I read the article you linked and have a few (very) quick comments. Firstly, the findings do seem to indicate that Canova has at least some level of efficacy, however the argument that the results are statistically significant is flawed. A sample of 20 (versus controls) would not provide a sufficient result from a power analysis to be at all conclusive. That's fine however... as the authors do concede this.
The second problem I have is that the study is not fully independent. I had a quick look on pubmed and no-one outside of the author's establishment have ever looked at Canova.
Lastly, I suggest that there may be a certain amount of bias (unconscious or not) instilled by the authors themselves. This was noted in a critical review of a very similar study and you can read about it in the link below:
(if you have access to a subsciption service, the paper is: Maddox, J.; Randi, J.; Stewart, W. (1988). ""High-dilution" experiments a delusion.". Nature 334 (6180): 287?291.)
If one day a full series of genuine double-blind trials are released proving that a homeopathic medicine is effective, I will certainly be open-minded to its use. I personally can't see it happening though. You will read regularly if you google homeopathy; 'it defies the laws of physics'...
