Hans Larsson is a fast walker and a fast talker. You need to be fit to keep up with him on the hills of the McGill University neighborhood in Montreal, let alone on the remote islands of the Canadian Arctic where he searches for fossils in summer fieldwork. He talks the way he walks, in a freely swinging fast-paced lope that ranges from the philosophy of science to genetic probes to the rich Cretaceous ecosystem he is exploring at another field site in Alberta.

Larsson is at the forefront of merging paleontology and molecular biology in an effort to connect major evolutionary changes—the development of new species and new characteristics, new shapes and structures, new kinds of animals—to changes in specific genes and in how those genes are regulated. He is interested in reactivating dormant genes or changing the regulation of active genes in embryos to bring back ancestral traits that have been lost in evolution.

Scientists can do this now because we have the fossils. We have the lessons of developmental biology. And we have the tools of molecular biology. All of these are being merged in the study of the history of life in evolutionary developmental biology, or evo devo.

Collecting and cataloging fossil bones, the heart of vertebrate paleontology, has been primarily a historical enterprise, one of unearthing ancient information and looking for patterns. Laboratory science has been conducted in a different fashion. You could suspect, say, that a gene that controls a particular growth factor is important in how five-fingered hands develop at the ends of arms. So the hypothesis might be that if that gene were absent or nonfunctional, the hand would not develop. You can engineer mice so that the gene is absent or silenced and see what happens to the development of the embryo. If the hand develops perfectly, your hypothesis is false. If the hand does not develop, you have good evidence that the gene in question does what you thought it did.

What about a significant change in form? What happens that allows something new to be introduced that hasn’t been seen before in evolution, something like the appearance of limbs, or hair, or feathers, or lactation? We can see it happening on a gross anatomical level throughout the fossil record, but how can we test in the laboratory our ideas about what went on at the molecular level?

There is a way that these ideas can be tested, but it takes us to the very edge of science. The trick is to run the tape of evolution over again: to intervene in the development of a chicken embryo, for instance, to reverse evolution, rolling back the clock to manifest earlier patterns of gene expression. This is a profound advance in the kind of experiment available to test evolutionary theory, and it depends entirely on the progress that has been made in evolutionary developmental biology. It is only because we can match developmental events to evolutionary events, only because we now have both the fossil record, which shows us the path that evolution has taken, and the developmental record of modern animals in extraordinary detail, that we can begin to link the two.

With birds, for instance, the absence of a tail, the difference between wings and grasping forearms, and the absence of teeth are all subtle evolutionary changes on a basic dinosaur plan. Why not rerun an ancient developmental process in a bird embryo, this time triggering the signals to produce a grasping forelimb or a long tail?

At last, we would have a truly experimental way of studying macroevolution, the kind of changes that lead to the creation of new species. If we think a particular process is responsible for the origin of feathers, we could rerun that process, changing the developmental pathway in the bird embryo as a way to test our proposed evolutionary pathway.

But why stop at one developmental process? Why not grow an entire dinosaur? At least that’s the thought that came to my mind. It seemed so obvious that if fairly small changes in development, which adjusted the timing and concentrations of growth and signaling factors, could have led to the evolution of birds from nonavian dinosaurs, we could readjust those changes in development and get a dinosaur from a chick embryo. Dinosaurs and chickens share a very similar skeleton. In the grand scheme the differences between the two creatures are just adjustments to a basic body plan, and such adjustments, we are assured by evo devo, are the result of changes in gene regulation, not a complete new suite of genes. Once I got the idea in my head that it could be done, I started talking to researchers who were truly fluent in the language, ideas, and techniques of both paleontology and molecular biology. That’s how I ended up talking to Larsson.

He was already researching how the tail in birds first got shorter and then disappeared over the course of evolution. I thought, why not look at it from the other direction? Suppose we were going to try to bring back the tail—what would we do then? I gave about 40 thousand dollars to pay for a postdoctoral researcher to work on this problem in Larsson’s lab. Larsson is continuing to pursue the research on his own now.

A very few experimental attempts to achieve this kind of thing have been done. Chicken embryos have been induced to grow teeth without any transplant of tissue from another organism. Changes in the presence of growth factors produced the teeth, which were consistent with those of archosaurs (the group that includes birds, nonavian dinosaurs, and crocodilians).

Primitive birds, like Archaeopteryx, had tails, but modern birds don’t. It seems a good bet that this was a simple evolutionary change at the molecular level, a turning off of the growth program that was keeping the tail going. Find the chemical switch, flip it the other way in embryonic development, and the result would be a bird with a tail. Larsson determined that in chicken embryos, a tail does begin to grow in the normal course of events. But then, he said, “I found that at a particular stage of development everything comes crashing to a halt.”

Larsson and his postdoctoral assistant began to try to make an embryo’s tail continue growing. They began snipping off the tip of a growing tail at one stage and stapling it, with fine tungsten wire, to a later stage, to see if the growth factors in action during early tail growth could override the stopping signals. The next step was to add retinoic acid, a compound that stimulates the release of a protein called sonic hedgehog, which is important in promoting growth during development.