NetNews Usenet Archive 1992 #31/NN_1992_31.iso/spool/sci/physics/fusion/3107

Text File | 1992-12-29 | 8.1 KB | 151 lines

Newsgroups: sci.physics.fusion
Path: sparky!uunet!paladin.american.edu!howland.reston.ans.net!zaphod.mps.ohio-state.edu!saimiri.primate.wisc.edu!ames!pacbell.com!tandem!zorch!fusion
From: Jed Rothwell <[e-mail]>
Subject: A Straw Man & Calorimeter Drift
Message-ID: <[e-mail]>
Sender: [e-mail] (Scott Hazen Mueller)
Reply-To: Jed Rothwell <[e-mail]>
Organization: Sci.physics.fusion/Mail Gateway
Date: Tue, 29 Dec 1992 23:38:06 GMT
Lines: 139

To: >INTERNET:[e-mail]

Steve Jones is not playing by the rules. He says:

"We do not discard earlier experimental observations, such as release of 3He
from earth's hotspots.  Rather, we build on sound existing data..."

No fair! You set up a straw man yourself, knocked it down, and you blame me.
*You* are the one who wants to throw away old data, not me! You are the one
who asserts that the excess heat violates e=mc^2. You say so, but many good
physicists like Schwinger, Hagelstein and Ikegami disagree. They say the
excess heat will fit into current physics without disturbing or overthrowing
any major theory. They say the old data was right, and so is the new data.
*You* are the one who wants to overthrow Einstein.

I don't know whether you are right, or Schwinger is right. I am neutral. I am
not arguing with your premises, I say your conclusion is wrong. You assert:

1. The data disagrees with Einstein, therefore:

2. The data is wrong.

I say that if statement 1 is true, than the only logical conclusion is:

2. Einstein was wrong.

Anyway, stop trying to prove or disprove experimental data by invoking
theory. To disprove an experimental result, your arguments must be restricted
to a discussion of the specific instruments and techniques used in that
particular experiment. If you assert that McKubre did not find heat, then you
have to show *specifically* what McKubre did wrong. I understand how flow
calorimetry works (because it is delightfully simple, straightforward and
mechanical), and I see no mistake in his work, or in the work of Ikegami,
Kunimatsu, Srinivasan, Mizuno, Storms or several dozen other people. If you
*do* see a mistake, the ball is in your court: tell us what McKubre did
wrong. If you cannot, then you lose, and the heat is real. Stick to the
rules, please.

And while we are on that subject: You ask people to take your Kamiokande data
seriously, and I agree, I think they should. I gather you have a couple dozen
neutrons, at sigma 2 or 3. (Correct me if I am wrong about these details.)
Okay, fine, everyone should take that seriously. However, you can't change
the rules. If your 2 or 3 sigma means something, than what about McKubre's 90
sigma? Do you want us to ignore him, and believe you? Why? It is much, much
easier for him to measure heat than it is for you to measure neutrons. When a
simple, trustworthy thermocouple shows a temperature Delta T, I am far more
inclined to believe it than I am to believe that the huge and complex
Kamiokande detector functioned flawlessly for weeks. The smaller and simpler
the instrument, the more likely it is to work, and the more believable it is.

For that matter, why should we believe in your handful of neutrons, and say
that Yamaguchi's are wrong? You got a few dozen, he got a million per second
for several seconds along with massive excess heat. It seems to me that he
has a lot more going for him than you do. The most logical conclusion is that
you are both right: Jones is seeing a tiny effect, and Yamaguchi, Takahashi
and many others are seeing a much more massive effect. You see a spark, they
see a bonfire. What's the matter with that?

Let us remember that neutron detection is an obscure, arcane scientific art,
whereas people measure temperature and heat in practically every industry on
earth, from automaking to bread baking, from doctor's offices to hospitals to
fishbreeding. It has been an exact science since 1847, when Joule measured
temperatures to the nearest 0.01 F; it is one of the most commonplace,
trustworthy, and practiced skills of all. To argue that Srinivasan cannot
measure 10 C is exactly like arguing that a weatherman cannot tell the
difference between winter and summer. Dick Blue and other experts can find
endless ways to quibble with neutron measurements, but they never argue that
thermocouples cannot measure 10 C. That is why Blue, Huizenga and others
never say a word about the high temperatures: they know perfectly well that
calorimetry proves they are wrong, so they avoid the issue.

Getting back to instruments that function flawlessly for weeks, let me
address this crucial point brought up by Tom Droege:

"I keep doing very long calibration experiments that would appear to show a
very stable apparatus.  Depending on whether I run at fixed room temperature
with the shell servo on or at variable temperature with it off I get
calibration runs with stability of 1 mw or 35 mw.  But from time to time,
there are changes in calibration that I cannot explain. These seem to always
come during an experiment and have always been in the direction of 'anomalous
heat.'"

Amen. I disagree with the last statement only: with other kinds of
calorimeters, I have seen drift in the direction of "anomalous cold,"
especially in Autumn. Maybe your calorimeter tends to drift up, but with
most, the drift is inherently random. After a month, a flow calorimeter can
go a little bonkers for all kinds of reasons, like changes in ambient
temperature, humidity, sand in the hoses... The point is well taken; this is
an important problem. Here are two clear and obvious solutions:

     Don't do long experiments.

     Ignore marginal results.

Do short experiments only. With a palladium Takahashi experiment, if you
don't see heat in two weeks, toss out the cathode: it ain't working. With
nickel, you should see heat in 30 minutes; if nothing happens after a few
hours, forget it. Clean up everything, get new electrolyte and new nickel,
and try again. If it still doesn't work, call me.

That is, of course, one of the many reasons nickel experiments are so much
better and easier than palladium experiments: you get results right away,
positive or negative. You can calibrate 5 points in the morning, let the box
cool down for half an hour, and run a conclusive experiment before lunch. A
calorimeter will not drift measurably in that amount of time. If a 1 watt
joule heater drives the temperature from 20 C to 30 C at 10:00 a.m., and a 1
watt CF device drives it up to 40 C at 11:00, you have a definite, 100%
excess. Naturally, it is a good idea to leave the gadget running a week, to
be sure it exceeds the limits of chemistry.

If you insist on doing a slow palladium experiment, I recommend an on-the-fly
recalibration with a auxiliary electric heater every week or so. Also, try
slowing the flow for an hour or two. And for goodness sake, measure the flow
with a simple liter bottle and a stopwatch. Precision flowmeters can lie like
any other instrument. Another good idea: after you see positive results for a
week or two, leave the cell running, tear the calorimeter apart, and replace
vital parts like the power supplies, thermocouples, flowmeter, and computer
interface board with backups and spare parts. As you replace each part, check
to be sure the heat is still there.

"Ignore marginal results" should be clear to all readers. Ignore
recombination -- toss out the gas and look for an excess like 1.7*I*V. Ignore
teeny, tiny excesses, like 3%. Other people are getting 70%, 200%, 1000%, so
why bother reporting if you can only get 3%? This is not 1989 any more. Drown
out the noise: put in lots of electricity, like 100 watts, and look for 170
out (or at least, 130). Look for BIG and obvious results. When you see a 70%
excess, you can write off all the nasty little marginal errors that creep
into measuring heat (or measuring anything else). Assume the worst case;
assume that every possible error is in the positive direction. Maybe that
pushes the excess down to 60%. So what? Who cares? Big is easier to measure
than small. You might be wrong about a 0.1 C temperature, but you can never
be wrong when the temperature climbs 10 C.

- Jed


Distribution:
  >INTERNET:[e-mail]