MAGNETIC SURVEY OF JAPAN FOR THE EPOCH 1895.0. 173 



net of infinitely thin thread to be placed symmetrically with 

 respect to the circulation ; microbes living on the net may find 

 that the velocity of the fluid on different parts of the net can be 

 expressed very approximately by a function of descending powers 

 of the radius vector, but they have no right to assert thereby, that 

 the flow must have been produced by a piston or pistons inside 

 the globe. If some of the bold animalcules make an adventur- 

 ous expedition inside the net in search of the cause of the 

 circulation, they may be perfectly disappointed to find no trace 

 of pistons ever having been there. To make this a magnetic 

 problem, we have only to substitute a double sheet of attracting 

 and repelling matter for the surface where the initial pressure 

 is supposed to have been exerted. 



The question of the seat of action, must be sought by con- 

 sidering physical circumstances, analysis can not settle it, for 

 the simple reason of the vanishing of the surface integral, unlike 

 that in the case of gravity or electrostatics. It is difficult to 

 understand, how some magneticians have come to conclusions with 

 regard to the seat of action from mere observations on the earth 

 surface. Of course there is the common sense judgment without 

 going into logical process ; but that is no proof. We may take 

 two convergent series one for an internal and another for an ex- 

 ternal distribution at the upstart, as Gauss did, and comparing 

 the coefficients with observed values, find that they fit very 

 nearly with the former. But the converse of this may or may 

 not be true as just shown. 



It is curious that the rigorous mathematician, while discussing 

 the possibility of the existance of the atmospheric current, and 

 after enunciating the admirable theorem with regard to magnetic 

 shells, confines his attension to the coefficients of scaler potential 



