of Edinhurgli, Session 1882-83. 
51 
basis of the next prediction. At first it was only necessary to use 
the lines h and F for comparison, since D + tt was very rarely < h dur- 
ing the months of June, July, August, and September. It was soon 
observed that rain seldom fell within twelve hours when the D-line 
at 9 A.M. appeared less than or equal to &, and that rain almost 
always followed when it appeared equal to or greater than F. . 
Intermediate* intensities were much more common, and in order to 
make predictions in these cases the degree of visibility of the thin 
lines in the green was taken into account. From the daily observa- 
tions in the month of June the following table for predicting was 
drawn up : — 
D + TT 
= or<5 >6<F = or >F 
I Thin lines | 
No rain | Eain 
. I Ti. 
Distinct Indistinct 
I I 
Probably Probably 
No rain Eain 
This table held closely during July. In that month the intensity 
was =or>F on fourteen occasions, and rain followed on thir- 
teen of these. The intensity was > 6 < F with thin lines indistinct 
seven times, and rain followed six times ; on six days the intensity 
was the same, but the thin lines were distinct, and rain (in each 
case only a slight shower) followed on three. Once the intensity 
was = 5, and it was followed by a slight shower. The table was also 
applicable, though less accurately, to August, September, and Oc- 
tober, but in November its predictive power completely broke down. 
The intensity was — or<& twenty-four times in that month, and 
rain followed on fourteen of these days. The probable reason of 
this is the low temperature which prevailed, in consequence of 
which a small quantity of water-vapour produced saturation, which 
was thus indicated by a fainter rainband than in previous months. 
A new table of predictions required to be framed for the winter 
months, in which the intensities D -f tt > E < 5, and = E were con- 
sidered. 
The following table (Table I) shows the number of times which 
