This is a nice example of why peer review is, although being the best process currently available to evaluate the quality of a scientific paper, still far from perfect. Here, a big name was all it needed to publish a paper. A superb take-down was given by professor PZ Myers. Since some people on this site might be fainthearted, I'll edit out his lucid language where appropriate. Visit the site for the full tour-de-force.
Let us begin with his summary of the work:
Montagnier claims in several papers that the DNA of pathogenic bacteria emits an electromagnetic signal, and further, that if you dilute that DNA homeopathically so that no DNA is actually present, the water continues to emit that same signal. Further, if you put two vials of homeopathically diluted EMS emitting water next to each other, the signal can move from one to another. And further, only bacteria and viruses pathogenic to humans produce this signal; ordinary E. coli does not. [...]
There is no sensible explanation given for this phenomenon, only some wild-eyed speculation that "water molecules can form long polymers of dipoles associated by hydrogen bonds" that may be "self-maintained by the electromagnetic radiations they are emitting. [...]"
Note: This has actually been disproved in the lab. Water has no long term memory. This question was also handled on this site
I'm not going to criticize the paper because it postulates a mysterious mechanism with no coherent physical cause, though. I read the paper and call it [bad] by virtue of the sloppiness of the work. I disbelieve it, not because I'm predisposed to find it unlikely (although I do), but because it's an appallingly bad paper.
What follows is a lengthy description of the experiments and data processing done by Montagnier, and PZ doesn't look to kindly at this. His conclusion:
There are a couple of other indicators that this is pathological science. They're looking at a minuscule, variable result that is prone to be picking up all kinds of irrelevant signals, yet nowhere in the entire paper can I find the word "blind". This is the kind of experiment that demands extreme rigor and care, yet the authors don't even bother to describe the protocols used. That's a warning sign.
There is another methodological problem that makes this paper suspicious:
They are claiming the existence of a truly remarkable phenomenon. A good scientist would focus on one fundamental observation, the claim that they can record species-specific bacterial signals with their crude apparatus, and nail that one down good and hard and believably. But no. They show off some very poor raw data and then rush off to dilute the experiment a trillion fold and claim to see the same signal. I found the first observation dubious, why are you showing me something even more unlikely?
Although the paper has been published in a peer-reviewed journal, there are even more problems:
And finally, another suspicious sign are the dates. This paper was submitted on 3 January 2009, revised on 5 January 2009, and accepted on 6 January 2009. That's an unbelievable turnaround, especially for a paper with such incredible results, and the revisions must have been trivial to be able to be whipped around in a day. Yet it's an awful paper that I would have shredded in a sea of red ink if it had come to me. Who reviewed this, the author's mother? Maybe someone even closer. Guess who the chairman of the editorial board is: Luc Montagnier.
The original reference:
Montagnier L, Aissa J, Ferris S, Montagnier J-L, Lavallee C (2009) Electromagnetic Signals Are Produced by Aqueous Nanostructures Derived from Bacterial DNA Sequences. Interdiscip Sci Comput Life Sci 1: 81-90.
Although PZ has not published his criticism in a peer-reviewed journal himself, the reservations he voices all make sense from a skeptical point of view. The methodology is poor, the proceedings of the publication are suspicious. What I recommend from a skeptical point of view is to withhold belief in Montagnier's results until they are repeated by other labs with more attention paid to a sound methodology.