The thought police: Brief Answers To Big Questions- 2. Can roses grow from thistles?

When botanists say that the rose is related to the thistle, are they reporting a discovery or making a decision? That is, are they telling us about the real nature of the world, or are they choosing to group certain things together because doing so is - to put it bluntly - useful?

If you think the former, you believe that things have an inner nature, which makes them what they are independently of our knowledge: on this view a rose and a thistle are essentially related. If you think the latter, you regard our taxonomies as the projection of our intellects: we class roses and thistles together because practical and theoretical convenience so invites.

These philosophical questions are freshly in the news. Scientists at Kew have recently turned botanical taxonomies upside down by taking DNA as the criterion for plant classification, in place of the system - originated by Linnaeus - which relies on appearances. Linnaeus's system says that one of Buddhism's holy plants, the lotus, is related to the water lily; DNA comparison says it is related to that familiar London inhabitant, the plane tree. Other such dramatic surprises abound in the new system.

Using DNA as the criterion seems to square with the view that botanical taxonomies succeed in carving nature at its joints, quite independently of human needs and interests. But does it? There are some problems in the way of thinking so.

The members of any collection can be grouped together in different ways, depending on the criteria chosen. One can for example classify books according to subject, or author, or title, or even colour or size. Which classification you choose depends on your needs: are you arranging a library, or trying to store volumes in an oddly shaped space? Scientific classifications appear less arbitrary because of their relation to scientific laws: they can be tested by how well they function in theoretical and experimental ways. But even here questions of interest apply: you group plants together in one way if you wish to identify species poisonous to humans, and in a different way if you wish to identify plants resistant to given sorts of pest.

The DNA test, however, seems to provide an objective yardstick which cuts across such classifications. Indeed it seems to give us the essence of botanical kinds, unmasking their inner nature. By the "essence" of something is meant that property which makes it what it is: if it lacked that property it would be either something else or nothing. Essential properties contrast with "accidental" properties, those that a thing might have or lack while still being itself. For example: it is an accidental property of a given potato that it is covered in mud or weighs a pound; were it cleaner or lighter it would still be a potato. But if it lacked whatever makes a potato a potato, it would obviously not be one. And the idea is: the essence of a botanical kind (more generally, any natural kind) is its DNA.

But DNA is very complex; it has a structure; it is made up of strands of proteins. Why choose the complicated DNA molecule as the essence, and not some more basic layer of structure down among the proteins? And if one did, might one not find a quite different set of comparisons to use as our classificatory criterion? Is it not possible that we might then find that at that level of structure, the rose is related to the lotus not the thistle? Either way, it is still we and not nature who are choosing what counts as the essence and therefore the criterion.

A C Grayling is Lecturer in Philosophy at Birkbeck College, University of London, and a supernumerary Fellow of St Anne's College, Oxford

Please send any comments on the Thought Police to Nick Fearn, email: n.fearn@independent.co.uk