may cost the farmer only about $30 more than the same amount for competing varieties. The farmer then profits by about $70. More broadly, economists have argued that competition between different patented products will often pass the benefit of the inventions on to the consumer; that result is less likely to occur with patents on process innovations.

Most developed nations accept that the benefits of a patent system outweigh its costs. The benefits are less obvious, however, for the developing nations. If there is little competition, the patent monopoly may impose serious social costs, especially when fundamental needs of the poor are at stake. Consequently, many developing nations have excluded pharmaceutical

and food-related inventions from full patent coverage. In many such countries, public sector research is also more important than the private sector research that benefits from patents. Even without a patent system, developing nations can usually still import products of technologies patented in other countries.

The attitudes of developing nations are changing, however. In part the changes are a response to political pressure from the U.S. in international negotiations such as the Uruguay Round of the General Agreement on Tariffs and Trade. (Those talks collapsed last December but are expected to resume.) More nations are also hoping to encourage their own industries, among them biotechnology, in which a

useful and not obvious. By those crite ria, a completely artificial gene may be patentable. If the protein that the new gene makes and the organism into which the gene is inserted are also novel and seem to have desirable qualities, the inventor may sometimes extend the patent claim to include them as well. Patents involving artificial genes and novel life-forms are therefore relatively straightforward.

Yet one of the most political issues in biotechnology concerns the patenting of useful genes found in nature. A wild plant may have a gene for disease resistance, for example, that can be transferred into commercial crops. Because such genes are arguably discovered rather than invented, they may not satisfy the novelty requirement of patent law. It would be ridiculous to ac cuse people of infringing a patent because they naturally carried a protected gene in their cells.

Patents on genes derived from organisms that are native to developing nations would also probably exacerbate the political tensions that already surround the international use of genetic materials found in nature. Imagine the political outcry if a company discovered a useful disease resistance gene in a natural Mexican weed and then sought to patent its use in commercial varieties of maize that would be sold back to Mexico.

The counterargument is that companies should be rewarded for undertaking the difficult task of making natural genes useful. A company faces enormous costs in identifying, cloning and sequencing a natural gene and in inserting it into an organism for commercial purposes. Furthermore, competitors may find it easier to imitate such a gene after the first company has pioneered the technology. This situation, in which the innovation costs are heavy and the imitation costs are slight, is precisely one in which patent protection can be most beneficial as an incentive.

Although the law in such cases is both confusing and changing, it is likely to evolve in a generally reasonable direction. The starting point in the U.S. is a long-standing doctrine that the purified form of a chemical can be patented if the chemical is found in nature only in an unpurified form. This rule permits the first person to isolate a pure protein, or the gene that encodes the protein, to patent it. In contrast, Great Britain defines its stan dards for innovation more severely. Its courts recently held that a naturally occurring gene sequence could not be patented.

Neither the U.S. nor Great Britain, however, is likely to grant a patent for the use of a gene in a species in which it evolved naturally or in a species to which it can be transferred by normal breeding: in those contexts the gene would not be novel. Both governments seem to permit the patenting of novel organisms in which a gene has been transferred from another species. As gene transfers become more common, however, a specific transfer may need to be more difficult or unusual to satisfy the patent law requirements that it not be obvious. These emerging guidelines seem reasonable.

F

air principles still need to be worked out for many of the ques

tions about the scope of a biotechnology patent. The most controversial of these issues-and one being argued before Congress now in the context of patents on animals-is whether a patent's claims should ever encompass progeny. The sale of a patented product used to take it out of the monopoly: the purchaser could use the product in any way without further restriction. Yet organisms are self-reproducing. Breeding or otherwise reproducing a patented organism amounts to copying the invention, which is normally prohibited without the consent of the patent holder.

Suppose, for example, that an inventor develops a way to insert an artificial gene into lambs, with the result that those lambs produce more, leaner meat from less feed. Under current U.S. law, the inventor can obtain a patent that extends the inventor's monopoly to all the lambs bearing the artificial gene. Any farmer who bought such a lamb would be entitled to fatten it (if that would still be the right term) and sell it for meat. It is nonetheless still uncertain whether the farmer could legally breed the animal and raise its offspring without the permission of the patent holder-after all, some of those offspring will have the artificial gene.

That problem has already been settled for plants covered by the plant variety protection act: a farmer does have the right to replant seeds harvested from a protected variety. Some firms, however, are now protecting seeds under the regular patent act, which is also the only one that applies to animals. Under that act, the issue is not clearly resolved and is likely to be contested fiercely by farmer organizations before the courts and Congress.

A realistic policy analysis is essential to ending the standoff between the logical arguments for both sides. Policymakers should also remember that the

expectations of the involved parties may plausibly differ in various cases. The sale of a patented brewer's yeast is meaningless, for example, unless the yeast can be allowed to reproduce during the brewing process. Sales of livestock for the purpose of fattening them, on the other hand, are less likely to include an assumption that the animals will be bred.

In general, however, a reasonable analysis will probably conclude that progeny should be covered under patent law. If the law does not allow patent holders to restrict the replication of their protected plants or animals, some of them will conclude that their most profitable option is to keep the organisms and to market only the ultimate products. Consequently, the production of related agricultural goods would probably become more consolidated under the control of single businesses. Such integration already occurs in the poultry industry, which carefully controls all its breeding materials.

The replication of some organisms cannot be limited practically. There is no realistic way to bar farmers from planting several generations of a protected nonhybrid seed. In those cases, however, it is unimaginable that the sellers of the organisms will not simply expect and permit such use and set their price or licensing fee accordingly.

should be baselines for further innovation is fundamental to patent law. For this reason, the plant variety protection laws clearly specify that protected varieties can generally be used as parents in breeding experiments. The regular patent laws of most nations also provide for research exemptions under which the use of a patented innovation for experimental purposes does not infringe the patent.

The U.S., however, does not have such a general exemption. It has only a narrow statutory exemption that applies to the manufacturers of generic pharmaceuticals and a history of weak statements taken from case law. The basic theme of the case law precedents is that experimentation is permissible only to satisfy academic curiosity and not for commercial purposes.

Such a principle is unfortunate. The progress of breeding and of science in general would be best served if all laboratories, commercial and academic, had the right to experiment more or less freely with patented organisms. That freedom would increase the incentives for follow-up research-and in biotechnology, follow-up research would very likely contribute at least as much economic benefit as the initial breakthroughs.

Again my hypothetical example of a patented lamb can prove useful. The DNA sequence of the artificial gene that affected the lamb's fat-storing properties would normally be disclosed in the patent and would also probably be protected by a patent claim. Under a broad research exemption, competing companies would have the right to experiment freely with that gene and the patented lambs in an effort to understand better the metabolic process at work.

ly different gene that accomplished the same result in a better way. Depending on the details of the claims, the new gene would probably not infringe the original patent, and the further improved lamb could be patented and sold by the second researchers without restrictions.

Another possibility is that researchers might develop a modified version of the patented gene, consisting of the original and a new supplementary sequence, that would permit the lamb to use more kinds of feed. The combined form of the gene probably would infringe the original patent, but the supplementary sequence might be patentable itself. If the combined gene were more valuable than the original, the two companies could negotiate arrangements allowing one or both of them to sell lambs carrying the combined gene, and each company would share in the monopoly profits according to the relative importance of its unique invention.

Under a narrow exemption, none of this research would have been possible without the prior permission of the original patent holder. The would-be follow-up inventor would have been in a very weak bargaining position because nothing new or valuable would have been developed yet.

The subject of variations on a patented property, as in the preceding example, is central to another hotly contested topic: the legitimate breadth of biotechnology patent claims. In some cases, a patent can cover the application of an idea to many different species. The key to permitting such cover

age is whether the idea could reasonably be expected to work more broadly than has been demonstrated in the examples of the patent disclosure.

An illustration of this point is provided by the basic Cohen-Boyer patent on the plasmid method of genetic engineering. The inventors, Stanley N. Cohen of the Stanford University School of Medicine and Herbert W. Boyer of the University of California at San Francisco, discovered a method of inserting genes into cells using microscopic rings of DNA called plasmids. Their patent disclosure detailed only how several different genes could be inserted into Escherichia coli, a common species of intestinal bacteria. Cohen and Boyer nonetheless obtained a patent that conveyed a monopoly over the use of the plasmid technique with a diverse number of host cells, including other bacteria and unicellular organisms. Thus, their rights went far beyond the use of E. coli.

and less settled questions

that the initial innovator should be entitled to claim and prevail against very similar gene sequences that perform essentially the same as the patented sequence. On the other hand, the competitor should be entitled to nearly full rights if the new sequence has any surprising or unexpected properties. At most, the competitor should have a duty to pay for a license from the inventor if the modified sequence includes the same properties as the original one.

Nevertheless, as reasonable as those principles may seem, they offer no guarantees that patent law will always afford innovators broad protection against similar biological products. Indeed, the first case to consider the issue seriously struck down the effort by an inventor for broad protection.

That case was part of a series of suits between Amgen, Inc., and other companies over patents relating to erythropoietin, a hormone secreted by the kidneys that stimulates the production of blood cells. Both Amgen and Genetics Institute, Inc., had filed applications for the patents at roughly the same time. In several decisions the courts eventually ruled that Amgen's patent covers the gene sequence for erythropoietin and a host cell that has been transformed by the addition of that gene but that Genetics Institute's patent covers purified erythropoietin itself. The decision on an appeal of these rulings is pending as this article goes to press.

Amgen had maintained that, according to its patent, its rights extended to any "amino acid sequence sufficiently duplicative of that of erythropoietin to allow possession of the biological property of causing bone marrow cells to increase production of reticulocytes and red blood cells and to increase hemoglobin synthesis or iron uptake." In ruling against Amgen, the court reasoned that the company had not disclosed enough about the "sufficiently duplicative" sequences to make such comprehensive claims.

That decision kept patent claims nar