Lyons — The Distrihution of Mean Annual Rainfall. 359 



give very consisteut aud satisfactory results for records of even a few years' 

 duration, especially when the fundamental stations with which they are 

 compared are not far distant. The assumption underlying the above 

 method is that, for stations within a limited area, the rainfall varies 

 similarly from year to year. This assumption is seen to be justified by 

 plotting on the same sheet the annual I'ainfall for such stations for a number 

 of consecutive years. Such sheets were prepared respectively for the stations 

 in each of the following five divisions — North Dublin, South Dublin, Central 

 Dublin, Wicklow, Meath, and Kildare. The graphs showed a marked 

 parallelism for stations close togetlier; and even for stations far apart and very 

 differently conditioned, there is found a similarity in the run of the annual 

 rainfall graphs. This is illustrated in the diagram, page 360, where the annual 

 rainfall graphs are given for a few selected stations for the period 1885 to 

 1905. 



In reducing to a 35-year mean the records of a short-period station, the 

 computation was independently made by reference to at least three long 

 records. These long records included one, and, if possible, more than one, of 

 the 35-year stations, or otherwise the longest records among neighbouring 

 stations. The greatest care and circumspection were used in selecting such 

 stations of reference, and special attention was paid to similarity of conditions 

 of the stations compared, and to the correspondence observable in their 

 annual rainfall graphs. In one or two cases where a value for an individual 

 year was missing in an otherwise complete record the value was estimated by 

 interpolation, or by careful consideration of correspondence with tlie graphs 

 of other stations. In only a few cases did any great discrepancy appear in 

 the means computed by reference to different fundamental stations ; and the 

 results in such cases were most carefully considered. 



Sources of Error and Accuracy attainable in Rainfall Measurement. 

 The values of mean annual rainfall submitted in the accompanying tables 

 are the results of computation based on actual returns, and no correction of 

 any kind has been attempted. The "catch" of a rain-gauge is subject to 

 several disturbing factors, nearly all connected with wind eifect, and 

 dependent on conditions of exposure. There exist in consequence many aud 

 serious sources of error in the measurement of rainfall ; but the effects are so 

 irregular that attempts at correction are necessarily purely speculative, and in 

 most cases injudicious. The wind by forming eddies and currents at the 

 mouth of the gauge tends to reduce the amount of rain caught. As the 

 height of the gauge above the ground is increased, the wiad and the above 

 effects are increased, aud the gauge will usually register less than if placed at 



