People ask how long it takes to make a frame and I always give them the wrong number. Nine weeks is the honest answer for the bicycle. For the frame itself it is ninety-six hours, and almost none of them look like work.
Monday, eight o’clock. The mould for a Col 54 comes out of the rack and goes on the bench under the lamp. It is two halves of 7075 aluminium, machined here in 2019 to replace the one I cut in Cran-Gevrier with a router and more optimism than skill. It weighs sixty-one kilogrammes and it has made four hundred and eleven frames.
Beside it, on the green mat, is the kit: between four hundred and sixty and five hundred and twenty pieces of pre-preg, cut on the plotter the previous Friday and kept at minus eighteen until an hour before we start. Each piece has a number written on its backing paper in pencil. The numbers correspond to a lay-up schedule that runs to fourteen pages for this size, and the schedule is the actual product. The mould is a way of holding it still.
What the first day is
The first day is the down tube and the bottom bracket, which is to say the first day is most of the frame. T1000 at ±45° for torsion, unidirectional M40X along the tube for the bending you feel when you stand up, and a local reinforcement at the shell that is eleven plies thick and about the size of a credit card. Get that wrong and the bicycle will still be a bicycle. It will just be a bicycle that nobody is happy on.
Nadia has laid up the down tube of every Col we have made since 2021. She does it in silence, with the lamp low, and she will not talk to you while she is doing it. I have watched her place a ply, look at it, take it off and place it again four millimetres to the left. When I asked her once what the difference was she said “nothing, probably”, and put it back the second way.
A frame that comes out wrong comes out wrong in the autoclave, and by then it is too late to be curious. So we are curious on Monday.
By Monday evening the lower half of the mould has a down tube and a bottom bracket in it and looks, to anyone who has not seen one before, like a mess of black cloth in an expensive tray.
Tuesday and Wednesday
The top tube and seat tube on Tuesday, the chainstays and the head tube on Wednesday. The chainstays are the fiddly ones: they are small, they are asymmetric, and the drive side takes a different schedule from the non-drive side because the chain pulls on one and not on the other. Every builder who has left us has said the chainstays were the part that took longest to learn.
Wednesday afternoon is the bladder. A silicone bladder goes down each tube, the two halves of the mould close on twenty-four bolts torqued in a sequence, and the whole thing is wheeled to the autoclave. This is the only twenty minutes of the four days that has any urgency in it.
The cure, and the cutting
Four hours at 125 °C, three bar inside the bladder, six outside. Then a ramp down that takes six hours because the mould is sixty-one kilogrammes of aluminium and it does not want to cool evenly. Thursday morning the bolts come off. The frame comes out warm, grey, slightly rough at the parting line, and 782 grams.
Then it is measured, faced, reamed, and it goes on the plate against the drawing. Eleven dimensions. If any of them is outside tolerance the frame is cut in half on the bandsaw and we look at the section, because a frame that is wrong is worth more to us as information than as a bicycle. We have cut nine frames in half this year. Four of them taught us something.
The other four hundred and two went to paint.
Céline Mareau · Annecy · 04 February 2026