and on the Ball-ended Magnets of Robison. 31 



larities and deviations from this ratio were much smaller." 

 Robison again refers to the magnets on pp. 121 and 123 and 

 mentions that " these experiments [with ball-ended magnets] were 

 first made in a rough way in 1769 and 1770." 



§ 6. Before I knew that ball-ended magnets had been devised 

 by Robison I was led to the same design 1 . I argued that if a 

 sufficiently easy inlet and outlet were provided by which the 

 magnetic induction might flow into and out of the ends of the 

 magnet, it would be content to flow in and out there, and would 

 consequently not enter or leave the magnet at intermediate points. 

 There would then be no free magnetism except at the ends of the 

 magnet. I conceived that, if two spheres of iron or steel were 

 fitted to the ends of a steel rod, the spheres, if their diameters are 

 large compared with that of the rod, would provide the necessary 

 easy inlet and outlet for the magnetic induction. I also hoped 

 that the flow of induction would be nearly uniform over the 

 surface of each sphere. In this case the magnet would act like an 

 ideal magnet with its poles at the centres of the spheres. 



The expectations which I had formed have been closely 

 verified in practice, the students at the Cavendish Laboratory 

 obtaining very definite results with the Robison magnets. The 

 magnets are made by drilling a hole in each sphere and then 

 driving it tightly on to the end of a thin steel rod. We have 

 generally used the steel balls supplied by the Auto-Machinery Co., 

 Coventry, softening them before drilling the boles. I have also 

 tried balls turned out of mild steel of high permeability and I 

 think that they give slightly better " poles " than the bought balls, 

 which are of hard steel. But the advantage is very slight. We 

 do not generally harden the rods ; rods of " silver steel " as pur- 

 chased, or knitting needles, retain their magnetisation quite well 

 enough for most purposes. When put together, the system is 

 magnetised in a long cylindrical coil by a powerful current. To 

 concentrate the magnetism upon the balls as much as possible, a 

 rod of mild steel is placed in the coil at each end of the system, a 

 spherical hollow being turned out of the end of the rod in contact 

 with the ball. To obtain a good magnetic contact, the radius of 

 the hollow is made equal to that of the spheres. A somewhat 

 similar plan is described by Sir W. Snow Harris in his Magnetism, 

 ed. 2, 1872, p. 109. 



§ 7. The magnets produced in this way behave very nearly as 

 if they had ideal poles at the centres of the spheres. As an 

 illustration, I give some results obtained in May, 1899. A magnet 

 formed of a steel knitting needle '15 cm. in diameter, with balls 



1 Natural Sciences Tripos, 1899, Part I., Practical Physics, question 16. 



