4^6 



NATURE 



\Marck 13, 1879 



tremely rare media light bodies are apparently repelled 

 by hot and attracted by cold bodies. At first this was 

 not recognised as a gaseous phenomena— in fact the non- 

 presence of gas was supposed to be essential to the 

 occurrence of the highest form of the action. 



That such motions might result from the inequalities 

 in the pressure of the residual gas caused by the com- 

 munication of heat between the objects and the gas, was 

 first shown by the author in May, 1874 {Proc. Roy. Soc. 

 1874, p. 402). 



It was shown that when heat is passing from a surface 

 to a gas whether by direct communication or by evapo- 

 ration the reaction from the receding molecules causes an 

 excess of pressure proportional to the heat communicated. 



The reasoning was definite so far as it went and, 

 although manifestly incomplete, the conclusion arrived at, 

 viz., that the phenomena resulted from the heat com- 

 municated to the residual gas appears very soon to have 

 been generally accepted as it was found to be verified in 

 various ways. Several papers appeared in which attempts 

 were made to render the explanation more complete, but 

 these attempts were for the most part based on a mis- 

 conception of the phenomena to be explained and were 

 altogether wide of the mark. 



In November of 1878 the clue to the step that was 

 wanting to complete the explanation occurred to the 

 author. It was then seen that although the surfaces with 

 which the phenomena occurred were of limited extent, 

 no account had been taken of this fact in the attempts to 

 explain the actions. Having once hit on this point the 

 deduction of a complete theory of the phenomena was 

 only a matter of labour. It was found that although the 

 excess of pressure is proportional to the quantity of heat 

 communicated, it also contains a factor which is pro- 

 portional to the divergence of the lines of heat flow ; 

 and hence the reciprocal of the density of the gas at 

 which the phenomenon would occur should vary inversely 

 .as the size of the vanes. So that by using vanes of 

 coaaparatively small size the phenomena'should be obtained 

 at proportionally greater densities of the gas. 



On considering how this conclusion might be experi- 

 mentally tested it appeared that in order to obtain any 

 results at measurable pressures the vanes would have to 

 be very small indeed, too small almost to admit of 

 experiment. It was while searching for some means to 

 overcome this difficulty that it became apparent that if 

 the vanes were fixed, then, instead of the movement of 

 the vanes, we should have the gas moving past the vanes 

 — a sort of inverse phenomenon — and then instead of 

 small vanes small spaces might be allowed for the gas to 

 paris. Thus the probable existence of phenomena of 

 Thermal Transpiration was suggested, and it was 

 obvious that the porous plug would furnish the means of 

 verifying the conclusions. This probable connection 

 ■between the phenomena of the radiometer, which may be 

 -called Impulsion, and the phenomena of transpiration 

 through porous plugs raised the question whether the 

 same extension of the dynamical theory of gas which 

 explained the radiometer would not include the results 

 obtained by Graham not then explained. This idea was 

 followed up, and a method was devised of extending the 

 dynamical theory of gases so as to take into account the 

 forces tangential, and normal, ;^arising from] a varying 

 condition of molecular gas. 



This theory appeared to explain fully all the results 

 established by Graham as well as the then known phe- 

 nomena of impulsion (the radiometer), besides definitely 

 indicating the phenomena of thermal transpiration to be 

 expected, as well as the effect of employing small vanes 

 in the radiometer. That this step had been accomplished 

 was intimated by the author in the following passage in a 

 letter published in Nature, vol. xix. p. 220. 



" N ow, however, I have arrived at a result, which, 

 although somewhat unexpected and striking, will, I hope. 



be found to reconcile what has hitherto appeared to be 

 anomalous in the phenomena already known, and to have 

 suggested certain hitherto unexpected phenomena which 

 now only await experimental verification." 



This experimental investigation which was at once 

 commenced, proved to be a much more serious under- 

 taking than had been anticipated, and was not completed 

 until the end of August, this was not so much on account 

 of difficulties, although these were considerable, but be- 

 cause it was found possible to do so much more than had 

 been expected. 



One of the results of the investigation was to show 

 that a diff'erence of pressure is maintained whenever two 

 chambers of the same gas are separated by a porous plate, 

 one face of which is hotter than the other. 



With a plate of meerschaum '44 inch thick, one side 

 of which was something less than 212°, while the other 

 side was about 50", the difference of pressure was "25 inch 

 of mercury with air at the atmospheric pressure, and "SB 

 of an inch with hydrogen at the same pressure. 



The existence of this thermal transpiration, al- 

 though a new phenomenon, is not considered to be the 

 most important result of this part of the investigation, 

 for it appeared on comparing the results' obtained with 

 different plates, and different densities of gas, that there 

 was a constant relation between the pressures for different 

 plates at which proportional results were obtained. Thus 

 comparing a plate of stucco with a plate of meerschaum, 

 it was found that the ratios of the difference of pressure 

 to the mean pressure, was the same for both plates so long 

 as the mean pressure with the meerschaum was six times 

 greater than with the stucco, and that this law held 

 although the mean pressures with the meerschaum ranged 

 from 30 to I "25 inches of mercury. This ratio of the 

 pressures at which corresponding results were obtained, 

 was the same whether the gas used were hydrogen or air ; 

 and for plates of different thickness, and hence it was 

 clearly shown to depend only on the coarseness of the 

 plates. 



Thus in thermal transpiration we have a phenomenon 

 of gaseous motion depending on a relation between the 

 density of the gas and the dimensions of the space 

 which it is constrained to occupy. The discovery of this 

 relation between the density of the gases and the coarse- 

 ness of the plates at which corresponding results of 

 thermal transpiration were obtained, suggested the possi- 

 bility of obtaining a like relation for corresponding 

 results when the gases were forced through the plates 

 by a difference of pressure. Graham had found that the 

 relative rates at which different gases transpired through 

 plates differed very considerably with the coarseness of 

 the plates and no explanation had been given of the 

 phenomena. On trying the experiments not only was it 

 found that with plates of different coarseness correspond- 

 ing results were obtained whenever the pressures of the 

 gas have a constant ratio but it was also found that with 

 the same plates the ratio was the same as in the case of 

 thermal transpiration. 



A successful attempt was then made to verify the 

 conclusion that the phenomena of the radiometer might 

 be obtained at higher densities by using smaller vanes. 

 By suspending fibres of silk and spider lines the repulsive 

 action of heat was rendered apparent at pressures 

 ranging up to the pressure of the atmosphere. 



These results, as well as the theory from which they 

 were deduced, have been fully described in a paper, an 

 abstract of which was read before the Royal Society on 

 the 6th inst. 



As regards transpiration and impulsion, the investiga- 

 tion appears to be complete ; most, if not all, of the 

 phenomena previously known have been shown to be 

 such as must result from the tangential and normal 

 stresses consequent on a varying condition of a mole- 

 cularly constituted gas ; while the previously unsuspected 



