320 Professor Marshall T Yard, On the Question, etc. 
and below, by several methods ; arguing that — since each method 
was consistently applied to every species examined — the results 
obtained by one would serve to check those obtained by another, 
and so lead to results of considerable approximate accuracy. 
The methods employed for obtaining the numbers of stomata 
per one square millimetre of leaf surface were as follows : 
Direct counting of the numbers found in the field of view, the 
area of which was known 1 . Both surfaces were examined, and in 
each case an average of ten observations was recorded. Pre- 
liminary countings were made on leaves decolourised in alcohol, 
and rendered more translucent by warming in Eau de javelle, 
and transferred through the alcohols to glycerine and glycerine 
jelly ; on leaves hardened in weak Flemmings solution and trans- 
ferred to balsam ; and on strips of the epidermis. The latter are 
difficult to obtain and yielded very unsatisfactory results. 
In the end I decided that the best results were obtained by 
transferring the pieces of leaf, hardened as before in absolute 
alcohol, through alcohol and glycerine to glycerine jelly. 
Having obtained the average numbers in the circular field of 
900 ya diameter (column 5) it was of course a mere sum in pro- 
portion to calculate the number per one square millimetre of 
leaf surface (column 6) — (see Table Y T .). It will be noted that 
throughout the table A = Above ; B = Below. 
I also tried the following method : 
Since the stomata are always disposed in longitudinal rows 
(column 2), with their long axes coinciding with that of their row, 
measurements of the average intervals (longitudinally) between 
the stomata (column 3) and of the distances (laterally) between 
the rows, and the measurements of the total breadth of the leaf 
(collected in a separate table) gave data which could be used to 
calculate the number per square millimetre, and these results 
(given in column 7) also afforded useful check-results, though in 
almost all cases the figures — derived from averages of twenty 
countings for each surface — are by this method a trifle too low. 
Another method was to compare the numbers of stomata on 
the upper and lower surfaces of transverse sections of the leaves 
(column 4). Averages of ten counts each gave useful results, and 
possessed the advantage that each surface was compared on the 
same section. The drawbacks to this method are two, at least. 
In the first place it was found impracticable to obtain the uni- 
formity of thickness and serial sequence of section 2 necessary to 
superpose them and so get the numbers per square millimetre of 
leaf surface direct. And, secondly, I frequently found that closely - 
1 With Zeiss Occ. 2 x Obj. C the diameter of the circular field was 900 /r. 
2 Ribbon sections were not used because I was obtaining averages from different 
parts of the area. 
