A FLUTE-CLOCK CAPER
by Robert Moore

My interest in automataphonics started as a small boy crawling under and into the workings of a player piano. What I discovered in that first encounter instilled a curiosity in me that has lasted for more than half a century.

I spent a large part of my working career as an Engineering Technician for the Canadian Government - in particular the Magnetics, Arctic and Ocean Acoustics departments of Defense Research Establishment Pacific (DREP). I made "toys" for science. I even got to go to the far arctic to play with them - drilling long holes through the sea ice for lowering the acoustic listening devices or other paraphernalia.

When our shop foreman, Glen May was to retire, I as the foreman-in-waiting, thought it only proper to build a parting gift as a token of our respect and, assurance that he was leaving the shop in capable hands.....But what? something reminiscent of the years of work with Glen: inventions that only the mother could love, ideas that crystallized through that indescribable process of concept to conceptualization, experimental bells and whistles that either worked or wilted in the sub zero temperatures and most certainly baffled the creatures under the ice. Glen was remarkable in that he could come up with an idea, make it and have it installed before an Engineer and Draftsman could hand you the drawing.

My vocation as a machinist and avocation as musical instrument maker were inseparable and at times this was apparent when my concept of an "acoustic array" better fit the description of a set of bagpipes than something to be lowered through a hole in the Canadian arctic ice. The answer was obvious: We would make a flute-clock. I had a picture of one in my Barrel Organ Book and although certainly not in the ranks of Captain Nemo's submarine organ, could be contrived discreetly in our home workshops.

Although the word "clock" is used to describe the device, historically it was used to describe not only a timepiece but also any mechanism driven by a heavy weight or spring, the potential energy of which was applied to drive the instrument through a train of gears. Yet it was also historically more explicit. Many flute-playing clocks not only had timepieces but would also unleash the potential energy at prescribed intervals to play little tunes by, for example, J. Haydn, Mozart or Salieri. The toys of kings, one might be found sitting in gilt splendor on a mantelpiece of Frederick the Great, King of Prussia (1740-86) or his brother Ferdinand. And as the "jukebox" of the people, innkeepers and hairdressers entertained clients with flute-playing clocks programmed with the music of Rossini and Schubert.

The idea of a barrel organ has been around for a few centuries. The concept was described in the ninth Century by the Banu Musa, but the conceptualization was first definitely established in the 16th Century. The Hohen-Salzburg organ, which used a barrel, was made in 1502. As well, a carillon in Mechlin (Belgium) in 1583 was mechanically programmed with a barrel. Complex orchestral music set on barrels reached its peak in the 1880's. Beethoven's Battle of Vittoria was written explicitly for Maelzel's Panharmonicon, a very sophisticated barrel instrument.

Our limited time framework prohibited much archival research for working drawings and information. I had some personal resources in my library: Barrel Organ, The Story of the Mechanical Organ and its Repair, by Arthur W.J.G. Ord-Hume (pub. 1978), The Mechanics of Mechanical Music, also, by Ord-Hume, and The Art of Organ-Building , by George Ashdown Audsley (first published 1905). However, information on the precise mechanical workings were few and in general the photographs of existing clocks required extensive imagination and interpolation. We chose history as our inspiration and experimentation as our tool of discovery. This, after all is what we, as a research lab, were all about. This flute-clock was to be a Mark I.

Fortunately for me, sometime during the first half of the eighteenth century the elves in the black forest region of Germany came up with a small barrel organ incorporated with a simple bracket clock, usually made very largely of wood. The idea of a musical clock for the people sounded good to me, and was more accessible as the photos were as close to working drawing as we were going to get. To make things simpler and of timely appropriateness for a man retiring from the proverbial time-clock, I decided to eliminate the timepiece.

The concept of barrel organs was something of long-standing intrigue with me, and our crew was capable. It turned out to be a challenge. It was a three month process of discovery to contrive the mechanism that finally tootled out the Monty Python theme of Liberty Bell with the breathless almost tuneful naiveté of a small boy auditioning for the church choir.

I took my annual holiday sequestered away on a small island with my Barrel Organ Book in hand and sheaves of paper. The first rough drawing resembled a schematic drawing from Ord-hume's book but this needed developing into a practical design.
Rough measurements were estimated from photographs and text in the book to fit the case dimensions we had chosen. The tunes we wanted required 11 pipes and an extra one for the whoopee cushion. Gear ratios were partly determined from the book, extrapolated by simple mathematics and tempered with what gears we had available. I came back to work with preliminary working drawings, and handed them out to my trusty comrades with instructions to come up with the goods in one month - discreetly, of course. Then the job of assembly and fine tuning would begin.

Drawing revisions evolved as the work progressed. The first working drawing depended largely on the wooden case as a framework for the mechanism. However, as we were working from the inside out in the conceptualization, I made a design decision early on: The wooden case would support the bushing for one end of the barrel as well as the reservoir, windchest and keyframe. However, because of the limitations of wood stability, the gear train would be integrated with its own framework. This would consist of three 1/8th inch aluminum plates, spaced apart using aluminum rods of appropriate length, thus holding all the gears shafts and bushings firmly in alignment.

In the initial drawing, the energy to drive the mechanism was from a cord wrapped around a drum with a weight on it. I made an improvement to this when I found that a piece of brass chain I had, just happened to fit a bicycle rear sprocket. I decided to add a simple ratchet. This drove the mechanism and when the weights reached the floor, a simple pull on the chain put the mechanism back into action.

I gave the job of making the wooden pipes to Rex Welland - his pertinent qualifications being that he had once made a whistle and he liked working in wood. We chose yellow cedar, a fantastic local wood with great acoustic qualities that machines well. Rex had little trouble coming up with the 11 pipes required. Essentially, these are wooden stopped diapasons, or basic organ pipes. An organ pipe is a coupled system, i.e., the edge tones at the mouth are coupled to the natural frequencies of the column of air in the tube. When the open end of the tube is stopped the pitch drops an octave and softens the tone. The stoppers do double duty as convenient tuning mechanisms for the pipes. Rex had some difficulty with the sealing of the tunable tops - the fit had to be just right in order for the leather to seal against the body. A moot point, but critical. The leather seal on the rectangular stopper did not seal the corners of the rectangular hole until he changed the leather wrapping strip to four separate strips. In 3 weeks, he had them finished beautifully and to a whistle.

Woody (Elwood Godlien) is a whiz at metal fabricating, so I gave him the job of making the keys, keyframe, and a few other parts. Being a welder he used aluminum welding rods for the push-rods that would open the palettes and sound the pipes - a good choice for keeping the lifting work of the palette to a minimum. His accuracy was appreciated. On assembling his keyframe in the case, the precise clearance on each key made the final positioning a simple job of adjusting the push-rod working length with leather nuts. When it came time to pin the barrel, the keys which read the music through the pins, operated smoothly and required no alterations.

I made the feeder and reservoir, windchest, barrel, case, gear train and a few other details. The feeder and reservoir were constructed with 4mm birch plywood. The outer boards are movable while the inner ones are fixed. They are covered with thin leather, stiffened by the addition of pieces of Bristol card with gaps at the folds, glued to the flesh side, enabling the leather to fold and unfold easily and not lose integrity by ballooning out with the pressure. A simple leather valve admits air into the feeder and another admits air from the feeder into the reservoir on a one way trip. At first the feeder valve was too heavy, being backed with a thick card and got into a flap. (Diagnosed by a fibrillating sound reminiscent of faulty bagpipe valve) I resolved this by eliminating the backing.

Music wire was used to make the springs which apply pressure to the reservoir's movable board. Stiffeners on the board additionally, provide a guide for the springs. The springs regulate air pressure in the reservoir and the relief valve protects the reservoir from excessive pressure. The relief valve has its own little spring that keeps the air contained until the tail of the valve reaches its limit, opens and dumps the excess air. For those with mathematical inclinations there is a Handbook of Mechanical Spring Design. However, previous experimentation on setting spring pressure for air reservoirs in some of my previous work had given me some guidelines. I estimated 3 1/2 inches of water pressure to sound the pipes based on reading in the Art of Organ-Building. Determining spring wire diameter, number of coils, coil diameter and spring arm length was largely an educated guess. Once made, I checked the air pressure with a simple manometer made from a small diameter, open ended plastic tube bent in a U, partly filled with water and connected to the windchest. My guess was close and further gratifying when the pipes sounded clearly when all was in position.

The windchest was made from maple. There were no acoustic requirements here and maple machines well. On top of the windchest, sit the pipes. Each pipe is self-locked into the windchest by a tapered foot which seats in a taper reamed hole. Gluing would be redundant and limiting for further alterations.

Below each pipe and covering each hole sits a spring loaded palette with a soft leather facing to ensure a proper seal. Each palette is opened at the required moment by the push-rod protruding through a small hole adjacent to the pipe. The rod takes a vertical descent each time the key rises off the face of the revolving barrel on the relative slope of the key-face against the pin. As the barrel continues to turn, the key drops off the pin and the palette's spring snaps it shut. The keys take the weight of the rods. We found the palette return-springs easily handled the rod's and their own weight and were ready for the next pin encounter.

The barrel was made of yellow cedar, like a whiskey-barrel, but cylindrical, with twelve staves glued to each other and to the two end-disks. This was turned and sanded on a lathe to remove the facets and make it smooth.

For the gear train I used off-the-shelf gears - Boston Gear Works - simply because they were there and they fit. Because the gears are such a fine pitch, the centre distance had to be carefully established to ensure smooth running. The three aluminum plates for the gear train housing were stacked, drilled and reamed for the bushings and spacers to ensure correct alignment. Everything fit together like "clock-work".

By this time we had developed a real working drawing, so making the wooden case fit was straight-forward. We chose 8mm Baltic Birch plywood for its strength and stability even though we had decided on an independent framework for the gear-train mechanism.
The barrel had a capacity of six short tunes, so I had to contrive a tune-change mechanism. The key-frame is lifted with a handled lever to clear the keys from the pins. At the same time, attached to the lever is a vertical indent lock which allows the repositioning and locking of the barrel shaft in the next tune position.

Next came the job of pinning the barrel. My friend Terry Miller didn't actually work in our shop, but in the computer wizardry department. He is very interested in programmable music and has made a Theremin, a key board synthesizer and several other electronic musical devices. I thought and was proved right in this, that he would have little trouble converting written music to the barrel of the flute clock.

Information for pinning the barrel was from the Barrel Organ Book and the Mechanics of Mechanical Music by the same author. To fully describe barrel pinning is beyond the limitations of this article. The literature describes complex devices for noting music onto barrels with great precision. However, an interesting and simple example given, is the bird organ or serinette: a simple French barrel organ in domestic use until the early part of this century used to teach the caged canary to sing popular or sacred tunes. The simple techniques described for pinning the bird organs suited us and required little more than a set of dividers and paper.

Pin positioning was a matter of converting the music onto the circumference of the barrel. The time for one barrel revolution, in our case, 15 seconds, was the critical determining factor in spacing the pins and establishing pin length. The term "pin' is somewhat misleading. It is more like a staple running the full duration of the note. The 3mm wide pins with pointed ends were cut from sheet metal, bent in staple fashion and driven into the barrel using simple hand-made gauges to maintain consistent pin-height and note-length. Whole notes and longer required support pins along the length.

The fabrication of the parts went smoothly and on schedule. However, the old adage that a flute clock is more than the sum of its parts proved itself once I began assembly. However, I was committed. I just hoped it would not be my undoing. Time was running out.

The gear ratio I had chosen was a bit slim. In hindsight, I would have given it a better ratio so that the barrel would turn slower enabling more music to be played per rev. Also the feeder would pump more often giving more air for a stronger sound. However, at this late point I decided to go with what I had - there were many details to work out for the thing to play without stopping and sound good. Because the whole thing turned, pumped and sounded the music using the energy of the chain weights, it was necessary to make everything happen with as little resistance as possible. The leather for the feeder and reservoir had to fold and unfold effortlessly , so care in selecting the leather, backing it in thin Bristol card and making the folds neat was very essential. Also, I found that the leather pleats in closed position were increasing the effort required. The connecting rod to the feeder was shortened to pull back on the positioning of the feeder travel. Another finicky detail was determination of the correct weights on the flywheel. I ended up using heavier fly-weights than I had first envisioned to increase momentum and ensure smoother running.

Well, Glen's retirement was fast approaching and the "clock" was still in the squeaks of infancy. I persisted, tweaking and tuning into the wee hours right up to the night before. The flute clock was speaking, although rather mechanical with the clicking of the keys and a slight respiratory problem when more than two pipes were required to speak at once. The respiratory disorder was considerably alleviated by decreasing the size of the holes in the foot of each pipe, decreasing the amount of air required to sing without taking its breath away. With the adjustments, the "clock" spoke freely in a soft whimsical voice. We were delighted with the sound of the pipes. However we were disappointed that the working pressure of the organ was insufficient to sound the whoopee cushion on the 12th key. More research is yet required to come up with a low pressure whoopee cushion.
There was space on the barrel for six very short tunes, but we had time for only two. The first, as mentioned, had to be 'Liberty Bell', the theme for Monty Python of which we were all fans. In repetition, the tune flowed well. The second was a simple contrapuntal J.S. Bach tune that also worked well in repeated cycles. We were ready for presentation.
The retirement party was memorable. After all the fine speeches, accolades and comradeship, Glen was liberated to the sweet refrain of "Liberty Bell" with the "Bronx Cheer" mechanism implemented manually by the flute-clock team. All this - the flute-clock adventure, the retirement party, my own retirement, are several years past, however the flute-clock team-work, in my mind, high-lights the achievements of the Defense Research Lab.