ON THE RAINFALL IN THE BRITISH ISLES. 



IRELAND. 



317 



scale-point of the gauge the area of which is known from the mean diameter 

 gives the last column, " Error at scale-point." An example will make this 

 clear. Gauge No. I. has a mean diameter of 7-995, therefore 0-1 of an inch 

 should be 1268 grains, but the glass only held 1260 grains, being 8 grains 

 too little, it therefore shows 0-001 too much ; that is to say, when rain falls to 

 the depth of 0-100 in. it shows 0-101 in., an error of no practical effect, being 

 less than the ordinary error of observing*. 



Concerning this Plate it is necessary to explain that Nos. I. and II. are 

 liable to considerable error if the rods are allowed to rise much above the 

 funnel, as they then intercept rain which would otherwise pass over ; they 

 are sometimes tied down to the cross piece, sometimes only dropped in when 

 an observation has to be made. No. VI., Crossley's gauge, registers on the 

 dials by an arrangement of wheelwork, but it is desii-able it should (when 

 used) always be provided with a cistern and tap so that the quantity may be 

 measured by a glass, and the clockwork-record thereby checked. Nos. Y., 

 VII., VIII., X., XL, and XII. are all provided with glasses similar to No. III., 

 but of course varying in capacity. No. IX. was designed by Mr. Marshall, of 

 Patterdale Hall, for use in the mountains of Cumberland and Westmoreland. 



* It may here be noted that though the errors are uniformly worked out to three places 

 of decimals, very much reliance cannot be placed on an error of -001 in a 5-inch gauge 

 seeing that it involves the correct reading of a water-surface to one hundredth of a linear 

 inch. 



