690 [June 18, 



If we assume, from the comparisons of long and short pipe, that 



Short pipe : Long pipe : : 1 : 3*1, 

 we have 



Station I. with short : Station II. with long : : 1 : 1-13 

 Portable with short : Station II. with long : : 1 : 9' 6. 

 All the numbers in my tabulated results are given in terms of 

 Station II. with long pipe, and the reductions have been effected by 

 means of a Traverse Table to single degrees on the following assump- 

 tions : 



Station I : Station II. :: Distance : Departure for 22 

 Portable : Station II. : : Distance : 10 x Departure for 18 



==Bist. : 10xDi ff .Lat.fo r2 lo. 



These ratios, when stated in numbers, are respectively 



1 : -375, 1 : 3-09, 1 : 1-14, 1 : 9'34 ; 

 whereas those above, designated as most probable, are 



1 : -365, 1 : 3-1, 1 : 1-13, 1 : 9'6. 



15. The cage of the Station Electrometer was taken out and 

 replaced on January 13th and April 10th, an operation which involves 

 a readjustment of the repelling plates, and may thus affect the sensi- 

 tiveness of the instrument. On the latter occasion a slight change 

 was also made in the position of the upper sight, tending to render 

 the instrument rather less sensitive ; but I do not think the effects 

 of these changes have been sufficiently great to be worth taking into 

 account. One at least of the electrometers seems to be affected by 

 some uncertain disturbing cause, which renders it more sensitive on 

 some days than on others. This is well exemplified by the com- 

 parisons of January 14th and January 19th, the mean results ob- 

 tained on the two days respectively differing by 8 per cent. Even 

 on the same day results obtained with different charges do not pre- 

 cisely agree, and the discrepancies are much greater than could arise 

 from errors of reading. They do not follow any easily ascertained 

 law ; and I am inclined to attribute them to imperfect elasticity in 

 one or both of the fibres a cause which may also have produced 

 discrepancies between results on different days, for the zero of torsion 

 has not always been tested when the comparisons were made ; and 



