How Do We Know Einstein's IQ? How Historical IQ Estimates Work

Why Einstein has no real number

Albert Einstein was born in Ulm on March 14, 1879. No record shows he sat an intelligence test. He is also absent from the largest scholarly study of historical IQ, because he was born 29 years after its cutoff. The figure of 160 that follows his name online has no identifiable author or method. Our profile of Einstein's IQ traces the claim in detail.

Einstein is the clearest case of a wider problem. The IQs attached to historical figures fall into three categories: scholarly estimates with a named method, informal guesses by writers, and numbers with no author at all. Only the first category deserves quotation, and it begins with one book.

From Galton to Terman

The idea of reading genius from the historical record is older than the IQ test. Francis Galton's Hereditary Genius (1869) was the first social scientific attempt to study eminence. Lewis Terman published the Stanford Revision of the Binet-Simon test in 1916 and, a year later, turned it back on Galton himself.

In "The Intelligence Quotient of Francis Galton in Childhood" (American Journal of Psychology, 1917), Terman compared Galton's documented childhood feats with his test's age norms. Galton was reading by 2 and knew some Greek, Latin and long division at 5. Terman concluded that the boy's mental age was about twice his actual age, a ratio IQ near 200. The method was born. For the wider story of testing, see the history of the IQ test.

How Cox built the study

Catharine Cox, a Stanford doctoral student of Terman's, scaled the method up. Her dissertation became Volume II of Terman's Genetic Studies of Genius in 1926. She started from James McKeen Cattell's 1903 list of the 1,000 most eminent people in history, ranked by the space biographical dictionaries gave them.

She then cut the list. Hereditary aristocrats and nobility went, since position rather than performance could explain their fame. So did everyone born before 1450 and everyone ranked below 510th. The main group that remained held 282 people; with a preliminary group of 19, the study covered 301.

For each subject Cox assembled a dossier on early interests, education, school standing, reading and production, with ages attached. Three raters, Cox, Maud Merrill and Terman, read every file and judged the mental age the evidence implied, using Stanford-Binet norms as the yardstick.

A1 and A2: two estimates per person

Each subject received two ratings. The first, A1, drew on the record from birth to age 17. The second, A2, added the evidence from 17 to 26, when university work and first publications appear. Both were ratio IQs: mental age divided by chronological age, times 100.

Cox also scored how far each estimate could be trusted. The naturalist Buffon's childhood record earned a reliability coefficient of just .11; Goethe's earned .82. The ratings below, taken from the book, show how wide both the results and the doubts could be.

Why the numbers were corrected upward

Cox saw the flaw in her own raw numbers. Raters treated each subject as a member of an unselected population averaging 100 and credited only what the records showed. Thin records therefore dragged estimates toward the average. "The more complete the record the higher the average IQ for any considerable group in our study," she wrote.

The records alone supported an average rating between 135 and 145 for the whole group. Cox then corrected each score by regression toward the mean of the subject's own professional group rather than toward 100. She concluded that the true mean IQ of her subjects was "not below 155 and probably at least as high as 165," and she named 14 people, among them Isaac Newton, Pascal, Goethe, Voltaire, Laplace and Michelangelo, who "probably rated at 200 IQ or even higher."

That is why one person can carry three or four different "Cox numbers" online: A1, A2 and the corrected version of each. Anyone quoting a Cox figure should say which one.

What the study found by field

Cox's comparisons between fields are sturdier than any single name. Her philosophers rated highest; for them the corrected figure was "not less than 175." Scientists came second. At the other end, "the Soldiers probably rated lowest in IQ, with the Musicians next above them," and military eminence was the one field in which she found no relationship between childhood IQ and achievement.

Those results frame our field rankings: philosophers' IQs ranked, physicists, mathematicians, composers and military leaders. Cox's own conclusion looked past the numbers. Eminent youths, she found, combined high intelligence with "persistence of motive and effort, confidence in their abilities, and great strength or force of character."

The case against the method

The sharpest criticism is built into Cox's own finding. If fuller biographies produce higher IQs, the estimates partly measure historians' attention, not the subject's mind. Joel Shurkin put it bluntly in Terman's Kids (1992): "She, of course, was not measuring IQ, she was measuring the length of biographies in a book."

Rater disagreement is the second problem. Three trained psychologists read the same file on young Mozart and returned 125, 140 and 190. The third problem is the scale. Cox's figures are ratio IQs; modern scores are deviation IQs with a mean of 100 and a standard deviation of 15. Even Terman's 1916 label of "near genius or genius" for scores above 140 was gone by the 1937 revision, and no modern test uses it.

The yardstick also moves. Average scores on IQ tests rose about three points per decade in the United States across the 20th century, as scaled by the Wechsler tests. James Flynn documented the pattern for Americans in 1984 and across 14 nations in 1987, and in several countries, Norway and Denmark among them, the rise has since stalled or reversed. Cox's raters judged children born centuries earlier against early 20th-century American norms. That was a reasonable research convenience. It is not a measurement that can be set beside a modern test result.

Simonton and the modern revival

The psychologist Dean Keith Simonton has done the most to test and extend Cox's approach. He practices historiometry, which he defines as "a quantitative method of statistical analysis for retrospective data." In 2006, in Political Psychology 27(4), he estimated IQs for 42 U.S. presidents. The study drew wide attention for placing George W. Bush between 111.1 and 138.5, with an average of 125, above only Warren Harding among presidents since 1900. Other researchers judged the approach reasonable, though imprecise.

Our profiles of the founding generation show how such numbers were assembled: Thomas Jefferson, Abraham Lincoln, Benjamin Franklin and Alexander Hamilton. In 2009 Simonton and Anna Song reanalyzed Cox's unpublished data on her 282 main subjects in Psychological Science. Eminence correlated positively with IQ, and IQ was related positively to mental health and negatively to physical health.

Tesla 310, da Vinci 220: numbers without authors

Most viral historical IQs fail the first question: who estimated this, and how? A claimed "Tesla 310" has no author, no method and no data. Nikola Tesla was born in 1856, outside Cox's range, and no scholarly estimate for him exists. On the modern scale, 310 would sit 14 standard deviations above the mean, a rarity of about one person in 10 to the 44th power, a number that dwarfs every human who has ever lived. See Nikola Tesla's IQ.

"Da Vinci 220" fails the same way, and Cox contradicts it: she placed Leonardo in her A1 group rated at 135, alongside Darwin and Kant. No test and no scholar produced 220. Numbers near 200 also collide with arithmetic. On a deviation scale, 200 is one person in roughly 76 billion, nearly ten times the world's population. Childhood ratio IQs escape that limit, which is how Marilyn vos Savant's listed 228 arose; IQ 228 explains the difference.

Five questions sort any historical IQ claim. Who produced it? By what method: biographical rating, historiometry or a disclosed test? Which figure: A1, A2 or corrected? On which scale: ratio IQ, or a deviation scale with a standard deviation of 15, 16 or 24? And is it possible at all? See IQ scale classifications and the IQ rarity chart.

Who falls outside Cox's window

Two large groups have no Cox estimate at all. The first is everyone born before 1450: Socrates, Plato, Aristotle, Archimedes, Pythagoras, Alexander the Great, Hannibal, Julius Caesar and Thomas Aquinas. Numbers attached to them are guesses.

The second is everyone born after 1850, which includes most of the modern pantheon: Sigmund Freud, Max Planck, Niels Bohr, Srinivasa Ramanujan, Kurt Gödel, John von Neumann, Alan Turing, John Nash and Carl Sagan. For them the evidence is the work itself, and occasionally a score the person disclosed, as Isaac Asimov did.

Editorial standard: IQ figures for historical figures are retrospective estimates and are labeled as such; scores for living people appear only when the person disclosed them or a reputable outlet reported them. Corrections: contact the editors.