It will help to appreciate the loss of trigonometric
relationships ‘hands-on’. Hold a triangle facing you and rotate it about the horizontal x-axis
in the direction of the line-of-sight (the z-axis), then repeat the process about
the vertical y-axis, hold the mental picture of both rotations superimposed.
The procedure demonstrates that two normal planes rotated about perpendicular
axes toward the line-of-sight, i.e. in the direction of an axis that is perpendicular
to both, become as if orthonormal (to both normal axes). (If the experiment could be carried out simultaneously with both
planes they would maintain an angular relationship- that would be orthogonal-
even though each has become orthonormal to both normal axes). Next turn the triangle again facing you and
rotate it about the horizontal x-axis in the direction of the line-of-sight
[see
animation].
Note that we cannot see its projection on any plane that lies in the line-of-sight
regardless of the triangle’s (depression) angle l toward
the line-of-sight, in fact the only projection we can see directly would be in
the normal view. The procedure can be
repeated with the triangle turned at any angle m
about the y-axis before rotating it in the direction of the line of sight to show
that m and l suffice to define any angle for the plane of rotation with
respect to the x- and y-axes. It is important
for our discussion that we see that the fact of vectors being perpendicular on
a plane in 3-space does not necessarily mean that their projection will be so
on another plane, and vice versa. The
exercise shows that for any angle q
between two vectors there is a set of angles l
and m on orthogonal planes (only l
and one plane is shown in the accompanying animation, m is held at 0º) in the direction of the line-of-sight that
will vary the projected angle between q and 180º.
In our example we find that the tangent that defines a right triangle becomes
a cosine in the triangle’s projection with the ordinate becoming a hypotenuse
(angle w12 = 90º when cotl = cotq1 + cotq2). In the limit case when l and m are both 0º, i.e. the triangle lays in the line of sight,
the reflected angle w12 is 180º regardless
of the angles q1 and q2. This loss of correlation is pivotal to our
discourse: any vector in an imaginary plane regardless of its direction will be
seen as orthonormal with respect to a normal (real) plane and conversely a vector
that appear as orthonormal may have any direction in an imaginary plane.