Candela (candle), lux, lumen, etc

["Andrei Broder" <broder@pa.dec.com>]

Light flux is measured in lumens.  Light sources are often labeled
with an output rating in lumens.  I posted a while ago a way to
convert wattage for various bulbs into lumens.  (Warning: knowing
the output is not enough since you don't know where the lumens
go. For DOT approved lamps they certainly don't go where you want
them :-)
 
Lumen is actually a "derived unit".  The basic international unit,
measures luminous intensity and is called candela.  Candles and
candlepower are the same thing but these are deprecated names.  It
tells how much flux is flowing through a solid angle,
which is measured in steradians. A point source that has intensity
of one candle puts out a lumen per steradian.  A unit area, all at
unit distance from a point covers exactly a steradian.

Illumination (illuminance) is the area density of incident luminous
flux: how many lumens per unit area.  A lux is one lumen per one
square meter.  Illumination from a point source falls off as the
square of the distance. So if you divide the intensity of a point
source in candles by the distance from it in meters squared, you
have the illumination in lux at that distance.  (Remember this
assumes a single point source in a sphere - no reflectors, lenses,
etc.  What we have are often multiple complex sources.)

Read the rec.photo FAQ for a good intro to all this. I cribbed from
it.

Ideally what you  would like to know when buying a headlight would be
a diagram that shows the illumination (in lux) at various distance
and at a various heights.  I never seen a catalog that shows this
clearly.  Hella has some nice diagrams but they don't give a lux
figure.  PIAA gives you a candlepower (candela) figure, presumably
they integrate the flux outside the headlamp.

The person asking about this units was interested in the rear fog
light.  The short answer is: I don't know.  The long answer is that
it is a 21W standard bulb.  I would guess it puts out around 13
lm/watt so let's say you get 250 lm at the source.  You lose at lot
(75% ?) because of the horizontal position of the filament and
through absorptions in the reflector and through the lens.  So my
guesstimate is that you can approximate the illuminance close to the
axis as if the source has a 750 candela intensity (300 * 4 * pi
/4).  I'd be surprised if I am wrong by more than an order of
magnitude :-)

- Andrei