53- 



NATURE 



[September 30, 1897 



much more compressible class of organic bodies, the rate would ture and pressure were made to vary in such a way (both in 

 probably be more nearly constant in the same body (siliceous or creasing) as to leave viscosity constant, it was inferred that heat 



organic) changed only as to temperature and pressure. This — -' — *:— '-i -' =- — ~~*-"^ ivt„... .u„ : ^.:_„ 



surmise was verified for naphthalene within an interval of 2000 

 atmospheres. 



The endeavour to interpret the change during fusion of the 

 volume of the chemical elements in terms of the periodic system 

 has been begun with much success by Max Topler for low teni- 

 peratures. It would be of great interest to complete this 



diagram for high temperatures in view of the specifically 

 molecular character of the fusion phenomenon, by repeating 

 such experiments as have just been described for rock 

 magmas. 



The heat conduction of rocks has been investigated in many 

 cases for temperatures lying below red heat. Among recent 

 observers we need only instance the extensive investigations of 

 Ayrton and Perry. No trustworthy experiments, however, have 

 yet been carried into the region of essentially high temperature ; 

 and yet, what is chiefly of interest in the geological applications 

 of such experiments is the change of conduction which accom- 

 jmnies changes of physical state, whether induced by pressure or 

 by temperature. 



Experiments in heat conduction are difficult under any cir- 

 cumstances. They become insuperably so when conduction at 

 white heat is to be studied under pressure, and that is what 

 the geological conditions actually imply. Some notion of a 

 body respectively solid and liquid at a given temperature 

 may be obtained by observing the behaviour of bodies 

 which are capable of being under-cooled. A number of such 

 bodies are known, thymol being a conspicuous example. Ex- 

 periments with this body were made by measuring the volume 

 expansion, specific heat, and heat conduction in parallel series 

 both for the solid and liquid state at like temperatures. They 

 showed, for instance, that the increment of absolute heat con- 

 duction, encountered in passing isothermally from the solid to 

 the liquid state, when referred to solid conductivity is about j convincing ; but in physical geology, for some time to come I 



conduction would also remain constant. Now the isometrics 

 or lines of constant viscosity of a viscous body for variable 

 pressure and temperature are much more easily found than the 

 isometrics of conduction. In fact, it has been shown that a 

 burden of at least 200 atmospheres would have to be brought 

 to bear in order to wipe out the decreased viscosity due to the 

 rise of a single degree (Centigrade) of temperature. The 

 depth at which this ratio is reached, as King points out, for 

 a given surface gradient of temperature and depth, depends on 

 the initial excess of the temperature of the earth considered, 

 and on the age of the temperature distribution resulting. But no 

 matter whether the Kelvin earth with an initial excess of 3900 

 and an age of 100 x 10'' years, or whether King's solid earth 

 with an initial temperature of fused platinum and 25 x 10* 

 years of life, be taken— in all cases the temperature effect pre- 

 dominates throughout those depths within which change of 

 temperature with depth is the marked feature of the temperature 

 distribution. In other words, if, for example, we consider the 

 Kelvin earth, the strata above 0035 of earth radius will be strata 

 of smaller conduction than the surface strata. From the sur- 

 face downwards as far as 0035 radius, conduction will decrease to 

 a minimum. Below this, conduction will increase again due to 

 preponderating pressure, finally to exceed the surface value. But 

 the computed temperature distribution of Kelvin's earth is such 

 that at "035 radius the initial temperature excess of 3900" has 

 been reached to within i or 2 per cent. Below this in depth, 

 Perry's correction would begin to apply, but the further changes 

 of temperature are so nearly negligible that the consideration 

 of conduction is superfluous. From this point of view, there- 

 fore, the staggering force of Perry's clever argument is removed. 

 Of course, I am fully aware that an argument from the supposed 

 parallelism of physical properties of a given body (in the 

 present case heat conduction and viscosity) is not inevitably 



13 per cent., and when referred to a liquid conductivity is about 

 15 per cent. Similarly, the change of thermometric conductivity, [ 

 under like conditions, is an increment of 36 per cent, and 56 per i 

 cent, respectively. Now, since in most questions relating to 

 thermal flow thermometric conduction enters exclusively, the | 

 importance of this large coefficient is obvious whenever a body \ 

 passes from the solid to the liquid state. 1 



Solid conduction is thus 40 or 50 per cent, in excess of liquid j 

 conductivity for the same body at the same temperature and 

 pressure. It is reasonable to infer that a corresponding decre- 

 ment of conduction will accompany any rise of temperature of | 

 a solid body. Measurements which have somewhat recently 

 been made for relatively small intervals at Zurich, at Glasgow, ! 

 and at Harvard upon typical rocks, all bear out this surmise. 

 The immediate incentive to these experiments was a strong ; 

 paper by Prof. Perry, in which Lord Kelvin's estimate of the I 

 age of the earth is shown to be insufficient for an earth in which ; 

 the interior conductivity is systematically greater than the sur- j 

 face conductivity. Indeed, he deduces tlie percentage increment 1 

 of the square root of the age of a Perry earth over that of a 1 

 Kelvin earth to be one-fifth of the percentage decrement of con- I 

 duction for each 100'' C. So far as the efl'ect of terrestrial tem- j 

 perature alone is concerned, the measurements just mentioned 

 show that Perry's correction is negative or that Perry's earth | 

 would be less long-lived than the 100 x 10" limit of years set 

 by Lord Kelvin.' 



To estimate the effect on heat conduction of the increase of 

 pressure which accompanies the increase of temperature with 

 the depth below the surface is a much more serious matter. In 

 the laboratory, pressure experiments are limited to 3000 or 4000 

 atmospheres ; compared with earth pressures, these scarcely 

 amount to a scratch on the surface ; yet even for this limit the 

 determination of heat conduction at high temperatures is out of 

 the question. A tentative method of arriving at a conclusion is 

 given by Clarence King in his discussion of the age of the earth, 

 the consequences of which have beenquite overlooked. What King 

 endeavoured to accentuate, long before Perry's contribution to the 

 subject, was precisely the fact we caimot assume greater conduc- 

 tivity for the interior than for the .surface. Since heat conduc- 

 tion decreases isothermally from solid to liquid,' it was assumed 

 that, in one and the same substance, the viscosity could be taken 

 as an index of the thermal conduction. Therefore if tempera- 



1 The text of Kelvin's recent address at the Victoria Institute, in which 

 an age of thirty million years is maintained, has not yet reached me. 



dare say, the question will be not one of rigorous proof, but 

 rather one of forming a rational opinion. 



In passing I will indicate the importance of an increased 

 knowledge of the isometrics of liquid ami solid matter, relations 

 which have thus far been found simpler in character than other 

 thermodynamic properties, as I shall again point out in the 

 course of the address. 



I want, finally, to add a few words on the electro-chemistry 

 of magmas. The physical chemistry of molten rock has 

 already been somewhat extensively considered, but I am hardly 

 competent to review it. In the United States, Joseph Iddings 

 and, more recently, George F. Becker have discussed the 

 natural diff"erentiation of magmas from different points of 

 view. Here I will merely include certain pyrometric experi- 

 ments on the electric conduction of fused glasses which seem 

 to give promise of throwing light on the chemical constitution 

 of complex silicates and to be suggestive in other ways. In 

 such measurements, if the magma is made to pass from the solid 

 to the liquid state, the observed electric conduction contains 

 no evidence either of a melting point or of polymerism. The 

 law of thermal variation is easily derived and it agrees closely 

 with the corresponding behaviour of a zinc sulphate solution, for 

 instance, observed through a range of temperature. In a 

 general way, electric resistance decreases in geometric progres- 

 sion when temperature increases in arithmetical progression. 

 Considered relatively to the composition of the magmas, electric 

 conduction shows a marked and regular increase with the 

 degree of acidity of the magmas. The less fusible acid magmas 

 are better conductors than the basic magmas at the same tem- 

 perature. Curiously enough, conduction thus runs in an oppo- 

 site direction to fusibility. However viscous a magma may be, 

 therefore, and however cogent the arguments such as those 

 launched by Becker against the differentiating importance of 

 ordinary diff'usion may prove, it is fair to conclude that a 

 thorough change of chemical structure through ionic diff"usion, 

 whether directed by an electric field or otherwise, rnust be an 

 easy possibility for a sufficiently hot, but otherwise solid, magma. 

 The results point specifically to the desirability of repeating 

 Hittorf s brilliant experiments on the migration of the ions for a 

 siliceous medium. This ought not to be difficult, seeing that 

 such a menstruum need not even be liquid to be compatible 

 with a high order of electric conduction. 



Further consideration of the subject shows the probable pas- 

 sage of conduction through a maximum when acidity is con- 



NO. 1457, VOL. 56] 



