Sixteen days after leaving the Galápagos for French Polynesia, I was woken up in the middle of the night by my partner, Guillaume Perret. Only this time we weren't changing shifts. Even half asleep, I knew something was wrong. Rebelle, our Philippe Briand-designed Kelt 30-foot sloop, was rolling wildly and had turned broadside to the waves.

Our passage up until that night had been going smoothly. We'd had steady, 20-knot winds and were averaging 160 miles a day. With any luck, we'd make landfall in French Polynesia in good time. But our luck took a turn for the worse on Day 16. At noon, an accidental jibe caused the boom to snap in two. At midnight, things got a whole lot worse. Conditions had deteriorated, and Guillaume was at the helm when the rudder sheered off. It quickly floated out of reach and out of sight. We were 650 miles from land and completely unable to control the boat.

Were we taking on water? We emptied the back cabin quickly and confirmed that it was dry. Once the adrenalin from our initial panic subsided, we had to come up with a plan. We'd seen pictures of a jury rig that used a spinnaker pole and a wooden board that was attached to the backstay and controlled with lines led to the cockpit winches. We'd start building such a rig at first light.

Since we didn't have any spare plywood with us, we used two locker doors to make the rudder. We bolted the doors together, lashed them to the end of our second spinnaker pole (the first one was being used as our new boom), led lines to the primary winches, and attempted to steer. The makeshift rudder kept jumping out of the water in the waves, so we attached diving weights to hold it down.

Steering with this rig was a lot harder than we thought it would be. We tried using the engine to help us steer and the triple-reefed main to keep us stable, but as soon as we managed to turn downwind, the boat would immediately round back up. We couldn't keep on a straight course, and we soon realized that our fix wasn't going to work. In hindsight, we could've lowered the mainsail completely, used the headsail to hold us on course, and towed warps behind us to keep the boat headed downwind. And steering with the sails-balancing the heavily reefed main and the jib-may have been the best solution. But we didn't try those techniques.

The following morning, Guillaume decided to dive to check out what was left of the rudder and to see if he could find a better solution. Going into the water in the open ocean is risky, but we had to do something. He tethered himself to the boat, went under with a mask and snorkel, and came up smiling. The rudder had broken away, but a portion of the rudderstock and a portion of a stainless-steel rudder reinforcement remained. With a bit of luck, it might just be strong enough to hold a temporary rudder for the next 650 miles.