Why won't you grow my potatoes?" a puzzled Monsanto scientist asked me.

His bioengineered potatoes carry a gene spliced in from a bacterium called Bacillus thuringiensis (Bt), which parasitizes a number of insect larvae. One strain of Bt is particularly fond of the relentless Colorado potato beetle.

In nature, Bt lurks in the soil and gets splashed up onto, say, potato leaves. If a beetle grub ingests Bt while chomping on a leaf, the bacterium multiplies in its gut. Within hours, thousands of Bt offspring have produced a specific toxin that kills the grub. Then the bacteria, having feasted on the grub's dissolved innards, pour out of its body. Not pretty, just nature at work. A potato beetle is Bt's way of making more Bt.

Maybe once or twice a growing season, organic farmers like me spray Bt to control potato beetles if grubs get out of hand - which they don't always do, because we never plant acres and acres of continuous potatoes. We rotate crops to keep these pests from building up. We don't need much Bt, especially since it's alive; once we've applied it, it reproduces itself for a while, until it can't find any more grubs.

Now comes Monsanto, snipping out the gene that tells Bt how to make that specific beetle toxin and sticking that gene into the Russet Burbank potato, the most widely grown potato in the world, the one that supplies all the fast-food fries. The spliced potato, trademarked NewLeaf, makes the toxin constantly in small amounts in every one of its cells.

My friend at Monsanto honestly sees this potato as a wonderful advance, saving organic farmers the trouble of spraying Bt and conventional farmers the danger of spraying beetle-cides. He can't understand why I wouldn't welcome it with praise and rejoicing and use it on my organic farm.

Sigh. The reasons seem so obvious to me. In order of increasing seriousness they are:

1. FOOD SAFETY.
When I spray Bt on my potatoes, its poison gets made only within the guts of beetle grubs. If it gets out onto the potato leaves when the grubs die and dissolve, it quickly washes away. The NewLeaf potato bears the toxin in every cell, even in the tubers we eat. We can't wash it out. The entire plant is a pesticide.

2. COMPANY GOOFS.
Monsanto revealed recently (and quietly) that huge quantities of another of its biotech products, a gene-spliced canola seed, had been mistakenly sold with the wrong gene in it, one that had not been tested or licensed. The problem here is not that companies make mistakes - of course they do - and not that the unlicensed canola gene was necessarily dangerous. The problem is that genetic engineering, like nuclear power, is not an arena where we want mistakes to be made.

3. PEST RESISTANCE.
Whenever a pest comes in contact with a poison, it's possible that a few members of its fast-breeding horde can survive, because they bear some genetic trait that allows them to detoxify, avoid, or defuse the poison. Those resistant pests are the ones that live to produce the next generation. The Colorado potato beetle is second only to the green peach aphid in its acquired resistance to hard-core pesticides. But it has never developed resistance to Bt. Organic farmers haven't blown Bt's cover, because they use it spottily and on the surface of the leaves.

But fields of potatoes carrying Bt toxin inside every leaf during the whole growing season are something new under the sun. I can't imagine a more perfect setup to select for resistance. The Diamondback moth, for instance, is already resistant to Bt's gut assaults. Experts, including Monsanto's own, estimate it will take five to 10 years before the NewLeaf potato will destroy both its own effectiveness and that of a good organic-crop-protection tool as well.

You might wonder why Monsanto would develop a product that is almost certain to render itself impotent. I wondered too, until I came across this quote from a company spokesman: "Resistance is unlikely to happen within five years, and within that time frame we'll offer new technology that will further reduce the likelihood of resistance." Don't worry. We'll destroy nature's tool for beetle control, but you can always come to us for a new one. It will even be a better one. Trust us.

4. FURTHER CONSEQUENCES IN NATURE.
In 1996, Danish scientists watched a gene for herbicide resistance in canola jump the farm fence to enter one of canola's wild relatives. Will the ability to make beetle toxin suddenly show up in, say, wild nightshade, which is a relative of the potato? Or could resistant beetles, no longer held in check by Bt, become more effective pests to other members of the nightshade family (which includes tomatoes, peppers, and eggplants)?

Another possibility is that cut-and-paste genes will travel via the very viruses that the gene splicers use to do their work. Viruses are routinely used as carriers to insert genes into the target cell of the potato or sheep or whatever. Wild viruses may - no one knows - be more likely to pick up a transplanted gene than a native gene. If a cut-and-paste gene has been engineered to fight a virus, that virus (if the gene fails to destroy it) or another one could acquire and spread the gene.

And then there's the already demonstrated food-chain effect. Aphids in Scotland ate Bt potatoes; when the "good guys" - lacewings and ladybugs - ate the aphids, they died from gut upset.

Once a gene has been loosed, it's way beyond our recall or can only be recalled through a massively expensive mobilization of people and resources (think of the smallpox virus).

5. "JUNK" DNA.
Genetic technology sounds like a precise science, but in fact it's primitive and messy. It doesn't matter whether plants are sexually crossed; or seeds are radiated or chemically induced to mutate; or genes are inserted into cell tissue; or chromosomes are shot with new genes by a micro-gun. No one is quite sure where the genes will go, how they will glom on to the chromosome, what sequences of "junk" DNA (meaning DNA that scientists can't see any use for) may come along with the desirable genes, or how the "junk" DNA may influence the cell. What outcomes might there be from this fairly random and uncontrolled process? No one - no scientist, no regulator, no activist hyping these threats, no gene-splicer making light of them - really has the slightest idea.

6. BREAKING THE SPECIES BARRIER.
Nature doesn't normally cut and paste single genes from bacteria to potatoes, toads to petunias, people to sheep. Though the DNA of a sunflower is essentially made of the same stuff as that of a chimpanzee, numerous physical, behavioral, and biological barriers prevent their specific genes from creeping, swimming, or leaping into each other's DNA. Contrary to the claims of biotech companies and consistent with the intuition of everyone else, the various critters within which nature packages various lengths and combinations of DNA do have some boundaries, which presumably have some evolutionary value. Moving single genes from any species to any other is not just a small extension of the age-old human practice of breeding roses or cattle. It's a whole new twist.

7. THE PACE AND THE SELECTION MECHANISM.
For several billion years evolution has proceeded in fits and starts, but generally slowly. In the the hands of biotechnicians, farmers and breeders, the rate of evolution speeds up enormously, and species are selected by their ability to fit not into nature, but into markets.

From a systems point of view these are two of the most profound interventions one can make in a system. Speeding up the rate of change relative to the rate of corrective feedback means a system can't manage itself, nor can it be managed. If there are terrible consequences from even one of our imaginative gene- tinkerings, we are unlikely to find out about it in time to cleanse nature of our mistake. We are not helping this problem by conducting our genetic-manipulation experiments in corporate secrecy, overseen by underfunded and politically compromised regulatory agencies, nor by putting these corporations under such stress from stock-market expectations that they must roll out their experiments by the millions of acres over just a few years.

Even more profound than destabilizing a system by changing it faster than its feedback mechanisms can function is derailing it by setting an entirely different goal around which those feedback mechanisms can "true." For billions of years nature selected species survival according to the ability to thrive and reproduce in the physical environment and in the presence of all neighboring species. For 10,000 years, farmers have selected for what can be manipulated by people in order to feed people. Now the criterion is what can be patented and sold in huge, global-market quantities.

Biotech companies love to talk about feeding the world, but their products must pay off in a market that measures dollar demand, not human need. By far the greatest effort has gone into the potato that makes fast-food fries, not the yam grown by folks with no cash. The corn that feeds America's pigs and chickens, not the dryland millet that feeds Africa's children. The diseases of the rich, not the plagues of the poor. There is some public funding and corporate charity directed toward gene manipulations that might conceivably help feed the world, but the vast majority of minds and bucks are working on caffeine-free coffee beans, designer tomatoes, seedless watermelons. They always will, if the market is the guide.

- Donella H. Meadows