AND NEIGHBOURING SEAS DURING 1897. 219 



It is seen from the preceding table that in only one out of ten cases of 

 drift does my calculated drift exactly coincide with the records of the 

 bottles (IX. 1) ; but in several other cases the results correspond in a 

 sufliciently close manner for all practical purposes. Some further 

 examination, however, is necessary before the reliability of my method 

 can be depended upon, because the estimated results depend upon the 

 assumption of open water, and this cannot always be conceded. 



In the case of II. 1 there is an angular error of 7" in a course of 180 

 miles, the estimated and actual distance of drift being practically 

 identical. In this instance there can be no question of any deflection 

 of the drift by the proximity of the shore. My calculations from the 

 winds at Prawle Point, from January 22nd to January 28th, give an 

 estimated drift of 34 miles in a S. by E. direction, which would thus 

 convey the bottles well into mid-Channel. The Prawle winds for the 

 ensuing period, January 29th to February 28th, give an estimated drift 

 of 99 miles in an E.N.E. direction, bringing the bottles to a position 

 about 26 miles S. by W. of the Needles, The direction of the drift 

 during March 1st to 16th was estimated as N.E. i E. in this part of the 

 Channel. The direction of the strong winds immediately before the 

 recovery of the bottle was as follows : — March 16th, S. to S.S.W. ; 

 March 15th, S.S.W. ; March 14th, S. to S.S.E. Clearly the bottle was not 

 driven to Eastbourne from the westward side of Peachy Head by a 

 current parallel with the shore, but must have been almost due S., 

 or even S.S.E., of Eastbourne when overtaken by the gale, as the 

 recovery of another bottle to the eastward a few days later also shows-. 

 The error iu my estimate is not due therefore to shore-deflection, but to 

 an excess in the southerly component of the wind records employed 

 This is just what one would expect under the circumstances, since the 

 records employed were those of Prawle Point, Hurst Castle, and Dunge- 

 ness. All of these stations are situated on the northern coast of the 

 Channel, and are consequently more exposed to the full force of southerly 

 than of northerly breezes. The course of the bottles, however, was in 

 mid-Channel for a considerable distance, and thus equally subject to the 

 influence of winds from north and south. That this is the true explana- 

 tion is seen from the fact that while the resultant wind for March 1st 

 to 16th was S.W. f S. at Hurst Castle, and S.W. ^W. at Dungeness, it 

 was nearly W.S.W. at Jersey (S.W. by W. h W.), the force at the three 

 stations being approximately the same. The error could have been 

 avoided by taking the mean between these three winds for the period ; 

 but I have preserved my estimate in its original form in order to show 

 by an example the precautions in this respect which must be taken when 

 very accurate results are desired. 



In the second case (III. 1) we have again an angular error to the 



KEW SKinES. — VOL. V. NO. 2. P 



