As in the case of chain letters, success among chemical replicators is simply synonymous with frequency in circulation. But that is just definition: almost tautology. Success is earned by practical competence, and competence means something concrete and anything but tautological. A successful replicator molecule will be one that, for reasons of detailed chemical technicality, has what it takes to get duplicated. What this means in practice can be almost infinitely variable, even though the nature of the replicators themselves can seem surprisingly uniform.

DNA is so uniform that it consists entirely of variations in sequence of the same four “letters” – A, T, C and G. By comparison, as we have seen in earlier chapters, the means used by DNA sequences to get themselves replicated are bewilderingly variable. They include building more efficient hearts for hippos, springier legs for fleas, aerodynamically more streamlined wings for swifts, more buoyant swim bladders for fish. All the organs and limbs of animals; the roots, leaves and flowers of plants; all eyes and brains and minds, and even fears and hopes, are the tools by which successful DNA sequences lever themselves into the future. The tools themselves are almost infinitely variable, but the recipes for building those tools are, by contrast, ludicrously uniform. Just permutation after permutation of A, T, C and G.

It may not always have been so. We have no evidence that when the information explosion started, the seed code was written in DNA letters. Indeed, the whole DNA/protein-based information technology is so sophisticated – high tech, it has been called by the chemist Graham Cairns-Smith – that you {151} can scarcely imagine it arising by luck, without some other self-replicating system as a forerunner. The forerunner might have been RNA; or it might have been something like Julius Rebek's simple self-replicating molecules; or it might have been something very different: one tantalizing possibility, which I have discussed in detail in The Blind Watchmaker, is Cairns-Smith's own suggestion (see his Seven Clues to the Origin of Life) of inorganic clay crystals as primordial replicators. We may never know for certain.

What we can do is guess at a general chronology of a life explosion on any planet, anywhere in the universe. The details of what will work must depend on local conditions. The DNA/protein system wouldn't work in a world of chilled liquid ammonia, but perhaps some other system of heredity and embryology would. Anyway, those are just the kinds of specifics I want to ignore, because I want to concentrate on the planet-independent principles of the general recipe. I'll go more systematically now through the list of thresholds that any planetary replication bomb can be expected to pass. Some of these are likely to be genuinely universal. Others may be peculiar to our own planet. It may not always be easy to decide which are likely to be universal and which local, and this question is interesting in its own right.

Threshold 1 is, of course, the Replicator Threshold itself: the arising of some kind of self-copying system in which there is at least a rudimentary form of hereditary variation, with occasional random mistakes in copying. The consequence of Threshold l's being passed is that the planet comes to contain a mixed population, in which variants compete for resources. Resources will be scarce – or will become scarce when the {152} competition hots up. Some variant replicas will turn out to be relatively successful in competing for scarce resources. Others will be relatively unsuccessful. So now we have a basic form of natural selection.

To begin with, success among rival replicators will be judged purely on the direct properties of the replicators themselves – for example, how well their shape fits a template. But now, after many generations of evolution, we move on to Threshold 2, the Phenotype Threshold. Replicators survive not by virtue of their own properties but by virtue of causal effects on something else, which we call the phenotype. On our planet, phenotypes are easily recognized as those parts of animal and plant bodies that genes can influence. That means pretty well all bits of bodies. Think of phenotypes as levers of power by which successful replicators manipulate their way into the next generation. More generally, phenotypes may be defined as consequences of replicators that influence the replicators' success but are not themselves replicated. For instance, a particular gene in a species of Pacific island snail determines whether the shell coils to the right or to the left. The DNA molecule itself is not right- or left-handed, but its phenotypic consequence is. Left-handed and right-handed shells may not be equally successful at the business of providing the outer protection for snail bodies. Because snail genes ride inside the shells whose shape they help to influence, genes that make successful shells will come to outnumber genes that make unsuccessful shells. Shells, being phenotypes, do not spawn daughter shells. Each shell is made by DNA, and it is DNA that spawns DNA. {153}

DNA sequences influence their phenotypes (like the direction of coiling of shells) via a more or less complicated chain of intermediate events, all subsumed under the general heading of “embryology.” On our planet, the first link in the chain is always the synthesis of a protein molecule. Every detail of the protein molecule is precisely specified, via the famous genetic code, by the ordering of the four kinds of letters in the DNA. But these details are very probably of local significance only. More generally, a planet will come to contain replicators whose consequences (phenotypes) have beneficial effects, by whatever means, on the replicators' success at getting copied. Once the Phenotype Threshold is crossed, replicators survive by virtue of proxies, their consequences on the world. On our planet, these consequences are usually confined to the body in which the gene physically sits. But this is not necessarily so. The doctrine of the Extended Phenotype (to which I have devoted a whole book with that title) states that the phenotypic levers of power by which replicators engineer their long-term survival do not have to be limited to the replicators' “own” body. Genes can reach outside particular bodies and influence the world at large, including other bodies.

I don't know how universal the Phenotype Threshold is likely to be. I suspect that it will have been crossed on all those planets where the life explosion has proceeded beyond a very rudimentary stage. And I suspect that the same is true of the next threshold in my list. This is Threshold 3, the Replicator Team Threshold, which may on some planets be crossed before, or at the same time as, the Phenotype {154} Threshold. In early days, replicators are probably autonomous entities bobbing about with rival naked replicators in the headwaters of the genetic river. But it is a feature of our modern DNA/protein information-technology system on Earth that no gene can work in isolation. The chemical world in which a gene does its work is not the unaided chemistry of the external environment. This, to be sure, forms the background, but it is quite a remote background. The immediate and vitally necessary chemical world in which the DNA replicator has its being is a much smaller, more concentrated bag of chemicals – the cell. In a way, it is misleading to call a cell a bag of chemicals, because many cells have an elaborate internal structure of folded membranes on which, in which, and between which vital chemical reactions go on. The chemical microcosm that is the cell is put together by a consortium of hundreds – in advanced cells, hundreds of thousands – of genes. Each gene contributes to the environment, which they all then exploit in order to survive. The genes work in teams. We saw this from a slightly different angle in chapter 1.