Before lasers were invented, light sources were like orchestras or choruses - even if all the parts were nominally at the same frequency, there were continual variations in phase, so that the parts were never quite in step.  Conversely, when the first electronic musical instruments were made, the tones were too pure, and had to be modified to achieve a passable chorus effect.

A laser can be made to produce a very narrow band in frequency and in wavelength, with all the parts of the wavefront in step.

To begin to understand how it works let's look at something completely different.  Every system has a distribution of energy that is greatest at lower energies and smallest at higher ones.  Click here if you want to know why.  Skip it if you don't.

What has this got to do with lasers?  Atoms emit energy when electrons in them all from a higher to a lower level.  The basic idea is to keep pumping electrons up so that they can fall down.  It is as if you pumped air up higher and used the energy of its falling down.  But there are major differences.  

Firstly, electrons in atoms can only have certain values of energy, and some of these levels are quite stable, so the electrons an stay there for a relatively long time.  One essential in a laser is to keep a high energy level well populated, so that there is a supply of electrons ready to lose energy by falling to a lower level.  Given that the electrons can have only certain energies, the amounts that they can lose is also limited to certain values.

Secondly, the light emitted when the electrons lose their energy is emitted in the form of photons with an energy equal to that lost by the electrons.  The light has a frequency and a wavelength that is related to the photon energy, so all the light from a given energy transition is emitted with the same frequency and wavelength.

Thirdly, if one of these photons strikes an atom that has an electron in the upper level, it has a good chance of stimulating the electron to fall to the lower level, making another photon emergr.

Fourthly, very important, this new photon is exactly in step with the stimulating one. 

If we could arrange a supply of energised atoms, and a supply of the right light, we could get a burst of extra light over and above the original burst.  Many lasers have two mirrors, so that the light bounces many times into the laser.  This gives a very high chance of stimulating new light, and so a chain reaction can build up.  In order to get the light out, the mirrors are made to be only partially reflecting.  The trick is to get the right balance between recirculated light and emitted light.  

A good trick is to change the reflectance of one mirror . It is kept high until a huge intensity has built up.  Then it is switched very suddenly to a low value so that the mirror is like plain glass.  The result is a great burst of light for a very short time.  The burst only lasts long enough for all the light to escape . The last bit of light to cone out is the bit that had left the switching mirror just before the switching occurred.  It has to go all the way to the other end and back before getting out.

The upper energy level should have a long life to enable plenty of energy to be built up.  The source of energy has the job of changing the natural energy distribution, biassed to low energies, to an artificial one with a lot of energy in the special energy level.  This often requires the use of another level that is higher than the wanted one.

There are also some tricks to shorten the burst of light after it has come out of the laser.  Bursts short as a femtosecond - 10^-15 second - are possible.  Since light travels about a foot, or 0.3 metre, in a nanosecond, a femtosecond burst of light is only about 0.3 mm long.

Laser light differs from other light in that the waves are in phase right across the beam.  This makes it easy to maintain a narrow beam over great distances.  Why is this the case?

These pictures simulate the output of 100 light sources in a vertical line at the left hand side of the picture.  The wavelength of the light is one tenth of the total width of the source array.  The degree of coherence of the sources ranges from zero in the first picture to complete in the last picture.  There are some artefacts in the pictures caused by the interaction of the pixels and the waves, but the trend is fairly clear.

In practice these patterns would never be seen.  For one thing size of a  normal light source is huge in terms of wavelength - a 2 mm laser beam of 700 nm light would span about 2800 wavelengths.  The sources, being atoms, would be much closer together.  Finally the phases would change rapidly and randomly.  For example, with a coherence length of 10 m, since light would take about 30 nanoseconds to travel that far, the patterns would flicker at very high random rates.