Sequences of Reversals. 61 



I shall call the old experiments the first ascending se- 

 quence (A) . Then I made a series which I shall call the 

 second ascending sequence (B). Then a final descending 

 sequence (C). 



The first values of B, which follow the old saturation, have 

 the resistance somewhat higher than that found when the bar 

 was fresh. After a few reversals, however, the effect of the 

 history seems to disappear as the values ascend, and the value 

 for mean inductions (minimum value of resistance) is almost 

 exactly the same as before. The saturation-value corresponds 

 to a higher induction than would be given by the old curve. 



The succeeding curve of descending values (C) shows in- 

 creased resistance, or less magnetism, in the mean inductions. 

 As the small inductions are approached the curve of C crosses 

 the old curve, and ends with a lower resistance, or greater 

 magnetism than the original initial value. 



The effects of ascending and descending sequences of re- 

 versals on the initial values would therefore appear to be 

 opposite in direction. At the same time the recent experi- 

 ments are few in number, and a much more extensive course of 

 work will have to be done before conclusions on these points 

 can be drawn with generality. 



I should like to say a few words on the relation of results 

 of this description to the molecular hypothesis, by means of 

 which I have represented a large number of experiments. 

 First, as to the position in which the hypothesis stands. 



I am quite unable to understand how Weber's hypothesis 

 can be applied to account for such laws as we are dealing 

 with, i. e. where the initial magnetic resistance is greater than 

 that for mean inductions. The hypothesis, which I have 

 worked out and applied in detail to a large number of experi- 

 mental cases of all sorts*, depends on two chief processes ; 

 the one of which accounts chiefly for the larger initial values 

 of the magnetic resistance, the other for the larger saturation- 

 values. 



First, as to the saturation-values. Each magnetic particle 

 is supposed to transmit the magnetism through a certain axis 

 and not otherwise. Using the analogy of a hole in a bead ; 

 packed with wires, the permeability so far is measured by the 

 portion of the hole left unoccupied, or by what I call the 

 defect of saturation. This alone would lead to a law of per- 

 meability similar to Frolich's supposed law of the conductivity 

 of magnets with ends. 



Next, to account for the initial values. The axes of the 

 particles being distributed uniformly in all directions, there 



* Phil. Mag. (1885) xix. pp. 73 & 333; xx. p. 318 ; xxii. p. 298 ; xxiii. 

 p. 350. 



