I'm writing this for Kjella, who can skip to the bottom if TL;DR concerning his entire future life.
It's long, detailed, and lucid for a good cause.
I free ran with a 25.5 hour period not so long ago—for three years straight, like a metronome. During the year I recorded most assiduously, I didn't deviate from my period by more than +/- 4 hours.
Note that my cycle was somewhat elliptical. I advanced more slowly during the day portion of my cycle, and more quickly during the night portion of my cycle, but over any 16.5 day interval (calendar days), my daily period averaged out to 25h25m. I couldn't even detect seasonal drift with a Canadian change in solar day length.
Before I free-ran, I had partially treated my condition with a small dose of melatonin taken mid-afternoon. This reduced my period to around 24h10m, which means I was drifting over an hour a week. Every fourth or fifth week I would cease taking the melatonin, put in a week of night mode, and in so doing, reset myself to a very early rise time, which would then inexorably advance until it became too much to bear (if I started the cycle rising at 0500, a month later I'm rising at 0900, and I'm rolling into the office at 09:45, which was as far as I could reasonably push things).
I guess I was a bit in denial about the 10 minute/day drift residue. I tinkering with every variable over a two year period: dose, time of day, coffee consumption, light exposure level, bedtime strategy (fixed or adaptive), etc. Mostly I kept my dose time in a fairly narrow window around 15:30. I suspect, in retrospect, that as my internal clock drifted my melatonin became less effective, because the dosing time became less optimal, and so there was an acceleration effect near the end of each three- or four-week compliance interval, because it was usually a day when I woke up around noon where I finally said "oh, fuck it" and suspended melatonin for a week.
This lifestyle gradually became unbearable because of a second straw: the three zombie hours every day an hour or two after taking the melatonin. Work—putting in longer hours than anyone else, because I seriously needed the social credit, return home—vegetate on couch, regain mind for one good hour before bedtime, then head punctually to bed (this period of my life was highly compliance oriented), to battle with ever-ramping insomnia, until it all fell apart again.
I put up with it while I retained hope that I was maybe one inspiring fiddle away from breaking either of the two straws. But I never did.
The free-running period was not gainfully employed, but by this point, my quality of life was near zero, so what the hell? I didn't really expect it to last more than six months and my theory was that I would be productive free-running, and that the whole thing would be massively inconvenient, but I'd finally have time for both a working life and a personal life. All I needed was some kind of work I could do during my long nights in the night phase of my cycle, and then pack my social life into the other half.
It didn't work out that way, because N24 was only half the diagnosis. This was the most important thing I've learned about my condition in thirty years.
My "formal" personal diagnosis now has the august title: disordered circadian rhythm—induced split-cognitive-modality syndrome.
What I learned was that I was in full possession of my mental apparatus for three or four days out of every sixteen, the days when I was best aligned with day mode. This shocked the hell out of me, because over thirty years, I'd never observed my period having any connection to the solar day. This included a three-week bicycle trip in my twenties through Washington and Oregon—in the month of June, during which we never once saw a cloud after rising at the crack of dawn every day—yet we finally had to cancel the California leg, because I simply couldn't get up before noon under any coercion for even one more day. I was probably a week into Tokyo time already, and unlike a normal person whose jet lag declines, mine continues to escalate, until I finally capitulate.
But there I was free-running, and clearly the elliptical component of my gyre was synchronised—by some unknown zeitgeber (not necessarily sunlight)—to the solar day.
The two drugs I had in my arsenal during this time were modafinil and nortriptyline. Certain circuits in my brain would go on hiatus during night mode, and was never able to work productively on any kind of STEM task. What I could do was read at a high retention level (though I couldn't deeply contemplate the material), and sometimes write (when you write, you just use what you've got, in some ways it's a forgiving medium).
The weirdest effect of all was that having read three to five books over the worst of night mode, when I finally swung around to day mode, all my suppressed subconscious thinking about the material would burst to the surface in a near-to manic crush. I recorded many of these days with my voice recorder, practically tripping over words, as the ideas surfaced faster than I could spool them out. This happened with great regularity at the beginning of each new cycle.
It was pretty clear that it was actually some manner of brain interconnection that was going haywire as my sleep became offset from the solar day. Mental functions that are still taking place, but which you can not access are still highly effective at suppressing remunerative activities that pay better than supervising a gas bar at 04:00.
What I began to suspect is that the amplitude of my sleep-wake cycle diminished when my phase was misaligned with the solar day. I'm not going to review my notes on this, but there's some kind of linear sleep pressure ramp for as long as you stay awake, then there's another ramp of something that makes your body ignore this mounting sleep pressure (so you don't potato right after dinner), and I think the DLMO response in the late evening (if you manage to avoid blue light) clears away this inhibitor, so that suddenly you feel the full force of all that built up sleep pressure, and down you go (modulo caffeine, which I think directly alleviates this). And then over night this sleep pressure chemical is cleared away, and there's some signal early in the morning that insists "hey, buddy, you might still feel tired, but forget about it—you're very awake now".
There's lots of ways to fuck this picture up, but when it's running in good trim, that's basically how it goes: a strong oscillation between "you are feeling sleepy" and "you are not feeling sleepy".
I suspect that I was experiencing 100% of this oscillation in my four good days, maybe 70% of this amplitude in my shoulder days, and perhaps 40% of the full amplitude at the worst of my cycle.
I always got eight hours of consolidated sleep (that part of my sleep architecture is rock solid), but depending on my phase, it didn't necessarily accomplish much.
What I discovered was that modafinil could boost the awake side of this signal to 100%, but then the sleep side of this signal was boosted so high, I actually couldn't sleep properly at all. A small dose of nortriptyline an hour before bedtime usually secured two hours of deep dream sleep right before waking (which was often later than normal), and then I would feel wonderfully rested, but nevertheless, my eyelids would never quite make it above half mast for the whole waking day.
Great, on day mode my sleep signal oscillates 0 to 100; in night mode on modafinil, my sleep signal oscillates between 60 and 100; in night mode on nortriptyline, it oscillates between 0 and 40. That's how it felt.
I actually found a tiny mitigation. I would alternate the modafinil and the nortriptyline, having one period where my sleep signal oscillated 0—40 following by another period where my sleep signal oscillated 60—100. Over fifty-two hours (night mode had 26 hour periods), I would experience something close to deep sleep on one side, and experience something close to be completely awake on the other side. While this decreased my misery substantially, it only increased my function modestly. It was also the closest I came to confirming my diminishing amplitude theory.
I was getting pretty sick of the free-running ordeal after three years, and was starting to contemplate my next desperate lunge into the void, when one day I wandered into a supplement store and saw sustained release melatonin on the shelf for the first time. I new that melatonin researchers used SR mel in their sleep studies all the time, but had never seen it in retail before, so I snatched it up.
I figured it would probably not be any worse than regular melatonin, and possibly better. I waited until my sleep cycle had me waking at 0030 and started taking these new pills at about a +9 hour offset after waking. Sure enough, it tracked my anticipated path for several weeks, until one day I woke up at 10:00 and I went to myself "oh, oh, same old, same old".
But then the next day I woke up at 10:00 again. And the day after that at 09:45. And an amazing week later—with steadily mounting amazing—at 09:00. And wow, for the first time in thirty years I was fully locked onto the 24-hour day.
Now these were 3 mg doses in tiny #4 capsules and I was manually shaking out approximately 2/3rds of the drug to get something closer to 1 mg, and this was by no means precise. After a few months I came up with a better way to trim the dose by mixing in a 1:5 ratio with another powder and repacking the capsules, but it still wasn't precise. The net result is that for the first year there was a lot of wobble, but it was mostly steady.
Then I ran out of my own supply of empty #4 capsules. When I tried to buy more (you get some severe looks) I was asked "why don't you get a compounding pharmacy to do this for you?"
Here's the thing. I hadn't ever had any useful advice from a sleep doctor and I had become so lone wolf in my hamster regime that I ceased to think that the medical establishment had anything to offer me. So my response was "oh, yeah, medical technology, I've heard of this, it might even work." I had custom SR mel capsules ordered to my own specifications from the most reputable compounding pharmacy in town within two hours.
I didn't know this, but the pharmacist was feeling generous, and gave me a very low price per pill for the first order of $0.20 per pill, plus a flat $30 order fee. They thought I was ask for 150 pills at most. But I thought to myself, "well I really don't know the dose, and it would be good to experiment in a scientific way, so maybe I should get small pills so I can try some doses that are closely spaced". Pretty soon I had mentally worked my way up from 200 to 400 to 600 and finally "you know what, just make it an even 1000". This was good, because he never again quoted a price of less than $0.60 per pill, and I've heard about that "whole day" filling my first order more than once since (he did once mention his rate of production afterwards and I think it was just a very tedious four hours).
My new pills were a tiny 100 ug each. I started this new regime at 300 ug (three at a time) and it sort of worked, but within a few weeks I wiped out. Back through a week of night mode, next stop 400 ug. A little more stable, but soon wiped out again. Back through a week of night mode, next stop 600 ug. Magic!
Seemed to be working 100%. For a couple of months. Then, to my surprise, I wiped out again. Damn! Back through night mode, next stop 800 ug (I had by then procured a year's supply to 600 ug pills, so I wasn't taking eight at a time of my precious quantums). Then I managed day mode for 18 straight modes, with a little scare in the month of June.
June. If only I had known. Because now my notes show that I've had a scare every June for three straight years, and I hazily recall very bad Junes long ago when I was on less effective melatonin formulations. Probably an interaction with my antihistamine, which I take in copious quantities for exactly five weeks every year, usually starting in the last week of May.
This year, in June, I blew up completely for the first time since forever. So I decided, maybe another 10% for safety, so I bumped 750 ug up to 825 ug (I had by now some 75 ug capsules in hand to experiment with taking a small extra dose at bedtime, but it didn't seem to help, and I had refilled at 750 ug instead of 600+100+100, figuring that was a safe decrement). I crashed and burned almost immediately. Okay, I thought, if 875 ug managed to blow me up, maybe I'm dangerously close already to the high end of the effective window. So I dropped back down to 600 ug, and now I'm feeling better than I have in years.
The only reason I left 600 ug (which had felt great at the time) was because of another June catastrophe. Man. That moment where I went "wait a minute, it was Mrs June, in the tomato garden, with the knife—all along!". Fooey. There went another year of my life with my condition less than optimally controlled because I failed to play enough Clue as a teenager.
I'm joking. This stuff is really hard. Self-observation, no matter how obsessive, is not a clean signal, and the clock rate on making any adjustment is one to three months (one month if something blows up spectacularly, or three months to decide it hasn't, and you feel stable again against a known baseline).
Most of my adjustments were made to control side effects, random midday sleep events between noon at 20:00 which I used to call "naps" but which I now refer to as "close encounters with comatose". In English, the metaphor is that we "fall" asleep. You feel sleep draping down heavily upon you. Coma doesn't arrive like that. Coma feels like you're in a tall building, and the lights go out on the ground floor, then the second floor, then the third floor, etc. So you're on the 14th floor, and nominally you still have consciousness, but floors 1-10 are all black (these parts of the brain did not wait for a coordinately "sleep" state to come along, they just shut off regardless), and you certainly can't do anything worthwhile feeling like this, it's taking a painful amount of energy just to stay vertical, and you know that two or three hours on a pillow getting sleep you don't need will fix the problem. This is why I stopped calling these events "unwelcome naps", because they arrive as a black hand reaching up from below. This is not like resisting normal sleep. It's like that old television trope where someone has a concussion and everyone is panicking and pleading and wheedling to prevent the consussed person from drifting off to the land of Old Yeller by falling into a final, permanent sleep before the paramedics finally arrive (if you're concussed so badly that there's also a cranial haemorrhage the coma was probably on it's way, regardless, and it never had much to do with a simple concussion in the first place). I add this just in case some first year college student presumes that their own sleep management has resulted in the One True Way to Suffer, and if you can do it, so can everyone else (news flash: there are more things in heaven and earth, Snowflake, than are dreamed of in your philosophy).
I've only had one of these coma rising-from-below events since returning to 600 ug. I'm exuberantly ecstatic about this. Pity about the two years I lost after I failed the Clue stick.
Concerning the SCN, my understanding is that there are two separate clocks (with all the same genetic mechanisms, but different entrainment factors), one that tends to track dawn and another that tends to track dusk. In mammals with seasonal mating cycles, mating season is triggered by a specific change in phase relationship between these similar, coupled oscillators (due to changing day length).
My best guess is that I have a genetic defect in one of these two oscillators, and that when I'm untreated, the crazy oscillator runs the show (mostly). However, the non-crazy oscillator does try to stick to the day/night cycle, causing the two oscillators to have a changing phase relationship throughout my 16 day gyre, which contributes to the flattening of my sleep-wake cycle during the portion of my cycle (night mode) when the two oscillators are maximally at dogs and cats.
This theory potentially explains everything, especially the four hour jump forward I usually experience coming out of night mode.
Imagine you have two axle pendulums such they can go around a full 360 degrees if you really hammer them. They are side by side, and have some kind of magnetic coupling between them, where they want to be 120Âdegrees out of phase. But the gravitational normal (weird planet) of one of the two pendulums starts to rotate around. Suppose the the afflicted pendulum begins to tug the unafflicted pendulum forward due to their magnetic coupling, which gets worse and worse as their gravitational normals further diverge (but never quite causes the unafflicted pendulum to break free from it's primary gravitational normal).
Then the afflicted pendulum passes through maximal divergence (gravitational normals 180 degrees apart). Suddenly the coupling between the pendulums is pulling the unafflicted pendulum back to where it wanted to go in the first place rather than further away, so now it jumps backward fast.
That would explain all the elliptical phenomenology I've observed in my free-running sleep cycle. But of course, it's hard to corroborate this theory beyond napkin speculation, even with thirty years of accrued data. Life is never a clean data stream.
These Nobel prizes are well deserved, but man is there ever a lot more to fill in, yet.
Exhibit A: the melatonin PRC.
This is a curve that purports to tell you whether melatonin will advance or retard your circadian rhythm based on when you take it. If you're thinking about the SCN, maybe you think this curve informs you about which hours of the day (measured relative to body temperature minimum, usually experienced about three hours before waking) that melatonin accelerates or retards the appropriate clock genes. Wrong! It's a trap!
Every melatonin PRC I've ever seen is a dose-response curve, where the effective on the circadian clock is measured relative to one specific formulation of exogenous melatonin.
One dose doesn't work right, and you want to math your way to a better answer? Fuggedaboutit. That information is not present in these curves. You can do this math, but only after you make a bunch of shit up about how you think maybe the larger family of curves might behave.
There is not curve I've ever seen relating blood melatonin concentration and the net transcription rate of your genetic clock (from such a curve, one would stand a far better chance of mathing out a good solution from several poor solutions).
It might not even be this simple, where melatonin levels behave like a predictable transcription gas pedal. (The simple model is actually rate=f(m,p) where m is melatonin concentration, and p is circadian phase, with p divided into two twelve-hour blocks, one block where df/dm has positive slope and the other block where df/dm has negative slope. Oh come on, you didn't think the marble would settle in the middle without a hill on either side, did you? The polarity flips at body temperature minimum, or close to it, and then again 12 hours later, in the late afternoon.)
This is why I haven't pursued my personal half-broken SCN hypothesis much further. Because the first thing you have to do is start guessing, as so much is still not known. The PRC you have (dose response) is not the PRC you want (blood concentration response). Or at least, not the last time I looked.
This field changes rapidly these days. Chronobiology is a going concern.
Maybe if I'm lucky, I'll live half my adult life on the right side of the moon yet. I'm recently at my highest point of optimism about this possibility in thirty years.
One final note for Kjella. In my twenties I experienced some abnormal "modes" where I would have a week with 24 hour waking periods followed by deep 12-16 hour sleeps. In my thirties I would often reset my circadian oscillator by pulling 28 hour waking days (with no fatigue experienced until the last hour). That stopped working in my forties (or maybe I just wasn't tough enough any longer). From my own long experience, I would suggest that these are probably abnormal overtones, rather than defining parameters. The younger you are, the more you can fuck up your self-observation with adrenaline. As that aspect of your physiology subsides with age, the true circadian component becomes a larger and more reliable part of the signal.
You can't possibly make melatonin therapy work until you figure out for yourself how to determine either your body temperature minimum or your dim-light melatonin onset (because you dose needs to be timed, every day, relative to one of these correlated reference points).
The reliable signal for me was waking with the feeling that I had just had an hour of deep, satisfying dream sleep. This happens for me about once every three days, on average, but it's enough to track the underlying curve. Wake time is about 300% more reliable for assessing true sleep phase than time of falling asleep. For me, the reliable indicator was waking naturally, after dream sleep, feeling refreshed. Yours could be completely different.
It could take you two years just to figure this one thing out. But I would stick with it because, trust me on this one, thirty years from now, you'll be glad you did.
Or maybe you have a better sleep clinic available than I had, and they'll give you a daily saliva test kit every day for a month, or something else that measures this key parameter directly.