From: Simon Richards <s.richards@physics.org>
Date: Tue, 20 Jan 2004 21:12:21 +0000
Subject: [TPIN] Cryogenics
- The Study
I have been in touch with Chip Jones, the author of the study in
question, and I have now read his Master's Thesis and a scientific
article he has published on the subject. They are available at the
following web address, together with a lay language article, and I urge
anyone with an interest in this to read them.
http://www.tuftl.tufts.edu/mie/Pubs/cryo_index.htm
In my opinion this study is scientifically sound (no pun intended) and
it shows that there is no statistically significant effect of cryo
treatment on the sample of trumpets studied (10 new Bach Strads). If
there is any effect it is so small as to be obscured by other sources
of variability (from player to player, or from day to day for a single
player, for example).
[pause to let that statement sink in]
You should read the experimental details for yourselves, but briefly
they randomly selected 10 new Bach Strads (Model 180, 37 bell,
lacquered) from the Selmer production line, and randomly selected 5 of
those to receive the cryogenic treatment. They had 6 players who each
played all the trumpets in a double blind trial. They recorded each
player playing the open harmonics of each horn and analysed the power
spectra of each note using the fast Fourier transfom (FFT) to analyse
the steady-state phase of the note and the short time Fourier transform
(STFT) to analyse the attacks. They also asked each player to provide a
qualitative analysis of each trumpet and, in the case of those who had
previous experience of playing cryo treated horns, they asked them to
guess which horns had been treated. Ultimately their results are based
on the 3 most proficient players in the group.
The results show no statistically significant differences between the
two groups of horns in any of these tests. There are some subtle
differences in some of the results, but these were inconsistent, in
some cases not repeatable, and not statistically significant.
Now, no real experiment is perfect and a key assumption in this study
is that the 10 trumpets are identical, aside from the fact that half
were given the cryo treatment. The authors acknowledge that this
assumption is not completely valid due to the individual
characteristics of hand-crafted instruments. They presume that subtle
variations will be averaged over the sample size. This is fair comment
as long as the sample size is large enough - so is it? Given 10
trumpets randomly organized into two groups, what is the probability of
there being a systematic difference between the two groups? It's
possible, but unlikely. The probability gets even smaller as you
increase the sample size, but in the real world we have limited
research budgets.
Personally I would have liked to have seen the testing and analysis
carried out on the 10 horns before half of them were treated, as well
as afterwards, but again budgetary constraints have to be taken into
account. Nevertherless I believe the results to be scientifically valid.
Another limitation of the study is that it is confined to the Bach
Strad. Now although there are apparently quality-control issues with
some Strads they are, in general, well-made high quality professional
trumpets. If there was an
effect and if it was due to
assembly stresses then perhaps if would be more apparent on cheaper
instruments. But I guess nobody is getting a cryo done on a Lark or a
Bundy?
Other proposed explanations for any perceived effect of cryo treatment
are the chem clean or placebo effect. This study takes the chem clean
out of the equation because it uses all new horns. I'll discuss placebo
in a separate post, as this one is getting long!
Dr. Simon Richards
