Seems the variations are very big, i.e., June compared to December 2010.
Is this correctly measured?
I've read here in avast forums it could be a biased measured market share (although not it is malicious, but just that the statistics are measured in a limited sample of computers).