226 



DR. W. GEOFFEEY DUFFIELD ON THE 



the lines are obviously divisible, but it is more pronounced in the case of the group 

 with the greater displacement. 



If the decrease of the displacement per atmosphere with the pressure were linear 

 throughout the whole range, it would lead to an equation between the displacement d 

 and the pressure p of the parabolic form d = A.p Bp 2 in which the constant B is small. 

 But though a linear relationship may reasonably represent the graph of djp and p 

 over the small range of pressure from 20 to 80 atmospheres, there is an indication that 

 the descent of the graph is more rapid at first and that with increasing pressure it 

 becomes more gradual, suggesting a curve of an exponential form. This is emphasized 

 by Table VII., in which are given the ratios of the displacements per atmosphere at 



Diagram 3. 



Ratio 



WAVELENGTH 



Ratio of displacements per atmosphere at 10 atmospheres to displacements per atmosphere at 



higher pressures. 



10 atmospheres to the displacements at higher pressures, the data being taken from 

 Table III. The former are generally greater than the latter, and the mean value of 

 the ratio is 1'8. In Diagram 3 these ratios are plotted against wave-length; if the 

 relationship between the pressure and displacement were precisely linear, the dots 

 would group themselves about the line marked I'O, but it is very obvious that the 

 readings at 10 atmospheres are too large for this relationship to hold. The diagram 

 also shows the curious fact that the departure from a linear relation is much more 

 pronounced for lines of small wave-length. For large wave-lengths the ratio is nearly 

 equal to unity. This is partly, but not entirely, due to the fact that the lines of great 

 wave-length have not been examined over the full range of pressures. There is a 



