Volume 3, Part 1, Slice 2 Part 31 (1/2)
(74) y/x = tan [eta].
so that [eta] is the inclination of the chord of the arc of the trajectory, as in Niven's method of calculating trajectories (_Proc. R.S._, 1877): but this method requires [eta] to be known with accuracy, as 1% variation in [eta] causes more than 1% variation in tan [eta].
The difficulty is avoided by the use of Siacci's alt.i.tude-function A or A(u), by which y/x can be calculated without introducing sin [eta] or tan [eta], but in which [eta] occurs only in the form cos [eta] or sec [eta], which varies very slowly for moderate values of [eta], so that [eta] need not be calculated with any great regard for accuracy, the arithmetic mean ([phi] + [theta]) of [phi] and [theta] being near enough for [eta] over any arc [phi] - [theta] of moderate extent.
Now taking equation (72), and replacing tan [theta], as a variable final tangent of an angle, by tan i or dy/dx,
(75) tan [phi] - dy/dx = C sec [eta] [I(U) - I(u)],
and integrating with respect to x over the arc considered,
(76) x tan [phi] - y = C sec [eta] [xI(U) - [Integral,0:x] I(u)dx],
But
(77) [Integral,0:x] I(u)dx = [Integral,U:u] I(u) dx/du du = C cos [eta] [Integral,x:U] I(u) {u du}/{g f(u)} = C cos [eta] [A(U) - A(u)]
in Siacci's notation; so that the alt.i.tude-function A must be calculated by summation from the finite difference [Delta]A, where
(78) [Delta]A = I(u) u[Delta]u / gp = I(u)[Delta]S,
or else by an integration when it is legitimate to a.s.sume that f(v)=v^m/k in an interval of velocity in which m may be supposed constant.
Dividing again by x, as given in (76),
(79) tan [phi] - y/x = C sec [eta] [I(U) - {A(U) - A(u)}/{S(U) - S(u)}]
from which y/x can be calculated, and thence y.
In the application of Siacci's method to the calculation of a trajectory in high angle fire by successive arcs of small curvature, starting at the beginning of an arc at an angle [phi] with velocity v_[phi], the curvature of the arc [phi] - [theta] is first settled upon, and now
(80) [eta] = ([phi] + [theta])
is a good first approximation for [eta].
Now calculate the pseudo-velocity u_[phi] from
(81) u_[phi] = v_[phi] cos [phi] sec [eta],
and then, from the given values of [phi] and [theta], calculate u_[theta]
from either of the formulae of (72) or (73):--
(82) I(u_[theta]) = I(u_[phi]) - {tan [phi] - tan [theta]}/{C sec [eta]}, (83) D(u_[theta]) = D(u_[phi]) - {[phi] - [theta]}/{C cos [eta]}.
Then with the suffix notation to denote the beginning and end of the arc [phi] - [theta],
(84) _[phi]t_[theta] = C[T(u_[phi]) - T(u_[theta])], (85) _[phi]x_[theta] = C cos [eta] [S(u_[phi]) - S(u_[theta])], (86) _[phi](y/x)_[theta] = tan [phi] - C sec [eta] [I(u_[phi]) - [Delta]A/[Delta]S];
[Delta] now denoting any finite tabular difference of the function between the initial and final (pseudo-) velocity.