The third, and by far the most efficient and most widely-used machine, is known as the Wetherill electro-magnetic separator. It depends on the principle, first applied by the American engineer Wetherill, that not only the iron minerals, but a large number of other minerals may be attracted if the magnetic field be sufficiently strong. In all types of this machine used in cleaning monazite concentrates, four magnetic fields of increasing intensity are traversed by the sand; the first removes magnetite, ilmenite, and the larger fragments of garnet; the second removes all the remaining garnet and ilmenite; the third removes the coarser, and the fourth the finer monazite, tailings of zircon, rutile, and silica passing on. Careful adjustment of the magnetic fields will readily give a 97-99 per cent. monazite.

Two types of this machine are in common use. In the first the magnetic fields are obtained by four successive electro-magnets, arranged so that a broad horizontally-moving belt passes between the poles of each in succession. The upper poles are ground down to a fine edge perpendicular to the direction of the belt, to secure a more powerful field. Just beneath these edges, and just above the broad belt are four rapidly driven horizontal belts moving at right angles to the first or main belt; these carry off and deposit in separate bins the minerals attracted by their respective magnets. This type is known as the Rowand separator.

In the second type four horizontal belts are arranged in the form of descending steps, as shown in the diagram (Fig. 2). The magnets are placed at the end of each belt, and within it. The attracted mineral is held to its own belt, whilst the remainder drops on to the next; the attracted mineral falls into a bin as soon as its belt carries it out of the magnetic field. The sand to be cleaned is fed on to the first belt by means of a hopper.

The almost pure monazite so obtained is now treated chemically for its thorium. The processes proposed and in use are described in Chapter XVIII.

As already stated, the extraction of monazite in the United States has practically ceased; but the processes outlined above, which were first brought into use in the Carolinas, have been adopted for the treatment of the Brazilian sands.

(b) The Idaho Deposits.

—Monazite was first observed in placer-gold deposits in the vicinity of Boise city near the Snake river. This deposit was a gold-bearing sand derived from granite. Later the gold-bearing sands of Oregon were also found to contain monazite; these sands are rich in zircon, and contain platinum and allied metals as well as gold. The sands of the Pacific slope are the so-called black sands, derived from hornblende, and augite-granites, usually porphyritic, which are much weathered at the surface. The soil is loose and is largely composed of granite fragments; the rain and streams constantly bring it down to the valleys, and continually renew the deposits. The concentrates obtained by washing are rich in well-crystallised zircon, with titanite and garnet.

In 1906 a company was formed to extract monazite from the black sand residues left after the extraction of gold. By 1909 they had erected plant and commenced operations at Centerville, and proposed to work the poorer auriferous sands for gold during the monazite washing. This, it was expected, could be done by washing the sands in boxes lined with amalgamated copper plates, which would retain the gold. Considerable amounts of monazite had already been extracted from the tailings when a disastrous fire put a stop to the operations in 1910.

Since then the production of monazite in the United States has practically ceased.

The Brazilian Deposits

Brazil first became a serious competitor in the world’s market with the United States, for the supply of monazite, in 1895. The greater percentage of thorium, the more even quality of the sands, and above all the occurrence on the sea-coast, rendered the Brazilian monazite cheaper from the beginning, so that it soon ousted the Carolina sand, and since 1910 has supplied the whole demand. The deposits at present worked lie along the coasts of Bahia, Minas Geraes and Espirito Santo, and whilst they are very rich in monazite, there is the disadvantage that their position and extent, and so also the possibility of working them, depend very largely on the variations in the tides, etc. The largest of these deposits is on the shores of a bay near the island of Alcobaca, on the southern coast of Bahia.

Monazite also occurs to a considerable extent in the diamond sands and gold-bearing sands of many of the interior provinces. In Minas Geraes it has long been known to occur at the celebrated mining centres of Diamantina and Ouro Preto, where xenotime and other rare earth minerals are also found; it is also known at various localities in the surrounding mining provinces of São Paulo, Goyaz, and Matto Grosso. More recently, extensive inland deposits have been found by Freise, in the province of Espirito Santo.[122] In the plateau-basin of the Muriahé and Pomba rivers he found a sand known locally as ‘catalco’ which carries an average of 2·1 per cent. of monazite and a gold-content of 1·75 grams per ton. In the Aymoré’s mountains he found monazite, both massive and granular, in pegmatite veins in granite; analysis showed a thoria content of 9·23 per cent., which is very high. These deposits would form a very valuable and extensive source of thoria, if the difficulties of transport could be overcome.

At present, as stated above, only the beach deposits are worked. The Brazilian Government has laid a very heavy tax on all monazite exported; it is stated[123] that the German Thorium Syndicate pays 50 per cent. of its profits in royalties to the Government. In spite of this, the high quality of the sand and the low cost of transport have enabled this combine to lower the price of thorium nitrate to a point at which the Carolina sands cannot be worked, and it appears probable that the world’s markets will be supplied for some time, at least, entirely from Brazil. The methods employed in working the sand are similar to those already described.

In the last few years monazite deposits have been found in various places, notably in Australia, India, and Ceylon. In the latter locality it occurs sparingly in the gem-gravels, in association with the much more valuable thorianite and thorite, but the supply is uncertain, and the minerals cannot be worked regularly. In Australia it occurs in Victoria and in Queensland. In Victoria the deposits are poor in monazite—about 0·025 per cent.—so that working is not profitable. In Queensland it occurs in beach sands on the southern coast, with gold, platinum, and cassiterite; there seems to be no reason why these deposits should not be profitably worked when sufficient labour is forthcoming. It also occurs in North Queensland, on the Walsh and Tinaroo mineral fields; here it is found massive and granular in veins in granite, associated with wolframite, molybdenite, and cassiterite.

Quite recently, deposits of considerable extent have been found near Travancore, India.[124] These sands contain about 46 per cent. of the mineral, which is itself very rich in thoria, containing about 10 per cent. of the oxide; the unconcentrated sand is therefore as valuable as a source of thorium nitrate as the ordinary Brazilian concentrates, which average 4 per cent. or less of the oxide.

Monazite has also been observed in the tin-bearing sands of Embabaan, Swaziland, South Africa, and in the province of Ottawa, Canada.

CHAPTER VIII
RADIOACTIVITY OF THE MINERALS

In the present chapter no attempt will be made to give a complete account of all the phenomena of radioactivity which have been observed in the mineral world. There are, however, a few problems of the highest scientific interest which centre about the rare earth minerals, and mention of these can hardly be avoided in a work which professes to give a general account of the rare earth group. It is obvious that a detailed treatment cannot be given without entering into phenomena which would be quite beyond the range of the present work, and an excuse is hardly needed, therefore, for the fragmentary and abbreviated account which follows. The reader’s acquaintance with the general phenomena of radioactivity is of necessity assumed.

Radioactivity (the spontaneous emission of special radiations) was first observed by Becquerel, in 1896, in the case of potassium uranyl sulphate, and was soon found to be common to all uranium compounds, and to the metal itself. Mme. Curie showed that whilst in uranium salts the degree of activity varies directly with the percentage of uranium, in minerals containing the element the same rule does not hold. The observation that pitchblende is considerably more active than the uranium it contains led to the discovery of polonium[125] and radium in 1898. Exactly analogous phenomena were shown to hold for thorium salts and thorium-containing minerals by Mme. Curie and Schmidt in 1898, and in 1905 Hahn separated Radio-thorium from thorianite. In 1899 Debierne discovered that the rare earths precipitated from the solution obtained on treatment of pitchblende are associated with another extremely active body, which he named Actinium; Giesel found that in the separation of the rare earths this remains with lanthanum. In 1903 Ramsay and Soddy experimentally confirmed the prediction of Rutherford and Soddy, that radium would be found to produce helium continuously. The discovery of these remarkable phenomena has modified many fundamental physical conceptions, and has opened up a new field of scientific enquiry, which is being developed with unexampled rapidity.

It has been mentioned, in the accounts of the rare earth minerals given above, that almost all these minerals are radioactive, i.e. have the property of emitting specific radiations. Moreover, radioactivity, to any considerable extent at least, is, with a few important exceptions, confined to the minerals which have been already described. It has been shown by many investigators, chief amongst whom are Strutt and Boltwood, that the activity is usually due to the presence of uranium or thorium, or both.[126]

After the discovery of helium in Cleveite (a variety of pitchblende, vide p. 13) in 1895 by Ramsay, a large number of minerals were examined for this gas, and it was found that almost all the rare earth minerals contain helium. The fact that these minerals are also for the most part radioactive, naturally suggested some relation between the activity and the presence of helium, and led directly to the discovery that radium is continuously producing helium; and it became apparent that helium has been accumulating in these minerals since their formation, by the decay of radioactive elements. The question of the origin of helium in minerals will be touched on again.

In 1904 Boltwood advanced the theory that radium is produced by the degradation of uranium, the parent-element having, however, a much greater half-life period. If uranium continuously produces radium, whilst the latter decays much more rapidly than the former, it must follow that in minerals containing uranium a state of equilibrium is reached between uranium and radium, and the ratio of these two in all minerals should therefore be constant, and independent of the geological age. Boltwood examined a number of the minerals of which descriptions have been given in the preceding chapters, and found the ratio to be surprisingly constant.[127] Strutt also examined a large number of minerals,[128] and whilst on the whole his results seemed to support the theory, his values for the ratio were by no means so constant as those of Boltwood. Strutt included in his examination the interesting radium-containing mineral observed by Danne at Issy l’Evêque.[129] This was a pyromorphite (lead chlorophosphate) containing neither uranium nor thorium. Danne suggested that the radium was not an original constituent, but had been introduced by the action of percolating waters. This view was confirmed by McCoy and Ross,[130] who found that the activity was entirely confined to the surface layer.

Mlle. Gleditsch has also examined the question of the uranium-radium ratio in minerals. Her earlier work[131] gave ratios which, whilst constant for each mineral species, varied in much the same manner as Strutt’s for different species, and afforded very little support to Boltwood’s theory. Her more recent results,[132] however, are much more closely in accord with the theory, which has been still further strengthened by the work of Pirret and Soddy[133] and of Marckwald and Russell.[134] It may now be regarded as firmly established that radium is in the line of direct descent from uranium.

Boltwood had assumed that the helium in radioactive minerals is produced from the uranium, during its disintegration. Strutt, however, disputed this; his experiments showed that very little helium is found even in the richest radium-uranium minerals unless thorium is also present. Thus pitchblende contains a very high percentage of uranium, but relatively little helium (there is usually a considerable thorium percentage here too, so that nothing conclusive can be deduced from this). Adams[135] found that carnotite, a mineral very rich in uranium, but containing no thorium, contains no helium at all; he explained its absence by the very loose texture and permeability of the mineral, which would allow the gas to escape. Strutt concluded that whilst helium is undoubtedly produced by disintegration in the uranium series, in minerals it is produced more by thorium or, as more recent work indicates, by radio-thorium, than by uranium.

The question of the origin of helium in minerals is, however, not definitely settled, for several anomalous cases are known. Thus the yttria silicate, Thalénite (q.v.), contains quantities of helium, but no uranium or thorium is given in the analyses. Similarly, Risörite contains a relatively large quantity of helium, but only traces of uranium and thorium. In the last mineral, the active constituent is precipitated with the lead, so that no radio-thorium appears to be present. Further, Thomsen analysed a fluorspar from Ivitgut in Greenland which he found to contain 27 c.c. of helium per kilogram. This specimen contains no uranium, but gives off the thorium emanation in quantities which suggest the presence of radio-thorium; moderate quantities of thorium are also present. Since the α particle has been definitely identified as a positively charged helium atom, it appears certain that disintegration in all three series (uranium, actinium, and thorium series) produces helium, and a mineral containing a member of any of these series (which gives α rays or α ray-giving products) would also contain helium.

Even so, there is a case in which the helium content is anomalous, if not altogether beyond explanation at the present stage. In examining a large number of minerals for helium, Strutt[136] found that some samples of beryl, a beryllium aluminium silicate, contain a relatively very large amount of helium, but only traces of thorium, and was altogether inactive. The absence of any active constituent renders untenable the ordinary explanations of the presence of such a surprising quantity of helium. Boltwood has put forward a suggestion which in the present state of our knowledge must be regarded as a provisional explanation. He conceives that in the concentration of beryllium from the parent magma, it may have become associated with some short-lived intermediate radioactive element, which had been altogether separated from its long-lived parent element in the process of concentration; this intermediate element, having collected in the crystallised beryl, decayed completely in the course of the great period which must have elapsed, leaving the helium to which it had given rise during its disintegration enclosed in the mineral. It is difficult to see how two substances which must be so intimately connected as a parent-element and its product could be completely separated in the process of cooling of a magma; but since so little is known of the process of crystallisation of minerals, the suggestion can hardly be rejected on geological grounds. In any case, we have here only one strongly marked exception to the very definite rule that in all cases in which helium occurs in minerals, it is accompanied by and undoubtedly produced from, a radioactive element or elements; and in the majority of cases, the helium in minerals is produced by disintegration of uranium or thorium and their products.

Strutt found that traces of helium are universal in the mineral world. His method of determining helium was approximate only. He obtained the gas content by heating the powdered mineral—a method which, as Wood has shown,[137] will only give all the gas when very high temperatures (up to 1000°C.) are employed. The gases were freed from oxygen and hydrogen by passing over a heated, partially oxidised, copper spiral, and from carbon dioxide by means of potash. Nitrogen was removed by sparking with excess of oxygen and shaking over potash; the excess of oxygen was removed by melted phosphorus. The inert gases so obtained were freed from all impurities by the use of the liquid alloy of sodium and potassium for the electrodes of the spectrum tube in which the gases were examined spectroscopically.[138] Argon, if present—it seems to be a universal constituent of igneous rocks, into which it may have been absorbed from the air—was removed by charcoal at a temperature of -80°C. The helium so left was examined spectroscopically, and measured in a MacLeod gauge.

As stated, helium was found in traces in nearly all minerals, and its presence is to be attributed to traces of radium, which also appears universal. In minerals containing uranium or thorium, or rare earths (the latter are almost always accompanied by uranium and thorium), helium is found to a much greater extent, and Ramsay considers it possible that some fraction of the helium content may arise from the rare earth metals. There is, however, no positive evidence to support the conjecture. He found that the helium ratio, i.e. the volume of helium per gram of uranous oxide, UO₂, varies with the amount of thoria present; but where the latter is absent the variations are much less marked. If helium were produced in a mineral from uranium alone, and none escaped, it is obvious that the helium ratio would depend only on the age of the mineral. For minerals of about the same age, and containing no thorium, the helium ratio would be roughly constant, if no disturbing factor required consideration.

In 1905 Strutt pointed out that in all the minerals he had examined, thorium was never present unless accompanied by uranium and radium, whilst uranium and radium often occurred without thorium. He suggested that the present atomic weight of thorium, 232·5, was too low, and that it was really the parent of uranium (at. weight 238·5); he further supposed that the next permanent member in the line of descent was one of the cerium metals. These suggestions have been negatived by later work of Boltwood and Holmes. The former pointed out[139] that it was far more likely that thorium is a disintegration product of uranium of considerably longer life. On the whole, however, there is very little positive evidence to connect thorium with uranium.

In the same year Boltwood (loc. cit.) drew attention to the persistent appearance of traces of lead, bismuth, barium, etc., in the radioactive minerals, and also pointed out that the variations of the ratio of helium to uranium in pitchblende might be used to determine the age of the mineral. In 1907 he suggested[140] that lead was the final product of the degradation of uranium, from which it follows that the ratio of uranium to lead should be constant for minerals of the same age (since, lead decays, if at all, at an infinitely slower rate than uranium). He collected all the available analyses, and classified the minerals dealt with into six groups according to the value of the ratio. The order given by the ratio was declared to be in accordance with the order of age as given by geological evidence.

Holmes[141] has further extended this work. He examined a number of rare earth and allied minerals from the Christiania district, which Brögger considers to be of approximately Lower Devonian age, and found the ratio of lead to uranium to approximate quite closely, for almost all the minerals examined, to 0·045. Representing the change in the usual way as

and using the data calculated by Rutherford and others for the rates of decay, he gives the age of Lower Devonian strata as about 370 million years. This figure is about twice as great as that deduced by palæontologists from the flora and fauna, and greater still than the times based on physical data, e.g. rates of cooling, precession and nutation, etc. His figures for pre-Cambrian rocks, based on the same ratio, range between 1000 and 1640 million years, the later being deduced from a thorianite from the Archæan rocks of Ceylon. Strutt’s figure for Archæan rocks is about 700 million years; this was derived from work on the helium ratio, which must now be considered.[142]

In 1898 Travers[143] had examined the effect of heat on cleveite and fergusonite, and found that about half the total helium, together with hydrogen, is given off at a bright red heat. He considered it likely that the helium was combined with a metal (though he recognised no distinction between occlusion and combination) and remarked: ‘The results of such experiments cannot therefore serve as a basis for speculation as to the origin or history of the substances in question.’ The chemical inactivity of helium, however, as well as the experiments of Moss and Gray, who showed that helium was evolved on grinding the materials,[144] indicate that the gas is mechanically bound only. This, however, introduces the difficulty, if an attempt be made to use the helium-uranium ratio to calculate the age of minerals, that the gas would be expected to escape from a porous material, so that its amount is never so great as it should be. Strutt himself found that helium escapes rapidly from powdered monazite, whilst even the solid mineral was found to evolve helium at a rate much in excess of the probable rate of production by radioactive changes. Similar results were found with thorianite, and the only conclusion, since helium is found in the minerals, is that under the conditions under which these minerals exist in the earth’s crust, this escape is checked or altogether prevented. It follows, however, that any age determined from the helium ratio must be a minimum age, since there is always the chance of loss; this of course is not the case—except where the minerals have suffered chemical changes—with the lead ratio, and may account for the discrepancies observed.

Strutt’s earlier work on the helium ratio was made with phosphate minerals (coprolites and fossil bones) of known ages. The ratios found were not in order of age, the minerals being very permeable, so that helium had probably been lost. He next turned his attention to igneous rocks, and selected zircon for the work. Here he obtained some sort of regularity in the order of age and the order given by the ratio, and assumed that if helium were lost at all, it must be lost in roughly proportional amounts by reason of the similarity in conditions. Geological criticism tends to lessen the trustworthiness of the conclusions; it is pointed out that the age of a specimen of zircon is not necessarily that of the rock in which it occurs, for zircon is an extremely stable mineral, and might survive unchanged several fusions and re-crystallisations of the magma. Strutt replies to this that at the temperature of fusion of a rock, zircon would certainly give up its accumulated helium, so that the age determined from the helium content would be that of the last fusion, i.e. the age as given by geological data. On the other hand, our ignorance of the real mechanism of the crystallisation of a magma, and especially of the amount and effect of the pressures obtaining, robs this reply of its force, and the objection must be counted valid.

In still later work Strutt used sphene and thorianite, and his results agree as well as can be expected. The sphenes used were all from Archæan rocks, except one, which was from a Tertiary volcanic deposit of the Laacher See, near Coblenz (the lake is in the crater of an extinct volcano). In this case the helium ratio was very much smaller (about ¹⁄₄₀₀₀ of the values for Archæan rocks) indicating the (comparatively) extremely recent formation of the deposit.

The most recent results in the study of radioactivity point to the conclusion that elements which differ in atomic weight and radioactive properties may be chemically identical, or at least chemically inseparable; such elements have been termed isotopes. The end product of the thorium series of radio-elements should have an atomic weight of about 208·4, and it has been suggested that the element actually produced in this series of changes may be bismuth. The latest results, however, rather point to the conclusion that disintegration in the thorium series gives rise to an isotope of lead. If this hypothesis be true, the lead derived from a mineral rich in thorium and poor in uranium should have an atomic weight appreciably higher than that of ordinary lead. Experiments to test this conclusion have recently been carried out by Soddy and Hyman.[145]

These authors have made analyses of Ceylon thorite, which they find to contain 0·35 per cent. of lead; from the ratio of thorium to uranium in the mineral, they calculate that the lead should have an atomic weight of 208·2, that of ordinary lead being 207·1. Preliminary comparative experiments on 1 gram of pure lead chloride extracted from the mineral point to an atomic weight for the thorite lead of 208·4, a result surprisingly in accord with theory. More extended experiments on this most interesting question are in progress.

The present chapter would be incomplete without a reference to the interesting work of Goldsmidt on radioactivity as an aid in identifying mineral species.[146] He describes a simple method by which the activity of a mineral may be rapidly and easily measured to a sufficient degree of approximation, and shows how the determination enables a line to be drawn on a diagram already mapped out; this line will intersect an area on the diagram which corresponds to the particular mineral. Owing to lack of analytical data, and to the great difficulty of determining with accuracy small quantities of uranium and thorium, the method is at present of scientific interest only; but it is capable of development, and its development would be of undoubted value in the further study of this branch of radioactivity.

In order to make this part of the subject as clear as possible, the chief points in this chapter are summarised as follows:

1. Radioactivity is only observed to an appreciable extent in some rather rare minerals. These minerals as a rule contain radium, uranium, thorium, rare earths, and helium.

2. The helium has been produced during geological time by the degradation of one or more members of the three series of active elements (the Uranium, Actinium, and Thorium series).

3. Radium is a degradation product of uranium, and itself is degraded continuously; the final product of degradation is probably lead.

4. The age of minerals has been calculated from the ratio of lead to uranium; the figures obtained are much greater than those put forward by geologists and physicists.

5. The helium ratio has also been used, but appears less trustworthy, owing to escape of helium, and uncertainty as to geological age of the minerals employed.

6. Some connection between radioactivity and the presence of the yttrium or cerium metals appears highly probable, but no satisfactory theories have been advanced on this point; it has been shown that actinium is very closely allied to lanthanum.

PART II
THE CHEMISTRY OF THE ELEMENTS

CHAPTER IX
GENERAL PROPERTIES OF THE CERIUM AND YTTRIUM GROUPS

The chemistry of the rare earth elements begins in the year 1794, with Gadolin’s discovery of the new oxide ‘Ytterbia,’ for which the name Yttria was subsequently proposed by Ekeberg, and generally adopted (see Chapter I, and under Gadolinite, p. 35). The discovery of Ceria followed in 1804 (see under Cerite, p. 32). The classical work of Mosander, carried out between 1838 and 1842, showed the complex nature of the new oxides. From ceria he separated three new earths, Ceria proper, Lanthana, and Didymia. Yttria was shown to be a mixture of at least three oxides, for which the names Yttria, Erbia, and Terbia were proposed. These oxides were believed to have the general formula RO, by analogy with the alkaline earths, which they were found to resemble in many respects, notably in their strongly basic character.

The properties of the new oxides were examined during the next twenty years by many chemists, the chief workers being Marignac, Rammelsberg, and Hermann, but the next important advance was the investigation of the absorption spectra of solutions of the rare earth salts, first suggested by Gladstone in 1856, and developed more fully by Bunsen and Kirchhoff in 1860 and the following few years. The introduction of the methods of spectrum analysis furnished a very delicate and valuable method of examining and identifying the various oxides, and so greatly assisted the laborious processes of separation.

Sixteen elements (excluding thorium and zirconium) are at the present time recognised as belonging to the rare earth group. With one or two exceptions, these show the closest resemblance to one another, both in chemical behaviour and in the properties of their compounds, so that the difficulties of separating and purifying them are very great. They may be said to form a series, in which the properties vary continuously but gradually from member to member, so that no sharp differences are anywhere perceptible. The method of division into groups is, therefore, almost entirely one of convenience, and has arisen from the course which the separations have followed.

The elements are divided into two chief families or groups, that of the cerium metals and that of the yttrium metals respectively. The cerium elements are separated by a process depending on the relative insolubility of their alkali double sulphates; in this group are included cerium, lanthanum, praseodymium, neodymium, and samarium. The yttrium family is further divided into four sub-groups: the first consists of scandium and yttrium; the second or terbium group of europium, gadolinium, and terbium; the third or erbium group of dysprosium, holmium, erbium, and thulium; and the fourth or ytterbium group of ytterbium and lutecium—the element celtium, recently discovered by Urbain, will also fall into this sub-group, but the discovery awaits confirmation. Whilst scandium and yttrium fall into somewhat abnormal positions, corresponding to their low atomic weights, the terbium elements occupy an intermediate position between the cerium elements and the remaining yttrium elements, or yttrium group proper, and so are frequently classified as a third or intermediate group.

This list does not include all the names which have been put forward to designate what have been claimed from time to time as new elements; whilst the individuality of some of those included is not yet fully established, and the homogeneity of others has been called in question. The uncertainty is more pronounced among the yttrium elements than among the cerium elements; owing to the opportunities for investigation furnished by the commercial treatment of monazite, the chemistry of the cerium group may be regarded as complete.

In the following table the elements are arranged in order of increasing atomic weight, and it can be seen at once how closely the division into groups follows this order:

In their chemical relations, the rare earth elements may be placed between the metals of the alkaline earths, and the trivalent metals iron, aluminium, and chromium. With the exceptions of cerium in the ceric salts, and of samarium and europium in the recently discovered dichlorides, they are uniformly trivalent, but the oxides are very strong bases, and the salts very slightly hydrolysed in dilute solutions; generally, therefore, they resemble the calcium family rather than the aluminium group. Among the common salts, the oxalates, phosphates, chromates, iodates, fluorides, carbonates, tartrates, and borates are almost insoluble; the sulphates are only sparingly soluble at ordinary temperatures. Among the double salts, the alkali double sulphates are of great importance from their employment for separations; the tendency to the formation of complex salts is greater among the yttrium than among the cerium elements, increasing with the atomic weight, and with the decrease in basic strength of the oxides.

The great similarity in chemical behaviour of the rare earth elements is apparent not only in the similarity in composition, solubility and chemical properties of the salts—which is so great that the general account of the compounds which follows applies almost in its entirety to each member of the group—but also in the crystallographic relations between corresponding compounds. Many of the salt hydrates form isomorphous series; the sulphate octohydrates, for example, appear to be isomorphous throughout the whole group, and probably the relation would be found to apply even more completely than is generally accepted, if the necessary data were forthcoming. Of great interest and practical importance is the isomorphism between the nitrates and double nitrates of the cerium elements and bismuth, which has been utilised with such valuable results in the processes of fractional crystallisation.

The Metals.

—The earlier attempts to reduce compounds of the rare earth elements to the metallic condition, by means of metallic sodium or potassium, did not yield pure products; nor did the use of aluminium or magnesium lead to results of practical importance. The metals were first obtained in a coherent physical condition by Hillebrand and Norton,[147] by electrolysis of the fused chlorides. These investigators obtained cerium, lanthanum, and the so-called didymium, and measured their specific heats; their results confirmed the atomic weights assigned to the elements by Mendelejeff, except in the case of lanthanum. Their method has since been elaborated by Muthmann, Hofer and Weiss,[148] who have prepared large quantities of the cerium elements in the pure state. More recently, Hirsch has prepared metallic cerium in large quantities,[149] and has studied its properties.