IN THE MEAN EFFECTS OF THE LARGER MAGNETIC DISTURBANCES. 127 
ances or variations previously recognized ; whilst the very large amount of a great 
portion of them affords a security that the character of their laws of periodicity, de- 
rived by a separation in which magnitude alone is regarded as the distinguishing cir- 
cumstance, is not likely to undergo any material alteration (although the mean nu- 
merical values may be slightly altered), should an improved method of discrimination 
lead hereafter to a more perfect classification. 
The magnitude of a disturbance being thus taken as a characteristic feature of the 
class to which it should be referred, it became a question where the line of separation 
should be drawn ; it was desirable that on the one hand the proportion of separated 
observations should be ample for the intended purposes of generalization ; and that 
on the other hand the line should be taken sufficiently distant from the mean or 
normal position, to exclude from the separated observations the disturbances which 
as far as could be judged might belong to another class. On examination, how- 
ever, it soon appeared that it was not necessary to be very particular, either at 
Toronto or Hobarton, in selecting the distance at which the line of separation should 
be drawn on either side of the normal ; the disturbances of very considerable magni- 
tude were so numerous, and so far exceeded in amount any changes of regular occur- 
rence, as to make it a matter of very minor importance whether a few more or a few 
less were comprehended at the lower end of the separated class ; and after one or two 
trials, in which the number of separated observations was varied, but the conclusions 
were found to be substantially the same, five scale divisions (or 3''6 in arc), on either 
side of the mean or normal position of the declination magnet at the same hour and in 
the same month, were adopted as a convenient distinctive value for a disturbance of 
the larger class at Toronto. 
The number of observations thus separated amounted in the three years (1843, 
1844 and 1845) to 1650. The number of hours at which observations should have 
been made in the three years, exclusive of Sundays, Christmas days and Good Fridays, 
is 22,392, and the number of hours at which observations were actually made is 
22,376, sixteen observations only in the three years having been either missed or 
being otherwise imperfect; the proportion which the number of separated observa- 
tions bears to the whole number observed is therefore one in 13'6; and we are 
thus furnished with an approximate measure, on the average of the three years, of 
the frequency with which the declination magnet is liable to be disturbed to a certain 
amount at Toronto, when we say that if observations are made at regular intervals, 
one observation in every 13’6 on the average may be expected to differ as much as 
3''6 from the true mean position corresponding to the hour and season. In 1843 
there were 472 disturbed observations; in 1844, 612; and in 1845, 566; whence we 
may infer that 1843 was the least disturbed year of the three, and 1844 the most so : 
and for the purpose of assigning numerical proportions, if the degree of disturbance 
in 1845 be taken as unity, that of 1844 will be expressed by T08, and that of 1843 
by 0‘84. It ought to be one of the results of the system of simultaneous observation 
