LXYTII. The Radioactivity of Terrestrial Surface Materials. By J. Joly, F.R.S [Communicated by the Author.] A SOMEWHAT considerable experience of the electric furnace as a means of determining the radium content of rocks f has led me to the conclusion that a revision of the subject of terrestrial radioactivity by its means is desirable. This view is founded upon two circumstances : (1) the method by fusion possesses features rendering it more reliable than any other method, and (2) in its results it differs considerably from the method by solution. As regards the first point, the reliability and searching character of the method, I have dealt with the subject elsewhere (loc. cit.). It may, however, be well to recall here that in this method the emanation is taken directly from the pulverized rock while this is being decomposed by the alkaline carbonates at temperatures rising to 1100° C. or more. In this process the melt possesses but small bulk and the exposed surface is very large. In fact the melt makes but a shallow film in the platinum boat in which it is being heated. (I re'er now to the use of the tube furnace.) There is complete decomposition and intense ebullition lasting for about half an hour. The gases evolved are led through a soda-lime tube ; the unabsorbed gas (for the most part air driven by thermal expansion from the furnace) is stored in a strong rubber bag, and finally carried again through the soda-lime tube into the electroscope ; a final current of air washes out both furnace and tube, and ensures that practically the whole of the emanation is brought into the electroscope. The completeness of the wash-out has been tested by filling a second electroscope through the entire system : when little, if any, residual emanation has been detected. It may, therefore, be assumed that, sensibly, all the emanation evolved in the furnace reaches the electroscope. There is no reason (o assume that any part of it leaks through the platinum. Such an occurrence, in view of the high atomic weight of emanation, would be contrary to expectation : moreover, tests carried out on varying quantities of material have been quite satisfactory. Hardly less important is the protection against contamination, which naturally arises in the simple nature of the procedure involved. The rock is pulverized till it passes a sieve of about 25 mesh to the centimetre. From seven to ten [header] 695 grams are weighed out and mixed with the alkaline carbonates or other fluxes required. The mixture is put directly into the furnace. The platinum boat and all parts o£ the furnace may be completely de-emanated by a preliminary heating, if thought requisite. When experiments are being made daily this naturally comes about. All the risky chemical operations of the method by solution — often involving a succession of filtrations and fusions — are here avoided. But the most important advantage which can be claimed for the method by fusion has still to be mentioned. The method by solution is exposed to a source of error in the liability of the radium present to assume the non-emanating form — by adsorption or precipitation or from some other cause. This is now fully recognized. It may be sometimes detected in the erratic nature of results obtained in successive experiments upon the one solution. Sometimes it appears as if the presence of sulphates or gelatinous silica was requisite. It is matter of satisfaction that Messrs. Eve and McIntosh, who first drew attention to this in the present connexion, are now investigating the subject [citation redacted]. It does not appear probable that any risk of this kind can beset the method by fusion. The second point to which I referred above — the fact that there is a substantial difference in the results of experiments made by solution and by fusion — will be illustrated in the data contained in this paper. A study of the published results obtained by solution shows much divergency. It seems certain that this largely arises in the actual nature of the material. But the means of several observers are very different. They appear to leave in uncertainty some of the most general questions arising : e. g,, the true mean for acid, intermediate and basic rocks and the difference between the radium content of primary and secondary rocks [f A recent investigator has actually questioned if there is any such difference, and points out that certain solution results on trachytes are actually lower than his own results on limestones. [citation redacted].]. In this paper — which is preliminary in its scope — I shall deal with general results rather than with details. Thus the question of " magmatic differentiation" as an influence in the segregation of radium I have put aside for the present ; on that account leaving out determinations upon differentiation products or what are believed to be such. The acid, intermediate, and basic magmas, as represented by undifferentiated plutonic, injected, or volcanic materials, are alone [footer] 696 [header] dealt with, because these are primarily involved in the wider geological applications of radioactivity. The method pursued is different from what has hitherto been adopted in such investigations. Individual rock specimens are not dealt with. A large number of typical rock samples are reduced to powder. A similar weight of each is taken and the added powders thoroughly mixed. The determination in the furnace is made upon one or more samples of this composite powder. If there is anything to suggest abnormality in the result, a second determination is made upon a composite powder compounded of about one half the original samples. If a high result was due to the presence of one very radioactive rock, this fact must be detected in this second experiment. If it is desired to eliminate such an excessively radioactive material it may be rapidly isolated by a few successive experiments upon one half, one quarter, &c. of the original samples separately compounded. This method has many advantages. At this stage of the inquiry it yields all the information required. The labour does not increase with the number of rock specimens ; on the contrary, the more rocks enter into the composite material the less need to consider individual abnormalities, or, we might say, the more the necessity for their inclusion. As the number of available rock specimens increases I hope to place the means arrived at in this manner upon a more secure basis. There is a certain small latitude of uncertainty in the results. The " constant " of furnace and electroscope working in conjunction, as determined by dealing with known amounts of radium, may vary somewhat. In a series of experiments carried out to ascertain the constant proper to the present series of experiments, variations of its value from 07 to 0*8 were obtained. The mean of these values has been taken. Comparison with results obtained when this method was first being developed (loc. cit.) will show that this value is well supported by other observations. In experiments upon the sedimentary rocks, carried out in my laboratory, a somewhat lower constant, i. e. 0'6, was arrived at [citation redacted]. The difference is probably due to the electroscopes used in the two cases. It is satisfactory that the results on composites of the sedimentary rocks, given further on, wherein the higher constant is used, almost exactly substantiate means arrived at on the detailed results calculated on the lower constant. This not only brings the latter measurements into line with those on [header] 697 the igneous rocks, but it supports the value of: the constant assumed above. In the following table of results the quantities of radium are, as usual, expressed in billionths of a gram per gram of rock, and must, therefore, be multiplied by 10 ~ 12 . Igneous Rocks. Acid. Intermediate Basic Additional [table redacted] In selecting rocks for the foregoing determinations I have endeavoured to include material as representative as possible. In the composites of the intrusive and volcanic series, differentiation products are excluded ; i. e. the chemical extremes of the groups. But it would seem from the results as if differentiation had operated to segregate radium in these rocks. The interesting question is raised as to how far such differentiation is indicative of more general effects of the kind, or may serve to indicate general differentiation. I hope to return to this question at a later date, when composites of the complementary differentiation products are being examined. The three composites of basalts serve to illustrate the divergence which exists in the radium content of these rocks. Thus the fourteen rocks, of which one only is from the great Hebridean outflow and ten are from the Antarctic and Deccan 698 [header] ureas, possess four times as much radium as the Hebridean basalt. The result for the latter confirms Strutt's low figure for this rock [citation redacted]. Finally, a carefully made up composite, in which basalts from the most widely separated areas are equally represented, gives a value approximating to a mean between the high and low results. The general mean for basic rocks includes the more representative series of basalts only. Of the Additional rocks, the gneisses not being in all cases of certain igneous derivation, are not available for general deductions. For comparison with the determinations by fusion I have classified below the results obtained in some of the leading investigations in which the method by solution has been employed. The numbers in brackets given after the results indicate the number of rocks dealt with. Granites, Acid Intrusives and Volcanics. Syenites Diorites. Trachytes and Andesites. Gabbros and Norites Gabbros and Diabases Basalts. [table redacted] [Fletcher, Phil. Mag. July 1910; Jan. 1911; June 1911. In the analysis given above the series of observations on the Leinster granite are treated as representing but one rock. Otherwise a considerable lowering of the mean for granite must arise. I have not dealt similarly with the series on the Andine lavas, as these appear to include distinct outflows and petrologically distinct varieties of trachytes and andesites.] When the means for acid, intermediate, and basic rocks are calculated from the above table, giving equal weight to each separate rock determination, we have: — Acid, Intermediate, Basic, [table redacted] [header] 699 It will be seen that these depart considerably from the means obtained on composite rock-powders, using the method by fusion : the intermediate and basic groups, more especially, are about 100 per cent, of their value too low if the method by fusion gives correct results. I have in the above analyses of results by solution omitted some results of my own [citation redacted] which afforded generally higher values than those obtained by other investigators using the same method. No clue to this difference is as yet forthcoming, and until the results by fusion were obtained it seemed best to assume that some difference in the preparation of the solutions was responsible, in which case the higher results would be the more reliable. However, many of my determinations by solution stand some fifty per cent, (or even more) of their value above the fusion results, and as it seems improbable that emanation can be retained when the latter method is employed, it seems safer to set aside for the present such determinations of mine as have not been confirmed by the use of the electric furnace. A remarkable fact is that just some of the highest of my results have been so confirmed. Thus, in the results obtained by composite fusions given above, the mixed powders of seven Vesuvian lavas give 12' 6. The results obtained by solution on these same lavas would afford a mean value of 10*8 [citation redacted] Again, eleven of the St. Gothard granites gave 7'2 by solution, and six other examples of this granite gave by fusion 6*0 [citation redacted]. I have also omitted some high values obtained by Gockel using the solution method §. These results give a mean of 74*6 for six rocks. His low results are obtained by a radiation method, and are not easily compared with the measurements by solution. In his first paper {loc. cit.) on the subject of terrestrial radioactivity, Strutt pointed out that the acid rocks appeared to contain a larger amount of radium than the basic rocks. The fusion results support this conclusion. There are extremes of radioactivity found in all the chemical groups. These may be regarded as exceptional, or set off one against the other. We see such extremes in the basic Vesuvian lavas and in the granites of the St. Gothard or of Karangan [citation redacted]. The fact 700 [header] remains that the great body of acid rocks are the richest in radium, and are over twice as rich as the typically basic group. I have shown that the thorium content of the acid rocks is also in excess of that of the basic rocks [citation redacted]. No explanation of this is forthcoming. It may arise in the primary differentiation of a parent magma into granitoid and gabbroid divisions. It has been maintained that these divisions represent in themselves fundamental magmas, and that they are present in about equal proportions in the lithosphere [citation redacted]. The connexion between radioactivity and silica content may turn out to be closer than a rough proportionality. If we take the three chemical rock-divisions as characterized by 74, 60, and 48 per cent, of silica and assign to the most acid division a radium content of 3, we find the radium in the other groups, if proportional to the silica, would be 2*4 and I' 9. The basic division departs most from the proportionality, but considering the extreme variations both in silica content and radium content which obtain in this division, the agreement seems sufficiently close to suggest some connexion between the segregation of uranium and of silica in the history of the magma. A similar proportionality is indicated in the thorium content [citation redacted]: but here the number of observations is insufficient. I hope to be shortly in a position to enter more fully into this question. Coming now to the availability for geological discussion of such results as the foregoing, we are met by the difficulty that we really do not know the quantitative distribution of the several rock families in the lithosphere. When, therefore, we seek to derive a general mean radium content our conclusion may be attended with a certain amount of error. On the result of a very considerable number of analyses of igneous rocks, Clarke § concludes that the lithosphere to a depth of ten miles probably approximates in composition to . that of a diorite or andesite ; or is, in short, " intermediate " in chemical character. Jf this is correct we should assign to it such a radium content as we find in intermediate rocks, i. e. 2*57. If we combine the gabbros and granites we get a mean of 2*0, and if basic and acidic generally are combined the average is 2'2. It would appear as if the radium content of average igneous rocks at the surface is not below 2'0, and may probably be about 2 5. If the entire number of 105 [header] 701 rocks entering into the fusion results are taken, the general mean is 2*60. Any one of these means is considerably in excess of what is deducible from the results obtained by solution. When the presence of thorium is taken into account we have the data for a complete evaluation of the number of calories developed per unit of time in average surface rock. The general average for the thorium content of igneous rocks appears to be about 2'60 x 10~ 5 gram per gram [citation redacted]. ]f we take 2 x 10~ 5 for purposes of calculation, we find 13'28 x 10~ 14 calorie as the heat developed per second per gram. Taking the radium as 2xl0~ v2 , we have to add 12-00 xlO" 14 calorie. The total is 25 X 10~ u gram-degree of heat per second. A material of this degree of radioactivity can only be supposed to extend downwards a relatively small distance. Assuming the surface gradient to be 1° in 35 metres, the conductivity to be 0*004, and the specific gravity of the rock to be 2*7, we find that a downward extension of 17 kilometres (10*6 miles) suffices to account for all the heat reaching the surface. This conclusion lands us in the difficulty of accounting for the high temperatures which are believed to extensively prevail deep in the crust. At the base of the 17 kilometre layer the temperature due to radioactivity would be no more than 246° C. We are from this driven to the conclusion that rocks of this degree of radioactivity do rot, in point of fact, extend downwards till all the radioactive materials present are exhausted. The low radioactivity of certain of the basalts probably affords an indication of the vertical distribution of rocks in the earth's crust. It seems a probable conclusion that such basic and feebly radioactive materials make up the deeper parts of the surface crust, and by carrying the heat-producing elements to a sufficient depth, account for such temperatures prevailing beneath as many geological phenomena seem to require. Thus it will be found on calculation that if basalts with a radium content of 0*5 x 10 " 13 and a thorium content of 1*0 X 10~ 5 carried most of. the radioactive materials, temperatures of between 500° C. and G00° (would, exist at depths of some 40 kilometres (25 miles) from the surface [In a recent determination Mr. Louis Smyth finds the thorium-content of the typical composite basalt to be 0*47 X 10- 5 gram per gram. This result would involve a greater extension downwards and higher basal temperatures than those cited above.]. It must be remembered, however, that the thermal state in the depths of the crust is largely 702 [header] matter of surmise. Most indications of deep-seated high temperatures are restricted to the geosynclines and over these sediment-laden areas the surface crust cannot be considered normal. The Secondary Mocks. The radium content of the secondary rocks has been dealt with by A. L. Fletcher [citation redacted], using the method by fusion as it is applied in my laboratory. The results are : — [table redacted] These rocks were treated in detail, the figures are the mean of the several observations in each group. The fact that the argillaceous rocks are inferior in radium content to the arenaceous is to be explained in the nature of the arenaceous materials dealt with. Many of these were by no means ultimate products of solvent denudation, that is were not purely quartzose. The importance of verifying the means for the two leading groups led me to treat composite sediments in the electric furnace. The agreement with the means given above is satisfactory: — [table redacted] It would appear from the comparison of the above with the few published results on sedimentary rocks obtained by the solution method that most of the latter are too low by about 50 per cent, of their value. The leading data known to me are : — [table redacted] [citation redacted] [header] 703 It must be remembered that the calcareous rocks are relatively o£ little importance in estimating the average radioactivity of the sediments. Estimates of their relative abundance concur in representing them as constituting but a few per cent, (less than 10) of the total sedimentary mass. We infer from this that 1*5 may be accepted as the average radium content of the sedimentary rocks. Results on the amount of thorium in the sediments show that they contain about 1*16 x 10~ 5 gram per gram [citation redacted]. From these data we get for the rate of development of radioactive heat in these rocks : — [table redacted] giving a total of 16*6 x 10 _u cal. The corresponding value for the igneous rocks is, as we have seen, 25 x 10" u cal. It is only in the geosynclines, wherein the great body of the sediments has collected throughout vast periods of geological time, that the geological importance of the radioactive heat of the sediments becomes conspicuous. The sedimentary deposits may amount to more than 10 kilometres in thickness. These great masses are laid down upon the normal igneous crust , and the intensification of the deep-seated temperatures thereby brought about would appear adequate to account for the leading facts of mountain elevation. The matter has been dealt with elsewhere [citation redacted]. The results obtained by the fusion and decomposition of rocks in the electric furnace assign to the sediments 1*5 and to the igneous rocks 2*5 as mean radium content. Of the existence of a substantial difference between these two classes of materials there appears to be no doubt. The denudative loss of radium (which, of course, stands for a measure of the parent substance, uranium) to the ocean is greater than appears at first sight from the foregoing figures; for there is a reduction of mass attending the processes of denudation amounting to about 33 percent, of the original rock. Hence in the sediments we have but 67 per cent, of the primary rock, and this containing but 60 per cent, of the original radium content. The total loss has been about 60 per cent, of the radium of the primary rock ; or for every tonne of igneous rock converted to sediment the loss of radium has been 15 x 10" 7 gram. This is carried into the ocean and remains in its waters and its underlying deposits. 704 [header] I have shown elsewhere that it is possible to use the amount of sodium in the ocean as a measure of the total quantity of primary rock which has been denuded in the course of geological history [citation redacted]; and later I calculated by its means the total mass of the sub-oceanic sediment [citation redacted]. These estimates are independent of the duration of geological time. They, in fact, do not involve the rate of denudation, but only the integral of its products. The security they possess may be judged from the fact that the only assumption involved in their derivation is that the sodium of the ocean is a product of solvent denudation which has been contributed to the ocean and which, practically, in its entirety now remains to our observation. The rest of the argument is based on our knowledge of the sodium content of average igneous and sedimentary rocks. The mass of igneous rock, parent to the sediments, proves to be 84*3 X 10 16 tonnes. The radium brought into the ocean has, therefore, been 84*3 X 10 16 x 15 x 10~ 7 or 1264 x 10 9 grams. This amount must be contained in its waters and in the sub-oceanic sediments. We may first make a deduction for the amount now in solution. Estimates of oceanic radium vary. If we take 4 x 10 -12 gram per kilogram or 4 x 10~ 9 per tonne, we have, multiplying by the mass of the ocean, i. e. by 1*32 x 10 18 tonnes, the quantity 5*3 x 10 9 grams as the total amount in solution. The deduction is trifling and leaves 1259 x 10 9 grains assignable to the sub-oceanic sediments. The mass of these in the aggregate is 19*5 x 10 16 tonnes, and hence we must ascribe to them an average radium content of 6'4x 10~ 12 gram per gram. The following measurements of the radium content of oceanic deposits have been made by the fusion method [They are considerably lower than the results which I formerly obtained by solution, but must be considered more reliable.]: — [table redacted] [header] 705 the measurements of the radium content of these materials. This result generally confirms the statistics of solvent denudation. But in a subsequent paper I hope to show that a full consideration of the data now at our disposal respecting the sub-oceanic sediments leads to the conclusion that the less radioactive deposits have been subject to considerable changes in chemical composition during recent geological times, and do not possess throughout their entire mass the chemical composition we observe in samples dredged from their surface. The subject is of sufficient interest to merit separate treatment. I desire to thank Mr. Louis B. Smyth and Mr. A. L. Fletcher for some experiments made in connexion with this paper, Iveagh Geological Laboratory, Trinity College, Dublin. Sept. 1912.