Sir

You report that incidents of cheating by students have increased significantly over three decades, amounting to more than two-thirds (73%) of all students polled in US universities without an ‘honour code’ and about a half (49%) in universities with an ‘honour code’ and punishments for its violation (Nature 390, 430; 1997).

At the beginning of this century, the record may have been even higher. Abram F. Ioffe (a solid-state physicist) reported in his book Meetings with Physicists (Nauka, 1983) that this problem was raised long ago by another prominent scientist, Gilbert N. Lewis (acids-bases theory), with whom he had worked at the University of California at Berkeley in the 1920s. At that time, students forced the university senate to allow them to take written examinations in the absence of instructors.

Lewis proved that 90% of the examination papers were written using unsanctioned sources. There was also an ‘honour’ angle. After Lewis's report, the senate ordered the students to add at the end of their work a declaration that they deserved to be trusted in examinations on their honour as American citizens. Lewis proved that this addition did not make a significant difference. (No punishment was incurred with this declaration.)

The senate was furious — with Lewis. It is tempting to suggest that the 90% rate of student cheating found by Lewis is a potential rate achieved under pure experimental conditions (in the absence of instructors and punishments), which can be reduced by instructor presence (to 73%) and further reduced by the application of punishment (to 49%).

Ioffe remarked that, on balance, when he examined his students orally after his course, he had only positive impressions. Thus the inclination to cheat did not seem to interfere significantly with the willingness and ability to learn.