In selecting a formula for the solution of geodesics it is of primary importance
to consider the length of the prograJn. that is the amount of core which jt will
occupy in the computer along with trigonOlnetric and other required functions.
It is advantageous to have on the computer system only one direct and one inverse
subroutine, both of which should give complete accuracy over lines of any length,
from a few centimetres to nearly 20 000 km.