496 
Philippine Journal of Science 
1920 
of letters — denoting uncertainty in the data — are treated as 
belonging to the medium group. 
It seems undesirable, considering the discrepancies as to de- 
tails, to attempt to combine these four diagrams into a single 
one, but the following remarks will draw attention to the main 
points on which such a combination might be based. We may 
begin by considering the two total concentrations separately, as 
to the degree of agreement between the data for tops and those 
for roots. For the lower of the two medium concentrations 
KHjPO, 
Fig. 4. Top-yield values for the thirty-iive different sets of salt proportions, with low-medium 
total concentration. 
