Effects of thread wrapping, Series 2:



In the previous article, we worked out a number of shapes that the bore of the Potter flute might have been originally.  In this article, we'll try to eliminate some of them as "impossible", and cast doubt on others as "improbable", in the hope that a process of elimination will lead us to the "most likely".

Just how bent?

Focus on the compression in the middle and lower tenons in the graph below.  In both cases, we see that the compression is not limited to the tenon itself, identified by the vertical dotted lines.  It also extends into the adjacent body wood, in both cases here to the left of the tenons.  But now focus on the upper tenon at image left.  Where is the compression to the adjacent body wood (which in this case will be to the right of the tenon)?  Is it that tiny flat spot, about 10mm wide, between X=40 and X=50mm?  Or is it that extended sloping bit down to about x=140mm?  In other words, is the section of bore between the top and mid tenon supposed to be bent like that, or is that a by-product of the strangulation at each end?

Check out the family

Imagine this flute was a human victim, mangled beyond recognition by perpetrator or perpetrators unknown.  Investigators need to know what the victim looked like.  What do they do?  Check out the family of course.  And we shall do the same.  Surely we can find some other Potter flutes that can provide some clues?  And the answer is .... yes!  Check out this graph:

Potter's previous model

The Bate Collection in Oxford holds a fabulously beautiful boxwood flute by Richard Potter, dating from 1782, just 3 years before he patented the Patent flute we're looking at in this series.  The Bate flute doesn't have a tuning slide, and has the older square keys with flat leathers for pads.  Because there is no slide, it needed some other way of dealing with the range of pitches in use at the time.  This flute has three "corps de rechange", left-hand sections of differing length, for that purpose.  So it doesn't just offer us one view of what a Potter flute LH should look like, but three.  Even more exciting is that I have the bore dimensions for it, kindly drawn up by another Australian, Ken Williams, in 1984.  Sometimes you just strike it lucky!

You can see that flute's bore illustrated above, as the pink, navy and brown traces in the LH, and the navy trace for RH and foot.  Because the three corps de rechange are of differing lengths, I've offset their starting points to keep them connected to the RH section.  The short gaps at the start of the RH and foot sections are because Ken measured in 0.2mm increments of diameter.  So we don't know exactly what diameter the start of those sections are, excepting they must be less than 0.2 larger than the next measurement point.  No problem for us - we're more interested in the trends of the bores rather than any one point.

The first thing we notice about the three LH traces is how alike they are, almost assuredly cut by the same reamer.  The second is how straight they are.  There is no sign of a sudden bend in the LH bore, as we wondered about in the strangled Potter, shown here in thicker yellow. The third thing we notice is there is only the most minor sign of bore compression and no other significant distortion.

This lead me to conclude that we are in the presence of a flute that has hardly ever been played.  And perhaps of a flute that had a shallow and loose thread wrap that had never been replaced or augmented.  If you don't play a flute, you have no need to tighten a loose wrap!  Acting on a hunch, I asked my colleague Dr Robert Bigio if he happened to have an image of it (Robert has taken images of a vast number of flutes at the Bate and other collections).  Sure enough he did and has kindly supplied it:


Richard Potter flute, boxwood & ivory, 3 corps de rechange
Image by Dr. Robert Bigio, London-based flute maker and researcher
Image reproduced here with permission of the Bate Collection, University of Oxford

(Robert is a bit scathing about this image, taken with his older photographic setup, although I suspect it hasn't sent you screaming from the room.  Or at least, you came back!  His newly-released book, Rudall, Rose & Carte: The Art of the Flute in Britain sets the standard for flute photography for the foreseeable future.

And while I'm in promotional mode, you can find out more about the Bate collection from the Bate website. And even order your very own copy of the detailed technical drawings of the Potter flute to mull over.)

As you can see, the image completely supports my prediction.  The flute is in magnificent, "as-new" condition, and the loose and shallow tenon thread is pushed back, on several tenons, exposing the wood below!  This is not a flute, it's a time capsule!  And a time capsule with an important message for us.  At least in 1782, Potter intended his LH bore to be straight, and to join the head diameter with the bore of the RH section.

He did appear to have quite a different idea of how to deal with the foot.  But that's a feature for a different study.

Potter's Son's flute

And our luck is holding.  The McGee Research Collection has another flute in the Potter family - this one made by Richard Potter's son, William Henry Potter, sometime after 1806.  William Henry had apprenticed with his father in 1774, took over in 1806 when Richard retired, and finally retired himself in 1837, dying in 1848, leaving a fortune of 30,000. 

Interesting to think that, given he started with his dad in 1774, he might have been involved in making all the flutes we've been looking at.  Here's the William Henry Potter flute:


You can see it's essentially the same as the Richard Potter we've been following, except in ebony and in slightly better condition.  Indeed, when we see the bore, we realise just how much in common they have -young Potter's flute in aqua, in the graph conveniently repeated here:

The only real clear differences we can see are that the LH top hasn't been cut off, and that the foot bore terminates in a most unusual curve.  Both flutes suffer from similar levels of strangulation at all three tenons, unlike Richard Potter's early flute in the Bate.  Because the flute under study had been shortened in the late 19th century, its upper tenon strangulation has moved downhill compared to the WH Potter, a beautiful proof that at least the compression under the tenons was not original.

So, in terms of finding out what the strangled Potter's bore looked like, we're not much further advanced.  We have one flute (or three, depending on how you regard the three corps de rechange) with a very straight bore, and two with the bent bore.  No clear winners here.  Did Richard Potter introduce a second slope when he invented his Patent flute in 1785, or is the bend in the LH section simply another symptom of severe strangulation?  Perhaps someone else out there has a Potter Patent flute in better condition they could measure and break the deadlock?

Reality Check I - Is there room for a straighter bore?

You can imagine that, if the LH bore were to be enlarged at both ends, the outside diameters of the two tenons would increase in diameter also.  Would they still fit in their sockets?  If not, then surely these theories are blown out of the water!  We need to check...

Seems like that isn't an issue for the conservative scenarios at the top, or any scenario at the bottom.  Both tenons, now unwrapped, are very free in their sockets, indeed to the point where you would say sloppy.  Measurements indicate that there is about 1mm clearance - more than a good maker would normally leave.  Potter was a very good maker, so that seems to suggest that that clearance has increased over the years.  As tenons collapse, clearance increases...

But 1mm clearance at the top tenon may not accommodate the 1.3mm increase in bore diameter we might expect to see under the boldest scenario. But there might be reason to still wonder.  This flute uses wide bands of ivory to reinforce the sockets.  Ivory seems to shrink even more than wood (it's often cracked because the wood under it didn't shrink as much), so it might be that any lack of clearance in the barrel socket could be partially due to shrunken ivory.  Certainly that socket is almost cylindrical, while the RH section's socket is tapered.  You can imagine how much the ivory has shrunk by seeing how much the barrel wood has shrunk, cracking to expose the slide inside.  Yet the ring is still firmly attached, suggesting it has shrunk at least as much as the wood, and probably more.  The crack is 1.15mm wide at the top shoulder, suggesting a reduction of about 0.37mm on diameter, or about 1.5% .  That would probably make the difference needed to accommodate an emboldened tenon.

Shrunken ivory is suggested also by the fact that the ivory on the other two sockets has been replaced by metal bands, probably after the ivory shrunk, cracked and fell off.

This stuff is tricky, isn't it.  If room-to-move proves to be a sticking point (so to speak), we may wish to test the possibilities by humidifying the barrel too!  Hmmm, indeed, why not?

Reality Check II - are the bold scenarios even feasible?

Hang on a minute - get real!  Are we really considering the bold scenarios?  Let's do a little computer modelling first to see if that kind of change is possible.... 

In our model, we'll assume the bore started as a straight taper, joining the head bore to the RH bore.  Pink in the graph below.  Then we expand the bore by a specified percentage due to humidity uptake, but let the thread constrain the area under the tenons to the original size.  We'll assume the section including the Bb and C blocks will be more-or-less immune from distortion because of the additional rigidity of these walls lent by the blocks, and because of the distance from the constrained tenons.  Then we'll assume the areas immediately adjacent to the constrained areas are only 50% constrained (because their wall thickness is about double the wall thickness of the tenons).  Then we taper the distortion between those and the immune region on the basis of proximity.  Finally, we'll reduce the damaged bore by the same percentage as we started with.

Note that there are a few other factors we could throw in, such as the thread retention grooves at the bottom of the trough, but these would be difficult to model and probably not make all that much difference to the overall picture.  They might well have a significant effect on the exact shape of the bottom of the thread trough though.

And what do we get?  Something truly astounding - a match to the as-found bore dimensions within less than 3%!  Compare the blue "as found" trace in the graph above to the brown "modelled" trace.  So even the boldest scenario does seem feasible. 

But what was that "specified percentage"?  I found in my model that I needed a 7% swelling and subsequent shrinkage rate to produce the match reported.  Is 7% change feasible?  Not in one sitting, perhaps, but in a case of serial strangulation, perfectly feasible.  Three cycles of just over 2% each would do the job nicely, Inspector. 

Now that we've seen the bold scenarios supported, doesn't that also cast some doubt onto the conservative scenarios?  Why would the damage caused by over-tight thread-bands be limited to just the tenon below?  Surely some distortion of the adjacent wood is inevitable.  Try this simple analogue.  Support a thin rule on two pencils, about 50mm (2") in from the ends, on your desk.  The pencils represent the tenon shoulders.  Now press down at each end, about 25mm (1") in from the ends of the rule.  This pressure represents the constraining thread band.  Watch the centre of the rule as you do it - does it remain still?  No, it rises as the pressure goes on.  This is because the rule has finite stiffness - it refuses to bend meekly around the pencils - the whole thing prefers to adopt a curved shape.  The same will apply to the flute section - the finite stiffness of the wood will ensure that the compression and thus the damage is taken over a broader area than just the point of pressure.  The conservative scenarios are not looking good...

How strangled?

Of course, the bolder the scenario we end up supporting, the more our "moderately strangled" flute is being reclassified as "seriously strangled".  The criteria I had proposed for the threshold of seriously strangled was 1mm reduction in diameter.  The aqua and violet traces would meet that criteria on the top tenon, and those plus the orange trace in the mid tenon.  And this is one I had classified as just a "moderately strangled" flute.  Sobering, eh?

A cold shower

Now, of course, we shouldn't necessarily expect that restoring a flute to its original shape will necessarily make it a better tuned flute for our modern purposes than it would have otherwise been.  Mr Potter didn't intend his flute to play at A 440, although with that very long tuning slide he was certainly keeping his options open.  The barrel extension on that flute is 27mm, so even if he never intended to let you see the naked metal of the head tube, he had allowed for about 27Hz variation in pitch.  So, even starting at the low pitch of the day, A440 might be within range.

By way of comparison, Ken Williams estimated that the three bodies on the Bate Potter were intended to play at ~418, 427 and 436Hz.  Our flute was a little shorter, and is now, because of the high pitch era modification it suffered, considerably shorter!  So it's a bit hard to know what to expect.  But, if we want to have any chance of understanding what he was aiming at, we need to get back to his dimensions.  And of course, the same applies to flutes by all the other makers.

What about the rings?

In these series, we've been concentrating on the compressive effects of thread bands.  But what about socket rings?  If a thread band can prove to be a barrier to expansion, surely a metal or ivory ring is even more?  Why don't we see the same compressive effects under the metal rings?

Answer is that we do.  But it's not so noticeable.  Look at the RH section in the "bore as found" graph, reproduced here.  It's the section from around X=210 to X=340mm.  At the 210 end of the section, we see a small amount of compression.  At the other end of the section, we see a thread-wrapped tenon from around X=310 to 330 with a lot more compression. 

Surely that makes no sense!  If a band of pliable thread causes problems, surely a strong band of unyielding metal has to create more?  It's a fair question, and one we need to investigate and be able to answer convincingly.  I have a few theories, half-baked at best, which I'll share with you until we can investigate more thoroughly. 

Theory 1 relies on the difference between playing and weather-induced swelling.  You'll remember from the first series that we tested two ways that moisture levels in a flute could be increased.  We found that simulated playing was a far more efficient way of getting moisture into the wood than putting the flute in a humidified environment.  The area under the tenon thread is bore, which gets nice and wet when you play.  The area under the socket ring is socket, protected from direct application of moisture by the tenon of the connecting piece.  So it only gets moist by conduction from the adjacent bore area, or from humidity in the air.

Theory 2 is that the compression under the ring is largely confined to the socket.  The socket wall is thin, and therefore relatively flexible, while the rest of the body wall is thick and therefore much stiffer.  The socket probably protects the bore by going a bit bell-mouthed.  We do see a little compression at around X=210, but nothing compared to that under the tenons.  We often come across bell-mouthing in old flute sockets.  That's the annoying situation where it's hard to get the tenon in, but once there, the joint is loose and floppy.  Now we can see why it happens.

Theory 3 relates to serial strangulation.  When the tenon shrinks, the joints get loose, and the owner adds a bit more thread to tighten them up, initiating the next cycle of serial strangulation.  But if the socket shrinks, the joints get tighter.  The owner might react by removing some thread off that tenon, but that won't affect the socket.

A little serial strangulation might apply to the sockets of this particular flute.  Note that, in both RH and foot, the original ivory socket rings have cracked off and got lost, and have been replaced by brass rings.  We don't know how well the repairer prepared those rings.  If, as I have reason to suspect, he made them undersized and hammered them on, they might well be a fresh source of compression.

Theory 4 goes to the differing nature of the thread band and ring.  When the socket wood shrinks under the ring, the ring suddenly exerts no more pressure, and shrinkage will probably stop.  The more elastic thread band probably continues to exert some pressure, so shrinkage continues longer each moist cycle.

And of course, Theory "n" tells us that all these things are involved.

But what about the rings on the foot?

Aha, reckon you got me, huh?  When we look at the first bit of the foot bore (X=330 to 350), we see no sign of compression at all.  So, how come?  Just take a look at that section of the foot in the image below:

That's where the bulge which forms the hinge block for all three foot keys, and which runs completely around the foot at this point.  Nothing is going to crush such a mighty block of wood!

That's not to say someone didn't try.  When I removed the brass ring, I found a cylindrical crack running back into that block at the diameter of the wood under the ring.  I suspect whoever replaced the missing ivory band with a brass one simply hammered it on to the tapered piece of wood remaining, causing the wood under the band to separate from the inside of the block. 

And, at this stage I'm not sure what to make of the very end of the foot bore.  It almost looks like it was intended to be horizontal at about 10.4mm for the last 45mm, from X=425 to 470.  Could that dip near the end be the result of compression under the foot ring?  We may have to look at that issue separately.

Involuntary Education

Let's imagine that we decide that the only damage this flute has suffered is to the tenons.  So we go with the ultra conservative scenario "just fix the tenons".  But let's say we were wrong.  Supposing the compression of the thread band squashed not only the tenons but a considerable part of the nearby body.  If we take the compression off the tenons, we also take it off the adjacent body area.  Any expansion to the tenons may cause the adjacent bore to enlarge.  So even just fixing the tenons might reveal that the bore has moved too.  So it's possible that, even in making the wrong call, we will still learn what really happened.  It's rather comforting isn't it - we can afford to press on even if we don't quite know what we might find.  What ever happened to that cornerstone of Tea Party ideology - the Right to Remain Ignorant?

Serial strangler clues

I had great hopes, when pulling the thread off the RH section tenon, that I might be able to find clues to a serial strangulation.  I most carefully sought out the end of the thread using a needle probe under the zoom dissecting microscope, and lifting the thread away gently with fine tweezers.  But the thread was so compressed, it broke at points where it crossed other threads, so I was not able to be sure how many separate lengths of thread were involved. 

But, when I finally reached the tenon, it became clear that the entire thread band had been replaced a number of times.  I counted at least 6 slashes in the wood that had been made with blades since the tenon had been combed.  Some straight and full length, some at wild angles, some in the form of tangential slices.  We can say with confidence that frequent vigorous thread removal and replacement was a part of its history.

A Plea for Moderation

Finally, let's here a voice in favour of a moderate expectation.  Check out the undamaged and lightly damaged flutes in the survey graph reproduced below.  The Geo Rudall Willis Fecit (dark green) seems perfectly happy to be pointing at about 18.1, but then, it's a small bore flute.  The Anon Pratten (light blue)would have come out around 18.3mm, if it weren't for the bridging taper, which might not be original.  The RC7120 (navy) seems from this graph to be heading for an opening of about 18.7mm.  Just as it's clear they have suffered little or no compression, there doesn't seem to be evidence that these flute bores were aiming for the head diameter.



At this stage, I think we have to accept we still don't know that the bore should look like, other than the conservative scenario, "flatten the tenons" in the graph reproduced below looks rather too conservative.  It might just be that we'll have to rely on recognising a credible bore shape when we see it, either from measuring and graphing it, or from its performance.

But it hasn't been wasted effort.  We've considered a lot of possibilities, and will be well armed to venture forward, antennas twitching.  Knowing that even if we achieve a moderate scenario (say the orange or red trace), our poor flute will be saved from strangulation.

But maybe you can see something I've overlooked?  Do feel free to get in touch!


In our next exciting episode, we'll look at ways of proceeding from here.  What are the options for action, and which option should we take?  We'll see that even that requires a lot of consideration.  You might remember me warning in the first series that realistic (as opposed to in extremis) studies would take a lot of thought and a long time.  Was I wrong?


My thanks to Dr. Robert Bigio for a copy of his image of the Bate Potter, and to Andrew Lamb, Curator, for permission to use it. 

On to: A Plan

or Back to McGee-flutes Index page...

  Created 5 May 2011