How Common is Multiple Invention?

Patent sketches of Gray’s (top) and Bell’s (bottom) telephones, via Hounshell.

When Alexander Graham Bell filed a patent for the telephone on February 14th, 1876, he beat competing telephone developer Elisha Gray to the patent office by just a few hours. The resulting legal dispute between Bell Telephone and Western Union (which owned the rights to Gray’s invention) would consume millions of dollars before being resolved in Bell’s favor in 1879.

Such cases of multiple invention are common, and some of the most famous and important modern inventions were invented in parallel. Both Thomas Edison and Joseph Swan patented incandescent lightbulbs in 1880. Jack Kilby and Robert Noyce patented integrated circuits in 1959. Hans von Ohain and Frank Whittle independently invented the jet engine in the 1930s. In a 1922 paper, William Ogburn and Dorothy Thomas documented 150 cases of multiple discovery in science and technology. Robert Merton found 261 examples in 1961, and observed that the phenomenon of multiple discovery was itself a multiple discovery, having been described over and over again since at least the early 19th century.

But exactly how common is multiple invention? The frequency of examples suggests that it can’t be particularly rare, but that doesn’t tell us the rate at which it occurs. In “How Common is Independent Discovery?,” Matt Clancy catalogues several attempts to estimate the frequency of multiple discovery, and tentatively comes up with a frequency of around 2-3% for simultaneous scientific discoveries, and perhaps an 8% chance that a given invention will be reinvented in the next decade. But the evidence for inventions is somewhat inconsistent, and varies greatly between studies. Clancy estimates a reinvention rate of around 8% per decade, but another study he found that looked at patent interference lawsuits between 1998 and 2014 suggests an independent invention rate of only around 0.02% per year.

The frequency of multiple invention is a useful thing to know, because it can give us clues about the nature of technological progress. A very low rate of multiple invention suggests that progress might be driven by a small number of “genius” inventors (what we might call the Great Man Theory of technological progress), and that it might be highly historically contingent (if you re-rolled the dice of history, maybe you get a totally new set of inventions and a different technological palette). A high rate of multiple invention suggests that progress is more a function of broad historical forces (that inventions appear when the conditions are right), and that progress is less contingent (if you re-rolled the dice of history, you’d get a similar progression of inventions). And if the rate of multiple invention is changing over time, perhaps the nature of technological progress is changing as well.

I wanted to create my own estimate of the rate of multiple invention, so I started with Wikipedia’s Timeline of historic inventions, which lists "nonincremental inventions that are widely recognized by reliable sources as having had a direct impact on the course of history that was profound, global, and enduring.” I narrowed this list down to inventions between 1800 and 1970, both to trim the list to a manageable number, and to restrict it to inventions where I could expect a decent amount of available documentation (earlier efforts may have more gaps in the history, and more recent efforts may not have had time to get the historian’s treatment). This yielded a list of around 190 inventions. For each one, I looked to see whether the invention had multiple inventors.

This required specifying what exactly we mean by “multiple.” I ended up using three different categories of multiple invention.

First, there may be multiple descriptions of the same basic idea. When Bell Labs went to file a patent for the transistor in 1948, it was discovered that physicist Julius Lillienfeld had filed a patent for a similar idea in 1926. Lillienfeld likely never actually built his transistor (it would have almost certainly been impossible to build a working one due to material limitations), but he nevertheless described the concept. Similarly, concepts for airships were described many years before a working one was built. I counted these descriptions as multiple inventions only if the description included a basic functional mechanism that could have plausibly worked: So Lillienfeld’s patent counts as a multiple invention, but someone idly stating that “it would be great if there was a solid state amplifier” wouldn’t. (In general, I don’t consider a previous description as meaningful in a “who invented X?” sense — I don’t think Lillienfeld should be credited alongside Shockley, Bardeen and Brattain as a co-inventor of the transistor — but I think it’s relevant if you’re trying to understand whether someone else would have invented something if the original inventor didn’t, so I’ve included descriptions in this exercise.)

Airship concept sketched by Francesco Lana de Terzi in 1670, via Wikipedia. Lift is provided by 4 hollow spheres that have had the air removed to create a vacuum.

The second category of multiple invention I used was cases where there were multiple serious efforts to actually create a given invention. With the Haber-Bosch process for synthesizing ammonia, multiple development efforts attempted to create an ammonia synthesis process using high-pressure and metal catalysts, though only Fritz Haber and Carl Bosch were actually successful. Similarly, there were numerous attempts to build an incandescent light bulb prior to Edison, which often worked but were impractical. As with multiple descriptions, I counted these as multiple inventions only if the efforts were based on some concept that could have plausibly worked. So Ostwald’s ammonia synthesis efforts based on using high pressure and metal catalysts count as a multiple because the concept was in the correct ballpark, even though he ultimately failed. Someone trying to transmute lead into ammonia using alchemy wouldn’t have counted.

Attempts to create a practical incandescent lightbulb, via Friedel.

The third category of multiple invention is if there were multiple successes or near-successes: inventions that actually work, or come close. These are things like Whittle and von Ohain’s jet engines (which first ran in 1939 and 1938, respectively), Edison and Swan’s lightbulbs, and Kilby and Noyce’s integrated circuits.

It wasn’t always straightforward to categorize multiple development efforts. Elisha Gray’s telephone efforts, for instance seem like they could be plausibly described as either a “serious development effort” or a “previous description”. And in some cases, it wasn’t obvious whether something should count as any sort of multiple. Vasily Petrov discovered the electric arc prior to Humphry Davy, but didn’t appear to try to make an actual lamp out of it. These and other cases were ultimately judgment calls.

Besides the type of multiple invention, another important consideration was determining whether efforts are “independent” or not. If multiple efforts are really just many people copying one person’s work, then conceivably an invention might not have happened if that first person didn’t put in the work, even though many people ultimately worked on it. Both Charles Goodyear and Thomas Hancock developed processes for vulcanizing rubber, but Hancock apparently only began working on his process after getting and reverse-engineering a sample of Goodyear’s rubber. Similarly, both Nicolas Appert and Peter Durand developed food-preservation methods, but Durand apparently just patented Appert’s process in England.

Unfortunately, it’s not straightforward to determine whether multiple invention efforts were independent. In some cases, we can be fairly confident that efforts were independent, such as the transistor, where the earlier effort was only uncovered much later. But in many cases it’s hard to be sure. For instance, Edison has been accused of stealing the work of Louis Le Prince for his early motion picture camera, and then having him killed (!) to cover it up, but it’s not clear if this actually happened or if it's just a wild conspiracy theory. Similarly, it’s been suggested that Werner von Siemens got the concept for an electric tram from Russian inventor Fyodor Pirotsky, but apparently all we really know is that Siemens’ had access to a publication that featured Pirotsky’s work. I also didn’t want to be too strict, and exclude a multiple invention merely because the inventor may have been aware of other efforts.

I’ve made my best effort to suss out the relationships at work with multiple inventions, and to exclude what seemed like purely derivative efforts, but I’m sure I haven’t gotten this 100% correct. If the literature asserts that multiple efforts were independent, I’ve generally trusted that without investigating too deeply.

(The full list of inventions and how I classified them is available on Github here.)

How frequent is multiple invention?

Overall, I found that within this sample of historic inventions, multiple invention was extremely common. Out of the 190 inventions I looked at, 105 of them had some type of multiple effort. In 72 cases, these were multiple successes or near successes.

A few lesser-known examples of multiple invention:

• There were many attempts to build an electric telegraph: “An early electrical telegraph based on Volta’s cell was proposed by Catalan physician and experimenter Francisco Salvá Campillo in 1804… Similar results were obtained five years later by the electrochemical telegraph developed by the German scientist Samuel Thomas von Sömmerring and the English scientist Francis Ronalds in 1816.” Samuel Morse’s efforts, which began in 1832, seem to have been undertaken without knowledge of these previous efforts.

• Cyrus McCormick’s mechanical reaper was preceded in the US by Obed Hussey’s, and in England by Patrick Bell’s. Between 1786 and 1831 (when McCormick first operated his reaper), there were more than 30 attempts to build a mechanical reaping machine in England and Scotland.

• Methods for taking photographs were developed by both Louis Daguerre (whose work was based on that of Nicéphore Niépce) and Henry Talbot.

• Stainless steel was discovered by “seven men, unknown to each other and living in four different countries.”

• Charles Parsons is often credited with the invention of the steam turbine in 1884, but Carl Gustav Patrick De Laval invented a different type of steam turbine in 1882, and the concept of a steam turbine had been described for hundreds of years prior.

• The electron microscope was independently invented by a team at TU Berlin in 1931 and Reinhold Rüdenberg in 1932.

We can also look at whether the rate of multiple invention has changed over time. The graph below shows the fraction of historic inventions each decade that had multiple, plausibly independent efforts.

While the graph is noisy, it doesn’t appear that the rate of multiple invention changed much between 1800 and 1970. The rate of multiple successes or near-successes may have declined in the 20th century compared to the 19th, but there’s enough decade to decade variation that it's hard to be sure.

We can also look at whether the frequency of multiple invention varies by the type of invention. I classified inventions into six different categories, based on the type of problem that had to be solved to create the invention.

• Electrical inventions - Like the telephone, the lightbulb, and electric motors or generators.

• Chemical inventions - New chemicals (such as mauve dye), materials (stainless steel), and new processes for producing existing chemicals or materials (the Bessemer process, the Haber-Bosch process).

• Mechanical inventions - New types of machines such as the steam hammer, the rock drill, and various types of aircraft.

• Electronic inventions - Things like vacuum tubes, cathode-ray tube-based television, and digital computers. I also put software inventions (like packet-switched networks) into this category.

• Medical inventions - New medicines (morphine), medical procedures (using hypodermic needles), or inventions to solve health-related problems (Pasteurization).

• Other - Inventions that didn’t fit into other categories. Often these were discoveries of new important physical phenomena (the photovoltaic effect, electrical induction, X-rays) that Wikipedia included on the list of historic inventions.

In a few cases, I put an invention into multiple categories if it seemed like it fit in both. If an engine seemed like it required solving both mechanical problems (building the engine) and chemical problems (experimenting with fuels), I classified it as both mechanical and chemical. Vapor compression-based ice machines were similarly categorized as both mechanical and chemical.

The graph below shows the frequency of multiple invention by category.

Frequency of multiples seems relatively constant across categories. There’s variation between categories, but not necessarily more than I’d expect from random chance given the relatively small sample sizes. Successes or near successes show especially little variation.

Conclusion

It was always clear that multiple invention wasn’t that uncommon, given the number of well-known multiples. But I was still surprised that over 50% of inventions in the time period looked at had some type of multiple effort to create, and that nearly 40% weren’t simply someone having the idea or working on the problem, but successes or near-successes. I was also surprised that this ratio didn’t vary much across time and across categories of invention.

My main takeaway is that the ideas behind inventions are often in some sense “obvious,” or at least not so surprising or unexpected that many people won’t think of them. In some cases, this is probably because once some new possibility comes along, lots of people think of similar things that could be done with it. Once the properties of electricity began to be understood, many people came up with the idea of using it to send signals (telephone, telegraph), or to create motion (engines and generators), or to generate light (arc lamps, incandescent lights). Once the steam engine came along, lots of people had the idea to use it to power various types of vehicles.

In other cases, multiple invention probably occurs because important problems will attract many people trying to solve them. Steel corrosion was a large problem inspiring many folks to look for ways to create a steel that didn’t rust, or notice the potential value if they stumbled across such a material. Lamps causing mine fires were a major problem, inspiring many people to come up with ideas for safety lamps. The smoke produced by gunpowder was a major problem, inspiring many efforts to develop smokeless powders. And because would-be inventors will all draw from the same pool of available technologies, materials, and capabilities when coming up with a solution, there will be a large degree of convergence in the solutions they come up with.

It’s also notable that this rate of multiple invention is much higher than any of the rates in Matt Clancy’s “How common is independent discovery?” piece. I suspect part of the answer here is that we narrowed our focus to only the most important inventions, which are presumably solving important problems that are much more likely to attract inventors. Clancy notes that previous studies on multiple discovery have found that problems recognized to be important tend to have greater rates of multiple discovery (though it’s hard to say by how much). It would be interesting to examine a sample of less-prominent inventions to see the rate of independent invention, but I think this would be hard: the inventions where there’s enough documentation on the various attempts at them are likely the most important ones. The inventions of lesser importance mostly don’t get written about at all.

One final thing to note: Clancy notes that a potential point of fragility in scientific or technological progress is in ideas or inventions that are only recognized as important after the fact. These will have fewer people working on them, and thus (presumably) a lower rate of multiple inventions and a greater chance that they get “missed”. The above analysis suggests that this may not be much of a risk; because it uses a list of “historically significant” inventions, it includes both inventions that were recognized as important at the time (the safety lamp) and ones that were only recognized as important later (electrical induction). Even including both types of invention, the rate of multiples remains very high. (To be more than weak evidence though, we would have to divide the list into inventions that were recognized as important prior to them being invented and ones that weren’t, and look at the rates of multiple inventions for each. But this seems like a very difficult categorization task.)

Thanks to Matt Clancy for reading a draft of this. All errors are my own.