A family of circles is coaxal if, taken two by two, they share the same radical axis.
Coaxal families of circles come in three varieties:
Families in which the circles have no common points (these are the Apollonian circles of a pair of distinct points.)
Families that share exactly one point.
Families that share exactly two points.
The Coaxal Circles Theorem asserts that the families come out in pairs: all circles in one family is orthogonal to all the circles in the other. The pairing is simple.
The Apollonian circles defined by two points are orthogonal to any circle through these two points, and vice versa.
The circles tangent at a point are orthogonal to all the circles in the second family of circles tangent at the same point, but for which the radical axis of the former serves as the axis of symmetry, and vice versa.
The second part is obvious as of the two perpendicular lines (axes of both families), one is tangent to all the circles in one family, the other is tangent to all the circles in the other family.
To prove the first part, consider the inversion t in the unit circle centered at, say, A, the way we did in the discussion on the Apollonian circles. Any circle through A inverts into a straight line. Any circle through A and B inverts into a straight line through t(B). On the other hand, the Apollonian circles invert into the family of circles centered at t(B). These circles and the lines through t(B) are obviously orthogonal. Since, an inversion, as is its inverse, is an angle preserving transformation the original families of circles are also orthogonal. (The applet showing the Apollonian circles provides a clear visualization of this property. Just choose N = 10 or more and rotate P.)