Matt Cutts: Here's a question from Stav Zilbershtein in Israel who asks: 'Hi Matt, what if there are external links to a URL that is not the canonical one? Does the power of links pass to the original, original URL or do links pointing to a non-original URL get lost when using a canonical tag on a website?'
Well, let's just review for a second.
Ideally you can have site architecture such that you don't have duplicate URL's.
Failing that, if you can do a 301 redirect from the extra or duplicate URL's to the preferred or canonical URL, that's the best thing.
If you have no choice whatsoever – you can't fix the site architecture and you can't generate 301 redirects -- then you can use rel equals canonical to say 'This URL is really the same as this URL'. In general, whenever Google combines duplicate URL's from the same website, we try to make sure that the links to the different pages get combined accurately.
But if you can fix the site architecture or use a 301, then you don't even leave it up to Google – you're just combining that on your side and it's much easier for all the search engines – not just Google.
So we do try to do a good job of taking the links to the non-original URL's and combining them – I wouldn't claim they were always perfect.
I think we do a very good job at it but I wouldn't say that we do 100% all the time.
So if you can fix it on your side, that's the ideal situation.
Will keep working because usually when you specify rel equals canonical, we can combine all of those links and we do a very good of that.
But the more we can find out duplicate URL's and resolve that earlier on in the process, either through 301's or better site architecture, the better it is for everybody.