One reason is that taxonomists lack clearlyachievable goals that are both realistic and rel-evant. Of course it would be great to describeevery species of organism on Earth, but we arestill monumentally uncertain as to how manyspecies there are (probably somewherebetween 4 million and 10 million); this goal isjust not realistic at present. There are variousprojects aimed at listing, for example, all thevalid described species of animal in Europe,or butterflies on Earth (see Box 1, overleaf).These aims are eminently achievable and veryworthwhile, but the results are like raw, un-annotated DNA sequences: unexciting and ofrelatively little value in themselves to non-specialists. Taxonomists need to agree ondeliverable projects that will receive wide sup-port across the biological and environmentalsciences, and attract public interest. A second problem is part of the legacy ofmore than 200 years of systematics. Manytaxonomists spend most of their career try-ing to interpret the work of nineteenth-century systematicists: deconstructingtheir often inadequate published descrip-tions, or scouring the world’s museums fortype material that is often in very poor con-dition. A depressing fraction of publishedsystematic research concerns these issues.In some taxonomic groups the past acts as adead weight on the subject, the complexsynonymy and scattered type materialdeterring anyone from attempting a modern revision. As Frank-Thorsten Krellpointed out in Correspondence (Nature415,957; 2002), “original descriptions haveto be referred to for ever, independent of thepaper’s quality”. The problems do not always lie in the past.Even today, many species are being describedpoorly in isolated publications, with noattempt to relate a new taxon to existingspecies and classifications. Many of these‘new’ species will have been described before,so sorting out the mess will be the headache ofthe next generation of taxonomists. It is notsurprising if funding bodies view much ofwhat taxonomists do as poor value for money.One of the astonishing things aboutbeing a scientist at this particular time inhistory is the vast amount of informationthat is available, essentially free, via one’sdesktop computer. I can download thesequences of millions of genes, the positionsof countless stars. Yet, with a few wonderfulexceptions, the quantity of taxonomic infor-mation available on the web is pitiful, andwhat is present (typically simple lists) is oflittle use to non-taxonomists. But surely tax-onomy is made for the web: it is an informa-tion-rich subject, often requiring copiousillustrations. At present, the output of muchtaxonomy is expensive printed mono-graphs, or papers in low-circulation jour-nals available only in specialized libraries.These are not attractive ‘deliverables’ formajor research funders.Two models of taxonomyThe taxonomy of a group of organisms doesnot reside in a single publication or a singleinstitution, but instead is an ill-defined integral of the accumulated literature on thatgroup. The literature is bound together andcross-references itself using the venerablerules of taxonomy encapsulated in the codes.But this is not the only way to organize a tax-onomy. The taxonomy of a particular groupcould reside in one place and be adminis-tered by a single organization. It could beself-contained and require reference to noother sources.commentaryNATURE|VOL 417|2 MAY 2002|www.nature.com 17H. Charles J. GodfrayTaxonomy, the classification of livingthings, has its origins in ancient Greece andin its modern form dates back nearly250 years, to when Linnaeus introduced thebinomial classification still used today. Lin-naeus, of course, hugely underestimated thenumber of plants and animals on Earth. Assubsequent workers began to describe moreand more species, often in ignorance of eachothers’ work, the resulting confusion andchaos threatened to destroy the whole enter-prise while still in its infancy. In today’s jargon, we might call this the first bioinfor-matics crisis. Using the tools then available,nineteenth-century taxonomists solved thiscrisis in a brilliant way that has served thesubject well since then. They invented acomplex set of rules that determine how aspecies should be named and associatedwith a type specimen; how generic and high-er taxonomic categories should be handled;and how conflicts over the application ofnames should be resolved. All these rulesrevolved around publications in books andscientific journals, and their descendantsform the current codes of zoological andbiological nomenclature.But today much of taxonomy is perceivedto be facing a new crisis — a lack of prestigeand resources that is crippling the continu-ing cataloguing of biodiversity. In the UnitedKingdom, a Parliamentary Select Commit-tee is currently conducting an enquiry intothe health of the subject for the second timein 10 years, and similar concerns are beingexpressed around the world. In this article I shall first explore why descriptive taxono-my is in such straits (in contrast, its sistersubject, phylogenetic taxonomy, is flourish-ing). Then, after this essentially negativeexercise, I will argue that taxonomy can prosper again, but only if it reinvents itself asa twenty-first-century information science.It needs to adopt some of the solutions thatmolecular biologists have developed to copewith the second bioinformatics crisis: thehuge explosion of sequence, genomic, pro-teomic and other molecular data. The problemWhy can’t descriptive taxonomy attractlarge-scale funds in the same way as other bigprogrammes like the Human Genome Pro-ject or the Sloan Digital Sky Survey? All threeprojects are enabling science: not in them-selves generating new ideas or testinghypotheses, but allowing many new areas ofresearch to be opened up. Challenges for taxonomyThe discipline will have to reinvent itself if it is to survive and flourish.This discipline ismade for the web:it is information-rich and often requirescopious illustrations.NATURAL HISTORY MUSEUM, LONDONFrom paper to screen: is it time for taxonomy tobreak with tradition and unify on the Internet?© 2002 Macmillan Magazines LtdMy main argument is that to address theproblems outlined above, and for taxonomyto flourish now and in the future, it has tomove from the first to the second model:from having a distributed to a unitary orga-nization. Such a massive task could only beaccomplished group by group, as resourcesbecame available. I believe a