The Nobel Prize in Physics: The papers

Each September, a few weeks before the announcement of the Nobel Prizes, the science analytics company Web of Science Group releases a list of Citation Laureates —researchers who are considered strong candidates for Nobel recognition because of “exceptionally high citation records.” The idea behind the annual exercise originated with science-citation analysts Eugene Garfield and Alfred Welljams-Dorof, who in 1992 identified researchers whose work was cited so frequently that they could be distinguished as “of Nobel class .” Since then numerous scholars have pored over citation data to investigate the research output of Nobel laureates.

Much less bibliometric analysis, however, has been done on the specific research that has earned laureates Nobel recognition. We set out to identify and examine the key papers that resulted in the Nobel Prize in Physics. Fortunately, Jichao Li, Yian Yin, Santo Fortunato, and Dashun Wang did a lot of the groundwork for us. Earlier this year, in the journal Scientific Data, they published a freely available database of Nobel Prize–winning papers. For some prizes they listed one pioneering paper; for some they listed multiple works, including five authored by 1965 laureate Richard Feynman.

We collected all the winning papers listed for the physics prize, added entries for the two most recent awards (Li and colleagues’ compilation went through 2016), and made several changes to the choices of paper. Then, in light of the rapid rise in scientific publishing that occurred following World War II, we decided to analyze only the papers of laureates who won the prize from 1946 onward (though note that numerous postwar prize recipients published their pivotal work well before the war). Finally, using the abstracting and indexing database Dimensions , we compiled year-by-year citation data for the remaining papers that have digital object identifiers.

As one might expect for a small sample of papers—we looked at 176—that includes both theoretical and experimental breakthroughs in diverse fields like astrophysics, particle physics, and condensed matter, there’s no single prototype of a Nobel-winning physics paper. Some papers racked up hundreds of citations within a few years. Others took longer to gain attention, and some continue to languish in relative citation obscurity. Still, Nobel-winning works generally attain far longer shelf lives than even the most-cited physics literature as a whole.

The changing citation landscape

Citations are a useful, if imperfect, way of tracking the impact of scientific papers. As Garfield and Welljams-Dorof demonstrated, counting citations can distinguish top researchers and their most influential work. So there’s certainly a place for using citations to help identify “the most important discovery or invention within the field of physics,” as instructed by the will of Alfred Nobel, and to recognize research that has had positive impacts on science and society.

But citations can also mislead, particularly when one considers only the raw numbers. Literature related to fields with an outsize number of researchers can easily rack up far more citations than arguably more impactful papers in smaller fields of study. Papers introducing a method or technique, such as density functional theory, can outpace those describing the most famous discoveries (see the article by Sid Redner, Physics Today, June 2005, page 49 ). And most important for this analysis, citation numbers are skewed by the past century’s rapidly evolving publishing landscape. To illustrate the explosion in science literature in the postwar period, consider the 1986 physics prize, which was awarded to Gerd Binnig, Heinrich Rohrer, and Ernst Ruska for advances in microscopy made a half century apart. The 1982 paper by Binnig and Rohrer, the inventors of the scanning tunneling microscope, has racked up 3300 citations; Ruska’s 1932 paper on the electron microscope has 11.

Publishing and citing practices have also changed more recently, particularly with the rise of the internet and publish-or-perish policies. “There’s been an increase not only in the number of papers but also in the size of reference lists,” says Fortunato, an Indiana University network scientist who has published multiple bibliometric analyses of Nobel laureates. “The value of citations is going down.” As shown in figure 1, Nobel-winning papers published in the 2000s have, on average, garnered more citations within the first five years after publication than those published in the 1990s, which have more than those in previous decades.

The big picture

Despite the multiple caveats, Nobel-winning papers do stand out from even well-cited physics literature as a whole. The inset graph in figure 2, taken from the 2015 study “Attention decay in science” by Fortunato and colleagues, illustrates a representative trajectory of a popular physics paper. Examining literature in the Web of Science database, the researchers normalized each paper’s citation life cycle by dividing the paper’s total number of citations by its peak annual number of citations. The trajectories shown depict the average life cycles of physics papers that rank in the all-time top 10% in terms of total citations; the colors represent different years of publication. The researchers found that the average paper attracts its peak number of citations a few years after publication, but then the citation rate rapidly falls. Over time the rate of post-peak decline has slightly increased, a change Fortunato attributes to literature getting lost in the shuffle due to publication proliferation.

Papers that result in a Nobel Prize in Physics often follow a different citation trajectory, as the main graph in figure 2 shows. Like other popular physics papers, they tend to reach a peak within a few years and then begin to steadily decline. But after bottoming out 20 years or so after publication, they experience a resurgence. Some of that bounce may be due to the spotlight the papers receive following the awarding of the Nobel Prize. Though there is some evidence of such a bounce—the 1985 paper on chirped pulse amplification by 2018 laureates Donna Strickland and Gérard Mourou, for example, is on pace this year to double the number of citations it received last year—quantifying the effect is difficult with such a small data set. Perhaps the simplest conclusion is that papers that result in a Nobel gain prestige over time because of the outsize influence they have on future research, with some of those impacts not obvious at the time the paper is published.

The comparison of Nobel-garnering papers’ year of publication with the year in which they attained their maximum number of citations, shown in figure 3, further illustrates the dual peaks in popularity. The dots situated just above the shaded region are the many papers that follow the trend described by Fortunato’s group: They peaked in citations within a few years after publication. But many others, including some a century old, reached their maximum only within the past decade. The cluster atop the graph may be the result of contemporary physicists’ increasing propensity to cite past pioneering research. With today’s flood of new research drowning out most papers within several years, the data suggest that researchers and referees may be placing increased emphasis on acknowledging classic papers that paved the way. A dozen or so outlier papers peaked decades after publication but prior to 2010.

Citation standouts

Examine the papers individually and their differences really come into view. Most notably, there’s a huge range of citation counts among the papers. Konstantin Novoselov and Andre Geim’s 2004 Science paper describing the discovery of graphene has more than 32 000 citations, with no signs of slowing down—it has garnered over 3000 citations annually since 2012. On the other hand, the two 1953 Physical Review papers on nuclear structure by 1975 laureates Aage Bohr and Ben Mottelson have just several dozen citations each. The median number of citations among the Nobel-winning papers is 664.

The list of most-cited Nobel papers is dominated by relatively recent experimental discoveries that took the physics community by storm. Georg Bednorz and Alex Müller’s discovery of cuprate superconductors in 1986 generated a surge of interest, as evidenced by the marathon Woodstock of Physics session at the American Physical Society’s 1987 March meeting in New York City. The discovery of dark energy by two teams that included Adam Riess, Brian Schmidt, and Saul Perlmutter in 1998 and of giant magnetoresistance by Albert Fert and Peter Grünberg in 1988 also produced flurries of excitement and thus citations. The lone theory paper in either top-five list is John Bardeen, Leon Cooper, and Robert Schrieffer’s now-eponymous BCS theory of superconductivity.

The discoveries of graphene and the cuprates are important developments but probably not the most important since the first Nobel Prize in 1901. So those who study networks and bibliometrics have proposed alternative ways to measure the impact of individual papers and researchers. One recently proposed metric is citation wake. In a 2014 PLoS ONE paper , David Klosik and Stefan Bornholdt count up not only direct citations but also the citations of second-generation papers, third-generation papers, and so on. The goal is to quantify the spread of ideas that results from a given study.

In their paper, Klosik and Bornholdt apply their metric to all the papers in the Physical Review family of journals through 2009. The list of top-10 papers by citation wake includes four Nobel-winning works. Atop the list is the BCS paper. In third place is Eugene Wigner’s 1934 “On the interaction of electrons in metals ,” which has received a not-quite-standout 2150 citations yet is foundational for modern condensed-matter physics (see the article by Erich Vogt, Physics Today, December 1995, page 40 ). The other two Nobel-winning papers are authored by Feynman. Although no single Feynman paper attained an extraordinary number of citations, the famous physicist certainly spurred new ideas and areas of research.

Paper profiles

After compiling the citation data, one of the first things we did was plot each paper’s citations over time on a single graph. The resulting cacophony of scribbles made it clear that Nobel-winning papers don’t follow a representative citation trajectory. Even so, some of the curves stuck out. Here we explore three papers with very different post-publication histories.

J/ψ: Samuel Ting et al.'s 1974 Physical Review Letters paper describing the detection of the J/ψ particle may have sparked the November Revolution in high-energy physics, but from a bibliometric perspective its impact was short-lived. Of the paper’s nearly 1100 citations, more than 600 came before 1980; it hasn’t racked up more than 31 in any year since. Fellow 1976 laureate Burton Richter’s J/ψ discovery papers have a similar citation trajectory. The pattern may be due to what sociologist Robert K. Merton termed obliteration by incorporation: A discovery becomes common wisdom so quickly that researchers don’t feel the need to cite it.

NMR: Edward Purcell et al.'s 1946 Physical Review paper on nuclear magnetic resonance has been cited more than 3600 times, yet it languished in relative obscurity in the first few decades after publication. The paper’s citation rate began picking up steam in the 1980s, soon after Peter Mansfield and Paul Lauterbur’s pivotal work on magnetic resonance imaging. The Purcell paper’s citation peak came in 2016, seven decades after it was published.

BCS: Bardeen, Cooper, and Schrieffer’s theory of superconductivity captivated the physics community when it appeared in Physical Review in 1957. Then, after two decades of sagging citations, the paper got a boost in 1986 when Bednorz and Müller discovered the cuprates, which aren’t sufficiently described by the theory. Continued interest in both conventional and unconventional superconductors has vaulted the BCS paper to about 8000 citations.

We encourage others to slice and dice the data and come to their own conclusions. Our data set, which has citation numbers that are current as of August 2019, is freely available on request.

Editor’s note, 14 October: The article and the Nobel paper data set have been corrected based on reader feedback. The two tables were amended, with the addition of a paper by Saul Perlmutter and colleagues and the removal of a paper by T. W. Hänsch that was not related to his Nobel Prize.