Normally each marker had a different batch of
scripts, but for the experiment one script was taken at random from each batch
and replaced after three copies of it had been made. The three copies
were sent to the other three markers who assessed them, while the
original was replaced and assessed in the usual way. Each of the four
copies was therefore assessed by a single marker, but the three markers
who had a copy knew that the script was part of the experiment, while
the person marking the original did not know it to be part of the
experiment. The experiment was repeated at another examination, with
the same examiners, but different scripts.