When Things Go Wrong: An Exploration of Error at Rockefeller

by Joe Luna

As anyone at the bench knows, it’s frustrating when a hypothesis doesn’t pan out. Sometimes a succinct and beautifully conceived idea, carefully fashioned to predict something testable, can in an instant be rendered useless by a muddy blot, a blank gel, or an otherwise completely healthy mouse. This prospect is part of the exciting—though slightly terrifying—feeling most anyone feels at the developer or computer screen, at the moment of discovery, when the response for important experiments can be summed up with either “It worked! Let’s have a drink” or “It didn’t work. I need a drink.” It’s little wonder the Faculty and Students Club is so well attended.

Having a hypothesis confirmed is one thing, although a refuted hypothesis is still useful since it obviously tells you what didn’t work. Historians and philosophers of science have long recognized this, and they have argued that this observation says something about our ability to practice science. Philosophers of science often write of something called the “pessimistic meta-induction” (PMI), which is essentially a fancy way to describe what history can teach us about being wrong. The basic idea is this: as history is littered with many scientific models that eventually failed, it’s problematic to assume that our present success rate is any different. Thus, it’s likely that some fraction of currently accepted scientific models will turn out to be wrong in the future. Combined with the specter of hindsight bias—whereby we only recall those ideas that were correct to begin with—it becomes easy to overlook which scientific paradigms could face eventual questioning absent the technological improvements or nurturing environments that make such questions testable. Such a situation can be thought of as the calm before a Thomas Kuhn-style scientific revolution.

In practice, this makes a lot of sense. There’s little reason for a scientist to have to know and catalog overturned scientific ideas, beyond their utility as disposable teaching tools and the occasional colorful anecdote. Somewhat counter to PMI, scientific realism offers a standard and more optimistic view. In general, since science self-corrects over time, absorbing and refining theories as new data warrants, while discarding pieces that don’t agree with observation, we can be confident that a scientific theory approximates some level of truth. Or, put another way, our “success rate” at getting things right gets higher as we build upon scientific advances that have withstood the test of time. Thus, it’s naïve to assert that our present success rate is the same as that of the past, when we knew much less.

There are examples in The Rockefeller University (RU) history that could be claimed by either camp. Those that would be claimed by the realists you’re likely already aware of because these tend to be the stories that are celebrated with repeated visits to Stockholm. But what about those examples where Rockefeller researchers got it wrong? What happened, and how can we learn from them?

There are numerous examples from RU history, though perhaps the most famous is the story of DNA research. Contrary to what you might think, this story doesn’t begin with Oswald Avery, Colin Macleod, and Maclyn McCarty, but with Phoebus A. T. Levene, a founding member of The Rockefeller Institute and well-known nucleic acid chemist. First with RNA, and later with DNA, Levene determined that both were long polymers composed of four chemical entities that he termed “nucleotides” (A, C, G, T, and U), names we can credit Levene with inventing. Over the course of the 1910s and 1920s, Levene’s lab correctly characterized their individual chemistries, determined the nature of their ribose or deoxyribose backbones, and correctly determined that nucleotides were linked via 5’-3’ phosphodiester bonds. All in all, these were stunning achievements. As for the uses of these long polymers, Levene reached an impasse. Because he measured approximately equal amounts of each nucleotide in biological samples, he proposed what became known as the “tetra-nucleotide” hypothesis, where each of the nucleotides in groups of four were connected via 5’-3’ bonds in stacked rings with their bases facing outwards as a structural support for proteins. There was nothing in his model that could suggest a means for information storage, let alone replication. In fact, similar to other chemists of the day, Levene considered nucleic acids to be far too chemically simple to store information. The basic gist, compared to the later Watson-Crick model, was that Levene got it completely backward. Only later could we call the idea wrong.

Being wrong is certainly not the end of the world. The problem, however, was that this work was considered to be definitive for many years. In an era when most thought that proteins carried genetic information, Levene’s model was considered compelling evidence that nucleic acids (with four paltry building blocks) couldn’t do the job. Surely a protein or a series of proteins, each composed of a mix of twenty amino acids, was better suited to explain the complex process of heredity.

Levene died in 1940. Thus, he never was able to grasp later work done in pneumococci by the Avery group that strongly pointed to DNA as the carrier of genetic information. Acceptance of this seismic result came slowly, as there were many holdouts of the protein-centric view, even among RU colleagues (Alfred Mirsky was one particular vocal opponent of Avery’s discovery.) Consequently, Avery’s result was largely overlooked in his lifetime and only later lauded as the monumental discovery we view it as today.

But what of Levene? It is unfortunate that his legacy has been overshadowed by the “tetranucleotide hypothesis,” despite the many key discoveries that make him an arguable great-grandfather of molecular biology. Though as much as Levene could be considered exhibit A for a PMI view of science—Levene needed to be wrong for Avery to be triumphant Avery—it is heartening to know that Levene was an impassioned realist. From his address, upon receiving the Gibbs award, Levene declared:

“Thus step by step, one mystery of life after another is being revealed. Whether the human mind will ever attain complete and absolute knowledge of and complete mastery of life is not essential. It is certain, however, that the revolt of the biochemist against the idea of a restriction to human curiosity will continue. Biochemistry will continue to function as if all knowledge, even that of life, were accessible to human understanding. The past has taught that the solution of some problem always opens up a new one. New discoveries in physics, in mathematics, in theoretical chemistry, furnish new tools to biochemistry, new tools for the solution of old problems and the creation of new ones. So long as life continues, the human mind will create mysteries and biochemistry will play a part in their solution.”

In one way or another, at the bench, we’re all either Levenes or Averys. We’re unknowingly crafting and testing incomplete hypotheses or we’re slaving away, ultimately on the right track, but with little more than fanciful interpretation to explain otherwise strange data. We’re either wrong or we’re right, and we just don’t know it.

For a nuanced and intimate portrait of Levene—the person—(and the source of the above quote,) see:

Van Slyke, D.D., and Jacobs, W. (1945) Biographical Memoir of Phoebus Aaron Theodor Levene, Vol. 23, pp. 75–86, National Academy of Sciences, Washington D.C.

July/August 2012