Spearman and the Importance of Archives
Spearman and the importance of archives
MOST of us enjoy reading about the history of our subject; some of us deem it necessary, if we are not to keep repeating what has gone before. But few of us are aware of just how useful archives of psychology might be to us. At the end of March 2006 the Society’s History of Psychology Centre (HoPC) archive quietly closed its doors, leaving Society members – psychologists and historians alike – largely unconscious of their loss. The archive contained a massive collection of papers by the social psychologist Henri Tajfel, manuals by the educational psychologist Philip Vernon, pamphlets, notes, reprints and correspondence by psychologists Charles Myers, Leslie Hearnshaw and Cyril Burt.
The archive also contained correspondence and extensive notes by,
and appertaining to, Charles Spearman (most famous for originating factor analysis). Taking Spearman as an example and centring on just a few areas of interest, I hope to demonstrate the usefulness of archives and some of what will be lost if access to them is restricted.
Why Spearman is important
Factor analysis, a statistical method for achieving parsimony from correlations, is substantially a product of psychology and has been evolving and developing for just over a hundred years. It should not, therefore, come as any surprise that people have begun to explore the origin and early uses of the technique, and the personalities behind its creation.
In 1904 Charles Spearman, who was working toward his PhD in Leipzig under the supervision of Wilhelm Wundt, introduced the concept of ‘factors’ when
he wrote two papers that revolutionised attitudes to assessing human differences (Spearman, 1904a, 1904b). The papers purported to have found a way of measuring ‘general intelligence’, an assertion that provoked instant indignation and even acrimony among many of the established psychologists of his day. Controversy has raged ever since, although the bases for disagreement are constantly changing.
Spearman’s early adult life had been spent enjoying the sport and privilege accorded to British army officers at that time. But his attempt to leave the army in his mid-thirties in order to study psychology, in Germany, was curtailed after three years by the outbreak of the South African (Boer) Wars and his posting to Guernsey (Lovie, 1995). After his release from the army in December 1902 Spearman returned to Leipzig with his Guernsey-born wife and, quite possibly,
all the raw data for his 1904 papers.
g = general intelligence
Many of the ideas about intelligence at the turn of the last century – the foundations of Spearman’s correlations – involved exam results and class placement; Spearman called these ‘present efficiency’. Spearman then derived an element he called ‘native capacity’ by weighting present efficiency against chronological age. Thereby someone who had achieved particular exam marks at an early age was considered more intelligent than an older child who achieved the same marks. Another type of intelligence accounted for by Spearman was ‘teacher assessment’, although he acknowledged that teachers were unable to classify one child without comparison to others within his or her class. If prompted, they could, however, produce a grading of assessment from brightest to dullest child in the class.
The final element that made up ideas of intelligence at this stage in history was ‘common sense’. Spearman derived a classification of this type for all of his reagents (as they were then known) by prompting the two oldest children in the school in the same way he had the teachers, obtaining similar graded classifications.
What Spearman tested
Judgement of musical ‘Pitch’ employed a chord (which Spearman thought was more accurate than a tuning fork, perhaps affording a finer degree of incremental change); his category called ‘Light’ was an amalgam of various sight, depth and tone tests; and his measures of discrimination of ‘Weight’ differences employed Galton’s easily manufactured apparatus (Galton, 1928) of rifle cartridges filled with different amounts of lead shot and sealed. Spearman was working without a laboratory.
Spearman’s (1904a) paper isolated the legendary ‘g’ by correlating ‘Pitch’, ‘Light’ and ‘Weight’, with cleverness in school and the two assessments of common sense. He then amalgamated the three ‘gradings’ of discrimination, and the three ‘intellective gradings’, and recorded the correlation between these two new categories: ‘general discrimination’ and ‘general intelligence’ (p.269).
Spearman has been heavily criticised for the small number of possible correlates to intelligence he appeared to have measured. However, even within the (1904a) paper he tells us that his new statistical method had allowed him to eliminate procedures like reaction times and pencil tapping from the intelligence equation.
Eliminating reaction times contradicted the findings of Gilbert, whilst Thorndike still hoped that teaching children to tap a pencil quickly would improve their intelligence (Thorndike, 1903). The arguments that ensued led Spearman
to develop a formula for checking the hierarchy of his calculations (the tetrad equation); but the animosity didn’t end there, and Thorndike will return to the story.
What the raw data adds to the picture
I believe we can get a fairly good idea of all the experiments Spearman actually subjected his reagents to by recourse to the Society’s archive data. Spearman’s (1904a) paper offers physical descriptions of some of the reagents, and these descriptions match those appended to raw data held in the (HoPC) archive.
The first easily identifiable reagent was ‘an ex-carpenter (at Guernsey)’. At the age of 78 he had ‘unusually pronounced calcareous degeneration of the arteries’ (Spearman, 1904a, p.292), which is commented upon in the paper and in the raw data.
This participant, it seems, was not very musical, and yet, Spearman tells us, he was able to distinguish pitch to within less than a 1/60th of a semitone. As a student of Wundt (who had studied just noticeable differences in pitch in detail: Wundt, 1902), Spearman was aware that his reagent’s discrimination was considerably more accurate than he could have expected. In the raw data Spearman commented that this reagent’s discrimination of pitch was ‘quite remarkably good’. However, Spearman conducted all his experiments several times with every reagent, and subsequent procedures with this participant were less than impressive.
This dataset also shows clear evidence of the results of ‘Muscular-sense’ tests and ‘Touch-images’ tests. It is clear from the raw data that Spearman did various tests of what he called ‘kinaesthesia’ with his reagents.
The reagent’s smell, apparently, was ‘excellent’, although, according to Spearman, he ‘never had smell images starting in his mind’. Smell, or olfactory sensation, attracted a lot of attention at the end of the 19th and beginning of the 20th centuries. Unfortunately, Spearman’s lack of a laboratory at this time meant that he was unable to carry out the standard tests for acuity of smell sensation, which required an olfactometer. However, he probably used two bottles containing the odours and passed them repeatedly in rapid succession before the nostrils, as described by Myers (1923).
The raw data also shows evidence of taste experiments. According to Wundt, by the end of the 19th century taste sensation was fairly thoroughly understood: indeed, the regions of the tongue responsible for recognising particular tastes had been mapped. Charles Myers (1923) outlined standard procedures for assessing acuity
of taste sensation. Accordingly, Spearman would have dried his reagent’s tongue and, with the aid of a finely pointed glass tube, dropped increasingly strong doses of surprisingly noxious substances on to it.
In the raw data Spearman seems to have been trying to look beyond sensory tests when he recorded the results of asking reagents, to recollect familiar visual stimuli.
The 21-year-old male ‘with bad epilepsy’ mentioned in the paper was
also identifiable in the raw data. Attitudes towards people with epilepsy at the beginning of the 20th century were invariably prejudiced as they were deemed ‘mentally defective’ under the law, but despite Spearman’s membership of the eugenics movement, he appears to have had a better attitude to disability than most people at that time.
This reagent’s answers, concerning discrimination of weight were, according
to Spearman, ‘always confident, but invariably wrong’. Spearman’s handwritten additions to his typed notes are often very revealing, and it is extremely interesting that he deemed it ‘excusable’ for this young man to be overconfident with regard to weight discrimination. Reference to the 1904a paper sheds light on this comment by revealing that Spearman was aware that previous studies had recorded inverse correlations between weight discrimination and teacher’s estimates of intelligence. Spearman also seemed to be impressed with his reagent’s ability to mentally recall a piece of instrumental music while ‘No movements of his voice-muscles accompany image’.
However, it might have been the reagent described as ‘an extraordinarily slow thinker’ (1904a, p.292), who gave Spearman the confidence to question Gilbert’s correlations between reaction times and brightness/dullness. In the raw data Spearman remarked: ‘still he seems to do his work in a satisfactory way, owing to his painstaking and methodical habits (he takes notes of everything in writing instead of trusting to his memory…)’. Spearman appears to have a lot of respect for this man (from Guernsey).
Finally, the ‘Psychologist experimental’ (1904a, p.292) appears to have been Spearman himself. The archive data recorded, amongst other things, the results of his attempts to remember a spoken remark, at different time intervals.
Spearman made meticulous calculations and graphs derived from all his studies and often his handwritten remarks seemed to be in case the data appeared abnormal when compared with data still to be collected. For instance, on one day he appended the words ‘first after a small storm’ to some data on sound afterimages.
Keeping quiet about Guernsey
Karl Pearson was joint author of the correlation coefficient known as Pearson’s product moment (which Spearman based his method on), and he is often identified as the originator of principal component analysis. He is possibly equally famous today for his views about racial differences and his arguments with many of his fellow psychologists and statisticians, most notably Fisher, but also Spearman. But in 1904 Pearson’s opinions were highly respected.
Spearman didn’t mention, in his papers, where the adult reagents came from. Could this have been because Pearson had just published a paper entitled ‘On the laws of inheritance in man’, in which he described the people of Guernsey as so different from other races as to render any data collected from them unusable?
What happened next
Having obtained his doctorate in 1906, Spearman left Germany in 1907 with a testimonial supplied by Wundt. Spearman’s handwritten translation shows that Wundt described Spearman as combining a ‘widely intensive knowledge of philosophy with an uncommon mastery of psychology’. He went on to outline Spearman’s strengths in the ‘physical mathematical sciences auxiliary to psychology’. It should be remembered that the discipline of psychology, at this stage in history, was subordinate to philosophy, and research methods were predominantly a matter of introspection. Spearman spent the rest of his working life at the University of London mostly in the Galton laboratory, where Pearson also worked.
The Binet-Simon scales (1905–8), which included a test of weight discrimination, together with many culture-dependent measures like ‘compare two figures from an aesthetic viewpoint’ (p.420), took the lead in intelligence testing.
Spearman replaced McDougall on the University of London faculty, but Spearman’s duties would have been familiar to him. Digging into the archives again we find a lesson plan for a first-year course, labelled ‘Lecture 10’ and with Spearman’s name on it. It reveals that taste experiments required that the students’ tongues should be painted with such things as zinc, copper, hydrochloric acid and hydrochlorate cocaine. Smell sensation procedures treated them to sulphuric ether, opium and decaying animal matter. That these lectures formed part of an existing curriculum is born out by a question on a practical exam paper, dated 1905 (two years prior to Spearman’s arrival), which asks, ‘With the apparatus provided, determine the threshold for the smell of camphor’.
In 1909 Thorndike launched his most vicious attack on Spearman’s 1904 papers (Thorndike et al., 1909). Less than a month later, Cyril Burt, who often asked for Spearman’s help – although he was not one of his students – sent Spearman the manuscript for his 81-page (1909) article, for comment. It is not surprising that it took Spearman several weeks to reply. Spearman’s reply was, though, not only thorough (covering more than five sheets of foolscap paper), but also provided solutions to dilemmas Burt was experiencing, including a mathematical calculation that addressed a particular problem he had concerning hierarchy.
Spearman’s most famous book The Abilities of Man was first published in 1927 and prompted the statistician E.B. Wilson to conclude that it read like the work of someone who was under considerable criticism. Wilson also took his first opportunity to meet Spearman (Lovie & Lovie, 1995). The impression that one gains from reading the published disputes that erupted between the two men is very much tempered by even the briefest reading of the many letters that passed between them. For example, following an unpleasant attack by Karl Pearson (for whom Wilson had little time) on Spearman in Nature, Wilson wrote to Spearman as a good friend offering comfort.
Spearman continued to work on his ideas for intelligence tests without cultural/educational bias, and there are copious notes and little drawings amongst the other paperwork in the Spearman boxes of HoPC. Spearman corresponded regularly with his collaborators, most notably Holzinger, to whom he often wrote twice a week over many years. He answered letters from students, people having problems analysing their data, admirers and detractors alike.
Spearman was under almost constant pressure to defend his ideas. Therefore, when Burt failed to mention Spearman’s help in connection with the history of the ‘proportional equation’, Spearman wrote to Burt and pointed out that the equation he had given Burt in 1909 might have led to his development of the aforementioned formula. Burt waited many weeks before returning a letter that was almost as long
as Spearman’s (1909) original review of Burt’s manuscript. Burt did not concede any ground; he defended his published attributions and, referring to Spearman’s formula, said ‘the equation I have in mind does not seem to follow from this’. The disputed figures and procedures are clearly stated in the letters, but the letters offer more than a possible mathematical solution to an interesting historical dispute: they convey a strong sense of the shift in power that had taken place during the 28 years that divided the sets of correspondence.
The Society’s place in history
A lot can be learnt about the history of a subject by reading the published material, but normally only the successes make their way into print. Archives bring history to life. They go beyond the restrictions imposed by publishers, beyond faceless, scientific writing conventions. They include jottings of ideas, communications between networks of scientists, experiments that were tried and dropped. They even give you the sense of communicating personally with someone (possibly long dead) who nevertheless was thinking about the same sorts of issues that you are grappling with yourself.
When a new London office was mooted by the Society, part of the reason given for its necessity was the need for more space for the valuable archives. The sentiment was correct; the archives are, without doubt, crucial to any thorough understanding of our discipline. They require space to allow them to evolve; a director who understands the material and who is able to be of practical assistance to those consulting them; and they have to be accessible. Currently, however, it is thought that the space can be more profitably used for extra meeting venues. Members should make their feelings known, to ensure that the valuable information is not simply consigned to history.
Julie Perks is a research student at Staffordshire University.
E-mail: [email protected].
A potted history of Spearman: human-nature.com/nibbs/03/spearman.html
A brief history of Pearson: tinyurl.com/jly5p
Lists of archives and a summary of their contents, including HoPC: nationalarchives.gov.uk/nra/default.htm
Full text, historically important articles, including Spearman’s 1904a: psychclassics.yorku.ca/index.htm
Discuss and debate
Do we, as members, expect our Society to care for and make available the archive material that is donated to it?
Is it useful to know the background to historic findings?
Does British, or even European, psychology merit a place in history?
Should the Society be able to switch the allocation of resources without consulting the membership?
Have your say on these or other issues this article raises.
E-mail ‘Letters’ on [email protected] or contribute to our forum via www.thepsychologist.org.uk.
Binet, A. & Simon, T. (1905–1908). The development of the Binet-Simon Scale: New methods for the diagnosis of the intellectual level of subnormals (E.S. Fite, Trans.) In D. Wayne (Ed.) Readings in the History of Psychology. New York: Appleton-Century-Crofts.
Galton, F. (1928). Inquiries into human faculty and its development. London: J.M. Dent & Co.
Lovie, P. (1995). Charles Edward Spearman F.R.S. 1863–1945. A commemoration on the 50th anniversary of his death. British Journal of Mathematical and Statistical Psychology, 48, 209–210.
Lovie, P. & Lovie, A.D. (1995). The cold equations: Spearman and Wilson on factor indeterminacy. British Journal of Mathematical and Statistical Psychology, 48, 237–253.
Myers, C.S. (1923). A text-book of experimental psychology with laboratory exercises: Part II Laboratory exercises (2nd edn). Cambridge: Cambridge University Press.
Spearman, C. (1904a). ‘General intelligence’, objectively determined and measured. American Journal of Psychology, 15, 201–293.
Spearman, C. (1904b). The proof and measurement of the association between two things. American Journal of Psychology, 15, 72–101.
Thorndike, E.L., Lay, W. & Dean, P.R. (1909). Relation of accuracy in sensory discrimination to general intelligence. American Journal of Psychology, 20, 364–369.
Thorndike, E.L. (1903). Educational psychology. New York: Lemcke and Buechner.
Wundt, W. (1902). Outlines of psychology. (C.H. Judd, Trans.) (2nd revised English edn. From 4th revised German edn.). London: Williams & Norgate.
BPS Members can discuss this article
Already a member? Or Create an account
Not a member? Find out about becoming a member or subscriber