future.

RENDEZVOUS 7 Placement and dating [105] of the tarsier family agrees with molecular [254] and morphological data.

RENDEZVOUS 8 Within strepsirhines, lemur interrelationships are disputed, although the aye-aye is often considered basal. Order and dating of the four other families is from molecules [322], scaled to place basal primate divergence at 63 Mya [105, 230 230]. However, other calculations place this divergence at 80 Mya [281], moving Rendezvous 9 Rendezvous 9, 10 10, and 11 11 backwards by up to 15 million years. backwards by up to 15 million years.

RENDEZVOUS 9 Placement of colugos and tree shrews is highly controversial (see accompanying tale), and is here based on recent molecular data [207]. Basal date then constrained by surrounding nodes to 6375 Mya.

RENDEZVOUS 10 Placement of Glires from robust molecular evidence [207]. Rendezvous date constrained by molecular clock dating of Rendezvous 11 Rendezvous 11 [ [207, 137 137], but may be up to 10 Mya or earlier [271]. Lagomorph placement uncontroversial [137, 207 207]. Rodent phylogeny debated. Hystricognath rodents (Hystricidae, Phiomorpha, Caviomorpha) generally accepted. Otherwise, 4 groups often found in molecular studies [e.g. 137 137, 202 202]: Muridae+Dipodidae, Aplodontidae+Sciuridae+Gliridae, Ctenodactylidae+ hystricognaths, Heteromyidae+Geomyidae. Branching order and rough dating of these groups from mtDNA and rDNA [202], but order is not robust [e.g. see see 137 137].

RENDEZVOUS 11 & 12 Phylogeny and dating from recent revolutionary molecular studies [207, 271 271].

RENDEZVOUS 13 Phylogeny and dating from molecular data [207, 271 271]. Morphology [177] and some molecules [205] agree on elephant/sirenian/hyrax split. However, there is uncertainty in the placement of the aardvark [205, 271 271], and morphological data may still conflict with the position of the Afrosoricida [177].

RENDEZVOUS 14 Rendezvous supported by old and recent data [208]. Placental-marsupial divergence at 140 Mya consistent with fossils and late molecular dates [7, 144 144]. Molecular studies find didelphids, then paucituberculates as sister to other marsupials [212, 272 272], consistent with morphology [251]. Other branches variably supported by molecular data [212, 272 272]: position of monito del monte particularly uncertain, here interpreted as sister to Diprotodontia [251]. Divergence dates based on molecular clock data, but also constrained by Gondwanan biogeography [212].

RENDEZVOUS 15 Phylogeny and dating from recent molecular, morphological, and fossil data [208].

RENDEZVOUS 16 Date estimates for Rendezvous 16 Rendezvous 16 average around 310 Mya [ average around 310 Mya [112], other early branch dates from fossil data [40]. Now-conventional branching within snakes and lizards [228]. Bird branching order from genetic studies [293] with dates from DNA hybridisation [265]: many orders grouped as Neoaves due to uncertain relationships.

RENDEZVOUS 17 Although disputed by some palaeontologists [40], molecular and morphological data strongly support lissamphibian monophyly, and hint at order of branching shown here [325]. Basal date from palaeontological evidence [4], others from maximum likelihood trees of mtDNA [325].

RENDEZVOUS 18 & 19 Phylogeny and dating from molecular [294] and morphological/palaeontological [326] studies.

RENDEZVOUS 20 Rendezvous date generally accepted [209]. Ray-finned fish phylogeny is currently in a state of flux [141, 199 199], although the traditional view followed here [209] is broadly supported. Divergence dates based on fossil data [40, 209 209]. Some groups deliberately omitted for simplicity, as phylogeny not robust.

RENDEZVOUS 21 Phylogeny based on morphological data [75, 263 263] [263]. Divergence dates based on fossil data [209, 252 252].

RENDEZVOUS 22 Agnathan grouping based on genetic data [97, 279 279] which contradicts most fossil-based phylogenies (but these specialised groups show secondarily character lost, making morphological data difficult to use). Rendezvous date tightly constrained by fossil data [264]. Lampreyhagfish divergence time suggested by molecular maximum likelihood trees [279].

RENDEZVOUS 23 Molecular clock data [315] places lancelet split close to basal deuterostome divergences here, estimated at 570 Mya according to medium-fuse dating of Cambrian Explosion (see the Velvet Worm's Tale). the Velvet Worm's Tale).

RENDEZVOUS 24 Rendezvous date constrained by surrounding nodes. Possibly closer to ambulacrarians than to lancelets [315].

RENDEZVOUS 25 Ambulacrarian grouping and basal divergences from recent genetic data [32, 97 97, 315 315], a.s.suming medium fuse Cambrian explosion. Genetic studies also give deep-branching Xenoturbellida Xenoturbellida [ [28], although exact placement not robust. Echinoderm phylogeny and dating from genetic, morphological, and fossil data [176, 297 297].

RENDEZVOUS 26 Rendezvous date (about 590 Mya) from recent molecular clock studies [8, 10 10], and broadly consistent with fossil data [291]. Protostome phylogeny recently revised [3]: here a single broad scheme has been followed [103], based on genetics and morphology. Three branches consist of several phyla grouped together. These are: Cephalorhyncha [103], Gnathifera [162] (including Acanthocephala and Myzostomida), and Brachiozoa (phoronids and brachiopods). Edysozoan phylogeny relatively robust [103]: main uncertainties are the onychoph.o.r.e/arthropod grouping and basal inclusion of chaetognaths, here placed according to morphological/genetic data [224]. Many ecdysozoan dates constrained by 'small-sh.e.l.ly' onychoph.o.r.e fossils (see the Velvet Worm's Tale). Lophotrochozoa branching order much more uncertain: annelid/mollusc/sipunculid group robust [ the Velvet Worm's Tale). Lophotrochozoa branching order much more uncertain: annelid/mollusc/sipunculid group robust [224], nemerteans probably sister to this [290], branching order of others unsure.

RENDEZVOUS 27 Phylogeny based on molecular data [247, 283 283]. These often weakly support a paraphyletic Acoela, but morphological data strongly supports acoelomorph monophyly; divergence date thus arbitrary. Rendezvous date based on genetic distance estimates [247, 283 283], a.s.suming protostome/deuterostome split of 590 Mya and bilaterian/cnidarian split of 700 Mya.

RENDEZVOUS 28 & 29 Order of branching of cnidarians and ctenoph.o.r.es is still uncertain [35]. Certain molecular data weakly support the order used here [191]. Within cnidarian phylogeny now conventional, dates from genetic studies [50] calibrated to timescale used here.

RENDEZVOUS 30 Trichoplax placement unsure [35], but possibly near the base of the Metazoa [Peter Holland, pers. comm.].

RENDEZVOUS 31 Sponges generally interpreted as basal metazoans, although occasionally molecular data hint that they might be paraphyletic [191]. Rendezvous date of 800 Mya based on molecular clock data [211], recalibrated using protostomedeuterostome divergence of 590 Mya; this conflicts with absence of fossilised sponge spicules before the latest Precambrian, although these may represent a derived character.