The CHRONOLITHE (Radiometric pendulum, 2001 / 2002) 

Could the force of light alone swing an 8-pound pendulum? And if yes, is it possible to use it as a clock? To try to answer those questions I created this clock, very different from all others. The pendulum beat is 1 second, two lamps, one on each side, turn on and "push" the bob alternatively. Crookes made some prototypes with pendulums, but it seems that no one ever used this type of propulsion for a clock. The principle was discovered by Sir Willam Crookes in 1873. The radiometric effect is used both as a propulsion force and as a braking force for the pendulum, allowing to accurately regulate the amplitude of the swing. Vacuum has been produced inside the tube, down to around 0.01 bar. The clock is made up of a glass tube, a pendulum, two reflectors (mica sheets), two relays and a quartz clock with the quartz removed. The initial swing is given from outside the tube with a magnet. The same magnet is used to fine-tune the period by turning the 4 auxiliary bobs. This avoids having to let air in, take the clock apart, and redo the vacuum for each adjustment. The whole clock is held together by the atmospheric pressure, without bolts an nuts. To take it apart, you only need to let air into the tube.

Full length view (1.75m high), December 2001

This clock idea occurred to me in April 2001. I ordered the parts in May and the very first test started on Wednesday November 7. This was the moment of truth: if the test failed I was left with a useless glass tube and large vacuum pump. The first proto? a pétanque ball for the bob, an incredible heap of 30 mica sheets, one side blackened with candle smoke. I placed a sensor outside the tube, did the vacuum, and launched the pendulum around 4 pm. As a full hour is needed for the pendulum to stabilise, I went for a drink to wait for the results. When I came back the pendulum was still moving. However I was not sure that this was not just the effect of inertia, I therefore went back for another drink. At 7 pm the pendulum was still moving with a constant amplitude, I went back to celebrate the success. At 10 pm of the clock and I one of us was still going straight...

This clock is made of a pyrex tube (190 mm wide, 9 mm thick) fitted on a quartzite block impregnated with epoxy compound
to make it airtight. The large sphere on the picture is the clock face with the hours, minutes and seconds hands. In January I
replaced it with a smaller one for aesthetical reasons. The first tests demonstrate an incredible accuracy, around 2 seconds
offset per month.

"This is the seventh and last version of the engine. The tuning was slow because every time three hours are needed
after the initial impulse for the pendulum to reach its final amplitude. I first had to work on the vacuum pressure, then
on the engine design to achieve the maximum possible gain. You need to realise that the clock will not operate if the
smallest amount of air leaks in. When this was done I was able to lower the power of the lamps from 35 watts down to
only 5 watts. This should expand their life span to 32,000 hours, a little less than 4 years. They are easily replaced, they
are standard halogen lamps."

This clock is self-adjusting, it adapts itself to unfavourable physical conditions to maintain its accuracy by compensating the factors that influence the amplitude of the pendulum. Here is how: assume that the lamp power increases, so will the swing amplitude. The bob will move faster toward the cell, will trigger the lamp sooner, which will slow down the pendulum a few microseconds, back to its normal amplitude.


(December 19, 2001) These last days have been spent fine tuning the isochronism versus the pressure inside the tube. But as
the clock is secured to my bench every time I move I create disruptions as large as 150 microseconds, translating into peaks on
the measuring computer screen. It has become impossible to continue working under these conditions, in order to make the
clock as stable as possible tomorrow I'll fix it against the wall near the display cabinet. The accuracy tests will then run on a
better basis.

(December 30, 2001) The accuracy tests have started. First finding: it is very easy to adjust the pendulum length with magnets.
Second finding: the clock is very accurate. No parasitic wave or unexplained fluctuations. The current air pressure changes
inside the tube will induce less than 5 microseconds fluctuation per beat. They will disappear when the tube which still has a few
leaks is assembled for good. I need to operate the vacuum pump every 9 hours, failing that the clock will stop. My current
concerns have to do with the way to get an accurate measurement through the 16 mm of pyrex glass. As the laser beam
reflection creates random inaccuracy, I do the measurement directly on one lamp. Each of those measurement can show up to 8 seconds variation per day but the average will be very reliable. The sample shown below, taken over 49 hours exhibits a
remarkable stability of the whole measurement series: 1.9 second offset per month. This measurement has since been confirmed
by much longer samples. This clock is therefore the most accurate of all those I built. So what started as a simple challenge to
build a pendulum activated by the pressure of light turns to be a very rewarding accomplishment. The adventure goes on...

               (4.9 hour sample; 0.7 microsecond delay per beat, translating into1.9 second per month)

(January 20, 2002) I received a new vacuum pump and a better pressure control and measurement instrument. This means
that the clock will operate in its optimal configuration by the end of the week... Stay tuned!

(January 28, 2002) This week's tests were all about time fluctuations as a function of air pressure. The results are
breathtaking. Here are some findings:
- The halogen lights start activating the pendulum at a pressure of 0.6 bar.
- It still works at 0.014 bar (maximum vacuum reachable by my pump).
- Inside those two values I could not observe any noticeable change in accuracy or any fluctuation in the pendulum swing.
- A half-degree turn of the auxiliary bobs induces a change of 0.7 microsecond per beat, around 1.9 second per month.
- The temperature changes in the workshop (around 4°C this week) do not seem to influence the pendulum operation. This is very surprising, they should. This is probably due to the self-regulating isochronism, to be confirmed.

(January 29, 2001) I was dumfounded this morning: when all accuracy recordings have always been flat this is what I
discovered: periodic and unexplained peaks, that started 12 hours ago and that appear every 2 minutes. The pendulum accelerates, reaching 1.995 second, then slowly goes back to normal. Each disruption lasts a total of 22 seconds.

The issue was quickly fixed: the clock base had moved during the night causing the pendulum to hit the glass wall every 2
minutes. I put it back in place and everything came back to normal. However this graph tells something much more important:
that's how even is the pressure on the pendulum. If the peaks had happened more randomly it would have meant that the
pressure was uneven and therefore a source of inaccuracy, which is not the case.

(February 3rd, 2002) In the next two weeks the tests will be about accuracy versus temperature, light and vacuum level. Here
is what the poor Chronolithe looks like in its torture room. The stone and the face have been removed, a multitude of sensors
record all parameters.

(March 8, 2002) Work is progressing and the Chronolithe is taking its final shape. Here it is, under different lighting conditions.
Click on a picture to download it with better definition (308 Ko, longer download time).

(Marc 24, 2002) Let's now talk about the limitations of this clock as they are far more interesting and instructive than its
qualities. For example what happens if a lamp is turned off? How does the lamp wear influence the pendulum swing? The
radiometric effect being strongly dependant on the light intensity the pendulum should quickly slow down. However the clock
self-adjusts by a large amount. This is precisely shown in the picture below. Half way down the recording, one of the lamps is
masked which means 50% less power. At this point I remove the screen. The graph then shows a 5 microsecond increment per
beat, which is less than 0.5 second per day. It will therefore be very easy to compensate for a possible time drift by adjusting
the lamp voltage. This will have the same effect as turning the smaller bobs without having to stop the pendulum.

This recording was done with a lamp tension of 7.3 volts and a pressure of 0.27 millibars. A laser is used to detect the
pendulum swing through the glass tube. Random fluctuations can be seen from sample to sample, up to 20 microseconds. It's
important to know where they come from. Contrary to what could be believed they are not caused by changes in the photonic
effect intensity. They are due to the very design of the Chronolithe because the sensors that monitor the pendulum's swing are
infrared sensors whose beam bounces back on a reflex reflector on the bob. This system is not accurate at all. If I connect the
measuring equipment directly to the sensor's output this is what I get, adjusted at the same scale...

What we have here looks like a random number generator, between 1.999300 and 2.000700 activating an otherwise very
accurate clock. We can also very clearly foresee how the next clock will be improved: by replacing the infrared sensors by
laser ones. They would however impact the Chronolithe's aesthetics. That's what was the biggest difficulty all along this design:
it would have been so much easier to use a wider tube, to get rid of the stone, to avoid all the airtightness issues... but no, the
Chronolithe exists as I had wanted it, consistent, complete, like drawn with one stroke of a pen.

If you care to know exactly what happens at the heart of the Chronolithe down to the microsecond, during almost two days at
the rate of one measurement every ten seconds, and if you have the Microset software click here.

The bob

(April 21st, 2002) The Chronolithe is still under test. These days, I sort of let it slowly suffocate by letting air in while the
computer records all the parameters. It still operates at 0.85 millibars. Other tests worth mentioning are full sun exposure to
check the self-adjustment capability, or inertial tests and motor gain. They should last for another month, time to harvest as
much data as possible. The clock will then be displayed for some time at the International Museum of Horology in La Chaux de

(July 1st, 2002) The Chronolithe won the first prize of the Kinetic Art Organisation international contest in the "Engineering
Ingenuity" category.
(September 23rd, 2002 The Cronolithe is displayed at the entrance of the International Museum of Horology in La Chaux de

                                                                         © Isabelle Favre


© Karine Denoual and Raphael Fiorina

         Crookes' radiometer

Everyone has seen this small instrument usually found in optical shops. It is a glass bulb with a device made of 4 mica sheets with rotates on a needle's tip. It will spin in the presence of a light source. The device is simple but its operation is not. People usually believe that light acts as water on the paddles of a water-mill. James Maxwell himself believed it, happily accepting this as an evidence of the radiation pressure predicted by his theory of electromagnetism. But according to his theory the vanes should have rotated in the other direction. The explanation was wrong and Maxwell regretted it publishing it for a long time. 
Other people believe that light heats the gas near the blackened face of the vanes and that the pressure created by its expansion makes them turn. This is the most widespread wrong explanation, still to be found today in the Encyclopaedia Britannica. Maxwell, that his first error had made cautious, demonstrated that this theory was in error but without being able to explain the rotation of the vanes.
The correct solution was only given in 1879 by Osborne Reynolds, best known for the "Reynolds number". He discovered that the rotating force was created by the flow of the gas at the edge of the vanes, from the hot side to the cold side. He called this phenomenon "thermal transpiration".

"Le Matin" "Télévision Suisse romande" "le Nouvelliste" "Horological Science Newsletter" "Orologi"
PORTFOLIO: The Chronolithe as seen by Karine Denoual
Video clip

All tests are performed using the excellent MicroSet, controller made by Bryan Mumford, Santa Barbara, California. All pictures are copyrighted. The English translation has been made made by Jean-marc Julia.Warm thanks to all those who supported me during the design, to those who helped me, greetings to those who constantly came to watch it grow on the website.