RicHarpson—Lines of Flow of Water in Saturated Soils. 309 
length to breadth, nor corners that are quite square. The former 
irregularity has been estimated by measuring the length and breadth 
of a number of chequers—say, a dozen—and finding the variability 
of the ratio in a rough way by observing what deviation from the 
mean has 2 of the observed deviations less than it, and } greater. 
This corresponds in a very rough way to the “ standard deviation ” 
of the ratio from its mean, and is recorded on the diagrams in the 
form 
reciprocal of chequer ratio = 1:18+°13, orelse 8S. D. =:18. 
But the effect of this error, and of the corners not being quite 
square, upon the value of = needs investigation. They must 
largely average out. I feel confident that the error is less than 
7 per cent. of ——, probably about 3 per cent. One’s best guide 
I gp 
is to attempt to improve the drawings, and to observe the con- 
y 
sequent changes in rie The value of Ge found is recorded on 
each diagram, so that the diagram may be applied to any 
uniform soil by putting in the specific value of A, and the 
absolute size may be neglected. 
§ 5. Discussion oF ReEsutts. 
Figs. 5 and 6 represent similarly placed ditches. It is inte- 
resting to notice how Bids is related to the height of the surface 
Kyp . 
above the ditch-bottom. The ratio of slid in fig. 5 to that in 
Kgp 
"226 
fig. 6 is 068 = 3°3, while the ratio of the heights runs as follows :— 
i Sue aes 
2 13 13 13 13 
mle 
cl © 
ol 
Ratio of heights, | 4:0 | 2°80 | 2°78 | 2°85 | 2°85 | 2:90 _ 2°80 
The ratio of the heights is remarkably constant about a mean of 
2°82. The reason why this ratio is less than 3:3 must be because, 
