June 22, 1905] 



NA TURE 



179 



the two curves thus indicate that either curve may 

 be taken to represent the flow changes. 



In the above mentioned reports the curve represent- 

 ing- tlie rainfall of the river basin from 1883 

 (curve iv.) was deduced from the statistics of twelve 

 stations covering a comparatively small area. It 

 happens, however, that these rainfall variations in 

 this valley are not restricted to this region alone, but 

 are similar to those which occur over a very large 

 area in the British Isles. By employing the Meteor- 

 ological Office records, and computing them accord- 

 ing to the present adopted system of grouping of 

 months, curves can be obtained which commence in 

 the year 1866. Investigation has shown that this 

 tvpe of variation is nearly common to 

 England S., Midland Counties, and 

 even the combination of Scotland E., 

 England N.E. and E., and the Mid- 

 lands, as can be gathered from the 

 curves in Fig. 2 (curves v., vi., 

 vii.). 



The other districts in the British 

 Isles (with the exception of Scotland 

 N., which is different from all other 

 districts in these isles) are of a type 

 similar to each other, but present 

 variations which, although not widely 

 different from the eastern and other 

 districts, are sufficiently unlike them 

 to be classified apart. 



The rainfall of the British Isles is 

 produced mainly by the passage of 

 areas of low pressure travelling over 

 the country in a north-easterly or 

 easterly direction. It should therefore 

 be expected that on the average the 

 greater the rainfall the more 

 numerous the cyclones, and conse- 

 quently the lower the mean value of 

 pressure. In the United Kingdom, 

 therefore, the rainfall variations from 

 year to year should correspond very 

 closely with the inverted pressure 

 changes. That this is so can be seen 

 by comparing the inverted Oxford 

 pressure curve in Fig. 2 (curve viii.) 

 with the rainfall curves underneath. 

 Instead of Oxford, any other town 

 in the United Kingdom, such as 

 Armagh, might have been taken 

 (curve ix.), for the pressure changes 

 are so remarkably similar over a very 

 wide area. 



It will thus be seen that the 

 pressure, rainfall, and river flow are 

 all intimately related, and any method 

 of forecasting pressure would make 

 it possible to determine beforehand 

 the rainfall. Since the Thames 

 flow has a lag of five months on both rainfall and 

 pressure, a means is possibly available of stating the 

 " expectancy " of excessive or deficient amount of 

 water in the river. 



It may here be pointed out that it does not seem 

 necessary to collect and discuss the data over the 

 whole of this region before any deduction for practical 

 purposes can be made regarding the flow of the 

 Thames. The barometric and rainfall observations 

 made at the Radcliffe Observatory, Oxford, exhibit 

 variations from i860 up to the present time so very 

 similar to those of the Thames flow that all three 

 curves are very ne.-Jrly interchangeable. 



So striking is the agreement that they are here 

 reproduced (Fig. 3). 



NO. i860, VOL. 72] 



The question now arises, can British pressure be 

 forecasted ? 



It has been previously pointed out in this Journal 

 (vol. Ixx. p. 177, June, 1904) that there exists a world- 

 wide barometric see-saw between two nearly anti- 

 podal parts of the earth, the one region about India 

 and its neighbourhood behaving in an inverse way to 

 that of South .\merica and the southern parts of_ the 

 United States. In some regions, and the British 

 Isles was one of them, the pressure variation curves 

 were found to be a distinct mixture of both the Indian 

 and South .American types, and it was difficult to 

 classifv them. 



To illustrate this, the accompanying figure is given 



SURBITDN 



3 YEAR 

 PEf?IOD 



I86OO 18700 



■ves illustrating the s 



(curves inverted) in t 



:present the epochs of su 



nilarity between the 

 = British Isles. [Th. 

 spot maxima and 



imes flow, and rainfall and 

 itinuous and doited vertical 

 respectively.] 



(Fig. 4). The upper curve represents an hypothetical 

 curve with a period of 3-8 vears, and beneath it the 

 South .American (Cordoba) pressure curve. At the 

 very bottom is given the inverse of this hypothetical 

 curve, and above this the Indian (Bombay) curve. 

 Between the Bombav and the Cordoba curves is given 

 that of Oxford. It' will be noted that the Cordoba 

 curve disagrees with its hypothetical curve in the 

 years 1892 and 1900 to 1903, while the Bombay curve 

 shows anomalies in 1892 and 1901 to 1903. 



If, now, the Oxford pressure be compared with 

 those of Cordoba and India, and a list made showing 

 the years in which high pressure at Oxford coincides 

 with years of high pressure at Cordoba or India, or 

 low pressure at Oxford with low pressure at Cordoba 



