462 Prof. A. W. Riicker and Mr. W. H. White. 



Sensitiveness and Accuracy. 



The errors arising from various sources have been carefully con- 

 sidered. They naturally fall into two groups 



First. Errors in the determination of the primary standard of sus- 

 ceptibility, and affecting all results equally. 



Second. Errors affecting individual specimens. 



1. Under the first heading come the probable errors in the mag- 

 netometer constant and readings. 



In the value of the constant there is an estimated probable error 

 of i 1 per cent. Combining with this several smaller errors due to- 

 temperature, variation of resistances, imperfect adjustment, external 

 magnetic disturbances, &c., we conclude that the susceptibility of 

 each standard jelly is subject to a probable error + 2 per cent. The 

 two jellies actually gave factors O000109 and O000106 (the error of 

 constant being the same in each case). Presumably the mean factor 

 0-0001075 is within 2 per cent, of the truth. 



2. Errors in the values of individual rocks arise from the limited 

 sensitiveness of induction balance, telephone, and ear. The values 

 of the magnetic liquids are not appreciably affected by this, but it 

 becomes important in the case of rock specimens of small size and 

 feeble susceptibility. 



In the earlier paper already referred to it was stated 0*00013 was 

 the lower limit to the range of measurement by the methods adopted 

 and experiments made since with magnetic and non- magnetic sub- 

 stances indicate about the same degree of accuracy. For moderate 

 susceptibilities, however, the accuracy now attained has been greater 

 than in the earlier work, and the errors cannot in most cases be much 

 greater than 1 per cent. The difference between two specimens taken 

 from the same rock is of course greater than this, as the composition 

 of the rocks is not uniform. 



It is not necessary to give a detailed example of an experiment. 

 Suffice it to say that in a particular case the susceptibility of the 

 rock (No. 2) was determined by comparison (1) with liquids H and 

 D of susceptibilities 0-00074 and 0*00357, and (2) with liquids 

 H and C of susceptibilities 0*00074 and 0*00250. The results 

 obtained were 0'00164 and 0*00179, the mean being 0*00172. 



The agreement of these two values is not particularly good, though 

 not exceeding the probable error given above. 



For some unrecorded reason this specimen was tested again, and 

 gave 



From D and H', x = 0'00174 



C and H', x = 0*00170 



Mean value = 0*00172 



The susceptibility was not in this case computed from mixtures D 



