Now, back to Pasteur's tartrate crystals. Pasteur noticed that when he left a solution of tartrate in water, two different kinds of crystal emerged, identical except that they were mirror images of each other. He laboriously sorted the two kinds of crystal into two separate heaps. When he separately redis-solved them, he obtained two different solutions, two kinds of tartrate in solution. Although the two solutions were similar in most respects, Pasteur found that they rotated polarized light in opposite directions. It is this that gives the two kinds of molecule their conventional names of left- and right-handed, since they rotate polarized light anticlockwise and clockwise, respectively. As you would guess, when the two solutions were allowed to crystallize out once more, each produced pure crystals that mirrored the pure crystals of the other.

Mirror-image molecules really are distinct in that, as {141} with left and right shoes, no matter how hard you try, you can't rotate them so that one can be used as a substitute for the other. Pasteur's original solution was a mixed population of two kinds of molecules, and the two kinds each insisted on lining up with their own kind when crystallizing out. The existence of two (or more) distinct varieties of an entity is a necessary condition for there to be true heredity, but it is not sufficient. For there to be true heredity among the crystals, left- and right-handed crystals would have to split in half when they reached a critical size and each half serve as a template for growth to full size again. Under these conditions we really would have a growing population of two rival kinds of crystals. We truly might speak of “success” in the population, because – since both types are competing for the same constituent atoms – one type might become more numerous at the expense of the other, by virtue of being “good” at making copies of itself. Unfortunately, the vast majority of known molecules do not have this singular property of heredity.

I say “unfortunately” because chemists, trying for medical purposes to make molecules that are all, say, left-handed, would dearly like to be able to “breed” them. But insofar as molecules act as templates for the formation of other molecules, they normally do so for their mirror image, not for their like-handed form. This makes things difficult, because if you start with a left-handed form you end up with an equal mixture of left- and right-handed molecules. Chemists involved in this field are trying to trick molecules into “breeding” daughter molecules of the same handed-ness. It is a very difficult trick to pull off. {142}

In effect, though it probably didn't involve handedness, a version of this trick was pulled off naturally and spontaneously four thousand million years ago, when the world was new and the explosion that turned into life and information began. But something more than simple heredity was needed before the explosion could properly get under way. Even if a molecule does show true heredity among left-handed and right-handed forms, any competition between them would not have very interesting consequences, because there are only two kinds. Once the lefthanders, say, had won the competition, that would be the end of the matter. There would be no more progress.

Larger molecules can exhibit handedness at different parts of the molecule. The antibiotic monensin, for instance, has seventeen centers of asymmetry. At every one of these seventeen centers, there is a left-handed and a right-handed form. Two multiplied by itself 17 times is 131,072, and there are therefore 131,072 distinct forms of the molecule. If these 131,072 possessed the property of true heredity, with each one begetting only its own kind, there could be quite a complicated competition, as the most successful members of the set of 131,072 gradually asserted themselves in successive population censuses. But even this would be a limited kind of heredity, because 131,072, though a large number, is finite. For a life explosion worthy of the name, heredity is needed but so also is indefinite, open-ended variety.

With monensin, we have reached the end of the road, as far as mirror-image heredity is concerned. But left-handedness versus right-handedness is not the only kind of difference {143} that might lend itself to hereditary copying. Julius Rebek and his colleagues at the Massachusetts Institute of Technology are chemists who have taken seriously the challenge of producing self-replicating molecules. The variants they exploit are not mirror images. Rebek and colleagues took two small molecules – the detailed names don't matter, let's just call them A and B. When A and B are mixed in solution, they join up to form a third compound called – you've guessed it – C. Each C molecule acts as a template, or mold. The As and Bs, floating free in solution, find themselves slotting into the mold. One A and one B are jostled into position in the mold, and they thereby find themselves correctly aligned to make a new C, just like the previous one. The Cs don't stick together to form a crystal but split apart. Both Cs are now available as templates to make new Cs, so the population of Cs grows exponentially.

As described so far, the system doesn't exhibit true heredity, but mark the sequel. The B molecule comes in a variety of forms, each of which combines with A to make its own version of the C molecule. So we have C1, C2, C3, and so on. Each of these versions of the C molecule serves as a template for the formation of other Cs of its own type. The population of Cs is therefore heterogeneous. Moreover, the different types of C are not all equally efficient at making daughters. So there is competition between rival versions of C in the population of C molecules. Better yet, “spontaneous mutation” of the C molecule can be induced by ultraviolet radiation. The new mutant type proved to “breed true,” producing daughter molecules just like itself. Satisfyingly, the new variant outcompeted the parent type and rapidly took over the test-tube world in which {144} these protocreatures had their being. The A/B/C complex is not the only set of molecules that behaves in this way. There's D, E and F, to name just one comparable triplet. Rebek's group has even been able to make self-replicating hybrids of elements of the A/B/C complex and the D/E/F complex.

The truly self-copying molecules we know in nature – the nucleic acids DNA and RNA – have an altogether richer potential for variation. Whereas a Rebek replicator is a chain with only two links, a DNA molecule is a long chain of indefinite length; each of the hundreds of links in the chain can be any one of four kinds; and when a given stretch of DNA acts as a template for the formation of a new molecule of DNA, each of the four kinds acts as a template for a different particular one of the four. The four units, known as bases, are the compounds adenine, thymine, cytosine and guanine, conventionally referred to as A, T, C and G. A always acts as a template for T, and vice versa. G always acts as a template for C, and vice versa. Any conceivable ordering of A, T, C and G is possible and will be faithfully duplicated. Moreover, since DNA chains are of indefinite length, the range of available variation is effectively infinite. This is a potential recipe for an informational explosion whose reverberations can eventually reach out from the home planet and touch the stars.

The reverberations of our solar system's replicator explosion have been confined to the home planet for most of the four billion years since it happened. Only in the last million years has a nervous system capable of inventing a radio technology arisen. And only in the last few decades has that nervous system actually developed radio technology. Now, an {145} expanding shell of information-rich radio waves is advancing outward from the planet at the speed of light.