32 EAINFALL AND FLOODS. 



In adopting the method I have suggested, it is necessaiy 

 to use some discretion in selecting the period to be taken 

 for the duration of the rainfall. In the case of an isolated 

 rainfall, such as that of March, 1889, there is little room for 

 discretion, but in a rainfall like that of last October the 

 case is different. If in that instance we had included in 

 the rainfall to be taken as the basis of calculation the rains 

 of the 13th, 14th, and 15th days, or those of the 20th and 

 21st, we should have increased the duration in a much 

 greater ratio than we should have increased the quantity, 

 thus reducing the average intensity of the rain, or the 

 average rate per hour, upon which the calculated rate of 

 discharge depends. 



The tables which I exhibit, apart from their primary 

 object, may serve to show the kind of rain which in the past 

 has occasioned floods in Bristol. Practically, any amount 

 of rain might have fallen in the course of a month without 

 flood if it had been pretty equally distributed. An inch 

 of rain in a day would not, even under former conditions, 

 have been a source of danger. Two inches in 24 hours, 

 or three inches in 48 hours, would generally have caused a 

 flood, and the flood would have been more serious in pro- 

 portion to the irregular distribution of the rain within that 

 interval of time. 



A word must be added with regard to certain rainfalls 

 which have considerably exceeded in intensity (not in 

 quantity) those that I have selected for comparison. In 

 July, 1875, 2 9 inches fell in 24 hours ; and in August, 

 1865, a still more remarkable fall occurred of 2j inches in 

 4 hours. It must be admitted that my method of in- 

 vestigation, if applied to rainfalls such as these, especially 

 to such as the last-named, would give results very different 

 from those we have seen. But contemporary records furnish 



