Now here is an important respect in which we have to be cautious about the river metaphor. When we think of the divide leading to all the mammals – as opposed to, say, the stream leading to the gray squirrel – it is tempting to imagine something on a grand, Mississippi/Missouri scale. The mammal branch is, after all, destined to branch and branch {9} and branch again, until it produces all the mammals – from pigmy shrew to elephant, from moles underground to monkeys atop the canopy. The mammal branch of the river is destined to feed so many thousands of important trunk waterways, how could it be other than a massive, rolling torrent? But this image is deeply wrong. When the ancestors of all the modern mammals broke away from those that are not mammals, the event was no more momentous than any other speciation. It would have gone unremarked by any naturalist who happened to be around at the time. The new branch of the river of genes would have been a trickle, inhabiting a species of little nocturnal creature no more different from its nonmammalian cousins than a red squirrel is from a gray. It is only with hindsight that we see the ancestral mammal as a mammal at all. In those days, it would have been just another species of mammal-like reptile, not markedly different from perhaps a dozen other small, snouty, insectivorous morsels of dinosaur food.

The same lack of drama would have attended the earlier splits between the ancestors of all the great groups of animals: the vertebrates, the mollusks, the crustaceans, the insects, the segmented worms, the flatworms, the jellyfish and so on. When the river that was to lead to the mollusks (and others) parted from the river that was to lead to the vertebrates (and others), the two populations of (probably wormlike) creatures would have been so alike that they could have mated with one another. The only reason they didn't is that they had become accidentally separated by some geographical barrier, perhaps dry land separating previously united waters. Nobody could have guessed that one population was destined to spawn the mollusks and the {10} other the vertebrates. The two rivers of DNA were streamlets barely parted, and the two groups of animals were all but indistinguishable.

Zoologists know all this, but they forget it sometimes when contemplating the really big animal groups, like mollusks and vertebrates. They are tempted to think of the divide between major groups as a momentous event. The reason zoologists may be so misled is that they have been brought up in the almost reverential belief that each of the great divisions of the animal kingdom is furnished with something deeply unique, often called by the German word Bauplan. Although this word just means “blueprint,” it has become a recognized technical term, and I shall inflect it as an English word, even though (as I am slightly shocked to discover) it is not yet in the current edition of the Oxford English Dictionary. (Since I enjoy the word less than some of my colleagues do, I admit to a tiny frisson of Schadenfreude at its absence; those two foreign words are in the Dictionary, so there is no systematic prejudice against importation.) In its technical sense, bauplan is often translated as “fundamental body plan.” The use of the word “fundamental” (or, equivalently, the self-conscious dropping into German to indicate profundity) is what causes the damage. It can lead zoologists to make serious errors.

One zoologist, for instance, has suggested that evolution in the Cambrian period (between about six hundred million and about five hundred million years ago) must have been a completely different kind of process from evolution in later times. His reasoning was that nowadays it is new species that are coming into existence, whereas in the Cambrian period major groups were appearing, such as the mollusks {11} and the crustaceans. The fallacy is glaring! Even creatures as radically different from one another as mollusks and crustaceans were originally just geographically separated populations of the same species. For a while, they could have interbred if they had met, but they did not. After millions of years of separate evolution, they acquired the characteristics which we, with the hindsight of modern zoologists, now recognize as those of mollusks and crustaceans respectively. These characteristics are dignified with the grandiose title of “fundamental body plan” or “bauplan.” But the major bauplans of the animal kingdom diverged from common origins by gradual degrees.

Admittedly, there is a minor, if much publicized, disagreement over quite how gradual or “jumpy” evolution is. But nobody, and I mean nobody, thinks that evolution has ever been jumpy enough to invent a whole new bauplan in one step. The author I quoted was writing in 1958. Few zoologists would explicitly take his position today, but they sometimes do so implicitly, speaking as though the major groups of animals arose spontaneously and perfectly formed, like Athena from the head of Zeus, rather than by divergence of an ancestral population while in accidental geographical isolation. [2]

The study of molecular biology has, in any case, shown the great animal groups to be much closer to one another than we used to think. You can treat the genetic code as a dictionary in which sixty-four words in one language (the sixty-four {12} possible triplets of a four-letter alphabet) are mapped onto twenty-one words in another language (twenty amino acids plus a punctuation mark). The odds of arriving at the same 64:21 mapping twice by chance are less than one in a million million million million million. Yet the genetic code is in fact literally identical in all animals, plants and bacteria that have ever been looked at. All earthly living things are certainly descended from a single ancestor. Nobody would dispute that, but some startlingly close resemblances between, for instance, insects and vertebrates are now showing up when people examine not just the code itself but detailed sequences of genetic information. There is a quite complicated genetic mechanism responsible for the segmented body plan of insects. An uncannily similar piece of genetic machinery has also been found in mammals. From a molecular point of view, all animals are pretty close relatives of one another and even of plants. You have to go to bacteria to find our distant cousins, and even then the genetic code itself is identical to ours. The reason it is possible to do such precise calculations on the genetic code but not on the anatomy of bauplans is that the genetic code is strictly digital, and digits are things you can count precisely. The river of genes is a digital river, and I must now explain what this engineering term means.

Engineers make an important distinction between digital and analog codes. Phonographs and tape recorders – and until recently most telephones – use analog codes. Compact disks, computers, and most modern telephone systems use digital codes. In an analog telephone system, continuously fluctuating waves of pressure in the air (sounds) are transduced into correspondingly fluctuating waves of voltage in a wire. A {13} phonograph record works in a similar way: the wavy grooves cause a stylus to vibrate, and the movements of the stylus are transduced into corresponding fluctuations in voltage. At the other end of the line these voltage waves are reconverted, by a vibrating membrane in the telephone's earpiece or the phonograph's loudspeaker, back into the corresponding air-pressure waves, so that we can hear them. The code is a simple and direct one: electrical fluctuations in wire are proportional to pressure fluctuations in air. All possible voltages, within certain limits, may pass down the wire, and the differences between them matter.

In a digital telephone, only two possible voltages – or some other discrete number of possible voltages, such as 8 or 256 – pass down the wire. The information lies not in the voltages themselves but in the patterning of the discrete levels. This is called Pulse Code Modulation. The actual voltage at any one time will seldom be exactly equal to any of the eight, say, nominal values, but the receiving apparatus will round it off to the nearest of the designated voltages, so that what emerges at the other end of the line is well-nigh perfect even if the transmission along the line is poor. All you have to do is set the discrete levels far enough apart so that random fluctuations can never be misinterpreted by the receiving instrument as the wrong level. This is the great virtue of digital codes, and it is why audio and video systems – and information technology generally – are increasingly going digital. Computers, of course, use digital codes for everything they do. For reasons of convenience, it is a binary code – that is, it has only two levels of voltage instead of 8 or 256.