388 Mr. cavendish’s Account of 
frame, may be done without any danger of altering the 
pofition of the needle or making it vibrate, and reads off 
the angle as before. The mean of thefe two is the true 
angle which the line p p makes with the fide ab, fup- 
poling the diviiion and vernier to be rightly conftrudted, 
even though neither the lines p p nor f/ fhould pafs 
through the center of the pin. Having done this, he 
takes off the cap and inverts the needle, and obferves by 
both indices ?is before. It is plain, that if the line p p is 
parallel to the direction of magnetifm in the needle, this 
mean will agree with the former, fuppofing that the 
magnetic variation has not altered between the obferva- 
tions. On the other hand, if it is not parallel to the di- 
rection of magnetifm, but makes the variation appear 
greater than it ought to do when the needle is upright, 
it will make it appear as much lefs when the needle is 
inverted; fo that the mean of the two abovementioned 
means is the true angle which the direction of magne- 
tifm in the needle makes with the fide ab ; that is, the 
true variation of the needle at that time and place, fup- 
pofing ab to be placed accurately in the meridian. Hav- 
ing thus found the true angle which the direCHon of 
magnetifm makes with ab, he fubtraCts that fhewn by 
the index p in the upright pofition of the needle; the 
difference is the error of the inftrument in the ufual 
manner of obferving. 
It was by this method that the error of the inftru- 
ment, at the time of the obfervations in 1774, v r as 
found to be io'. For example, by a mean of the obfer- 
vations 
