Those palaeontologists who name the most new dinosaur species are the least likely to get it right, a survey of nearly two centuries of research has found. The trend is as true for modern researchers as it was for their nineteenth-century forebears.

Michael Benton, a palaeontologist at the University of Bristol, UK, analysed the work of 321 authors active between 1824 and 2004. Of these, the 23 most prolific, who each named more than 10 species, were responsible for 665 of the 1,400 dinosaur and prehistoric bird names designated in the period.

But sometimes these names turned out to be attached to already known species, as in the case of Brontosaurus, named in 1879 but later discovered to be the same as Apatosaurus, named in 1877.

More common still is for subsequent researchers to decide that the original material is inadequate to determine what the species might be. And some species turn out not even to be dinosaurs.

Names given by the 23 most prolific palaeontologists have proved especially vulnerable to such pitfalls. Only 274 out of 665 names, or 41%, are still in use.

The remaining 735 dinosaur names were the work of almost 300 authors. These have been more robust: 444, or 60%, are still in use.

"I would have expected that more prolific, experienced authors might be better at recognising genuinely new species, yet they were less successful than authors who name only a few dinosaurs," says Benton. "It is hard, and maybe impossible, to construct a case that experience in naming dinosaurs makes one better at the job." His study is published in the Journal of Vertebrate Palaeontology1.

The author with the worst record is Edward Cope. In the late nineteenth century, Cope and fellow palaeontologist Othniel Marsh vied to be dinosaur king of the United States. Cope named 64 dinosaur species, only 9 of which are still considered valid, giving a success rate of 14%. Marsh has fared better — 35 of his 98 names are still in use, a 36% hit rate.

Among authors still active in 2004, Dong Zhiming, now retired from the Institute of Vertebrate Paleontology and Paleoanthropology (IVPP) in Beijing, has the lowest success rate. Dong, one of China's most famous palaeontologists, named 37 dinosaurs, 22 of which are still valid (59%). He could not be reached for comment.

Benton suggests two reasons for the trend. First, prolific namers may have been, to use the taxonomic slang, 'splitters' rather than 'lumpers', meaning that they tended to subdivide species rather than ascribing small differences between similar fossils to, say, sex differences or individual variation.

Second, dedicated dinosaur hunters may have been especially driven by the kudos or extra funding that can come with naming new species and so may have jumped at each chance. Authors with wider interests, who named fewer species, may have felt less pressure and so might have taken more care, suggests Benton.

There are "unique difficulties" in naming dinosaurs, says Matthew Carrano, curator of dinosaurs at the Smithsonian Institution in Washington DC. There are no genomes to sequence, and often only a few bones are known, so it can be difficult to distinguish between species.

Some mistakes, such as those resulting from changing definitions of what counts as a dinosaur, could not have been anticipated, says Carrano. "Some 'early dinosaurs' are no longer considered dinosaurs by definition. They are still valid, just not dinosaurs." But, he adds, dinosaur names are not particularly less reliable than those of any other group.

Zhonghe Zhou, director of the IVPP, said that the low success rates seen for some scientists did not diminish their greatness. "The greatness of a researcher is in the mind of his colleagues, not in the statistics," he said.