[mikeash]

Can you tell me where to find the original information about this 12/12 LENR NASA run? I must not know what search terms to use.

Edit: to be clear, I'm wondering what the experimental setup was. What sort of device were they testing, how were inputs and outputs measured, etc. Not just the results, but the experiment from which they got those results.

[Recurecur]

Well it appears I get to eat a little crow. It was not NASA work, but it was a peer-reviewed paper cited in the NASA materials:

http://arxiv.org/abs/1305.3913

The NASA material is worth reviewing in its own right though - and the Navy has conducted quite a few LENR experiments of its own.

[mikeash]

I find this paper to be extremely confusing.

They state that they couldn't weigh the device because it was already running when the test began. Why would they start it before doing basic tasks like seeing how much it weighs?

Oh, because the test was conducted at the company that makes the devices, and I guess the company set up the device and turned it on before the scientists got there.

Then one naturally wonders, who set up the equipment to monitor the electrical power consumed by this device? As far as I see, the paper does not say.

One further wonders, why does it need a continuous input of electrical power at all, if that power is only connected to resistor coils in the device as claimed? The paper says that electric heat is needed to start the reaction, but why doesn't the reaction sustain itself on its own heat after starting? Why can't they just unplug the thing at that point?

Finally, the experiment was conducted with the device sitting out in the open, radiating and convecting into the air. Output power was estimated by pointing an IR camera at it (measuring only half of it!) and calculating power based on the observed temperature. All of this in an environment controlled by the company that makes the device.

I hope you can see how this experiment is 1) nothing like the definitive test I describe and 2) suspicious as all hell.

The test I describe is not difficult to set up and would not be difficult for a real, working device to pass. So again, why don't they do it? That they don't doesn't automatically mean it's a scam, but then what is the answer?