296 



DR. L. N. G. FILON ON THE 



5500 there exists a definite deviation from the straight line law, which deviation 

 takes the shape of an undulation, with a crest at about 4700 and a trough l)et\veen 

 5000 and 5100. 



To trace this effect more exactly, a new set of readings were taken with the glasses 

 3453. The readings were taken with special care at intervals of load of half a kilogramme. 

 The residuals from the best straight line were computed as before, and they are 



4?00 



4500 



5000 5500 



TMTH-MTftS 



1st order obs. 



6000 



6500 6700 



- Und 



- Illrd 





 Jl > 



Tension obs. 

 Pressure 



Fig. 9. Glasses 3413. Diagram of residuals. 



shown on fig. 10. Here, the observations being much more numerous, the residuals 

 indicate a distinct curve. The dotted curve in the figure has been drawn freehand 

 through the points to give some idea of the general shape of this curve. 



It is seen that this curve amply confirms the previous set of results, although the 

 observations were taken at several months' interval, with different adjustments, and 

 probably different personal equations. 



On examination, only 3296 shows anything like so marked an effect. 1809 and 

 2783 show the effect in much the same place, but weakly, and some sets of 

 observations do not confirm it. 935 and 3749 are hopelessly irregular. Nothing 

 definite can be asserted about them. As to 3413, the pressure observations do not 

 show this effect at all, and the tension observations show it only very doubtfully. 



