18 REPORTS ON THE STATE OF SCIENCE.—1917. 
The ordinary observatory, however, has only one set of magnetographs, and 
accidents will occasionally happen. There may be loss of the whole or 
part of the trace of one or more of the five selected days of a month, and 
there is no general agreement as to the course to be then adopted. The 
omission of one day represents the absence of 20 per cent. of the material. 
On the average ‘quiet’ day at Kew, the n.c. (non-cyclic) change in H in the 
24 hours is a rise representing some 10 per cent. of the total range, but it 
varies much on different ‘ quiet ’ days, and has not even an invariable sign. 
Thus the absence of one or two days from the selected five may make an 
enormous difference in the n.c. change. Then there is the allied question of 
whether or not to apply an n.c. correction. The theoretical aspect is 
complicated by the fact that most, if not all, force magnetographs have an 
instrumental drift. In some it is comparable with the true n.c. change on 
‘ quiet’ days, and sometimes it is even larger, especially in vertical force 
instruments. Everyone will allow that what is a purely instrumental 
effect should be eliminated, but the difficulty is to say what is instrumental 
and what is not. The question of how to deal with n.c. changes has an 
importance which has generally been somewhat imperfectly appreciated. 
A second question in connection with diurnal inequalities arises at 
stations which do not confine themselves to the international ‘ quiet ’ days. 
The natural view to take, especially for a theorist, is that all days should be 
included in the diurnal inequality. If we take, however, a station like 
Sitka or Eskdalemuir, when a really big storm comes along it is largely a 
matter of chance whether the trace remains on the sheet. When the limit 
of registration is exceeded, the hourly value is quite unknown, Even the 
rigid moralist in such a case recognises the necessity of omitting the hour, 
and the average physicist will concede the omission of the whole day. But 
the omission or inclusion of even one day of very large disturbance may 
produce a large effect in the diurnal inequality forthe month. Recognising 
this, and also the great uncertainty in measurements made on highly 
disturbed curves, those brought into intimate contact with magnetographs 
have usually decided to omit highly disturbed days, and others have gone a 
great dealtfarther. Disturbance is immensely greater at some stations than 
others, and even at a single station, in estimating disturbance, the personal 
element counts for a great deal. Thus, without some systematised scheme, 
a common choice of days cannot be hoped for. This would not so much 
matter if disturbances were absolutely erratic phenomena, exercising no 
systematic effect on the diurnal inequality. Disturbance, however, 
influences both the type and the amplitude of the diurnal inequality, and 
the influence may be much greater in one magnetic element than in 
another. At stations in high latitudes, a diurnal inequality, based on 
selected disturbed days, may have double the amplitude of one based on 
selected ‘ quiet ’ days. 
§6. Another question of some practical importance may be mentioned. 
The practice of recording declination (D) and H changes was once nearly, 
if not absolutely, universal, but there are now at least four observatories 
(Batavia, Potsdam, Eskdalemuir, and Greenwich) which have departed 
from this practice, the three former recording changes in two rectangular 
components. ‘This departure may have a balance of advantages at a very 
fully-equipped observatory, though that is a matter of opinion, but at the 
average magnetic observatory it has some very ‘obvious practical dis- 
advantages. A declination magnetograph requires less attention and less 
