Home > Experiments > Archimedes death ray: idea feasibility testing > MythBusters
2.009 Archimedes Death Ray: Testing with MythBusters

The first showing of this Mythbusters episode was January 25, 2006 on the Discovery channel. You may also want to read about and see video of the original experiment at MIT in 2.009.

On Friday, October 21, 6 students from the current and previous two years of 2.009 and the course instructor went to San Francisco to burn a real boat with the MythBusters. October 21 was used for final mirror fabrication and to setup the stands for the array.

Click on images to see high resolution versions.

condition before


condition after

On Saturday, October 22, the weather was good. We were able to cause charring and smoldering in a 1-2 foot wide swath along much of the boat's length. After three passes over the boat, the hull was penetrated and a small open flame was achieved. This burning area was self-sustaining and, about 2 hours after the test had ended, it was still burning with intermittent flame and had created a roughly 10 inch diameter hole in the hull of the boat.

We did not, however, achieve a large flash ignition as we did in the MIT test. We believe that this was due to the high moisture content of the boat that was used as a target. In my view, it is an open question as to whether the boat was representative of a seaworthy Roman ship... but at least this boat should not be considered as too easy a target!

A more detailed description of what happened follows.

Overview of test configuration

The test employed 300, 1 foot square, bronze mirror tiles (510 phosphor bronze alloy). The number of mirrors was determined using the first MIT test as a benchmark and accounting for the different mirror material, along with the different conditions in San Francisco.

The tiles were polished using Brasso by the MythBuster and MIT team. The polish was quite high on most mirrors. About 30% of the mirrors had a variety of flatness defects that would reduce their performance to some degree.

The tiles were aimed in a manner similar to the MIT test, with the exception that there were 4 tiers of mirrors instead of just two rows. The mirror array was about 110 feet long, similar to the MIT test. The goal was to make an array that was big enough to start a fire, but small enough to implement within reasonable budgetary and time constraints (my guesstimate is ~3-4 person weeks, mostly spent on polishing the bronze).



Our understanding was that the target boat was a 30 foot, 1924 wooden fishing boat. It was purchased from its prior owner after it had sunk and then been retrieved and brought back to port. We were told that it was much lower in the water before the engine and other heavy items were removed a few days before the test.

It is believed that the boat planks were made of douglas fir. The planking was estimated to be 2-3 inches thick, dependent upon location. The boat had been partially stripped of paint and then coated with pitch.

If you are interested in what a more authenic ship would be like (both in design and condition) please review this slideshow. For estimates of how both moisture content and wood type would affect ignition time, please see the FAQ.

The boat was positioned so that it was roughly 150 feet from the center of the array. Since the array was in a straight line along the edge of a dock, the furthest mirrors (at the ends of the array) were roughly 160 feet from the boat.

The test (Saturday, October 22)

The test began at roughly 12:15 and the 300 bronze mirrors took 10 minutes to aim (14 people working in pairs) using an aiming process similar to the MIT experiment. The aiming mirror was initially pointed at the stern of the boat since the motion of the the earth around the sun caused the point projected by the fixed mirrors to move towards the bow of the boat over time.

Adam was an awesome mirror aiming machine.

Almost immediately after uncovering the mirrors, a considerable amount of smoke was visible. Black charring was visible on the side of the boat in a few minutes. The bright spot was estimated to be 4-5 feet in diameter... not as tightly focused as we had anticipated based on preliminary tests, but not terrible either.

As the sun spot moved along the side of the boat (at about 9 inches per minute), it created a roughly 1-2 foot wide charred band along the side of the boat, but without open flame. We believe that a significant amount of moisture was being boiled out of the wood.

When the sun spot had moved to 3/4 along the length of the boat (leaving an estimated 20 feet long x 1-2 feet wide charred strip on the boat), we attempted to continually adjust the position of the mirrors to keep the bright spot on a single location, giving the wood more time to dry and heat. In doing so, we were not able to keep a focused beam, and the bright spot became much larger, reducing the power per square foot. Thus, the effect of the beam was reduced.

At roughly 12:40, we repeated the process, again using the bronze mirrors. The targeting mirror was aimed at the stern of the boat and the entire array was re-aimed. This allowed us to make a second pass over the band partially dried and charred in the first attempt. There was an even more substantial amount of smoke and the charring deepened. The beam was better focused than the first attempt (~3-4 feet in diameter). There were glowing, smoldering embers in a number of locations along the side of the boat, and in some locations there was white ash on the surface of the planks... but no open flame. There was a modest sea breeze at the time.

After letting the beam traverse the length of the boat it was requested that we try to ignite a sail, again using the bronze mirrors. This created a bright spot on the sail and no charring whatsoever. The dominant reason for this is probably that the sea breeze was rising and the thin sail had substantial convective cooling on both sides of the cloth (see FAQ xii). Also, the lighter color of the sail may be an important factor.

During the time that the beam was focused on the sail, charred areas on the side of the boat continued to smolder. After the test with the sail, we decided to switch to silver/glass mirrors to see if they could take us from smoldering embers to open flame. The time was around 1:30 PM (estimating).

The silver mirrors provided a tighter beam that we could more easily focus (~2-3 feet in diameter). We repeated the process, making a third pass along the side of the boat, starting with the aiming beam at the stern of the boat, and letting the bright spot move its way along the side of the boat (again, at roughly 9-10 inches per minute due to the motion of the earth). There was substantial smoke, and at one point near midship, there was a period of near-calm wind. The smoke pattern looked very much like what we saw just prior to flash ignition in the test at MIT. However, no open flame appeared, even though there were substantial glowing embers.

At this point, we decided to bring the boat closer for one last attempt to achieve open flames. The rationale was that, at a shorter distance, we would be able to more tightly focus the beam. The boat was brought to a distance of 75 feet. The time was roughly 2:00 PM.

In hindsight, this was a poor decision on my behalf. It would have been better to use the time to just make another pass over the boat since, as became apparent later, the coals were on the verge of reaching an open flame (the glowing coals can be seen right under Jamie's feet in the picture at left, which was taken while the boat was being repositioned).

You may want to check the close up image. If we had applied the sunlight to these areas once more, we probably would have achieved a much more substantial flame.

Regardless, we brought the boat to a distance of 75 feet and began the aiming process once again, starting from the stern using the silver mirrors. However, while the beam was still near the stern, a modest open flame broke out mostly on the inside of the hull at midship, near a scupper hole. The sea breeze was really picking up and presumably fanned the burning embers to obtain an open flame.

The open flame was caused by the efforts at 150 feet since the beam at the closer distance had not yet reached this point on the boat (Jamie pointed this out to us when the flame broke out). It is not clear if the switch to silver mirrors was necessary. My opinion is that the switch was not required.

Before the closer range beam reached the burning area on the boat, a substantial wind gust blew over several of the mirrors, creating a rather spectacular effect and ending the experiment. We had modeled that this would occur when the winds reached 14-15 knots. We did not have the instrumentation needed to obtain a wind speed measurement. Thus, we could not make a comparison with our estimated value. We had planned to finish before 2 PM and thereby avoid the afternoon period when the sea breeze becomes quite strong.

The fire at midship continued to smolder with glowing embers and also intermittently burn with open flame. Open flame appeared to be dependent on fanning by the wind. In the picture on the right you can see a very small flame (you may want to look at the large image).

About 2 hours after ending the test, the fire was still smoldering with intermittent flame. There was a roughly 10 inch diameter hole in the side of the ship. The fire was extinguished with water.

Final thoughts:
We were able to cause charring and smoldering along the length of the ship. After three passes along the boat, we were able to penetrate the hull in one location and achieve an open flame. This burning area was self-sustaining as a smouldering fire, and about 2 hours after the test had ended it was still burning and had created a roughly 10 inch hole in the boat.

We were not able to achieve the large flash ignition as we did with the 2.009 experiment at MIT. Most likely, the difference was the moisture content of the wood, which was very high. Certainly wood in any boat on the water would have higher moisture content than kiln dried lumber but, given the history of the boat, it may still be an open question as to whether the moisture content (above the water line) was representative of a typical wooden boat (or Roman ship) in good seaworthy condition. In hindsight, it would have been prudent to have taken a moisture content reading so that this issue could have been addressed.

All said, we feel that the power estimates and models used to design the array were quite close to the actual performance, but we did underestimate the effect of moisture content.

For larger distances, one would want to improve our design to allow a fixed spot on the boat to be tracked... and inevitably, some fairly simple ways to do this have already come to mind! Too much fun...

Still curious? Check out the FAQ or the original MIT 2.009 experiment.