The Government is running the risk of provoking fresh boycotts of national tests by teachers unhappy at the way the first primary league tables are likely to be compiled.
Ministers are believed to have been advised that they will not be able to introduce what many teachers regard as fairer league tables until at least 1999.
Gillian Shephard, the Education and Employment Secretary, announced this week that primary league tables based on this summer's tests of 11-year-olds will be published in January. It seems certain that for the next three years they will be confined to raw test results.
English teachers who led the successful boycott of national curriculum tests two years ago have already questioned the validity of the 1995 tests and are deeply unhappy at the prospect of the results being used for tables.
When Mrs Shephard took over as Education Secretary she promised to delay publication of results - a major factor in ending the widespread boycott of tests.
The London Association for the Teaching of English whose action sparked off the boycott, insisted that crude results should not be published without being put in context. "Value-added" tables, which try to measure the difference schools make to their pupils, are considered to be fairer.
Specialists say that the results of individual pupils, and not just schools, need to be tracked in order for such measures to be meaningful. However few local authorities have saved this information from Year 2 tests in 1992, which would be needed to show in 1996 the progress during the junior years .
It is understood that the School Curriculum and Assessment Authority only located one such local authority - Avon.
Attempts to collect the data from previous years from all 20,000 primary schools would be be virtually impossible. The first possible year for a "baseline" of seven-year-olds' results is likely to be 1995, especially since tests were boycotted in many schools in 1993 and 1994.
Michael Barber, dean of new initiatives at the University of London Institute of Education, said: "Value added in primary schools without individual pupil level data would be totally unreliable."
Individual pupil data is needed because primary schools are small institutions. Changes to each year group caused by families moving could make big differences to results. In some inner-city areas, up to a quarter of pupils change between Years 2 and 6.
It also emerged this week that thousands of lower-ability pupils and 11-year-olds with special needs were double counted when the test results published by the Government last month were calculated. This has raised doubts about the results. The error appears to mean top juniors did slightly better in maths and science than official figures show.
But the SCAA insists the mistake will not be repeated next year because the system of collecting results has been changed.
Peter Coles, chief education officer in Hampshire, which discovered the problem, believes "tens of thousands" of pupils with low scores were counted twice. No one as yet knows the national figure, but Hampshire statisticians found 1,000 cases.
Mr Coles said: "As a CEO I am very concerned that there was not proper validation of the KS2 results before the information was reported to Parliament."
At the same time, an independent evaluation of this year's tests from Bath University and the University of Wales highlights inconsistencies in the way teachers carried out the tests. Although there was no evidence of cheating, some teachers helped with spellings and allowed pupils more time on the maths test.
Testing conditions also varied, with some pupils sitting right next to classmates doing the same paper, and others more spaced out.
David Hawker, a spokesman for the SCAA, said these problems had been ironed out for the 1996 tests.
SCAA was confident that the tests themselves "have reached a stage in 1996 where they are of high quality", said Mr Hawker.