Hi, JJ!

The Tyndall Effect is very handy
to check up on the CS
production process. But it can
yield the two types of erroneous
conclusions to which statistical
inferences are prone.

The false-negative type:
Silver is generated but TE is weak
or absent.

The false-positive type:
Silver is not generated but
a TE is present.

Others have discussed the
occurrence of the false-negative
outcome. I have done some
experimental batches using
low currents running for several
days, with bubbler mixing. As
a control experiment, I shut off
the current and allowed the
bubbler to mix the water.
After several days I noticed a
Tyndall Effect, even though no
CS was produced.

The Tyndall Effect is a remarkable
indicator and I would not do without
it. It could be made into a
quantitative procedure. Tolman and
Vliet invented a "tyndallmeter"
described by Strutt in 1918
(Strutt, Proc.Roy. Soc., London,
94A, 453, 1918.) But it is liable
to false-negative and false-positive
conclusions when used alone.
A false-positive would be more
serious than a false-negative, it
seems to me. It's better to have
an overly potent brew than a 
weak brew thought to be potent,
but is not.

If false-positive conclusions can
be prevented by switching to
another mixing method, I would
do it. But I would continue using
a qualitative Tyndall Effect as a
handy check.

Best regards,

Matthew