^6 JOURNAL OF THE [April, 



spots and come upon a grating running diagonally across the 

 valve — as in slides Nos. 3, 6, and 8 — and it will be seen that the 

 holes of the diagonal grating lie immediately under the larger 

 interspaces left between the fibrils. I believe these round white 

 patches to be only ghosts, or, in other words, images of the inner 

 layer thrown on the outer as on a screen, in proof of which I sub- 

 mit slides Nos. 7 and 8. In No. 7 the fibrils are distinctly seen 

 running parallel to the median line, but only some of the inter- 

 spaces are filled up, and on focussing down you discover the 

 cause. In the under layer the holes are irregular in size, and it 

 is only from the larger ones that the images are thrown. The 

 outer layer often appears as if formed of definite squares, but I 

 have no reason to believe there are any actual cross-bars, for 

 directly the torn fibrils begin to diverge from each other, as in 

 parts of slide No. 1 1 already referred to, cross-lines disappear. 



There is nothing to say, that I know of, against the idea of a regu- 

 lar grating; but when we get positive evidence to the contrary, a 

 merely reasonable interpretation of appearances must give way to 

 actual fact. It may be easily seen, however, on referring again 

 to Fig. 4, how an appearance of cross-lines may be brought 

 about. 



Having done my best to establish the nature of the structure 

 of the different layers of P. formosum, it remains to see how far 

 the structure of the valves of the minute forms of this genus are 

 identical with it. Similarity of form in itself should almost be 

 enough to establish a unity of structure, and it satisfied myself a 

 long time before I had obtained positive evidence to support it. 

 But proof to one's self does not mean proof to others, and it was 

 necessary to complete the evidence by offering sufficient examples 

 of torn structure from each typical species to prove the identity. 

 But before going on to this I would mention a few salient points 

 common to all the Fleurosigma, and which hitherto seem to have 



