510 Philippine Journal of Science 1920 
corresponding green weight, as both approached the zero value 
where all growth was inhibited. 
Turning next to the relation between the discrepancies shown 
in Table 12 and the various sets of salt proportions, it may be 
remarked that there are just three cases in which a discrepancy 
is recorded for two or more total concentrations of the same 
set of salt proportions. One of these cases refers to solution 
TiR3S1, for which disagreement between the two criteria is 
shown for the lowest three total concentrations, but not for the 
highest. The second case refers to solution T2R1S3, for which 
disagreement is recorded for the two medium total concentra- 
tions. The third case refers to solution T4R1S1, for which 
there is disagreement for the lowest and for the highest total 
concentration. Altogether, it appears that there is no evident 
relation between salt proportions, which supports the generaliza- 
tion stated on page 508, to the effect that these two criteria 
generally agree in their indications, and that (for the present 
study) one is to be regarded as about equivalent to the other. 
Considering only the two medium total concentrations, the 
distribution of the low, medium, and high values of the green 
weight of tops, with reference to salt proportions, may be 
brought out diagrammatically as in the case of the dry yields. 
This is done by fig. 8, which is constructed in a manner similar to 
that followed for figs. 4 to 7. The letter designations shown for 
green weight and for the two medium total concentrations (series 
16 and 19) in Table 12 have been placed on the tetrahedral 
diagram (HL being replaced by M, as for the dry-yield diagrams) 
and the high and low areas have been demarked and distin- 
guished by shading. For this particular diagram the HM-points 
are included in the high regions, and the LM-points are included 
in the low regions. 
From this diagram it appears at once that this sort of sum- 
mary of the green-weight data leads to the same general con- 
clusions as were reached from the study of the dry-yield data. 
It appears that the most promising set of salt proportions on 
this basis is T1R1S1, and the least promising ones are T2R1S4, 
T3R1S3, T4R1S2, and T5R1S1 (the top apex). It seems safe, 
however, to allow the general conclusions derived from dry 
yields to stand as they are, and merely to remark that the 
points there emphasized are supported and substantiated by the 
indications derived from green weight of tops. 
It is clear from what precedes that the thirty-five sets of salt 
