The boards are currently holding meetings to decide the final grade boundaries for this year’s exams, the process which was at the centre of last year’s regrading controversy.

Ron McLone, chief executive of the Oxford, Cambridge and RSA board, last year ordered changes to grade boundaries on certain papers, provoking an inquiry which eventually led to the results of 1,945 students being upgraded.

Roger Porkess, who wrote the first modular A-level syllabus in the Eighties and has been a critic of the present grading system, warned last weekend of fresh “downgrading” problems this year.

A new code of practice, brought in to prevent a repeat of last year’s controversy, allows boards to consider students’ results across a whole subject before setting grade boundaries for particular modules.

But, writing in a Sunday newspaper, Mr Porkess said he believed that the code leaves open the possibility that individuals could be penalised if the boards believe that too many students are getting high grades in a subject.

Under this scenario, the boards could simply make it harder for students to do well in particular modules, to keep down the pass rates and the numbers getting higher grades overall.

Mr Porkess said the code of practice should rule this out. Boards should only consider how students have done within a particular module, before setting the grade boundaries for that module.

However, Paul Sokoloff, director of policy and qualifications at the Edexcel board, said that boards needed to consider whether some papers were proving more difficult than others.

But this, Mr Sokoloff added, was only one piece of information available to the boards, including crucially, the reports of senior examiners on the overall quality of students’ work.

This year, the code of practice has been tightened to give less weight to purely statistical information, and more to these qualitative judgments of examiners.