386 



NA TURE 



\Sept. lo, 1874 



THE BRITISH ASSOCIATION 

 Reports 



Report of the Committee oh Mxtetiments to determine the 

 Thermal Conduetivities of certain Rocks, by Prof. Herschel and 

 Mr. Lebour. 



In the introductory notes on these experiments, published as an 

 appendix (p. 233) in the last volume (for 1873) of the British Asso- 

 ciation Reports, the list of rocks selected and the manner of experi- 

 menting on them were described. With the exception that sections 

 of Calton trap rock, of a great pyramid casing stone (nummulitic 

 limestone), Caen stone (or Normandy building limestone), Cannel 

 coal, chalk, and red brick were added to this list, and that the 

 apparatus received some small but very important improvements 

 to make it heat tight, the material of the experiments as well as 

 the method of working them remained substantially the same as 

 last year. Instead of a conical tin vessel with i lb. of water, a 

 cylindrical one holding 2ilbs., with an internal agitator and 

 thermometer, was used as the cooler. The opposite faces of the 

 heater and cooler were lined with velvet and each clasped by a 

 caoutchouc collar, which, projecting a little above them, clips 

 tlie circular edge of the rock plate when it is placed between 

 them, and two small notches cut in each collar also allow the 

 wires of a thermocouple to be introduced, touching the rock- 

 surfaces while the rock is being heated. With the view of tra- 

 versing the plate with the thermopile in different directions, the 

 piece of stout palladium wire (about iS gauge) used as the electro- 

 motive element between the two iron wire branches of a delicate 

 reflecting galvanometer, was silver-soldered to the iron wires at 

 its two ends, all the wires being first rolled thin and flat to some 

 distance from the junctions. The scythe or scimitar-blade shape 

 most easily given to the wire in rolling it thin was advantageous 

 in the construction, because, instead of uniting the wires continu- 

 ously in ore straight length and folding the points of junction on 

 opposite sides of the rock (thus confining their range upon it to 

 a single diameter or to one straight line), advantage of the curva- 

 ture was taken to connect the wires by superposition, instead of 

 by prolongation at their junctions, without overlying each other, 

 into two flat ogee arches, or merry-thought-like blades, between 

 which the rock is held as in a forceps. The straight unrolled 

 piarts of the wires are bound very firmly to a square piece of 

 wood, which acts as a handle to guide the points of the forceps 

 to various parts of the rock-faces, while it keeps them securely in 

 their places, and thus allows the small elastic pressure of the 

 wire5 to clasp the rock gently between the points of the thermo- 

 electric pincette without assistance from the velvet covers. After 

 thus inserting a rock section in the apparatus, protecting the 

 rock and cooler from below with a stout wooden screen, and 

 from loss or gain of heat in other directions by a suitably thick 

 case of woollen stuff and a few bandages of similar materials, the 

 rate of rise of temperature in the cooler, when agitated, was 

 noted by the average number of seconds taken by a delicate 

 thermometer contained in it to rise \° F. (one graduation on its 

 stem) as soon as this rate of rise was found to have become 

 sensibly constant. About twenty minutes were usually occupied 

 in the beginning of an experiment W'ith waiting for a steady con- 

 di ion of the thermometer readings, and ten or twelve minutes 

 more were required to ensure it and to obtain the average rate of 

 their increase for the rock specimen under observation. The 

 temperature difference shown by the galvanometer at the same 

 time at first rose r.ipidly to a high maximum and then descended 

 very gradually to a fixed lower reading. The pincette was tra- 

 versed to and fio over the rock surfaces while the thermometer 

 was being noted, and exhibited during these motions fluctuations 

 answeringto about 1° or 2" F. on either side of an average posi- 

 tion ; corrected for zero of the scale and reduced by trials for this 

 purpose between every two or three experiments to Fahrenheit 

 degrees, the temperature dilTerence thus formed, divided by the 

 quantity of heat tiansmitted to the cooler per minute, gave the 

 apparent thermal conductivity of the pla'e. The results in 

 Peclet's units were scarcely more than one-third of what Peclet 

 and other earlier experimenters had obtained. It was nbvious 

 that, instead of maiking the temperature difference between the 

 two solid contact surfaces of the rock and velvet which they 

 touched, the points of tlie thermo-electric forceps showed the 

 temperature ol the fluid air-baih in which those two surfaces 

 are immersed. 1 he ex'ix-me mobility of this integument, enabling 

 ii to I eni-lrate thiongli ihi; \< l\et to the ]>lates uf the he.iler and 

 the cooler, while it equally insinuates itself between the rock sur- 

 face and the thermopile that can only enter into actual solid con- 



tact with each other (at least theoretically) at three points, 

 controls the temiierature of the metallic thermometer far more 

 powerfully than the rock face that it touches, and the real tem- 

 perature differences between the rock faces are accordingly com- 

 pletely masked. It is very probable that if the velvet covers 1 

 fitted on the instrument were replaced by soft wash-leather, the 

 source of this error would be very much reduced ; and although 

 it is certain that the confronting surfaces of the rock and leather 

 faces will nowhere have actually the same temperature, from the 

 existence of a sensible quantity of resisting air between them, so 

 that, as before, the thermopile will not mark the true rock tem- 

 perature difference, but a mean between that difference and a 

 similar difference for the leather faces, yet the range of this error 

 will be incomparably smaller than in the experiments already 

 made with velvet covers, whose loose texture precludes the possi- 

 bility of regarding the comparative results now obtained as posi- 

 tively correct, or more than first approximations, from which the 

 errors arising from surfice characters of the rock sections tested 

 have not yet been removed. 



To obtain the true rock temperature differences, mean^ were , 



taken to cement the thermopile points to the rock with plaster, 1 



a course it would be desirable to adopt with as fcw samples as i 



possible as standards of correction for the rest, on account of the 

 tediousnessot the process and the injury that it necessarily entails 

 to the beautifully worked surfaces of many of the plates. If 

 the correction so found to be required can be restricted by the 

 mode of operating to a range of such small limits as to be appli- 

 cable generally, wiihout appreciable influence of the surface cha- 

 racters, in making its occasional departures from a mean value 

 very sensible, then the reduction factor found by absolute exferi- 

 ments on a few rocks of characteristically rough and smooth or 

 polished surfaces to obtain the true temperature difterence for a 

 given heat-flo»v from the apparent one shown by the thermo- 

 couple placed simply between the rock and leather faces will be 

 admissible within the limits of error of the observations to con- 

 vert a list of apparent conductivities as just supposed to be ob- 

 tained from a mere comparative table of relative conducting 

 powers to a table of absolute thermal conductivities, in which 

 the errors of the valuts given will certainly not be greater than 

 would in all probability have been committed had the direct 

 method of absolute measurement been applied separately to each 

 specimen of the list, instead of only to a few rocks, to furnish 

 data on which calculations of the remainder may be founded. 

 Circular discs of linen, well wetted with plaster of Paris, mixed 

 with a little glue or white of egg, w-ere laid over the surfaces of 

 two or three of the rocks, enclosing under them and against the 

 rock (to which they were also plaster-wetted) the two branches 

 of the thermopile pincette. When these had set quite haid 

 under pressure and were thoroughly dried by a gentle heat, they 

 were placed in the apparatus, and a measurement of the abso- 

 lute tempera'ure difference and accompanying heat-flow was thus 

 ob'ained, affording the real conductivity and a means of com- 

 paring it with the apparent one found by similar observations of 

 the same rock whirn no planter was used, and when the points of 

 the thermopile merely pressed against its surface. Thus the 

 thermo-electric difference obtained with the wire couples merely 

 touching the surfaces of white statuary marble between velvet 

 faces was 16°, while for the same heat-flow when the arms 

 of the thermopile were firmly plastered to the marble plate, 

 the temperature difference observed was only l6°'2* — being 

 more than i\ times as large a difference in the former 

 as in the latter case. With whinstone the corresponding 

 temperature dilTerences were 26° and 8°'5. — in the propor- 

 tion of very nearly 3 ; i. A similar experiment was made 

 with cannel coal, of which the conductivity is much less than 

 that of the last-mentioned rocks, the temperature differences 

 obtained being for the same heat-flow in the plain and plastered 

 plate 53°'4 and 39°7 ; in the proportion of only I -37 : i— a far 

 smaller reduction than was observed in the two foregoing cases. 

 Care is, however, necessary to introduce wet plaster under as 

 well as over the points of the thermopile in cementing them to 

 the rock, that air may be excluded and the junction may be solid, 

 a precaution which was omitted in this case, as plaster without 

 size was used, which in drying sometimes flakes ofl' |fiom the 



* The he.it-flow through the plate was actually grenlcr in this latter thon 

 in the former case in the proportion of about 5 : 4, showing that the rough 

 plaster-washed linen surface received and delivered heat to the velvet 

 covers much more readily th.m the smoothly-dressed surface of the stone, 

 and rhe whole r.sist.ance was less in the latter than in the former case, al- 

 thougli the rock plate itself had been made thicker. The same diminution 

 of the total resistance occurred also in the experiment with plastered whin- 

 stone. 



