I decided to see what my jpeg only digicam can capture in the way of light
in stops in real terms, in a real situation

After reading:
“With 8 bits there are 256 possible values corresponding to decimal numbers
from 0 to 255. So with an 8 bit A/D converter you can take a range of analog
(sic)
numbers and convert them into a digital number with a range from 0-255. Now
let's ask the question "what's the dynamic range of an 8 bit A/D
converter?".

“Well, it's the ratio of the smallest signal it can record to the largest
signal. The smallest signal is "1" and the largest signal is "255", so the
dynamic range is log(255/1) = 2.4.”
(ref = http://www.photo.net/learn/drange/ )

so the 65k colours from a 16 bit image therefore have a dynamic range of 4.8

it is then extrapolated that 2.4 / 0.3 (stops) = 8 stops.. 8 bit jpegs have
a dynamic range of 8 stops (..and therefore 16 bit images must have a
dynamic range of 16 stops?)

Using this logic it is clear that jpegs have less dynamic range than other
formats.

hmmm..

from
http://www.luminous-landscape.com/tutorials/blended_exposures.shtml

"Colour transparency film has a useful dynamic range of about 4-5 stops
between the lightest area with any detail and the darkest. Beyond that is
found featureless black or transparent white. Colour negative film has a
couple of additional stops and B&W film another stop or so. But, nature can
present the landscape photographers with scenes having a dynamic range of 10
stops or more — double that which slide film can handle"

it seems 'dynamic range' can mean a multitude of things..

So I got out the old Minolta V with the spot meter attachment and the Gretag
colour chart and the lowly Sony f717 which is capable of only shooting jpegs
and set to work to find what the dynamic range of the Sony actually is in
stops..

I spot metered the white and the black patches at ISO 100, 1/60th and got
readings of f8, f45 bang on.  The other patches were a little off, but given
there are 6 of them, then the white and black points having those readings
looks good.

I then shot a sequence of images (in the shade *) at f8 across 6 stops as:
1/500th , 1/250th , 1/125th  , 1/60th  , 1/30th  , 1/15th

* which is why the values below show a little more blue than a neutral
colour should.

I then fired up Irfanview and cropped then resized the images, then used
ColorCop to measure the RGB values, recorded in the table below.  Note I
applied a 50% blur to each cell to compensate for the variation in tones
between individual pixels as they were not (cannot!) be a uniform colour.

1/500th  1/250th    1/125th     1/60th       1/30th        1/15th
88,94,91 97,108,111 148,152,151 202,210,218  237,240,245   255,255,255
63,64,68 73,82,91  119,125,126  170,183,196  218,223,230   250,251,253
39,42,44 46,52,64  91,95,98     138,150,164  190,202,208   229,234,243
23,23,26 27,31,36  55,63,63     103,115,125  151,161,170   197,210,222
11,12,12 16,14,19  28,29,33     57,69,80 104, 113,125 149,  162,178
4,4,4    5,5,7     10,12,12     20,26,31     42,55,64      82,96,115

Just a note on that final value of 4,4,4 – I decided to do a shot at 1/500th
f8
With the lens cap on to record the values for the dark current which came
out to be 1,1,1 so 4,4,4 is still significant.

What does this tell us?  Firstly that my Gretag colour patches are a little
out of whack.  Second it suggests that my sensor clearly can capture more
than TEN stops of light - clearly more than the *limit* of 8 stops a jpeg is
theoretically capable of in digital terms.  It also indicates the
sensitivity curve is a little different from what you'd expect ;)

Now to do this with a REAL camera that can produce jpegs and raw / tiff file
formats to see what we can see

k