Forgive me if I am misreading your post, but it seems that what you wish to do is orchestrate an ordered sequence of curve objects such that the last control point of curve m is constrained in some fashion with respect to the first control point of curve m + 1. Perhaps the constraint is just positional: it suffices if the two points merely coincide and it does not matter if the join fails to be fair. Or perhaps the tangents must align, so that one end flows to the next without an obvious kink. Better yet, lets match up curvature across the join. Constraints vary, but the overarching construct is that of a Spline, an ordered sequence of piecewise elements constrained to join at their endpoints in accordance to characteristic rules. It seems to me that the end point coordination you are manually performing among the distinct curve objects t, r, b and l has its analog within the geometry data block of one curve object - but in this latter setting, Blender manages the mathematics for you - accepting some input from you on the end formulation of the constraint: perhaps setting the w parameter of a NURBS curve or specifying the types and positions of Bézier control handles. So what is the gain of replicating spline machinery among a sequence of curve objects when it already operates automagically at the geometry level of one curve object – or am I thoroughly misunderstanding your aims?