DE. T. E. EOBINSON ON THE SEDUCTION OF ANEMOGEAMS. 
413 
1857 
. . . W =12-59 
S= 9-81 
1858 
... - 6-27 
22-14 
1859 
. . . 7-78 
6-31 
1860 
. . . 6-80 
20-91 
1861 
... - 3-91 
- 1-21 
1862 
. . . 1-29 
- 2-70 
1863 
. . . 14-62 
22-65 
It is evident that here there is no regular succession ; and equally so that little 
dependence can be put on even the mean of the seven as representing the hour 0 for 
that day. But if, as is probable, these discordances are casual, we may expect they 
will disappear from the mean of a large number of observations — how large may be 
estimated from the Probable Error of these observations, though, on account of the mag- 
nitude of their discordances, this cannot be determined with great precision. There is 
also this difficulty in the process of finding the Probable Error, that the coordinates 
undergo daily and monthly variations, which must not be confounded with the casual 
errors. It is therefore necessary to confine ourselves to the observations of each individual 
hour during the seven years, and combine any number of these groups of seven. This 
is effected by the simple means of using as the divisor n — m instead of n — 1, n being 
the number of terms in the entire set, and m the number of groups. I have only 
thought it necessary to make the computation for W in January and June, and I find 
P E of a single observation . . + 5-901 + 3-913 
P E of mean of seven .... +2-230 +1-479 
P E of W in Table, mean of 217 +0-401 ±0-266 
P E of mean of month ±0-082 ±0-054 
The discordancy in summer is only two thirds of that in winter, and in both is so 
great that the mean of seven is not to be relied on ; and even the numbers of Table III. 
are not sufficiently certain. Perhaps these seven years may have been exceptionally 
irregular. The discordancy of S is still greater than that of W. Evidently single hours 
were out of the question ; I therefore took for each hour the mean of the month in the 
first instance ; I then grouped these means for every ten days, but ultimately adopted 
the entire month as the group. 
Before discussing these means individually, it may be useful to give their means for 
the entire period of seven years. Supposing the winter from October to March inclusive, 
the summer from April to September, the day hours from 7 am. to 6 p.m., the night 
from 7 p.m. to 6 A.M., we find : — 
Winter Day. 
Sum W=7899 m, 315 ; Sum 8=11239-92 ; Ann. Translation =137 38 ; D=35° 6'. 
Winter Night. 
Sum W=7264 m, 43 ; Sum S=11527-75 ; Ann. Translation=13812 ; D = 33° 25'. 
