The title of the paper may not have been particularly enticing, but its contribution was important: in “Negative results of an attempt to detect nuclear magnetic spins” (Physica 3, 995–998; 1936), Cornelis J. Gorter (pictured) reported calorimetric experiments aimed at detecting nuclear magnetic resonance in crystals of lithium fluoride and in alum. He had failed.

Credit: © KAMERLINGH ONNES LABORATORY

With his idea of irradiating a magnetic dipole transition between two Zeeman levels, Gorter was precisely on track to make the first observation of nuclear magnetic resonance (NMR), but not with his choice of samples. Stimulated by Gorter's work, two years later Isidor Rabi used the resonance technique to successfully measure nuclear magnetic moments in molecular-beam experiments. And in 1945 and 1946 — by which time Gorter had published further negative results (Physica 9, 591–596; 1942) — groups led by Edward Purcell and by Felix Bloch independently reported the detection of NMR signals from solid paraffin and from liquid water, respectively. Rabi, Purcell and Bloch all went on to Nobel honours, but Gorter was left to wonder, as he reflected later in his career on this and other instances, about his “bad luck in attempts to make scientific discoveries” (Phys. Today 20 (1), 76–81; 1967).

Gorter's reporting of negative results may seem laudable, but is something less likely to be found in the scientific literature of today, with competition for funding and recognition so fierce. In a systematic study of the disappearance of negative results from the literature, Daniele Fanelli has found that the phenomenon is indeed widespread and getting more pronounced (Scientometrics http://dx.doi.org/10.1007/s11192-011-0494-7; 2011).

Fanelli has perused 4,656 papers published between 1990 and 2007, sampled from over 10,800 journals from across the physical, biological and social sciences. All of the papers claimed to have tested a hypothesis, and Fanelli noted whether the reported findings supported the hypothesis, or not. Averaged over all of the papers studied, Fanelli found that the odds of reporting a positive result increased by around 6% every year over the period considered.

The trend was most pronounced in the social sciences, and more so in the biological sciences than the physical. But a tendency to report positive results was evident in most disciplines, physics included. And there were regional trends too: across Fanelli's sample, corresponding authors based in Asian countries (and in Japan particularly) tended to report more positive results than in the USA, who in turn reported more positive results than their colleagues in Europe (especially more than those in the UK).

Fanelli concludes that research “is becoming less pioneering and/or that the objectivity with which results are produced and published is decreasing.” This is not only a loss for science, but also invites the question of how, on a global scale, resources for research should be best utilized.