IN SCIENCE, IT DOESN'T REALLY COUNT if you see it coming. The genuinely novel discovery is the one that takes researchers by surprise. Who knew that in 1964 a couple of unknown astronomers, trying to silence the static coming out of a radio telescope--static they suspected was caused by pigeon droppings--would discover the echo of creation? The crackle turned out to be microwave radiation that bathes the cosmos, and was the first solid evidence that the universe began in a big bang. And who knew that, after vaporizing some carbon in 1985, chemists would discover a strange spike on the readout of molecular weights in their sample, and that it would be buckyballs? These previously unknown soccer-ball-shaped molecules of 60 carbon atoms, and their "buckytube" cousins, promise to revolutionize materials science, producing structures 100 times stronger than steel but with only one sixth the weight. Who knew?

Which is not to say that novel ideas pop out of nowhere. It's just that it's easier to retrodict than to predict. In other words, once the breakthrough breaks through you can usually identify its intellectual roots. Even Darwin's theory of evolution can be traced back, to work by Jean-Baptiste Lamarck and Georges Buffon, who had proposed that species change through time. They just didn't know how. "If you search the scientific literature, you can reconstruct the ancestry of any idea," says linguist Frederick Newmeyer of the University of Washington. "But that doesn't mean that if you asked linguists, the year before Noam Chomsky published his theory of deep grammar, what might happen, that they would have predicted it."

Despite the difficulty of predicting, researchers in sciences from genetics to astrophysics to linguistics all consult crystal balls for one pragmatic reason: hitching one's fortune to the Next Big Thing is the surest way to keep the grant money flowing. Some of what scientists foresee is sure to come about. The 21st century will see gene "pharmers," for instance. By the year 2000, biotech firms will be splicing human genes into the DNA of female rabbits. The genes will trigger production of certain proteins whose absence causes fatal disorders. People born without those crucial genes can then obtain the necessary proteins with a dose of rabbit's milk. The new millennium will also bring the discovery of genes for specialized bits of language. Already, researchers have found a genetic mutation that shows up as an inability to put suffixes onto words: people who carry the gene cannot add "-s" or "-er" or "-ed" to words, explains Newmeyer. "In the next century we will locate other aspects of language in the genes," he believes. Could a gene for the subjunctive be far behind? Next time you don't know whether it's "if she was" or "if she were," you'll be able to blame your DNA.

Wilder forecasts have stiffer odds. Will the 21st century see nano-assemblers? These microrobots would break down the chemical bonds of cheap ingredients--grass and water, say--and reassemble the carbon, nitrogen, hydrogen and other molecules into, for instance, a sirloin steak. You scoff? It is not much more incredible than a cow's ability to do the same. And scanning tunneling microscopes can already manipulate single atoms, which is what the assemblers would do. Will the millennium bring ways of downloading the contents of a human mind into a computer? Our memories, personalities and thoughts are but bits and bytes of information. All of those could conceivably be transferred to a CD-ROM. And copied. Or slipped into a robot.

Long before that identity crisis ("Which CD is the real me?"), we will know who we are on the molecular level. The Human Genome Project will be completed six or seven years before the original target date of 2010. The $3 billion undertaking will have produced the molecular blueprint for a human--that is, the exact sequence of the four chemical letters that make up each one of our 80,000 genes, and the exact location of each of those genes on the 23 pairs of chromosomes. (The string of letters would take up no more than 750 megabytes of your hard drive, less with a good data-compression program.) And then what? "Just as the discovery of the periodic table of the elements in the late 19th century set the table for the 20th, giving rise to everything from the chemical industry to theories of quantum mechanics," says geneticist Eric Lander of the Whitehead Institute for Biomedical Research in Cambridge, Mass., "so the genome project will lay the table for the 21st century."

Though not right away. The output of the genome project-- a string of 3 billion chemical letters--will at first be as illuminating as a broken flashlight. "What we need is a Yellow Pages, a directory that tells us the job of each gene," says Gerald Fink, director of the Whitehead. One way of determining what a gene does is to compare its sequence of chemical letters with genes of simpler creatures, like yeast, using powerful search engines to scan computerized DNA databases. A match will show, for instance, that this gene builds receptors for neurochemicals. Another way to figure out what a gene does is to discover what protein it makes and what that protein looks like, says Peter Kim of the Whitehead. This is called "the protein folding problem." Once scientists know the job of every gene, they should be able to discover the causes of many diseases. "You can take this huge genetic database of, say, people with heart disease," says Lander, "and determine which form genes associated with the heart have taken. That will tell you what went wrong, and then you can make therapies based on that."

The genome project will produce the blueprint of a "standard" human (in fact, it analyzes DNA from 10 people). But what about genes that make each person unique? "Once the genome project is completed we'll be able to find how an individual's DNA differs from the reference genome," says Francis Collins, director of the National Institute for Human Genome Research. You'll know that you're sensitive to cigarette smoke but not to eggs and bacon, to hair dye but not to pesticides. The tests will be performed by a single silicon chip like Affymetrix's "GeneChip" (page 64). "And if we sequence other organisms," says Collins, "we will find out what makes us human. Our DNA differs from chimps by just 1 percent. But which 1 percent? And why does it produce such drastic differences?"

Throughout the sciences, the years 2000 and beyond promise answers to old questions and new--and also to questions that no one has even thought of yet.

Cosmology: With data taken by satellites set for launch in 2001 and 2005, "I think we will know whether the universe will expand forever" or, one day, come crashing back on itself, says cosmologist Craig Hogan of the University of Washington. The fate of the cosmos has plagued scientists ever since Edwin Hubble discovered, in the 1920s, that the universe is expanding. "And if we're lucky," says Hogan, "we will know what most of the universe is made of." Most of the matter in the cosmos--upwards of 90 percent--is dark; it does not shine and so is not detectable by any telescope. Astronomers have no inkling what it is. Well, actually they have many inklings; when there is a dearth of data, theories rush in to fill the data vacuum. Is the missing matter baby black holes? Jupiter-size planets? Hypothetical particles called WIMPs?

Particle physics: At CERN, the European physics lab outside Geneva, experiments smashing protons into protons will create, in 2005, conditions not seen since a few fractions of a second after the creation, says physicist Chris Quigg of Fermi National Accelerator Laboratory. In the detritus created by these collisions at the Large Hadron Collider, physicists hope to find a never-detected particle called the Higgs boson. Just after the big bang, 12 or so billion years ago, when all the forces in the cosmos were one unified, symmetric field, the Higgs supposedly acted like a sledgehammer on a mirror, shattering the perfect symmetry and doling out this shard of mass to one class of elementary particles, like electrons, and that shard to others, like quarks. At least, physicists think it did. If they don't find it, their standard model of forces and matter will be in trouble.

Gravity: When black holes collide, the calamity should flood the cosmos with gravitational radiation, physicists predict. The waves would create ripples in the very fabric of spacetime. They would make space itself contract or expand. No one has yet detected such a wave, but in 2001, physicists may catch such a wave at two elaborate systems of mirrors called LIGO, one in Louisiana and one in Washington state. LIGO should detect when a gravity wave makes space shrink by a little bit.

Neuroscience: New brain scans will take almost instantaneous pictures of brain activity. "It will tell you what happens, step by step, when a schizophrenic has a psychotic episode, and when you imagine something or recall a vivid memory," says neuroscientist Bruce McEwen of Rockefeller University. And new MRIs, hardly bigger than a salon-size hair-dryer, will tackle the most challenging question of all, says brain-imaging pioneer Marcus Raichle of Washington University: how the brain develops. The MRIs will safely image the brains of children, and show step by step how the most complicated info-processing device in the world becomes wired. Not far off: using the discoveries of neuroscience to retool education. Does musical training prime the synapses for learning math? How do some brains hold thousands more vocabulary words than others? The answers may point the way to new teaching techniques.

Chemistry: No crystal-ball gazing would be worth the glass it uses without a prediction of "making life in a test tube." Nobel laureate Dudley Herschbach of Harvard University foresees making molecules that self-assemble and self-replicate, sometime in the next 35 years. Biochemists are close to doing it. And they have a good idea of what to do with these creations: make those turn-grass-into-sirloin nano-assemblers. In 1997 that seems like so much science fiction, while genetic discoveries, for instance, seem like sure bets. But sometimes in science, the dark horse comes in before the favorite.

PHOTO (COLOR): THE NEXT BIG THING: Peter Kim grapples with the 'protein-folding problem'

PHOTO (COLOR): MORE THAN MEETS THE EYE: The Milky Way and other visible bodies make up less than 10 percent of the cosmos

GENECHIPS, AHOY! READING YOUR DNA

Affymetrix's GeneChip system may one day unravel a person's unique genetic code. The chip can identify, quickly and cheaply, which forms genes take--such as those that protect against or cause cancer. Even farther in the future, it may identify genes that let you eat red meat safely or those that mean you should stick to tofu.

1 DNA is extracted from a blood or cell sample, separated into single strands and treated with a flourescent dye. It is then put on the GeneChip for testing.

2 The GeneChip contains 64,000 squares ("features"), each holding millions of identical strings of DNA. Each square holds a different set of these identical DNA strings.

3 The cartridge is inserted into an analyzer. A laser scans the GeneChip, causing colorful flourescence in a sea of dark blue wherever a DNA match occurs.

4 If the sample DNA matches the strings of known tumor-suppressor gene mutation that is built into the chip, it could mean that the patient is at greater risk for cancer.

DIAGRAM: GeneChips, Ahoy! Reading Your DNA

A CASE OF RUMPLED GENES

DISPATCHES FROM THE FUTURE

Dec. 15, 2012 Anxious college-bound students sent off the last of their GPs (Genetic Profiles) and MRI/PET (brain) scans to universities just as the application window closed. After slowly phasing out the old SATs and Advanced Placement scores starting in 2006, admissions offices now require GPs and MRIs for all applicants. But the huge demand for tests overwhelmed genetic and neuro labs. Rumors were flying that some labs had provided inaccurate profiles. Students carrying the gene for the neurochemical receptor that governs risk-taking were instead identified as carrying the alcoholism gene, complained one distraught father in Palo Alto, Calif. Those carrying a receptor gene that causes neurosis were reportedly labeled as carrying the gene for placidity; according to unconfirmed rumors from MIT, the admissions office there is concerned that its incoming freshmen will not be as angst-ridden as they need to be. At Juilliard, deans suspect that some students admitted to its theater program do not carry the extroversion gene at all; the school may now reinstate the requirement for auditions.

The brain profiles were reportedly even more riddled with error. Students whose frontal-cortex activity on PET scans during math tests had always been low--indicating that their brain needed to exert little effort to crack the problems--were mistakenly labeled as having high activity, a sign of neurons struggling.