I marked the dropout with a marker pen so that when I marked the required clearance around the nut and derailleur, my scribed lines would be easier to see.

I then scribed a line to round off the inside corner.

I had to file carefully in the vice to avoid nicking the other surfaces.

scribing detail.

Closer..

nearly there..

almost finished.

I tried to get a smooth curve.

Checking to see if it will pass - the nut I used here is quite large, over 19mm. I thought it would be best to check the fit with the "worst case scenario" size.

The 1st drive side dropout became a template for the non-driveside dropouts, and vice-versa, with flat plate against flat plate. I scribed a line tracing the corner.

the line is marked.

Marking up a batch to be filed.

Before (top) and after (bottom). I'll clean up the stray brass a little and try and remove that broken bolt in the rack mount. Never use a blunt tap.. I bought a much higher quality tap to finish all the other threads properly.

Filing in the vice to the template dropout.

Almost matching now.

Will she pass? Yes!

I filed the inside of the track ends flat and parallel, smoothing out any lumpy brass. I used a round file in the curve.

On a few of the dropouts I cheated and touched in a little brass from inside if there were any small gaps (come on, I am a newbie). This obviously had to be cleaned up with files too.

A flock of filed dropouts.

Tapping M5 threads for the rack & mudguard mounts.

Finished threads

After tapping all the threads with cutting oil in the drill press chuck, I got pretty grubby!

I then had to degrease, wash and dry the dropouts ready for brazing.

I dried them with paper towels after using detergent and warm water to remove the degreaser.

I sanded the mating faces so the brass would take as well as I hoped it would.

Pre-braze cleanup completed.

I took this opportunity to file off the marks from the waterjet cutting process, as it would be much more difficult after brazing.

More filing.

..It took a while..

I took a 2mm drill and cleaned out all the abrasive paper debris from the pin holes by hand.

Next, I fluxed up all the dropouts with Tenacity 125 flux, the same stuff they use in the autobrazers at work. I added some more after this photo was taken.

My NASA-grade Wilkinson FX497DBJ dropout brazing jig is set up in the vice. This hi-tech model allows all-round flame access.

I tacked the ends first to try and eliminate lifting of the track faces. This was about midnight.

Moving the flame around the dropout.

It was a little challenging to control the heat whilst looking for the brass to flow all the way through.

I moved the torch around underneath to try and spread the heat evenly.

My workshop-mate Ryan took these pictures. That's his torch and glasses too. Should've remembered that dust masks don't work with tinted glasses - some of your breath is channeled up around the sides of your nose and steams them up. At least in London in January it does.

It took some time to braze 12 of these..

..But I seemed to get better at it as I went along.

Nearly finished..

Filing the track slot flat and square after brazing.

I decided to get the dropouts shotblasted as the flux is not water soluble and there was a little blackening here and there.

Trial-fit to my road bike wheel.

I need to carefully file a new curve profile something like this so that the nut can pass the derailleur mount. A bit of weight loss won't hurt..