É«ºÐÖ±²¥

The study of the ¡°half-life¡± of journal articles (¡°Tried and tested: for how long do research papers remain useful?¡± News, 30?January) should have included Doctor Who as a co-author, because it certainly involved time travel.

The reality is more mundane, but it should induce a little caution in any use made of these data. The study did not arrive at half-life figures. Instead, it posed and answered a?different question, finding a lower bound for the half-life by finding the median date for citations in an arbitrary window of time. The choice of time interval in some disciplines will determine the answer, so it is unclear how to interpret this.

A modest sample (of my most recent journal article) illustrates the extent of the problem in mathematics: the publication dates of the references cited in it range from 1915 to 2014, with the median around 1986. This is not atypical in maths, and makes one suspicious of the claimed median of four to five years in maths. Indeed, there is nothing in the study that precludes the possibility that the half-life in some disciplines is no more finite than is the universe itself.

Thomas Ward
Pro vice-chancellor (education)
Durham University