418 
DR. T. R. ROBINSON ON THE REDUCTION OF ANEMO DRAMS. 
scarcely be deemed casual. In the 744 septimates of January there are only seven in 
which all the W and S are positive ; in the 720 of June there are none. 
It might be expected, from the mechanism of the polar and equatorial currents, that 
both components would change signs simultaneously ; but it is not so. I find that the 
proportion of the combinations is : — 
In January . . ^^0*589; ^^0*141; ^~0*184; 0*086. 
In June . . . „ 0*406; „ 0*220; „ 0*198; „ 0*176. 
The combination of -f-W with — S may arise from the influence of a continent to 
the east of Ireland, and that of — W and ~f S from a north-east current whose north 
component has been destroyed by friction ; but I looked for a greater frequency of 
— W and — S. If we confine ourselves to consider +W, — W, +S, and — S sepa- 
rately, we find : — 
For January . . . Sum (-f-W) 
Sum (+S) 
For June .... Sum (+W) 
Sum ( + S) 
32121; Sum (-W)= -8315; 
:42733 ; Sum (-S) =-6372. 
13298; Sum (— W)= — 9002 ; 
15978; Sum (— S) =-7163. 
The amount of negative components does not differ very much in the two months, 
but that of the positive is nearly triple in January what it is in June. Were we to 
attempt to develop separately these + and — values, we should be embarrassed by the 
different numbers of them belonging to each hour. Thus in J anuary the number for 
— W is 47 at 2 h , 69 at ll h ; for — S is 41 at 2 h , 56 at l h . In June, for — W it is 63 
at 3 h , 91 at 15 h ; for — S it is 65 at 2 h , 98 at 9 h . Supposing them developed in terms 
of the time, we should still be unable to obtain any absolute values of the components 
at a given epoch unless we knew the causes which produce these negative values and 
the laws of their action. It is evident that the equatorial current predominates here, 
but that there coexists with it a polar one, probably above, possibly collateral, which is 
occasionally mixed with the other by some disturbing force — probably barometric. It 
seems also that the monthly variation of the components is in a great measure limited 
to the positive values. For these reasons I have confined myself to the simple means 
of the entire set. But I think it might be well, in a series extending to several periods 
of five or seven years, to keep them so far separate as to be able to examine whether the 
occurrence of the negative values has any relation to time. 
A Table like this, whose data refer to dates separated by considerable intervals, will 
not suffice to give the components generally without some process of interpolation ; and 
we proceed to consider this. The form universally adopted where the quantities con- 
