Effects of thread wrapping, Series 2:

A Test Tube




The bores of the left hand sections of old flutes often display a bulge near the middle.  That wouldn't be a straight-forward shape to achieve with simple reamers.  We need to know whether that bulge was intentional, or could it be a by-product of tenon thread wrapping compression? 

A test flute tube was designed and built, wrapped at both ends with thread and submitted to humidity cycling.  The experiment confirmed that thread wrapping causes distortion beyond the tenons, with distortion detectable right to the centre of the piece.  Further, the nature of the distortion fits the general pattern seen in the flute under test, and old flutes in general.  The experiment also suggested that tenon thread wrapping may be responsible for causing ovality.


It's taken some time and effort to prove that thread wrapped around tenons can cause dramatic local bore compression, but that has been now shown beyond doubt.  (If any quibbles remain in your mind, you should backtrack and read the earlier articles in this and the previous series.)

But once that is accepted, we need to find out where that distortion ends.  Is it limited to under the wrap itself?  Does it extend to the areas immediately adjacent?  Does it extend to the middle of the piece, or even beyond the middle?  And, in the case of the LH section with tenons at both ends, could it be responsible for the bulge we often see near the middle of the section?  This experiment is intended to explore these questions.

The Bulge

So what is this bulge?  You can see it peaking at around X=140mm, on the thick navy trace on the graph below.

Intentionally making such a bulge in the bore would require the use of a compound reamer, several short straight reamers or other special tools.

The Experiment

For the experiment, I decided to make a simplified model of a flute LH section.  Like the test tenon in Series 1, it would be made lightly and treated harshly, in order to get first indications within an acceptably short timeframe.  But, unlike the test tenon, it would have a central body with a tenon at both ends, just like a real LH section.  To make it easier to track and see changes, the bore and outside would be plain cylindrical.

I chose a leftover 200mm (8") piece of an Australian acacia, either Brigalow or Lancewood from southern Queensland.  I often use these species for experimental purposes, such as new models.  They are a little heavier than boxwood, but not as dense as African Blackwood, cocus, gidgee etc.  "Averaged flute wood", if you like.  I have made a few full flutes from such acacias; they work fine and behave as you'd expect.

I bored the piece to head-joint diameter (19mm, 3/4"), then turned the outside to a nominal 23mm, thus leaving 2mm walls, half the typical 4mm of a typical flute.  That alone should speed up response to humidity cycling!  I turned 21mm OD, 30mm long tenons on each end.  Woah, that's a tenon wall thickness of only 1mm!  On one end I cut a very shallow trough for thread, and combed the wood beneath.  On the other I merely did the combing.  That difference might tell us something.

I wrapped both ends with 200 turns of light polyester-cotton sewing thread.  While 200 turns sounds a lot, it's probably less than would fit in a typical 19th century flute trough.


As I measured up the test flute tube for the lab book, I realised I needed to tighten up on the loose words we often use to describe flute parts.  Let's see if these work for you.  Visualise the LH section of a flute, with a tenon on each end.  In this article, and starting at one end, I'm calling them:

  • the Tip (of the tenon), abbreviated in the graph coming later as T,

  • the Lapping, abbreviated as L,

  • the Neck, being the last part of the tenon before it joins the body, N,

  • the Shoulder, being the first part of the body adjacent to the neck, S,

The only new expression here is Neck, although I'm also proposing to reserve the word Shoulder for the start of the body, rather than the neck of the tenon, which is sometimes referred to as the tenon shoulder.  I'll be interested in your reaction to this terminology.  Please feel free to offer alternatives - we could really benefit from standardising terms for these parts.

Since this piece has two tenons, we have TL & TR, LL & LR, etc, where the L and R mean left and right.  That leaves the remaining parts of the main body of the piece.  Apart from the two Shoulders, I've identified three other measurement places, the middle, "1/2", and the two quarters "1/4" and "3/4".  That gives us a total of 11 measurement points.


It became very clear in our previous experiments that ovality is an important part of any wood movement in flutes.  Consequently, I decided to measure both the maximum and minimum at each point.  That now makes it 22 measurements. 

Management data

In addition to the diameter measurements, we need to log some management data:

  • Time of measurement, in case I need to correlate with the datalogger recording RH and Temperature.

  • Date, to keep track of days elapsed

  • Status - what's been going on, and

  • Weight, being our measure of moisture content.

That brings the daily ritual to 26 data points.  When we multiply that by 18 days of measurements, it becomes 468 physical measurements.  Plus designing a spreadsheet to accept, process and present the data, all that data entry, analysis and writing up.  I sure hope you're appreciative of all this effort! 

Oh well...

The outcome

For greater clarity, I decided to manipulate the mass of data to make it physically represent the object.  In the graph below, you'll see a crude representation of the piece, with the tenons at each end, and the larger body in the middle.  As discussed previously the L and R tenons are shown as having Tips, Lapping and Necks.  The body has a Shoulder at each end, and between them we have the 1/4, 1/2 and 3/4 data points.

You'll see that, even despite the simplified representational presentation, it's a mess of data.  Don't panic, we'll cut to the central story in a minute.  Looking down the Legend on the right of the graph you can see what we did to the poor thing:

  • On day 1, the thick navy blue trace illustrates what it looked like for a start.

  • Then followed three days of humidification, in air saturated with water vapour.  Weight rose by 7%.  The aqua trace illustrates the result.  Note that the body, tips and shoulders swelled as you'd expect, but the thread actually compressed the two lapping areas.

  • Then two days of drying in the desiccator, taking weight down to -4%.  Thin brown trace shows the body curved, with the tenons now heavily compressed.

  • Followed by two more days of humidification.  As the thin cobalt blue trace shows, the tenons are now in real trouble, but the body is still trying its best to swell.

  • Drying again to -4% in the desiccator, and then airing for two days to let the piece return to its original weight and moisture content.  The yellow trace confirms the tenons are now permanently compressed.

  • I decided at this point to up the ante by rewinding the tenon threads.  My first experience as a serial strangler!

  • Two more days drying, followed by 3 days of humidification.  Then nudging the moisture content just below zero in the desiccator and a day to air brings us up to Day 18.

  • Then, to be absolutely sure it had re-equilibrated to atmosphere, a further 17 days lying around on the bench!

The tenons

You'll notice some asymmetry between the Left and Right tenons - the left being much more compressed.  It was the one in which I had cut a thread trough, whereas the right had one I'd just combed the surface of the tenon.  It confirms the probable - the thinner the wood under the thread lapping, the more compressed the tenon will be.

(You might be wondering about the intense V shape the tenons appear to have taken on.  That is illusory of course.  We only had three data points in that area - the Tip, Lapping and Neck, so a V shape is inevitable.  In real life, the tenon distortion would be more a wide shallow U shape.)

It's hard to make out where the tenons ended up in that mash of data, so let's zoom in on just the tenons and just the first and last days.  Both started out around 21mm as shown in navy in the graph below.  At the end of the experiment, the maximum diameters, minimum diameters and average diameters are shown in pink, violet and orange respectively.

Just as in the test tenon experiment, we can see that the thread has compressed the wood under the lapping (LL and LR columns).  But we can also see that the tenons have also gone quite oval (the maxima are considerably larger than the minima).  Is that an artefact of the tenon wrapping, or has something else caused it?  I think we might want to delve into the matter of ovality in greater depth.

And the body?

Now let's turn out attention to how the central body section fared.  Again we zoom in, and just show the first and last days.  The original dimensions are in navy; the other three traces show how it ended up.  You'll notice that at this zoom level, we can see that my hand turned cylinder wasn't exactly a perfect cylinder!  Boy, it was quick though!

The pink trace shows the average of the maxima and minima and tells us what we've suspected.  Yes, the compression on the tenons has been transmitted to the central body and is causing it to now taper mildly in both directions, from the centre outward.  So the bulge in the middle of the section as we see in the Potter flute can be explained by tenon compression at both ends.

Indeed, note that the maximum and average measurements at the centre are now greater than the original diameter at that point, and the minimum diameter is not much less than the original.  Remember the analogy I'd speculated on in an earlier article?  If you support a thin rule on two pencils 30mm in from the ends (simulating the tenon shoulders), and then press down 15mm from the ends (simulating the tenon wrap), what does the centre of the rule do?  It bulges upwards, because the finite stiffness of the rule won't allow it to just bend meekly over the pencils.  So, not only is our cylinder becoming barrel shaped, the widest point of the barrel is now larger than originally.  It's therefore reasonable to assume our flute LH might have done the same.

Note also that the left Shoulder (SL) has ended up more ovalled (orange vs violet) than the right shoulder, SR.  Because the Left tenon was thinner under the wrap and therefore more affected than the right tenon.

You might object that the degree of bulge shown in this graph is not great, but then the test tube has experienced only a total of 8 humid days and 7 drying days, the rest of the 35 days being airing.   By comparison, Potter's flute is now up to around 73,000 days.  None the less, we can already see the trend.


So, getting back to Mr Potter's flute, what does our experiment tell us?  It encourages us to consider the boldest original scenarios, such as "Ignore the Shrinkage" or even the outlandish "Join the Ends" shown in the graph below.  It suggests to us that the bulge in the middle of the LH section and the sloping compressions joining the bulge with the two tenon shoulders are also artefacts of tenon thread compression.  And this would mean that we have seriously underestimated the degree of strangulation this "moderately strangled" flute has suffered. 


This experiment, in conjunction with what we've learned before, allows us now to explain all the bumps and grinds we see in the LH section of the Potter flute.  Thread wrapping, in conjunction with moisture cycling, has compressed the tenons dramatically, compressed the shoulders considerably, and caused a bulge in the middle of the piece.  Thread wrapping is not just a risk to tenons, but to whole flutes.

It also appears to have induced considerable ovalling at all the points of compression (not just under the tenon wraps).  We need to confirm and explain the ovalling mechanism, and to consider its significance.  That might be best done in a separate experiment.

What will prove interesting to see is what happens to that bulge and that ovalling if we can find a way to reconstitute the LH fully.  This story still has a long way to run.

On to Ovality.htm

or Back to McGee-flutes Index page...

  Created 28 June 2011