MERIDIAN ON THE COAST OF COROMANDEL. 177 



ought to be, no further notice was taken of it; 

 but should tlie sum of the three angles be nearer 

 the truth by taking it into the account, and that 

 there appeared an irregularity in the other two ob- 

 served angles, I have made it a rule to take each 

 observed angle as a correct one, and divide the ex- 

 cess or defect between the other two, and then 

 compute from the given side the other two sides; 

 and after doino: the same thino- with each of the 

 angles successively, a mean of tlie sides thus brought 

 out was taken, which, to certain limits, will al- 

 ways be near the truth. I then varied the selec- 

 tion of the observed angles, rejecting such as I 

 had reason to doubt; and by correcting them, and 

 computing the two required sides of the triangle, 

 those which gave the sides nearest to what had 

 been brought out by the other method, were adopt- 

 ed, let the error be what it would. This, however, 

 has rarely happened ; and when it did, great pre- 

 caution was used ; and no angle was rejected with- 

 out some reason appeared to render it doubtful. 



In correcting the observed angles to obtain those 

 made by the chords, I have used the formula given 

 by the Astronomer Royal, in his demonstration of 

 ]\I. De La MB re's prol)Iem, which appears in the 

 ■Philosophical Transactions for 1797. The spheri- 

 cal excess is of course had from the well known 

 method of dividing the area of the triangle in 

 square seconds, by the number of seconds in the arc 

 equal to radius, where the number of feet in a se- 

 cond may be had by using the degree as has been 

 commonly applied to the mean sphere, or the mean 

 between the degree on the meridian and its per- 

 pendicular. This being of no further use than to 

 check any error that might happen in computing 

 the corrections for the angles. 



In converting the sides of the triangles into arcs, 

 Vol. VIII. N 



