220 Messrs. Lodge and Clark on Dusty Air in the 



majority of cases we used a simple round rod, and its appear- 

 ance has been already described. A flat horizontal plate or 

 spade, a centimetre or less in width, shows a coat all over its 

 under surface, very thin coats at its edges where the currents 

 are violent, and a dark plane rising from a broad base, conti- 

 nuous with the dark coat on its upper surface. On a horizontal 

 flat copper plate, 6 centim. wide, the coat on the upper sur- 

 face is very thick, being little disturbed by currents. Its 

 outline is not perfectly sharp, and it hovers or wavers about, 

 some portions being thicker than the rest, and appearing 

 likely to form upward streamers, though they seldom do. 

 There is a certain amount of up-streaming from the coat, 

 mainly from near the edges. On perforating the plate with 

 a hole" of about 0*07 centim. in diameter, a thin line of du t- 

 laden air rises through the black coat on the under surface 

 of the plate, and emerges into the black coat on the upper 

 surface like a miniature volcano. If it emerge into dusty air 

 the small smoke-column is sometimes seen thinly edged with 

 black from having passed so near a solid. 



A vertical flat plate has good coats and a plane. With thin 

 mica or copper, '003 inch thick, the coats are rather thicker 

 near the top of the plate than near the bottom, and the dust 

 touches the plate near the bottom, as proved by a deposit of 

 dust which forms there. Blackening the surface increases the 

 thickness of the coats. 



A hemicylinder of sheet-copper was examined (a) with the 

 concave side turned towards the light, (b) turned upwards, 

 and (c) with the concave surface turned downwards. Its ap- 

 pearance in position a is shown in fig. 2; its interior is lined 

 with a good dark coat. The coat was first seen on this piece. 

 In position b & coat surrounds the surface, and from the two 

 upper edges rise dark planes. We illuminate this from above 

 by a 45° mirror. On the inner surface the definiteness of the 

 boundary of the coat is not so good. Position c gives a dark 

 plane from the top much as a round rod does ; and inside 

 also the dust-free coat is thick and well-marked. Wishing to 

 make the air inside more stagnant, we glazed in its ends with 

 mica, so as to stop even longitudinal currents, and now the 

 gravitational settling of the dust broadened the inside dark 

 coat ; but the coat narrowed to its customary width when con- 

 vection-currents were caused to circulate within the cavity. 

 The different thickness of coat on curved surfaces is well shown 

 by using the hemicylinder of glass, shown in fig. 4. The 

 coat is evidently thickest where gravitation assists it, and 

 thinnest where it opposes it. 



The behaviour of very thin films is curious. Vertical 



