Tom Roberts wrote:
> Eric Gisse wrote:
> > Tom Roberts wrote:
> >> [cahill article]
> >> Dead wrong. In particular, he does not understand error analysis, and
> >> thinks that Michelson and Morley, and Miller, both "detected absolute
> >> motion". In fact, they did not, and their measurements are fully
> >> consistent with SR. But one must do an error analysis to fully
> >> understand this, and Cahill did not do one.
> >
> > Do you know why Cahill thought it was best to use the data from a
> > century old experiment rather than more recent incarnations of the same
> > experiment which have tighter error bars?
>
> Because the "signal" in just about all of these experiments is
> proportional to their resolution or errorbars. So the older ones have
> bigger "signals". He simply does not understand that a statistically
> insignificant "signal" is useless and a figment of his imgination.

If that is the case, the older ones have lower signals. For example, if
you are using a high resolution encoder to measure position and then
calculate velocity, the signal is bigger (sum of counts) and your
accuracy better provided your counter's clock can keep up. The lower
the resolution the lower the signal and the higher the noise.

But a signal of 300 Km/s can be hardly associated with noise. It is
just to big. Increasing the resolution will probably increase the
figure than lower it.

>
> As I say: amateurs look for patterns, professionals look at errorbars.
> Cahill is an amateur.

This gum you chew constantly about errorbars is a childish red herring.
The signal is WAY to BIG to be affected by resolution significantly.
That could happen if we were talking about mm/s or nm/s.


> > I have always thought it was odd that he did that.
>
> It's not merely "odd", it's dishonest and just plain wrong.

Dishonest is the attitude of the scientific community for failing to
declare SR and GR garbage.

Mike