Out of 350 undergraduates asked to take part in the test, usually given to children to assess their understanding and knowledge of Earth, more than 35 ended up drawing the Earth flat or as a hollow.

For the past 20 years educationists and psychologists have thought children do not have a good understanding of the Earth until they reach 10 to 12 years old.

Then psychologists Gavin Nobes, Georgia Panagiotaki and Katharina Dworzynski, at the University of East London, decided to use the same test on undergraduates.

Questions asked included:

* “Draw a picture of the earth and then show where the sky is and show where people live.” The correct answer would show a circle with a blue ring round it and people all over the earth. But 25 per cent of people drew a flat earth, or a circle with only sky and people at the top.

* “If you walk for many days in a straight line where would you end up?”

requires the answer back where you started. But some said at the end of the earth, in another country or by the sea.

Mr Nobes said: “This doesn’t show that undergraduates are stupid, but that the questions we have been asking children are ambiguous.”

Dr Nobes said: “We found an amazing proportion of well-educated adults totally misinterpreted the questions.

“Seeing how adults got some questions wrong makes us believe that the reason children have weird and wonderful ideas of the Earth is because they are trying to answer strange questions.

“The questions are too confusing. We think children know more than they have been given credit for.”

The next stage of the research will now be to improve the questions.

Researchers hope the results will help the science education of children.

The results of their research were presented at the British Psychological Society’s annual conference last week.

Other research presented included work by Kate Nation and Lucy Cragg at Oxford university on children who are able to read accurately but have poor understanding of what they have read.

These children, known as poor comprehenders, and a control group of youngster with good understanding, were asked to write a story based on pictures.

Dr Nation said: “The two groups produced stories of equivalent length and grammatical complexity. However, the poor comprehenders’ stories achieved lower semantic content scores indicating that their narratives included fewer main ideas from the story.”

Professor Dorothy Bishop, from Oxford university, studied identical and non-identical six-year-old twins to see if reading and writing disorders are genetic.

She found learning difficulties in children were only genetic in some of the most severe cases. Mild reading difficulties could be attributed to environmental influences such as children not being encouraged to read at home.