been nearly the same for all the substances on which the experiments were made. The apparatus made use of was sufficiently simple. The heat was derived from a stove, the fire within which could be elevated, depressed, or entirely withdrawn at pleasure. A very shallow pan of mercury was placed over the stove, the fire being so regulated as to preserve the mercury at any constant required temperature. A cylindrical block of any substance, the conductive power of which was to be determined, was so placed as to rest with its base just in contact with this mercury, from which it derived its temperature (t1). Its upper end was also covered with sufficient mercury just to cover the small bulb of a thermometer. The temperature of this latter mercury gave t1. Careful arrangements were made for observing these temperatures, as well as that of the air into which the heat radiated from the upper mercury. Precautions were also taken to prevent the lateral transference of heat through the sides of the block, and any influence of radiation from the heated stove which might affect the results of the experiments. When the temperature (t) of the upper mercury became stationary, the experiment was completed, and the substitution of this stationary value of t1, together with the values of t, and 7 in the above formula, gave the numerical results required. T 2. The following were some of the results obtained for conductive powers as measured by the ratio Chalk.. k -:- ⚫056 These substances were all in the state of very dry powder. In the last case the sand and clay were in equal quantities. Substances in the state of rock-masses. (1) Calcareous rocks. Chalk (same block from a dry state to a state of saturation with water) from •17 to 30 ⚫30 to 40 .... 50 to 55 (2) Argillaceous substances. (3) Siliceous rocks. New red sandstone (same block dry to satu- 23 to 37 25 to 60 Freestone 33 to 45 Hard compact sandstones (Millstone-grit). . . . 51 to 76 (4) Hard, compact, old sedimentary rocks. 50 to 61 ⚫53 to 1.00 (5) Igneous rocks Effect of Pressure. 3. This effect was not appreciable for a pressure of 7500 lbs. per square inch in such substances as bees'-wax and spermaceti. Nor was there any sensible effect with chalk between a pressure of 4300 lbs. and 7500 lbs. per square inch. Clay which when incompressed had a conducting power="26, had when compressed with 7500 lbs. per inch, a power =' =33; and the conducting power of a mixture of sand and clay in equal quantities rose from 36 to 378 by an increase of pressure from 4300 lbs. to 7500 lbs. per inch. Generally the effect of pressure is much less than might have been anticipated. Effect of Discontinuity. 4. When the conducting mass consists of a number of strata superimposed on each other, the mathematical problem presented to us requires a distinct investigation, which is here given under a very general form, together with the experiments necessary to determine the effect of this kind of discontinuity. The result is that if a mass of sandstone consisted of a number of strata, the conducting powers of which should be about 5, the mean conductivity of the whole would not be diminished by more than about th part, supposing the average thickness of strata to be 1 foot; or by about th, if that average thickness should be 6 inches. This effect is much less than might possibly have been anticipated. Effect of Moisture. 5. This effect was very considerable in those rocks which are great absorbents of water. The maximum effect appears to be produced = by a quantity of moisture which falls considerably short of producing complete saturation. The conducting power of a piece of dried chalk was 19, but became 30 when the substance was very moist. That of a well-dried piece of new red sandstone was =25, but became as much as 60 when saturated. Both these substances absorbed a large quantity of water. Ancaster oolites absorbed considerably less, and their conductivity was affected in a smaller degree. For a block of dry clay the conductive power was 23, and became 37 when well moistened. Close indurated sandstone, paleozoic rocks of close texture, and igneous rocks are bad absorbents, and are very little affected in their conductive powers by moisture. Comparison of Deductions from Theories of Terrestrial Temperature with the Results of Observation. 6. It has long been established by mathematical investigation, that if a large globe like the Earth be heated in any manner and in any degree, its temperature at points not too remote from its surface, and after a sufficient lapse of time, will necessarily become such that the increase of temperature in descending along a vertical line will be proportional to the increase of depth. In this enunciation, however, it is assumed that the conductive power throughout the mass, or at least throughout its more external portion, is uniform. The difference of conductive power between the unstratified and sedimentary portion of the earth's crust, or that between one sedimentary portion and another, has not hitherto been taken into account *. The author has investigated the problem assuming the crust of the globe to consist of any number of strata of different conductive powers and bounded by parallel surfaces, the problem being much simplified by considering their surfaces as plane instead of spherical. Then, assuming the temperature of the crust of the globe to be due entirely to the transference of heat from its central portions to its surface, it is shown that the increase of temperature in descending vertically through any two strata, ought to be in the inverse ratio of the conductive powers of those strata, whether the two strata belong to the same group of stratified beds, or to two different groups in different localities. Such at least must be the result unless we introduce very * Except in the case in which Poisson investigates the state of temperature of a sphere surrounded by a single concentric spherical shell of different conductivity. VOL. VIII. 2 R arbitrary and, as the author conceives, entirely inadmissible hypotheses into the problem. For the purpose of testing this theory in its application to our own globe, four or five cases of Artesian wells and vertical shafts are especially referred to, in which the temperature has been carefully observed at greater depths than at any other places in Western Europe, or probably in any other part of the globe*. The cases spoken of are the following: (1) An Artesian well near Geneva.-Depth=225 metres; increase of depth for 1° (F.)=55 feet. (2) An Artesian well at Mondorff in the Grand Duchy of Luxembourg.-Depth=730 metres; increase of depth for 1° (F.)=57 feet, (3) An Artesian well at New-Saltzwerk in Westphalia.-Depth =644.5 metres; increase of depth for 1° (F.)=54 feet. (4) The Puis de Grenelle at Paris.-Depth=546 metres; increase of depth for 1° (F.) 60 feet. (5) A coal shaft at Duckenfield, near Manchester.-Depth=1400 feet; increase of depth for 1° (F.)=65 feet. (6) A coal shaft at Monkwearmouth.-Depth about 1700 or 1800 feet; increase of depth for 1° (F.) about=60 feet. The general rate of increase of temperature in our own deeper coal-mines is that of about 1° (F.) for 60 feet in depth; and the same result has been obtained for many parts of the chalk in Northern France. These cases present a remarkable approximation to uniformity, whereas the conductive powers of the strata which have been penetrated are very different. Cases (4) and (5) present the best means of comparison. The Puis de Grenelle passes through nearly 500 metres of chalk, the conducting power of which is estimated by the author at somewhat more than 25, while the mean conducting power of the rocks through which the coal shaft at Duckenfield passes, is estimated, by means of experiments performed on specimens of these rocks, at rather more than 5. This is about twice as much as in the former case, whereas the depths corresponding to the same increase of temperature are only as 65 to 60, instead of being in the ratio of about 65 to 35, as they ought to be according to the In a great majority of instances observations of this kind have not been made with sufficient care to be relied on. theory here considered. In all the other cases the conductive powers of the masses penetrated are doubtless greater than that of the chalk at Paris, though, for the most part, they present a more rapid increase of temperature in descending, instead of a less rapid increase (as this theory would prescribe) than the Puis de Grenelle. Within the region comprising the cases above cited, there are many local variations as to the rate of increase of terrestrial temperature in descending below the earth's surface. The author conceives that these phenomena cannot be accounted for according to this theory without the introduction of arbitrary hypotheses. Upon the whole, he believes that in the present state of our knowledge of terrestrial temperature, it is impossible to account for its phenomena by regarding them as the consequence simply of heat, not generated in, but transmitted through the crust of the globe from some deep-seated central source. The discrepancy between the actual terrestrial temperatures and those which would be assigned by the theory here discussed, may be illustrated perhaps by placing the subject in a rather different point of view. It is assumed in the theoretical investigation, that the isothermal surfaces at depths sufficiently great (as 50 or 100 miles for example) are approximately concentric with the earth's external surface, or, speaking with reference to areas not too large, parallel to that surface, in which case it is proved that the isothermal surfaces at comparatively small depths (not much exceeding that of the sedimentary beds) cannot be parallel to the external surface. For example, the depth of an isothermal surface of given temperature, which should be some 3000 feet at the Puis de Grenelle, ought to be nearly 6000 feet at the coal shaft at Duckenfield; and at other places it ought to be very nearly proportional to the conductive power of the terrestrial mass lying above it. But the observations above cited demonstrate that, independently of local irregularities, such an isothermal surface is nearly at equal depths throughout the whole region of Western Europe. No theory of terrestrial temperature, then, can meet the requirements of observation which does not account for isothermal surfaces approximately parallel (with local variations) to the earth's external surface at comparatively small depths beneath it. Moreover, it is easily shown that the quantity of heat transmitted from such a surface to the external surface, must be proportional to the conductive power |