AND NEIGHBOURING SEAS DURING 1897. 223 



to produce a drift of 24 miles in an E.N.E. direction (actually N.E. by 

 E, h K). The distance of Seaford from Selsea Bill is 34 miles. This 

 suggestion as to the course of the Seaford bottle, therefore, is sufQciently 

 consistent with the conditions prevailing at the time, and with the 

 remainder of our records, to render it the probable explanation of 

 the conspicuous lack of correspondence between actual and estimated 

 drift in this case. Another alternative is that the bottle may have been 

 driven ashore between the 16th and 19th March on the east coast of the 

 Isle of Wight, or on the Sussex shore west of Selsea Bill, as occurred in 

 the case of III. 4, and that it remained ashore until the latter end of 

 March, when it resumed its eastward drift ; but I consider this theory 

 much less probable. What I hold to have established is that the course 

 of the Seaford bottle was quite exceptional, and that its position on the 

 17th April does not indicate the position of the remaining bottles of the 

 same batch on that date ; for there was a general tendency for bottles 

 put out near the Eddystone between the 16th and 18th of February to 

 arrive on the French coast between Calais and Dunkerque between the 

 loth and 19th May, and I have shown that this could be achieved, pro- 

 vided the bottles passed through the Straits of Dover in the last week of 

 March. It was, on the other hand, impossible, if the bottles were in the 

 neighbourhood of Seaford in mid-April. The distance from Seaford 

 to Calais is 70 miles, the direction E. 12° N.; but the estimated drift for 

 the period April 17th to May loth (based on the winds at Dungeness) 

 is only 35 miles, and the direction E. 30" S., an error both of direction 

 and distance which is too serious to be due to the method of computa- 

 tion, and which admits of no explanation from the nature of the winds 

 prevailing at the time. 



The two last cases in the table now require consideration. The first 

 of these (XXX. 1) is the drift of two bottles from the Manacles to 

 Sennen Cove, a distance of 35 miles, during the first week of April. 

 The fact that two bottles pursued the same course shows that the causes 

 of the drift were very constant. The case is the more remarkable as two 

 headlands — the Lizard and the Land's End — had to be rounded during 

 the drift. Now my estimated drift for this period, though in the right 

 direction, is remarkably deficient in distance. Indeed, so variable were 

 the winds in this part of the Channel during this period that it is quite 

 impossible to attribute the drift in this case to the action of the wind. 

 As a check upon my estimate based on the Scilly records, I have also 

 made a calculation as to the direction and distance of drift based on the 

 winds at Falmouth, using the complete set of hourly records of the 

 anemograph for the period. The result is ecj^ually inadequate. I have 

 also estimated the drift on the assumption that the rate of drift is 

 proportional to the velocity of the wind, which only made matters worse, 



