One more year of raw data from tritium decay experiment now available

Is there any reason not to calculate an intermediate result? This experiment was looking for seasonal changes in something? Was it gravitational impacts on the decay rate?

Subtract off the best fit exponential curve first and give it a try. The implicit discontinuity across the start end boundary of the FFT will otherwise create far too many artefacts to see anything interesting.

If you do see anything seasonal it will most likely be ambient temperature related. You should also look for a weekly 7 day signal for weekday vs weekend ambient variation which will also likely be present.

You have nothing to lose by FFTing it now. You might get slightly better sensitivity to anything that is really there but a marginal detection by computing the FFT amplitude only for each year and averaging that (ie throw away the phase). FFTW should do it fairly easily either way.

For example, if your data is half a cycle of cos(x), starting at x=0, removing a trend line produces something very close to sin(2x).

With a known functional form, it should work better.

Cheers

Phil Hobbs

On a sunny day (Wed, 13 May 2020 17:23:54 -0700 (PDT)) it happened Ricky C wrote in :

Well somebody did a paper on the data (see the webste), years ago. Anyways I did do the Fourier transform last night and have solid data now that there ARE NO seasonal changes. Cheating on the transform by leaving out the slow changes and increasing gain in the display that way shows interesting almost exactly the same patterns in both tritium light detectors, but a factor 8000 below the decrease in brightness due to the tritium half-life. So conclusion is solid: No change

Writing the paper now, and it will have to be apple -, pear -, or cherry reviewed and will then hopefully go on the site. As to the FT: never seen such a clear solid line :-) No spectra.

Good hardware!

Here is the detector 1 data: 70017 datapoints, one per hour starting in 2012 until May 13 2020:

:-)

Yes, in decay time.

No, sun-earth distance See the website, this paper made me test:

Others have also tried to repeat their results and found no such effect. Maybe they had local interference on their PMT or whatever it was setup. That is why I tried the tritium light output, as PMTs are sensitive to just about everything.

On a sunny day (Thu, 14 May 2020 12:14:34 +0100) it happened Martin Brown wrote in :

Thank you for the suggestions, I did the transform yesterday, see my reply to Rick The thing is in an electronics controlled thermostat, so no problem there. If you want to have a go at the data:

PS Faster than I can keep up, feedback comes in via email. Somebody just did a complete analysis of the data, and gave me permission to publish it:

Now linked from

Nice work!