RAINFALL IN DORSET. 187 



The average rainfall for the past 57 years is 33' 843 inches, 

 whereas at the end of 1903 it was 33" 753 inches. This 

 raising of the average has the effect of lowering the ratio of 

 each year to the average, as will also be seen from Table 5, 

 the figures in brackets being the ratios previously given. 



I was unable to go back farther than 1898, as I have no 

 records beyond this year ; but this information can be 

 obtained, I think, from the appendix to that year's report. 



August was by far the wettest month in the year, and will, 

 I think, be long remembered by most of us. 



The average fall during that month was 7' 28 inches on 27 

 days ; at some stations rain fell on 30 days. 



April was by far the driest month, and at three stations 

 no measurable rain fell. Taking the average, namely, '11, 

 it proved to be the driest month since February, 1891, when 

 '04 was the average of 34 stations. The wettest day generally 

 appears to have been the 29th September, as it is so recorded 

 at 36 stations, followed by the 16th January at 10 stations 

 and the 17th August at six stations. 



The 2 - 68 inches recorded at Swanage on the 29th September 

 is the greatest rainfall recorded by any observer throughout 

 the county, the observers at Parkstone and Branksome record- 

 ing 2 - 38 inches and 2*20 inches respectively on the same day. 



The maximum number of wet days was recorded at 

 Broadstone, namely, 269 ; and the minimum of 138 days at 

 Fleet House, Chickerell. 



One inch, or more, was recorded on seven days at three 

 stations, six days at three, five days at five, four days at 

 seven, three days at 22, two days at 15, and one day at 11, 

 whilst at two stations the rainfall appears not to have 

 amounted to one inch on any day. 



Referring to Table 3 we find that the month of May is the 

 driest month in the year 011 the average of the past 57 years, 

 and October the wettest month, the proportionate fall for the 

 two months being 58' 5 and 123 respectively. 



In a large proportion of the returns sent in I have found 

 errors which have simply been caused through incorrect 



