I've spoken before about the trials and tribulations of diagnosing the dreaded 'intermittent fault' - and I never ceased to be amazed at the sheer variety of ways in which sod's law wreaks havoc.
Whenever a client who's suffering an excess of sod's law turns up in the workshop the opening gambit is always the same - the client has a problem, evidently, but as soon as they start to look for it, it disappears. The moment they think that all is well again - it reappears.
It's a safe bet that when they eventually get around to showing the problem to someone who might be able to find it and fix it, it cannot be found.

The latest example was almost breathtaking in its mechanical deviousness.
The problem, as described by the player, was that every now and then the horn, a vintage Martin tenor, would lose everything. One moment it would be fine - but a trip up to the top notes would sometimes result in an awful howl. More curiously, it would sometimes suddenly recover in mid squeal.
Clearly this was not a tenable situation for a horn player on a gig - but there appeared to be no apparent fault with the instrument.

It seemed most likely that the problem lay around the top end of the horn, perhaps with the octave key mechanism...a sticky or bent key perhaps - although this wouldn't stop the horn dead, you'd at least be able to get a few notes out of it.
I could see nothing wrong, so I contemplated a loose crook socket. A leak here could be significant, and it could also open and close as the horn was moved to and fro...but then it would have to be a very significant leak to completely stop the horn, and by significant I mean that the socket would have to be practically falling off.

I found no such problem, so I began an examination of the tone holes. On a Martin these are soldered on, and there's a curious malady that can affect the integrity of the soldered joint (known as Selective Galvanic Corrosion) whereby the solder oxidises and the joint fails. A decent sized gap under a tone hole might be prone to opening and closing as a key is pressed - but again it would have to be quite a fair old gap to have such a drastic effect.
I couldn't spot any problems, and a general inspection of the action's regulation revealing nothing untoward.

It was time to blow the horn to see if I could duplicate the problem.
So, I blew...and I blew....and I blew. All I got out of the horn was that typically rich, vintage Martin sound.
I concentrated on the top notes, running chromatic licks over the palm keys - still no problems.

And then it happened. Halfway through a lick the horn simply stopped dead in its tracks and emitted a most awful noise.
I carefully moved the horn out of my mouth and swung it so that I could peer at the action. I couldn't see a thing wrong with it. I put the horn back in my mouth and, as expected, it blew fine.
Back to blowing then.
It took me a good few minutes before I was able to reproduce the problem, and as soon as it occurred I stopped blowing and very, very carefully moved the horn out of my mouth - keeping my fingers in the exact positions they were in when I'd stopped blowing.
As I slowly moved the horn I felt a slight knock. Something had moved! But what?
This narrowed down the field of possibilities. What I was looking for now was a mechanism that was sticking, but one that was capable of unsticking itself...which hinted at it being sprung. A stiff key then.
Well, I waggled keys this way and that. I poked, I prodded, I wiggled - but nothing stuck.

Casting my eyes over the top key stack I now noticed that the stack rod screw was protruding from the top pillar.
My first thought was that this meant it had come unscrewed from the lower pillar, and that the end of whatever key was positioned there would be flapping about in the breeze. It hadn't come unscrewed though, not fully at least, and the key that sat there seemed quite snug up against the pillar.
'Self extracting' rod screws are a common problem. You often find them, surprisingly enough, on brand new horns. The action is new, and therefore unworn and relatively tight - and if a grease has been used to lubricate the action it's often the case that the friction of the tight action and the grease combined is enough to work the rod screw loose over a period of time.
There could be other factors too - vibration, dirt in the key barrel, or even a screw that simply wasn't done up tight in the first place.
On older instruments the reason for friction is far more likely to be down to worn key barrels being able to twist, and thus rub on the screw, or slightly bent key barrels.

Either way, this appeared to be a simple problem to fix (just tighten the screw back up) and not really related to the problem - so I reached for a screwdriver.
I found my access to the screw head blocked though, by the arm of the side F key. In order to get to the screw head I'd have to first remove the F key - and the screw head was really rather close to the key arm. In fact...almost touching??

I love a good Eureka moment!
Nothing beats that wonderful feeling of visualisation. Just as thrilling is the subsequent examination of the circumstances that have to come together in unison to create 'the problem'.
In this particular case there were several factors in play.
The first was that the top stack rod screw had worked its way out and that it was touching against the side F key arm.
This wasn't the nub of the problem though; had that been all it was then the F key would have stuck once and stayed put, until such times as the stack screw was tightened up. No, it was much more complicated than that.
The head of the screw was somewhat chewed up. Years of use and abuse had left the head rather uneven, with a distinct rough burr on it. The screw had worked itself out until it had touched the F key arm, and then stopped - or almost stopped. As I pressed various keys down on the stack I could see the head of the screw turning....but every so often, as I released keys, it would turn the other way...as though re-tightening itself.
The chewed up tip of the screw was sharp enough to catch the F key arm and hold it open, but the moment you pressed and released certain keys at random the screw moved backwards, released the grip the and the F key could fall home.
Even more dastardly, when the screw had locked the F key open the slightest stress on the horn's body was enough to give the F arm just enough space to free itself, and close...which was why the problem could suddenly disappear halfway through the awful howl created by an unexpectedly open side key - and I doubt you can get any more intermittent than that!