Erudite essays that explore the pros and cons of reductionism in science. There have been rumblings in the halls of academe that reductionism--in which the whole is explained by a dissection of its parts--will not remain the dominant mode of doing science. Biologists speak of the need for ``integrative biology''; medicine has its adherents of ``holistic'' approaches. To explore the issue, Cornwell, director of the Science and Human Dimension Project at Jesus College, Cambridge University, assembled a stellar cast of scientists and philosophers for a 1992 symposium. Physicist Freeman Dyson leads off with an essay describing scientists as artists striving against a given culture; he also stumps for multiple visions rather than a single vision of science. There follow a number of essays describing how the turn-of-the-century vision to reduce all of mathematics to a few pure axioms was blown out of the water by the intricate theorems of Kurt Gîdel. Astronomer John Barrow casts doubt on theories of everything and the concept that the universe is a continuum, while Roger Penrose states themes that are later repeated in a series of essays on neuroscience and artificial intelligence. These have to do with whether mind equals brain, whether the brain is a computer, and what is meant by computability. There is steep sledding with Nobelist Gerald Edelman and colleague Giolio Tononi's exposition of the theory of neuronal group selection--steep enough to require a following explanatory essay by none other than Oliver Sacks. The upshot is that, with the exception of a couple of spirited voices championing reductionism, the authors declare, ``The king is dead, long live the replacement''--which Edelman, in a concluding essay far more pithy than his first, describes as a ``second'' enlightenment that can celebrate human freedom. Some nuggets here for philosophers of science, neuroscientists, mathematicians, and computer folks--but one wonders if maybe Cornwell didn't stack the deck a bit and if a different cast might have come up with a different consensus.