Two spheres A, B intersect in a circle, obviously, so a 3rd sphere C intersecting both A and B does so in 2 different circles.
It seems to me that the circle of the AC intersection must intersect the circle of the BC intersection in exactly two points, but a proof — even an informal one — eludes me. I can't think of a way (in a proof) to force the two circles AC and BC to intersect at all.
A useful trick: if $f(x,y,z) = 0$ and $g(x,y,z) = 0$ are the equations of two spheres $S_1$ and $S_2$ with distinct centers, then $h = f - g$ is linear, and therefore $h = 0$ is the equation of a plane $P$. Further, if any two of $f, g, h$ vanish then so does the third, or in other words, the intersection of any two of $S_1, S_2, P$ is contained in the third, ie equals the intersection of all three.
Now, the intersection of two spheres is the intersection of a plane and a sphere, and if you restrict the equation of a sphere to a plane you get a possibly degenerate circle (ie could be a single point or empty). If you intersect three distinct spheres, then that's the same as intersecting two planes with a sphere (or one of the pairs has the same center and therefore empty intersection). If we first intersect the two planes, we can get a line, a plane, or an empty intersection (if they are parallel and not coincident). Now the intersection of that with a sphere has to be one of: a circle, two points, one point (if the line or plane is tangent to the sphere) or the empty set.
All of these are possible, and all but the two points case are possible with all the centers being collinear.