The stained bands of the samples to be compared are lined up and compared by eye. The eye is an unreliable instrument, which gives plenty of room for error. Juries, typical as they are of the population as a whole, are bad at understanding risk (which is why the National Lottery does so well) and are much more likely to convict if told that the chance of a random match between defendant and sample is o.i per cent than the (identical) figure of one in a thousand. These arguments are the stuff of legal dispute and are no different from the controversies about other forensic tests that hit the headlines. However, forensic genetics faces a deeper problem that arises from evolutionary history.

DNA fingerprints are made up of short sequences of the message which are repeated again and again. The number of repeats and the position in which they occur varies from person to person. A sample from the scene of the crime is compared with one from the suspect and with others from a panel of innocent donors. Rather like an identity parade, witnesses pick out the criminal from a group known not to have committed the crime.

In the earliest days of ONA fingerprinting the FBI set up a reference group of donors made up of white police officers. To some jurors, if the suspect's fingerprint was more similar to that at the scene than to that of each member of the panel, the case seemed indisputable.

This simple approach faces an evolutionary problem. If an eyewitness had seen — say — a white man committing a crime, and then had to pick out the alleged criminal from an identity parade of blacks, legal eyebrows would be raised. The ethnic group of any suspect must be matched with that of the group with which he is compared.

DNA fingerprints evolve quickly. Those from people of African ancestry are somewhat different from those of Europeans (although the overall racial divergence for this character is not much greater than that for enzymes and blood groups, with nine-tenths of total diversity due to differences among individuals). Imagine a black suspect who is wrongly accused of a crime in fact committed by another black man. His DNA fingerprint is compared to that left at the scene and to those of a panel of whites. Genetic divergence between blacks and whites means that the innocent suspect's DNA may be more like that of the criminal than that of any European; and the innocent man is found guilty.

This has led to controversy in the world of DNA fingerprinting and it is right that it should. In the United States, where legalised murder by the state is common, the issue is one of life and death. The rule in American courts is  that scientific evidence may be rejected if it is not generally accepted in the scientific community. Appeal courts threw out convictions for murder and rape because they are not satisfied that DNA fingerprinting is 'generally accepted*. Now, the scientists are ahead, with a survey of individual variation in DNA sequence so extensive that the small racial differences pale by comparison. Even so, the tale of' genetics and the law is another reminder that objective knowledge can soon be hijacked by those with a subjective view of how it should be used.

People from different parts of the world may differ but the idea of pure races is a myth. Much of the story of the genetics of race, a field promoted by some of the most eminent scientists of their day, was prejudice dressed up as science; a classic example of the way that biology should not be used to help us understand ourselves. Most of today's biologists feel that the moral issues raised by our own biology — racism, sexual stereotypes, and claims that selfishness, spite and nationalism are driven by genes — are issues of ethics rather than science and that science has nothing to do with how we perceive our fellows. Although it may comfort the liberal conscience to find that genetics reveals few differences among the peoples of the world, this is irrelevant to the issue of racism, which is a moral and political one.

As a result, those determined to dislike one race or another are not much impressed by scientific arguments. I once gave a lecture on race when I was teaching in Botswana. The class was delighted to learn that they were almost the same as the white South Africans who so despised them. At the end of the lecture there was just one question. Surely, a student asked, what you are saying can't be true of the Bushmen; they are obviously different from us.

I admit to a certain despair at that; but it was a useful reminder that although biology may tell us a lot about where we come from it says nothing about what we are. The dismal history of racial genetics strengthens that belief.

Evolution is now a practical subject in its own right although many who use it do not realise what they are doing. Inventors once used an approach close to that of the natural world. For gadgets and life, tinkering works; and can be the means to an unexpected end. Just like the engineers who designed stone tools or steam engines with no understanding of physics, the first farmers developed new crops with no knowledge of heredity at all. Pragmatism led, as always, to progress.

Nowadays, technicians in concrete or metal have a different attitude. They design what is needed with as much scientific theory as is necessary. Applied biology, from agriculture to medicine, has adopted this approach only in the last few years and has begun to advance as much as has transport since Stephenson's Rocket. For biology, a new steam age (albeit not yer a space age) is upon us.

A fusion of Mendelism and Darwinism has made agriculture much more productive. The amount of food available per head, worldwide, has gone up in the face of the greatest population explosion in history. In the developing world there is still room for progress as half of all crops are lost to weeds (a figure last seen in Europe in the Middle Ages) and disease can lead to the loss of entire harvests. In Africa, indeed, such is the rate of population growth that — against the world trend — the amount of food produced per person is decreasing. Third-world fanning has a long way to go before it catches up. General economic weakness is much to blame, but some of its failure is because it lacks the technology used elsewhere.

Darwin or Mendel would each feel quite at home with most modern agricultural research. In Illinois in 1904 an experiment started in which, each generation, the maize plants most rich in oil were bred from. The work still goes on and, a hundred generations later, the amount of oil has gone up by a hundred times with no sign of any slowing of progress.

Such straightforward applied evolution can do remarkable things, as any cattle-breeder can attest. The 'Green Revolution' took a step further down the genetic road. Its success came from crosses between new and productive stocks of rice and wheat, bred in the Darwinian way, and other lines with stiffer and shorter stalks. Just a few genes were involved. Dwarf varieties were crossed with others with rigid stems. Their descendants were mated with stocks that contained genes for high yield and rapid growth. Plants which combined the best qualities of their parents were chosen and the process continued for several generations. Sex — genetic recombination — did the farmers* job by making new mixtures of genes. It solved a major problem of tropical agriculture, the tendency of rice and wheat to grow tall when fertiliser is used, but to fall over in high winds. One simple trick transformed the rural economies of India and China. In fifty years, planned gene exchange gave a six-fold boost in yield, a figure as great as that at the origin of farming ten thousand years before.