March 1991
AT&T Bell Labs 600 Mountain Ave.
Murray Hill N.J. 07974
Copyright © 1991 AT&T All Rights Reserved

Why is the sky blue?

     This famous question was first answered correctly by Lord Rayleigh in 1881. He recognized that the air is full of particles (e.g. dust, gas molecules, etc.) which scatter light. The kind of scattering depends on the particle size. If the particles are much larger than the size (wavelength) of light (about one millionth of a meter, which is about 1/25th the diameter of a human hair), the light is scattered uniformly in all directions independent of the wavelength.This is why clouds, which consist of water droplets, are white and fluffy.

     If the particles are about the size of the wavelength of light or smaller, the amount of scattering depends on the wavelength of the light. (And thus the color. Blue light is "smaller" than red, by about a factor of 1/2). If the wavelength is a little longer than the dust particle is wide, not too much scattering occurs. A good analogy is a wave at the beach hitting a piling. A gentle, long period wave is barely disturbed, and just raises and lowers the water level at the piling. However, a wave about the size of the piling is scattered sideways.

Thus, the effect of dust is to let the long wavelength (RED) light pass through, while scattering to the sides the short wavelength (BLUE). In this way dust takes the  white light from the sun and reorganizes it by wavelength. The background color of the sky comes from scattered light,  and is blue, while the image of the sun comes from direct light, and is yellow-red.


         If the number of scatters is very large, the sun will appear blood red. This usually occurs on a dusty day, after volcanic eruptions (the sun set red around the world for a year after the great explosion of Krakatoa in 1883), or at sunrise or sunset, when the light from the sun is almost tangent to the ground and so goes through a longer length of the atmosphere. An excellent resource on all aspects of color and optical illusions is the charming book "The Nature of Light and Color In The Open Air", by Minnaert, and there is a clear discussion by Jackson in his "Classical Electrodynamics" book. A good resource for experiments is "Clouds In A Glass of Beer- Simple Experiments in Atmospheric Physics" by C.F. Bohren.

     The traditional demonstration illustrating the effect of "Rayleigh Scattering" is to shine a 35 mm projector through the side of a rectangular fish tank filled with water. Milk is slowly mixed into the water. As the milk is added, the circle of light on the wall will turn red, and the tank of water, blue.

What is the Difference Between A Glass, a Solid and a Liquid?

     Glass is sometimes referred to as being a "liquid". This is a little bit confusing, in that glasses usually appear to be quite rigid just like any other solid material. In fact, almost all glasses can quite accurately be described as being a liquid which has been cooled so far below its melting point that it is extremely viscous and thus has the mechanical properties of a solid. The reason glasses are described as being "liquids" is to distinguish them from other solids that are "crystalline" in nature. In a crystal every atom is precisely arranged. Each atom is a certain distance from its neighbors, and the bonds between atoms always have the same angles with other bonds. In a liquid the molecules that make up the liquid are constantly moving with respect to one another, so that if you wait a fraction of a second all of the molecules will have rearranged themselves. A glass is like a snapshot in time of a liquid. The distances between atoms are not always the same, nor are the angles between them. But, like a crystal, the position of the atoms rarely change over human time scales.

     Here is a simple experiment using freezing to illustrate the distinction between a crystalline phase and a glass. Put some water and some corn syrup into the freezer. As the water gets colder, needles of ice form and grow, eventually taking over the whole volume.There is always a sharp boundary between the ice and water, because ice (a crystal) and water (a liquid) have different structures. The corn syrup just gets thicker and thicker until it gets so thick that it won't noticeably flow. At no time is there a sharp boundary between fluid and solid phases, because the syrup has not gone through a change of structure. It remains "glassy" at all times.

     In a glassy material like a window pane (made up in large part by silicon and oxygen atoms) the atoms do not appreciably rearrange themselves, unless you are willing to wait millions of years. A silicon atom will remain firmly attached to a certain oxygen atom which will in turn be firmly attached to another silicon atom.The only disorder is that which was described earlier, in which bond lengths and angles are somewhat varied from atom to atom in the glass.

       There is an old story that "proves" glass is a liquid because it was "observed" to flow over the centuries. In colonial houses, the window panes are often found to be thicker on the bottom than the top. People then erroneously drew the conclusion the glass, being a liquid, had sagged under the pull of gravity. In fact, old glass is thicker on one edge when first made, and does not change shape. In the past, glass sheet was made by blowing a ball of hot glass into a hollow globe, and then slowly forming into a cylinder. The cylinder was cut apart and laid out flat into a sheet. The resulting window panes have small air bubbles, wrinkled, picked up dirt when cooling, and tended to have one thick-end. The window installer would systematically place the thick-end down (where are today's craftsmen?), thus setting off centuries of idle, and ultimately, wrong-headed speculation on the behavior of glassy material. If glass flowed by an eight of an inch over two centuries, then scratches would seal-over in days- and that never happens.... See this article from the Corning Glass Museum for more details.

Are Zebra Mussels a Problem in Europe?

     Zebra mussels were first described in rivers emptying into the Caspian Sea in the 18th. century. They were introduced into European waters roughly 150 years ago, where they underwent a population explosion comparable to the one now occurring in this country. The population levels of the zebra mussel in European waters have fallen due to adjustments in behavior and population levels of their predators, e.g., fish, birds and crustaceans. Zebra mussels are now integrated into the food chain in European waters but they are still considered a pest species and still cause significant fouling of pipes and boats in European waters. In the U.S., the mussels have recently invaded the Great Lakes. They form dense colonies of more than 30,000 mussels per square meter. Since each mussel can filter one liter of water in 24 hours, in some locations the volume of the lake is completely filtered each day by it's bivalve invaders! Needless to say, the ecological impact is enormous.

     An excellent source of further information on all aspects of zebra mussel ecology and economics is The Zebra Mussel Clearing House at the State University of New York at Brockport. The co-ordinator is Susan Moore (716 395-2516). They have a great deal of free literature, including a 10 page fact sheet, 3 complimentary issues of a newsletter, photocopies of published papers, a bibliography($2.00) and specialists to answer technical questions.

Why do we divide the day into 24 hours, an hour into 60 minutes, etc.?

     The 24 hour day and 60 minute hour originated with the Babylonians, who based their calendar on the earlier Sumerian and, perhaps, Egyptian calendars. The Encyclopedia Britannica has a number of good articles on this subject, and there is a wonderful chapter by Daniel J. Boorstin in "The Discoverers". As Boorstin explains, the history of timekeeping is itself lost in time.

     The origin of a counting system based on 60 is related to astronomical cycles in the earth-moon-sun system. The earth takes about 360 days to orbit the sun, and over the course of a year the moon goes through about 12 cycles. Our life today is not very well keyed to these astronomical periods, but following and understanding the flow of the seasons was a great deal more important in earlier societies.

     The Sumerians were probably the first to develop a calendar based entirely on the phases of the moon. The Babylonians refined this idea to form a calendar of which several details survive in our"modern" calendar.-A Babylonian month began on the first day of a new moon.-There are slightly more than 12 Babylonian months in a year (about 12.37 to be more precise). [Every two or three years, the royal astronomers would insert an extra full month into the that the calendar based on the moon's cycles would stay in better sync with the year]-From the fact that there are about 12 lunar months in a solar year, the Babylonians decided that the number 12 is the logical number by which time should be further divided.-From about 300BC, the Babylonians used a hemispherical sundial-type device to keep time during the day. The shadow cast by a thin rod would travel past 12 equally spaced marks on the inside of a hemisphere. The time it took for the shadow to pass from one mark to the next was ONE HOUR. Thus there were 12 hours during daylight. The night time was also divided into 12 parts, as timed most likely by water clocks...the other type of Babylonian clock. (Note that this Babylonian system yields hours which are shorter during winter.when the days are shorter.and longer in summer. As a result, in Babylon, in modern-day Iraq, the daylight hours in mid-winter were only 2/3 as long as the mid-winter night time hours with the opposite situation occurring in mid-summer.

     The invention of accurate mechanical clocks at the end of the 13th century finally made the variable-length hour less useful than the modern hour of unchanging duration.-The Babylonians divided the hour into 60 minutes (60 = 5 x 12....that preferred number 12 appears again). Additional Questions:

  1. 2000 years ago, how would you determine if the year was 365 or 364 days long?
  2. The day was not always 24 hours long. Half a billion years ago, a day (one complete rotation of the earth) lasted about 20 of our present hours. Since the day is set by rotation of the earth, and angular momentum must be conserved, how did the day change in length?
  3. If the year were 100 days long, with 10 months to the year, we might have adopted the decimal system a little earlier (especially with the attraction of 10 fingers and 10 toes to count on). But, what effect would a three times longer day have on our climate, tides, etc?

Fruit Flies and Power Lines

     A student called in asking how to set up a 110V circuit  to mimic the electric and magnetic fields around a high voltage line. They are working on a science fair project,  and wanted to see if these fields might effect the growth and death of a colony of fruit flies. Now, setting up a  solenoid and a capacitor plate to give similar field  distributions is straightforward, but designing an  experiment to detect possible biological effects is not. It  is a good illustration of how hard experiments are to perform when the effects are small and the experimental conditions are out of your direct control.

  The first problem you encounter is the same one  epidemiologists must overcome in studies on humans. Although  the fields right next to a power line can be high, they drop off rapidly with distance. Measurements show the distribution of field strengths in homes close to power  lines overlaps the distribution of electric and magnetic  fields in homes miles away from any power line. Thus, simply trying to correlate the address of people's houses with the cancer rate is likely to give spurious results. People near  the line who use gas heat, mechanical clocks, and in general  very little electricity could get a lower field exposure than those in an electrically heated apartment building miles away. Similarly, the students could set up their experiment in the basement, with the "control" population of flies near the main power feed to the house. These fields might be greater than those in their solenoid encased "exposed" population. Careful experiments (including monitors worn by people to measure their actual exposure)  will be needed to settle the issue in human populations. The cover story in April's issue of Radio Electronics is on  magnetic fields and "how-to-build" simple devices to measure  your exposure (however, their choice of references is a little biased in favor of possible links to cancer ). 

     The second problem you'll encounter is a matter of  statistics. No reputable investigation has found that power lines cause big effects; your fruit flies won't just drop  dead when you put the wires nearby. After all, people don't drop dead when they walk under power lines either. For  people, the argument is about 1% or so (possible) extra  deaths from cancer over a period of years. So, if we assume that fruit flies behave like people (not always a good assumption), you should expect that most flies will live a normal fly lifespan, and die at about the same rate whether the field is on or off. Now let's think about some results we might get. Assume 12 flies died in bottle A, and 11 in  bottle B. Does this mean that there is a real difference between the two bottles? Not necessarily. One of the flies could have been less healthy in bottle A, perhaps. You might as easily find two flies which lived longer in a bottle near  the power line. Does this mean we should recommend power  lines as a health tonic?   The problem is, you have to deal with large enough  numbers so that individual, random differences between the bottles are less important. If you made the experiment 10 times bigger, so that 120 flies died in bottle A and 110 in B, you'd have a better result. One unhealthy fly can no  longer tip the scales - you have to have 10 sick flies in  the same bottle. If you make the experiment even bigger so that A had 1200 and B had 1100 - then you'd really have something.The odds of getting 100 extra sick flies into  bottle A would be pretty small. You have to have enough flies so that even a small difference between the two groups  is still much more than the random fluctuations in the death  rate. So, if you want to look for a small difference, you're going to have to count a lot of fruit flies (dead and  alive).

     A lot also depends on how the electromagnetic field  affects the death rate. If fruit flies never die before one  month of age, and with exposure to the field 1% start dying at one week, a field induced effect could be claimed.  However, it is more likely flies die all the time, and we  are forced to look for an increase in the average death rate by 1%. Lets say 10% of the flies normally die in the first month, so a 1% increase in the death rate corresponds to a new death rate of 11%. If we had 1,000 flies in the bottle,  we need to tell the difference between 100 dead flies and  110 dead flies at the end of a month. This sounds simple (but tedious) enough, except that the 10% figure is an average number. You expect in some months more flies would die than average, and some month less (that's what average  means). In many cases, the square root of the expected  number is a good measure of the fluctuations. In this case,  the square root of 100 is 10, so you might expect 100+-10 to  die even without the field. This means you cannot tell if  the 110 represents additional deaths due to the field, or just a random fluctuation above the average. Here is how the confidence level increases as you increase the number of  flies in the bottle.


Number of Flies    

10% Deaths (with no field)     

11% Deaths (with a field)  

Deaths Due to Field        

Deaths due to random statistics         





















     You can see that it becomes a very hard experiment if you want really accurate answers, or if you want to see a small  difference. (With 100,000 flies, the difference in death rate with and without the field is 1,000 flies, compared to  random fluctuations of 100 flies. This would be considered a  "statistically significant" experiment). That's exactly why  there is still argument about the effect of power lines on  people. You'd have to keep track of a LOT of people to be sure that a difference is real, and not just chance.

     You also have to be sure the flies ate the same food, got  the same lighting, and so on. With people you have to compensate for diet, smoking, workplace related disease, etc. Otherwise, when looking for small effects, uncontrolled (sometimes called "systematic") errors can swamp your  results. Ultimately, science is done by people in a real and  challenging environment. Mistakes happen, even to people  with the best training and intentions. Over the long term, however, we like to believe the best and most accurate results are reproduced, survive, and are finally taught to our children and the public

In the next newsletter:

  • What happens when electrons scatter off electrons?
    How fast does the earth cool?
    What are imaginary numbers good for?


   If you have any comments on this newsletter, call

             1-800-CLEARUP or write

       c/o AT&T Bell Labs
       600 Mountain Ave.
      P.O. Box 636
      Murray Hill, N.J. 07974


Contact Greg Blonder by email here - Modified Genuine Ideas, LLC.