226 Dr. C. R. A. Wright on the JS number of 



E.M.F. of the cell, Mr. Shida used the following formula, 



E = R.H.tan«.K, 



where H=b + p, H is the horizontal component of terrestrial 

 magnetism, a is the angle of deflection of the tangent-galva- 

 nometer, and K is a constant calculable from the dimensions 

 of the instrument &c. With the object of obtaining several 

 values of R . tan a and averaging them, Mr. Shida caused R 

 to vary within certain limits : unless considerably different 

 currents were made to flow, no sensible alteration in the 

 E.M.F. through polarization would occur; so that the product 

 R . tan a should be sensibly constant, barring unavoidable 

 errors of observation. Mr. Shida states, in each of his original 

 papers, that R=# + £ + ?', where g is the resistance of the gal- 

 vanometer (measured as 30*86 ohms), b is the internal resist- 

 ance of the cell (found to be 2*02 ohms), and r is a variable 

 resistance introduced to vary the current and to bring the 

 deflection to near 45°. Mr. Shida then gives in each paper 

 the following table of numbers obtained — 



a. t, H. 



45 15 80 ohms 107*88 



42 45 100 „ '112-88 



51 39 50 „ 82-88 



and thence concludes that the mean value of R tan a. = 

 104*73 x 10 9 . 



Now the above numerical values are irreconcilable. Whether 

 Mr. Shida has copied down the wrong figures from his note- 

 book, or how he came to print these numbers, I cannot say ; 

 but the fact remains that they cannot possibly be all right. 

 That the accidental omission to correct errors of the press is 

 the cause of the discrepancies seems improbable, not only from 

 the nature of the errors themselves, but also from the fact that 

 the second (Phil. Mag.) paper is not an exact replica of the 

 first (B. A.) paper, various sentences having been altered, so 

 that an additional revise must have been made. 



First, if ^=80, since H=g + b + r 3 and since the galvano- 

 meter and cell were necessarily used in all the determinations, 

 it follows that if # = 30*86 and 5 = 2'02 as stated, R must 

 = 112-88 instead of 107-88. Secondly, if r=100, it similarly 

 results that R= 132*88, not 112-88. If it be supposed that in 

 these two cases the values of R are correctly quoted and those 

 of r wrongly given, then r must have been 75 ohms in the 

 first case and 80 in the second, instead of 80 and 100 respec- 

 tively. Thirdly, whether it be assumed that the values of R 

 or those of r are correctly given, in neither case is the mean 



