Introduction After an early application of thermodynamics in metamorphic petrology V. M. Goldschmidt (1888 1947) took the lead in explaining the rules governing the distribution of the elements in minerals and rocks. From structures and compositions of REE compounds, he recognized lanthanide contraction and derived the basic knowledge of REE distribution. Based on the separately measured size of O2-, Goldschmidt calculated the ionic size of numerous elements from oxide structures, and explained their behavior in geochemical cycles. The partition of elements between coexisting meteorite minerals was used to derive their accumulation in the core or silicate shell of the earth. His use of greatly improved minorelement analysis displayed interelement relations.

Background

The name geochemistry, coined by the Swiss chemist Schonbein [1] , was almost 100 years old when Goldschmidt added a theoretical background to this discipline. Until his time, data on the majorelement composition of the accessible units of the earth's crust, the waters, and the atmosphere had been collected, having been compiled and clearly arranged by Frank W. Clarke in several editions of his Data of Geochemistry from 1908 to 1924 [2]. Goldschmidt sought the principles according to which the major and minor elements are accumulated in representative minerals and rock units, and consequently in the total earth as a cosmic product. Mason [3] has recently edited biographic information on the life and work of Goldschmidt as observed by his contemporaries. Here we consider the impact of his thinking on the two generations following him. Plate tectonics and planetology, with their farreaching transformation of geologic concepts in the latter third of our century, have their roots in basic discoveries in physics and chemistry made between 1900 and 1930.

Petrological Studies in Southern Norway

Victor Moritz Goldschmidt was born in 1888 in Zurich as the only child of a Germanspeaking family open to international relations and scientific thinking. In his boyhood, the family moved from Zürich to Amsterdam, then to Heidelberg, and finally to Kristiania, now Oslo. The move to Oslo in 1901 was the consequence of the decision of his father, Heinrich Jacob Goldschmidt, to succeed Peter Waage in his university chair in chemistry, specializing in physical chemistry. The young Goldschmidt entered the gymnasium, and, switching to the Norwegian language, graduated from this school in 1905, the year in which Norway separated from the union with Sweden. As a teenager, he was more interested in objects in nature than in adventures with other boys. From observations made during a holiday trip early in his university days, he published a small paper on the pyroluminescence of quartz in the periodicals of the Norwegian Academy of Science. Goldschmidt's Ph.D thesis in 1911 [4] Die Kontaktmetamorphose im Kristianiagebiet opened a completely new view of the petrological interpretation of progressive metamorphism by the application of the Gibbs phase rule to rocks. In his case, hornfelses were products of contact metamorphism in aureoles around the Permian syenite intrusions of the Oslo area. This new view was initiated both by his father's, having introduced him to physical chemistry and by Waldemar Brogger's advanced petrographic field work in that area.

Stimulated by his results on mineral reactions in silicous limestones and marly shales, progressed from the temperature of shallow magmatic intrusions, Goldschmidt turned his attention to metamorphic reactions at greater crustal depth. With scientific contact to Friedrich Becke and Paul von Groth during visits to Vienna and Munich, he signaled that his future interest would be located somewhere between petrology and crystallography. At the age of 26, he was appointed director of the Mineralogical Institute of the University of Oslo; in those days, a formal position such as this was required to be able to set up a scientific program. He undertook detailed field work in the Caledonides of southern Norway, where he recognized principles of mineraltomineral and mineraltomelt reactions in regional metamorphic rocks and magmatic intrusions. His observations, mainly from the area between Stavanger and Trondhjem, were reported in five monographs with the title Geologisch-petrographische Studien im Hochgebirge des südlichen Norwegens [5] published from 1912 to 1921. Goldschmidt was probably unaware that Barrow, as early as 1893, had already formulated principles on which particular index minerals characterize zones of progressive metamorphism in the Scottish Highlands [6]. His experiences in Norway were amalgamated with those of Pentti Eskola from Finland, who spent almost a year in Oslo (1919/20) and conceived the final version of his metamorphic facies concept. Goldschmidt's publication in 1922 [7] on the evolution of magmas through fractional crystallization based this important principle, that explains the generation of intermediary and felsic melts from mafic magmas, on the experimental results of Norman L. Bowen at the Geophysical Laboratory of Washington and on his own observations. We must bear in mind that the importance of partial melting of crustal rocks to form granitic magmas had not yet been recognized in those days; but experimental results in magmatic petrology were far ahead of those required to explain regional metamorphic products. Goldschmidt was apparently aware of the large gap in the calibration of conditions of metamorphic reactions at high pressures and temperatures, and probably for this reason, turned his interest in a new direction.

Early Geochemical Work at Oslo, Invention of Crystal Chemistry

The lack of raw materials for the Norwegian economy, that became noticeable towards the end of World War I, induced the Norwegian government to set up a Commission on Raw Materials associated with a related laboratory in 1917. Goldschmidt was appointed chairman of both the commission and the laboratory, in addition to his university duties. These duties could be successfully combined, as both institutions were housed in the Geological Museum. The major findings from the rawmaterials research were white titanium oxide pigments, the separation of apatite from the Fen carbonatite for use as a fertilizer, and the application of dunite as a refractory material in pyrometallurgy. The analytical instrumentation of the laboratory for raw materials was soon superior to comparable institutions abroad.

In 1912, the discovery by Max von Laue of Xray diffraction on crystal lattices had a major impact on Goldschmidt's future research. Von Laue's discovery allowed both calculation of the atomic arrangement and distances in crystals and analysis of the chemical composition of materials from diffracted Xray intensities and wavelengths. For Goldschmidt's research, both applications were indispensible. The first generation of instrumental and mathematical tools was available in the early 1920s in a few advanced laboratories. William Henry Bragg and his son William Lawrence had calculated the structures of some simple ionic crystals soon after the discovery of Xray diffraction. Almost at the same time, Karl M. Siegbahn pioneered Xray spectrum analysis, and A. Hadding adopted the same tool for mineral and rock analyses. Goldschmidt and his research associate Thomassen used their own Xray spectrograph to search for element No. 72 in zirconium minerals. George von Hevesy and D. Coster [8] were a little faster and discovered the element hafnium in zircon, calling it after the Latin name for Copenhagen, where they worked. The chemical similarity between zirconium and hafnium was recognized as an effect of the socalled lanthanide contraction, which Goldschmidt had concluded from his studies of lanthanide oxide structures and von Hevesy from molecular volumes, using specific gravity data.

The unsuccessful search for element No. 61 in another gap of the Periodic System of Elements stimulated Goldschmidt's broader interest in the group of the lanthanides, which, at least in those days, were unattractive for the majority of inorganic chemists. Beside the analysis of REE minerals, nearly 200 compounds (mainly oxides) of these and related elements were prepared for Xray structural investigations. For this comprehensive task he needed a group of very able coworkers, among whom T. Barth and W. Zachariasen later became wellknown scientists. The relationship between the sizes of ions and their association in minerals was first discovered during Goldschmidt's investigations on the minerals of the REE (rare earth elements) and of the crystal structures of their sesquioxides [9, 10]. He realized that the occurrence of these elements in REE minerals was a product of a relatively simple element fractionation and that elements with even numbers were more abundant than those with odd numbers, as Oddo and Harkins had foreseen. The lattice dimensions of the 6coordinated modification of the sesquioxides of Sc, Y, and Tb form a progressive series due to increasing ionic size at an increasing number of electron shells. From lanthanum to lutetium, however, a regular descrease in lattice dimensions in this modification of the sesquioxides can be observed. Goldschmidt suggested the name lanthanide contraction for this behavior, which was caused by the increasing number of electrons in lower shells being attracted by the positive charge of the atomic nucleus. Between the elements Dy and Ho, the lanthanide contraction exactly balances the increase in lattice dimensions from Y to La, thus explaining the close crystal structural relations between Y on the one hand and the elements Dy and Ho on the other, and the difficulties in separating these elements chemically.

It is the close chemical and crystal chemical coherency between all threevalent REE and their ability to form Ce4+ and Eu2+ separately from the regular threevalent state of the REE, indicating oxidizing or reducing natural conditions, which makes this group of elements so attractive for modern geochemistry. Goldschmidt [11] already realized the stability of Eu2+ and Ce4+ within the crust and at the earth's surface, respectively. He knew the close relation of Eu2+ to Sr2+ in minerals and observed a negative Eu anomaly in many REE minerals and a positive anomaly in potash feldspar. He anticipated that "the study of assemblages of rare earth elements is one of the most fascinating in the field of geochemistry". For the calculation of absolute values of ionic radii from crystal structures of oxides, Goldschmidt [12] needed to know the size of the oxygen ion, which in those days was derived from ion refractivity according to Wasastjerna [13]. A short time later, Pauling published a comparable set of ionic radii based on the electronic theory of the atomic structure of alkali halides.

In the 1920s, using exclusively the Xray analyses of REE in sesquioxides, monazites, xenotimes, epidotes, alkali feldspars, alkali pyroxenes, etc. Goldschmidt drew conclusions on the geochemical behavior of these elements, and could inspire his Japanese colleague, E. Minami, to preconcentrate the REE in abundant rocks chemically for Xray spectrography. These analyses on three composites of European Paleozoic and Japanese Paleozoic and Mesozoic shales were accomplished in the early 1930s during Goldschmidt's Gottingen period [15]. We present the results of Minami in the modern form as normalized to chondrites, to avoid the zigzag shape of the curves following the OddoHatkins rule (Fig. 2). For comparison, we have analyzed the same set of 36 European shales after chemical preconcentration with modern Xray fluorescence equipment [16]. Haskin and Haskin also analyzed this set by neutron activation [17]. Figure 2 demonstrates that, with the exception of La to Pr, Minami gave the correct order of magnitude of REE abundances in the shale composites and even the negative Eu anomaly characteristic for the majority of upper crustal rocks. His oddnumbered heavyREE data obviously deviate distinctly from the "true" values because of their low concentrations in the order of I ppm. Because of the analytical problem with the light REE (probably due to difficulties in the calibration of the photographic registration of Xray intensities), Goldschmidt reported a too small accumulation of the light in comparison with the heavy REE in the continental crust. In Goldschmidt's days, the definition of the continental crust as the layer above the seismic crust to mantle discontinuity given by Mohorovicic early this century was not generally accepted. He used the term upper lithosphere for the compartment which we now call upper continental crust.

The melting behavior in the system ironiron sulfidesilicate by Tammann [18] inspired Goldschmidt from 1922 onward, to estimate, the partition of certain elements between the major units of the earth's interior [19]. For this purpose he used the iron, troilite, and silicate proportion of common meteorites as products of different cosmic environments and as models of the major units of the earth's interior. To calculate the siderophilic, chalcophilic, and lithophilic behavior of the elements, he adopted a thermodynamic procedure and invented mineralmineral and mineralrock partition coefficients. Such partition coefficients are helpful in modern geochemistry to model the generation of certain magmas from assumed source rocks, etc. Goldschmidt emphasized the importance of analyzing common meteorites and meteoritic minerals for genetic conclusions, so that these materials were no longer exclusively on display in museums. It was not generally accepted in his time that the majority of meteorites are fragments of microplanets in the solar system.

In his book, edited posthumously by A. Muir [20], Goldschmidt presents a comparatively modern view of the composition of the solar system and stellar atmospheres, and states that the majority of parent bodies of meteorites were too small for chemical segregation by gravity. In his various studies, he overestimated the abundance of sulfur in the parental cosmic substances for the earth, which led him to assume the segregation of a sulfide melt from the earth's core as forming a separate layer above the core. Later seismic investigations could not support this assumption. Goldschmidt realized quite early in his geochemical studies that he needed reliable data on the abundance of several siderophilic elements in meteoritic and terrestrial materials. Such key elements were Au and the platinum group. His coworkers Lunde and Johnson [21] adopted a microassay method which had been previously used by Haber [22] for Au in seawater. The Pt metals and Au were extracted from a fluxed rock into a lead or silver melt, which later became the sample for Xray or optical spectrography. The analytical equipment in Oslo was not adequate to analyze many minor elements in abundant rock types, but suitable for crystal structure work. Goldschmidt's first Oslo period led to the discovery of the correlations between ionic size and atomic shell structure and of the rules for interelement replacement in minerals and artificial compounds, which were the basis for the field of crystal chemistry [23-26].

Basic Minor-Element Geochemical Investigations at Göttingen

At the end of this period, Goldschmidt was called to the chair in mineralogy and petrology at the university of Göttingen in Germany. He accepted and moved to Göttingen in 1929 because the German government provided enough money and space for the new equipment and staff required to solve many of the problems in minorelement geochemistry. Goldschmidt converted an old school building into an institute with modern analytical laboratories. The economic situation in postwar Germany was not good, but still inspiring enough for mathematics, physics, physical chemistry and other branches of science to be flourishing in Göttingen. Theoretical physics here was one of the centers where the laws of quantum mechanics, which explain the energies of atomic spectra, were established. Goldschmidt got R. Mannkopff from the physics department as a coworker. He constructed spectrographs and heat sources for sensitive spectrum analysis, the favored heat source being the carbon arc, highenergy flame and Xray fluorescence not having been invented at that time. At this period, geochemistry took the lead in the introduction of sensitive instruments for chemical analysis of large sample series.

The new spectrographic tools were soon available and Goldschmidt and his coworkers published a series of papers on the geochemical behavior of numerous minor elements in the continental crust. He demonstrated that Sc behaves geochemically differently from REE, and that Ga and Ge are related to Al and Si, respectively. The Pt-group metals and Au could be quantitatively analyzed by optical spectrography in the fireassay concentrates of iron meteorites, troilites, and some chromites. Compared with modern data by neutron activation [27], the concentrations of Ru, Rh, Pd, Ir, and Pt in iron meteorites reported by Goldschmidt and Peters [28] exceed the former data by factors of 1.3 to 2.5, and Os is too low. Problems of calibration, contamination, and Os volati lizati on du ring preconcentration might be responsible for the discrepancy. Our present laboratories have the advantage of cleaner grinding tools and chemicals, dustfree sample preparation, and the availability of international reference standards.

Goldschmidt [19], using the fact that certain clastic sediments contain the wellmixed weathering products of abundant upper crustal rocks, estimated the majorelement composition of the upper continental crust from tillites, which consist of the mechanical erosion materials collected by glaciers [29]. From the balance between average magmatic rocks and their weathering products, he attempted to calculate the amount of limestones in the crust and the mass of sodium delivered to the oceans. His result is not very precise because he neglected the large amount of metamorphic rocks, which was not even roughly known in those days, when knowledge of the lower continental crust and the earth's mantle was very slight. Goldschmidt [19, 20] assumed, from the composition of chondritic meteorites, from rock density, and from the abundance of eclogite xenoliths in certain kimberlites, that the earth's mantle consists of eclogite and peridotite. There were no data available on global heat flow to exclude a major proportion of eclogite in the mantle, which is distinctly higher in heatproducing elements than peridotite. Even the discovery of his coworker Ernst [30], that peridotite inclusions in basalts are xenoliths because of their fabrics, did not convince Goldschmidt of the major abundance of ultramafic rocks beneath the crust. In this crystal chemical investigation of Mg2GeO4 [31], he observed that this compound occurs in a spinel structure and not in the normal Mg2SiO4 modification. This observation led 40 years later to the discovery of a highpressure spineltype modification of Mg2SiO4 by Ringwood [32].
The very productive period at Göttingen, when Goldschmidt motivated a relatively large group of coworkers, lasted only 7 years. The Nazi party had meanwhile seized power in Germany and the persecution of Jewish citizens had started. Within these 7 years, he and his colleagues investigated the geochemical behavior of all rare and stable alkali and earth alkali elements, of B, Sc, Ga, Ge, As, Se, and continued the investigation of Y, REE, Ag, the Pt group, and Au. Such work was based on spectrum analyses of meteorites abundant terrestrial rocks and their rockforming minerals, as well as on compositional and structural analyses of minerals and oxides of the specific elements. From the time of his publication with Peters [33] on rare elements in coal, Goldschmidt realized the importance of plants in the cycle of certain minor elements. He observed that Ge with a crustal abundance of a few ppm can be accumulated in certain coal ashes to a level of 1% .

The reliability of the analytical basis of his investigations can be judged from a comparison of his upper crustal element concentrations [34] with those from a recent compilation [35]. Both sets are listed in Table 1. The ratios B/A of the respective data in this table show the agreement or discrepancy between the two sets. Ratios from 0.75 to 1.3 indicate a good correspondence, which occurs for the elements B, Na, Mg, K, Ca, Ga, Se, and Cs. Reasonably good agreement is to be observed for Sc, Ag, and Au; Be and Ge are in the correct order of magnitude. The largest discrepancy occurs in the case of Sn, Pt, Ba, Sr, Rb, and Li. The lack of agreement for Sn, Ba, Rb, and Li is mainly caused by a composite of granites in which one out of 14 samples represents not a common but a pneumatolytically altered granite. This sample contains 80 ppm Sn, 830 ppm Rb, and 160 ppm Li [20]. The error for Pt is due to sample contamination. Discrepancies of this order were to be expected at a time in which new methods were being invented, and additional data on the same elements from other laboratories were not yet available. Crustal averages are only a part of the wealth of the new data explaining the cosmochemical and geochemical properties of almost half all naturally occurring elements. One of the important recognitions in Goldschmidt's minorelement investigations is the deficiency in Li, Be, and B in terrestrial materials, which was inherited from stellar interiors, where these elements are fuels for nuclear processes [20].

Return to Oslo and the Tragic End of an Outstanding Scientist

Totally dedicated to his work, and living within the microcosmos of his fellow workers, Goldschmidt underestimated the dimension of pressure and danger arising from the ruling Nazi power. He first became aware of this danger from a poster addressed to him on his daily walk to the institute in the summer of 1935, whereupon he immediately arranged for his return to Oslo in September 1935. It was not easy for him to accommodate himself again to the smaller scientific community of the Oslo University. Spectrographic equipment for the rawmaterials laboratory could still be transferred to Oslo by his Göttingen colleagues, but Goldschmidt's scientific vitality had suffered, and his capacity for innovation was apparently diminished by the conditions of his forced move. In 1937 he published Part 9 of the Geochemische Verteilungsgesetze der Elemente, the most concise and comprehensive report on the geochemistry of minor elements to be written in the 1930s [34]. He himself called this his Ninth Symphony.

Conditions became worse with the German occupation of Norway in 1940. Goldschmidt, after being arrested and released twice, was finally rescued in December 1942 from Oslo harbor, where the ship Donau was ready to leave for Poland with the Norwegian Jews. The Norwegian resistance arranged his transport to Stockholm, from where he was flown to England. There he enjoyed the hospitality of his British colleagues in London, Aberdeen (Macauley Institute for Soil Research), and Harpenden (Rothamsted Experimental Station). Serious health problems however, compelled him to spend a good part of his time in hospitals and nursing homes. Due to his reduced working capacity, he concentrated on the problems of metal separation from coal ashes and silicosis, as well as drafting and compiling his experiences of 20 years crystal chemical and geochemical research for his book Geochemistry [20]. In this, he envisaged the future importance of geochronology and isotope geochemistry, but the tools and research capacity to initiate such investigations were not yet available. In June 1946, he returned a second time to Oslo, but as a very sick man. Friends who had come to the airport to meet him hardly recognized this shadow of a formerly robust figure. Goldschmidt died on March 20 th 1947 at the age of 59 from a cerebral hemorrhage after six cancer operations.

Victor Moritz Goldschmidt was an outstanding scientist, able, by his talent and his early professional education, to initiate new concepts and guide earth science in new directions. He was a fast, disciplined, and profound thinker of a generous, honest, but sensitive nature. His knowledge and memory were phenomenal, and it was apparently not easy to work with a senior scientist who had strong likes and dislikes, and who was very demanding in the accomplishment of projects. He expected that both he and his fellow workers should write papers "as concisely as a Scotsman sending a telegram". It is still enjoyable to read his papers and hear reports of his sarcastic humor. The last part of his life was a tragedy.

References

1. Schonbein, C. F.: Poggendorffs Ann. Phys. Chem. 45, 263 (1938)
2. Clarke, F.W.: U.S. Geol. Surv. Bull. 770, 1 (1924)
3. Mason, B.: Victor Moritz Goldschmidt: Father of Modern Geochemistry. Geochem. Soc. Spec. Publ. 4, San Antonio 1992
4. Goldschmidt, V. M.: Norske Vidensk. Selsk. Skrifter I Mat. Naturv. Kl. 1, 1 (1911)
5. Goldschmidt, V.M.: ibid. 18, 19 (19l2); 10 (1915); 2 (1916); 10 (|920)
6. Barrow, G.: Quart. J. Geol. Soc. London 49, 330 (1893)
7. Goldschmidt, V. M.: Norske Vidensk. Selsk. Skrifter I Mat. Naturv. Kl. 10, l (1922)
8. von Hevesy, G., Coster, D.: Z. Anorg. Chem. 150, 68 (1925)
9. Goldschmidt, V. M., Thomassen, L.: Norske Vidensk. Selsk. Skrifter I Mat. Naturv. Kl. 5, 1 (1924)
10. Goldschmidt, V.M., Barth, T., Lunde, G.: ibid. 7, 1 (1925)
11. Goldschmidt, V.M.: ibid. 4, 1 (1937)
12. Goldschmidt, V.M.: Naturwissenschaften 14, 477 (1926)
13. Wasastjerna, J. A.: Soc. Fenn. Comm. Phys. Math. 38, 1 (1923)
14. Pauling, L.: J. Am. Chem. Soc. 49, 705 (1927)
15. Minami, E.: Nachr. Ges. Wiss. Gottingen IV N.F. 1, 155 (1935)
16. Herrmann, A.G., Wedepohl, K.H.: Z. Anal. Chem. 225, 1 (1967)
17. Haskin, M.A., Haskin, L.A.: Science 154, 507 (1966)
18. Tammann, G.: Z. anorg. Chem. 110, 17 (1924)
19. Goldschmidt, V. M.: Norske Vidensk. Selsk. Skrifter I Mat. Naturv. Kl. 11, 3 (1922)
20. Goldschmidt, V.M.: Geochemistry (A. Muir, ed.). Oxford 1954
21. Lunde, G., Johnson, M.: Z. Anorg. Allg. Chem. 172, 167 (1928)
22. Haber, F.: Z. Angew. Chem. 40, 303 (1927)
23. Goldschmidt, V. M., Ulrich, F., Barth, T.: Norske Vidensk. Akad. Skrifter I Mat. Naturv. Kl. 5, 1 (1925)
24. Goldschmidt, V. M., Barth, T., Holmsen, P., Lunde, G., Zachariasen, W.: ibid.1, 1 (1926)
25. Goldschmidt, V. M., Barth, T., Lunde, G., Zachariasen, W.: ibid. 2, 1 (1926)
26. Goldschmidt, V.M.: ibid. 8, 1 (1926)
27. Crocket, J.H.: Geochim. Cosmochim. Acta 36, 517 (1972)
28. Goldschmidt, V. M., Peters, C.: Nachr. Ges. Wiss. Gottingen Math. Phys. Kl. IV, 26, 377 (1932)
29. Goldschmidt, V. M.: Fortschr. Miner. Krist. Petrogr. 17, 112 (1933)
30. Ernst T.: Nachr. Ges. Wiss. Gottingen Math. Phys. Kl. 13, 147 (l935)
31. Goldschmidt, V. M.: ibid. 1, 84 (1931)
32. Ringwood, A. E.: Composition and Petrology of the Earth's Mantle. New York 1975
33. Goldschmidt, V. M., Peters, C.: Nachr. Ges. Wiss. Gottingen Math. Phys. Kl. III 38, 371 (1933)
34. Goldschmidt, V. M.: Norske Vidensk. Akad. Skrifter I Mat. Naturv. Kl. 4, 1 (1937)
35. Wedepohl, K.H.: Geochim. Cosmochim. Acta 59, 1217 (1995)