At the Bluebell, the Signals & Telegraph department asked us to make 6 right angled brackets to hold signal wire pulley wheels on the Ouse bridge, as part of the work to replace the point from the single line at Sheffield Park. We got a specification, with sizes and positions of various holes, including two holes 2½" apart through which the pulley wheels were to be bolted with ¼" bolts. So last time we were there I cut the metal, marked out and pilot drilled the holes, drilled the base holes to full size, and then we heated up and bent the metal as instructed. ("Cut out the tabs and then fold along the dotted lines" is far more fun when you're doing it with ¼" steel rather than mere cardboard)
Today I finished off rounding the tops off with an angle grinder, and was taught a new trick, something I'd never even thought of doing. Apparently it's considered acceptable practise to take a shortcut by chopping the corner off the metal with the angle grinder. You just push the edge in slowly and gouge your way through. It's not quite as easy as a knife through butter, but you do end up with the chunk you chop off glowing red hot, which is more fun than just the normal shower of sparks, and a lot faster and less tiring than than brute force grinding your way down.
So then my dad presented me with a pulley wheel assembly, which the S & T had given him that morning as a sample.
So what had been nice routine metalworking job turned into quite a taxing exercise which might even fail, because the revised location of the lower hole would now intersect the existing pilot hole. Drills don't like this. They try to fall into the existing pilot. So for starters it involved a lot of jiggling of the sample to try to push the tolerances as close as possible on the top bolt hole, to move the bottom hole as clear as possible, to work out what could be done. Then committing to a course of action, by drilling out the top hole to the exact size of the bolt, bolting the sample to each piece in turn, spotting through the bottom hole with as off centre as possible, then placing each on the pillar drill and coaxing the drill through where I wanted it, not where the drill might feel like. One the first, the drill got part way through then drifted off and jammed. Only one more chance - drill from the other side. Fortunately did get through (just), and the sample bolted on, admittedly with the bolt askew, but it worked. The other five were a lot calmer, and all went through first time.
In the end what was surprising to me was the amount of error on my original pilot holes, revealed because the final holes were effectively drilled via jig, rather than measurement. I'd taken great care with my measuring and drilling, yet I'd managed to produce errors of about 3/8ths of an inch in my original inter-hole distance. I figure that next time I'll make a jig first, then use that for all hole spaces. Heck - next time I'll insist on a regression test, er sample, from the customer, and work to that.