How Catharine Cox estimated the IQ of historical figures in 1926

A study from Terman's team

Lewis Terman of Stanford adapted Binet's scale for the United States in 1916 and spent the following years studying children with very high scores. The project became known as Genetic Studies of Genius. The second volume, published in 1926, was Catharine Cox's work, and it asked the question in reverse: instead of following children to see whether they became eminent adults, it started from eminent adults of the past and looked for signs in their childhoods.

Terman himself had tried something similar in 1917 with the scientist Francis Galton, working from letters and accounts of his childhood. Cox extended the method to people born between 1450 and 1850, chosen from a list of eminence compiled earlier by the psychologist James McKeen Cattell, who ranked names by the space they took up in biographical dictionaries.

How the estimate was made

For each person, Cox gathered biographies, letters, school records and family accounts of their childhood and youth. Using this material, raters (Cox, Terman, Maud Merrill and others) estimated what IQ those records would indicate on the scale of the time, the 1916 Stanford-Binet, which was based on the ratio of mental age to chronological age. If the sources said that a boy was doing at 8 what was typical at 12, for example, that pointed to a high ratio between the two ages.

Cox published two values for each person:

• AI: the estimate for childhood, up to age 17.
• AII: the estimate for youth, from 17 to 26, when there were more records.

Each estimate came with a data reliability index, from 0 to 1, indicating how much that person's sources allowed for an estimate. The averages were low: about 0.46 for childhood and 0.54 for youth. When data were missing, the raters pulled the estimate toward 100. That is why the values vary little from one another, and none of the figures we use has a youth estimate below 110.

The obtained value and the “corrected” one

If you look up Newton's IQ online, you will find 190. In Cox's book, the obtained value for Newton's youth is 170. The difference comes from a correction Cox herself proposed: since cases with little data had been pulled toward 100, she applied a statistical group adjustment that pushes each estimate toward the average of the person's field (poets with poets, scientists with scientists), more strongly the less reliable the data.

Cox called this adjustment “no more than an approximation.” It assumes that people in the same field were alike, and the corrected values come out 5 to 25 points above the obtained ones. Those are the numbers that circulate in popular lists, such as Goethe at 210.

IQNorth uses the obtained value for youth, AII. First, because it is an individual estimate, not a group adjustment. The value is printed in each case study in the book, which means it can be checked. And in a section that already leans toward flattery, picking the larger of the two numbers would mean choosing the effect over the data.

Newton has 170 in Cox's book and 190 in internet lists. The difference is a group correction that the author herself called an approximation.

The limits of the sample

Cox's list reflects whom the reference books of the early 20th century considered eminent: almost exclusively European men, many from families with access to education and written records. Women rarely appear (among them, Madame de Staël and George Sand), and people from outside Europe hardly appear at all. The sample shows who was recorded and remembered, and says little about the talent of those left out of the books.

There are other limits. A famous figure's childhood is usually told by biographers who already knew how the story ended, and anecdotes of precocity tend to be remembered and embellished. The scale is different, too: a Cox 140, calculated from the ratio of ages with 1916 norms, is not point for point equivalent to a 140 on a modern test with a mean of 100 and a standard deviation of 15.

In 2009, Dean Keith Simonton and Anna Song went back to Cox's 282 main cases and reanalyzed the estimates alongside each person's eminence, physical and mental health, and field of achievement. Their work shows that the 1926 data still raise questions, as long as you take seriously how much each estimate depends on the information available.

How the values appear on IQNorth

On the historical figures page, we show some of Cox's values as a curiosity: Napoleon, Beethoven, Lincoln and Darwin at 140, Leonardo da Vinci at 150, Mozart at 155, Newton at 170 and Goethe at 200. They always come with the caveat: Cox's estimates (1926), retrospective approximations, not real measurements.

In the report, you see figures whose estimate falls within 10 points of yours, with no ranking and no badge. We never say that anyone has the intelligence of a historical figure: nobody took a test, and the scales are different. If your estimate falls below 110, we show no names, because the study has no figures in that range, and we do not pull names from higher up to make up for it.

See also: Historical figures.

Read next: What was Einstein's IQ?