on January 6, 1892, and most carefully photographed by Mr. Lafayette, and made into lantern slides.
The bottles were placed near a window in a room in the building of the Leeds Philosophical Society, i. e., quite in the center of Leeds. The materials tested were: canvas; bunting; ordinary flannel; domette flannel, rough side in; domette flannel, rough side out; cotton-wool, one inch thick.
The results of the experiments show that as a consequence of eight months' exposure, including a week of the worst fog I ever knew in Leeds, three of the filtering tissues admitted a very appreciable amount of dust, viz., coarse canvas the most, bunting coming second, ordinary flannel admitting less than either. The other three bottles were screened, one with thick domette rough side in, one with domette rough side out, and one with cottonwool about an inch in thickness. The last three show hardly a trace of dust. Curiously, the cotton-wool shows a trace more than the domette flannel. The explanation of this I suspect to be that the cotton-wool was not tied firmly enough round the neck of the bottle, which had no rim, and that some air passed between the bottle and the wool, instead of through the wool.
Another experiment which I tried was to fit up a cupboard with panels of double domette flannel. After the fog, to my surprise, the inner screen had become more or less black, showing that black particles had passed into the cupboard, but with this remarkable difference: whereas the outer flannel was almost uniformly black from top to bottom, the inner flannel was divided into four squares of different shades of blackness, corresponding to four spaces between shelves. Of these four, the lowermost was almost as black as the outside, and the uppermost was almost clean. I just mention this as a fact which needs an explanation, but without suggesting one.
There is one error which I think has been committed in the screens made for me, and it was pointed out by my friend Mr. White, the architect, of Wimpole Street. The filtering material is likely to act more effectively if left loose and not stretched tight, as when tense the interstices are stretched and made larger, and when out of sight it might be very loose, almost baggy, with advantage.
Hoping to get some hints as to the comparative value of the various textures under trial, I placed specimens of each under the microscope. It is obvious that both canvas and bunting are of too open a texture, having numerous small holes unguarded by delicate fibers. Judging by the microscope, one would conclude that of woven textures, probably flannel, and still more, domette flannel, are the best, and this judgment seems to be borne out by the experiments with the bottles.
