So nobody actually told me this, I had to work it out — we run the tunnel after the surface goes quiet because the rock holds the cold. Down here the air sits at maybe eleven, twelve degrees all night, and it does not move unless we move it. That sounds boring. It is not boring. It means the only heat in the whole system is the heat we make, and once you see that you start to see everything else.

Up top, a car fights the weather. The sun's been baking the asphalt, a gust pushes the nose around, the track temperature drifts forty degrees over an afternoon and your setup is a moving target you'll never quite hit. Down here? The track is the track. The walls are the walls. The fungi glow the same dim green at lap one and lap ninety. So when something changes — when the car suddenly wants to understeer into the long left — I know it's the car. It's the tyres, or the fuel weight, or me getting tired. The tunnel is honest with you. That's the whole pitch.

You'd think the dark would be the hard part. It isn't, really — your eyes give up on detail and start reading motion instead, which is what you wanted anyway. Yasu says the old runners used to tape over half their mirrors on purpose. I don't believe him but I also kind of do.

Anyway. Night running isn't romance. It's just the cheapest way to get a quiet, repeatable lab. Everything in this notebook assumes you've got one too — a stretch of road that doesn't argue back.

A tyre is a chemistry experiment you're standing on. The compound only does its job inside a window — maybe eighty to a hundred and ten on the surface, depending who poured it — and below that window it's literally a different material. Harder. Glassy. It slides where it should bite. People say "cold tyres" like it's a temporary inconvenience; it's not, it's a physical state change and you have to build out of it, lap by lap, with the only heater you've got, which is friction.

Here's the bit nobody draws for you: the heat doesn't come from the road being warm. We just covered that — the road is eleven degrees. The heat comes from the rubber deforming. Every time the contact patch squashes and springs back it loses a little energy as warmth — hysteresis, the books call it, which is a fancy way of saying rubber is bad at giving back what you put in. So a tyre warms up when you work it: brake hard, lean on it through a corner, let it squirm. Drive it gently and it just stays cold and confused.

Which means — and this took me three nights of getting it wrong — the warm-up lap is not a gentle lap. It's a weird, scrubbing, deliberately-clumsy lap. You're not lapping, you're cooking. Hmm. Wait, that's not quite it either; you can overcook the fronts and pull the rears along behind, freezing, and then lap one is a coin flip. There's a balance and I haven't found it.

The honest summary: temperature is the hidden variable under almost everything else in here. When a later chapter says "the car does X," assume "the car at the right temperature does X" and that I'm quietly panicking about it.

Okay. The tyre is not pointing where the car is going. That's it. That's slip angle — the gap, measured in degrees, between the direction the wheel is aimed and the direction it's actually travelling across the ground. And the strange, beautiful fact is: that gap is where the grip comes from. A tyre pointed perfectly along its path makes almost no cornering force. You need it to be a little wrong on purpose.

I drew this maybe ten times before it clicked, so let me try the version that finally worked. Picture the contact patch as a stack of tiny rubber bristles touching the road. As the tyre rolls and you turn, each bristle gets dragged sideways, bends, builds up a little spring force, then snaps free at the back of the patch. Add up all those bent bristles and that's your cornering force. More steering angle → more bend → more force… up to a point. Past maybe six, eight degrees the bristles at the back are just sliding, not gripping, and the whole patch starts to let go. That's the peak. That's the edge.

So the good drivers aren't the ones holding the wheel still. They're the ones living right at that little peak — feeding in just enough wrongness, feeling the force stop growing, backing off a hair. You'd think you'd want the tyre as "correct" as possible. Watch what happens: the more correct it is, the less it does. Hmm — wait, that's only true while you're still climbing the curve; once you're over the top, more angle = less grip and a sliding mess. Both things are true on opposite sides of the same hill.

I still can't feel the peak reliably. Yasu says you stop trying to feel it and start trusting that it's there. Note to self: ask him what that even means.

First way, the honest one: a gear ratio is a trade. You can have force at the wheels or speed at the wheels, never both at once, and the gearbox just lets you slide that trade around while you drive. Low gear, lots of force, not much speed — good for hauling the car off a slow hairpin. High gear, the opposite — good for the one long straight where the engine's already shouting. Everything else is bookkeeping.

Second way, the picture: imagine the engine's useful power as a hill with a flat top — it pulls hardest in a band of revs, weak below it, falling off above. Each gear takes a slice of road speed and maps it onto that hill. Pick your ratios badly and you spend the straight in the weak part of the hill, revs too high or too low, the car just… not pulling. Pick them well and every gear change drops you neatly back onto the strong band. Down here the straights are short and the corners are tight, so I run the whole box closer together — more changes, but the engine never falls off the hill.

Third way, the one I actually use at 2am: count the changes between two corners you know. If it's three flicks of the lever, the box is probably about right. If it's one long strained pull, you're geared for a track that isn't this one. Crude. Works. Yasu hates that I do this. He's not wrong, it's not a method, it's a vibe — but the vibe has never sent me into a wall, so.

Note to self: stop "drawing it three ways" and just do the maths once. (I won't.)

A corner is not a curve in the road. A corner is a negotiation about where the speed is allowed to be. The road just suggests; you decide. You can take the same bend forty different ways and they're forty different corners, and exactly one of them is fast, and it almost never looks like the one you'd guess.

The thing that took me too long: the corner starts before the corner. The braking, the little settle of weight onto the front tyres, the patient bleed off the brake as you tip in — that's all part of it. By the time you're actually turning, the corner's already half-decided. Get the entry wrong and there is nothing — nothing — you can do in the middle to fix it. You just get to choose which kind of slow you'd like.

And the apex isn't a place on the kerb, it's a moment: the single instant the car is most loaded up, most committed, most "this is as much as we get." Touch it too early and you're running out of road on the way out, lifting, throwing the lap away. Touch it late and you've left grip on the table the whole way in. You'd think later is safer. Watch what happens — later means slower entry means a slow, sad straight afterward. Hmm. Wait — not always; on a corner that opens onto a real straight, late-and-patient wins because the exit speed multiplies down the whole straight. So: tight twisty bit, hit it early-ish. Big corner onto a long pull, be patient. Same word, "apex," two opposite jobs.

Note: I keep "fixing" the middle of corners. The middle is not where the problem is. Write that on the wheel.

Brake bias is the dial that says how much of the stopping happens at the front versus the rear. And here's the cruel part: there is no right number, because the right number changes during the braking event itself. When you first hit the pedal the weight's still settling forward; a fraction of a second later it's all on the nose and the rear is light as a feather. A bias that's perfect at the start is wrong at the end. You're aiming at a target that moves while the arrow's in the air.

So what you actually do is pick a compromise and then drive around its lies. Too far forward: the fronts lock, the car goes straight when you wanted it to turn, you've got a passenger seat view of the wall. Too far rear: the back steps out under braking, which on a good day is "rotation" and on a bad day is the car deciding to lead with its boot. Down here, in the cold, with tyres that take forever to warm — I run it a touch rearward of "safe," because the corners are tight and I'll take a nervous, rotating car over a numb, ploughing one. Yasu thinks that's reckless. Yasu also brakes in straight lines like a librarian, so.

And the tunnel adds its own twist: the air's still and cool, the discs don't get the wind to shed heat, so over a long run the rears — smaller, working harder than the bias number suggests — fade first, and the balance you set on lap one is a different balance on lap thirty. I move the dial mid-run now. It feels like cheating. It is not cheating.

Honest verdict: I've never once set brake bias and thought "that's it, solved." It's a guess you keep editing. Ask Yasu if the fast guys feel the same. (They do. He won't admit it.)

Up top, aero is a religion — wings, floors, the whole car bent toward shoving itself into the road with the air it's pushing through. Down here it mostly just… isn't. The straights are too short to ever go fast enough for downforce to matter much, so a tall rear wing is basically a sail you're dragging through a cave for no reason. The first thing the tunnel teaches you about aero is to want less of it.

But — and this is the bit I didn't expect — the tunnel itself does aero things to you. The walls are close. The air can't get out of your way sideways like it does in open country; it has to squeeze past in the gap between the car and the rock, and squeezed air speeds up, and faster air is lower pressure, and lower pressure beside you is a gentle, constant suck toward the wall you're nearest. On a wide section it's nothing. In a tight bore at speed it's a real hand on the car, and it's not symmetrical unless you're dead centre, which you never are. So the tunnel quietly steers you, a little, toward whichever wall you drift near. You learn to feel it as a softness in the wheel and to not fight it so much as account for it.

Also: still air doesn't cool things. We said this about brakes; it's true of everything. No headwind to carry heat off the radiators, the bodywork, the discs. So "aero" down here ends up being mostly about getting air to the bits that need cooling, not about downforce at all. Ducts, not wings. Hmm — wait, that's not quite "aero," that's plumbing. Fine. It's plumbing that happens at speed.

Note to self: stop bolting on wing for "stability." It's drag with a press release. Ask Yasu what he actually runs back there. (Nothing. He runs nothing. The smug git.)

A fuel map is just a table that says: for this much throttle, at these revs, give the engine this much fuel — and the spark this much timing. Get it right and the engine pulls cleanly and doesn't melt. Get it wrong and you've either got a sluggish, rich, fuel-flooded lump, or a lean, hot, knocking one that's quietly damaging itself a little every lap. There's a narrow good zone and a lot of bad either side, same as everything else in this book.

The tunnel changes the recipe in two ways nobody warns you about. First: it's cold and dense down here, eleven degrees, and cold dense air has more oxygen packed into the same volume — so a map tuned up top arrives slightly lean underground. You're getting more air than the table expects. You add a touch more fuel across the board, or you live with a hot, edgy engine. Second: the still air, again — yes, again — means less cooling, so you can't lean it out for economy on the straights the way a surface car would, because "lean" also means "hotter" and you've got nothing to carry the heat away. Down here, rich is cool, and cool is alive.

So my map runs fatter than the textbook and flatter than it should be — I've given up the last sliver of straight-line power for an engine that finishes the run. You'd think you'd want the spicy map for qualifying at least. Watch what happens: one banzai lap on the lean map, and laps two through ten are a derate-y, heat-soaked apology. Hmm — wait, on a really short sprint, sure, take the spice. But "short" down here is rarer than you think. The tunnel is long. The tunnel is always longer than you think.

Note: stop chasing the dyno number. The dyno isn't eleven degrees and the dyno isn't ninety laps long. Ask Yasu for his cool-running map. (He'll trade it for the brake-bias thing I won't admit works.)

Down here the only timekeeper is a wristwatch in somebody's chalk-dusted hand, and that's almost the point. The number it gives you at the end is true and also nearly useless. It tells you the lap was 3.2 seconds slower than your best. It does not tell you where, or why, or whether the "slow" lap was actually the better-driven one with a bad tow off a wall, or the messy one that just happened to flow. The stopwatch is a verdict with no evidence attached.

What I've slowly learned to do instead: time the pieces. Pick three spots — the brake into turn one, the apex of the long left, the moment the throttle's flat onto the back straight — and feel each one as its own little event with its own quality, good or bad, independent of the others. A lap is not one thing. It's a string of maybe a dozen small decisions, and the watch just sums them and hides the working. You'd think a faster total means a better lap. Watch what happens: chase the total and you start "making time" in the safe bits and bleeding it in the scary ones, and the lap gets uglier even as the number gets prettier. Hmm — wait, sometimes the prettier number really is the better lap. The trick is you can't tell which from the number. You have to know what you did.

So I've started writing a sentence after each run — not a time, a sentence. "Turn one was late and panicked, the long left was the best it's ever felt, the back straight I lifted for no reason." That sentence is worth ten stopwatches. The watch goes in the log because Yasu makes me. The sentence goes in the notebook because it's the thing that's actually true.

Note to self: when the watch says you were fast and you know you drove like a sack of spanners — believe yourself, not the watch. The watch has been wrong before.

For the longest time I thought the engine was the thing at the end of the throttle — input here, noise there, done. Then one night in the bore, lights off, eyes useless, I realised I'd been steering by sound for half a lap without noticing. The engine isn't downstream of you. It's talking, constantly, and the dashboard is just the least interesting thing it says.

What it says: the pitch tells you the load — a flat, fat note means it's happy in the band; a thin, climbing whine means you've over-revved and you're just making heat now; a lumpy stumble low down means you asked for too much, too early, and it's drowning. The little hesitation before it pulls clean tells you the fuel map's a touch off for this temperature. A faint metallic edge under hard load — that's the knock you really, really don't want, and you learn to lift before your brain even names it. None of this is on a gauge. All of it is in your ears, if you stop filling them with worry.

And it's a conversation, not a readout — you do things, it answers, you adjust. You'd think the goal is silence, a smooth quiet engine. Watch what happens when you chase quiet: you short-shift everything, baby it, and it goes cold and sulky and gives you nothing. Hmm — wait, not loud either; loud-and-thin is the over-rev, that's bad too. What you want is loud-and-round. A full note. The engine sounds like that when it's in the place where it makes power and stays alive, and your whole job, really, this entire notebook, is just: find that note, and stay near it.

So that's the last page. Run at night, keep the tyres in their window, live at the slip-angle peak, gear it close, never fix the middle of a corner, guess the brake bias and keep editing, skip the wing, run it rich, ignore the stopwatch — and underneath all of it, listen for the round note. The car will tell you everything. You just have to be quiet enough to hear it. — C.