Uncovering how new tools trigger new breakthroughs leads us to explore the most puzzling and mysterious feature of discovery: serendipity. Many think breakthroughs are serendipitous—an unexpected and fortunate observation. From the discovery of x-rays to superconductivity. In fact, many believe discovery itself is unexplainable and unpredictable precisely because of serendipity. But what role do chance and serendipity actually play in discovery? Are they what we commonly think they are? And has discovery been more random than it needs to be, just because we have not been looking in the right place? By studying science's over 750 major discoveries, we uncover a surprising and powerful pattern that has been overlooked until now. Discoveries often labelled as serendipitous are actually sparked by using—for the first time—a powerful new tool that makes the unexpected observation possible that we are not even looking for: an improved microscope revealed cells, a discharge tube uncovered x-rays, gel diffusion revealed the Hepatitis B virus and a cutting-edge spectrograph detected the first planet outside our solar system. What seems like chance is often just using an entirely new lens—just as in each of these discoveries. In other words, breakthroughs that seem serendipitous follow shortly after we create the new tool—with other researchers able to arrive at the same findings applying these same discovery tools. While a serendipitous moment was reported in about one in six major discoveries, there is a pattern behind the surprise: using new tools to make the breakthrough finding commonly through exploratory research. Why is this important? It changes how we think about discovery: we can generally better understand individual serendipitous discoveries as tool-triggered discoveries that multiple researchers can make. And most crucially, developing powerful new tools helps reduce chance and predict new discoveries—whether we label them serendipitous or not. So here we reframe discovery: because new tools are what commonly unlock surprise, we can actually design and accelerate discovery—shifting it from a more passive outcome of serendipity.
Are many discoveries serendipitous and unpredictable—or can we actually learn to better anticipate them? Are they more accidental—or more designed and engineered? Many think discoveries emerge as a sudden serendipitous moment, or a flash of insight, that we cannot study and measure systematically. The term serendipity itself originally comes from a Persian fairy tale, The Three Princes of Serendip, who were 'always making discoveries, by accidents and sagacity, of things they were not in quest of ' . Seeing discoveries as arising serendipitously, defined as a moment when we make an observation or finding we are not searching for, is attractive. It is a concept that captures the imagination of both scientists and the general public. The narrative of serendipity is compelling and appears in science textbooks, in studies of scientific discoveries and in science news outlets. An editorial in Nature argues that 'data are not available to track it in any meaningful way … [so largely] academic research has focused on serendipity in science as a philosophical concept … [and so] giving curious minds free rein to explore nature may well be the best way to generate discoveries' . It seems natural to think we cannot study serendipity: after all, how could we analyse something that happens by chance? So is serendipity—by its very nature—unpredictable? Some scientists, psychologists and philosophers have tried to study the elusive and difficult-to-measure role of serendipity. Scientists examining the careers of scientists suggest that serendipity plays a key role in breakthrough moments. Studies of hundreds of thousands of scientific publications suggest that a scientist's most impactful research may be randomly distributed across their career—and appear at unexpected moments. Scientists stress 'the profound unpredictability that pervades many aspects of scientific discovery' . Psychologists explore how scientists reason, arguing that unexpected discoveries are surprisingly common in science—and can be linked to creativity. Even philosophers like Karl Popper argue that 'there is no such thing as a logical method of having new ideas … every discovery contains "an irrational element"' . The influential physicist Henri Poincaré also famously said that 'It is by logic that we prove, but by intuition that we discover' . And Nobel laureate Carlo Rubbia highlighted that 'Scientific discovery is an irrational act' . Other researchers also believe that thinking that 'the process of discovery has a distinct logic may have been vastly overstated' . Yet arguing that discovery is irrational or unpredictable does not actually explain it at all—and leaves no direct way to intentionally spark it. It is also misleading—and can waste researchers' time and resources. A keyword search of 'serendipity' up to 2025 in the journals Nature and Science reveals about 2000 hits. Yet despite the interest on the topic, no large-scale study of science's major discoveries yet exists on serendipity. Researchers exploring the topic commonly focus on the unexpected observations made by discoverers—not what sparks those surprising observations across discoveries. So have we been looking at the wrong side of the coin? Here we test that hypothesis: is there a logic to how discoveries emerge, whether they seem serendipitous or not? We will see that we can in fact establish a common cause that precedes new discoveries and makes them possible. But to do this, we need to systematically analyse science's major discoveries and serendipity at scale. Only then can we uncover a common hidden pattern that links seemingly random discoveries together. So which discoveries do scientists describe as classic examples of serendipity? Galileo famously discovered the moons of Jupiter serendipitously in 1610, just like Hooke did when uncovering cells in 1665 and Röntgen when revealing x-rays in 1895. But these scientists were only able to unlock these unexpected discoveries by deliberately using a powerful new tool: Galileo's recently invented telescope in 1608, Hooke's improved microscope built in 1662 and Röntgen's discharge tube created in 1875. These extraordinary tools enabled these three surprising observations they were not even looking for—rather than just a flash of insight. Take Arno Penzias and Robert Wilson who unexpectedly discovered cosmic background radiation in 1964—providing strong evidence for the Big Bang theory and transforming cosmology. But they actively applied a new highly sensitive radio receiver—a 20foot horn antenna built in 1961—that made the surprising finding possible in the first place. Even Robert Hooke himself pointed to serendipity when saying that: 'the greatest part of invention being but a luckey bitt of chance' . But he was using his new advanced microscope to unexpectedly discover cells—the basic building blocks of life. Many also think of Kamerlingh Onnes' groundbreaking discovery of superconductivity in 1911 as unintentional. But Onnes used a liquefier he created in 1906 and a sensitive thermometer he built in 1907 to run systematic experiments. These tools allowed him to measure temperatures as low as absolute zero and ultimately made discovering liquid helium and superconductivity possible. What do these classic serendipitous discoveries have in common? Could it be that they all happened when scientists used—for the first time—a new tool? They were not expecting what they saw, but could only make the surprising observations because they were using a tool that had never been used in that way before. So what if that is how serendipitous discovery generally works? Could the key to understanding serendipity be the tools we use to explore the unknown—rather than chance? What if the reason why discovery is commonly thought of as mysterious and unpredictable is because researchers have not yet studied it in a comprehensive way spanning across science's discoveries and fields to reveal what they share in common? Framing discoveries as serendipitous makes for exciting stories of individual discoverers. We are fascinated by discoverers, often seen as geniuses. It seems natural to focus on the personal side of science, where discovery can seem random. And we are more intrigued by surprise than design. From the perspective of an individual discoverer, a moment of surprise can seem very unsystematic. But what if framing discoveries as triggered by serendipitous moments makes them seem much more blind than they actually are? By studying serendipity among science's major discoveries, we uncover a key pattern here: breakthroughs that seem serendipitous are often made soon after we develop the new enabling tool. We will see how these tools make the surprising observations possible—and make them increasingly likely (before the discovery) and replicable by other researchers (after the discovery). The irony is that serendipity is seen as illustrating no structure in the discovery process. But by analysing these moments of serendipity across science and identifying the pattern driving them, we find that they actually reveal strong evidence of the mechanism behind discovery. The key to these unexpected moments is generally the tool—one that allows us to look at the world in a completely new way. Yet serendipity suggests a passive process—something that happens to us rather than something we can actively shape. But tool development highlights a systematic strategy that is far from passive: scientific progress depends on designing better tools that maximise the odds of uncovering new observations. It shifts the focus to creating the conditions where surprising moments become possible in the first place—and increasingly likely. Thinking of discoveries as serendipitous is not only unhelpful in creating new breakthroughs, it also offers little guidance for policy on how to best fund science and incentivise researchers to spur innovation (Box 2.1). So what if the unpredictability associated with discovery does not stem from researchers, scientific institutions or even timing? What if it comes from not yet realising that the very tools we use— and their blind spots—are largely shaping what we can see, and what we miss? What if the best and fastest strategy we have to generate new discoveries—whether they appear serendipitous or not—is to innovate and improve our tools? Could that be the missing key to reducing chance in science and sparking more new unexpected discoveries—from CRISPR to exoplanets? Box 2.1 The mystery of discovery: Karl Popper's view on the illogical nature of breakthroughs The Austrian philosopher of science Karl Popper published his seminal book in 1959: The logic of scientific discovery . Ironically, there he argues that identifying a logic of how scientific discovery emerges is not possible. Popper's core argument is that discovery arises through a scientist's unexpected intuitions and chance observations—moments that cannot be predicted because each discovery involves an irrational element. In his view, chance and serendipity play a strong role in discovery and make it unexplainable. As he points out, Einstein also argues that for universal laws of science: 'There is no logical path … leading to these … laws. They can only be reached by intuition' . The consequence for Popper is to shift the focus from how discoveries happen to testing whether theories and discoveries hold up to scrutiny. In fact, for Popper, 'The question how it happens that a new idea … [or] scientific theory [arises] is irrelevant to the logical analysis of scientific knowledge' . For him, scientific knowledge grows when scientists propose theories and then test and attempt to falsify (disprove) them through experiments. Other philosophers of science—from Reichenbach to Carnap—also think discovery is a deeply subjective process and highlight its unpredictable nature. For them, discoveries are a mystery of human psychology. These philosophers do not focus on the process of discovery, but on assessing an output—scientific theories. After all, testing theories is easier to measure and evaluate than how we create them. In short, they neglect the fascinating process of how discoveries actually emerge. As we reveal here, we can ultimately uncover measurable features of discovery—and establish a logic of how we trigger breakthroughs.
In statistical studies, we generally cannot make a census or gather data on all individuals in a population because the numbers are too large. Instead, we take a sample—a smaller subset of the population—and use it to draw conclusions. Yet, collecting data as comprehensively as possible is crucial to draw general insights about discovery and the role of serendipity. Here we explore science's over 750 biggest discoveries— from those made by Marie Curie, Nikola Tesla and Charles Darwin, to Louis Pasteur, Rosalind Franklin, James Maxwell and Richard Feynman. This big-picture approach enables us to move from unpredictable features of a few select discoverers to a more predictable pattern across the total population of major discoverers. To explain the extent and nature of serendipity, a discovery is categorised here as involving a serendipitous moment—including an unexpected observation or finding—if the discoverers described it as such in their original discovery publication. Discoveries are also classified as including a serendipitous moment if the discovery is described as such later in prize ceremonies and interviews of the discoverers—within nobel-prize documents, one of six encyclopaedias of science or the seven mentioned science textbooks. In fact, the description for most discoverers comes from two sources, Nobel prize speeches and the discovery-making paper. Through this search strategy, we confirmed that we captured all well-known and classic cases of serendipity found in scientific publications. Discoverers' reported serendipitous moments are the only way to study whether a discoverer experienced such a moment as only they had access to the entire discovery process. It is important to note that serendipity is not luck. Far from it, serendipity depends on our ability to use the needed tool to make the discovery and recognise it is a discovery. Luck means blind chance or passive stumbling—that we do not find. A serendipitous moment reflects one element in the broader discovery process that involves leveraging scientific methods and tools and often systematic research and experiments. We can also assess how effective tools are by applying both general criteria like empirical power and how generalisable they are—and tool-specific criteria like resolution, sensitivity, speed and scope. Tool development is precisely where science is most intentional and design-driven in response to bottlenecks in our tools—while serendipity, in contrast, is not about planning and strategy.
discoveries How common is a serendipitous moment in science's major discoveries? About one in six. In total, 84% of discoveries have been sparked using just tools and methods and 16% using tools and methods and involve a serendipitous moment. So for discoveries with a serendipitous element, there is one thing they have in common: using a new method or tool—whether an electron microscope, an advanced x-ray method or sophisticated computer simulations—that generally made the unexpected breakthrough possible. No major discovery—nobel-prize or major non-nobel breakthrough—has been made by accident, without tools and methods, or by nonresearchers. Surprisingly, when we explore trends over time, we find that discoveries with a serendipitous moment are becoming less common. We also uncover that 15% of nobel-prize discoveries involve a serendipitous moment while 18% of major non-nobel discoveries. So the results for both groups of discoveries are comparable and robust—and not just driven by one set of discoveries. But do these overall trends mask important differences across fields? Indeed, discoveries with a serendipitous moment are concentrated most heavily in astronomy—both for nobel-prize and major non-nobel discoveries. In fact, they represent about a quarter of nobel-prize discoveries in astronomy. And yet, what is key is that all of these astronomical breakthroughs were made possible by newly developed instruments. These tools enabled us to spark the surprising observations about the universe in a way we never could without them. Take pulsars for example—unexpectedly discovered in 1967 using an immense new radio telescope built the same year at the Mullard Radio Astronomy Observatory. The serendipity was not in stumbling upon something by chance—it was in deliberately using this new tool that allowed revealing something completely new and surprising in the universe.