The results of a third-party investigation of Rossi's E-CAT reactor have appeared on the Cornell arxiv, and the conclusions of the tests are at the very least startling:

An experimental investigation of possible anomalous heat production in a special type of reactor tube named E-Cat HT is carried out. The reactor tube is charged with a small amount of hydrogen loaded nickel powder plus some additives. The reaction is primarily initiated by heat from resistor coils inside the reactor tube. Measurement of the produced heat was performed with high-resolution thermal imaging cameras, recording data every second from the hot reactor tube. The measurements of electrical power input were performed with a large bandwidth three-phase power analyzer. Data were collected in two experimental runs lasting 96 and 116 hours, respectively. An anomalous heat production was indicated in both experiments. The 116-hour experiment also included a calibration of the experimental set-up without the active charge present in the E-Cat HT. In this case, no extra heat was generated beyond the expected heat from the electric input. Computed volumetric and gravimetric energy densities were found to be far above those of any known chemical source. Even by the most conservative assumptions as to the errors in the measurements, the result is still one order of magnitude greater than conventional energy sources.

Now, if the study is correct, one needs to first of all figure out what those additives are, and then understand whether the energy output comes from nuclear sources or chemical ones. A model is direly needed, I would say; the secrecy behind the project does not help figuring out whether this is a very elaborate scam or a Nobel prize worthy discovery.

I continue to believe in the scam hypothesis, but I must admit that this study impressed me for its reported result. Here is an important passage in the longish report:

For a further confirmation of the fact that the E-Cat HT2’s performance lies outside the known region of chemical energy densities, one can also calculate the volumetric energy density of the reactor, by referring to the whole volume occupied by the internal cylinder, namely 1.52π33 = 233 cm3= 0.233 l. This is the most conservative and “blind” approach possible.
Taking the figures from the worst case, we get a net power of 800-354=446 W; by multiplying this by (3600 · 116), we find that 185 Mj where produced. Thus, we have a volumetric energy density of 185/0.233 =(7.93 ± 0.8)10 2 Mj/Liter, meaning that even by resorting to the most conservative and “worst case scenarios”, where the total volume of the reactor is comprehensive of the 5- mm thick steel cylinder, we see that we are still at least one order of magnitude above the volumetric energy density of any known chemical source.

I don't think I am going to read the paper with more attention than I already used with it; this is not my field of research so I would not learn much more anyway. But I must say I will from now on follow more closely the developing story of Rossi's E-CAT...

In the meantime one can give a look at the story as reported in Wikipedia: Already in the introductory notes we find statements that appear to settle the issue:

An international patent application[1] has received an unfavorable international preliminary report on patentability because it seemed to "offend against the generally accepted laws of physics and established theories" and to overcome this problem the application should have contained either experimental evidence or a firm theoretical basis in current scientific theories.[11]

The device was demonstrated to an invited audience several times, and commented on by various academics and others, but was not independently tested. Mark Gibbs of Forbes commented: "until a verifiably objective analysis is conducted by an independent third party that confirms the results match the claims there's no real news".[12]

Well, I would say that at least that step is cleared now, if the study reported in the arxiv preprint is believable...