Lovering and Bond on Magnetic Observations at Cambridge. 69 
We are now able to add one more to the list of 99 places for which 
Gauss has compared the computed and observed values of the ele- 
ments. The difference which appears in all cases between the two 
is produced by several causes. The observations are not cotem- 
poraneous: and they are vitiated by accidental errors and the strange 
anomalies of the magnetic force. The coefficients which depend 
upon the grouping of the observed values must suffer from the same 
influences; and hence the computed places by no fault of the theory 
are involved in uncertainty. In some cases the two errors may bal- 
ance each other; at other times they will conspire to produce a 
great difference. Thus we explain those considerable discrepancies 
which occasionally appear between the results of observation and 
calculation. So far as declination is concerned, Cambridge suffers 
particularly from these causes; there are only 3 out of the 97 places 
for which Gauss has made the computation where the difference is 
so great between the computed and observed element. In these in- 
stances, it amounts respectively to 5° 45’, 4° 42’and 5° 15’. In regard 
to inclination the case is more favorable, as there are 40 places in 
Gauss’ catalogue where the difference is greater than at Cambridge ; 
the maximum difference being 4° 38’, or 5 times that of the latter 
place. Out of the 98 places for which the declination has now been 
calculated the difference is plus in 52 instances and minus in 46 ; and 
out of 100 for which the inclination has been computed the difference 
is plus 66 times and minus 34 times. This is satisfactory proof that 
the error proceeds from the observations and not from the theory. 
Before sentence can be fairly pronounced upon the latter, better ob- 
servations must be possessed for comparison and the determination 
of the arbitrary coefficients. It is especially to be desired that cotem- 
poraneous observations of great accuracy should be made in every 
quarter of the globe, that the calculated values of the coefficients may 
