This week, mathematicians at a conference in Hong Kong will discuss a radical idea: that maths should accept theorems that are very, very, very probably right - but not proved definitively.
Since maths has always prided itself on absolute proof, this is a startling theory. It is the more so because excitement and controversy are not generally associated with the traditionally dry and dusty subject of numbers and algebra.
Yet it may be just the sort of intriguing work needed to beguile students into returning to the subject.
Mathematics is seen as the foundation of the sciences, and indeed of the world. Mathematical law describes how the Earth turns and how cells divide. But just as buildings win accolades while the foundations they rely on are unseen so, generally, sciences win the prizes built on mathematical bases: biology is credited with findings about cells and geology with knowledge about the Earth.
And while those sciences struggle to keep students more interested in taking up business and finance, maths is in danger of dying for want of new blood. Princeton, one of the world's most prestigious universities, this year has three undergraduates studying maths as their main subject - though staff also teach maths for engineers and scientists.
'Most universities have about two students per professor. They have 10 professors for every (full-time) student,' says Professor Manuel Blum, formerly of the University of California at Berkeley, now a visiting professor in computer science at City University, where the conference starts today.