LOWEST TEMPERATURE OF JANUARY, 1838. 243 



Naturalist, some notes on the effects of this extreme cold upon vegetation ; at 

 present it is impossible to ascertain the injury done to many species whose time 

 of growth is not yet come. In my garden hundreds of evergreen shrubs appear 

 to be destroyed, unless the main stems or roots may be still alive. 



The paper in the Monthly Chronicle, which has called forth these remarks, is 

 occupied with observations on Mr. Murphy's predictions of the weather, which 

 are so frequently altogether erroneous. The fixing the day of lowest temperature 

 was certainly a very lucky guess, if a guess only ; but there is an explanation 

 which I have not seen referred to by his critics. Taking the average of a long 

 series of years, the coldest weather of London occurs in the three days following 

 the lS'h of January. Hence the probability that the lowest temperature may 

 ^ occur on the night between the 19th and 20th. The day of actual greatest cold 

 in single years is not uncommonly remote from the 20th of January ; although 

 in a series of years we shall find that it falls most frequently between the 13th 

 and 23rd of January. According to Howard on the Climate of London, first 

 edition, in the ten years following 1806, the lowest temperature occurred seven 

 ■times in January; and six of these seven times were between the 13th and 

 23rd ; the other being on the 29th. From this and other evidences I am disposed 

 to think that Mr. Murphy's prognostications are partly founded upon the em- 

 pirical averages from past observations ; that they are predictions a posteriori, 

 and not really derived from principles which enable him to determine weather 

 a priori. Most enthusiasts in science start from a few ascertained facts ; and 

 having imagined a theory in correspondence with these facts, they can never be 

 wholly destitute of coincidences in favour of their imaginary principles. 



The insertion of these remarks in The Naturalist will have two useful results ; 

 first, that of correcting an error published on influential authority ; and secondly, 

 it will show that prognostications of temperature may be made as probabilities, 

 without thereby proving the prognosticator to have discovered any new and cer- 

 tain principles for calculating this beforehand. 



Thames Ditton, Surrey, 

 March 4, 1838. 



