differs by two hours from that given by actual observation ; 
but the phase of (6) differs by ten hours, and, if we are to 
choose between (4) and (5), I must say, that I prefer a 
difference of two hours to a difference of ten hours. If it 
was considered necessary to enter into greater detail, we 
should have to divide the observed vertical force variation 
into two terms, one having sin(^+X) and one having 
cos(^ + X) as factor. The first term only would have to be 
considered in a comparison of (4) and (5) with observation, 
for the second term would not belong to that part of the 
potential which is expressed by (4). If this is done we 
should find that all over the world, as far as we have any 
observations, the sign of the factor multiplying cos(^ + A) 
agrees with (5) and not with (6), and this is a sufficient 
proof that the greater part of the disturbance, whose hori- 
zontal components are. derived from (4), must have their 
seat outside the earth. 
The results of my calculations I had expressed, by saying 
that more detailed investigations “can hardly upset the 
conclusion arrived at in this paper, that the greater part of 
the diurnal variation is due to disturbing causes outside the 
earth’s surface.” As Mr. Chambers charges me with drawing 
conclusions unwarranted by evidence, I now formulate my 
result more definitely, thus : — 
In the general expansion of the variable g)cirt of the 
magnetic p>otential on the earth’s surface, there is an im- 
portant term V= — asin'i^cosi^sin(^-l-X), vjhere X is the 
longitude reckoned from Greenwich towards the east, and 
t Greenwich time reckoned from three o’clock in the after- 
noon. The greater part of this term, and possibly the vdiole 
term, is due to causes outside the earth’s surface. 
No subsequent work can affect this result, unless there 
is some blunder in my calculations. 
In the last part of my paper, I shewed how to calculate 
