ÖйúAƬ

You characterise Thomson Reuters' success rate for predicting Nobel prizewinners "at about 15 per cent, with seven of its 45 predictions made between 2002 and 2008 proving to be accurate" ("Brits tipped for Nobel glory", 24 September). You calculate this rate in terms of correctly identified Nobel prizes, not in terms of the number of scientists who were named and later won (11 over this period).

Of course, because a scientist has not yet won the Nobel prize does not mean they will not do so in the future. It is more meaningful, in our view, to consider the improbability of predicting a Nobel prizewinner, whatever method is used (we analyse citations).

Take chemistry, for example. Our data indicate that there are about 700,000 publishing chemists today. Even if one were to limit the pool of potential Nobel laureates to the elite, say the top one tenth of 1 per cent, that leaves 700 plausible contenders.

In 2008, we said that Roger Y. Tsien might win the Nobel Prize for Chemistry for his research on green fluorescent protein - and he won that year. We would characterise the odds of a successful prediction in this case as 1 in 700. In other words, the odds are long but Thomson Reuters has had significant success owing to the power of citations to reveal judgments of peer esteem.

David A. Pendlebury, Thomson Reuters, Philadelphia, PA.