DB. T. E. ROBINSON ON THE REDUCTION OE ANEMOGRAMS. 
431 
Having no data to guide me in detecting the most favourable Lunar hours, I began by 
comparing the Ws for 0 h , 6 h , 12 h , 18 h , and 3 h , 9 h , 15 h , 21 h . I soon, however, found that 
this involved too much labour, and confined myself to the last hour. 
Calling C' the current, C'=|(W-W'\,=|/W-W'\. In this I made no attempt to 
\ 21 3 / \ 9 15 / 
allow for the sun’s elongation from the moon, or for their declination, nor for the 
horary changes of the coordinates, as the selected lunars are nearly uniformly distributed 
in each of the 24 common hours. 
By a Table with the moon’s hourly motion in Bight Ascension for argument I found 
the time which should be added to the Greenwich time of its culmination to obtain the 
common time of the above-named lunar hours at Armagh, and entering the Journal with 
these I obtained for each day two values of ^(W — W'), belonging to the upper and 
lower culminations. From the irregularity of these values it might seem hopeless to 
get any result ; but I pursued the inquiry in hopes of ascertaining the limits within 
which the mean of a considerable number of observations (even though very discordant) 
might be depended on. 
I only took the first six months of the year, as the results which they gave were quite 
satisfactory. 
Table X. 
Month. 
Current. 
No. 
PE 
Weight. 
CxW. 
Januarv 
0*2289 
404 
±1-939 
1-000 
0-2289 
The mean according to the 
weight. 
C'= 0-0906. 
Not differing much from the 
single mean. 
February 
0-1549 
376 
±1-809 
1-020 
0-1580 
March 
-0-0414 
415 
±1-886 
1-082 
— 0-0459 
April 
Q* 
o 
© 
© 
396 
±1-737 
1-190 
0-0084 
May 
0-0844 
423 
±1-345 
2-175 
0-1837 
June 
0-1125 
404 
±1-247 
2-416 ■ 
0-2720 
0-0911 
2418 
±1-661 
8-883 
0-8051 
The weights are proportional, the least, that for January, being taken as unity. It 
will be observed that these probable errors are far less than those given in page 415 ; but 
it should be recollected that here the variations can only occur within 6 lunar hours, 
while in the other case they range through months and years. Even so there are occa- 
sionally very great and startling changes when a gale bursts out suddenly or suddenly 
ceases. There were two values of W— W' above 40, and three above 30. Yet with all 
this I think the result is very remarkable. I do not pretend to assert that this value of 
C' really represents the tidal current at these hours, though it is in the right direction 
and of not improbable amount ; for it may be some uncompensated residue of the horary 
changes. But it is of great importance, as giving what must be a close approxi- 
mation to the real value of the average air-tidal stream, and as verifying my former 
