An outgrowth of genetic engineering, the technique has been branded "pharming." Rather than manipulating plants to make a firmer tomato or a sweeter peach, "pharmers" insert genes that instruct a plant to manufacture pharmaceutical compounds. In the future they envisage, flu shots will be replaced by bananas. Prozac, anyone? Try this corn puff. Pharmers dream that all drugs will ultimately be delivered in snacks.

First out of the pipeline will be vaccines. In August, industry leader ProdiGene began Phase I clinical trials for a vaccine against traveler's diarrhea. Resulting from a dismal species of E. coli, the condition is also a prime cause of infant mortality in many poorer nations. Instead of pursuing the regular path of cell-culturing and purification, Texas-based ProdiGene hopes to deliver a vaccine in a simple kernel of corn.

To understand what is at stake here, consider the case of hepatitis B. Worldwide, that virus kills more than 900,000 people a year, many of them in China, where the disease is at almost epidemic levels. A dose of hep-B vaccine costs around 50 cents, yet even that -- in quantity -- is beyond the budgets of many developing countries. Besides the cost of the drug itself, vaccination is hampered by the additional expense of needles and by lack of refrigeration. Vaccines produced the traditional way cost thousands of dollars a gram, but corn can be grown for 5 cents a pound. Hoping to tap into the huge hepatitis market, ProdiGene is currently conducting field trials on a strain of transgenic corn that has been spliced with hep-B antigens.

ProdiGene even has its eye on AIDS. Two years ago the company received a $300,000 grant from the National Institutes of Health to research the possibility of a plant-based vaccine against the HIV virus. As a test of the concept, company scientists are beginning with the primate version, simian immunodeficiency virus, by splicing into corn the SIV genes responsible for producing a protein called GT120, which is known to trigger antibodies against the deadly invader. The NIH itself will conduct the clinical trials, and if all goes well it will move on from there to a human version.

In theory, just about any drug could be engineered in a kernel of corn or a grain of rice. Earlier this year ProdiGene announced it was scaling up transgenic production of aprotinin, a protease inhibitor used in cardiac surgery. Traditionally extracted from bovine lungs, aprotinin reduces the need for blood transfusions in patients undergoing bypass surgery. Once the initial splicing is done, such transgenic crops can be grown anywhere, from Nebraska to Nigeria.

Pharmers are also looking to actual farms. Every year millions of animals have to be vaccinated at enormous expense to their owners and considerable stress to the creatures. Transmissible gastroenteritis virus (TGEV), for example, is a highly contagious disease that kills infant pigs. Clinical trials by ProdiGene have shown that in principle plant-based vaccines can be effective against this pathogen. As with many biotech products, the development of transgenic vaccines will be driven initially by the demands of animal husbandry.

To protect our health, we have fluoride in our water and iodine in our salt; why not deliver codeine in corn flakes, Wellbutrin in Ho Ho's? The problem, says Norman Ellstrand, a plant geneticist at UC Riverside, is that transgenic crops are incredibly difficult to isolate. Scientists now know that genes are routinely passed among plant species, and "gene flow" from genetically manufactured (GM) organisms to wild varieties has been documented all over the world. That's bad enough when a gene involved conveys herbicide resistance, but when you're talking about genes for proteins and hormones, the potential for disaster is enormous -- both for human health and the environment. "We need to be assured of zero tolerance," Ellstrand says, but that's almost impossible to guarantee.

Genes aren't the only things that are hard to contain. Remember the Starlink debacle, when GM maize intended for animal feed found its way into taco shells? Imagine if the modified grain had contained a drug -- any drug. With vast amounts of grain being shipped around the world, Ellstrand believes it won't be possible to prevent such mix-ups. Sooner or later, innocent folk chowing down on corn chips or sesame buns are going to find their bloodstreams coursing with aprotinin or swine vaccine or God knows what else. According to Jane Rissler of the Union of Concerned Scientists, "The food industry is apoplectic about the possibility of this stuff getting into the food supply." Rissler and Ellstrand argue that pharming should be strictly limited to nonfood crops -- to, say, tobacco or castor beans.

IN THE DEVELOPED WORLD, WHERE DRUGS CAN be delivered in so many other ways, it seems hard to justify the risk of pharming -- as with nuclear power, we really do have alternatives. But in the developing world, millions of people die each year from preventable diseases for lack of very basic drugs. That at least is the argument pushed by the biotech industry. Yet just as people are challenging the GM solution to Africa's food crisis, so, Rissler says, the Third World's health problems are not going to be solved by cutting-edge technology.

The idea of helping the Third World with transgenic vaccines is little more than "a ruse," Rissler believes. "It's selling biotechnology on the back of the poor," by attempting to make it palatable to well-off folks like us. Rissler points out that to be medically effective drugs have to be delivered in the right dose. How would people know how much they were supposed to eat? A whole banana, half a banana? Who's to say? More critical, how could you be sure that people wouldn't overdose? How would you even know you were eating the right variety? After all, a genetically modified banana looks the same as a regular one. Rissler is skeptical of the medical miracle promised by companies like ProdiGene and suspects that a lot of the blue-sky ideas being bandied about will "never see the light of day as commercial products."

Behind the hype about cheap drugs, Rissler and Ellstrand note that the pharming industry is quietly pursuing a much bigger goal -- engineering into plants genes that encode for all manner of industrially useful compounds, from enzymes to solvents. Since these don't qualify as drugs, they are not regulated by the Federal Drug Administration, and very little information is publicly available about what is going on here. Earlier this year the U.S. Department of Agriculture updated its guidelines for industrial pharming, but many scientists believe these are grossly inadequate.

USDA spokesman Jim Rogers acknowledges that "Nobody's going to know all the possible risks. But, he says, "We mitigate these risks to what we feel is appropriate." In the department's view, "There are adequate safety provisions in place." Not according to Rissler, who opines that "The USDA's oversight is way too lax." Given the enormous potential dangers, Rissler insists there ought to be external scientific oversight as well. What most appalls Ellstrand, who sat on a National Academy of Sciences committee that reviewed the regulations for GM crops, is that companies do not have to disclose what genes they are adding, or even what organisms the genes derive from -- that's "confidential business information."