The whiteheart process is not something anyone would describe as enjoyable. Heat the billet to white-hot. Pull it out and bury it in rust dust. Push aether through to drive vibration, force the carbon out as the oxygen binds to it and strips it away, blue flames rising where the reaction is hottest.

  Hold the temperature, keep the vibration even, don't let it cool too fast or the carbon creeps back in. Shave the face off when the glow dies, check the grain, repeat. It took me most of a day to get enough material.

  My arms were complaining by the third billet and the courtyard smelled like rust smoke and scorched iron for hours after I'd moved inside. But there was no way around it—the coil gun needed pure iron in its cores. Same reason as the motor, same physics.

  The coils would sit around those cores and when the electricity pulsed through them the magnetic field had to form and collapse fast, clean, in under a heartbeat. Impure iron resisted alignment. It would bleed the field out as heat, burning the silk, instead of concentrating it, and in a launch mechanism that wasn't inefficiency—it was the difference between a projectile and a weak shove.

  The whiteheart process was the only way to get where I needed to be with the materials I had. After the cores came the laminating. Each sheet dipped in resin, stacked, pressed under weighted boards overnight.

  The same logic as before: layered iron interrupted by thin resin breaks eddy currents before they can form. A solid core in a pulsed application would eat itself in heat. Laminated, it stayed cool and kept the field honest. I left the boards weighted and went to sleep.

  Three days of assembly after that. Fitting the coils to the cores, seating everything into the casing, running the contact checks between the palm points on the gloves and the mechanism. Each step was something I'd designed in theory and was now discovering the gap between theory and physical reality.

  One coil sat half a mark off-center and had to be reseated twice. The brush contacts needed adjustment. The retraction motor caught on the rope channel until I shaved a sliver off the guide. Small problems. The kind that meant the design was basically right and needed fitting rather than rethinking.

  On the fourth day I put it on and walked out to the spinning dummy. I set the dummy spinning with a kick, then kept kicking until it reached full speed. The blade was seated at the back of the barrel, iron striking face pointing forward.

  I pushed aether through the glove contact point deliberately—a specific push, separate from the passive flow, felt the silk convert, heard the faint crack as the coil discharged, and the blade was in the dummy's shoulder before I'd consciously registered it had left the casing.

  The dummy swung hard on its post, rocking back and straining against the anchor chain. I held still. The retraction trigger next. The motor wound the rope in—smooth, no jam, no tangle—and the blade pulled free with a resistance that said the hit had seated deep.

  As the chain came back the blade followed, and the flared mouth of the barrel caught it—the outward curve I'd shaped into the bone at the opening, wide enough to accept the blade coming in at a slight angle and funnel it back into the cavity without catching. It seated fully, flush, silent. From outside there was nothing to show a blade had ever been there.

  I ran it four more times. Two center-mass, one at the base of the neck fitting, one low. All five struck. All five retracted cleanly. 'There we go.' I unstrapped the case and set it on the bench and looked at it for a minute without touching it.

  The outside was still rough—tool marks, bands not fully flush, bone surface unpolished. It looked like a prototype because it was a prototype. But the inside worked and the inside was the part that mattered. One functional coil gun.

  The second would go faster now that I'd worked out the fitting problems on the first. The rope was the one piece I hadn't been satisfied with. Spider silk was the strongest thread I'd worked with. Stronger than anything available in the market, stronger than most manufactured cordage, and with the particular property of converting aether directly into electricity—the mechanism that made it useful in the gloves and the reason I'd used it for the coils in the first place.

  Length wasn't the issue. What I kept thinking about was the load case at the end of a full-extension launch. The blade goes out at speed. The chain pays out behind it. The rope reaches its full extent and goes taut in an instant.

  Every bit of momentum the blade carried transfers back through the rope and into the case and into my forearm in a fraction of a second. That's not a sustained load—it's a shock, a spike, and spikes were what broke things. A single thread of silk would handle it today.

  The story has been illicitly taken; should you find it on Amazon, report the infringement.

  I was reasonably confident it would handle it next week. I was not confident it would handle it on the hundredth repetition, or when I eventually understood the mechanism well enough to push more power through it and increase the launch speed.

  Three threads twisted together changed the failure math. A single thread failed at its weakest point—wherever there was a flaw in the silk too small to see, the load concentrated there and the thread parted. Three threads twisted into a braid distributed the load across all three simultaneously.

  Any flaw in one thread was backed by the other two. The braid also had a small amount of elastic give that a single thread didn't. That give absorbed the peak of the shock load at full extension—less force into my arm. But the energy didn't disappear.

  It stored in the rope for a fraction of a second and released back into the blade as it rebounded, snapping it into whatever came next with more force than the retraction motor alone would give it. A sword coming in to intercept the retraction would meet a blade that accelerated partway through its return arc instead of just being reeled back passively.

  Harder to time, harder to read, more dangerous at the moment an opponent thought the exchange was over. It would take time to train around. The snap changed the blade's return path in ways I couldn't fully imagine yet and I wasn't going to pretend I had.

  But the concept was right and the training problem was a training problem, not a design problem. There was one other calculation I'd been running and not saying out loud. If the blade seated in something solid enough—stone, heavy timber, a structural beam—and I needed to move vertically, the rope would hold my weight.

  I weighed less than most people my age, I was not bulky, having built myself for explosive bursts and endurance. There would be situations where getting up a wall fast was worth more than anything else I could do.

  A single thread of silk under that load combined with impact shock wasn't a thread I'd be trusting my life to. Three twisted threads were a different proposition. I cut three equal lengths and started working. The weaving had to be consistent—too loose and it just coiled without truly braiding, too tight and the threads kinked against themselves and lost their individual strength.

  I kept even tension on all three, rotating together, working slowly down the length. An hour to do it properly. My fingers were stiff at the end and the joints in my right hand were complaining in the particular way they did after fine work, but the result was a rope that looked right and behaved right when I tested the join by hauling against the anchor point I'd fixed to the tower wall.

  The connection held. The chain held. The splice where rope met chain held. I stretched my fingers out against the workbench and looked at the window. The afternoon light was cutting in at a low angle, the particular quality of early spring when the sun had started to work again but the air was still carrying winter's reluctance.

  Five weeks until the tournament.

  Magnar appeared at the gate while I was still at the bench. Cassia was with him—not beside him exactly, half a step behind and to the side, the slight separation that was just enough to say they hadn't walked over together even though they clearly had.

  I'd noticed this about her lately. The careful management of appearances—which things she applied it to and which she dropped the moment something annoyed her enough. I'd noticed the pattern. Magnar looked at the casing on the bench and then at the dummy, which still had the mark of five strikes in its shoulder and torso.

  "It works," he said. It wasn't a question.

  "Five for five."

  He nodded with the particular approval of someone who'd watched the design develop long enough to have his own investment in whether it worked.

  Cassia came to the bench and looked at the case. She didn't touch it. Her eyes moved over it the way they did when she was actually thinking about something rather than forming a response—the focus she saved for things she found genuinely interesting and wasn't about to admit to immediately.

  "How far?"

  "Three meters on the dummy. Could push it further. I want to understand the mechanism better before I change the power."

  She turned that over.

  "Can it miss?"

  "Yes. It's not a guided throw, it's a launched one. Angle and distance have to be right, but it's far easier to use than a bow."

  She nodded. Not satisfied, just processed.

  "Lyra asked if we wanted to go to the market tomorrow." Magnar said to me. "The southern vendors come through before the proper season starts. She wants to leave early."

  Cassia had sat down on a stool somewhere behind us. Not participating. Something she did when she didn't like an idea.

  "What time?" I asked.

  "Before it gets crowded." He placed his hand on the bench and drummed his fingers.

  Cassia finally looked at me. "You don't have to."

  I'd learned by now that 'you don't have to' from Cassia meant something different from the same phrase from other people. From other people it was social cushioning. From her it was either genuine removal of obligation—she was constitutionally incapable of asking for something she thought she wasn't entitled to ask for, so she removed the frame of the request entirely and left only the information; or she was expressing her disapproval.

  "I'll come," I answered frowning at Cassia's resistance. She probably didn't want me around other girls much, even if Lyra had her eyes on Magnar. The shift was small. A slight easing in the line of her shoulders, visible for about a second before she reassembled her composure and looked back at the casing.

  "What does the finished version look like?" She changed the subject, bringing attention back to the launcher. I turned towards her and started explaining. She listened without interrupting, asking one question partway through about the contact geometry that was better than most questions I'd gotten about the design from anyone.

  Behind us Magnar sat quietly on another stool and did nothing, listening to my explanations as well. Outside, the sky was the pale color of early spring evenings when the cold was losing its argument with the season but hadn't admitted it yet.

  Five weeks.