Re: Dayton Miller's Data have no Real Signal

David Thomson wrote:
Tom Roberts wrote: 
I do indeed have the data in an Excel workbook. But it is not mine to
publish. I will inquire about redistributing it. I do not have MathCAD
nor know how to use it.

Maybe if you could explain the exact functions you use, I might be able
to coax it out of MathCAD.

About the only math function I use is a Digital Fourier Transform with 320 points. Excel only implements an FFT, which is limited to lengths of a power of 2. I suspect MathCAD will be the same, but don't know.


For a project of this importance,

In the overall scheme of physics, this is not important at all. This is just clearing up a minor anomaly overlooked by most physicists. Of course it will be devastating to those few who are emotionally invested in Miller having a non-null result.


Excel does not have the 320-point DFT that is needed,

I'm not sure what you mean here,

A digital Fourier Transform (DFT) is necessaily a sum over an array of samples of the signal. Miller took data in runs of 320 points (20 turns, 16 markers per turn). So to avoid discarding data, one needs a DFT routine that can handle 320 points[#]. I didn't find one after a 5-minute search, so I wrote and debugged one in Java (took ~5 minutes to write and ~20 minutes to test[@], IIRC; much easier than reading the quirky .csv files exported from the Excel file).

	[#] How quaint (:-)). Today of course one would naturally
	pick 256 points as a unit.

	[@] I am a professional, and know the necessity of testing.


I do have a
question about your comment on Maurice Allais' work. It is easy to say
that he was seeing "shapes in the clouds," but his work was peer
reviewed and published. He stated with absolute certainty that the
data was not offset by thermal anomalies.

I have only the English translations from his website. In none of those papers does he justify that claim, except to assert it. In particular, he cites no errorbars or error analysis, he just marvels at the patterns he claims are there.


I understand your "error
bars" point, but he seemed to be aware of that, too.

Where? In which paper does he discuss errorbars? Is there an english translation available?


Can you be more
specific as to how Maurice Allais made the great error of "seeing
shapes in the clouds?"  That is a very strong position for you to take
when his paper has already been reviewed and published.  You will need
to address this.

AFAIK he just marvels that the glorious patterns he sees in the data, without justification or error analysis. That is not science.

	It was acceptable in Miller's day. But not today.


Tom Roberts	tjroberts@xxxxxxxxxx
.

Quantcast