1901.] 



on Polish. 



569 



smaller ones being, as it were, eaten up by the bigger. The division 

 line? between the cells are ridges, raised above the general level, and 

 when seen in good focus appear absolutely sharp. The general 

 surface within the cells shows no structure, being as invisible as if 

 highly polished. 



That each cell is in fact a concave lens, forming a separate image 

 of the source of light, is shown by slightly screwing out the object- 

 glass. Fig. 4 was taken in this way from the same surface, the 

 source of light being the flame of a paraffin lamp, in front of which 

 was placed a cross cut from sheet-metal. 



The movement required to pass from the ridge to the image of 

 the source, equal to the focal length (/) of the lens, may be utilised 

 to determine the depth (t) of a cell. In one experiment the necessary 



Fig. 5. 



movement was ■ 005 inch. The semi-aperture (y) of the " lens " was 

 •0015 inch, whence by the formula y 2 — ft, we find t = "00045 inch. 

 This represents the depth of the cell, and it amounts to about 8 wave- 

 lengths of yellow light. 



The action of the acid seems to be readily explained if we make 

 the very natural supposition that it eats in everywhere, at a fixed rate, 

 normally to the actual surface. If the amount of the normal corro- 

 sion after a proposed time be known, the new surface can be con- 

 structed as the " envelope " of spheres having the radius in question 

 and centres distributed over the old surface. Ultimately, the new 

 surface becomes identified with a series of spherical segments having 

 their centres at the deeper pits of the original surface. The construc- 

 tion is easily illustrated in the case of two dimensions. In the figure 



