16 KU.HTH RKPORT — IS.^8. 



No. of Station Miles from Height greater by 



in Minute book. Portishead. 2iul tiiau 1st Levels. 



Feet. 



1527 .... 12 ... . 0-3501 



1278 .... 15 ... . 0-.3796 



1229 .... 18 ... . 0-4591 



1178 .... 23 ... . 0-5339 



1128 .... 27 ... . 0-5734 



759 .... 30 ... . 0^352 



1 .... 33 ... . 0-6888 



45 .... 37 ... . 0-6956 



63 .... 39 ... . 0-7170 



114 .... 43 ... . 0-7532 



177 .... 49 ... . 0-8237 



210 .... 52 ... . 0-8622 



246 .... 56 ... . 0-9021 



248 .... 59 ... . 0-9208 



402 .... 63 ... . 0-9373 



462 .... 68 ... . 0-9714 



656 .... 74 ... . 1-0294 



After the most careful examination of every circumstance 

 which could possibly tend to occasion these curious differences, 

 I am inclined to believe that they arise principally from rapid 

 variations in the amount of atmospheric refraction which occur 

 durin(( the time that elapses in a single observation, and that 

 the progression of the error is in some way or other connected 

 with the progressive changes of the average temperature during 

 the course of the day, from about eight in the morning till six 

 or seven in the evening, — the usual limits of my working hours. 

 These variations in the refraction are much greater and more 

 sudden in summer than in winter, especially during the forenoon 

 of a hot and sultry day, when there are frequent alternations of 

 cloud and sunshine, and copious exhalations of moisture from 

 the ground. On such occasions I have sometimes known the 

 sudden clearing away of a cloud from the sun followed almost 

 in an instant, by a change in the apparent height of the vane 

 amounting to y^^^^ ^^ ^" inch, or more, on a distance of only 88 

 yards. At other times the change has been more gradual, so 

 that several successive readings, taken at intervals of two or 

 three minutes, have all either increased or diminished progress- 

 ively. Different seasons or states of the weather may therefore 

 fully account for the more rapid increase of these differences at 

 certain times than at others, such as the above table presents, in 

 which the errors are fovmd proportionablygreaterbetvveen Portis- 

 head and Bridgewater, than between Bridgewatt>r and Axmouth ; 



