The first real attempt was not the first attempt.
He spent two nights practicing the diagnostic pulse on forest creatures before he used it on Lyra. The technique was clear enough — he had built it from the boar's memory and he understood the mechanism. What he didn't yet have was calibration. He needed to know what the return felt like before he pointed it at his sister, before he put something he didn't fully understand into contact with her body and tried to read what came back.
The Bark Lizard's interior was healthy. He confirmed this by cross-referencing the diagnostic pulse against his healing sense, checking one against the other. They matched. The pulse was showing him what was actually there. After two sessions he trusted that enough.
Three nights after finding the memory fragment in the boar's absorption, on a night when the house was deeply asleep and Lyra's breathing had the settled quality of someone not in active distress, he sent Shadow under her bed.
He extended the pulse.
It required a different quality of attention than healing did. Not the focused outward push of healing application — something more open than that, a receiving quality, a willingness to let the return arrive without anticipating its shape. He had compared it in his own internal vocabulary to the difference between looking at something and listening for something. One was directed outward. The other was open and waiting. He had not been naturally practiced at the second mode. He had been working on it deliberately for three days.
The pulse left Shadow's position beneath the bed and moved upward through the mattress and blankets and the small body of his sleeping sister. The quality of it surprised him in a way he hadn't predicted from his practice sessions. He had expected something biological and informational. What he found was that the interior of a sleeping person — a body working quietly on its own maintenance, all the small functions continuing without any attention from their owner — had a quality to it he didn't have a clinical word for.
He let himself acknowledge this for exactly as long as it was true. Then he attended to what the return was telling him.
◆ ◆ ◆
The return took forty minutes to fully arrive and two days to fully interpret.
The diagnostic information was dense in the way complex conditions were dense — layered and distributed, requiring him to cross-reference readings from different areas to build a complete picture. He was not a trained diagnostician. He had no reference materials, no mentor, no clinical vocabulary to make interpretation faster. He was reading a language he had taught himself from a single partial observation of someone else using it.
He was also, by the end of the second day, fairly confident he was reading it correctly.
The picture assembled itself with a completeness and internal consistency that seemed unlikely to be wrong. The pieces fit together too well. The distribution of the infection, the locations where it was most concentrated, the pattern of scarring that mapped onto the infection's history over time — it was too coherent to be misinterpretation.
Persistent bronchial infection. Deep-seated, long-established, adapted to its host the way chronic infections adapted when they had been in place long enough: not maximally aggressive, because maximum aggression killed the host and the infection with it, but stable and self-sustaining. It had found a balance with Lyra's immune system that kept it alive at the cost of slow, steady damage to the surrounding tissue.
The damage was not dramatic in any single year. Small enough to fall within normal variation, attributable to weather or season or the quality of the last harvest. Over a decade the accumulation would have been significant. Over two decades, given the rate of scarring he had mapped, significant was too gentle a word.
He held the complete picture in his mind and sat with what he felt.
Relief first, because he knew now. He had gone from suspecting something general and serious to knowing something specific and addressable. A problem you can define is a problem you can work on. That was not a small thing.
Anger, briefly, at Healer Vance — who had either missed this or found something like it and not communicated it fully to his parents. He entertained the more charitable interpretation for about thirty seconds before filing it.
And underneath both of those, settling as the others resolved: the specific quality of determination that came when a long campaign finally had a map.
He knew where the infection was concentrated. He knew the scarring pattern. He knew the architecture of the damage. He knew now where to direct the healing, how finely to calibrate it, and what he was trying to accomplish with each session. The campaign that had been running blind for months had coordinates.
He could do this. Not quickly — quick would produce changes visible enough to raise questions no one was ready to have yet. Not simply — simple didn't apply to any part of this problem. But thoroughly, and carefully, and with the patience that had gotten him this far inside a body that most people would have written off as a significant limitation.
◆ ◆ ◆
The treatment protocol had three phases.
Phase one was stabilization. Not improvement — he was deliberately conservative in his ambitions for the first four weeks. He targeted only the most acutely active infection sites from the diagnostic scan, working to reduce the infection's current activity level without attempting to touch the structural damage it had already caused.
His reasoning was simple: he didn't fully trust his precision yet. He had spent months healing forest creatures, and he was good at it, but there was a real difference between the tissue of a Bark Lizard and the tissue of his sister. The consequences of an error were not in the same category. He needed to develop calibration for this specific patient before he attempted anything more ambitious.
Four weeks of nightly sessions. Shadow beneath the bed, still and patient in the dark. The healing magic extended through their connection in the finest thread Arthur could manage, directed to the primary infection sites with the care of someone doing delicate work under conditions they couldn't entirely control.
The thirty percent potency loss through the connection was a constant he had been compensating for long enough that it was intuitive now — adjusting his output to account for the transmission loss the way you adjusted the force of a note to reach the back of a hall. He didn't have to think about it anymore. It was built into his working method.
The most active infection sites reduced. He confirmed this with a diagnostic pulse at the end of the first month, checking the new reading against the baseline. The changes were small — small enough that he might have doubted them if he had less confidence in his diagnostic precision. He chose to have that confidence, because the alternative was discounting real data, and discounting real data was not a practice he intended to develop.
Phase one was working.
Phase two would go further — not just reducing the infection's activity but beginning to address the structural damage, the scarring that years of chronic inflammation had quietly deposited in the tissue surrounding the infection sites. This was the longer work, the slower work, the work that would require the regeneration capability he had been developing on forest creatures and had not yet trusted to use here.
He was not ready for phase two yet.
He went back to the forest creatures and kept practicing.