MacTutor
With the death on March 18, 1989, of Sir Harold Jeffreys, FRS, science lost the last of the great founders of quantitative seismology and, indeed, of modern geophysics. Jeffreys' life of almost 98 years spanned the most heroic age of accumulation of knowledge on the physical properties of the Earth since Newton's. The central building stone was Jeffreys' The Earth. First published in 1926, it is now in its 6th edition. Among his major research contributions, which number over 300, on the nutations of the Earth-Moon system and constitutions of the planets, is the crucial argument that established the fluidity of the Earth's core.
Harold Jeffreys was born on April 22, 1891, in the schoolhouse at Fatfield, a mining village in County Durham, England. His father was the schoolmaster. He wrote that he was a keen naturalist at the age of 9 and, as a teenager, had an active interest in botany and photography (Two research papers on photographic matters appeared in the British Journal of Photography in 1910.) His undergraduate years were spent at Armstrong College, then part of the University of Durham and now the University of Newcastle-upon-Tyne, from where he graduated with distinction in mathematics. In 1910 he was admitted to St. John's College, Cambridge, which was to become his permanent academic home. After his Cambridge degree, he began research in dynamical astronomy and discussed it with Newall, Eddington, and Larmor. In 1912 he was awarded the Adams Memorial Prize by the College for an essay on Precession and Nutation, a subject to which he returned many times. His postgraduate years saw the award of the Smith's Prize (in 1915, bracketed with J. Proudman) and in 1927 the Adams Prize for 'The Constitution of the Earth'.
He was elected a Fellow of the Society on January 8, 1915 (and shrewdly compounded on election!) He served on the Council for three periods, 1919-28, 1929-31, and 1955-60: he was President from 1955 to 1957 and Vice-President from 1957 to 1959. Jeffreys was created Knight Bachelor in 1953. He shared the Vetlesen Prize of Columbia University (with F.A.V.Meinesz) in 1962. He received many high honours, including Gold Medals from the Royal Astronomical Society (1937) and the Royal Statistical Society (1963); the Royal Society awarded him its Royal Medal (1948) and its Copley Medal (1960). In the United States, the American Geophysical Union presented him with the Bowie Medal in 1952. He was awarded a number of honorary degrees and was elected to several foreign scientific societies, including the U.S. National Academy of Sciences.
Jeffreys was a professional assistant at the Meteorological Office from 1917 to 1922, and then became a Fellow and Lecturer at St. John's College, Cambridge until 1931, when he was appointed University Reader in Geophysics. In 1946 he was elected Plumian Professor of Astronomy and Experimental Philosophy. (Some have argued that the term 'Experimental Philosophy' best classifies Sir Harold's overall interests; 'Natural Philosophy' or 'Applied Mathematics' are also apt.) Jeffreys held a number of professional offices, apart from that of President of the Society, including President of the International Association of Seismology of the Earth's Interior (1964) and Honorary Director of the International Seismological Summary (1946-57). With H.H.Turner and R.Stoneley, he promoted geophysics and the Geophysical Supplement (later the Geophysical Journal) and his papers dominated its numbers for 35 years.
After World War II, he travelled widely and lectured on a wide range of topics to mathematics, statistics and geophysics students and research groups. In the United States he spent five months at Lamont Geological Observatory in 1964 and five months at Southern Methodist University in 1967. There are many warm anecdotes from students who knew him at those institutions.
Jeffreys wrote that the work of Sir George Darwin (a past Plumian Professor whom he never met) was his inspiration, but the empirical British tradition of Kelvin and Rayleigh is clear. His scientific modus was highly creative mathematical modelling, usually from classical mechanics and then crucial comparison with observations. He was an accomplished math-ematician, who wielded operational and approximation skills with rare flair and confidence. The comprehensive treatise Methods of Mathematical Physics (first published in 1946, and later in paperback edition) was written jointly with his wife Dr Bertha Swirles Jeffreys. The topic selection and mathematical style are strikingly original and it has been a well of information for generations of students.
A central geophysical work (1916 onwards) was his theory of the thermal history of the Earth and its internal strength; this was the basis of his global-contraction theory and robust view on the limited role of terrestrial convection - both of which were to clash with the plate-tectonics paradigm of the 1970s. Because his preferred time-dependent strain law did not permit large-scale mantle convection in recent geological time, he was an opponent of Wegener's theory of continental drift and remained unconvinced of its later manifestations, even though some of his distinguished Cambridge contemporaries were at the vanguard of the successful assault on his views. As a pioneer in planetology in 1923 he argued correctly, contrary to the then current general belief, that Uranus and Neptune have surface temperatures controlled mainly by solar radiation, about 120 °C or less. Further, from their densities, he argued that the constitutive materials must have molecular weights similar to methane and ammonia.
Both R.D.Oldham (in 1906) and B.Gutenberg (in 1914) gave seismological evidence that the major part of the core of the Earth was much less rigid than the mantle. (Jeffreys always preferred to speak of the Earth's 'shell'.) Jeffreys, using rough estimates of density and elastic parameters based on Wiechert's Earth model, calculated in 1915 that the resulting theoretical free period of the Eulerian nutation was too short. By 1926, after including the effect of compression, he had demonstrated that the seismological radius of the core, obtained by Gutenberg, was compatible with the observed free period if the core had negligible rigidity. In celebration of this accomplish-ment, surely of the first magnitude, Jeffreys sent a postcard to his closest colleague and friend, Robert Stoneley: "Die Weichertsche und Oldham-Gutenberg Kern identisch sind."
Later, with R.O. Vicente, he further investigated the nutations of the Earth's axis. This heavy computational work allowed for mechanical properties of the mantle and the fluid core and yielded fair agreement with the observed amplitude of the 19-yearly nutation. In the judgment of W.H. Munk and G.J.F. MacDonald in 1960, "the subject (of the rotation of the Earth) was reopened in the light of modern geophysical knowledge by Jeffreys. His contributions dominate the subject."
Jeffreys did seminal work on the reconstruction of the theory of the Earth's gravitational field and, in 1939, produced a plot of the deviation of the Earth's figure from hydrostatic shape, which showed gravity highs over the North Atlantic and the Pacific, and gravity lows over the Caribbean and Indian Ocean. Although, as in other studies, his then unfashionable results met with skepticism, they were eventually vindicated by observations of perturbations of artificial satellite orbits
Jeffreys was also a most distinguished statistician. Here his reputation rests on his two lucidly written books, Scientific Inference (first edition, 1931) and The Theory of Probability (first edition, 1939). In a review, I.H. Good judged the latter to be "of greater importance for the philosophy of science... than nearly all the books on probability written by professional philosophers lumped together." His earliest work, beginning with joint papers with Dorothy Wrinch, was motivated by a concern that he could discover no satisfactory methodology to guide the logical progression from observation to inference. He rooted his scientific method on Bayes's principle of inverse probability and suggested a useful prior distribution form ("The Jeffreys Prior") Influenced by Karl Pearson's Grammar of Science, his theory set out to explain from a series of axioms how scientific laws can acquire high probabilities at all. His views were not popular in statistical circles during the fertile period when R.A. Fisher, J. Neymann and others were developing the maximum-likelihood and significance-testing approach of the modern school. While paying compliments to Fisher's statistical contributions, even in 1974 Jeffreys would write that (Fisher) "never understood the theory of probability". Today, Bayesian methods are widespread in technological and risk studies. On a separate issue, Jeffreys' insistence that all estimates be accompanied by a statement of formal uncertainties is now much more widely adopted in geophysics than four decades ago, but the requirement has still been avoided in recent constructions of Earth models in seismology. Measures for discrepancies between distributions are now commonly called "Jeffreys (or J) divergences". A book on statistics in his honour was edited by A. Zellner in 1980.
Jeffreys's curiosity never ran out, and the whole planetary and geological world around him, with the perhaps significant exception of electromagnetic phenomena, was fair game for his mathematical skills: relativity, quantum theory, dynamics, the Moon's librations, tides, meteorology, hydrodynamics (particularly water-wave generation by wind), radiometric age dating, denudation, stream-sediment transportation, differential equations, tensors, epistemology, and psychology. On this canvas, worthy of Da Vinci, must not be missed published papers on "Digging in Rowing," "The Draining of a Vertical Plate," and "Turbulence in a River."
When Jeffreys began his seismological work, the special mathematical aspects were in their infancy. Key mathematical contributions had been made by Poisson, Green, Raleigh, Love and others, and the first important reliable observational work on the travel of seismic waves through the Earth was just being reported by John Milne, Oldham, Mohorovičić and others.
His 1926 inference that the main part of the core is fluid was based on early seismological measurements. Further progress depended on improved tables of travel times of seismic waves through the Earth. Such tables would not only allow the more precise location of remote earthquake sources, but also allow a mathematical inversion for estimating with better resolution the seismic velocity structure in the Earth's interior. Jeffreys set about correcting the generally used Milne-Zoppritz-Turner tables in 1931 and was soon joined by K.E. Bullen as a research student. By 1939, the research program, which produced the famous Jeffreys Bullen tables and a reliable velocity model for the Earth, was complete. Contemporaneously, Gutenberg, with help from C. Richter, pursued a similar program using rather different methods. Jeffreys excelled in statistical reduction of published times measured by others; Gutenberg in personally correlating wave recordings. Jeffreys remarked on the closeness of their final results, given the complete independence of their analyses.
The Jeffreys-Bullen travel times remain in standard use for most teleseismic locations today, even though the corresponding derived velocity structure has been modified in important ways. Attempts have been made to drastically refine the Jeffreys-Bullen tables, so far without great success, as differences in mean times at most distances are within the standard errors estimated by Jeffreys. At present, another international revision effort is underway, but validation is not yet assured.
After World War II, Jeffreys put much effort into developing travel times appropriate for various regions of the Earth, contrary to claims that he was now well aware of the non-radial structural variations in the Earth
Much fundamental knowledge emerged from his long series of travel-time papers, including smoothing techniques, the use of weights on long-tailed distributions ('the method of uniform reduction'), and the effectiveness of the chi-squared test. He showed extraordinary insight into avoiding the removal of significant discontinuities, such as the preservation of the 20° discontinuity in the upper mantle suggested by P. Byerly's study of the 1925 Montana earthquake.
The Earth structure that resulted from the Jeffreys-Bullen tables had several important discontinuities which Bullen used to classify the Earth into radial shells or 'regions'. An inner core, or Region G, proposed in 1936 by Dr. Inge Lehmann, was incorporated by Jeffreys after he rejected the earlier diffraction explanation of the early branch of PKP waves by a crucial test. He calculated, from Airy's theory of diffraction at a caustic, that diffracted amplitudes could not carry the observed energy into the shadow zone.
One major difference in the solutions of Gutenberg and Jeffreys was that Gutenberg's model had an anomalous gradual velocity increase at the boundary of the inner core, whereas Jeffreys' model had a small decrease, outside a sharp boundary. Resolution of this old dichotomy came only in the 1960s when alternative allowable models and crucial PKiKP reflections were found.
His geological interests made Jeffreys well aware that in the Earth seismic waves do not propagate through a homogeneous, perfectly elastic medium. Because rock discontinuities and strong velocity gradients lead to serious inadequacies in the application of ray theory, numerical approximations of waveforms are needed near cusps, caustics, and shadow zones. Further realistic deviations from perfect conditions involve the effects of viscosity and scattering. His seminal work on these problems made incisive use of the Airy integral and rediscovered the Green-Liouville solution of the harmonic equation with a variable coefficient. (The technique is now known as the JWKB approximation, after Jeffreys, G. Wentzel, H.A. Kramers, and L. Brill-ouin.) His application of Rayleigh's principle to compute approximate wave eigenfunctions was also a landmark contribution.
A major part of Jeffreys' legacy is that he set down in his papers all the essential ingredients of his argument. Basic observations, statistical methods, mathematical development, and the inferential process are all present. Here there is a contrast with some recent papers published on Earth structure. An example of his critical approach was his refusal to accept early inferred Earth models based on observed free oscillations of the Earth, available after the great 1960 Chilean earthquake. Discrepancies between observations and eigenspectra predicted by models led him to suggest in 1965 that the effects of damping on the oscillations must be considered. Follow-up work by others had confirmed this criticism by 1976 by showing that, for example, the viscosity effect changes the eigenfrequencies of the fundamental torsional and spheroidal terrestrial oscillations by about 0.5 and 0.25 percent (13 and 28 seconds), respectively
The clarity of Jeffreys' arguments was enhanced by his use of significance testing in making adjustments. He also dealt with the fundamental question of convergence involved in inverse processes. In 1936, he investigated whether a vicious infinite regression was involved in constructing new travel-time tables. The present seismological emphasis on mapping non-radial variations in the deep interior by tomographic methods is very much in Jeffreys' spirit of seeking anomalies. He would, however, insist that significance tests accompany claims of such structural variation
Jeffreys was often not easy to converse with, but his shyness was much mitigated by Lady Jeffreys's warm personality. Over the years, many found him helpful and unpretentious, with high professional loyalties. Although his lectures were low-key and sometimes difficult to follow, at best his writings were brilliant, with flashes of ironic humour. His results often struggled in an adverse climate, but he met these challenges with scientific integrity and courage. Jeffreys loved detective stories and his college life. He listed his chief recreation as cycle touring and "the sort of mountain climbing that does not need a rope." He happily participated in choral singing and botanical field studies. Jeffreys tackled problems with the widest variation in difficulty, from straightforward observational reduction to the hardest nonlinear dynamics. He worked prodigiously, but his reward must have been repeated deep creative satisfaction.
The collected Jeffreys papers have been published in six volumes by Gordon & Breach, London (1971–77).
Bruce A. Bolt
Harold Jeffreys was born on April 22, 1891, in the schoolhouse at Fatfield, a mining village in County Durham, England. His father was the schoolmaster. He wrote that he was a keen naturalist at the age of 9 and, as a teenager, had an active interest in botany and photography (Two research papers on photographic matters appeared in the British Journal of Photography in 1910.) His undergraduate years were spent at Armstrong College, then part of the University of Durham and now the University of Newcastle-upon-Tyne, from where he graduated with distinction in mathematics. In 1910 he was admitted to St. John's College, Cambridge, which was to become his permanent academic home. After his Cambridge degree, he began research in dynamical astronomy and discussed it with Newall, Eddington, and Larmor. In 1912 he was awarded the Adams Memorial Prize by the College for an essay on Precession and Nutation, a subject to which he returned many times. His postgraduate years saw the award of the Smith's Prize (in 1915, bracketed with J. Proudman) and in 1927 the Adams Prize for 'The Constitution of the Earth'.
He was elected a Fellow of the Society on January 8, 1915 (and shrewdly compounded on election!) He served on the Council for three periods, 1919-28, 1929-31, and 1955-60: he was President from 1955 to 1957 and Vice-President from 1957 to 1959. Jeffreys was created Knight Bachelor in 1953. He shared the Vetlesen Prize of Columbia University (with F.A.V.Meinesz) in 1962. He received many high honours, including Gold Medals from the Royal Astronomical Society (1937) and the Royal Statistical Society (1963); the Royal Society awarded him its Royal Medal (1948) and its Copley Medal (1960). In the United States, the American Geophysical Union presented him with the Bowie Medal in 1952. He was awarded a number of honorary degrees and was elected to several foreign scientific societies, including the U.S. National Academy of Sciences.
Jeffreys was a professional assistant at the Meteorological Office from 1917 to 1922, and then became a Fellow and Lecturer at St. John's College, Cambridge until 1931, when he was appointed University Reader in Geophysics. In 1946 he was elected Plumian Professor of Astronomy and Experimental Philosophy. (Some have argued that the term 'Experimental Philosophy' best classifies Sir Harold's overall interests; 'Natural Philosophy' or 'Applied Mathematics' are also apt.) Jeffreys held a number of professional offices, apart from that of President of the Society, including President of the International Association of Seismology of the Earth's Interior (1964) and Honorary Director of the International Seismological Summary (1946-57). With H.H.Turner and R.Stoneley, he promoted geophysics and the Geophysical Supplement (later the Geophysical Journal) and his papers dominated its numbers for 35 years.
After World War II, he travelled widely and lectured on a wide range of topics to mathematics, statistics and geophysics students and research groups. In the United States he spent five months at Lamont Geological Observatory in 1964 and five months at Southern Methodist University in 1967. There are many warm anecdotes from students who knew him at those institutions.
Jeffreys wrote that the work of Sir George Darwin (a past Plumian Professor whom he never met) was his inspiration, but the empirical British tradition of Kelvin and Rayleigh is clear. His scientific modus was highly creative mathematical modelling, usually from classical mechanics and then crucial comparison with observations. He was an accomplished math-ematician, who wielded operational and approximation skills with rare flair and confidence. The comprehensive treatise Methods of Mathematical Physics (first published in 1946, and later in paperback edition) was written jointly with his wife Dr Bertha Swirles Jeffreys. The topic selection and mathematical style are strikingly original and it has been a well of information for generations of students.
A central geophysical work (1916 onwards) was his theory of the thermal history of the Earth and its internal strength; this was the basis of his global-contraction theory and robust view on the limited role of terrestrial convection - both of which were to clash with the plate-tectonics paradigm of the 1970s. Because his preferred time-dependent strain law did not permit large-scale mantle convection in recent geological time, he was an opponent of Wegener's theory of continental drift and remained unconvinced of its later manifestations, even though some of his distinguished Cambridge contemporaries were at the vanguard of the successful assault on his views. As a pioneer in planetology in 1923 he argued correctly, contrary to the then current general belief, that Uranus and Neptune have surface temperatures controlled mainly by solar radiation, about 120 °C or less. Further, from their densities, he argued that the constitutive materials must have molecular weights similar to methane and ammonia.
Both R.D.Oldham (in 1906) and B.Gutenberg (in 1914) gave seismological evidence that the major part of the core of the Earth was much less rigid than the mantle. (Jeffreys always preferred to speak of the Earth's 'shell'.) Jeffreys, using rough estimates of density and elastic parameters based on Wiechert's Earth model, calculated in 1915 that the resulting theoretical free period of the Eulerian nutation was too short. By 1926, after including the effect of compression, he had demonstrated that the seismological radius of the core, obtained by Gutenberg, was compatible with the observed free period if the core had negligible rigidity. In celebration of this accomplish-ment, surely of the first magnitude, Jeffreys sent a postcard to his closest colleague and friend, Robert Stoneley: "Die Weichertsche und Oldham-Gutenberg Kern identisch sind."
Later, with R.O. Vicente, he further investigated the nutations of the Earth's axis. This heavy computational work allowed for mechanical properties of the mantle and the fluid core and yielded fair agreement with the observed amplitude of the 19-yearly nutation. In the judgment of W.H. Munk and G.J.F. MacDonald in 1960, "the subject (of the rotation of the Earth) was reopened in the light of modern geophysical knowledge by Jeffreys. His contributions dominate the subject."
Jeffreys did seminal work on the reconstruction of the theory of the Earth's gravitational field and, in 1939, produced a plot of the deviation of the Earth's figure from hydrostatic shape, which showed gravity highs over the North Atlantic and the Pacific, and gravity lows over the Caribbean and Indian Ocean. Although, as in other studies, his then unfashionable results met with skepticism, they were eventually vindicated by observations of perturbations of artificial satellite orbits
Jeffreys was also a most distinguished statistician. Here his reputation rests on his two lucidly written books, Scientific Inference (first edition, 1931) and The Theory of Probability (first edition, 1939). In a review, I.H. Good judged the latter to be "of greater importance for the philosophy of science... than nearly all the books on probability written by professional philosophers lumped together." His earliest work, beginning with joint papers with Dorothy Wrinch, was motivated by a concern that he could discover no satisfactory methodology to guide the logical progression from observation to inference. He rooted his scientific method on Bayes's principle of inverse probability and suggested a useful prior distribution form ("The Jeffreys Prior") Influenced by Karl Pearson's Grammar of Science, his theory set out to explain from a series of axioms how scientific laws can acquire high probabilities at all. His views were not popular in statistical circles during the fertile period when R.A. Fisher, J. Neymann and others were developing the maximum-likelihood and significance-testing approach of the modern school. While paying compliments to Fisher's statistical contributions, even in 1974 Jeffreys would write that (Fisher) "never understood the theory of probability". Today, Bayesian methods are widespread in technological and risk studies. On a separate issue, Jeffreys' insistence that all estimates be accompanied by a statement of formal uncertainties is now much more widely adopted in geophysics than four decades ago, but the requirement has still been avoided in recent constructions of Earth models in seismology. Measures for discrepancies between distributions are now commonly called "Jeffreys (or J) divergences". A book on statistics in his honour was edited by A. Zellner in 1980.
Jeffreys's curiosity never ran out, and the whole planetary and geological world around him, with the perhaps significant exception of electromagnetic phenomena, was fair game for his mathematical skills: relativity, quantum theory, dynamics, the Moon's librations, tides, meteorology, hydrodynamics (particularly water-wave generation by wind), radiometric age dating, denudation, stream-sediment transportation, differential equations, tensors, epistemology, and psychology. On this canvas, worthy of Da Vinci, must not be missed published papers on "Digging in Rowing," "The Draining of a Vertical Plate," and "Turbulence in a River."
When Jeffreys began his seismological work, the special mathematical aspects were in their infancy. Key mathematical contributions had been made by Poisson, Green, Raleigh, Love and others, and the first important reliable observational work on the travel of seismic waves through the Earth was just being reported by John Milne, Oldham, Mohorovičić and others.
His 1926 inference that the main part of the core is fluid was based on early seismological measurements. Further progress depended on improved tables of travel times of seismic waves through the Earth. Such tables would not only allow the more precise location of remote earthquake sources, but also allow a mathematical inversion for estimating with better resolution the seismic velocity structure in the Earth's interior. Jeffreys set about correcting the generally used Milne-Zoppritz-Turner tables in 1931 and was soon joined by K.E. Bullen as a research student. By 1939, the research program, which produced the famous Jeffreys Bullen tables and a reliable velocity model for the Earth, was complete. Contemporaneously, Gutenberg, with help from C. Richter, pursued a similar program using rather different methods. Jeffreys excelled in statistical reduction of published times measured by others; Gutenberg in personally correlating wave recordings. Jeffreys remarked on the closeness of their final results, given the complete independence of their analyses.
The Jeffreys-Bullen travel times remain in standard use for most teleseismic locations today, even though the corresponding derived velocity structure has been modified in important ways. Attempts have been made to drastically refine the Jeffreys-Bullen tables, so far without great success, as differences in mean times at most distances are within the standard errors estimated by Jeffreys. At present, another international revision effort is underway, but validation is not yet assured.
After World War II, Jeffreys put much effort into developing travel times appropriate for various regions of the Earth, contrary to claims that he was now well aware of the non-radial structural variations in the Earth
Much fundamental knowledge emerged from his long series of travel-time papers, including smoothing techniques, the use of weights on long-tailed distributions ('the method of uniform reduction'), and the effectiveness of the chi-squared test. He showed extraordinary insight into avoiding the removal of significant discontinuities, such as the preservation of the 20° discontinuity in the upper mantle suggested by P. Byerly's study of the 1925 Montana earthquake.
The Earth structure that resulted from the Jeffreys-Bullen tables had several important discontinuities which Bullen used to classify the Earth into radial shells or 'regions'. An inner core, or Region G, proposed in 1936 by Dr. Inge Lehmann, was incorporated by Jeffreys after he rejected the earlier diffraction explanation of the early branch of PKP waves by a crucial test. He calculated, from Airy's theory of diffraction at a caustic, that diffracted amplitudes could not carry the observed energy into the shadow zone.
One major difference in the solutions of Gutenberg and Jeffreys was that Gutenberg's model had an anomalous gradual velocity increase at the boundary of the inner core, whereas Jeffreys' model had a small decrease, outside a sharp boundary. Resolution of this old dichotomy came only in the 1960s when alternative allowable models and crucial PKiKP reflections were found.
His geological interests made Jeffreys well aware that in the Earth seismic waves do not propagate through a homogeneous, perfectly elastic medium. Because rock discontinuities and strong velocity gradients lead to serious inadequacies in the application of ray theory, numerical approximations of waveforms are needed near cusps, caustics, and shadow zones. Further realistic deviations from perfect conditions involve the effects of viscosity and scattering. His seminal work on these problems made incisive use of the Airy integral and rediscovered the Green-Liouville solution of the harmonic equation with a variable coefficient. (The technique is now known as the JWKB approximation, after Jeffreys, G. Wentzel, H.A. Kramers, and L. Brill-ouin.) His application of Rayleigh's principle to compute approximate wave eigenfunctions was also a landmark contribution.
A major part of Jeffreys' legacy is that he set down in his papers all the essential ingredients of his argument. Basic observations, statistical methods, mathematical development, and the inferential process are all present. Here there is a contrast with some recent papers published on Earth structure. An example of his critical approach was his refusal to accept early inferred Earth models based on observed free oscillations of the Earth, available after the great 1960 Chilean earthquake. Discrepancies between observations and eigenspectra predicted by models led him to suggest in 1965 that the effects of damping on the oscillations must be considered. Follow-up work by others had confirmed this criticism by 1976 by showing that, for example, the viscosity effect changes the eigenfrequencies of the fundamental torsional and spheroidal terrestrial oscillations by about 0.5 and 0.25 percent (13 and 28 seconds), respectively
The clarity of Jeffreys' arguments was enhanced by his use of significance testing in making adjustments. He also dealt with the fundamental question of convergence involved in inverse processes. In 1936, he investigated whether a vicious infinite regression was involved in constructing new travel-time tables. The present seismological emphasis on mapping non-radial variations in the deep interior by tomographic methods is very much in Jeffreys' spirit of seeking anomalies. He would, however, insist that significance tests accompany claims of such structural variation
Jeffreys was often not easy to converse with, but his shyness was much mitigated by Lady Jeffreys's warm personality. Over the years, many found him helpful and unpretentious, with high professional loyalties. Although his lectures were low-key and sometimes difficult to follow, at best his writings were brilliant, with flashes of ironic humour. His results often struggled in an adverse climate, but he met these challenges with scientific integrity and courage. Jeffreys loved detective stories and his college life. He listed his chief recreation as cycle touring and "the sort of mountain climbing that does not need a rope." He happily participated in choral singing and botanical field studies. Jeffreys tackled problems with the widest variation in difficulty, from straightforward observational reduction to the hardest nonlinear dynamics. He worked prodigiously, but his reward must have been repeated deep creative satisfaction.
The collected Jeffreys papers have been published in six volumes by Gordon & Breach, London (1971–77).
Bruce A. Bolt
Harold Jeffreys's obituary appeared in Journal of the Royal Astronomical Society 31:2 (1990), 267-271.