CAN WEATHER EE PREDICTED IN THE BRITISH ISLES? 167 
•overdraining exerts a similarly pernicious effect to the clearing 
of woodland. The upper proprietors demand that the land 
shall be drained as perfectly as possible. Thereby the water 
which should have lain in the soil for months escapes at once, 
and causes the river level to rise at the lower part of its course, 
washing the unlucky residents on its banks out of house and 
home. 
This is, however, a digression, and as to the prediction of 
weather we may say that it has not at the present time been 
shown to be feasible to forecast weather for one short week, 
except on the principle, which affords us scanty consolation, 
that weather, when once well established, takes a long time to 
change. This comes out very clearly from the researches of 
M. Koppen, at present engaged at the Deutsche Seewarte at 
Hamburg. He has shown in a paper in vol. ii. of the “ Eussian 
Eepertorium fiir Meteorologie ” that if we investigate weather 
changes by the laws of probability we find that, as regards tem- 
perature, if a cold five-day period sets in after warm weather we 
may bet two to one that the next period will be cold too ; but if 
the cold has lasted for two months, we may bet nearly eight to 
one that the first five days of the ensuing month will be cold 
likewise. This does not mean that the chances are in favour of 
the weather never changing, but are only against its changing 
on a definite day, and increase with the length of time the 
existing weather has lasted. The problem is somewhat similar 
to that of human life : the chance of a baby a year old living 
another year is less than that of a man of thirty living to be 
thirty-one. 
There is one very interesting idea which, if it could be worked 
out, would be of great value to us, and that is that we could fore- 
see the coming of such a winter as we have just passed through 
if we had regular and early information as to the state of sea 
temperature off the coast of Portugal during the late autumn. 
Franklin, in 1776, traced the Grulf Stream right across the 
Atlantic in November, and found a temperature as much as 5 0, 5 
above the mean for the month at one spot in about 10° W. and 
45° N. Now in January 1822, H.M.S. Iphigenia , nearly on the 
same spot, found a temperature 3°*2 above the average in 
44° 30' N. ; the difference increased to 6° in 39° N., and again 
diminished to 4° in 32° 20' N., while at the same period the 
; general temperature in the adjoining parallels, both to the 
northward and the southward, even as far as the Cape Verde 
Islands in 19° 40', was colder by a degree and upwards than the 
usual average. We quote from Sabine’s a Pendulum and other 
Experiments,” p. 430. The same writer goes on to say: “Ncr 
is the probable meteorological influence undeserving of attention 
of so considerable an increase in the temperature of the surface 
