Sir

Your reports of the suicide of Jason Altom, a student at Harvard University, crystallized some thoughts about the frustrations of many graduate students ( Nature 395, 823; 1998 and & 397, 291; 1999).

Tragedies like this are known at almost all research universities. It is surprising how much control supervisors can exert over their students' careers, and bad relations with supervisors are nightmarish. There are many factors that individually may seem innocent, but which can have disastrous cumulative effects. One of these is the system of recommendations (references).

Throughout a postgraduate scientific career, recommendations are required, and that of one's PhD supervisor is almost indispensable. Applicants will obviously try to submit only positive recommendations, so this system is inherently biased and subjective. As long as one has a friendly and objective supervisor, all is well. But a student who has strained relations with their boss is at an unfair disadvantage.

I know of many students who cannot speak out against their mentors even after finishing their PhD, largely because they will need references in the future. This breakdown might be read as an inability to maintain professional relations, but such relations are a two-way street.

References should be less important, and hiring choices should be decided by more objective parameters such as test scores, quality of work and publications, and personal interviews. Sympathetic professors and graduate programmes should try to put this into practice as soon as possible, and encourage others to do so.