Terman’s Termites: what happened to 1,528 gifted children

01 Key findings

1. The youngest pupil in class was a better tip than the teacher’s pick. Of the 447 Termites found through teachers’ blanks in 1921-22, 88 (19.7%) were named only as the youngest in class and 70 (15.7%) only as the most intelligent.Terman et al. (1925), Vol. I, pp. 31-33
2. 121 boys for every 100 girls in the main group of 643 (1921-22), against 104.5 in the schools searched; yet the three highest IQs were girls’.Terman et al. (1925), Vol. I, pp. 45-53
3. 29.1% of the fathers were professionals (560 fathers, 1921-22), ten times the 2.9% share in Los Angeles and San Francisco in the 1910 census; one father was a labourer.Terman et al. (1925), Vol. I, p. 63
4. By 1955, 70% of the men and 67% of the women had a college degree (827 men, 643 women), when under 8% of their generation did; 80 men and 17 women held a doctorate.Terman and Oden (1959), Vol. V, pp. 65-67
5. Their 1,525 children averaged IQ 132.7 (tested 1939-52): well above 100, but below their parents’ 151, as regression to the mean predicts.Terman and Oden (1959), Vol. V, pp. 3, 141

02 Statistics to quote

Terman’s own data said a class register beat the teacher: being the youngest pupil in class found more of his gifted children than being the one the teacher rated most intelligent.

Leon Falkner, IQ Metrics Research

03 What was Terman's study of gifted children?

It was the first large study to pick children by IQ score and follow them for life. In 1921 Lewis Madison Terman, the Stanford psychologist who had published the Stanford-Binet intelligence test in 1916, began a search of California schools for children who scored in the top 1%. He kept 1,528 of them, most chosen in 1921-22, and reported on them in four of the five volumes of Genetic Studies of Genius (1925 to 1959) ^{[1] [5]}. The children called themselves, and are still called, the Termites.

Terman was born on 15 January 1877 in Johnson County, Indiana, and died at Stanford on 21 December 1956 ^[6], so 15 January 2027 is his 150th birthday. His team kept 95% of the living members active for more than 30 years ^[5], and others carried on after his death. This page rebuilds the study’s own published tables into charts and one open dataset. For how his test worked, see our guide to the Stanford-Binet IQ test.

Stanford-Binet (1916)

Terman’s American revision of the Binet-Simon scale from France, and the test that chose most Termites ^[1].

Ratio IQ

The IQ of Terman’s day: mental age divided by real age, times 100. A 10-year-old who passes the tests for age 15 scores 150.

Deviation IQ

The modern IQ: a score placed against people of the same age, with 100 as the average and 15 points as one standard deviation. Old ratio IQs at the top are not the same as modern scores.

Corrected IQ

Terman added months of mental age for older children who passed almost every test, because the 1916 scale ran out of hard items ^[1].

Termites

The nickname of the 1,528 people in the study.

Concept Mastery Test

A hard adult test of word meanings and analogies that Terman’s team built for the follow-ups of 1939-40 and 1950-52 ^[5].

Table 1

The study, wave by wave

Years	What happened	Who took part	Reported in
1921-22	The main search in Los Angeles, San Francisco and the East Bay, plus a high-school search and volunteers elsewhere	643 in the main group; 1,444 on file by May 1924	Vol. I (1925) [1]
1927-28	First field follow-up, average age 16 to 17; 58 younger siblings added		Vol. III (1930) [3]
1936	Mail survey	About 90% located	Vol. IV (1947) [4]
1939-40	Second field follow-up, average age 29.5	About 98% reached	Vol. IV (1947) [4]
1950-52	Third field follow-up, average age about 40	1,367 of 1,437 living (95.1%)	Vol. V (1959) [5]
1955	Mail survey, average age 44	1,326 of 1,424 living (93.1%)	Vol. V (1959) [5]
1956	Lewis Terman dies on 21 December; the study goes on		Boring (1959) [6]
1972	Survey of the women, average age 62	430 women (64%)	Shurkin (1992) [11]
1986	Mail survey, average age 76		Holahan and Sears (1995) [10]; Shurkin (1992) [11]

04 How were the Termites chosen?

Teachers named candidates, a group test sifted them and the full Stanford-Binet decided. In each class of 30 to 50 pupils the teacher wrote down the one to three most intelligent pupils and the youngest. Those nominees sat a group test, and the best of them took the individual Stanford-Binet. A child qualified at IQ 140, or a little lower for ages 11 to 14 ^[1]. The main search covered about 168,000 pupils in Los Angeles, San Francisco, Oakland, Berkeley and Alameda and found 649 children, about 1 in 258 ^[1].

Figure 1

How the 1921-22 main search narrowed about 168,000 pupils to 643 children

Public schools of Los Angeles, San Francisco and the East Bay, grades 1 to 8. The first three counts are Terman's own round estimates. Between the bars: what happened at each step.

Pupils covered by the searchabout168,000all

Teachers named the brightest and the youngest; 6 to 8% were tested

Different children nominatedabout; from some 20,000 names10,0006.0% of all

A group test; about the top 5 to 10% for age went on

Took the group test (National B)about8,0004.8% of all

IQ 140 on the Stanford-Binet; 30% came by other routes (Figure 2)

Qualified on the Stanford-Binetby every route6490.4% of all

Parents of 6 children refused

Joined the main groupparents agreed6430.4% of all

Nearly a third of them did not come through the teachers’ blanks at all. Some had scored high on an earlier test, some were recommended, some were brothers or sisters of a child already chosen, and 15 by accident or by means not recorded. In one school a teacher meaning to name the youngest pupil wrote down the name next to it on the roll; that child was the only one of the building’s 300 pupils who reached IQ 140 ^[1].

Figure 2

How the 644 children of the main search were found

Each teacher's blank asked for the most intelligent pupils and the youngest in the class. Orange: children named only as the youngest; dark blue: named only as the most intelligent. Hover or tap a ribbon.

Main search, 1921-22FoundHow they were foundHowWhat the teacher's blank saidThe blank said

Found in the main searchAll found 644

Named on a teacher's nomination blankTeacher's blank 447Found another wayOther way 197

Youngest in class, named only for thatYoungest only 88Most intelligent in class, only thatBrightest only 70Second most intelligent, only that2nd brightest 42Third most intelligent, only that3rd brightest 18Named by last term's teacher, only thatLast teacher 17Named in two waysTwo ways 157Named in three waysThree ways 55High score in an earlier testEarlier test 83Special recommendationRecommended 48Brother or sister already inSibling 41By accident or not knownAccident 15Second search of two schoolsRe-search 10

The blanks held the study’s most quoted surprise. More Termites were found only because a teacher named them the youngest in class (88, or 19.7% of the 447 found through the blanks) than only because she named them the most intelligent (70, or 15.7%). Terman, whose figures rest on blanks from about 6,000 teachers, concluded that the class register was a better guide to the brightest child than the teacher’s opinion ^[1]. He also estimated that the search missed 10 to 20% of the children who would have qualified ^[1].

05 How high were the Termites' IQ scores?

The 643 children of the main group averaged IQ 151 on Terman’s corrected scores, and the 1,070 Termites tested on the Stanford-Binet averaged 151.0, with a range of 135 to 200 ^{[1] [5]}. Seventy-seven of them scored 170 or more ^[5]. Boys and girls had almost the same average (151.4 and 151.2 in the main group), and the three highest scores were girls’ ^[1].

Figure 3

IQ scores of the 643 children in Terman's main group, boys and girls

Stanford-Binet IQs as Terman corrected them for older children who hit the top of the test, in 5-point bands. The cut-off was 140; 22 children below it were let in for special reasons. The three highest scores were girls'.

Boys352 boys, mean IQ 151.4IQGirls291 girls, mean IQ 151.2

0200–2041

0195–1990

0190–1942

1185–1891

8180–1842

2175–1796

11170–1749

17165–16910

26160–16417

29155–15935

85150–15449

76145–14974

85140–14475

12135–1399

0130–1341

2550750255075

Table 2

How the 1,528 Termites qualified

Test that qualified them	All	Boys	Girls	Mean IQ	Average age when chosen
Stanford-Binet (individual test)	1,070	577	493	151.0	9.7
Terman Group Test (high schools)	428	265	163	142.6*	15.2
National Intelligence Test or Army Alpha	30
All Termites	1,528	857	671		11

Terman set the bar at 140 so that each child would be about the brightest in 200; he did not claim they were geniuses ^{[1] [5]}. For what scores like these mean today, see what an IQ of 140 means, what an IQ of 150 means and our guide to gifted cut-off scores.

06 Why were there more boys than girls in Terman's study?

The extra boys appeared at the test, not in the teachers’ names. The teachers’ blanks that produced a Termite named 1,010 boys and 920 girls, close to the 104.5 boys per 100 girls in those schools; among the children who then qualified, the ratio jumped to 135 ^[1]. The main group ended at 121 boys per 100 girls, and the high-school group, chosen less carefully, at 212 ^[1].

Figure 4

Boys per 100 girls at each step of the search

100 = as many boys as girls. Ratios computed from Terman's counts; he cut them to one decimal, so two print 0.1 lower in his book (109.7 and 212.3).

Pupils in the cities searched (below high school)104.5

Pupils named on the blanks that found a Termite109.8

Qualified through those blanks135.3

Main group (643 children)121.0

Main group with 33 later siblings (676)116.0

Binet group from other cities (356)123.9

High-school group, group test (378)212.4

All 1,528 Termites127.7

100125150175200

Terman weighed four explanations: biased teachers, family patterns, miscarriages and greater spread in boys’ scores. He ruled out teacher bias, since the names on the blanks were almost balanced, and leaned toward the last ^{[1] [5]}. A 2019 review agrees that a slightly wider spread of boys’ scores is enough to tilt the very top, though the girls’ scores here were just as spread ^[15]. Private schools, which might have changed the mix, were never searched ^[1].

07 Did the Termites come from richer families?

Mostly, yes. In 1921-22, 29.1% of the main group’s fathers worked in the professions, ten times the 2.9% share of professionals in Los Angeles and San Francisco, and only one father in 560 was a labourer ^[1]. A third of the fathers had a college degree ^[5]. Terman read this as evidence that intelligence is inherited; today it reads as a sample tilted toward comfortable homes.

Figure 5

Fathers' jobs: the gifted children's families against their cities

Percent in each census job group. 560 fathers of the main group, 1921-22, against the population of Los Angeles and San Francisco in the 1910 census, as Terman printed it.

Los Angeles and San Francisco, 1910 censusFathers of the gifted children

Professional2.9%29.1%

Public service3.3%4.5%

Commercial36.1%46.2%

Industrial (trades and labour)20.2%57.7%

0%10%20%30%40%50%60%

Only one father (0.2%) was a labourer, against 15.0% of the census population.

The search also left whole groups out. It skipped private and parochial schools and the segregated schools that served Chinese-American children, and examined rural schools in only one county ^{[10] [15]}. The group that resulted was urban and mostly white, with very few Black, Hispanic or Asian-American children ^{[12] [15]}. That is a limit of the sample, and this page makes no comparisons between racial groups.

08 Were gifted children weak, odd or one-sided?

No. On almost every measure Terman took, the gifted children were at or above average: taller and heavier than the norms of the day, healthier on the doctors’ checks, and rated by their teachers at or above their classmates’ average, more often than not, on 23 of 25 traits for boys and 24 for girls ^{[1] [5]}. Their lead was largest on intellectual traits and smallest on social and physical ones.

Figure 6

Where teachers rated the gifted children above their classmates, 1921-22

Percent of gifted children a teacher rated at or above the average of a control group of the same sex, averaged over the traits in each of Terman's trait groups (our averages). Dashed: 50%, no different from classmates.

Gifted boysNo different from classmates

ShowGifted boysGifted girls

Intellectual (4 traits)91%Classmates 50%Will and drive (4)81%Classmates 50%Emotional (3)68%Classmates 50%Aesthetic (2)63%Classmates 50%Moral (4)64%Classmates 50%Physical (2)58%Classmates 50%Social (5)56%Classmates 50%Mechanical (1)48%Classmates 50%

Teachers rated 98% of the gifted boys and 96% of the girls at or above the control group’s average in general intelligence, but only 53% of boys and 59% of girls in popularity, and 48% and 55% in fondness for large groups ^[1]. The lowest trait for both sexes, mechanical ingenuity (48% and 46%), Terman put down to teachers’ error, since tests of mechanical skill showed the opposite ^[5].

Table 3

All 25 traits: percent of gifted rated at or above the control average

Trait	Trait group	Gifted boys	Gifted girls
General intelligence	Intellectual	98%	96%
Desire to know	Intellectual	93%	87%
Originality	Intellectual	87%	88%
Common sense	Intellectual	86%	83%
Will and perseverance	Volitional	83%	86%
Desire to excel	Volitional	81%	88%
Self-confidence	Volitional	81%	82%
Prudence and forethought	Volitional	79%	83%
Sense of humor	Emotional	72%	77%
Cheerfulness and optimism	Emotional	66%	62%
Permanence of moods	Emotional	65%	61%
Musical appreciation	Aesthetic	64%	68%
Appreciation of beauty	Aesthetic	62%	66%
Conscientiousness	Moral	71%	74%
Truthfulness	Moral	72%	70%
Sympathy and tenderness	Moral	59%	58%
Generosity and unselfishness	Moral	55%	55%
Health	Physical	58%	62%
Physical energy	Physical	58%	66%
Leadership	Social	67%	73%
Sensitiveness to approval	Social	58%	56%
Popularity	Social	53%	59%
Freedom from vanity	Social	52%	53%
Fondness for large groups	Social	48%	55%
Mechanical ingenuity	Mechanical	48%	46%
All 25 traits combined		70%	72%

Character tests pointed the same way: on the total of seven tests, 86% of gifted boys and 84% of girls scored at or above the average of unselected children aged 10 to 14 ^[5]. Terman warned that the tests could partly reward children who guessed the answer the tester wanted ^[5].

Table 4

Seven character tests, ages 10 to 14: percent of gifted at or above the control average

Test	Gifted boys	Gifted girls
Overstatement A	57%	59%
Overstatement B	63%	73%
Book preferences	74%	76%
Character preferences	77%	81%
Social attitudes	86%	83%
Cheating test	68%	61%
Emotional stability	67%	75%
Total score	86%	84%

09 Did the Termites stay ahead as adults?

Yes, on average. When retested on a hard adult test in 1939-40 and again in 1950-52, the same 768 Termites scored higher the second time, and so did their husbands and wives ^{[5] [7]}. Terman’s team read this as evidence that this kind of intelligence keeps growing to at least age 50; the gains are approximate, because the two test forms were linked through a small sample.

Figure 7

Adult test scores rose between 1940 and 1952, for the Termites and their spouses

Mean score on the Concept Mastery Test, a hard adult test of word meanings and analogies, for the same people tested twice, 11 to 12 years apart.

TermitesTheir husbands and wives

1939-4019401950-521952

Men 98.6Women 94.9Husbands 65.3Wives 59.8

114.9 Men110.4 Women80.2 Husbands71.2 Wives

422 men, 346 women, 143 husbands, 191 wives. The 1950-52 form was converted to the 1939-40 scale from 148 students and staff who took both forms, so treat the gains as approximate.

Childhood IQ still predicted adult scores, though loosely: the correlation was .29 ^[5]. The 146 Termites who never went to college averaged the same adult score (118) as a group of advanced graduate students at the University of California ^[5]. For a modern test of the same question, see our story on whether high-scoring children at 7 stay high.

Table 5

Childhood IQ and adult score 30 years later

Childhood Stanford-Binet IQ	Tested 1950-52	Mean Concept Mastery score	Spread (SD)
170 and above	48	155.8	23.1
160-169	70	146.2	26.2
150-159	200	136.5	29.0
140-149	344	131.8	28.6
135-139	41	114.2	33.3

10 What happened to the Termites as adults?

They went much further in school and work than their generation. By 1955, 70.0% of the men and 66.7% of the women had a college degree, when under 8% of people their age graduated; 80 men and 17 women had a PhD or other doctorate ^[5]. By 1955, 86% of the employed men were in professional or managerial jobs ^[5].

Table 6

How far the Termites went in school, by 1955

Schooling	Men	Women
College degree	579 (70.0%)	429 (66.7%)
One to four years of college, no degree	139 (16.8%)	105 (16.3%)
High school plus special training	38 (4.6%)	28 (4.4%)
High school	69 (8.4%)	72 (11.2%)
Did not finish high school	2 (0.2%)	9 (1.4%)
Highest degree: PhD or other doctorate	80	17
Highest degree: medicine (MD)	49	6
Highest degree: law	90	2

Table 7

Men's jobs at three follow-ups (percent of employed men)

Job group (Minnesota Occupational Scale)	1940	1950	1955
I. Professional	45.4%	45.6%	45.6%
II. Managerial, official, semiprofessional	25.7%	39.4%	40.7%
III. Small business, clerical, skilled trades	20.7%	12.0%	10.9%
IV. Agriculture	1.2%	1.8%	1.6%
V. Semiskilled	6.2%	1.2%	1.2%
VI. Slightly skilled	0.7%	0	0
Employed men counted	724	762	757

By 1955 the group counted 77 people in American Men of Science (70 men, 7 women), 33 in Who’s Who in America and three men in the National Academy of Sciences; the men had published about 2,000 scientific and technical papers and held at least 230 patents ^[5]. Termites named in later accounts include the physiologist Ancel Keys, the Los Alamos director Norris Bradbury, I Love Lucy creator Jess Oppenheimer and the psychologists Lee Cronbach and Robert Sears, who later ran the study ^[11].

Many of the women, born around 1910, were homemakers: in 1955 half had no paid job ^[5], a pattern Terman’s team put down to the expectations of the time. A 1940 comparison of the 150 most and 150 least successful men found almost no difference in IQ; the gaps were in persistence, working toward goals, self-confidence and freedom from feelings of inferiority ^[5]. Later work on another group of gifted children, the SMPY study, asks a similar question: see does IQ above 120 stop mattering? ^[16]

11 How smart were the Termites' children?

Bright, but less extreme than their parents. The 1,525 children who were tested averaged IQ 132.7, with boys and girls exactly level; about a third scored 140 or more and about 2% under 100 ^[5]. That fits regression to the mean: children of parents chosen for very high scores tend to score closer to the average, which Terman’s team said matched what Galton’s law predicts ^[5].

Figure 8

IQ scores of 1,525 children of the Termites

Stanford-Binet tests given 1939-40 to 1951-52 to the children the field workers could reach. About a third scored 140 or more; about 2% scored under 100.

76241011693493682621485524102Average: 100Their mean: 132.7

7090110130150170190

Stanford-Binet IQ

Two cautions. The children were tested between 1939 and 1952 against norms set years earlier (the team compares them with the 1937 revision), and scores on ageing norms drift upward over time; and only the children the field workers could reach were tested ^[5].

12 Did the Termites live longer?

Up to their mid-40s, yes. By 1955, 6.9% of the 1,500 Termites still in touch had died, against 8.0% expected from the 1939-41 US life table for white Americans who reached age 11 ^[5]. The comparison is with the whole white population, not with people of the same class and schooling, so part of the gap is likely wealth and education.

Figure 9

Fewer Termites had died by their mid-40s than the life tables expected

Percent dead by about age 44, of people alive at 11. Termites: deaths to 1955 among the 1,500 still in touch (62 of 846 men, 42 of 654 women). Expected: the US white population, 1939-41 life table.

Expected (US white population, 1939-41)Termites, by 1955

Men7.3%9.1%

Women6.4%6.8%

All6.9%8.0%

0%3%6%9%12%

The book prints 6.3% for the women; 42 of 654 is 6.4%. Against the older 1929-31 table, 11.7% were expected.

Later researchers used the Termites to study long life. Children whom parents and teachers rated as conscientious in 1922 lived longer; children rated cheerful, against expectation, did not ^{[8] [9]}. In 1955 about 9% of each sex had at some point had a serious mental health problem, a rate Terman’s team judged about the same as the general population ^[5]. Our study of when the brain is sharpest shows how different skills change with age.

13 Did Terman's study miss two Nobel Prize winners?

One for certain, the other probably. William Shockley, co-inventor of the transistor, was tested twice as a boy and scored 129 and 125, below the cut-off ^[12]. The claim that Luis Alvarez was also tested rests on interviews in a 1992 book and has no published score ^[11]. No Termite is known to have won a Nobel Prize.

Table 8

The two Nobel physicists: the claims and the evidence

Claim	What the record shows	Our verdict
William Shockley (Nobel Prize in Physics, 1956) was tested for the study and not chosen	His biographer Joel Shurkin, working from the Shockley family papers, reports two tests: IQ 129 at about age 8 and 125 a year later. His mother kept “her son’s Terman IQ test”, and Shockley joked about failing it [12]. Born in 1910, he was tested before the 1921 main search.	Supported
Luis Alvarez (Nobel Prize in Physics, 1968) was tested and not chosen	Shurkin (1992) cites “interviews with Shockley and Alvarez”, undated [11]; a 2009 National Academy of Sciences memoir repeats it without a source [14]. Alvarez’s own 1987 memoir never mentions Lewis Terman [13], and no score has been published.	Plausible, not proven
No Termite ever won a Nobel Prize	No member is known to have won one [11] [14]. By 1955 three men had been elected to the National Academy of Sciences [5].	True as far as known

Missing them says less about IQ tests than it seems. Russell Warne points out that with about 168,000 children screened, a future Nobel winner was a one-in-tens-of-thousands case; catching one would have needed a near-perfect test and a near-perfect link between IQ and scientific prizes ^[15]. For what is known about the measured IQs of Nobel winners, see IQ of Nobel Prize winners; for record scores, the highest IQ ever recorded.

14 What is wrong with Terman's study?

Its sample, its lack of a true comparison group and its author’s politics. Terman was a eugenicist: he sat on the boards of eugenics groups and pushed for the forced sterilization of people labelled feebleminded ^{[17] [15]}. He wrote that test-score differences between races were inborn, and late in life told friends he regretted some of those statements ^[17]. In 2018 Palo Alto renamed its Terman Middle School over his eugenics ^[18]. Our article on eugenics and the history of IQ testing covers that record in full.

15 Where can I read the original volumes?

Scans of all five volumes are on the Internet Archive; Volumes I to III are in the US public domain. Volume II, Catharine Cox’s estimates of the childhood IQs of 301 historical geniuses, is a separate study that this page does not cover ^[2].

Table 9

The five volumes of Genetic Studies of Genius

Volume	Year	Authors	Covers	Where to read it
I. Mental and Physical Traits of a Thousand Gifted Children	1,925	Lewis M. Terman and others	The search and the children around age 10	archive.org (US public domain)
II. The Early Mental Traits of Three Hundred Geniuses	1,926	Catharine M. Cox	Estimated childhood IQs of 301 historical figures; a separate study, not used here	archive.org (US public domain)
III. The Promise of Youth	1,930	Barbara S. Burks, Dortha W. Jensen and Lewis M. Terman	The 1927-28 follow-up	archive.org (US public domain since 2026)
IV. The Gifted Child Grows Up	1,947	Lewis M. Terman and Melita H. Oden	The 1936-46 follow-ups	archive.org (lending copy)
V. The Gifted Group at Mid-Life	1,959	Lewis M. Terman and Melita H. Oden	The 1950-52 and 1955 follow-ups	archive.org

16 How was this study built?

1. 01

   Read the volumes

   We read Volume I (1925) in two scans, the 1925 and 1926 printings, and Volume V (1959), and typed each number from its printed table, with the page ^{[1] [5]}. Every table was checked against its own printed totals by script.

2. 02

   Compute only what is missing

   Ratios of boys to girls, shares, the trait-group averages in Figure 6 and the women’s death rate were computed from Terman’s counts and are marked as ours. Two printing quirks are noted where they occur.

3. 03

   Check the famous claims

   The Shockley and Alvarez stories, Terman’s dates and the later findings were checked against the best source we could find, including search-inside of the books on the Internet Archive ^{[11] [12] [13] [6]}.

4. 04

   Recompute every figure

   A separate script rebuilt every number in every figure from the transcribed tables and matched it to the page.

5. 05

   Publish the data

   Every number on this page is in one CSV under CC BY 4.0, with the volume and page it came from. No individual records were used. Our standards are on the methods page.