There have been plenty of evolutionary surprises in recent years, things that Darwin never would have guessed. The number of genes a species has doesn't correlate with how complex it is, for example. With some 37,000 genes, rice has almost twice as many as humans, with 20,000. And genes aren't passed only from parent to offspring; they can also be passed between individuals, even individuals of different species. This "horizontal transfer" of genetic material is pervasive in bacteria; it's how antibiotic resistance often spreads from one strain to another. Animals rarely acquire whole genes in this way, but our own DNA is packed with smaller bits of genetic material picked up from viruses during our evolutionary history, including many elements that regulate when genes are active or dormant.

Do these surprises challenge the central idea of Darwinian evolution? "Absolutely not," says David Haussler, a genome scientist at the University of California at Santa Cruz. "I am struck with the fact daily that the more information we accumulate, the more validation we find of Darwin's theory." Once new material has nestled into a host's genome via horizontal transfer, the genetic material is as subject to natural selection as ever. Truly one of the most remarkable traits of Darwinism itself is that it has withstood heavy scientific scrutiny for a century and a half and still manages to accommodate the latest ideas. "So far the data sets we've looked at and the surprises we've found show that the essence of the idea is right," Haussler says.

Another growing field of biology is shedding further light on the origins of variation. Evolutionary developmental biology, or evo-devo, focuses on changes in the exquisitely choreographed process that causes a fertilized egg to mature. Behind one series of such changes are the so-called homeotic genes, which dictate where legs or arms or eyes will form on a growing embryo. These central-control genes turned out to be almost identical even in animals as different as worms, flies and human beings. Many researchers now think that much of evolution works not so much through mutations, or random errors, in the major functional genes, but by tweaking the ways by which developmental genes control other genes.

"The building blocks of squids and flies and humans and snakes are stunningly similar," says Carroll, of the University of Wisconsin at Madison, one of the founders of evo-devo. "It kind of upsets your worldview at first," he adds, "but then you see that it bolsters the Darwinian view a thousandfold. These kinds of connections were at the heart of descent with modification."

Carroll says he thinks Darwin would be thrilled with the evolutionary details scientists can now see—how, for example, changes in just a small number of regulatory genes can explain the evolution of insects, which have six legs, from their ancestors, which had even more. From there, it's a short step to solving some of the mysteries of speciation, working out the mechanics of exactly how one species becomes many, and how complexity and diversity can be built up out of very simple beginnings. "I think this is a new golden age of evolutionary science," says Carroll. "But what we're really doing is fleshing out Darwin's idea in ever greater detail."

Perhaps the most surprising discovery in recent years has to do with one of Darwin's predecessors in evolutionary theory. Jean-Baptiste Lamarck, a French naturalist, developed his own theory of biological evolution in the early 19th century. He suggested that acquired traits could be passed along to offspring—giraffes that stretched to reach leaves on tall trees would produce longer-necked offspring. This "soft inheritance" became known as Lamarckism and soon proved susceptible to parody: Would clipping the tail off a rat lead to tailless pups? Of course not, and in time soft inheritance was dismissed, and Lamarck became a textbook example of shoddy thinking.

Then, in the early days of genetic engineering more than two decades ago, researchers inserted foreign genes into the DNA of lab animals and plants and noticed something strange. The genes inserted into such host cells worked at first, "but then suddenly they were silenced, and that was it, generation after generation," says Eva Jablonka, an evolutionary biologist at Tel Aviv University in Israel. Researchers figured out that the host cells were tagging the foreign genes with an "off switch" that made the genes inoperable. The new gene was passed to an animal's offspring, but so was the off switch—that is, the parent's experience influenced its offspring's inheritance. "Mechanisms that were at the time hypothetical proved to be real," says Jablonka, "and of course much more complicated than anyone thought, which is natural."

All sorts of changes in cellular machinery have shown up that have nothing to do with the sequence of DNA but still have profound, and heritable, impacts for generations to come. For example, malnourished rats give birth to undersized pups that, even if well fed, grow up to give birth to undersized pups. Which means, among other things, that poor old Lamarck was right—at least some acquired traits can be passed down.