"The mind that would hold many arts must be ordered as a library,
else knowledge lieth scattered as loose leaves and profiteth the scholar nothing."
— De Custodia Mentis, attributed to Magister Petrus of Prague, 1522
The problem of measurement occupied me for three days before I put it aside and worked around it.
Three days of looking at the same row of boards on the desk and arriving at the same conclusion: whatever was moving through the channels when I pressed my thumb against a carved rune, I had no way to see it, and no way to quantify it, and therefore no way to know with certainty whether any adjustment I made was producing a real change or only producing a different kind of result I was misreading. A thermometer would have resolved half of my questions in an afternoon. I did not have a thermometer. The next best thing I could construct, given the tools available to me and the materials I could get without arousing suspicion, was a method of direct observation, and direct observation of magical flow required that the magic disturb something physical enough to record.
I went back to the runes text and read the section on Sensory Runes twice over. The list included Calor, a detection rune for heat, and Magus, listed as a detector of magical energy itself. The text's description of Magus was less precise than I would have liked. It noted that the rune registered the presence or absence of magical energy without indicating quantity or direction, which made it useful for wards and alarms but not for the kind of continuous, directional observation I needed. A smoke alarm was not the same thing as a weather vane. Both detected something, but only one told you where things were going.
What I needed was a physical medium that would respond to the passage of magic, much like iron filings respond to a magnetic field. I spent a Wednesday evening sketching out possibilities on a scrap of parchment before I settled on the simplest one: fine powder in a carved channel. If magic moved through the channel the way I believed it did directionally, from high potential to low, favoring the path of least resistance, then fine dry particles sitting in that channel ought to shift in the direction of flow when the sequence was activated. The movement would be slight. Probably too slight to see from a distance. But being slight and visible was better than nothing.
I borrowed a stub of charcoal from the supply cupboard in the art of inscription class, which used charcoal for practice lettering, and brought it to the classroom the following afternoon. I also brought a small scrap of smooth pine from the carpentry supply, chosen for having close-set grain rather than the wider-grained boards I had been using. I spent the first half of the session carving a simple three-rune sequence into it: Fehu, Kenaz, Thurisaz, with channels cut at the depth I had established as the reliable working range from earlier experiments. Then I ground the charcoal stub against a flat stone until I had a small pile of fine black dust, which I tipped carefully into the carved channels and distributed along their length with the edge of a folded parchment scrap, trying to get an even coverage without packing it into the cuts.
The first attempt told me very little. I activated the Fehu rune and watched the charcoal, and the charcoal sat there. The Kenaz position warmed as expected after a few seconds, with a duration around six or seven counts, and I watched the dust in the channel between Fehu and Kenaz the entire time, looking for any movement, and saw none. The thermal effect faded. I set the board down. The charcoal was exactly where I had put it, without any change I could detect by eye.
I cleaned out the channels and tried again, this time spreading the dust thinner, barely more than a dusting across the channel surface, on the reasoning that a lighter mass would take less energy to move. The activation produced the same thermal effect at Kenaz and the same apparent absence of movement in the dust. I sat back and thought about that.
The channel was carved through wood, running from one rune to the next. If magic moved through the channel, it moved through the interior of the cut — through the air above the channel floor, or through the wood itself along the grain, or through both, with the carved geometry only shaping the path rather than constituting it. If the flow was primarily through the wood rather than across the air inside the cut, then charcoal dust on the channel surface would not register it, because the dust was sitting on top of the path rather than in it. I had been looking in the right place with the wrong medium.
I spent the next attempt covering not the channels but the flat wood surface around the Fehu rune and between the runes, an area of perhaps two square inches around each carved symbol, dusted lightly. The reasoning was that if magic spread outward from each rune position into the surrounding wood, even slightly, the disturbance ought to be detectable in the material immediately adjacent to the carving rather than inside it. When I activated the Fehu rune this time, I held the board flat and kept still and watched.
The dust moved. Not much, nothing dramatic, nothing I would have caught if I had not been looking directly at it, but in a radius of perhaps an inch around the Fehu carving, the charcoal particles shifted perceptibly, settling toward the channel edge nearest the Kenaz rune rather than distributing evenly in all directions. The movement was over in under two seconds, well before the thermal effect at Kenaz had built to its usual intensity, and afterward the dust lay still. But the displacement was real: the powder was in a slightly different arrangement than I had placed it in, skewed toward the direction of the next rune in sequence.
I wrote in my notebook: Charcoal dust displacement observed around Fehu position, radial shift toward channel connecting to Kenaz, duration under two seconds, preceding thermal effect at Kenaz by approximately three seconds. Suggests magical flow occurs before thermal manifestation, not simultaneously with it. Flow appears to pass through wood rather than along the surface of the channel. Effect too brief and small to track continuously through the full sequence.
Then, underneath: First observable evidence of directional flow. The instrument is crude. Repeatable.
I ran the same setup four more times across three sessions, varying the dust coverage and the activation duration. The displacement appeared in all four attempts, always at the source rune and always directional rather than radial. By the end of the week, I had enough consistent results to call it a reliable observational method, if not yet a precise one. It did not tell me how much magic was moving. It told me which way it was going, and confirmed that it was going somewhere at all. That was the first useful research instrument I had made, and I did not exaggerate its value to myself: it answered one question and left about fifteen others untouched, but one was more than zero.
The second problem I had been circling for weeks was the carving itself.
Pine was a cooperative material in the sense that it was soft enough to work with a knife, but soft wood had grain, and grain meant the blade did not always follow the line I intended. The rune forms required precise geometry: consistent channel widths, clean symbol edges, and corners that met at the angles the text specified. In hard pine with close grain, I could usually manage the straight cuts well enough, but curved sections in containment boundaries and the internal arcs of certain rune forms pulled at the blade, and a slight catch in the grain at the wrong moment produced a wobble that was small to the eye and apparently significant to the activated sequence. I knew this because I had carved two identical three-rune boards, one clean and one with three small wobbles in the containment ring, and the clean board produced a duration of nine seconds at Kenaz while the other produced five and a half, with a faint vibration that the clean board did not show.
Geometry mattered. My blade control was the limiting factor, and my blade control was a function of the material's resistance. There was not much I could do about the blade. There was something I could do about the material.
We had covered partial Transfiguration in class, temporary alteration of a material's surface properties, the kind of work Professor Blackwood described as preparation rather than transformation. She had used the example of softening wax before pressing a seal into it, the brief relaxation of the surface allowing a clean impression that the hardened material would then hold permanently. The Transfiguration did not change what the wax was, only its state for a short interval. The impression remained when the wax returned to hardness.
I spent a Tuesday evening in the common room working through the application on a practice scrap before trying it on carving stock. The spell was one I had attempted in class without much difficulty, the same one used to temporarily soften wax or clay for impression work. Applied to pine, it produced a discernible change in resistance within about thirty seconds: the knife moved through the grain with noticeably less pull, the cuts cleaner and more consistent, the curved sections of containment boundaries coming out without the grain-catch that had been producing wobbles. I carved a full three-rune sequence while the Transfiguration was active, confirmed the softening was still present throughout, and then set the board aside and waited.
The wood returned to its normal hardness in under ten minutes by my estimate, the surface feeling exactly as it had before. I activated the sequence the following afternoon, once the board had sat overnight, and the result was the longest clean duration I had recorded: eleven seconds at Kenaz, no vibration, no irregularity in the charcoal displacement pattern around the source rune.
I ran the same preparation on six more boards over the following week and found one condition that mattered: the Transfiguration had to be fully resolved before activation. Two boards I carved while the softening was still active and then activated too soon, before the wood had returned to its natural state, produced unstable results — one a thermal spike that faded immediately rather than building, the other a faint smell of scorching from the Kenaz position that was gone before I could find its source. Both failed to produce the characteristic Kenaz warmth at all. The material needed to be back to its normal properties before the runic sequence would treat it as a stable conductor. A transfigured substrate apparently did not channel magic the same way an untransfigured one did, even if the two looked identical to the eye.
I added that to the notebook as a working rule: Transfiguration as preparation is viable. The board must be fully returned to its natural state before activation. Carved geometry produced during the softened interval is retained and functional after return to hardness.
With cleaner carving and the charcoal displacement method giving me directional readings, I went back to the containment geometry problem that had been producing inconsistent results since the beginning.
The runes text was clear enough that containment mattered and specific about the basic forms: circle for stability, triangle for amplification. What it did not address directly was what happened when neither alone produced the result you wanted. The triangular containment boundary amplified output noticeably — the Kenaz thermal effect ran hotter and shorter, a sharp burst rather than a sustained warmth — but under sufficient input, the triangular boards cracked along the angles. The circular boundary produced a stable, sustained effect that ran longer without risk of structural failure, but reduced the intensity of the output compared to the triangle. I had run both enough times that these differences were consistent rather than incidental.
The question was whether the two geometries could be combined without canceling each other's advantages.
I tried the obvious arrangement first: a circle inscribed around the outer edge of the board with a triangle inside it, the runes placed at the three points of the internal triangle, the channels running along the triangle's sides. The first two boards with this layout produced results close to the circular arrangement alone, stable, mild, sustained, without the amplification I had hoped the internal triangle would contribute. Looking at the charcoal displacement pattern around the source rune, the outward movement was directed toward the nearest channel along the triangle rather than toward the containment ring, which meant the circle was not participating in the flow at all. It was containing, but passively — catching whatever dispersed past the internal triangle rather than actively shaping the path.
The arrangement that worked came from a different direction. Instead of placing the containment ring outside a fully enclosed triangle, I used the circle as the primary containment and arranged the internal channels along three arcs inside it, each arc running from one rune position to the next but curving inward rather than running straight across. The channels did not form a triangle; they formed something closer to a three-bladed shape, each blade curving toward the center before bending toward the next rune. The circle enclosed the whole arrangement, and the curved internal channels directed flow toward the center before it reached the next rune, which I reasoned would both sustain the flow longer within the containment and direct it more efficiently toward the terminal rune.
The first board cut with this layout produced a Kenaz effect that lasted fourteen seconds at an intensity I had not recorded before, not the sharp burst of the triangle, but a sustained, even warmth that spread from the Kenaz carving in a slow radius rather than a quick pulse. The Thurisaz terminal rune showed a faint warmth for the first time, brief and much weaker than the Kenaz effect, but present. The charcoal around the Fehu position showed the strongest directional displacement I had recorded; the powder shifted clearly toward the channel curve rather than spreading.
I sat looking at the board for a while after the warmth faded.
The terminal rune had finally responded. It had taken two months of failed boards and careful single-variable changes to get there, but the flow had reached the end of the sequence. I wrote the result down carefully and then tested three more boards cut to the same geometry before I allowed myself to conclude that the layout was repeatable. All three produced the same sustained Kenaz effect and faint Thurisaz warmth. The hybrid containment held.
The warming token came out of a practical calculation rather than a theoretical interest.
December in the castle was cold in a way that crept in from the stone floors and the drafty passages rather than arriving all at once, a slow accumulation of chill that the common room fire managed well enough but that the classroom did not. The classroom faced north, admitted gray light and very little warmth, and had no fireplace. By mid-afternoon, I was often working with cold hands, which was a nuisance with a knife in them, and by the last working hour, when I had lit the candle stub, my fingers were stiff enough that fine carving was not worth attempting.
The self-warming token was the first device I built because I actually needed it, which I mention because everything before it had been experimental: boards built to test principles, to observe failures, to bracket working ranges. The token was the first thing built to be used.
The design was straightforward by that point. I had the hybrid containment geometry, I had Fehu for sustained power input, I had Kenaz for heat, and I had a rune I had not yet used in a working device: Uruz, associated in the text with amplified flow and sustained magical force, classified among the power runes as a reinforcing element rather than a source. The text's description suggested it did not add energy but extended the duration of a flow that was already established. If Fehu was the supply and Kenaz was the load, Uruz was something placed between them to keep the current from dissipating before it reached the terminal.
I placed Uruz between Fehu and Kenaz in the sequence, used the hybrid containment ring with curved internal channels, and carved the whole arrangement into a disc of pine roughly the size of a large coin, the kind of size that would fit flat against the palm or sit in a coat pocket without bulk. The carving took two sessions, the first done with Transfiguration preparation and the second used only for cleanup and checking the channel depths against my earlier, reliable range.
When I activated it, the Kenaz warmth appeared at about four seconds as usual, and kept going. I counted out loud in the empty classroom, something I had started doing for duration measurements after the breath-count method proved variable. The warmth was present at twenty seconds, at thirty, at forty, still steady and mild, the disc warm against the palm but not uncomfortably so, the heat distributed across the whole disc rather than concentrated at the carved position. At sixty seconds, I set it on the desk and watched the charcoal dust. No displacement, which meant the internal flow had stabilized into a steady state rather than pulsing.
At two minutes by my count, the intensity had dropped somewhat, but the warmth remained. At three and a half minutes, it was perceptibly fading. By four minutes, it was gone.
Four minutes of steady, mild warmth from a disc small enough to keep in a pocket. I wrote the result in my notebook, set the disc on top of the notes, and looked at it for a moment, sitting there on the desk in the gray afternoon light. The warmth was already gone, but the fact of it wasn't. The first runic device I had built for use rather than observation, and it had performed exactly as designed on the first attempt.
I built a second one that afternoon.
The second disc matched the first in geometry, channel depth, and rune placement, and produced a duration of three minutes forty seconds on first activation, slightly shorter than the first, the difference attributable to small variations in carving I had not been able to eliminate. Both were within the range I would call reliable. I kept them in my coat, one in each inside pocket, and activated them before the coldest part of the afternoon session.
By the fourth week of November, the notes had grown into a problem of their own.
The notebook was nearly full. The pages held channel-depth tables, containment geometry diagrams, duration records, activation conditions, failed board descriptions, and the charcoal displacement observations, all written in the cramped hand I had developed from years of fitting too many thoughts into whatever writing surface was available. Reading back through it required either remembering which section came from which week or turning pages until the relevant material appeared. Neither was efficient when I was in the middle of an experiment and needed to check an earlier result.
More than the physical organization, there was the problem of carrying the active theoretical framework in working memory while doing something with my hands. A rune sequence with more than three elements required me to hold the whole arrangement in mind, including the containment geometry and the channel routing, at the same time as I was carving one part of it. Whenever I shifted attention to the knife work, something else slipped, and whatever slipped usually made itself known in the test results two sessions later, when I could not account for a deviation.
I had run into this before in a different form: the engineering mind in me said that any system complex enough to require documentation should be documented, and any operator working from documentation should be able to consult it without interrupting the work. The notebook was the documentation. The problem was that consulting the notebook interrupted the carving, and consulting my memory interrupted the carving in a different way, and both interruptions introduced errors.
What I needed was a way to hold structured information in a form I could review without looking away from the work.
The idea that the mind itself could be organized like a physical space was not new in 1609. The art of memory, as the classical and Renaissance scholars called it, held that information could be fixed in the mind by associating it with a location in an imagined space, then retrieved by mentally walking through that space and encountering the associations in order. It was how some people had tricks to memorizing others' names by associating them with items in a room or memories they have. I remember some fictions that described a mindscape that allowed one to freely shape their memory storage and how their mind functions, though I highly doubt you could make your brain a quantum computer by meditating or using some sort of mind magic. If that were the case, you would've seen everyone seek this out already. I just need better memorization, not to be a biological computer. .
The imagined space was always available. The question was whether I could construct one detailed enough to be genuinely useful rather than merely a trick for remembering lists.
I picked up a library book on magical mental discipline the following Saturday. It was a general text on the subject, covering Occlumency primarily, with one chapter on what the author called the structuring of the scholar's mind, a short section compared to the chapters on defensive technique, clearly written as an afterthought for readers who had acquired the book for reasons other than defense. The author's description of mental organization was thin on practical instruction, but contained one paragraph that was worth the library trip:
The practitioner who would order his memory constructs an interior space and places there what he would retain, in fixed stations, returning to the same stations by the same paths upon each retrieval. The space must be consistent, familiar, and detailed in its construction, for an ill-imagined space will shift and dissolve at the first distraction, and an ill-placed object will be found nowhere when sought.
It was rather similar to a sermon I heard on the differences of men and women. A woman's brain is like a ball of yarn or a thing of spaghetti with each thought connected to a different area, so on and so forth in a way that no man could follow. A man's brain is like a set of boxes and a man has a box for everything and he goes into a box and follows that line of thinking and only enters another line of thought after exiting the previous one. There is also a nothing box which men go into and think of literally nothing. It has been scientifically proven that men can actually do this; just sit there and have nothing going on up there. It's like when a guy watches TV, someone calls out, and they go, "Huh?" Not because they didn't hear or understand the person but because they had been thinking of nothing at all. It's actually quite common for me, sitting outside and just taking in nature, with not a thought in my mind. But women can never not think, which is why any man who wants to read women's minds is delusional, I imagine it would be like the old television static sound, just noise you can't follow. Like a professor who talks too fast for you to follow.
The first attempts at constructing the space were less promising than the theory.
I tried it in the dormitory after Thomas was asleep, lying flat with a candle on the bedside table burned down to a stub. The initial idea was simple enough: a stone room, roughly the proportions of the empty classroom on the third floor, with shelves along one wall. I would place runic principles on the shelves in a consistent order: the three main principles on the top shelf, the individual rune functions on the middle shelf in the sequence they appeared in the text, and the containment geometry rules on the bottom. Walk in, find the shelf, find the item. Simple, right?
What actually happened was that I got the room sketched in adequately, the floor, the north window, the shelves on the left wall, and then a door knocked somewhere in the castle and the whole construction scattered, and I was looking at the ceiling of the dormitory with nothing in mind but the sound of the door.
I tried again. The room held for longer. Then a board creaked somewhere above me, and I lost it again. The third attempt lasted until I tried to place the first item on the shelf and found that the shelf itself had lost its position relative to the window, swapping sides when I was not looking.
I lay there and accepted that this was going to take longer than one evening.
The sessions after that were approached the same way I had approached the first carving experiments: one thing at a time, in order. The first week I worked only on the room itself, spending ten minutes each evening with the candle lit and the curtain drawn, building the same space in the same detail, the stone floor, the north window with its gray light, the left-side shelving, the desk below the window, until the room arrived consistently when I entered it without the walls or the furniture shifting when my attention moved.
The second week, I placed objects. Not runic principles, yet I started with things I already knew without effort, historical dates and geography facts I had learned in the tower, placing them on the shelves in fixed positions and retrieving them in order. The retrieval worked reliably enough that by the end of the week, I was walking the room without losing items to distraction. A loud sound could disrupt the session entirely, and I had not found a remedy for that beyond waiting until the castle was quiet. But within a quiet period of ten minutes or more, the space held.
By the third week, I began placing runic material. The Three Principles went on the top shelf as planned: the First, Correspondence, at the left end; the Second, Ordered Flow, at the center; the Third, Containment, at the right. Each principle appeared not as text but as an arrangement of carved marks, the way I had come to see them in practice, the containment ring as a groove cut in a small stone block, the flow sequence as connected marks in pine, the correspondence principle as a rune form paired with its function, carved side by side. The physical representation seemed to stick better than text.
It was slow work, slower than the carving experiments, and less satisfying in the short term because there was nothing measurable to record except the quality of the retrieval itself, which I could only assess by trying it. I added a note to the bottom of one notebook page: Memory chamber, progress unremarkable but consistent. Patience required.
Margaret came by the classroom three or four times through November and December. On the first visit, she brought her Potions notes and sat across from me for an afternoon without asking about the boards, which I appreciated. The second, she arrived on a Thursday with a question about Transfiguration homework that could have been answered by the textbook, though I didn't say that, and we spent twenty minutes talking through it before she settled in and did her own reading.
The third time, she looked at the warming tokens, one of which I had left on the desk after activation, and picked it up.
"This is warm," she said.
"It is."
"Hast thou put it near the candle?"
"No."
She turned it over in her hand, feeling the disc, the carved surface, the weight of it. "'Tis thy rune work," she said.
"Yeah. First thing I've made that works the way I intended it to."
She set it down on the desk and looked at the row of boards behind it. "All these others did not work?"
"Some did. Just not the way I intended. I'm still trying to figure it out but I need to make it easier to carve the runes themselves before I can get anywhere."
She considered this and appeared to accept it as a reasonable distinction. "What dost thou intend to make next?"
"Haven't decided. Something more complex. But I am trying to finish working out the containment geometry first before I add more variables."
She nodded and went back to her book, which meant she was satisfied with the answer rather than disinterested, those being two different expressions with her. She stayed for another hour, left before the supper bell, and came back twice in December before the holiday preparations made the afternoons busier and the classroom harder to use.
The days shortened through December, and the castle grew colder by degrees, the stone floors overnight pulling heat from the air in the passages until the torchlight in the morning was lit against a chill that had settled in the night. The work assignment continued on its weekly schedule. Classes did not slow down. The Great Hall grew louder in the evenings as older students argued cheerfully about nothing in particular in the way people do when a holiday is close enough to make ordinary routine feel temporary.
I cut back the afternoon sessions in the final two weeks, not because the work had stopped, but because materials were harder to obtain with the castle occupied differently and the carpentry supply less accessible when the work assignment schedules shifted for holiday preparation. Instead I spent the free hours on the mental chamber, extending the placed material into the second shelf individual rune functions in sequence, each one a carved form with its English function beside it, placed at measured intervals along the shelf's length and on the runes text itself, reading ahead into sections on more complex containment geometries that I had not yet tried in practice.
The last session of real carving before Christmas week was on the twenty-first. I cut a new warming token as a third copy, using the same hybrid containment layout and the same three-rune sequence, and the result came in at three minutes fifty-five seconds on first activation, the longest duration of the three. I recorded it, noted the channel depth for that session, and set the token with the others on the corner of the desk.
On the twenty-third, I sat in the classroom in the last of the afternoon light and went through the notebook from the beginning. It started with the first activation, the day in early November I had pressed my thumb against a Fehu-Kenaz-Thurisaz sequence and felt warmth appear at the middle rune. Forty pages later, it ended with the third warming token's duration record. Between them: the channel-depth tables, the cross-line failures, the ring board anomaly that had never repeated, the charcoal displacement observations, the containment geometry tests, and the first appearance of a terminal rune responding. The hybrid containment worked. The preparation method worked. The warming tokens worked consistently. The mental chamber was holding material reliably after three weeks of daily construction.
I closed the notebook and sat for a while, the classroom going gray and quiet around me, the candle not yet lit. The three warming tokens sat on the corner of the desk, inert at the moment. Outside the narrow north window, the light was the color of used pewter, the clouds low and thick enough that there would be no stars tonight.
I activated one of the tokens, pressed my thumb against the Fehu mark, and set it on the desk in front of me. The warmth arrived at the usual interval, mild and even, spreading outward from the carving across the disc in the way I had recorded it doing every time.
Nothing that lay ahead was going to be simple. More complex sequences meant more variables, more opportunities for the kind of unexplained result that could not be corrected without understanding its cause. The measuring problem was still unsolved. The ring board anomaly was still unexplained. I had a list of things I intended to study that was longer than what I had actually studied. The shape of it was always the shape of it. The reading produced more questions than the experiments resolved, and the experiments produced results that required more reading.
But the system had stopped feeling like a thing operating on entirely unknown principles and started feeling like a system I could take apart, work on in sections, and put back together in a different arrangement. That change in the quality of understanding was not something I could have recorded in the notebook with any precision, but it was real, and it was enough to make the work feel like it was going somewhere specific rather than just somewhere.
The warming token cooled toward the end of the third minute, the warmth dropping off gradually until the disc was the same temperature as the desk it sat on. I gathered my things, tucked the tokens into the inside pocket of my coat, and went to supper