Cheaper than printing it out: buy the paperback book.

Natural selection is a very grim natural reaper. Darwin made the bold claim that, at the very heart of evolution, many small deletions in bulk -- many small wanton deaths -- feeding on the throwaway optimism of minor variation, could, in a counterintuitive way, add up to something truly new and meaningful. In the drama of traditional selection theory, death plays the star role. It works single-mindedly by attrition. It is an editor that knows only one word: "No." Variation counterbalances the one-note song of death by giving birth to the new in cheap abundance. It too knows only one word: "Maybe." Variation cranks out disposable "maybes" in bulk, which are immediately mowed down by death. Bulk mediocrity is dismissed by wanton death. Occasionally, the theory goes, this duet produces a "Yes!" -- a starfish, kidney cells, or Mozart. On the face of it, evolution by natural selection is still a startling hypothesis.

Death gives room for the new, it eliminates the ineffective. But to say that death causes wings to be formed, or eyeballs to work, is essentially wrong. Natural selection merely selects away the deformed wing, the unseeing eye. "Natural selection is the editor, not the author," says Lynn Margulis. What, then, authors innovation in flight and sight?

Evolution theory, from Darwin on, has had a dismal record in dealing with the origin of innovation. As his book title made clear, the question of the origin of species was the great riddle Darwin hoped to solve, not the origin of individuality. He asked, Where did new kinds of creatures come from? He did not ask, Where did variation among individuals come from?

Genetics, which began as a distinctly separate field of science, did pay attention to variation and origin of innovation. Early geneticists like Mendel and William Bateson (Gregory Bateson's father and the man who coined the term "genetics") struggled with explanations of how variations arose and were passed on to descending generations. Sir Francis Galton showed that for statistical purposes -- the main bent of genetics until bioengineering came along -- the propagation of variation within populations could be considered to have a random origin.

Later, when the mechanism for heredity was discovered to be a code of four symbols strung on a long chain of molecules, the random flip of a symbol at a random point on the thread was easy to visualize as a cause of variation and easy to model in mathematics. These molecular flips are generally attributed to cosmic rays or thermodynamic noise. A monstrous mutation, once implying freakish severity, was newly seen as simply a flip, a mere deviation from the average variation. It was not long before all variations in an organism -- from freckles to cleft palates -- were treated as statistical degrees of mutational error. Variation thus became mutation and "mutation" became inseparably compounded into "random mutation." Today, the term random mutation seems redundant. What other kind of mutation could there possibly be?

In computer-intensive artificial evolution, mutations are manufactured by electronic, pseudo-random generators. But the exact nitty-gritty origins of mutations and variations in biology are still uncertain. We do know this: variation is emphatically not due to random mutation -- at least not always; it has some measure of order. This is an old idea. As early as 1926, theorist Jan Christaan Smuts gave this genetic semi-order a name: internal selection.

A plausible scenario for internal selection allows cosmic rays to produce supposedly random errors in the DNA code, which are then corrected in cells by a known self-repair apparatus working in a discriminate (but unknown) fashion -- correcting some and passing others. There is a high energetic cost to the correction of errors, a cost which must be weighed against the possible benefit of the variations. If the error occurred where it is probably opportune, it stays; if it occurs where it is bothersome, it is corrected. For a hypothetical example, the Krebs cycle is the basic fuel plant in every cell of your body. It has worked fine for hundreds of millions of years. There is simply too little to gain, and far too much to lose, in fiddling with it now. When a variation is detected in the code for the Krebs cycle, it is quickly extinguished. On the other hand, body size and body proportions might be worth tweaking; let's leave that area open to variation. If this were how it worked, differential variation would mean that some randomness is "more equal" than others. One fascinating consequence of this setup is that a mutation in the regulatory apparatus itself could have a large-scale effect far beyond a mutation in the strings it governs. I'll get back to that later.

Because genes interact and regulate each other so extensively, the genome forms a complex whole that resists change. Only certain areas can vary at all because most of the genes are so interdependent upon each other -- almost grid-locked -- that variation is not a choice. As evolutionist Ernst Mayr puts it, "Free variability is found only in a limited portion of the genotype." The power of this genetic holism can be seen in animal breeding. Breeders commonly encounter undesirable side effects triggered when unknown genes are activated in the process of selecting for one particular trait. However, when pressure for that one trait is let up, organisms in succeeding generations rapidly revert to the original type, much as if the genome has sprung back to its set point. Variation in real genes is quite different than we imagined. The evidence suggests that not only is it nonrandom and parochial, but it is difficult to come by at all.

The impression one gets is of a highly flexible bureaucracy of genes managing the lives of other genes. Most astounding, the same gene bureaucracy is franchised throughout life, from fruitfly to whale. For example, a nearly identical homeobox self-control sequence (a master-switch gene which turns hunks of other genes on) is found in every vertebrate.

So prevailing is the logic of nonrandom variation that I was at first flabbergasted in my failure to find any biologists working today who still believe mutations to be truly random. Their nearly unanimous acknowledgment that mutations are "not truly random" means to them (as far as I can tell) that individual mutations may be less than random -- ranging from near -- random to plausible; but they still believe that statistically, over the long haul, a mass of mutations behaves randomly. "Oh, randomness is just an excuse for ignorance," quips Lynn Margulis.

This weak version of nonrandom mutation is hardly even an issue anymore, but a stronger version is more of a juicy heresy. It says that variations can be chosen in a deliberate way. Rather than have the gene bureaucracy merely edit random variations, have it produce variations by some agenda. Mutations would be created by the genome for specific purposes. Direct mutations could spur the blind process of natural selection out of its slump and propel it toward increasing complexity. In a sense, the organism would direct mutations of its own making in response to environmental factors. Ironically, there is more hard lab evidence at hand for the strong version of directed mutation than for the weak version.

According to the laws of neodarwinism, the environment, and only the environment, can select mutations; and the environment can never induce or direct mutations. In 1988 Harvard geneticist John Cairns and colleagues published evidence of environmentally induced mutations in the bacterium E. coli. Their claim was audacious: that under certain conditions the bacteria spontaneously crafted needed mutations in direct response to stresses in their environment. Cairns also had the gall to end his paper by suggesting that whatever process was responsible for the directed mutations "could, in effect, provide a mechanism for the inheritance of acquired characteristics" -- a bald allusion to Darwin's rival-in-theory Jean-Baptiste Lamarck.

Another molecular biologist, Barry Hall, published results which not only confirmed Cairns's claims but laid on the table startling additional evidence of direct mutation in nature. Hall found that his cultures of E. coli would produce needed mutations at a rate about 100 million times greater than would be statistically expected if they came by chance. Furthermore, when he dissected the genes of these mutated bacteria by sequencing them, he found mutations in no areas other than the one where there was selection pressure. This means that the successful bugs did not desperately throw off all kinds of mutations to find the one that works; they pinpointed the one alteration that fit the bill. Hall found some directed variations so complex they required the mutation of two genes simultaneously. He called that "the improbable stacked on top of the highly unlikely." These kinds of miraculous change are not the kosher fare of serial random accumulation that natural selection is supposed to run on. They have the smell of some design.

Both Hall and Cairns claim that they have carefully eliminated all other explanations for their results, and stick by their claim that the bacteria are directing their own mutations. However, until they can elucidate a mechanism for the way in which a stupid bacterium can become aware of which mutation is required, few other molecular geneticists are ready to give up strict Darwinism.