322 Professor Marshall Ward, On the Question of 



packed stomata were cut through twice, or two stomata on the 

 same row occurred in the thickness of a section, or a stoma was 

 merely shaved by the razor, and such were apt to be missed in 

 the counting, and so on. The numbers thus obtained were only of 

 use in establishing generally the fact that those above are more 

 numerous than those below. 



On the whole, useful though the above methods were for 

 checking results, there were considerable divergencies, and the 

 best way was found to be that of carefully counting the stomata 

 on a given area somewhat larger than one square millimetre, and 

 plotting them out on squared paper. On then placing over the 

 chart a glass plate, covered all over except a square repre- 

 senting, on the same scale as the paper, one square millimetre, it 

 was easy to count the stomata in five positions of the square and 

 obtain more correct results (see Table IV.). 



I may add that I checked this last method, so far as B. mollis 

 is concerned, by repeating it on a seedling germinated in December, 

 and at a much lower temperature and under very different con- 

 ditions as regards illumination, &c. On the upper surface the 

 number of stomata, 35 per square millimetre, was the same as 

 before, though that of the hairs was fewer than the average. 



It only remains to add — so far as Table V. is concerned — that 

 the measurements of the lengths of the stomata (column 1) were 

 obtained from 10 measurements on each surface, and that I have 

 given the extremes as well as the commonest lengths. The 

 variations in breadth were too small to measure accurately with 

 medium powers, and I concluded that as all the stomata appeared 

 shut (except in the transverse sections) no data of value could be 

 obtained from them. 



The next part of the plan adopted was as follows : 



Assuming that the number of experimental infections — 1846 

 were employed — is large enough to warrant the procedure, I con- 

 struct a curve of percentages of successful infections for each 

 species with each kind of spore (see Table III.). Thus, with the 

 spores obtained from B. mollis, B. mollis gave 706 °/ , B. velutinus 

 56'4 7 > & secalinus 50*7 % °f successful infections, and so on. By 

 joining the upper ends of ordinates proportional to these percent- 

 ages, a diagram-curve is obtained which we may call the infection- 

 curve. See Table VI. 



I then take the measurements of — say the sizes of the stomata 

 — and construct a diagram-curve in similar manner of each of the 

 measurements — extremes, commonest sizes, above and below — by 

 joining the upper ends of ordinates proportional to the lengths, 

 and taking*the species in like order. 



It appeared that in this way it ought to be possible to express 

 graphically any correspondences or discrepancies between the 



