What about 21.k12 22.l11 23.l10 24.j15 25.n11?

That would work. Unfortunately for me, I didn't spot -- until it was too late! -- the fact that a k12 link would allow white to connect on *both* sides of blacks k13-n14 link. Earlier, the 21.r10 move was misguided because I did not recognize -- at first -- that my planned setup from t14 to t18 along the right side could be circumvented by black being able to connect from r14 to p18. By that point, black's line from p10 to r14 prevented white's potential k12 move from working out successfully.

MEDCALF DEFENSE comment -- Early on, after black's excellent 6.p18 move, I realized that I was already in trouble. I could not see any workable response except 7.t18, and I knew that Axel could (and probably would) use the Medcalf defense (see-- http://twixt.wetpaint.com/page/The+Medcalf+Defense) to successfully get past the 7.t18 block. However, although black's supporting peg at 4.m14 was sufficient for executing the Medcalf defense, I realized that it was only just *marginally* sufficient, so I varied the usual response to the Medcalf defense to try to defeat it by creating an option to proceed up either the right side or the left side of the board. When Axel responded to 7.t18 with 8.t20, playing the Medcalf defense, I responded with 9.r19 and then Axel followed up with the usual 10.r21. However, at this point, I played 11.p16 instead of the usual 11.p20. From there, I eventually developed a position that was probably "winnable" (as per Alan's comment above), even though (as described above) I failed to actually see the winning 21.k12 move until it was too late!