Using the same tests over two decades allows a comparison of standards in literacy and numeracy. There are many problems in making such comparisons over time, chiefly because some items no longer carry the same meaning. However, many of the questions - which were primarily multiple choice - covered national curriculum attainment targets.

We chose in the end to compare children in Years 4, 5 and 6 at the end of the school years 1976-77 and 1996-97.

As figure 2 demonstrates, the scores on maths, language and reading have declined significantly. Overall, as in 1976, performance was lowest in language skills, although generally punctuation improved. The decline in performance was greatest in reading.

In 1976 the Richmond tests were given in September, whereas in 1996 they were administered in June. In comparing sub-test scores over the two decades, like was not compared with like since pupils tend to do less well on tests after the summer holidays. We therefore adjusted the 1976 September scores as if the test had been taken in June.

Figure 3 shows the percentage change in scores after this adjustment. It can be seen that only in Year 6 are differences generally positive for mathematical concepts. Mathematical problem-solving shows a decline, particularly for Years 4 and 5.

Improvement in reading was greatest in Year 5. In language, spelling has worsened except in Year 5, but both punctuation and use of capital letters appears to have improved in all three-year groups. Only in Year 6, possibly because of national curriculum tests, does language usage show overall improvement.

In our judgment, informal teaching approaches cannot be blamed for the fall in standards, if only because our results - and those of other researchers - identify a continued shift from individual teaching towards whole-class activity since the mid-1980s.

Indeed, when we used the same tests, admittedly in small rural schools in the mid-1980s, we found that scores increased. The fall appears to have occurred in the late 1980s and throughout the 1990s. The one factor which stands out in this period of rapid change is the national curriculum.It has reduced the time given to hearing children read and the amount of immediate feedback pupils receive while writing.

Teachers said they were under pressure to get through the curriculum, emphasising instruction and content, rather than teaching for understanding. This is reflected in our observation data.

Furthermore, there has been continued emphasis on subject specialism and a discouragement of topic work involving more than one subject area. Teachers told us, despite denials from the Office for Standards in Education, that it is easier to pass inspections if you have a secondary-style timetable to demonstrate that the requisite number of hours are given to the core subjects.

Twenty years ago children were often required to do research for a project, write out their findings and present them to the teacher who would listen and correct them. The decline in such activities was very marked in our study.

Hearing children read now rarely takes place outside a formal English lesson. It is perhaps ironic that those who have criticised primary teaching most vehemently, such as the Chief Inspector, helped to encourage this form of the national curriculum.

Serious thought clearly needs to be given to the revision of the primary curriculum. We should start by looking at the evidence of what has happened to primary teaching since the 1970s.

Maurice Galton is professor of education at the University of Leicester