These are findings of a report by Marie Duru-Bellat and Alain Mingat of the Dijon education research institute, who studied how 212 state coll ges (lower secondaries) organised a total of 20,000 pupils into classes.
They found that practice varied widely between schools. A fifth graded "very rigorously" by ability, while another fifth divided pupils randomly. In other schools there was at least one single-ability class.
The researchers compared streamed and unstreamed systems and found that random allocation "allowed the weakest pupils to progress while affecting the brightest pupils only very marginally". They calculated that low-ability pupils benefited twice as much as the best were penalised.
In an interview with a teachers' union, Marie Duru-Bellat of the Federation de l'education nationale, which commissioned the report, gave two reasons why unstreamed pupils performed better overall.
She said teachers tended to adapt to the assumed level of a class and organise their time and lessons differently for "slow" and "bright" pupils. Also teachers' expectations and demands varied depending on the ability of the group, and pupils responded accordingly.
Unstreamed classes, she said "constitute the most efficient solution to make all pupils progress. However, it is the most difficult to manage."
Many schools deny that they operate selection, but disguise the practice by the choice of options they offer or by labelling classes of slow learners as remedial. Sharp-elbowed, middle-class parents know how to decipher certain codes; they commonly pick German as their child's first language on entry to coll ge, as this was traditionally reserved for the brightest pupils and some schools tacitly still do so. Similarly, schools might group those taking Latin into the same class.
The Paris chief education officer recently reminded one coll ge that excessive selection is forbidden, and he ordered the dismantling of intensive language options for bright pupils. In an exam taken at the school last spring by the six forms of the fourth and final year, results showed all the pupils in the top stream passed, but all those in the lowest failed. In the intermediate classes the pass rates followed the order into which pupils had been streamed.
The college unique - unstreamed comprehensive secondary education - was introduced in the 1975 education Act, known as the Loi Haby after the education minister of the day. Under the legislation all pupils followed the same basic curriculum, though slow learners could take an extra year - three instead of two - to complete the first stage of secondary school.
Further remedial measures have been introduced recently to help coll ge pupils in difficulty. These are defined as those who have repeated one or two years at primary school, whose test results reveal problems in French, maths or both, or who are below a certain standard after the first two years of college. Latest figures show 8 per cent of pupils entering secondary school have severe difficulties.
Jean-Paul Roux, general secretary of the federation, said the report demonstrated the anti-democratic character of selection, and called for the restoration of truly comprehensive education as part of the fight against social exclusion.