Children of different ages took the pilot tests

It said there were aspects of the results that needed to be better understood before they were released.

Children are entered for these "single level" tests when their teachers think they have reached a certain standard.

The tests are to confirm this judgement - so should result in a high pass rate. They were piloted in about 400 schools.

About 40,000 children were tested in December, spread across Years 3 to 9.

The results should have been sent to the schools last Friday.

Instead head teachers received a letter from Sue Hackman, chief adviser on school standards at the Department for Children, Schools and Families (DCSF).

Apology

It said there had been some very positive feedback about the tests and the response of pupils and teachers to them.

"However, the marking and level-setting process has revealed some unexpected patterns in the results which we need to understand further.

"This is not unusual in piloting new tests, but it does mean a delay whilst we check the results.

"We have asked the National Assessment Agency (NAA), who develop and deliver the tests for us, to look into this for us. I apologise for this delay."

Ms Hackman added: "We aim to get results to you as and when we can - this may mean that you will not receive results for all levels in all subjects at the same time."

A spokeswoman at the DCSF said the delay was expected to be several weeks.

"All I can say is that we're still analysing the results, so don't want to comment in detail at this stage, but there are some differences between subjects, levels and performance in different key stages that we need to understand better before we are confident about releasing results," she said.

Asked whether the pass rates had been lower than expected, she said: "No - just that there are some unexpected patterns in the results which we need to understand further."

'Professional discussion'

Officially there was no expected pass rate.

In an interview with the BBC News website when the tests began the head of the National Assessment Agency (NAA), David Gee, would not be drawn on what it might be but agreed it might be expected to be high, if schools were correctly selecting children who were ready to take the tests.

The general secretary of the National Association of Head Teachers, Mick Brookes, said his experience was that if a teacher assessed 10 pupils as being at a certain level it was likely that eight of them would attain that level in tests.

This would equate to a pass rate of about 80%.

"Where the test and the teacher assessment is the same you can pretty well tick the box but where it's different and the results are higher or lower then you need to have a professional discussion about what's happened," he said.

Professor Alan Smithers of the University of Buckingham said: "The Department for Children, Schools and Families would not have pulled the results unless something had gone quite badly awry.

"An educated guess is that they are disappointed with the results. This could be because the tests are not yet pitched at the right level."

One of the reasons children might have found these tests difficult, if that is what has happened, is that they are unlike others they would be used to, and unlike the usual "Sats" they will be taking in May.

Usually tests have a range of possible outcomes, and have what experts call "compensatory" questions at the start - relatively easy ones that settle candidates into the test, progressing until they cannot answer any further.

But in single level tests, by definition, the questions are all at the same standard and children can either manage them or not, which could leave them struggling.

As an incentive to take part in the pilot scheme, schools were told they could keep as their official score the higher results from either the normal Sats or these single level tests.