4i8 



NA TURL 



ypeb. 28, I J 



holding rouge in suspension, will) a daik i-ilane rising from it. It 

 is not always easy to obtain the dark coat in liquids, however, and 

 its thickness is enormously less than it is in air. Moreover, the 

 results are less definite and satisfactory. In gases the thickness nf 

 the coat may be anything below the eighth of an inch, accord- 

 ing to the circumstances of the case, but its commone.'t width 

 is more comparable with the hundredth of an inch. On a car- 

 bon rod in an electric be.im the coat is about h.ilf a millimetre 

 thick, no other heat being applied. Glass shows a perfect coat 

 and dark plane, but for some reason or other very thin films of 

 glass ('0003 inch thick) behave differently, and it is sometimes 

 by no means easy to see any coat at all. It may be that such 

 thin films are unable to absorb enough radiation, or it may lie 

 that the cause is more deeply seated. It can hardly be that they 

 give off their heat too rapidly, because the convection-currents 

 set up by Ihem are very sluggish. It is pretty certain that ihey 

 fail to absorb radiation ; for a plate of rock-salt in a perfectly 

 dry atmosphere behaves in the same way. In ordinary air, a lump 

 of rock-salt is able to absorb sufficient radiation to give a satis- 

 factory dark plane. The behaviour of thin films is under further 

 investigation. Covered with lampblack they act perfectly well. 

 Incidentally it has been noticed that films of freshly-blown glass 

 adhere together though cold, giving the black spot; hut ihat 

 when films are a day or two old they refuse to adhere, doubtless 

 because of the condensed air-sheets wiih which they have coated 

 themselves. The slow formation of these condensed .sheets, as 

 studied recently by Bunsen ( ;?'/></. Ann. February 1884) is of 

 great interest. 



The effect of electrifying rods from which a dark plane is 

 streaming is not marked except when the potential is high ; 

 100 volts or so produce a little effect, poiitive potential broad- 

 ening the coat, negative potential narrowing it. As soon .as a 

 brush discharge occurs, the effects are violent and the air is 

 rapidly cleared of smoke, the particles being deposited on all the 

 surfaces near. Various electrical phenomena can be conveniently 

 examined by means of smoky air aid a strong beam of light. 

 Thus a flake of mica, on being examined for its coat one day, 

 showed a curious phenomenon. The dust aggregated on its sur- 

 face in little bushes or trees, and its edge became fringed vith 

 long aggregations of dust particles. Our first thought was that 

 mica was photn-electric, but we now think that it h.ad been ]ier- 

 haps electrified by casual pressure. This also is still under 

 investigation. Tourmaline shows all its pyro-electric properties 

 exceedingly well by being simply illuminated in dus'.y air. If 

 mica be written on with a blunt point, a sheet of paper inter- 

 vening, the writing becomes manifest when it is exposed to dust. 

 We find, however, that a brass plate is capable of acting in a 

 similar way, and we are not prepared to be content with a mere 

 electrical explanation. We are probably here dealing with pheno- 

 mena allied to those known as Haifclibihlcr, which are supposed 

 to be connected with the conden-ed air-sheet on the surface of 

 solids ; though their explanation may also be associated with 

 vapour-condensing nuclei on solid surfaces. The phenomena 

 connected with the settling of dust on surfaces by gravitation 

 have also been investigated, and it is found that so long .as a 

 body is warmer than the air it keeps itself free from dust ; except 

 that just at the top, where the air is st.agnant, the excess of tem- 

 perature being only small, a large particle or two may drop on. 

 The dust-free coat is not an absolute barrier to dust: it m.arks 

 the region into which dust is not carried by [ovvation-currciits ; 

 but other causes may drive dust into this region. Thus it may 

 be blown into it either from outside or ihrough a hole in the rod 

 itself if it be hollow; or the rod may give off smoke, or the 

 dust may, as stated, occasionally drop into the dark region liy 

 common gravitation. The persistence of the dust-free plane at 

 a distance from the body which produced it is dependent on the 

 motion of the dust particles tc/.'/; the nir streamlines ; whatever 

 drives du-t across stream-lines interferes with and tends to 

 obliterate detached dust-free regions. All dark .streaks in smoky 

 air are commonly the wiped-off coats of bodies. 



We have been led to a fairly complete explanation of 

 the whole phenomenon, and though' it is impossible to attribute 

 every case of dust-freeness to one single well-defined cause, we 

 see reason to believe that the main causes in ordinary operation 

 are two, viz. : — 



1. Molecular bombardment. 



2. Gravitative settling. 



We were long under the impression that the sheets or films of 

 condensed ga.ses which are known to exist on the surface of all 



bodies v\ere connected with the dark coats, and had some share 

 in their production ; and this view was pressed home by an 

 observation of the surface of warm water in dusty air. The 

 evaporation of the water drives back the dust and keeps a clear 

 space of some thicluiess above the water ; and if the water be 

 linearly heated by a pla'inum wire stretched just beneath its 

 surface and warmed by a current, the dark coat streams upward 

 in a fine and well-defined dust-free plane. The up-streaming of 

 the portion aljove the wire causes the remainder to become 

 thinner, until there i- an evident equilibrium between the rate 

 at which the evaporation reproduces the dark coat and the rate 

 at which the convection current carries it off. That the dust 

 is kept off a solid, say a warm copper wire, by an evaporation 

 and continued renewal of its condensed air-sheet, we think 

 decidedly improbable, but v/e are convinced that the dust 

 particles are driven away from the solid by some form of molecu- 

 lar bombardment, possibly such as goes on from the vanes of a 

 t rookes' radiometer. There is, however, a very great difference 

 between the two phenomena : the Crookesian layer is supposed 

 to correspond with the mean free path, and this is enormously 

 less than the thickness of a dust-free coat. A possible sugges- 

 tion is that the dust-free coat represents something more like the 

 extreme free path of the molecules, the dust particles being so 

 easily moved that they are driven away by the blows of 

 even a few molecules. A simpler and more satisfactory mode of 

 puiting the matter is this. The temperature of the air near a 

 warm solid decreases gradually as we recede from its surface. 

 Consequently a dust particle in the neighbourhood of the solid 

 has warmer air on one side of it th.an on the other ; in other 

 words, it receives heavier and more numerous blows on one side 

 than on the other, and accordingly is driven away from the 

 warm body. Whenever the temperature of air is steadily dif 

 ferent in succes-ive layers, there the dust particles must get 

 driven in the direction of decreasing temperature at a rate de- 

 pending on the temperature slope. This is not complete, how- 

 ever, because the extra temperature really shows itself as a 

 diminution of density, not .as an increase of pressure. The ex- 

 planation is further elaborated in the paper, which will appear 

 in the Philosophical Magazine for April. The conduction of heat 

 across the air near a hot body is itself an interesting problem. 

 So also is the distribution of up-streaming velocities. The 

 maximum velocity of convection occurs at some distance from 

 the body, bting often distinctly outside the dust-free coat. 



Some few ca-es of the dust-f.ee coat and plane can hardly be ex- 

 plained in the manner now indicated. We point out, however, that 

 gravitation is an effective cause certainly in operation, which of 

 it elf is competent to account for the formation of dust-free 

 spaces when circumstances are favourable. Dust is always 

 settling or falling downsvard relatively to the air in which it is. 

 The velocity of relative fall depends on the size of the particle, 

 on the density and viscosity of the gas, but not on the motion of the 

 gas. Immediately I elow a solid body round which gentle currents 

 are rising, there is a small region of nearly stagnant air ; out of this 

 dust slowly falls, leaving it free, and if then part of it is dragged 

 round the body by the currents, it contributes to tlie dark coat 

 an<l to the ascending dark plane. Underneath horizontal plates, 

 also, this gravitation-settling of the dust assists and broadens the 

 coat. But there is also a coat on the upper surface ; and this 

 coat gravitation, so far from producing, does its best to spoil. 

 A few of the larger particles are in fact seen to fall occasionally 

 through it on to the surface of the body, their weight being too 

 great for the bombardment to sustain. In the case of a cold 

 body the down-streaming currents deposit a good deal of dust 

 on the upper surface of the body, and so that portion of air w hich 

 has grazed ihe surface passes on dust-free. The tendency now is 

 for the warmer air outside the dust to bombard it on to the cold 

 surface. This goes on all over the upper half of the body, but 

 over the lower half a coat is visible when the cold is not too 

 great, but it is only fairly thick at the bottom of the body when 

 it forms the base of the inverted dark plane. A smooth vertical 

 surface of ice gives no dark coat, and the maximum velocity of 

 the particles in the descending current is apparently little, if at 

 all, distant from the actual surface. Finally the writers call 

 attention to a paper just read at the Royal Society of Edin- 

 burgh by Mr. John Aitken. They have only seen an abstract of 

 tl»is paper at present, but it appears that Mr. Aitken has been 

 travelling over much of the same ground as they have, and that 

 he has arrived on the whole at fairly the same conclusions. 

 The abstract of Mr. Aitken's interesting paper was printed in 

 Nature of January 31 (p. 322). 



