TRANSACTIONS OF SECTION A. 913 
seventy years. In order to make practical use of this large mass of material, it 
was arranged in the following manner :—Taking out first all the Januaries, then the 
Februaries, and so on, [ arranged each month, not in chronological order, but in 
the order, first, of its mean temperature, and afterwards of its rainfall. The same 
thing was also done for each meteorological season (winter consisting of the three 
months, December to February, and the other seasons in order), 
In accordance with the principle which I had the pleasure of explaining to the 
British Association in 1879, and which is described at page 277 of the Report for 
that year, I divided each of these arranged lists as nearly as possible into five equal 
sections. Those referring to temperature are termed respectively, very cold, cold, 
average, warm, and very warm, and those referring to rainfall are named, very 
dry, dry, average, wet, very wet. We have thus a fair division of this long series 
of years, as regards the important characters of warmth and moisture. This 
simple arrangement is of the greatest utility in enabling us to generalise upon the 
mass of facts that have accumulated to our hand. 
Taking out first all those months that were ‘very cold,’ there is no difficulty in 
writing down against them, in order, the character as regards mean temperature: 
and rainfall of the months that came next after, and in saying how many of these 
latter were very cold, how many were cold, and so on, through the several sections 
of temperature and rainfall. Then, to ascertain the influence of high temperature, 
or of very dry or rainy weather, we take out all those months that were very 
warm, or very dry, or very wet, and proceed in the same manner. Supposing, for 
example, that we wish to inquire in what way great warmth in January affects 
the following month. We take all the instances (fifteen in number) within the 
last seventy years in which January was ‘ very warm.’ Obviously there must be 
fifteen Februaries to be examined. We find that three of these were ‘ very cold,’ 
none were ‘cold,’ five were of ‘average’ temperature, two were ‘ warm,’ and five 
were ‘very warm.’ Therefore, as regards the temperature of February it is safest to 
say that the result is zndefinite, because the three very cold and the five very warm 
ones too nearly balance each other to enable us to assert that there is a distinct 
tendency toward either side. But as regards the rainfall of these same Februaries, 
we find that seven of them were very dry, one was dry, two were average, three 
more were wet, and the remaining two were very wet. These facts exhibit a 
strong tendency for a very warm January to be followed by a dry February. Out 
of fifteen cases, seven Februaries were very dry against two that were very wet. 
The probability, therefore, of a very dry rather than a very rainy February may 
be conveniently expressed by the formula—7 to 2 out of 15. 
Proceeding in the manner thus briefly indicated, it is remarkable what a 
harvest of interesting results promises to reward the inquirer. It is, indeed, sur- 
prising in how many cases the warmth or moisture of one month or season is 
discovered to be influenced by some unsuspected law of association with the month 
or season which it succeeds. 
Omitting all those results which are of an ambiguous or indefinite character, I 
would venture to state the following definite propositions :— 
1. A very cold spring tends to be followed by a cold and wet summer. The 
probability of a very cold summer is decidedly strong, being 6 to none out of 15,} 
and that for a very wet one is 5 to 1 out of 15. 
2. A very cold summer tends to be followed by a cold autumn, the probability 
being 6 to 1 out of 14. 
vil wery warm summers are prone to be succeeded by warm autumns, 5 to 1 
out of 14. 
So much for the influence of the seasons on those which come directly after 
them. Except in these three instances I perceive no definite law. Now let us 
see the apparent effect of one month upon another following it :— 
4, In seven out of the twelve months we find that very Jow temperatures tend 
to be prolonged into the succeeding months. Thus, a very cold January gives a 
likelihood of a cold February, the probability being 4 to 1 out of 14; a very cold 
' That is to say, of fifteen very cold springs, six were followed by very cold 
Summers, and none by a very warm one. 
* 1885. 3N 
