KShanahan. What’s that story about the time you were trying to dispute some ‘cold fusion’ findings by showing a non-correlation between two factors, but ballsed up the analysis, and ended up unknowingly proving it? Or something. Abd used to write about it. Never heard your side of it. Maybe something about a horizontal line on a graph?
In my 2010 J. Env. Monitoring paper, there is a slight error in my discussion
of a specific figure. Abd has tried to use that to discredit everything I write
in a ‘throw the baby out with the bathwater’ style. I replied to him here on
lenr-forum, but in brief…
I find it sad that Kirk hasn’t learned anything, apparently, and his skill at mind-reading sucks. I do not use that error to discredit everything. It could illustrate, however, how his eagerness to debunk can overwhelm his understanding of fact. (And how his Letter was not carefully edited.) He calls it a “slight error.” It reverses the sense of his discussion there.
Kirk digitized a graph from Storms The Science of Low Energy Nuclear Reaction (2007), Figure 47, page 87, wherein Storms presents results from ten heat/helium measurements by Miles, and three from Bush and Lagowski.
The graph is, in my opinion, mislabled, in a way that made Shanahan’s error easier to make.
To study this, Shanahan digitized the plot, measuring the position of each point. In fact, the data is in table 7 on the next page.
The correct comment on the figure in question (which shows #He atoms/
Watt-sec which can be related to the MeV/atom He purportedly released in the unknown LENR (the value is used to draw the horizontal line you reference)) is that the Figure shows nothing because the noise level of the results is so high it includes zero and values up to 2X the maximum postulated for the LENR.
If it shows nothing, why did Kirk feature it in his article? The Miles work was indeed quite noisy; however, Shanahan clearly exaggerates the significance of that. The purpose of the plot was to show scatter, and it shows how the scatter declines with increased power (and helium). There is, as Shanahan points out, a flyer, but this flyer is the result with the least reported power, 30 milliwatts. That sample also had the least helium, 0.61 x 10^14 atoms per 500 ml (of collected electrolytic outgas).
(There are many details to this experiment that I’m not going into now, but what is correlated with produced helium is energy, not power, but … the way the experiment was done and data was collected makes those equivalent. I.e., the power plotted is for a certain period.)
Background helium was 0.51 x 10^14 atoms per 500 ml, so that flyer was very close to the noise.
Abd thinks the Figure shows the reaction is related to D-D fusion, but to get that he (a) has to ignore a high flyer datum,
I don’t ignore that datum, and one of the great features of Miles’ work is that he reported all his experiments. This is not cherry-picked data. That the ratio of helium to heat was within an order of magnitude of the theoretical deuterium to helium conversion expectation was considered amazing by Huizenga, in 1993. This is a strong result, compared to the much less specific excess heat reports, but this is far from the best data. It was simply enough for its time. It can be seen on the Bush and Lagowski reports that the ratio is much more stable with better helium measurement accuracy. There is a broad agreement across many experiments other than Miles that the observed ratio is consistent with a retention of about 40% of the helium — if the reaction is deuterium conversion with no escapes. Of course, there is an escape, trapped helium, apparently, and that is certainly reasonable as an explanation for the “missing helium.”
There is more. Perhaps Shanahan should read my paper. The work that I recommend in it is being done, the heat/helium ratio is being measured with increased precision, and I expect they will capture all the helium, it is now known how to easily do that.
and (b) has to assume some large fraction (~40-50%) of the produced He atoms are trapped in the solid electrode.
I make no such assumption. I observe what has been found experimentally. I do not know what the true ratio is. I am aware of a level of evidence that it is within 20% or so of the “fusion” value, i.e., 23.8 MeV/4He, even that does not “prove” that the reaction is some kind of fusion. My hope is that we will know, in fairly short order, a tighter value.
The trap, by the way, would be near-surface, based on substantial evidence going back to Morrey et al. The location of trapped helium — it is not found in the bulk — is a clue as to the reaction site, for if helium were formed in the bulk, it would stay there indefinitely, absent extraordinary measures. Helium stays put in palladium lattice, and this is known from the behavior of 3He produced in the lattice by tritium decay.
Ignoring the flyer is not a good idea. It is a real data point derived from the
system under study. It shows you how spread the data can be, thereby supporting the assertion that the method is too inaccurate to be useful. The % thing has no technical support, it is just handwaving designed to ‘explain’ the result that the average value comes in well below the theoretical one (the horizontal line).
There are many other reports that align with Miles and Bush and Lagowski, done with increased precision. Shanahan is, here, handwaving himself, not looking at the actual experimental data.
I don’t really care all that much about that “horizonatal line.” Rather, what is observed experimentally? It is obvious that with a helium background of 0.51 x 10^4 atoms, a measurement of 0.61 is not likely to be tight, so of course that would generate flyers. Shanahan is ignoring the more solid results, turning a feature of Miles (reporting all the results) into a bug (which isn’t a bug, because Miles also reports no-heat results, which uniformly came in around the background value. If we were to report those results with high (and unjustified) precision, we would really have flyers. I’m actually appalled by this argument.
No, I didn’t ‘unknowingly prove’ anything. The ‘CCSH’ idea authors however, in proving their straw man false, unknowingly supported my ATER-CCS proposition.
First of all, this is not a discussion of whatever errors those authors allegedly made. It’s a discussion of what Shanahan wrote, and, now, how he is explaining his error away.
So what mistake did Shanahan make? Here is his response to Krivit and Marwan.
Storms also presents another heat He plot as Figure 47 in his book. However, this plot shows no correlation such as presented by K&M or Hagelstein.
Well, it does, that is, if we were to plot helium atoms vs accumulated energy, we would see them as a line, like the Hagelstein plot (which is from McKubre and the Case experiment, and presents other mysteries, but not this one). Perhaps I’ll do that plot and present it. The Storms plot was designed to show the scatter, not the correlation itself.
In fact, digitizing the data of Figure 47 and neglecting the one obvious flyer at the lowest excess power value produced a correlation coefficient of 0.0995. This is a highly statistically significant number indicating strong confidence that in fact no correlation exists.
Indeed. And if we look at, say, crude measurements of the diameter of a circle vs the ratio of the diameter to the circumference, we would find a low correlation coefficient, because that ratio is a constant, and a low correlation coefficient indicates, in this case, that the ratio of helium to heat does not vary with the heat, it is at least roughly constant. Hence Shanahah did show (not “prove”) the opposite of his intention, by misreading what he was studying.
Including the single flyer produces R ¼ 0.38, which is indeterminate as to whether a correlation exists or not. This plot was constructed from data from two different laboratories, one from 1998 and the other from 2003.
The plot was from Miles (2003), but the experiments were much earlier, that was a review, and it was from Bush and Lagowski, 1998 (and I haven’t checked, but I think that was earlier as well). Miles’ early helium measurements were quite crude.
Apparently, it depends on where and when one gets the data as to whether or not a correlation is observed. This is a typical problem observed when one attempts to plot two truly uncorrelated variables in a correlation plot.
Except that all roads lead to the correlation. There really isn’t any significant contrary evidence; and disputes exist about the value of the ratio, hence the value of more precise measurement. Back when Shanahan’s Letter was published, I noticed his error and wrote him (before I published it). He responded with an insult.
Just because I can, I will present the Marwan et al response to Shanahan on helium. They did not notice his blatant error, my guess is that it simply made no sense to them, and they focused on the Case SRI plot. I used that plot in my own paper, perhaps against my better judgment; the editors wanted some eye candy, but there are “issues” with the Case work. It’s still of high interest, to be sure, but … it was never formally published, for one thing, and there was some unexplained anomalous behavior. It showed a quite different retention ratio, perhaps because it was a very different experiment (gas-loaded, coconut charcoal coated with palladium)
2.5 Temporal correlation between heat and 4He
The China Lake experiments on the correlation of heat and helium-4 production carefully ruled out contamination.[5,21] Control cells were run in the same manner as the cells that produced excess power. Excess helium-4 was measured in 18 out of 21 cells that produced excess heat. None of the 12 control cells yielded excess heat or showed excess helium-4 production. The random probability of obtaining the correct heat/helium-4 relationship in 30 out of 33 studies is 1:750,000 (see Appendix C of Ref. 5). Furthermore, it is very unlikely that random errors due to contamination would consistently yield helium-4 production rates in the appropriate range of D + D fusion of 1011–1012 atoms/s per watt of excess power.[5,21] Case reported production of extra energy by nano-particles of palladium on the surface of charcoal when the material was exposed to D2 gas at temperatures up to 175°C. McKubre et al. replicated the claims. The results of this experiment are shown in Fig. 3a. This is the Figure 6 in the 2004 report prepared by Hagelstein et al.24 that Shanahan discusses. This plot illustrates the real-time correlation between excess heat and the growth of 4He concentration in a metal-sealed, helium leak-tight vessel that was observed in the SRI replication of the Case experiment. In his critique, Shanahan briefly touches on the quantitative and temporal correlation of excess heat and 4He production with an odd argument posed as a rhetorical question: “If in fact there is no excess heat, then what exactly is being plotted on the Y axis?”
Where does the “fact” that “there is no excess heat” come from? It comes from the strained logic that the CCSH “explains all excess heat results.” As discussed above, CCSH has no validity. Plotted on the X-axis of Fig. 3a is the increased level of 4He measured in samples drawn from a helium-leak-tight vessel. Again in his critique, Shanahan asks: “If there is no
proof that the observed He is not from a leak, then how does one know that is not what is being plotted on the X axis?” This is easily explained. The shape of the measured 4He vs. time curve is quantitatively different from that of a convective or diffusional leak of ambient 4He into the closed cell. The measured and plotted [4He] first remains constant (no leak), then rises approximately linearly to roughly twice the ambient air background level. A shape consistent with the hypothesis Shanahan proposes would be exponential with greatest slope at time zero and rising asymptotically to the environing background level (5.22 ppmV). So an
explanation invoking an in-leak from the ambient can be seen to fail quantitatively.
That rhetorical question is vintage Shanahan. What is plotted is independently calculated heat. Miles measured helium blind. The lab he sent the samples to did not know which cells had produced heat and which did not.
Come up with a final word for this post better than “error,” and you will win 50 quatloos, redeemable in advertising credits. Your message here!
“Better” means “more fun.” This is a blog post, not a scientific journal article. The intention of the title was playing on the claim that the error was “slight.” I thought, “itsy bitsy,” and then came the obvious result of pattern recognition. That song hit #1 on the charts in August, 1960, when I was at the University of California, Santa Barbara, for a summer program for high schoolers going into their senior year.
All submissions become my property. The winning entry, as judged by the chief cook and bottle washer at CFC (moi!), will be published as the revised title of this post. The authors of losing entries will be subjected to natural public ridicule by being published in comments here. In fact, that is how to submit a contest entry, by commenting on this post. Offensive comments will be deleted without mercy, but mild offensiveness may result in winning. Have fun!