This first experiment has some obvious limitations that Schmeck and her team sought to address in a follow-up involving 168 more pupils (average age 14). Most importantly, the researchers showed that drawing each paragraph of the to-be-learned text led to superior test performance (63 per cent correct on multiple choice, on average), even when compared to a condition in which pupils were instead provided with drawings produced by the author of the scientific text (53 per cent correct). The better the pupils’ drawings, the more successful they were in the tests afterwards.