Re: Cahill's recent paper readdressing MMX et al

From: Tom Roberts (tjroberts_at_lucent.com)
Date: 01/23/05 

Date: Sun, 23 Jan 2005 20:15:20 GMT

I have just posted an article on this in a new thread with subject:
   An Analysis of the Resolution of the Michelson-Morley Experiment.

Paul Stowe wrote:
> It seems that M&M DID indicate a cyclic pattern during rotation
> see page 340, Figure 6 of http://www.aip.org/history/gap/PDF/michelson.pdf

Of course they did! By subtracting off the assumed-to-be-linear
systematic shift per turn they _FORCED_ the data to have a "cyclic pattern".

>>I doubt that the quantity of data is sufficient to draw any hard
>>conclusions, [...]

This depends upon what type of "conclusion" one attempts to draw. The
following are hard conclusions drawn from the data in the link Paul
Stowe gave above:
  1. The MMX measurements are solidly consistent with a null result.
  2. Estimates for their basic resolution in measuring the position
     of a fringe are 0.14-0.15 fringe width. This is based on a
     statistical analysis of the data in that paper.
  3. Conclusions that they saw a non-zero effect are bogus, because
     the variations in the data are below their resolution.
See my post in the other thread referenced above for details.

> If I plot actual data on a grid cyclic patterns are rather easy to spot
> and recognize. Of couse one MUST set the axes such that the data spread
> is sufficient to fill a perferably square area. If it looks like someone
> to a shotgun to the page then, of course, randomness is also apparent.

But you also must plot the errorbars. If the error bars are larger than
the variations in the data, any conclusion that the data does vary is
unwarranted.

> The human eye is quite good at pattern recognition.

Yes. And quite good at finding "patterns" that are demonstrably not in
the data itself. Experimenter's bias comes in many forms....

> if you have collected tens of thousands of data
> points over several days, a scatterplot WILL illuminate any cyclic processes
> that might be present.

You are naive. The human eye will _always_ find patterns in a
randomly-populated scatterplot that is reasonably dense. Physicists have
known for many decades that one _MUST_ perform a statistical analysis on
both the data and the errors/resolutions.

> For M&M, even their few showed a cyclic
> pattern according to their paper referenced above.

But the "cyclic pattern" is forced by their analysis technique, and is
smaller in amplitude than their resolution. You are looking at errors
and proclaiming them to be significant. In the figure on p340, the
errorbars are taller than the entire plot (0.057 \Lambda). See my post
in the other thread referenced above for details.

Mountain man wrote;
> However Cahill correctly points out that the original analysis
> (used to arrive at the 6-8km/s) was flawed, not taking into
> account the refractive index of the air, for a start.
>
> Cahill is correct, and the first person
> to point out this oversight since 1870.

Cahill's analysis is fundamentally flawed -- he imposes his persoal
hopes and dreams onto the data. A proper comparison of this "theory" to
  the MMX data, using a realistic estimate of its resolution obtained
from the data, would indicate the experiment is consistent with his
"theory" for ANY speed relative to the ether between zero and several
thousand km/s, in any direction. Like Stowe above, Cahill is looking at
errors and proclaiming them significant; he then divides by a tiny
number to obtain agreement with his personal prejudices.

        I put "theory" in quotes because he invokes MAGIC to
        exclude experiments that otherwise refute his "theory"
        (any measurement using a solid, such as Shaumir and Fox).

> Michaelson and Morley, Miller and others performing
> the gas mode interferometer experiments did not know
> the physics of the apparatus, and could not therefore
> understand their results (6-8 k/s).

They understood them perfectly well in the theoretical context of their
time. Nobody could reasonably ask for more.

> Either Cahill is correct about the physics
> of the gas mode michaelson interferometer
> requiring admission of a term for the
> refective index of the gas, or he is not.

His current papers are fundamentally flawed in their "analysis" of
various experiments. And his invocation of magic is just ridiculous.

Tom Roberts tjroberts@lucent.com