I wrote:
> It is not precise, but it is reliable, and accurate enough to prove the > point. > The point is, this is a huge effect. It runs at high temperatures and it is at least three times input. McKubre needed a high precision flow calorimeter because he was trying to measure an effect that usually occurs at about a third of a watt and sometimes at 3 W with maybe 5 W of input. That is difficult. You need high precision and accuracy to be highly confident of the result. When there are 300 W going in a 900 W coming out and the cell is so hot it is sometimes incandescent you do not need flow calorimetry. Using a method that is more precise or more accurate than the task calls for does not increase mathematical confidence in the results, or my mental confidence. On the contrary, it decreases my confidence. It shows that the person doing the tests does not understand how to do an experiment. You should always select the simplest and most direct method that will work with adequate precision and accuracy. Never make things more complicated than they need to be. When digital thermometers became widely available in the 1970s, I saw some medical research from a grad student in Japan in which the temperature of lab rats was measured and reported to four digits of precision. Obviously, the temperature of the body of a rat is not uniform, and it varies from moment to moment. A medical researcher who would report that the body temperature was 99.6873°C does not inspire confidence in his ability. He looks like someone who does not understand biology, instruments, error bars, or gradeschool arithmetic. Meaningless extra digits of precision prove nothing. - Jed
