With this demonstration, I won the Best Demonstration competition at the British Interactive Group's annual Event, held at the Magna Centre in Rotherham from 25th-27th July 2007.
I first saw the demonstration at the European Conference on Visual Perception in Edinburgh in 1993, when Alan Gilchrist used it in an invited talk. It doesn't seem to be known to the science communication community, and I think that it deserves greater fame.
What you see
At the front of a dim room, you see a small bright rectangle. It appears to be white. Then a second rectangle is brought into place beside the first. This one is brighter, and so it seems white, and the first one now seems pale grey. A third rectangle is brought into view. It's even brighter than the second one, which therefore starts to appear grey. When a fourth rectangle appears, it is, of course, even brighter than any of the others, and it appears white, while the others appear various shades of grey. Surely nothing can be brighter than the fourth rectangle? Wrong! Up comes the fifth one, dazzling white and outshining all the others. By now the first rectangle appears mid-grey.
The picture on the right is an attempt to photograph the sequence of events. Somehow the camera manages to underplay the contrast between the rectangles, while overplaying the rather uneven illumination of the left-hand rectangle. In reality, audiences spontaneously see the left-hand rectangle as white.
By now, most watchers will have figured out that the rectangles are being lit by a bright, hidden light, and probably now think that they know what colour each rectangle really is. How wrong they are. The next step is to turn off the hidden light, and turn on the room lights.
It is now plain to all that the first rectangle is actually black.
Why does it work?
This demonstration highlights the problem of ambiguity in vision. The initial glowing rectangle could be a dark surface brightly illuminated, or it could be a light (ie highly reflective) surface dimly illuminated. In fact there is any number of combinations of illumination and surface reflectance that would result in the same amount of light being sent to the eye. Unless you know how bright the light is, you can't tell how reflective the rectangle is. In this case, your vision makes the wrong guess, and an object that is covered in black paint is perceived as being white.
Similar arguments apply to each of the succeeding grey rectangles, but as they appear, you have more and more information about the brightness of the illumination, and the earlier rectangles no longer appear white.
Still, it's clear that your vision is unable to comprehend how brightly lit the rectangles really are. It's only when the hidden light is off, and the room lights are on, that you have reliable information about the illumination of the rectangles. Finally, the first rectangle appears as it was painted: black.
The dark side of the Moon
Everyone has seen this effect in action. The Moon looks as if it is made of very bright material - we make jokes about it being made of cheese. In fact, the Moon is made of rather dark rocks, which on average reflect only 7% of the light that falls on them. However, watching the Moon from Earth, we have no information about how brightly it is lit by the Sun - more brightly than the sunniest day on Earth. Our vision makes the wrong guess about the illumination, and we perceive the Moon to be made of light material.
How to produce the effect
The aim of the apparatus is to illuminate the rectangles brightly, whilst keeping stray light down to a minimum. Therefore the hidden lights are enclosed in a box, in such a way thay they are masked from most other things than the rectangles.
Each rectangle is lit from below by a 50-watt 240 V quartz-halogen spotlight. The lights point steeply upwards so that stray light is largely projected inconspicuously onto the ceiling. The rectangles seen by the audience are in fact the top ends of much longer pieces of board. These are hinged so that they can stand upright or fold back out of view. The wall in front of the boards obscures the unlit parts of them. This arrangement (for which I thank Steve Owens) gives satisfactory views of the rectangles for viewers at a range of heights (as in a raked auditorium).
This picture shows some of the lights inside the box. I used the aluminium-backed type (as opposed to the dichroic type) so that spare radiant energy was, as far as possible, ejected from the enclosure. You don't seem to be able to get a simple batten lampholder for this kind of lamp. I mounted them in ceramic lampholders, which were bolted to pattress boxes, which in turn contained the wiring.
This picture also shows that the lowest few centimetres of each rectangle (below the hinge) blocks the light from the spotlight when the rectangle is lying horizontally out if sight of the audience. As each rectangle is raised, its light is exposed.
I painted the black rectangle with matt blackboard paint, and the white one with white emulsion paint. For the three grey rectangles, I used mixtures of matt white emulsion paint and ordinary cheap black liquid paint in a bottle from an art shop.
Having tried black and then grey, I found that a white box is best for making the first rectangle appear white. The room should be dim, but it doesn't have to be anywhere near pitch-black.
Development
The halogen spotlights are easy to deal with, and are efficient in the sense that they cast most of their light forward. However they are extended sources (5cm diameter) which makes it impossible to mask off all stray light without at the same time blocking some light from reaching the rectangles. It would be worth trying bare halogen lamps instead, which are much smaller sources.
The other disadvantage of the spotlights is that their light distribution is visibly uneven. Again, bare halogen lamps would go a long way to alleviating this problem.
The main problem with bare lamps is that, having no reflectors, they would probably have to be of higher wattage to get the same illumination of the rectangle. This (and the absence of reflectors per se) would make the problem of heat buildup in the enclosure more severe.