Determination of the Magnetic Susceptibility of Rocks. 463 



and C taken together, for that would introduce extrapolation, which 

 we avoid as much as possible. 



In the case of the three specimens of susceptibilities greater than 

 O'Ol we are compelled to extrapolate beyond our strongest mixture, 

 and with the eighteen specimens below O'OOl we make our weakest 

 mixture a factor in the extrapolations. 



In some instances (e.g., No. 45) there is a very distinct difference in 

 the strength of two pieces broken off the same hand specimen. 



A few of the specimens used in the earlier investigations were 

 re-examined, and gave considerably larger susceptibilities than those 

 previously measured. This is probably due to an error in the deter- 

 minations of the absolute value of the susceptibilities of the liquids 

 employed, but in the case of any individual rock this error would not 

 exceed the difference often observed between two different fragments 

 of the same specimen. The fact also that the mean value now obtained 

 is almost exactly the same as that formerly given (viz., 0*00255 as 

 against 0*00245) justifies the conclusions based upon the use of this 

 or the smaller number (0*00160) which was used in many of the 

 calculations on the magnetic effects of basaltic rocks. The accuracy 

 of the absolute values now given has been tested by experiments on 

 ferric chloride. Although the susceptibility of this substance is 

 much smaller than that of most of the rocks, the result of our 

 measurements was in close accord with the means of values deduced 

 from the experiments of Quincke and Townsend. 



