The first thing you learn as a night assistant is that nobody thanks you for a good night. A good night is one where the astronomer wakes up, looks at the data, and has no questions. Nine years of that teaches you a great deal about instruments and almost nothing about photography, which took me an embarrassingly long time to find out.
I ran the telescope while other people slept. That is the job, almost literally: the run is four nights, the astronomer cannot be awake for all of it, and somebody has to sit with a two-metre instrument and keep it pointing at the right piece of sky. That somebody knows which of the four axes is lying tonight, and that somebody is not on the paper.
In 2021 I moved into a third-floor flat with a balcony, a concrete parapet, and a television aerial belonging to the flat below that I have never once got out of a frame. The sky over it is what the Bortle scale calls an eight: on a good night you can see four stars, on a bad night you can see the aerial. I had a small refractor, a secondhand mount, and nine years of professional experience, and I assumed the experience would carry.
It did not carry, for a reason specific enough to be worth saying: a research instrument removes every variable I was about to meet. The mount is a pier sunk into bedrock, the optics were collimated by someone whose whole job is collimation, and the sky is a mountain. What I was expert in was noticing — and noticing is worth very little when you have no idea which of the eleven things you are noticing is the one that matters.
Nine years of running a two-metre telescope taught me nothing that survived a third-floor balcony, and it took me about five weeks to accept that.
Six weeks of failing differently
The first night produced nothing. So did the second. I kept a log out of professional habit, and it is not a story of gradual improvement — it is six weeks of one photograph failing in different ways, which is much harder to learn from than failing the same way twice. Three of those failures are the three I now see most often in other people's images.
Night nine. Four hours on polar alignment, two hundred frames, every star trailed into a short comma. I concluded the mount was faulty and priced up a replacement. The counterweight bar was a centimetre off balance, so the drive was fighting itself on one side of the meridian and coasting on the other. Eleven days to find, because I had already decided what the answer was.
Night nineteen. Clear, no moon, everything working, and a gradient across the stack like a thumb smudge. I spent four sessions taking better and better flat frames to correct it. It was skyglow — I was calibrating out a thing that was not there and leaving the thing that was.
Night twenty-nine. Four hundred and ten frames, all technically fine, and I threw away every one. That was the correct decision and I could not have told you at the time why, which is the part that bothered me.
What the forty-fourth night did differently
What changed on night forty-four was not equipment; I bought nothing between the two. I stopped starting at the top. Every session until then had begun with the thing I most wanted to fix; every session after began with the lowest thing that could be broken, checked in order whether or not I thought it was the problem — mount, focus, calibration, target, then the frames themselves. Most nights the answer is two rungs below where I would have looked.
The sky was the same on night forty-four. Everything else was in a different order.
That order is the whole method, and it is what the nine modules are. I called it the Integration Ladder because integration is the word for accumulated exposure time, and because naming it made me actually follow it.
The galaxy, in case it matters, was not a difficult one — the kind of target people photograph in their first season. That is rather the point. The difficulty was never the object and it was never the sky; those weeks were spent learning the order to check things in, from a starting position of thinking I already knew.