This is an often entertaining book, but not as well written as it thinks it is, Paul Stammers writes.

Gould, the American palaeontologist who died a year ago, attempts in this modest-sized volume to reconcile what he sees as a clash between science and the humanities. His argument is that in science, as with much else, history is written by the winners.

He believes that a false impression has been given by scientists over the past 150 years that they have won a battle for objective truth against ill-informed people who prefer the arts and humanities as guiding lights, when, as Charles Darwin admitted, all experiments are carried out with the aim of proving or refuting an idea already taking shape in the researcher's mind.

Gould -- who was a well-matched opponent of Oxford University's Richard Dawkins -- believes some scientists prefer to lampoon those who disagree with them, rather than engage in debate and become adept at studying the social context of science. He also thinks they spend too much time attacking the "anti-science" fringe movements, who should not be taken seriously.

Gould has written previously on how he sees no conflict between science and religion as long as their practioners do not stray into each other's territory, and continues this theme here, although his accounts of how Renaissance and Enlightenment figures pushed the boundaries of learning reveal how many struggled to reconcile faith with observation of the material world.

Gould can be a pithy, witty and illuminating writer. But there are two main flaws with this book.

One is that it is too broad a subject to cover in 265 pages, so the reader has to flit from the likes of Erasmus to Newton in a few pages. The other difficulty is that Gould frequently lapses into a convoluted style that becomes tiresome. In particular, a better editor would have insisted on fewer clauses per paragraph.

However, the book makes a useful contribution at a time when concerns about scientists' 'agendas' has threatened to erode public trust in their findings.