Evolution is based on the idea that random genetic mutations produce changes in individuals. If these changes are beneficial, the genes spread to the rest of the population. Over time, therefore, the species "evolves". More recent versions of the theory suggest that evolution takes place in isolated pockets of individuals, where it is easier for a gene to spread throughout the whole population.

Unfortunately, while this explanation looks good in overview, it falls to bits as soon as you look too closely. It is not enough to suggest ways in which an eye or a wing could have evolved in stages. One must demonstrate that such intermediate stages actually existed, and also that they were beneficial.

But the fossil record does not show intermediate forms, nor does it show a continuous variation in species. Instead, it shows clearly demarcated species, with nothing in between. Nor is there any evidence for evolution taking place today. Animals fall into discrete species, with no intermediates.

Different races of dogs may seem to have "evolved" through breeding; however, all dogs are the same species. All that has happened is that groups have been bred which lack some of the dominant genes present in the natural population, thereby allowing recessive genes to be expressed. Nothing new has been created.

It is not surprising that there are no intermediate forms. What use is half a wing? Or an eye with no optic nerve? Or an optic nerve that doesn't quite reach the eye? How can two independent structures evolve piecemeal when neither is useful by itself and each is worse than useless until it is fully functional? Darwin himself admitted that the evolution of the eye seemed "absurd in the highest possible degrees."

But evolution requires much more dramatic changes. How could an animal who lays eggs evolve into one who keeps her offspring in a uterus? A uterus would not evolve in an egg-layer, since it would be a huge disadvantage to have one. Conversely, an animal without a uterus could not give birth to viable offspring. But an animal must either lay eggs or nurture offspring in her womb: she cannot do a bit of both.

Moreover, all these changes are supposed to come from random mutations. Random changes in a computer program almost never produce improvements. For the same obvious reasons, genetic mutations are virtually never beneficial. How many mothers hope that their child will have a genetic abnormality?

Another serious dilemma for evolution is that there is no way by which lessons learnt during one animal's life can be passed on to the next generation: the chromosomes in the sperm and egg cells are already formed before the individual is born, and these are either passed on intact or lost completely.

How could any kind of complicated behaviour develop in such a system? This would be like developing a massively complicated computer program by getting someone who could not read or write to add pieces of code at random, and then testing whether the program as a whole was better or worse after the change.

Termite structures present an even bigger challenge. Termites live underground in colonies. Some species grow fungus in underground farms, and build elaborate structures above ground to provide ventilation.

The animals face two problems: first, how do they know the overall structure of the mound? They are blind, and cannot realistically understand its purpose, let alone the complicated relationship between its structure and its function. This would be like a group of blind humans constructing a skyscraper with no plans, no tools, and no apparent means of communication.

One of the biggest problems is co-ordinating each other's activities. The South African naturalist Eugtne Marais showed that if you stick a steel plate through a termite mound, the termites are able to rebuild the structure correctly. When the plate is removed, columns and arches match up perfectly. Yet neither side can see, smell or touch the other.

It is difficult to see how pheromones would be enough to pull this one off. So what are we left with? The idea of a Christian God who is all-loving and all-powerful hardly fits in with the mess that we have today.

However, if one separates out the dross (created in six days) from the rest (created by an intelligent agent) one comes close to creation myths found in many other religions.

When a common idea is found in religions which developed separately, one needs to take note. The thought of more complicated species evolving from simpler ones seems to make sense. What is missing is the mechanism for directing this. If there is no external God, perhaps this self-organisation is an emergent property of complicated systems, which results not from any external reality, but as part of the nature of matter.

In a sense, God would evolve bit-by-bit. Rupert Sheldrake, a dissident scientist from Cambridge, has a theory which works much like this. According to Sheldrake, when things happen, similar things are more likely to happen because a Morphogenetic field has been created. A good example of this is that when a new chemical is first synthesized it is difficult to crystallize it. However, when it has been crystallized once, anywhere in the world, the same chemical is easier to crystallize, everywhere in the world.

If we accept the idea of Morphogenetic fields, the evolution of behaviour becomes much easier to explain. Spiders of a particular species are linked into a Morphogenetic field for creating the appropriate kind of web. The Morphogenetic field will change, based on the experiences of individual spiders. This means that the experiences of individual spiders can be passed down to future generations. If a piece of behaviour is inappropriate, spiders who follow that field are less likely to survive.

So Morphogenetic fields for inappropriate behaviour will gradually die out. In contrast, a spider who latches onto a Morphogenetic field coding for a successful piece of behaviour will live to repeat that behaviour, thereby reinforcing the field. Gradually, the best fields become the strongest ones, and the animals follow these guidelines automatically.

This would work like evolution. However, unlike evolution, the organizing principle lies outside the animal. In addition, it can learn from experience and can pass on much more complicated things, such as having a mental model of a web.

Because there is a link between the structure of the animal and the structure of the Morphogenetic field, spiders will automatically latch onto fields appropriate to their own species. Morphogenetic fields would also help to make sense of termite behaviour. Marais showed that when the queen termite dies, activity in the colony ceases.

This suggests that there is a Morphogenetic field associated with the queen which co-ordinates the activity of the individual termites.

What has this got to do with God? Well, in the more advanced religions, such as Taoism, a common organizing principle is the non-local nature of reality - or, the "oneness" of creation. This isn't such a big jump from a Morphogenetic field.

Rupert Sheldrake, the mavarick scientist at Cambridge University whose theory of morphogentic fields challenges our dearest assumptions about evolution.