The first half of this report of the Institute's 1991 symposium appeared in the previous issue of the Newsletter.
Dr Edwards described the statistical concepts which originated with Galton and their subsequent development. The chief contributions were:
In 1872, in his "Statistical Enquiries into Efficacy of Prayer", Galton set out the need for comparison of matched groups so as to evaluate factors one at a time; this method held sway until Fisher’s methods of factorial analysis became avail able.
In 1874 Galton published a paper on extinction of families which, although strictly about probability, was really the start of his influence in statistics.
In 1875 Galton realised that a normal mixture of normal distributions is a normal distribution.
In 1877 he introduced the idea of regression (or "reversion", as he originally called it). He asked friends to grow sweet peas to help ascertain how overall variance remains constant from generation to generation - he found that increased variance in some lines is exactly matched by regression towards the mean in others.
Although Karl Pearson was deeply devoted to Galton and produced a mathematical explanation of regression towards the mean, his dominance of statistics in the early 20th century actually held back the development of Galton’s ideas - particularly the development of analysis of variance. In fact Pearson’s great antagonist, R A Fisher, was the real inheritor of Galton’s mantle, achieving a coherent synthesis of Mendelian and statistical ideas.
Galton was initially interested in fmgerprints (or dermatoglyphs) for the purpose of identifying prisoners. He proved that they have three notable characteristics:
they do not change after formation;
they are not affected by the environment; and
the detailed patterns of ridges are extremely variable from one individual to another.
His classification, which is still the basis of all present work, built on Purkinje’s analysis of three main patterns -the whorl, the loop and the arch. The digits are categorised according to the pattern each pos sesses and they are then listed in a specific order - index, middle, ring of right hand, ditto left, thumb, little of right, ditto left - and the index fmger pattern further refined by reference to the direction of loops. This is the basis of primary retrieval and detailed analysis of minutiae of broadly matching loops can then take place. Galton was convinced of an here ditary tendency and contemplated an inquiry into the units of heredity using the minutiae of fingerprints. He noted strong resemblance between twins but didn’t distinguish monozy gotic and dizygotic pairs. He noted the greater maternal influence which is now known to be due to early intra-uterine effects. He was also interested in differences between populations but found they were negligible; with better data differences are now detected - eg Chinese have the highest incidence of whorls in the world.
Francis Galton’s long standing interest in meteorology resulted from his interest in maps which in turn came from his involvement with the Royal Geographical Society. Weather maps were his particular interest but when he began, around 1860 just 5 years after the meteorological office had been set up by Vice Admiral FitzRoy (Darwin’s companion on the Beagle), there were no conventional symbols. Galton chose December 1861 as a specimen month and collected all the data he could from many European countries. The first half of the month was dominated by a classic cyclonic system, the second half by a classic anticyclone, thus allowing Galton easily to recognise the distinction. He also observed significantly the simultaneity of wind changes over vast areas of observation and the importance of the Alps and the mountains of east Germany in breaking up wind patterns. Although his results, published at his own expense in "Meteorographica" in 1863, showed what could be done and secured wide acceptance for his symbols, meteorology remained descriptive for another half century. In 1991 we still lack information from many parts of the world, especially the Southern oceans where the position is actually worsening. We can understand even better than Galton the interconnectedness of weather systems and we have a better idea of their precariousness. We do not know if Galton ever turned from his concern for daily weather observation to thoughts of major climatic change, despite the debate in the 1840s and 1850s over glaciations. However, he would surely have been interested in the methodology by which we have now established the pattern of climatic change over the past 6000 years, which Sir Crispin went on to describe. Predicting future trends is more difficult but there is every possibility that the emission of greenhouse gases will lead to a major climatic disaster, and indeed there have been disasters in the past, although the problem is not so great as that of growth in the human population.
Professor Maynard Smith addressed three aspects of Galton’s contributions to the development of evolutionary theory. The first was his experiment to test the theory of pangenesis - the notion that hereditary particles are drawn from all parts of the body for incorporation into the germ cells, reflecting all the individual’s bodily characters including those acquired as a result of environmental influences. Galton’s method was to transfuse blood from one variety of rabbit to observe the effect on the offspring of the recipient. As with Weismann’s famous experiment involving the removal of mice’s tails, the results were totally negative; unlike Weismann, Galton had hoped for positive results. The second aspect considered was Galton’s emphasis on discontinuous variation. His discovery of regression towards the mean led him to seek explanations in terms of selection of "sports" - individuals showing discrete differences from the parental type. His rejection of selection acting on continuous variation as the main evolutionary mechanism held back the development of evolutionary biology for many years until R A Fisher and J B S Haldane clarified the concepts. Thirdly, Professor Maynard Smith observed how close Galton came to a discovery of Mendel’s laws as a result of his analysis of hybrids; if only he had extended his analysis beyond the Fl to the F2 generation he would probably have succeeded, although he was also held back by his suspicion of theoretical models. It is also possible that Galton’s lack of interest in natural history reduced his motivation to pursue these questions to the degree necessary to resolve them.
Galton’s interest derived primarily from his desire to improve the intellectual and physical quality of the population by encouraging some groups to produce more children than others. He therefore sought to establish the baseline from which improvement could be measured, to identify already well endowed groups and to find the laws underlying growth in different groups. He sought statistics through schools, apparently unaware of the data already collected by the Belgian pioneer of statistics Quetelet, in the factory surveys in 1833 & 1837 and in the immense survey of 10,000 boys and girls by Charles Roberts in 1872-73. Only public schools responded to Galton’s requests and he later collaborated with Roberts. In 1884 Galton set up a unit at an Exhibition in South Kensington where visitors paid 3d to pass through and have various physical and psychometric traits measured. 10,000 adults were measured and the results expressed in percentiles for the first time. Galton later sent out questionnaires seeking data on heights of parents and offspring and from the responses he demonstrated a regression of 0.67, exactly that which subsequent mathematical analysis would predict as a consequence of regression towards the mean. Thus Galton’s legacy was the demonstration that mathematical methods could be applied to anthropometry but he had no real successors in the field in the UK and he never fully pursued the lines of research which he began.
Dr Mascie-Taylor noted that Galton is credited with the first use of twins in science. However, it is not certain that he truly understood the concept of twinning and it is certain that subsequent writers have incorrectly attributed to him the "twin method" - the comparison of monozygotic and dizygotic twin pairs. In Europe roughly one third of twin conceptions are of opposite sexes, one third are of the same sex but fraternal and one third identical. The incidence of dizygotic twinning is related to maternal age and parity with a genetic component; there are no identifiable influences on monozygotic twinning. The dizygotic twinning rate is declining (fewer older conceptions?) whereas the monozygotic rate is rising (conception shortly after ceasing oral contraception?). Diagnosis of zygosity is critical in twin studies. One method is the fetal membrane method - monochorionic twins are always monozygotic but dichorionic twins could be either. Genetic methods are not completely reliable but DNA fingerprinting is nearly so and should in time supersede other methods. For discrete characters (eg schizophrenia) concordance methods are used; for continuous characters (eg height, weight, IQ) correlation methods. For schizophrenia there are substantial genetic factors but environmental factors still predominate. Some workers believe that identical twins have more similar environments than fraternal twins so some studies have involved mono and dizygotic twins reared apart. However, there are only 122 such pairs in four studies, one of which was conducted by Sir Cyril Burt. There is also doubt about the randomisation of environments where twins are reared by relatives. A recent study shows high genetic influence for IQ and other amazing similarities between twins reared apart. Although Galton started the idea, he did not follow it through and progress had to await a clearer concept of twinning.